Vous êtes sur la page 1sur 892

Biotechnology for Fuels and Chemicals

The Twenty-Fourth Symposium

Presented as Volumes 105-108


of Applied Biochemistry and Biotechnology
Proceedings of the Twenty-Fourth Symposium
on Biotechnology for Fuels and Chemicals
Held April 28-May 1, 2002, in Gatlinburg, TN

Sponsored by
US Department of Energy' s Oftice of Fuels Development
and the Oftice of Industrial Technologies
Oak Ridge National Laboratory
National Renewable Energy Laboratory
Applied CarboChemicals
BBI International
Cargill, Inc.
Cargill Dow LLC
CIFAR- University of California, Davis
Corn Refiners Association
Diversa
E.l. DuPont de Nemours and Company, Inc.
Idaho National Engineering and Environmental Laboratory
Iogen Corporation
Katzen International, Inc.
Natural Resources Canada
Novozymes NS
Pacific Northwest National Laboratory
Tate and Lyle PLC
Tembec, Inc.
Tennessee V alley Authority Public Power Institute

Editors
Brian H. Davison and James W. Lee
Oak Ridge National Laboratory
Mark Finkelstein and James D. McMillan
National Renewable Energy Laboratory

* Springer Science+Business Media, LLC


Applied Biochemistry and Biotechnology
Volumes 105-108, Complete, Spring 2002
Copyright ©Springer Science+Business Media New York 2003
Originally published by Humana Press Inc. in 2003
Softcover reprint of the hardcover 1st edition 2003
ISBN 978-1-4612-6592-4 ISBN 978-1-4612-0057-4 (eBook)
DOI 10.1007/978-1-4612-0057-4
All Rights Reserved.
No part of this publication may be reproduced or transmitted in any
form or by any means, electronic or mechanical, including photocopy,
recording, or any information storage and retrieval system, without
permission in writing from the copyright owner.

Applied Biochemistry and Biotechnology is abstracted or indexed


regularly in Chemical Abstracts, Biologica[ Abstracts, Current Contents, Science
Citation Index, Excerpta Medica, Index Medicus, and appropriate related
compendia.
Introduction to the Proceedings
of the Twenty-Fourth Symposium
on Biotechnology for Fuels and Chemicals

BRIAN H. DAVISON
Oak Ridge National Laboratory

MARK FINKELSTEIN
National Renewable Energy Laboratory

The Twenty-Fourth Symposium on Biotechnology for Fuels and


Chemicals was held April 27-May 1, 2002 in Gatlinburg, TN. The sympo-
sium continues to have an interdisciplinary focus on bioprocessing, with
multidisciplinary interests; and remains the preeminent forum for
exchange of information and technology, for updating of current trends in
biotechnology, and for meeting active participants and organizations in
the area of sustainable fuels and chemicals production. This annual sym-
posium focuses on advances in biotechnology to produce high-volume,
low-price products from renewable resources, as well as to improve the
environment. Session topics include advanced feedstock production and
processing, enzyme and microbial biocatalysts, bioprocess research and
development, opportunities in biorefineries, commercialization of bio-
based products, and a number of other special topics sessions.
This year was particularly exciting due to the number of projects and
concepts that are being commercialized. These large commercialization
projects (e.g., lactic acid and propanediol, as well as ethanol) demonstrate
the commitment to this technology and aid in the excitement for the newer
concepts still in the research phase. Hurdles remain to the larger entry of
the application of biotechnology in commodities-in particular, the need to
move beyond corn sugar as the primary feedstock into lignocellulosics.
Participants from academic, industrial, and government venues con-
vened to discuss the latest research breakthroughs and results in biotech-
nology to improve the economics of producing fuels and chemicals. A
total of 295 attendees represent a 50% increase over the 2000 conference
in Gatlinburg. Of this total, 49% were from the academic sector (which
included 69 students), 32% from industry, and 19% from government.
III
iv Introduction

The increased attendance required concurrent sessions for the 48 oral


presentations and 190 submitted posters (for more details see Website:
www.ct.ornl.gov/symposium).
Attendees came from Australia, Austria, Belgium, Brazil, Canada,
China, Denmark, Finland, Germany, Hungary, India, Japan, Korea, Mexico,
The Netherlands, Russia, South Korea, Spain, Sweden, Turkey, and Ven-
ezuela, as well as from the United States.
This international perspective was continued in a Special Topic Ses-
sion sponsored by the International Energy Agency (lEA) Bioenergy Pro-
gram on Biofuels and chaired by Jack Saddler and David Gregg from the
University of British Columbia. Several of the 10 member countries in this
network are approaching Demonstrations of the Biomass-to-Ethanol pro-
cess and have a range of more fundamental projects that look at various
aspects of pretreatment, enzymatic hydrolysis, fermentation, and lignin
utilization. Presenters from several of the participating countries described
their country's biomass-to-ethanol projects, and differential factors such as
the type of biomass available, the maturity of the wood or agricultural
processing industry, and the willingness of government to bear the risk/
cost of development and demonstration.
The challenges other than technical for the sustained growth in this
area were also recognized by a Special Topic Session on Educational Needs
for a Biobased Bioprocess Industry (BPI), organized by R. Mark Worden of
Michigan State University. This workshop (1) solicited input from industry
on training needs to update existing employees and to prepare new employ-
ees for jobs in the BPI; (2) discussed new educational models being used,
including Internet course offerings and multidisciplinary training pro-
grams; and (3) identified new opportunities for university-lab-industry
interactions in support of a biobased economy.
The 2002 Charles D. Scott award for Distinguished Contributions in
the field of Biotechnology for Fuels and Chemicals was presented to Sharon
P. Shoemaker. Dr. Shoemaker has contributed toward our understanding
of cellulytic enzymes for over 20 years, with benchmark research at both
Cetus and Genencor. In 1991, she joined UC Davis as founder and executive
director of the California Institute of Food and Agricultural Research
(CIFAR), where she now brings together a wide variety of organizations
and individuals, forming collaborations and project teams to conduct
applied research and conducting conferences to disseminate state-of-the-
art knowledge in emerging technology areas. She is involved in state,
national and international biomass projects involving a number of feed-
stocks, microbial and enzyme systems, separation systems, and products.
This award was created to honor Dr. Charles D. Scott, founder of this Sym-
posium and its chair for the first 10 years.
Introduction v

Session Chairpersons
Session 1: Feedstock Production, Genetic Modification,
and Processing
Chairs: Sharon Shoemaker, University of California, Davis, CA
Lynn Wright, Oak Ridge National Laboratory, Oak Ridge, TN
Session 2: Microbial Catalysis and Metabolic Engineering
Chairs: Nancy Ro, Purdue University, West Lafayette, IN
Francisco Roberto, Idaho National Engineering
and Environmental Laboratory, Idaho Falls, ID
Session 3: Bioprocessing Research
Chairs: Mark Eiteman, University of Georgia, Athens, GA
Amit Vasavada, Diversa Corporation, San Diego, CA
Session 4: Genetics and Genomics in Bioenergy and Bioproducts
Chairs: Thomas W. Jeffries, USDA Forest Products Laboratory,
Madison, WI
James McLaren, Inverizon International, Inc.,
Chesterfield, MO
Session 5: Industrial Bio-Based Product Development

Session 6: Enzyme Biocatalysis and Production


Chairs: Joel Cherry, Novozymes Biotech, Inc., Davis, CA
Robert DiCosimo, DuPont, Wilmington, DE
Session 7: Downstream Processing of Bioproducts
Chairs: Michael Cockrem, KiwiChem International, Inc.,
Madison, WI
Cesar Santana, UNICAMp, Sao Paulo, Brazil

Organizing Committee
Brian Davison, Conference Chair, Oak Ridge National Laboratory,
Oak Ridge, TN
Mark Finkelstein, Conference Co-Chair, National Renewable Energy
Laboratory, Golden, CO
Bill Apel, Idaho National Engineering and Environmental Labora-
tory, Idaho Falls, ID
Doug Cameron, Cargill, Minneapolis, MN
Tom Jeffries, USDA Forest Service, Madison, WI
James Lee, Oak Ridge National Laboratory, Oak Ridge, TN
Lee Lynd, Dartmouth College, Hanover, NH
Jim McMillan, National Renewable Energy Laboratory, Golden, CO
Dale Monceaux, Katzen International, Inc., Cincinnati, OH
vi Introduction
Mark Paster, US Department of Energy, Washington, DC
Jack Saddler, University of British Columbia, Vancouver, British
Columbia, Canada
Valerie Sarisky-Reed, US Department of Energy, Washington, DC
Sharon Shoemaker, University of California, Davis, CA
David Short, E. l. DuPont de Nemours & Co., Newark, DE
Jeff Tolan, Iogen Corporation, Ontario, Canada
Nancy Watlington, Oak Ridge National Laboratory, Oak Ridge, TN
Liz Willson, National Renewable Energy Laboratory, Golden, CO
Charles Wyman, Dartmouth College, Hanover, NH
Guido Zacchi, Lund University, Lund, Sweden
Gisella Zanin, State University of Maringa, Maringa, PR, Brazil

Acknowledgments
The continued success of the Symposium is due to the many partici-
pants, organizers, and sponsors, but is also a success and pleasure due to
the diligent and creative staff. In particular, Nancy Watlington of ORNL,
the conference secretary and Liz Willson of NREL, the assistant conference
secretary, provided advice, persistence, and unfailing good humor.
Dr. John Barton also contributed greatly to the website design and imple-
mentation. Other ORNL staff included: Norma Cardwell, Angie Fincher,
Tony McBee, Norm Kurtz, Sandie Jones, and Sadie Drescher.
Oak Ridge National Laboratory is operated for the US Department of
Energy by UT-Battelle, LLC under contract DE-ACOS-000R2272S.
The National Renewable Energy Laboratory is operated for the US
Department of Energy by Midwest Research Institute, Battelle, and Bechtel
under contract DE-AC36-99GOI0337.
The submitted manuscript has been authored by a contractor of the
US Government under contract DE-ACOS-OOOR2272S. Accordingly, the
US Government retains a nonexclusive, royalty-free license to publish or
reproduce the published form of this contribution, or allow others to do
so, for US Government purposes.

Other Proceedings in this Series


1. "Proceedings of the First Symposium on Biotechnology in Energy Production and
Conservation" (1978), Biotechnol. Bioeng. Symp. 8.
2. "Proceedings of the Second Symposium on Biotechnology in Energy Production
and Conservation" (1980), Biotechnol. Bioeng. Symp. 10.
3. "Proceedings of the Third Symposium on Biotechnology in Energy Production
and Conservation" (1981), Biotechnol. Bioeng. Symp. 11.
4. "Proceedings of the Fourth Symposium on Biotechnology in Energy Production
and Conservation" (1982), Biotechnol. Bioeng. Symp. 12.
5. "Proceedings of the Fifth Symposium on Biotechnology for Fuels and Chemicals"
(1983), Biotechnol. Bioeng. Symp. 13.
Introduction vii
6. "Proceedings of the Sixth Symposium on Biotechnology for Fuels and Chemi-
cals" (1984), Biotechnol. Bioeng. Symp. 14.
7. "Proceedings of the Seventh Symposium on Biotechnology for Fuels and Chemi-
cals" (1985), Biotechnol. Bioeng. Symp. 15.
8. "Proceedings of the Eigth Symposium on Biotechnology for Fuels and Chemi-
cals" (1986, Biotechnol. Bioeng. Symp. 17.
9. "Proceedings of the Ninth Symposium on Biotechnology for Fuels and Chemi-
cals" (1988), Appl. Biochem. Biotechnol. 17,18.
10. "Proceedings of the Tenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1989), Appl. Biochem. Biotechnol. 20,21.
11. "Proceedings of the Eleventh Symposium on Biotechnology for Fuels and Chemi-
cals" (1990), Appl. Biochem. Biotechnol. 24,25.
12. "Proceedings of the Twelfth Symposium on Biotechnology for Fuels and Chemi-
cals" (1991), Appl. Biochem. Biotechnol. 28,29.
13. "Proceedings of the Thirteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1992), Appl. Biochem. Biotechnol. 34,35.
14. "Proceedings of the Fourteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1993), Appl. Biochem. Biotechnol. 39,40.
15. "Proceedings of the Fifteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1994), Appl. Biochem. Biotechnol. 45,46.
16. "Proceedings of the Sixteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1995), Appl. Biochem. Biotechnol. 51,52.
17. "Proceedings of the Seventeenth Symposium on Biotechnology for Fuels and
Chemicals" (1996), Appl. Biochem. Biotechnol. 57,58.
18. "Proceedings of the Eighteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1997), Appl. Biochem. Biotechnol. 63-65.
19. "Proceedings of the Nineteenth Symposium on Biotechnology for Fuels and Chemi-
cals" (1998), Appl. Biochem. Biotechnol. 70-72.
20 "Proceedings of the Twentieth Symposium on Biotechnology for Fuels and Chemi-
cals" (1999), Appl. Biochem. Biotechnol. 77-79.
21. "Proceedings of the Twenty-first Symposium on Biotechnology for Fuels and
Chemicals" (2000), Appl. Biochem. Biotechnol. 84-86.
22. "Proceedings of the Twenty-second Symposium on Biotechnology for Fuels and
Chemicals" (2001), Appl. Biochem. Biotechnol. 91-93.
23. "Proceedings of the Twenty-third Symposium on Biotechnology for Fuels and
Chemicals" (1997), Appl. Biochem. Biotechnol. 63-65.

This symposium has been held annually since 1978. We are pleased to
have the proceedings of the Twenty-Fourth Symposium currently pub-
lished in this special issue to continue the tradition of providing a record of
the contributions made.
The Twenty-Fifth Symposium will be May 4-7, 2003 in Breckenridge,
Colorado. For more information on the 24th and 25th Symposia, visit the
following Websites: http://www.ct.ornl.gov /symposium and http:/ /
nrel.gov /biotech_symposium. We encourage comments or discussions
relevant to the format or content of the meetings.
Applied Biochemistry and Biotechnology Vols. 105-108/ Spring 2003

CONTENTS

Introduction
Brian H. Davison and Mark Finkelstein ................................................ iii

SESSION I-FEEDSTOCK PRODUCTION, GENETIC MODIFICATION, AND PROCESSING

Introduction to Session 1
Sharon P. Shoemaker and Lynn L. Wright ............................................... 3
Rapid Biomass Analysis: New Tools for Compositional Analysis of Corn
Stover Feedstocks and Process Intermediates from Ethanol Production
Bonnie R. Hames, * Steven R. Thomas, Amie D. Sluiter,
Christine J. Roth, and David W. Templeton ....................................... 5
Xylem-Specific and Tension Stress-Responsive Expression
of Cellulose Synthase Genes from Aspen Trees
Chandrashekhar P. Joshi * ........................................................................ 17
Microbial Pretreatment of Biomass: Potential for Reducing Severity
of Thermochemical Biomass Pretreatment
Fred A. Keller, Jenny E. Hamilton, and Quang A. Nguyen* ................ 27
Physical Separation of Straw Stem Components to Reduce Silica
J. Richard Hess, *David N. Thompson, Reed L. Hoskinson,
Peter G. Shaw, and Duane R. Grant ................................................... 43
Application of a Depolymerization Model for Predicting
Thermochemical Hydrolysis of Hemicellulose
Todd Lloyd and Charles E. Wyman* ....................................................... 53
Dilute-Sulfuric Acid Pretreatment of Corn Stover
in Pilot-Scale Reactor: Investigation of Yields, Kinetics,
and Enzymatic Digestibilities of Solids
Daniel J. Schell, * Jody Farmer, Millie Newman,
and James D. McMillan ......................................................................... 69
Hydrothermal Pretreatment Conditions to Enhance
Ethanol Production from Poplar Biomass
Maria Jose Negro, Paloma Manzanares, Ignacio Ballesteros,
Jose Miguel Oliva, Araceli Cabanas, and Mercedes Ballesteros* ..... 87
Estimation of Temperature Transients
for Biomass Pretreatment in Tubular Batch Reactors
and Impact on Xylan Hydrolysis Kinetics
Suzanne L. Stuhler and Charles E. Wyman* ........................................ 101
*For papers with multiple authorship, the asterisk identifies the author to whom corre-
spondence and reprint requests should be addressed.

ix
x Contents
Effect of Sulfuric and Phosphoric Acid Pretreatments
on Enzymatic Hydrolysis of Corn Stover
Byung-Hwan Um, M. Nazmul Karim, * and Linda L. Henk .............. 115

Combined Use of H 2S04 and S02 Impregnation


for Steam Pretreatment of Spruce in Ethanol Production
Johanna Soderstrom, Linda Pilcher,
Mats Galbe, and Guido Zacchi* ........................................................ 127

Effect of Lignocellulosic Degradation Compounds from Steam


Explosion Pretreatment on Ethanol Fermentation
by Thermotolerant Yeast Kluyveromyces marxianus
Jose Miguel Oliva, Felicia Saez, Ignacio Ballesteros,
Alberto Gonzalez, Maria Jose Negro,
Paloma Manzanares, and Mercedes Ballesteros* .......................... 141

Enzymatic Hydrolysis of Ammonia-Treated Rice Straw


Betzabe Sulbaran-de-Ferrer, Marielena Aristiguieta,
Bruce E. Dale, Alexis Ferrer, *
and Graciela Ojeda-de-Rodriguez .................................................... 155

Effects of Temperature and Moisture on Dilute-Acid


Steam Explosion Pretreatment of Corn Stover
and Cellulase Enzyme Digestibility
Melvin P. Tucker, * Kyoung H. Kim,
Mildred M. Newman, and Quang A. Nguyen .................................. 165

Sugar Monomer and Oligomer Solubility:


Data and Predictions for Application to Biomass Hydrolysis
Matthew C. Gray, Alvin O. Converse,
and Charles E. Wyman* ...................................................................... 179

Influence of Pressure in Ethanol/Water


Pulping of Sugarcane Bagasse
Adilson R. Gon~alves* and Denise S. Ruzene ..................................... 195

Post-Harvest Processing Methods for Reduction


of Silica and Alkali Metals in Wheat Straw
David N. Thompson, * Peter G. Shaw, and Jeffrey A. Lacey ............. 205

Composition and Ethanol Production


Potential of Cotton Gin Residues
Foster A. Agblevor, * Sandra Batz, and Jessica Trumbo .................... 219

Wood-Ethanol for Climate


Change Mitigation in Canada
Peter J. Graham, * David J. Gregg, and John N. Saddler ..................... 231
Contents xi
SESSION 2-MICROBIAL CATALYSIS AND METABOLIC ENGINEERING

Introduction to Session 2
Nancy W. Y. Ho and Francisco F. Roberto ............................................ 245
Saccharification of Marine Microalgae
Using Marine Bacteria for Ethanol Production
Mitsufumi Matsumoto, * Hiroko Yokouchi, Nobukazu Suzuki,
Hiroshi Ohata, and Tadashi Matsunaga .......................................... 247
D-Xylose Transport by Candida succiphila
and Kluyveromyces marxianus
Boris U. Stambuk, * Mary Ann Franden,
Arjun Singh, and Min Zhang .............................................................. 255
Molecular Characterization of a Gene
for Aldose Reductase (CbXYL1) from Candida boidinii
and Its Expression in Saccharomyces cerevisiae
Min Hyung Kang, Haiying Ni,
and Thomas W. Jeffries* ...................................................................... 265
Changing Flux of Xylose Metabolites by Altering Expression
of Xylose Reductase and Xylitol Dehydrogenase
in Recombinant Saccharomyces cerevisiae
Yong-Su lin and Thomas W. Jeffries* .................................................... 277
Effect of Media on Spore Yield and Thermal Resistance
of Bacillus stearothermophilus
Thereza Christina Vessoni Penna, *
Irene A. Machoshvili, and Marina Ishii ........................................... 287
Cassava Flour Wastewater as a Substrate
for Biosurfactant Production
Marcia Nitschke* and Glaucia M. Pastore ......................................... 295
A New Oxygen Sensitivity and Its Potential Application
in Photosynthetic H2 Production
James W. Lee* and Elias Greenbaum .................................................... 303

SESSION 3-BIOPROCESSING RESEARCH

Introduction to Session 3
Mark A. Eiteman and Amit Vasavada .................................................. 317
Optimization of S02-Catalyzed Steam Pretreatment
of Com Fiber for Ethanol Production
Renata Bura, Rodney J. Bothast, Shawn D. Mansfield,
and John N. Saddler, * .......................................................................... 319
xii Contents
A Comprehensive Kinetic Model for Dilute-Acid
Hydrolysis of Cellulose
Qian Xiang, Jun Seok Kim, and Y. Y. Lee* ........................................... 337
Effect of Agitation on Removal of Acetic Acid
from Pretreated Hydrolysate by Activated Carbon
Sarah A. Priddy and Thomas R. Hanley* ............................................. 353

Enzymatic Digestibility of Used Newspaper Treated


with Aqueous Ammonia-Hydrogen Peroxide Solution
Sung Bae Kim* and Nam Kyu Moon ..................................................... 365

Continuous Production of Butanol


from Starch-Based Packing Peanuts
Thaddeus C. Ezeji, Marisa Groberg,
Nasib Qureshi, * and Hans P. Blaschek ............................................ 375
Measurement of Rheological Properties
of Com Stover Suspensions
Natalia V. Pimenova and Thomas R. Hanley* ................................... 383

Effect of Bacillus circulans Dl Thermostable Xylanaii V

on Biobleaching of Eucalyptus Kraft Pulp


Daniela A. Bocchini, Valquiria B. Damiano,
Eleni Gomes, and Roberto Da Silva* ................................................ 393

Fungi Allergens Produced by Solid-State Fermentation Process:


Optimization and Allergen Characterization
Salah D. M. Hasan, Walderez Gambale,
Ricardo L. Zollner, and Maria H. A. Santana* ............................... 403
Hybrid Model for an Enzymatic Reactor:
Hydrolysis of Cheese Whey Proteins by Alcalase
Immobilized in Agarose Gel Particles
Ruy Sousa Jr., Miriam M. Resende, Raquel L. C. Giordano,
and Roberto C. Giordano* .................................................................. 413
Preliminary Investigation of Fungal Bioprocessing
of Wheat Straw for Production
of Straw-Thermoplastic Composites
David N. Thompson, * Tracy P. Houghton, Jeffrey A. Lacey,
Peter G. Shaw, and J. Richard Hess .................................................. 423
Simulation of Aerated Lagoon Using Artificial Neural
Networks and Multivariate Regression Techniques
Karla Patricia Oliveira-Esquerre, * Aline C. da Costa,
Roy Edward Bruns, and Milton Mori ............................................... 437
Contents xiii
Simplistic Modeling Approach to Heterogeneous Dilute-Acid
Hydrolysis of Cellulose Microcrystallites
Par O. Pettersson, * Robert W. Torget, Robert Eklund,
Qian Xiang, Y. Y. Lee, and Guido Zacchi ......................................... 451
Cellulosic Fuel Ethanol: Alternative Fermentation Process Designs
with Wild-Type and Recombinant Zymomonas mobilis
Hugh G. Lawford* and Joyce D. Rousseau .......................................... 457
Effects of Pressure Pulsation on Oxygen Transfer Rate
Measured by Sulfite Method
Wei-Cho Huang, * Cheng S. Gong, and George T. Tsao ...................... 471
Permeation Associated with Three-Phase Partitioning Method
on Release of Green Fluorescent Protein
Thereza Christina Vessoni Penna, * Eb Chiarini,
and Adalberto Pessoa Junior ............................................................. 481
Comparison of Growth Characteristics of Panax ginseng
Hairy Roots in Various Bioreactors
Gwi-Taek Jeong, Don-Hee Park, * Baik Hwang,
and Je-Chang Woo ............................................................................... 493
Heterogeneous Aspects of Acid Hydrolysis of a-Cellulose
Qian Xiang, Y. Y. Lee, * Par O. Pettersson, and Robert W. Torget ..... 505
Characterization of Molecular Weight Distribution of Oligomers
from Autocatalyzed Batch Hydrolysis of Xylan
Xia Li, * Alvin O. Converse, and Charles E. Wyman .......................... 515
Conversion of Sugarcane Bagasse to Carboxylic Acids
Using a Mixed Culture of Mesophilic Microorganisms
Piyarat Thanakoses, Nagat Abd Alla Mostafa,
and Mark T. Holtzapple* .................................................................... 523
Alternative Approach for Utilization of Pentose Stream from
Sugarcane Bagasse by an Induced Flocculent Pichia stipitis
Heizir F. de Castro, * Samuel C. Oliveira,
and Sandra A. Furlan3 .......................................................................... 547
Hydrogen Production from Paper Sludge Hydrolysate
Zs6fia Kadar, Truus de Vrije, Miriam A. W. Budde,
Zsolt Szengyel, Kati Reczey, and Pieternel A. M. Claassen* ....... 547
Single-Stage Anaerobic Codigestion for Mixture Wastes
of Simulated Korean Food Waste and Waste Activated Sludge
Nam Hyo Heo, Soon Chul Park, * Jin Suk Lee,
Ho Kang, and Don Hee Park .............................................................. 557
xiv Contents
Biosorption and Desorption of Copper (II) Ions by Bacillus sp.
Waihung Lo, '" Lau Mei Ng, Hong Chua, Peter H. F. Yu,
Shirley N. Sin, and Po-Keung Wong ................................................. 567
Flow Field in a Shrinking-Bed Reactor
for Pretreatment of Cellulosic Biomass
Yinkun Wan and Thomas R. Hanley'" ................................................... 581
Lactic Acid Production Through Cell-Recycle
Repeated-Batch Bioreactor
Hurok Oh, Young-Jung Wee, Jong-Sun Yun,
and Hwa-Won Ryu'" ............................................................................ 603
Limits for Alkaline Detoxification of Dilute-Acid
Lignocellulose Hydrolysates
Nils-Olof Nilvebrant, '" Per Persson, Anders Reimann,
Filipe de Sousa, Lo Gorton, and Leif J. Jonsson ............................. 615

SESSION 4-GENETICS AND GENOMICS IN BIOENERGY AND BIOPRODUCTS

Introduction to Session 4
James S. McLaren and Thomas W. Jeffries ........................................... 631

SESSION 5-DEVELOPMENT OF BIOBASED PRODUCTS

Introduction to Session 5 ............................................................................ 635


Production of Lactic Acid from Food Wastes
Kwang II Kim, Woo Kyung Kim, Deok Ki Seo,
In Sang Yoo, Eun Ki Kim, and Hyon Hee Yoon'" .............................. 637

Microbiologic Oxidation of Isosafrole into Piperonal


Alberdan Silva Santos, Nei Pereira Jr., Iracema 1. da Silva,
Maria Ines Sarquis, and Octavio A. C. Antunes, '" .......................... 649

Breathing Air from Protein Foam


Douglas M. Ackermann Jr., David N. Jewell,
Matthew L. Stedman, Vorakan Burapatana,
Prabhani V. Atukorale, Michelle L. Pinson,
Alison E. Wardle, Wenyan Zhu, and Robert D. Tanner'" .................. 659

SESSION 6-ENZYMATIC PRODUCTION AND CONVERSIONS

Introduction to Session 6
Joel R. Cherry and Robert DiCosimo ..................................................... 675
Contents xv
Partial Purification and Kinetic Characterization
of Acid Phosphatase from Garlic Seedling
Begiim Yenigiin* and Yiiksel Giivenilir ................................................. 677
Automated Filter Paper Assay
for Determination of Cellulase Activity
Stephen R. Decker, * William S. Adney, Edward Jennings,
Todd B. Vinzant, and Michael E. Himmel ....................................... 689
Adsorption of Components of Enzymatic Synthesis
of Ampicillin on Different Hydrophobic Resins
Marcelo F. Vieira, Marlei Barboza,
and Raquel de Lima C. Giordano* .................................................... 705
Xylanase Production by Trichoderma reesei Rut C-30 on Rice Straw
Alejandro Colina, Betzabe Sulbaran-de-Ferrer,
Cateryna Aiello, and Alexis Ferrer* .................................................. 715
A Different Method of Measuring and Detecting Mono-
and Dioxygenase Activities:
Key Enzymes in Hydrocarbon Biodegradation
Roberto Zazueta-Sandoval, * Vanesa Zazueta Novoa,
Hortencia Silva Jimenez, and Roberto Cabrera Ortiz ................... 725
Xylanase Production by Penicillium canescens
IO-IOc in Solid-State Fermentation
Yasser Bakri, Philippe Jacques, * and Philippe Thonart .................... 737
Protease Production by Streptomyces sp. Isolated
from Brazilian Cerrado Soil: Optimization of Culture
Medium Employing Statistical Experimental Design
Luciana A. I. de Azeredo, Leda R. Castilho, Selma G. F. Leite,
Rosalie R. R. Coelho, and Denise M. G. Freire* .............................. 749
Synthesis of Monocaprin Catalyzed by Lipase
Marco A. M. Da Silva, Vitoria C. Medeiros,
Marta A. P. Langone, and Denise M. G. Freire* ............................. 757
Effect of Dose of Xylanase on Bleachability
of Sugarcane Bagasse Ethanol/Water Pulps
Denise S. Ruzene and Adilson R. Gon~alves* ..................................... 769
CelF of Orpinomyces PC-2 has an Intron
and Encodes a Cellulase (CeIF) Containing
a Carbohydrate-Binding Module
Huizhong Chen, Xin-Liang Li, * David L. Blum,
Eduardo A. Ximenes, and Lars G. Ljungdahl .................................. 775
xvi Contents
Partition Behavior and Partial Purification of Hexokinase
in Aqueous Two-Phase Polyethylene Glycol/Citrate Systems
George G. G. Oliveira, Daniel P. Silva, Ines Concei~iio Roberto,
Michele Vitolo, and Adalberto Pessoa Jr. * ....................................... 787

Effect of Aeration on Lignin Peroxidase Production


by Streptomyces viridosporus T7A
Leda M. F. Gottschalk, Ronaldo Nobrega,
and Elba P. S. Bon* ............................................................................... 799
Evaluation of Supports and Methods for Immobilization
of Enzyme Cyclodextringlycosyltransferase
Keli A. Sobral, Regina o. Rodrigues,
Rogerio D. Oliveira, Jose E. Olivo,
Flavio F. de Moraes, and Gisella M. Zanin* ................................... 809
Active Nuclear Shuffling System Using a Swollen
Conidium of Trichoderma reesei
Hideo Toyama, * Makiko Yano, Akane Gisushi,
Takeshi Hotta, and Nobuo Toyama ................................................. 821

SESSION 7-DoWNSTREAM PROCESSING OF BIOPRODUCTS

Introduction to Session 7
Michael Cockrem and Cesar C. Santana ............................................... 827

Phenylboronate-Chitosan Resins for Adsorption


of I3-Amylase from Soybean Extracts
Eduardo J. Arruda and Cesar C. Santana* ............................................ 829
Ester Fuels and Chemicals from Biomass
Edwin S. Olson,' Ted R. Aulich,
Ramesh K. Sharma, and Ronald C. Timpe ....................................... 843
Purification of Glucose-6-Phosphate Dehydrogenase
from Baker's Yeast in Aqueous Two-Phase Systems
with Free Triazine Dyes as Affinity Ligands
Yan Xu, Michele Vitolo, Cristina Northfleet de Albuquerque,
and Adalberto Pessoa Jr. * ................................................................... 853
Continuous Clavulanic Acid Adsorption Process
Renata M. R. G. Almeida, * Marlei Barboza,
and Carlos o. Hokka ........................................................................... 867
Extraction of Nutraceuticals from Milk Thistle:
1. Hot Water Extraction
Jose F. Alvarez Barreto, Sunny N. Wallace,
Danielle Julie Carrier, and Edgar C. Clausen* ................................ 881
Contents xvii
Extraction of Nutraceuticals from Milk Thistle:
Part II. Extraction with Organic Solvents
Sunny N. Wallace, Danielle Julie Carrier,
and Edgar C. Clausen* ......................................................................... 891
Effect of Lidocaine on Ovalbumin
and Egg Albumin Foam Stability
Vorakan Burapatana, Ernest E. Butler, Gaurav Chauhan,
Sean Hartig, Helen Kincaid, Tong Wang,
Shayrizal Samsudin, and Robert D. Tanner* .................................. 905
Estimation of Solubility Effect on the Herbicide Controlled-
Release Kinetics from Lignin-Based Formulations
Felix M. Pereira, Adilson R. Gonfalves, Andre Ferraz,
Flavio T. Silva, and Samuel C. Oliveira* ........................................ 913

Author Index ................................................................................................... 921


Subject Index ................................................................................................... 925
SESSION 1
Feedstock Production,
Genetic Modification, and Processing
Session 1

Feedstock Production,
Genetic Modification, and Processing

SHARON P. SHOEMAKER 1 AND LYNN L. WRIGHT2

University of California, Davis, CA; and


1
20ak Ridge National Laboratory, Oak Ridge, TN

This session covered a very wide range of topics from an overview


of the reorganization of the DOE Biomass Program to chemical pretreat-
ment alternatives and discussion of a kinetics model for pretreatment.
The majority of the oral presentations addressed ways of improving or
identifying feedstock quality. Reports related to improving feedstock
quality included (1) identifying the effects of corn variety on starch to
ethanol conversion, (2) identifying the genes that control cellulose char-
acteristics in aspen trees, (3) use of physical separation methods to im-
prove wheat straw quality, and (4) use of fungal pretreatment during
storage and handling to improve digestibility of lignocellulosic feed-
stocks. The report of a feedstock composition analysis technique, Near
InfraRed (NIR), offered the promise of being able to select feedstocks
based on quality either at the plant gate or, better yet, in the field where
it can enhance genetic improvement programs.
Valerie Sarisky-Reed of DOE noted in her overview that the market
potential for ethanol from corn stover is as high as 10 billion gallons if cost
of production can be reduced. Reduction of cellulase enzyme costs was
identified as the number 1 priority for advancing sugars platform tech-
nology. Dr. Bothast of USDA reported that some varieties of corn grain
convert more efficiently to ethanol than others due to differences in per-
centage and availability of starch. He also reports that utilization of corn
fiber could increase the amount of ethanol per bushel of corn by 10%.
Bonnie Hames of the National Renewable Energy Laboratory presented
information on a Near InfraRed (NIR) spectroscopy for rapid biomass
analysis to identify composition of biomass resources. NIR could both
facilitate purchase of feedstocks with preferred qualities and help breed-
ers develop plants with preferred qualities. Work now is focusing on
calibrating the NIR techniques against standard wet-chemistry tech-
niques, a process that must be repeated for each type of feedstock to be

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 3 Vol. 105-108, 2003


4 Introduction to Section 1

analyzed. Chandrashekhar Joshi of Michigan Technological University is


working to identify the suite of genes responsible for producing the differ-
ent types of cellulose found in aspen trees with the ultimate goal of creat-
ing genetically modified trees with desirable feedstock quality. Work
reported by Richard Hess of Idaho National Engineering and Environ-
mental Laboratory is focusing on improving feedstock quality of wheat
straw by physical separation of feedstock components and by genetic
selection. Quang Nguyen and others from the National Renewable Energy
Laboratory work on modifying feedstock quality during the storage and
handling phase by using fungal pretreatment. Studies on the effectiveness
of different fungi are being conducted. The final two presentations
addressed more traditional methods of pretreatment, with Todd Lloyd
and Charlie Wyman from Dartmouth developing a kinetics model based
on a depolymerization mechanism to describe how hemicellulose may be
degraded and solubilized. Peter van Walsum and collaborators of the Uni-
versity of Texas described experimental and model results suggesting that
pretreatment with carbonic acid eliminates some of the problems associ-
ated with sulfuric acid with comparable energy and capital costs.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0005/$20.00

Rapid Biomass Analysis


New Tools for Compositional Analysis of Corn Stover Feedstocks
and Process Intermediates from Ethanol Production

BONNIE R. HAMES,* STEVEN R. THOMAS, AMIE D. SLUITER,


CHRISTINE J. ROTH, AND DAVID W. TEMPLETON
National Bioenergy Center, National Renewable Energy Laboratory,
1617 Cole Blvd., Golden, CO 80401, E-mail: bonnie_hames@nrel.gov

Abstract
New, rapid, and inexpensive methods that monitor the chemical compo-
sition of corn stover and corn stover-derived samples are a key element to
enabling the commercialization of processes that convert stover to fuels and
chemicals. These new techniques combine near infrared (NIR) spectroscopy
and projection to latent structures (PLS) multivariate analysis to allow the
compositional analysis of hundreds of samples in 1 d at a cost of about $10
each. The new NIRjPLS rapid analysis methods can also be used to support
a variety of research projects that would have been too costly to pursue by
traditional methods.
Index Entry: Corn stover; near infrared spectroscopy; projection to latent
structures; feedstock; multivariate analysis.

Introduction
Robust analytical methods are needed to support and enable biomass
conversion processes because of the heterogeneity that is an inherent prop-
erty of biomass. The chemical composition of a biomass feedstock varies as
a function of many factors including plant genetics, growth environment,
harvesting method, and storage. Many biomass feedstocks are residues of
another process, which introduces the varying efficiency in the original
process as an additional source of compositional variance. All of these
sources of compositional variance are difficult to control; however, the
composition of a given feedstock can be measured in real time and that
information can be used to adjust process conditions for optimal conver-
sion. The rapid, inexpensive compositional analysis methods described
here are examples of new tools that will be needed for the commercializa-
tion of processes that convert biomass into fuels and valuable chemicals.
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 5 Vol. 105-108,2003


6 Hames et al.

Standard wet chemical methods for the chemical characterization of


biomass feedstocks and biomass-derived materials have been validated
through the International Energy Agency and are available through the
American Society for Testing and materials (ASTM) (see Website: www.
astm.orgl cgi-bin/SoftCart.exe/index.shtml?E+mystore) (1). However,
these methods are not applicable in a commercial setting because they are
very expensive (labor intensive) and cannot provide the analysis informa-
tion in a time frame useful for process control. For example, a complete
analysis using standard wet chemical methods costs $800-$2000 per sample
and results are not available for days, sometimes weeks. By contrast, the
new methods reported here can perform the same analysis for about $10 per
sample and the results are available in a time frame relevant for process
control, meaning that the information can be used to make the process
adjustments necessary for steady-state production. One approach to reduc-
ing the time and cost of compositional analysis is the development of rapid
analysis methods that use multivariate analysis software to extract chemical
information from easily obtained spectroscopic data. Rapid analysis meth-
ods match the precision and accuracy of their calibration methods so the
savings are obtained without loss of precision or accuracy (2). New tech-
niques, such as rapid analysis, are needed to provide analytical support for
large-scale processes that convert biomass to fuels and chemicals.
As shown in Fig. 1, rapid biomass analysis can be useful at many
stages of an industrial process. The methods described here involve the
compositional analysis of biomass feedstocks and biomass-derived solids
produced during ethanol production. These tools can be used to character-
ize feedstock as it enters the reactor. If necessary, rapid analysis can be used
to guide feedstock blending. Monitoring chemical changes during the pro-
cessing of biomass provides feed-forward and feed-backward information
that can be used to ensure that the process maintains a steady state in spite
of the feedstock variability. Finally, process residues and products can be
evaluated to assess overall process efficiency. As more samples are ana-
lyzed, information can be obtained about the composition of an "ideal
feedstock." Field-mobile instruments can be calibrated for use as purchas-
ing tools. Buyers can obtain compositional information about a biomass
feedstock at the point of purchase. Feedstock prices can be based on quality
instead of weight. Living plants, young plants, and perhaps even seeds can
be evaluated and selected for desirable characteristics and production
potential. This article describes the steps necessary to develop rapid biom-
ass analysis for all of these applications.

Development of Rapid Analysis Method


Several steps are involved in the development of rapid analysis
method, including gathering appropriate calibration samples, chemical
characterization of the calibration samples, developing spectroscopic meth-
ods for the rapid technique, projection to latent structures (PLS) regression

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Rapid Biomass Analysis 7

Process Product

Milled Feed Pretreatment


Intermediates
Residues
Fig. 1. Applications for rapid biomass analysis in biomass conversion processes.

and the validation of the PLS algorithm, and the development of quality
assurance/ quality control procedures including guidelines for appropri-
ate application of the new methods.
Calibration Samples
The first step in developing a new method is the gathering of appro-
priate calibration samples. A minimum of 30 unique samples is needed for
preliminary methods, and calibration sets for robust methods usually con-
tain 100-300 well-characterized samples. Collecting and characterizing a
good calibration set costs about $300,000, which is by far the most expen-
sive and time-consuming step in method development.
Calibration samples should have compositions similar to the samples
to be analyzed. If possible, the calibration set should include samples that
represent all known sources of compositional variance for that material.
Another feature that is essential to a good calibration set is independent
variance of the different constituents. Calibration sets should be checked
for strong correlations among major constituents. Since the composition of
plants is determined by physiology and plant viability, biomass feedstock
samples with independently varying constituents are sometimes difficult
to find. Several calibration samples can often be made from one feedstock
by manually adding or removing specific plant tissues such as leaves or
nodes. The extremes of a corn stover calibration set can be found within a
single plant, in the various plant tissue types. For example, no whole corn
stover feedstock will have as much protein as a sample consisting only of
leaves. A research environment is ideal for the generation and collection of
process calibration samples since reaction parameters are often skewed
for experimental purposes. Larger-scale processes run at steady state, so
development of a rapid analysis method to support these processes should
begin during the reactor shake-down phase. Hundreds of samples may
need to be evaluated to identify 30 unique samples for a preliminary cali-
bration. The preliminary method can then be used to identify the best
samples for expanding and improving the calibration set.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
8 Hames et al.

Rapid biomass analysis methods require calibration using robust and


accurate methods. Corn stover is a whole-plant feedstock whose major
constituents include cellulose, hemicellulose,lignin, protein, chlorophyll,
and structural inorganics (silica). The hemicellulose in corn plants is a
complex network consisting of an acetylated xylan backbone with
branches incorporating galactose, arabinose, and uronic acids. The lignin
in corn stover contains both syringyl and guaiacyl aromatic rings, and the
side chains are enriched in ester groups. Standard methods for the deter-
mination of cellulose, hemicellulose, and lignin content in biomass are
available from ASTM (see Website: www.astm.org/cgi-bin/SoftCart.exe/
index.shtml?E+mystore). Standard methods optimized for the composi-
tional analysis of corn stover are available from the National Renewable
Energy Laboratory (NREL) (see Website: www.ott.doe.gov /biofuels/
analyticaLmethods.html). Methods are available for the determination of
total solids, structural inorganic matter, soil, nonstructural sugars, chloro-
phyll, protein, structural carbohydrates, acid-soluble lignin, and acid-
insoluble lignin. Standard procedures are also posted for sample prepara-
tion and representative sampling of bulk feedstocks. Following these
methods, a complete compositional analysis can be obtained with mass
closures between 97 and 103%.
Rapid analysis methods look for patterns in spectroscopic data
that correlate with changes in composition. A diverse calibration set with
quality calibration data is required for an accurate and robust rapid analy-
sis method.
Spectroscopic Methods
Quality spectroscopy is the second essential component of method
development. Many different spectroscopic techniques can be used in
the development of rapid analysis methods. For biomass samples, near
infrared (NIR) spectroscopy offers several advantages. A wide variety of
instruments are commercially available including rugged spectrometers
designed for use outdoors or for process monitoring (see Website:
www.FOSS-Nirsystems.com and http://asdLcom/index.htm). Sample
handling devices have been developed specifically for the analysis of bulk
biomass samples and biomass-derived liquids and slurries. Newer ver-
sions of field-mobile instruments include spectrometers that fit in a back-
pack and reduced wavelength versions that resemble digital cameras and
run from palm pilots. These instruments collect spectroscopic data from
wavelengths in the visible and NIR regions of the electromagnetic spec-
trum from 400 to 2500 nm. Spectra in these regions contain signals from
C-H, O-H, N-H, and C=O bonds, so compositional information for cellu-
lose, hemicellulose, protein, and lignin is contained in the NIR spectra of
biomass samples. For biomass solids, NIR spectroscopy is usually done in
reflectance mode. This is a surface technique capable of penetrating only
about 150}l into the biomass surface. To obtain a representative spectrum
of biomass, samples must be ground to a uniform particle size or a large
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
Rapid Biomass Analysis 9

~ whole stover
0.9
-leaves
0.8
-leaf sheaths
0.7
-pith

-I>- internode

0.3

0.2

0.1

o~----------~----------~------------~----------~~
400 900 1400 1900 2400
wavelength (nm)

Fig. 2. NIR spectra of anatomic fractions of corn stover.

sample volume must be scanned during spectral collection. In most cases,


30-100 spectra are collected and averaged prior to PLS analysis. Reflec-
tance spectroscopy separates the detector from the NIR light source,
allowing sunlight to be used as the NIR source in field measurements. The
spectroscopic technique selected must contain the desired information
about the chemical composition of each sample. Since inorganic materials
do not absorb IR light, nonstructural materials such as soil cannot be
measured using NIR wavelengths. Structural inorganics can be measured
indirectly if they perturb the absorbance of adjacent organic compounds.
Figure 2 shows the NIR spectra of nine corn stover anatomic fractions.
These spectra contain all of the necessary information to determine the
chemical composition of these samples.
The selection of an appropriate spectroscopic method is the key to
cost reduction and speed of analysis. The final rapid analysis method will
extract chemical information from spectroscopic patterns. Once calibra-
tion is complete, compositional analysis becomes as fast and inexpensive
as the spectroscopic method.

Multivariate Analysis
Multivariate analysis connects the compositional data with the spec-
troscopic data. The PLS (sometimes called partial least squares) method
regresses the matrix of compositional information against NIR spectral
data collected from 400 to 2500 nm. In simplified terms, PLS analysis solves
hundreds of equations in thousands of variables to obtain a linear equation

Applied Biochemistry and Biotechnology Vol. 105-108,2003


10 Hames et al.

that translates spectroscopic data into compositional information. Multi-


variate analysis methods were designed for complex systems like those
found in biomass compositional analysis. Rapid analysis techniques using
NIR spectroscopy are common in industrial applications monitoring the
composition of animal feed, cheese, beer, and grain. The concept of NIRj
PLS analysis of biomass is not novel. Only the application of this proven
technology to biomass conversion and the development of calibrations
specific for energy feedstocks is new. Method development in this area has
great potential for high impact with low risk.
In support of the biomass conversion industries, rapid analysis
methods are being developed at NREL using a variety of spectrometers
and multivariate analysis software packages. Procedures are being devel-
oped for the standardization of spectra collected on different instruments
and for the accurate transfer of PLS equations between instruments and
software packages.

Rapid Analysis Methods for Corn Stover


The US Department of Energy has selected corn stover as a model
feedstock for its enzyme sugar platform work. In support of enzyme sugar
platform work, NREL has developed laboratory-based rapid analysis
methods for the compositional analysis of corn stover feedstocks and
corn stover-derived intermediates from a dilute-acid pretreatment pro-
cess. In development are calibrations measuring the composition of liq-
uid process intermediates and growing corn plants. The details and
calibration statistics for those methods will be published separately. In
this article, we describe applications of rapid analysis methods to genetic
screening and bioprocess development.
Feedstock Method
The calibration set for the compositional analysis of corn stover feed-
stocks described here contains 47 samples representing five locations and
three harvest years. The calibration set includes aged stover samples and
hand-separated anatomic fractions. The calibration range and the SE of
cross validation for each constituent are presented in Table 1. The accuracy
of the NIRjPLS method is illustrated in Fig. 3, which compares the compo-
sition determined by wet chemistry with that estimated by the NIRjPLS
equation. The diagonal line represents perfect agreement. The NIRjPLS
data are from a full cross-validation model, which provides a conservative
estimate of the accuracy of the final equation. The distribution of points
parallel to the line represents the calibration range. The distribution of the
points perpendicular to the line represents error in the NIRjPLS method.
Figure 3 demonstrates that the NIRjPLS methods provides a complete
compositional analysis for a wide range of samples with precision and
accuracy that match the wet chemical methods used to obtain the calibra-
tion data.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Rapid Biomass Analysis 11

Table 1
Calibration Ranges and SEs of Cross Validation for NIR/PLS
Method for Compositional Analysis of Corn Stover Feedstocks
Min Max SECV
Constituent (% drywt) (% dry wt) (% drywt)a

Glucan 26.9 48.0 1.445


Xylan 14.5 25.3 0.949
Lignin 10.8 27.4 1.120
Protein 0.0 9.2 0.828
SCinorg 0.0 13.6 1.029
Acetyl 0.0 4.5 0.200
Galactan 0.3 3.0 0.327
Arabinan 0.8 4.7 0.199
Mannan 0.0 1.5 0.136
Uronic_acids 2.2 4.4 0.1886
aSE of cross validation.

predicted vs measured for stover5b.eqa


50

40

35

~ 30
d.
IE
z 25. :091uoa n
.!' :Oxylan
lII:
!: 20 i61ignin
i- prolsin
!ucetyl ,
15 :0 8TI!.~.i.~n..l

o 5 16 20 25 30 35 40 50
wt"!. bv wet chemistry
Fig. 3. Comparison of corn stover feedstock composition as determined by wet
chemical and NIR/PLS methods.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


12 Hames et al.

rrh"'rvpti~1 Maximum Potential Ethan (gal per dry ton)

114

.. acetyl

80
• lignin

"#- C structural Innrrl~n""' 1


i 60

• minor sugars

40
cXylan

20 • gtucan

o
Colorado, 2000 lIinols, 2000 Iowa. 2000A Iowa, ZOOIA 10_ 20018

Fig. 4. Comparison of theoretical ethanol yields and composition of bulk corn stover
feedstocks.

Application of Feedstock Method


In addition to significant savings in time and money, rapid analysis
methods can be used to provide levels of information that were not previ-
ously available. For example, feedstock assessment and genetic studies
require the screening of hundreds, sometimes thousands, of samples. Such
studies would be too costly to pursue without the savings in time and cost
provided by a rapid analysis method, which permits approx 200 samples
to be analyzed from a bulk corn stover feedstock in 1 d at a cost of about
$2000. This enables large sample studies to be used to develop protocols for
the representative sampling of fields, bales, totes, and bags of feedstock.
The ability to accurately sample a bulk feedstock and to analyze
hundreds of samples for about $10 each provides a new tool that is being
used at NREL to assess the compositional variability of corn stover in the
United States. as a function of variety, geographical location, harvest time,
and collection method. Changes in feedstock composition during storage
can also be monitored. Feedstock composition will be reported as a range
of expected normal values. Significant differences have already been seen
in the composition of bulk corn stover feedstocks. Figure 4 compares com-
positional information for three samples from the 2000 harvest and two
from the 2001 harvest. The potential value of these bulk feedstocks for
ethanol production is indicated by a theoretical maximum ethanol yield.
The theoretical maximum yield assumes that 100% of the sugars are con-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Rapid Biomass Analysis 13

Table 2
Calibration Ranges (% dry wt) and Standard Errors
of Cross Validation (SECV, % dry wt) for a NIRjPLS Method for
the Compositional Analysis of Pretreated Corn Stover Solids
Constituent Range (% dry wt) SECV (% dry wt)a
Gluean 39.6-68.9 1.549
Xylan 1.0-23.2 1.458
Lignin 22.1-33.5 1.506
Protein 1.1-4.9 0.479
Ash 0.5-16.6 1.467
·5E of cross validation.

verted to ethanol and can be calculated from the carbohydrate content


(for a calculator for theoretical ethanol yields, see Website: http:/ /
www.ott.doe.gov /biofuels/ ethanoLcalculator.html). The theoretical
maximum ethanol yields for these feedstocks vary from 105 to 119 gal! dry
ton. Clearly, weight alone is not a good measure of feedstock value. The
differences shown in carbohydrates, lignin, inorganics, and acetyl content
are of a magnitude that could significantly impact process economics.
Compositional studies are under way to improve our understanding of
feedstock value and compositional variance as a function of location;
genetics; and cultivation, harvest, and storage methods.
The NIR/PLS rapid analysis method for feedstock characterization is
also being used to screen thousands of corn plants to identify interesting
cell-wall mutations. One such study quickly selected 44 individual plants
from a population of 2000 for further DNA analysis (not shown). Each of
these individuals had a xylan content that was significantly different from
the general population. These individuals belonged to 11 families and rep-
resented the expected 1:4 ratio of an otherwise normal family. No unusual
compositions were reported for members of the four inbred families used
as negative controls. The rapid analysis method provided a means of select-
ing unusual samples based on their unusual cell-wall chemistry.
Field-based methods for the analysis of live plants are being devel-
oped for use in large-scale genetic screening projects. These methods will
eliminate the costly and time-consuming milling and drying steps. The
ability to identify interesting plants before flowering may improve the
efficiency of cross-breeding experiments.
Process Intermediate Methods
The calibration set described here for the compositional analysis of
solid process intermediates from the dilute-acid pretreatment of corn sto-
ver contains 96 samples produced using three bulk feedstocks. The calibra-
tion range and the SE of cross validation for each constituent are given in
Table 2. The accuracy of the NIR/PLS method is illustrated in Fig. 5, which
Applied Biochemistry and Biotechnology Vol. 105-108,2003
14 Hames et al.

compares the composition determined by wet chemistry with that esti-


mated by the NIR/PLS equation. The diagonal line represents perfect agree-
ment. Figure 5 demonstrates that the NIR/PLS method provides a complete
compositional analysis for a wide range of samples with precision and
accuracy that match the wet chemical methods used to obtain the calibra-
tion data.

Application of Process Intermediate Method


This NIR/PLS rapid biomass analysis method is being used to evalu-
ate the performance of dilute-acid pretreatment of corn stover. Figure 6
shows compositional differences in pretreated samples of corn stover as
measured by NIR/PLS on process samples. Rapid analysis allows process
operators to determine when the sample is depleted in xylan and is ready
for the next process step, enzymatic hydrolysis. Stopping pretreatment
early will leave hemicellulose in the sample, which decreases enzyme
efficiency (3). Continuing pretrea tment beyond the point of xylan removal
will result in the loss of fermentable sugars, which lowers potential etha-
nol yields.
Methods are being developed for the rapid compositional analysis of
liquid and slurry process streams. When used together, the feedstock, solid
process intermediate, and liquid/ slurry methods can be used as feed-for-
ward or feed-backward tools for process control.

Conclusion
For biomass analysis, rapid techniques based on NIR/PLS can pro-
vide significant savings in time and money with no loss of precision or
accuracy relative to the calibration methods. Spectroscopy-based compo-
sitional analysis methods are applicable to a wide variety of biomass and
biomass-derived materials. NREL is uniquely situated for the develop-
ment of these methods for several reasons. First, years of experience in
method development for biomass analysis and experience with a wide
variety of biomass types provide the necessary data quality. Second, the
availability at NREL of unique biomass samples produced in a research
environment enhances calibration range and depth. These new rapid meth-
ods for biomass analysis can support and improve research and develop-
ment by providing levels of information that would have been too costly to
pursue using traditional wet chemistry-based compositional analysis
methods. The development of rapid analysis methods for the composi-
tional analysis of biomass and biomass-derived materials is a key element
in enabling the commercialization of processes that produce fuels and
chemicals from biomass.
Acknowledgments
This work was funded by the US Department of Energy: Office of
Biomass Programs and the NREL Biomass program.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Rapid Biomass Analysis 15
stovint3.eqa calibrat

l
&J
'"
~ o Ash
"0
01)
U " Protein
1J
~ o Lignin
Ul
;:[
"- DGlucan
'"
:!
lIXylan

o 10 20 30 40 50 60 70
Wet cherristry values (%)

Fig. 5. Comparison of composition of pretreated corn stover solids as determined


by wet chemical and NIRjPLS methods.

~ 60 _ Xlyan
2: _ Ash
'j'" _ Lignin
D GIu:;an
40

2 3 4 5 6
sample

Fig. 6. Composition of pretreated corn stover samples demonstrating optimization


of pretreatment for xylose removal.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


16 Hames et al.

References
1. Milne, T. A., Chum, H. L., Agblevor, F., and Johnson, D. K., (1992), Biomass Bioenergy
2 (1-6), 341-366
2. DiFoggio, R. (1995), Appl. Spectrosc. 49(1), 67-75.
3. Grohmann, K., Torget, R., and Himmel, M. (1989), FY1988 Annual Report, Document
ID 1191, BIC no. B00463, Solar Energy Research Institute, Golden, CO, pp. Bl3-B32

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0017/$20.00

Xylem-Specific and Tension


Stress-Responsive Expression of Cellulose
Synthase Genes from Aspen Trees

CHANDRASHEKHAR P. JOSHt
Plant Biotechnology Research Center,
School of Forestry and Wood Products,
Michigan Technological University, 1400 Townsend Drive,
Houghton, M149931, E-mail: cpjoshi@mtu.edu

Abstract
Genetic improvement of cellulose biosynthesis in woody trees is one of the
major goals of tree biotechnology research. Yet, progress in this field has been
slow owing to (1) unavailability of key genes from tree genomes, (2) the inability
to isolate active and intact cellulose synthase complexes and, (3) the limited
understanding of the mechanistic processes involved in the wood cellulose
development. Here I report on the recent advances in molecular genetics of
cellulose synthases (CesA) from aspen trees. Two different types of cellulose
synthases appear to be involved in cellulose deposition in primary and second-
ary walls in aspen xylem. The three distinct secondary CesAs from aspen-
PtrCesA1, PtrCesA2, and PtrCesA3-appear to be aspenhomologs of Arabidopsis
secondary CesAs AtCesA8, AtCesA7, and AtCesA4, respectively, based on their
high identity I similarity (>80%). These aspen CesA proteins share the trans-
membrane domain (TMD) structure that is typical of all known "true" CesA
proteins: two TMDs toward the N-terminal and six TMDs toward the C-termi-
nal. The putative catalytic domain is present between TMDs 2 and 3. All signa-
ture motifs of processive glycosyltransferases are also present in this catalytic
domain. In a phylogenetic tree based on various predicted CesA proteins from
Arabidopsis and aspen, aspen CesAs fall into families similar to those seen with
Arabidopsis CesAs, suggesting their functional similarity. The coordinate
expression of three aspen secondary CesAs in xylem and phloem fibers, along
with their simultaneous tension stress-responsive upregulation, suggests that
these three CesAs may playa pivotal role in biosynthesis of better-quality cel-
lulose in secondary cell walls of plants. These results are likely to have a direct
impact on genetic manipulation of trees in the future.

Index Entries: Aspen; cellulose biosynthesis; cellulose synthase; trees;


wood development.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 17 Vol. 105-108, 2003


18 Joshi
Introduction

Understanding the biosynthesis of cellulose, a major macromolecule


in forest trees, is essential for developing important strategies needed to
produce genetically improved trees. Such an approach is also necessary
to meet our current and future demands for cellulose-based wood and
paper products. Wood is a major agricultural commodity of which cellu-
lose and lignin impart strength and rigidity to timber products, but lignin
needs to be removed from paper pulp to produce high-quality paper prod-
ucts. Therefore, our long-term goal is to develop genetically improved
trees with balanced proportions of cellulose and lignin that will benefit
global forest products industries. To achieve this goal, we have been
acquiring basic knowledge about the underlying genetic mechanisms con-
trolling cellulose, and lignin biosynthesis, which will assist us in building
better trees in the future. Here, I will review the current state of knowledge
regarding the molecular mechanism of cellulose biosynthesis with special
emphasis on trees.
Cellulose is by far the most abundant organic material on Earth (1).
One would therefore ask, is it necessary to improve cellulose production
in trees? However, augmentation of cellulose content and improvement of
cellulose quality in specific tree tissues such as in wood fibers definitely
holds a greater promise. Cellulose is a homopolymer of glucose that forms
unbranched B-1,4-D-glucan chains of which every alternate glucose mol-
ecule is flipped by 180 0 (2). It has been suggested that subsequent to syn-
thesis, many single cellulose chains instantly associate to form microfibrils,
and many such microfibrils come together to form macrofibrils in the sec-
ondary walls (3). The number of glucan chains in each such bundle is
known to vary from 36 to more than 1200 in plant cell walls (2). In addition,
the number of glucose residues per cellulose chain, or degree of polymer-
ization (DP), is another trait that shows significant variations from about
500 to 2000 in primary cell walls to about 14,000 in secondary cell walls (4).
The cellulose content also varies between primary and secondary cell
walls. The primary walls contain only 1-10% cellulose, to perfectly fulfill
its function in expanding and growing plant cells, whereas cellulose con-
tent in secondary walls exceeds 50%, to provide rigidity and strength
to plant cell walls. In nature, cellulose is naturally aggregated in highly
ordered crystalline form alternating with less-ordered amorphous form.
The crystalline form of cellulose has better tensile strength and wood cel-
lulose is about 60-70% crystalline (5). Thus, plant cellulose is a highly
heterogeneous material.
Cellulose synthesis occurs on the plasma membrane of plant cells (4).
In freeze-fracture replicas from angiosperm tissues, discrete particle
rosettes or terminal complexes (TC) coinciding with cellulose synthesis
have been observed (2). Kimura et al. (6) have recently used immunogold
labeling of rosette terminal complexes to confirm that cellulose synthases
(CesAs) are physically localized within these complexes. Since improve-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Aspen Cellulose Synthase Genes 19
ments in secondary wall cellulose are economically more desirable and
quantitative as well as qualitative variations in primary and secondary
wall cellulose are postulated to be the result of different types of CesAs
(7), it is natural to search for such secondary CesAs. The first breakthrough
in CesA gene cloning did not come from plants that make an abundant
amount of cellulose, but from bacteria, such as Acetobacter, that produce
extracellular cellulose. Active CesA complexes could be isolated from
these bacteria, and the first Acetobacter CesA gene was cloned in the early
1990s (8).
Through intensive search for conserved D, D, D ... QXXRW signature
that was proposed to be present in many processive glycosyltransferases
such as bacterial CesAs (9), Pear et al. (10) isolated the first cotton CesA
cDNAs (GhCesAs) from a cDNA library prepared from cotton fibers that
are highly active in secondary wall cellulose synthesis. Subsequently, with
completion of Arabidopsis genome sequencing, at least 10 distinct CesA
genes (AtCesA-1 to AtCesA-lO) with similarity to cotton CesA genes were
identified (11). Why so many CesAs are present in plant genomes is still an
open area of question, but at least some members, such as AtCesA-4, AtCesA-
7, and AtCesA-8, have been associated with secondary wall development
(12-14), indicating that different CesAs may indeed be involved in cell- or
tissue-specific cellulose biosynthesis. Many reviews are available on these
topics and interested readers may refer to them for further information
(1-3,11,15). Understanding the process of cellulose biosynthesis during
wood development is one of the major goals of tree biotechnology research.
Yet, the factors that control cellulose content and quality in wood cellulose
are still unknown.
In response to tension stress, many angiosperm trees develop tension
wood on the upper side of their leaning trunks and nonvertical branches
(5). A newly developed yet highly specialized gelatinous layer (G-Iayer)
replaces one of the inner secondary walls (52 or 53), filling the entire lumen
of the tension wood fibers (16). This thick G-Iayer, present in both xylem
and phloem fibers, consists almost exclusively of cellulose (98.5%), has a
small percentage of xylose (1.5%), and is devoid of lignin or pectin materi-
als (17,18). The G-Iayer cellulose is highly crystalline (>95%) and has a
high degree of tensile strength associated with parallel orientation of
microfibrils. This is the purest form of cellulose occurring in any woody
tissues, and it rivals only with cotton fiber cellulose. Since cellulose-based
forest product industries are vital to our economy and they mainly use
wood as their raw materials (see Website: http:/ / www.afandpa.org/news/
COP_6_index.html), we are also interested in understanding the molecular
mechanism of highly crystalline cellulose production during wood devel-
opment of trees. Work has focused on quaking aspen (Populus tremuloides)
trees for all the studies in which tension wood formation is well docu-
mented (18). For the last 10 yr, researchers from our laboratory have suc-
cessfully used quaking aspen as a model tree system for many molecular
genetic explorations (see Website: http://forestry.mtu.edu/iwr/pbrc/),
Applied Biochemistry and Biotechnology Vol. 105-108,2003
20 Joshi
and it is the most appropriate species for research in cellulose biosynthesis
in trees.
As stated before, recent molecular genetic evidence from a variety of
sources indicates that the CesA enzymes play an important role in cellu-
lose biosynthesis in plants, and that at least two different types of CesAs
are likely to be involved in cellulose biosynthesis of primary and second-
ary cell walls (1,13,14). As compared with primary wall, secondary wall
cellulose shows higher DP, higher crystallinity, and specialized orienta-
tion of microfibrils (4,5), and these characteristics may be attributed to
activities of secondary CesAs. To investigate the molecular basis of cellu-
lose production during wood development as well as tension stress
response in aspen trees, our group recently started experiments in cloning
and characterization of CesA genes from aspen. We reported the first full-
length cellulose synthase (PtCesA)(designated as PtrCesA1 hereafter, as
suggested by Delmer [11) cDNA from developing xylem of aspen (19).
Northern blot and in situ hybridization analyses of PtrCesA1 gene tran-
scripts in various aspen tissues conclusively demonstrated that PtrCesA1
expression is confined to developing xylem cells during normal stem
development. Moreover, predicted PtrCesAl protein showed a high
degree of similarity (approx 90%) with other known secondary CesAs
from developing cotton fibers (GhCesA1) and Arabidopsis xylem (AtCesA8
=irxl), indicating that PtrCesAl may also playa role in secondary cell wall
synthesis (10,14). These observations were further confirmed by PtrCesA1
gene promoter-~-glucuronidase (GUS) fusion analysis in transgenic
tobacco plants that showed xylem-specific GUS expression in all plant
tissues examined. During tension stress conditions, however, PtrCesAl
promoter-regulated GUS expression continued in xylem on the tension
side and GUS expression was turned off in tissues undergoing compres-
sion on the opposite side of the bend. Our results suggested a unique role
for PtrCesA1 in cellulose biosynthesis in both tension-stressed and normal
tissues in aspen (19). Although transcriptional regulation of PtrCesA1 gene
in tension-stressed and normal xylem tissues was conclusively shown in
our earlier work with transgenic tobacco, tobacco plants do not make ten-
sion wood in nature. We therefore transformed aspen tissues (which nor-
mally produce tension wood) with PtrCesA1 promoter-GUS fusion. These
transgenic aspen trees showed the same GUS expression profile, confirm-
ing that PtrCesA1 gene is indeed xylem specific and tension stress-respon-
sive in aspen background. Additional confirmatory data have been
recently collected using an in situ mRNA hybridization technique in which
gene-specific probe from PtrCesA1 gene was shown to be confined to
developing vascular tissues. Moreover, PtrCesA1 gene expression was also
observed to be upregulated on the tension-stressed side of aspen stem
with simultaneous downregulation of PtrCesA1 gene on the compressed
side of the aspen stem (unpublished observations).
We recently cloned and characterized two more full-length CesA
cDNAs from aspen xylem-PtrCesA2 and PtrCesA3-that showed a high
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Aspen Cellulose Synthase Genes 21
Table 1
Percentage Identity /Similarity Among Various
Secondary CesA Proteins from Arabidopsis and Aspen'.
AtCesA4 AtCesA7 AtCesA8 PtrCesA1 PtrCesA2 PtrCesA3

AtCesA4 69/77 69/76 69/76 69/78 79/85


AtCesA7 67/74 66/74 87/92 70/77
AtCesA8 82/88 67/75 68/76
PtrCesA1 64/74 68/75
PtrCesA2 69/77
PtrCesA3
aHomologous members from Arabidopsis and aspen are shown in bold.

D D D QVLRW

Fig. 1. Diagrammatic representation of PtrCesA proteins. Various domains are


indicated as follows: Zn = zinc-binding domain; HVR 1= N-terminal HVR; 1-8 = TMDs;
sub domains A and B =highly conserved (70-90%) part of catalytic domains in relation
to other CesA proteins; HVR II =central HVR. 0, 0, 0, and QVLRW motifs are shown
at the top.

degree of similarity to two other secondary CesAs from Arabidopsis-


AtCesA7 (irx3) (12) and AtCesA4 (13), respectively (unpublished obser-
vations) (GenBank accession nos. AY095297 and AF527387). PtrCesA2 is
3280 bp long and contains an open reading frame of 3096 bp encoding a
protein of 1032 amino acids. PtrCesA3 is >3.4 kb long and encodes a protein
of 1043 amino acids. A comparison among Arabidopsis secondary CesAs,
AtCesA4, AtCesA7, and AtCesA8 with aspen PtrCesAl, PtrCesA2, and
PtrCesA3 is provided in Table 1.
Based on these data it can be concluded that PtrCesA1, PtrCesA2, and
PtrCesA3 are aspen homologs of Arabidopsis AtCesA8, AtCesA7, and
AtCesA4, respectively. All three CesA proteins from aspen share the same
transmembrane domain (TMD) structure that is typical of all known "true"
CesA proteins (15): two TMDs toward the N-terminal and six TMDs toward
the C-terminal (Fig. 1). Moreover, the first hypervariable region, HVR I,
resides in the N-terminal region immediately following the putative zinc-
binding domain that is suggested to be involved in protein-protein inter-
action with other accessory proteins involved in cellulose biosynthesis (20).
The putative catalytic domain is present between TMD 2 and 3. All signa-
ture motifs of processive glycosyltransferases (9) are also present in this
globular region, as shown in Fig. 1. The second hypervariable region, HVR
II, which may have some specific function in regulating the quantity and
quality of cellulose produced, further interrupts the catalytic domain into
sub domains A and B.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
22 Joshi

Fig. 2. Phylogenetic analysis of 13 full-length CesA proteins from Arabidopsis and


aspen. DNA and protein sequence analysis was performed using various program rou-
tines from Genetics Computer Group (GCG) package version 10.2. Multiple sequence
alignment of various CesA proteins was done with Pileup program from the GCG pack-
age. Cladograms were developed using PAUP 4.0blO version (Phylogenetic Analysis
Using Parsimony; Sinaur, Sunderland, MD) with parsimony analysis and heuristic
search algorithm. Bootstrap analysis was done with 1000 replicates, and bootstrap val-
ues above 50 were considered for the development of unrooted tree. GenBank accession
numbers (underlined) for CesA genes to deduce the protein sequence included in this
figure are as follows: AtCesAl, AF027172: AtCesA2, AF027173: AtCesA3, AF027174:
AtCesA4, AB006703: AtCesA5, AB016893: AtCesA6, AF062485: AtCesA7, AF088917:
AtCesA8, AF267742: AtCesA9, AC007019: AtCesAlO, AC006300: PtrCesAl, AF072131:
PtrCesA2, unpublished; PtrCesA3, unpublished (the nomenclature used was proposed
by Delmer (1) in which At = Arabidopsis thaliana; Ptr = Populus tremuloides). Bold lines
indicate CesAs that are implicated in secondary cell-wall synthesis.

Since DNA sequence information regarding 10 full-length CesA genes


from Arabidopsis and 3 three full-length CesA genes from aspen is currently
available from which encoded proteins could be easily obtained by using
in silico methods, we compiled a total of 13 deduced CesA proteins for the
reconstruction of evolutionary events. The unrooted phylogenetic tree
developed from these 13 sequences using 1000 bootstrap values is shown
in Fig. 2. The most important observation is separation of the secondary cell
wall-specific clade, as indicated by the thick delimiting lines. AtCesA7,
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Aspen Cellulose Synthase Genes 23
which is shown to be associated with secondary wall development in
Arabidopsis (12), now pairs with PtrCesA2, forming the first distinct branch
of secondary CesAs. PtrCesA3 from aspen xylem groups with vascular
tissue-specific Arabidopsis AtCesA4 (13) and forms the second branch of
secondary CesAs. The third branch of the secondary clade consists of
PtrCesAl from aspen-developing xylem (19), and AtCesAB is associated
with secondary wall development in Arabidopsis (14). Thus, members of
these groups seem to be associated with the secondary wall development
in plants. The remaining three groups may be associated with primary cell
wall development.
Another interesting conclusion can be drawn from this sequence analy-
sis. Assuming that this phylogenetic tree reflects structural (and possibly
functional) differences among various CesA gene family members, one can
trace the possible minimum number of functional classes of CesA genes
in plants. The 10 Arabidopsis CesAs actually form only six phylogenetic
groups (group 1: AtCesAl and AtCesAIO; group 2: AtCesA3; group 3:
AtCesA2, AtCesA5, AtCesA6, and AtCesA9; group 4: AtCesAB; group 5:
AtCesA4; and group 6: AtCesA7). It must be acknowledged that other alter-
native interpretations of this phylogenetic tree are also possible. The three
known aspen CesA genes cluster around the same secondary CesA groups
(groups 4-6) among a total of six groups. Considering the hypothesis that
presence of more than one CesA is involved cellulose synthesis in plants
(14,21) and six subunits together make up a rosette complex where each
subunit further contains six CesA proteins (2), it is tempting to speculate
that three primary CesA groups (groups 1-3) form functional rosettes in
primary walls and three secondary CesA groups (groups 4-6) form rosettes
in the secondary walls, as shown in Fig. 3. This would also explain why even
a slight alteration in just one CesA gene could seriously impact the process
of cellulose biosynthesis in Arabidopsis (12,14,22).
In the near future, we would like to examine whether expression pat-
terns of PtrCesA2 and PtrCesA3 genes are also xylem specific and tension
stress-responsive in aspen similar to PtrCesAl. Taylor et al. (14) have re-
cently shown that AtCesA7 (irx3) and AtCesA8 (irxl) coordinately express
in xylem tissues of Arabidopsis and thatthey both may cooperatively control
cellulose synthesis in secondary cell walls of xylem tissues. Both these genes
have also been genetically proven to be involved in secondary cell-wall
formation in Arabidopsis (12,14). We would therefore like to explore if
PtrCesAl and PtrCesA2 proteins that show a high degree of identity / simi-
larity to AtCesA7 and AtCesAB proteins, respectively, are also coordinately
expressed in xylem tissues in a tension stress-responsive manner. Further-
more, expression of the third secondary CesA gene from Arabidopsis,
AtCesA4 has recently been shown to be associated with vascular tissue
development (13) and has a high degree of similarity to aspen PtrCesA3, as
shown in Table 1. Although cooperation of this third group of CesA with
the other two secondary CesA groups has not yet been previously sug-
gested, we would like to explore if this third group of secondary CesA from
Applied Biochemistry and Biotechnology Vol. 705-708,2003
24 Joshi

Fig. 3. Cross-sectional schematic diagram of rosette structures consisting of six


subunits each made up of three types of CesAs shown as open circles, closed circles,
and checkered circles. Both primary and secondary walls may have different groups
of three CesA proteins.

aspen, PtrCesA3, is also xylem specific and tension stress-responsive. Pre-


liminary data with in situ mRNA hybridization of gene-specific probes
from PtrCesA2 and PtrCesA3 to young aspen stem tissues have suggested
that both of these genes are also expressed in the xylem tissues similar to
PtrCesAl (unpublished observations).
More experiments with tension stress-responsive behavior of aspen
stems are currently in progress. These investigations may help us to obtain
a more accurate picture of molecular events occurring in normal xylem
development as well as in response to tension stress in angiosperm trees.
We hypothesize that secondary CesAs from aspen, PtrCesAl, PtrCesA2,
and PtrCesA3 coordinately express at the transcriptional and translational
levels in developing xylem tissues of aspen and work cooperatively to
produce cellulose with high DP in secondary walls of normal xylem. In
addition, we believe that they may also regulate the high crystallinity of
cellulose during tension stress conditions through compensatory meta-
bolic shifts between cellulose and lignin biosynthesis.

Acknowledgments
I wish to thank Dr. Vincent Chiang and Dr. Glenn Mroz for their con-
stant support and encouragement for my laboratory's work in cellulose
biosynthesis. I am specifically grateful to Dr. Luguang Wu for his initial
contribution to aspen CesA gene cloning. I also wish to thank Dr. Priit
Pechter, Dr. Xiaoe Liang; and my graduate students Rajesh Chavli, Anita
Samuga, and Udaya Kalluri for their assistance in preparing the manu-
script and sharing their unpublished data. This work was partially sup-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Aspen Cellulose Synthase Genes 25
ported by the USDA-NRI Grant Program (99-35103-7986), USDA-McIntire
Stennis Forestry Research Program, and Research Excellence Fund from
the State of Michigan.

References
1. Delmer, D. P. (1999), Annu. Rev. Plant Physiol. Plant Mol. BioI. 50,245-276.
2. Brown, R. M., Jr., Saxena, I.M. and Kudlicka, K (1996), Trends Plant Sci. 1, 149-156.
3. Delmer, D. P. and Amor, Y. (1995), Plant Cell 7, 987-1000
4. Haigler, C. (1985), in Cellulose Chemistry and Applications, Nevell, T. P. and Zoronian,
S. H., eds., Ellis Horwood, Chichester, UK, pp. 30-83.
5. Timell, T. E. (1986), Compression Wood in Gymnosperms. Springer-Verlag, Berlin,
Germany.
6. Kimura, S., Laosinchai, W., Hoh, T., Cui, X., Linder c.R., and Brown, R.M. (1999),
Plant Cell 11, 2075-2085.
7. Haigler, C. and Blanton, R. L. (1996), Proc. Natl. Acad. Sci. USA 93, 12,082-12,085.
B. Saxena, I. M., Lin, F. c., Brown, R. M. (1990), Plant Mol. BioI. 15,673--683.
9. Saxena, I. M., Brown, R. M., Fevre, M., Geremia, R. A, and Henrissat, B. (1995), J.
Bacteriol. 177,1419-1424.
10. Pear, J. R., Kawagoe, Y., Schreckengost, W. E., Delmer, D. P., and Stalker, D. M. (1996)
Proc. Natl. Acad. Sci. USA 93, 12,637-12,642.
11. Richmond, T. A (2000), Genome BioI. 1(4), Rev. 3001.1-3001.6.
12. Taylor, N. G., Scheible W.-R., Cutler, S., Somerville, C. R., and Turner, S. R. (1999),
Plant Cell 11, 769-779.
13. Holland, N., Holland, D., Helen~aris, T., Dhugga, K, Xoconostle-Cazares, B., and
Delmer, D. P. (2000), Plant Physiol. 123, 1313-1323.
14. Taylor, N. G., Laurie, S., and Turner, S. R. (2000), Plant Cell 12,2529-2540.
15. Joshi, C. P. (2003), in Molecular Genetics and Biotechnology ofForest Trees, Kumar, S. and
Fladung, M., eds., Howarth, Howarth, NY.
16. Dadswell, H. E. and Wardrop, AB. (1955), Holzforschung 9, 97-103.
17. Norberg, P. H. and Meier, H. (1966), Holzforschung 20, 174-178.
lB. Timell, T. E. (1969), Svensk Papperstidning 72,173-181.
19. Wu, L., Joshi, C. P., and Chiang, V. L. (2000), Plant J22, 495-502.
20. Kawagoe, Y. and Delmer, D. P. (1997) in Genetic Engineering, vol. 19, Setlow,J. K, ed.,
Plenum, New York, NY, pp 63-87.
21. Fagard, M., Desnos, T., Desprez, T., Goubet, F., Refregier, G., Mouille, G., McCann,
M., Rayon, c., Vernhettes, S., and Hofte, H. (2000), Plant Cell 12, 2409-2424.
22. Arioli, T., Peng, L., Betzner, A S., Burn, J., Wittke, W., Herth, W., et al. (1998), Science
279, 717-720.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0027/$20.00

Microbial Pretreatment of Biomass


Potential for Reducing Severity
of Thermochemical Biomass Pretreatment

FRED A. KELLER, JENNY E. HAMILTON,


AND QUANG A. NGUYEN*
National Renewable Energy Laboratory, 1617 Cole Boulevard,
Golden, CO 80401, E-mail: qnguyen@bioenergy.abengoa.com

Abstract
Typical pretreatment requires high-energy (stearn and electricity) and
corrosion-resistant, high-pressure reactors. A review of the literature sug-
gests that fungal pretreatment could potentially lower the severity require-
ments of acid, temperature and time. These reductions in severity are also
expected to result in less biomass degradation and consequently lower
inhibitor concentrations compared to conventional thermochemical pre-
treatment. Furthermore, potential advantages of fungal pretreatment
of agricultural residues, such as corn stover, are suggested by its effective-
ness in improving the cellulose digestibility of many types of forage fiber
and agricultural wastes. Our preliminary tests show a three- to five-fold
improvement in enzymatic cellulose digestibility of corn stover after pre-
treatment with Cyathus stercoreus; and a ten- to IOO-fold reduction in shear
force needed to obtain the same shear rate of 3.2 to 7 rev / s, respectively,
after pretreatment with Phanerochaete chrysosporium.
Index Entries: Microbial pretreatment; fungal pretreatment; corn stover;
enzymatic hydrolysis.

Introduction
In many enzyme-based biomass conversion processes to ethanol or
other chemicals, a thermochemical pretreatment step is required to disrupt
the lignocellulosic structure of biomass, and partially solubilize polysaccha-
rides (1,2). Pretreatment improves the susceptibility of holocellulosic
polysaccharides to enzymatic hydrolysis. Thermochemical pretreatment
has been identified as a unit operation that is the second highest (after feed-
stock) cost component in enzyme-based conversion of com stover to ethanol
(3). The cost centers are steam, chemicals, and expensive corrosion-resistant

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 27 Vol. 105-108,2003


28 Keller et al.

pretreatment reactors. A major reduction in cost of pretreatment would


have a significant impact on the overall cost of ethanol production.

Literature Highlights on Fungal Pretreatment


of Hardwood and Softwood Feedstock
Eriksson and Vallander (4) and Messner and Srebotnik (5) reported as
much as a 23% reduction in pulp-refining energy required by incubating
spruce and pine chips for 2-wk with the white-rot fungus Phanerochaete
chrysosporium at 35-40°C. It was suggested that fungal pretreatment dis-
rupts the wood structure apparently by partially breaking down the lig-
nin/ carbohydrate complex.
Akhtar et al. (6,7) found an average 47% reduction in refining energy
required in mechanical pulping after a 4-wk treatment with Ceriporiopsis
subvermispora sp. on aspen wood chips and a 30% reduction in extractable
resin compared to controls. Energy savings of 37% on loblolly pine were
obtained. The weight losses on each were about 6% on aspen and 5% for
loblolly pine after 4 wk of fungal treatment.
Kashino et al. (8) reported the effects of treating coarse mechanical
pulp with an as-yet unidentified white-rot fungus, IZU-154. Seven-d treat-
mentwith whole culture reduced refining energy of beech pulp by 50-66%
in 7 d for bleached and unbleached pulp, respectively, and by about 33% for
softwood pulp (spruce and pine), which was treated for 10-14 d. The hard-
wood lignin loss was 17.5%, with a weight loss of 4.3%, and the softwood
lost 5.3% lignin and 1.5% weight. The lignin loss for both Coriolus versicolor
and P. chrysosporium was about half of that for IZU-154: about 6-7% for
Coriolus, and about 9 % for Phanerochaete. The refining energy was mea-
sured by the number of revolutions in a PFI mill (Process Facilities Inc.,
Philadelphia, PA) required to develop a measured amount of freeness", as
/I

well as by electrical refining energy consumed. The inoculum for the coarse
pulp was the mycelium of potato dextrose broth-grown fungi fragmented
in water for 30 s in a sterile Waring blender, or the whole broth culture. It
was found that whole broth, or adding additional glucose to the blended
mycelium inoculum, provided no added benefit. The moisture content rec-
ommended for each pulp is 75% beech, 89% spruce, and 86% pine.
Sawada et al. (9-11) investigated the effects of fungal pretreatment
and steam explosion pretreatment on enzymatic saccharification of beech
wood meal. Their work indicates that fungal pretreatment followed by less
severe steaming maximizes enzymatic saccharification. The investigators
suggested that the lignin network covering the holocellulose (cellulose and
hemicellulose) is broken down by successive fungal pretreatment and
steam explosion pretreatments, which together maximize subsequent
enzymatic saccharification of beech wood meal. P. chrysosporium (ATCC
34541) was incubated without agitation at 37°C with 5.0 g of meal (unspeci-
fied dryness), 15 mL of Kirk's broth (2.0 g of glucose, 0.02 g of ammonium
tartrate, 0.02 g of yeast extract, 5.0 mL of 0.4 M phthalate buffer pH 4.5,
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Microbial Pretreatment of Biomass 29
1.0 mL of Kirk's salt solution, and 100 mL water) in 300-mL Erlenmeyer
flasks. The type of closure on the flasks was unspecified. One gram of
fungal pretreated biomass was then subjected to steam explosion at 170-
230°C for 0-30 min.
The maximum saccharification of beech wood meal (9.5%) occurred at
28 d of fungal pretreatment and became less thereafter, possibly, Sawada
et al. (9-11) explained, because degraded lignin combined with or coated
the remaining holocellulose. Fungal pretreatment combined with steam
explosion improved saccharification, compared to steam explosion alone
by 20-100 % of the polysaccharide in the wood. However, 17 % of the
holocellulose degraded during fungal pretreatment, and there was an
unspecified holocellulose loss during steam explosion at optimum 215°C
for 6.5 min.

Literature Highlights on Fungal Pretreatment


of Agricultural Wastes, Grasses and Forage Crops
Karunanandaa et al. (12) studied the biodegradability of crop residues
colonized by white-rot fungi. Colonization of maize (Zea maize L.) and rice
straw for 15 and 30 d by three fungi and a cellulaseless mutant of
P. chrysosporium were compared for dry matter (DM) digestibility. Cyathus
stercoreus improved the in vitro cellulose digestibility of maize and rice
straw by 37 and 45%, respectively, and DM loss was only 3.3%. The wild
and mutant strains of Phanerochaete indiscriminately degraded both hemi-
cellulose and cellulose. No direct correlation was found between lignin
degradation and improved digestibility.
Akin et al. (13) found that microbial delignification with white-rot
fungi improves forage digestibility relative to nonfungal treated controls.
C. subvermispora Jp-90031 improved in vitro ruminal digestion of Ber-
muda grass by 80% in 72 h. Of four other fungi-Phellinus pini RAB-83-19;
Phanerochaete chrysosporium K-3; and two cellulaseless mutants,3113, and
85118, derived from K-3-only K-3 improved the digestibility of Ber-
muda grass and produced losses of lignin. The other three organisms did
not improve the digestibility of the forage. C. subvermispora extensively
removed ester-linked p-coumaric and ferulic acids as well as non-ester-
linked aromatics from growing (parenchyma) and more recalcitrant,
nongrowing (sclerenchyma) plant cell walls. Further, it appears that the
fungi improve mass transport into the biomass.
Hadar et al. (14) reported that biodelignification of cotton straw by
the edible "oyster mushroom", Pleurotus ostreatus, followed by 36 h of in
vivo ruminal digestion removed 2.2 times more organic material than
nonfungal pretreated controls.
The concept that fungal pretreatment lowers the energy requirements
of thermomechanical pulping of lignocellulosic biomass could potentially
be applied to enzyme-based biomass conversion processes to lower the
thermochemical pretreatment severity. Lower pretreatment severity is
Applied Biochemistry and Biotechnology Vol. 105-108,2003
30 Keller et al.
expected to translate directly into lower chemical consumption, and
because of lower temperature requirements, possibly lower steam costs.
These reductions in severity could also reduce the capital costs and result
in lower biomass degradation and consequently lower inhibitor levels. The
reason for the success of fungal pretreatment, or "seasoning" as it is called
in the pulp industry, is that such pretreatment of wood chips reduces elec-
trical power consumption of mechanical pretreatment, as well as the chemi-
cal requirements. This potential for fungal pretreatment can be explained
by the ability of certain fungi to disrupt the plant cell wall chemistry, result-
ing in partial breakdown of the lignin I carbohydrate complex. Further-
more, fungal pretreatment can readily be incorporated into the feedstock
handling area of biomass-to-ethanol processes.
From this summary of relevant literature, we present preliminary
experimental results of the effect of fungal pretreatment on the enzymatic
cellulose digestibility of corn stover, and a concept of applying fungal pre-
treatment to bioconversion of lignocellulosic biomass to ethanol.

Materials and Methods


Fungal Collection
Based on our interest in the articles that we have reviewed, we col-
lected two fungi for evaluation: (1)C. stercoreus NRRL 6473 (delignification
of maize), and (2)P. chrysosporium BKM-F-1767; FPL (reference wild strain,
delignification)

Handling of Fungal Strains


Two possible approaches for handling fungal strains are to apply the
key fungi directly to the biomass, and to produce the enzymes and then
apply enzymes to the biomass. We applied the former approach since it has
been most successful in the literature.
Culturing and Preservation
Preliminary procedures were developed for propagation and preser-
vation of the fungal cultures. As soon as the fungal strains arrived from
either the Forest Products Lab or the Agricultural Research Service, they
were grown on potato dextrose agar (PDA, Difco, Detroit, MI). Once the
cultures covered most of the PDA plate, they were transferred vegetatively
by cutting pieces out of the agar with a sterile scalpel, or using a sterile,
sharp cork borer.
Preservation of Fungus in Glycerol
The PDA pieces were placed into a sterile test tube and resuspended
using 1 part culture and adding at least 2 parts 50% (w Iv) sterile glycerol,
making at least 25% (w Iv) glycerol. The mixture was well blended with
a glass rod so glycerol covered the fungus. A small amount of glycerol
was added to rinse the rod. A drop of Tween-SO was added if the glycerol

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Microbial Pretreatment of Biomass 31
Table 1
Media Preparation for Fungal Pretreatment"
Chemical Quantity (g)
NaN03 3.0
KCI 0.5
MgS04·7H20 0.5
FeS04·7 H 20 0.5
KH 2P04 1.0
Glucose 20.0
Distilled and deionized water Make up to 1 L (resulting pH 6.0)
aAdapted from Demain and Solomon (15).

did not wet the fungus and was reblended. The mixture was vortexed to
mix well, and 1.5 mL was serologically pipeted into cryovials, and frozen
at -70 DC. Freeze preservation was evaluated; all cultures survived freez-
ing and thawing, and grew well when plated on PDA. The cork borers
were kept from corroding by washing and draining them quickly and
rapidly drying in a lOODC oven. Sterilize fast at 121 DC and dry.
Preparation of Stover
Coarsely milled corn stover was washed to remove dirt, drained, and
air-dried to about 30% moisture. It was then well mixed (coned and quar-
tered), portioned, and frozen. To obtain homogeneous material, approx
2 kg of frozen, coarsely milled corn stover was mixed with crushed dry ice
(using 1 part crushed dry ice to 1 part biomass); the mixture was then milled
in a laboratory Wiley mill through a 2-mm mesh. The milled stover was
stored unsealed in plastic bags in a freezer. When all the dry ice had sub-
limed (when it was observed that the bag was not expanding), the bags
were sealed and kept frozen until needed.
Preparation and Inoculation of Corn Stover Biomass
Mycelial cultures were grown on PDA agar. Approximately 1 in. 2 of
a fresh agar culture was minced and suspended in 5 mL media at pH 6 (see
Table 1, ref. 15) and then homogenized using a sterile glass rod until
uniformly dispersed in the buffer. Large quantities were prepared in a
small, sterilized Waring blender. Mycelial culture homogenates were
used promptly. Spore suspensions could be kept for up to 1 wk in a refrig-
erator (4 C).
D

Approximately 100 g (dry basis) of Wiley-milled corn stover was au-


toclaved for 20 min at 121 DC. The two fungi, P. chrysosporium (wild-type)
and C. stercoreus were compared for their ability to pretreat corn stover vs.
a control (i.e., corn stover incubated without fungi). The fungi pretreatment
was carried out in triplicate using 250-mL DeLong Erlenmeyer flasks
capped with stainless steel Morton closures (Belleo, Vineland, NJ). Three
control flasks (without fungal inoculum) were also run simultaneously.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
32 Keller et al.
Ten grams of washed and frozen-milled com stover was added to each
flask. The total solids (TS) of stover added was 70.0%, so each flask contain
approx 7.0 g of bone-dry stover. Water was added to bring the stover up to
about 88% moisture. The flasks were autoclaved at 121°C for 1 h, and water
was added to replace evaporated water. To each flask was added 20 !J.L of
12 mg/mL tetracYcline-HCI (prepared in 70% ethanol) to prevent bacterial
contamination. Five milliliters of a 1:1 PD agar:buffer homogenate was
used as inoculum. No extra nitrogen was added. Initial flask weights were
obtained, and the flasks were incubated at 27°C at 150 rpm. These cultures
still lost moisture, even though the rest of the incubator was filled with
flasks containing only water to increase the ambient relative humidity. The
weight of the flasks was periodically monitored and distilled deionized
water was added every 2 to 3 d to maintain constant weight.
After 29 d, the flasks were weighed and distilled and deionized water
was added to give 90% moisture. The flasks were incubated for 1 h. Then
the cultures were harvested (the contents of three flasks of each culture
were combined) and centrifuged. Free liquid was pipeted off, filtered
through a 0.45-!J.m filter, and analyzed for soluble protein and glucose. The
solids were washed twice, and samples were obtained for dry weight and
cellulose digestibility assay.
Roller Bottle Tests
A roller bottle test was done to evaluate the performance of roller bottles
as the next scale-up from shake flask. One hundred grams of frozen-milled
com stover (30% TS) was weighed out into each of two 1-L sterile plastic
roller bottles (Belleo). One was inoculated with 10 mL of Phanerochaete wild-
type homogenate, and the other with 10 mL of distilled deionized water. The
bottles had Kraft-wrapped foam caps and were rotated at about 5 rpm. After
35 d, weight loss was replaced by distilled deionized water. Each bottle was
emptied into a 1-L beaker, using hooked spatulas to retrieve as much of the
contents as possible. Free water was used when possible to rinse the bottles.
No additional water was added. The consistencies and viscosities were mea-
sured for each batch.
Assays
Enzymatic cellulose digestibility and viscosity assays were conducted
to estimate the effectiveness of biopretreatment and potential reduction in
the severity required for thermochemical pretreatment.
Enzymatic Cellulose Digestibility
Two types of control samples were included in the enzymatic cellu-
lose digestibility evaluation. The fungal pretreatment control sample was
incubated at the same temperature for the same residence time but with-
out fungal inoculation. Another type of control sample, the enzyme
digestibility control, is the frozen-milled com stover without incubation.
The first two control samples and the fungal-pretreated corn stover

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Microbial Pretreatment of Biomass 33
samples were washed with water to remove soluble components by
blending 1 part, by weight, of corn stover with 4 parts of warm distilled
deionized water, and incubated at 45°C for 1 h. The slurry was vacuum
filtered using 0 AS lAm-pore filter.
The cake was then stored in sealed containers and frozen. A corn
stover sample containing 0.10 g of cellulose (dry wt basis) was added to a
20-mL labeled glass scintillation vial, with plastic-lined caps. In addition,
a milled corn stover sample as is (i.e., not incubated) containing 0.10 g of
cellulose was added to another vial for a digestibility control. To each vial,
5.0 mL of 0.1 M, pH 4.8, sodium citrate buffer and 40 !AL (400 !Ag) of tetra-
cycline were added to control the pH and minimize the potential for a
bacterial infection. Distilled and deionized water was added to make up
a total of 9.93 mL. The vials were capped and kept at 50°C until cellulase
enzyme was added. Commercial cellulase enzyme (CPN) was added to
the vials at dose equivalent to approx IS, 25, and 60 filter paper units
(FPU) / g of cellulose.
The vials were capped and incubated with gentle shaking (100 rpm) at
50°C until the release of soluble sugars leveled off, usually between 72 and
168 h. A 0.5-mL sample was taken daily using large-mouth serological
pipet, filtered through OA5-!Am filters. The glucose concentration was mea-
sured by YSI (Yellow Springs, OH).
Spectrophotometric Estimate of Soluble Protein
The second portion of free liquid pipeted off, and filtered was read on
a Milton Roy Spectronic Genesys 5 UV spectrophotometer, at 260 and 280
nm, using a quartz cell, to get an estimate of enzyme protein concentration
(less nucleic acid), based on the original method of Warburg and Christian
(16) (see ref. 17). The following equation was used to estimate the enzyme
protein concentration, by measuring absorbance at 280 and 260 nm:
Protein (mg / mL) = 1.55A 280 - O.76A 260

Viscosity of Corn Stover Slurry


A number of articles have reported that less mechanical energy is
required for pulp refining after fungal pretreatment. Because the particle
size of corn stover was visually reduced after fungal pretreatment, viscos-
ity measurement was thought to be a quick, but less rigorous, test than
digestibility to determine the extent of the pretreatment.
Thomas-Stormer Viscometer
The TS range for the Thomas-Stormer viscometer (Thomas Scientific,
Swedesboro, NJ) is from a few percent to as high as 30%, depending on the
suspended solids and substance tested. If the material exhibits true viscos-
ity, then the time in s is measured for 100 revolutions of the rotor, and the
viscosity is read from graphs of standard viscous substances, determined
in the same apparatus. Whether the material exhibits true viscosity can be

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


~
[
OJ

3-
Ib
3<;;.
Table 2
~
.,
::.
Enzymatic Cellulose Digestibility of Fungal-Pretreated Corn Stover
0..
OJ Flask Enzyme loading Cellulose digested (%) Dry Insoluble

fii
n no. (FPU / g cellulose) 16h 22h 112h 136 h solids loss (%)h
::.-
::.
o
Digestibility control: 1 15 0.3 1.2 10.2 9.5 N/A
~ no fungal culturing or no incubation 2 25 0.7 1.4 7.2 9.4
3 60 8.4 9.6 18.1 16.1
w Incubation control: 4 15 1.2 1.6 3.8 6.9 5.6 ± 0.7
..j::.
no fungal culture present during 5 25 2.2 2.6 5.0 6.1 5.6 ± 0.7
29-d incubation at 27°C 5 60 6.3 6.7 11.0 12.1 5.6 ± 0.7
Cyathus: pretreatment, 6 15 8.3 10.2 24.6 24.6 8.3 ± 1.1
29-d incubation at 27°C 8 60 20.5 22.1 34.0 35.7 8.3 ± 1.1
Phanerochaete: pretreatment, 9 15 0.3 0.5 4.9 3.4 20.0 ± 1.8
29-d incubation at 27°C 10 25 2.0 2.3 4.5 5.8 20.0 ± 1.8
11 60 5.6 6.1 9.8 12.0 20.0 ± 1.8
aNot available.
~ bInsoluble solids loss as a result of incubation control or fungal pretreatment before enzymatic hydrolysis.

Cl
-~
$0
N
Cl
8
Microbial Pretreatment of Biomass 35
determined by plotting a graph of laminar flow rates of the fluid vs force
on the fluid. This can be done for the Stormer viscometer by plotting revo-
lutions per second vs weight applied, for at least four weights. For a true
viscous substance, the plot is a straight line through the origin. Other
graphic forms represent various types of "plastics." If a straight line is not
produced, viscosities need to be determined at more than one force
(weight). Known amounts of distilled deionized water may be added to
low-moisture biomass, in both samples and controls, to reduce the TS to the
range for measurement of viscosity with the Stormer.
Bostwick Consistometer
The Consistometer (CSC Scientific, Fairfax, VA) is used to determine
the consistency of viscous materials such as jellies and sauces by measuring
the distance that the material flows under its own weight in a given time
intervaL This may not work for a slurry if the solids dewater. It can be
calibrated against known standards.

Results and Discussion


Shake Flask Tests
At least three major conditions, and their ranges, need to be satisfied
to support fungal pretreatment of biomass: aeration, carbon-to-nitrogen
(C:N) ratio, and moisture leveL These conditions usually vary widely,
depending on the biomass and organisms. It is well known that fungi
require a certain amount of air to disrupt lignocellulosic bonds. A range
of CN ratios is important to favor disruption of lignocellulosic bonds
over fungal growth, where fungal growth is the utilization of the carbon
and nitrogen to support the increase in fungal cell mass. Third, the correct
ranges of moisture level, or water activities are required not only to sup-
port fungal pretreatment, but also to disfavor bacterial growth. Certain
bacteria may be desirable for optimal mixed cultural communal activity,
but this topic is beyond the scope of this first phase of study.
Table 2 compares the cellulose digestibility of fungal-pretreated corn
stover with control samples at varying enzyme loadings. The cellulose
digestibility of the incubation control stover (using cellulase loadings of 15,
25, and 60 FPU / g of cellulose) ranged from 1.2 to 12.1 % of theoreticaL The
digestibility of Cyathus-pretreated corn stover ranged from 8.3 to 35.7%, or
between 6.9 and 3.0 times that of the controls, respectively. Phanerochaete
pretreatment did not increase the cellulose digestibility. The digestibilities
of washed raw stover were similar to those of the controls. The average
insolable dry solids losses (±2 SDs) after 29 d of incubation without culture,
with Cyathus and with Phanero-chaete, respectively, are shown in the last
column of Table 2. The liquid that was pipetted off was divided into two
parts and analyzed for glucose and enzyme protein. The glucose concentra-
tion was not detected. The enzyme protein concentratiosn (less nucleic
acid) are given in Table 3.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
36 Keller et al.
Table 3
Estimated Protein Concentration in Liquid Obtained
from Fungal Pretreated Corn Stover
Sample source Protein concentration (mg/mL)
Corn stover control (incubated without fungi) 40
Cyathus-pretreated corn stover 34
Phanerochaete-pretreated corn stover 166

These protein values are expected from the literature since Phanero-
chaete is known to be a prolific enzyme producer. Cyathus produced no net
protein relative to the control. The lowest moisture level they dropped to
before water was added back during the 29 d was about 77%. The fungi
appeared to be present when viewed under the microscope. No bacterial
contamination was observed in any of the flasks.
The flask weight loss discussed earlier contains CO2 loss from fungal
activity on corn stover. As shown in Table 2, the 29-d incubation with
Cyathus caused an 8.3% dry-wt loss, and Phanerochaete caused a weight loss
of almost 20%. However, the stover controls lost an average of 5.6%. This
is an expected weight loss owing to water-soluble extractives dissolved out
of the stover during the 29-d incubation. Compared to the controls, the net
weight loss owing to the Cyathus was only 2.7%, and about 14.5% for
Phanerochaete. Shorter incubation time and selection of the appropriate
fungal species should reduce the dry weight loss. Furthermore, one would
also want to know what component(s) make up the weight loss since loss
of carbohydrates results in lower ethanol yield and lignin loss could mean
less boiler fuel.
However, the problem here, which will need to be solved before accu-
rate work can be done in the future, is that we are not able to detect moisture
loss simply by weighing the flasks, unless we know the DM loss, and fungal
cell mass gain. Studies in the literature solved this problem by controlling
the humidity so that there is no significant moisture loss. We could estimate
the DM loss from CO2 emission. That is possible, but difficult to do in small
shake flasks while they are being aerated. One could use a mass spec if one
obtained a uniform and high enough air flow rate. Or, one could attempt
to trap the CO2, The DM loss could at best only be estimated unless one
knows the solubilization rate of each: hemicellulose, cellulose and lignin,
and cell mass gain.
At small scale, studies in the literature use an environmentally con-
trolled incubator, or they make them by enclosing the flasks in a closed
chamber, or even plastic bags, while sparging the flasks by pumping water-
saturated air through tubing to each flask. At large scale, the biomass is
stacked in piles over tarpaulin-lined depressions. Drainage is pumped out
and sprayed over the piles, at frequent intervals, occasionally adding
makeup water, or the piles are covered with tarpaulins.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Microbial Pretreatment of Biomass 37
Table 4
Force vs Time Measurements of Fungal-Pretreated
and Control Corn Stover Slurries
Force(g) Time (s)
Phanerochaete-treated corn stover 5 29.7
10 28.0
30 23.7
60 18.8
80 14.3
Corn stover control 250 No flow
(incubated without fungi) 300 21.3
400 16.6
500 13.0
600 11.0

Roller Bottle Tests


A roller bottle test was done at the same time as the second shake flask
test to evaluate the performance of roller bottles for fungal pretreatment.
The consistencies and viscosities were measured for each batch.
Consistencies
The stover control took 20 s to travel 2.5 cm, and 40 s to travel 3.5 cm at
12°C on the Bostwick Consistometer, before it began to noticeably dewater.
The Phanerochaete pretreated material ran the full course (24 cm) in 0.5 s. The
slurry of fungal-pretreated corn stover became noticeably more fluid before
adding any water back, as compared to the control. The approx 80-fold
difference in flow velocities between the two consistencies is quite dramatic
to the unaided eye.
Viscosity
Viscosity of the stover slurries (diluted with distilled deionized water
to 10% [w fwD was measured at 15°C using the Thomas-Stormer viscom-
eter. Experimental data of force vs time per 100 revolutions for Phanerochaete-
pretreated corn stover and corn stover with no fungal pretreatment are
presented in Table 4 and Fig. 1. Figure 2 shows the data transformed into the
apparent viscosity, or fluid flow characteristic, which shows the type of
"plastic" that the slurry flow describes. The unpretreated stover behaves
like an "inverted plastic," while the Phanerochaete-pretreated stover flows
like a pseudoplastic. More important, Fig. 2 illustrates the great reduction in
force required to agitate the Phanerochaete-pretreated stover as compared
with the unpretreated stover. The reduction in force ranges from almost a
10-fold reduction at the higher shear rates to a greater than 100-fold reduc-
tion at the low shear rates.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


38 Keller et at.
Tlme/100 revs (sec)

100 I!!'!"I

~1 ...... Untreated Control


No Flow
...... Phanerochaete
80
pretreated

60 --

40 -

20 ~
~
\"
~

o
o 100 200 300 400 500 600 700
Force (g)

Fig. 1. Viscosity measurements of 10% (w /w) corn stover slurry using Thomas-
Stormer viscometer.

Proposed Performance Criteria for Fungal Pretreatment


We expect that application of fungal pretreatment to achieve the
objective of reduced steam pretreatment severity requires the following
general criteria:
1. Low mass loss of feedstock, especially carbohydrates.
2. Low costs-particularly nutrients, air and a reasonably short resi-
dence (storage) time.
3. Robustness; that is, the fungi must compete with the natural flora.
4. Lower thermochemical pretreatment severity (Le., significant reduc-
tion in one or more of the following parameters: temperature, acid or
alkali concentration, and residence time).
5. Improved enzymatic cellulose digestibility.
6. Noninhibitory to fermenting organisms.
Figure 3 shows a conceptual configuration for incorporating fungal
pretreatment in bioethanol production. Simulation of process and equip-
ment options should be carried out to evaluate the potential process
improvements with application of fungal pretreatment. Possible process
improvements include the following:
1. Lower pretreatment severity requirements result in lower operating
costs (e.g., size reduction, acid, steam) and pretreatment reactor cost
(e.g., lower temperature, pressure, and acid concentration, result in
less expensive material of construction).
2. Higher enzymatic cellulose digestibility and higher ethanol fermen-
tation yield because of lower level of inhibitors as a result of lower-
severity pretreatment.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Microbial Pretreatment of Biomass 39
Shear Rate (rev/sec)

10.-------------------------------------~
9 +-----------.
8+-------------------------------~~--------_4

7 ---------------------~~-------------I

6
5
4
3
~ Untreated Control
2
No flow _ _ Phanerochaete
1 +--------;:-,-,,-1----
~ pretrea
ted
O+---~--~~~~--~===r==~==~
o 100 200 300 400 500 600 700
Shear Stress (g)

Fig. 2. Flow characteristics of 10% (w /w) corn stover slurries.

Biomass
Feedstock
Fungal
Inoculum Air
Lignin Residues

Fungal Pretreatment Recycled Water & Nutrients


(Incorporated into 1+---------------1
Feedstock Storage)
Enzymes & Fermenting
Acid or
Steam Organism
alkali
Recycle

Water

Leachate

Filtrate containing enzymes and nutrients


Ethanol

SandlGravel1 Fermentation Inhibitors

Fig. 3. Conceptual block flow diagram of bioethanol production using fungal


pretreatment.

3. Lower enzyme requirements owing to the combined fungal and ther-


mochemical pretreatments and supplemental enzyme produced in
the fungal pretreatment step.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
40 Keller et al.
As practiced in the pulp industry, onetime high-dose inoculation of
the biomass feedstock is required to establish a colony of desirable fungus
(or fungi). The fungal population could be maintained by blending a small
portion of the fungal-containing biomass with incoming feed. The process
simulation of various process options would set the targets and priority for
research activities to confirm the technical and economic feasibility.

Conclusion
Based on the preliminary results, there is not a clear correlation
between digestibility improvement and viscosity reduction. However,
we believe that these impressive fungal pretreatment improvements
should translate into significant reduction in size-reduction (e.g., milling)
energy requirements or the severity of steam pretreatment needed to
solubilize corn stover polysaccharides. Recommendations for further
work include fine tuning the fungal pretreatment conditions and time
needed to result in significant reduction in viscosity and improvement
in cellulase digestibility at small scale, with minimal carbohydrate loss;
scaling up of the fungal pretreatment to 5-gal pails or 55-gal drum to
generate sufficient material for evaluation of the impact on thermochemi-
cal pretreatment severities required to achieve high hemicellulose solu-
bilization and cellulose digestibility; optimizing a symbiotic fungal
pretreatment using two or more fungi; and since the key requirement for
the small-scale work is a controlled, constant-humidity incubator, main-
taining constant humidity while providing mixing, controlled tempera-
ture, and airflow for the aerobic fungi.

References
1. Hsu, T. A, (1996), in Handbook on Bioethanol-Production and Utilization, Wyman, C.
E., ed., Taylor & Francis, Washington, DC, pp. 179-212.
2. Zerbe, J.I. And Baker, A J. (1987), in Energy from Biomass and Waste X, Klass, D. L., ed.,
Elsevier, London, UK, pp. 927-947.
3. (2002), NREL Enzyme Sugar-Ethanol Platform Project, National Renewable
Energy Laboratory, Golden, CO; (Website: http://www.ott.doe.gov /biofuels/
esp _background.html.)
4. Eriksson, K E. and Vallander, L. (1982), Svensk Papperstidning 6, R33-R38.
5. Messner, K and Srebotnik, E. (1994), FEMS Microbiol. Rev. 13,351-364.
6. Akhtar, M., Attridge, M. c., Blanchette, R. A, Meyers, G. c., Wall, M. B., Sykes, M.
S., et al. (1992), Proceedings ofthe 5th International Conference on Biotechnology in the Pulp
and Paper Industry, Kuwahara, M. and Shimada, M., eds., Uni Publishers, Tokyo,
Japan, pp. 3-8.
7. Akhtar, M., Attridge, M. c., Meyers, G. c., Kirk, T. K, and Blanchette, R. A (1992),
TAPPI J. 75(2),105-108.
B. Kashino, Y., Nishida, T., Yoshimasa, Y., Fujita, K, Kondo, R., and Sakai, K (1993),
TAPPI J. 76(12), 167-17l.
9. Sawada, T., Nakamura, Y., Kobayashi, F., Kuwahara, M., and Watanabe, T. (1995),
Biotechnol. Bioeng. 48(2), 719-724.
10. Sawada, T., Kuwahara, M., Nakamura, Y., and Suda, H. (1987), Int. Chern. Eng. 27,
686-693.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Microbial Pretreatment of Biomass 41
11. Sawada, T., Nakamura, Y., Kobayashi, F., Mohammed M., Watanabe, T., and
Kuwahara, M. (1991), Proceedings of 6th International Symposium on Wood and Pulping
Chemistry, Melbourne, Australia, pp. 463-469.
12. Karunanandaa, K, Fales, S. L., Varga, G. A., and Royse, D. J. (1992), J. Sci. Food Agric.
60, 105-112.
13. Akin, D. E., Sethuraman, A., Morrison III, W. H., Martin, S. A., and Eriksson, K E.
(1993), Appl. Environ. Microbiol. 59(12),4274-4282.
14. Hadar, Y., Kerem, Z., and Gorodecki, B. (1993), J. Biotechnol. 30,133-139.
15. Demain, A. L. and Solomon, N. A., eds., (1986), Manual of Industrial Microbiologyand
Biotechnology, American Society for Microbiology, Washington, DC.
16. Warburg, O. and W. Christian (1942) Biochem. Z. 310:384-421.
17. Alexander, R. R. and Griffiths, J. M. (1993), Basic Biochemical Methods, 2nd ed., Wiley-
Liss, New York, NY.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0043/$20.00

Physical Separation of Straw Stem


Components to Reduce Silica

J. RICHARD HESS,*,l DAVID N. THOMPSON,l


REED L. HOSKINSON, l PETER G. SHAW, l
AND DUANE R. GRANT 2
lldaho National Engineering and Environmental Laboratory,
PO Box 1625, Idaho Falls, 1083415, E-mail: jrh@inel.gov; and
2Grant 4-0 Farms, 600N 707E, Rupert, 10 83350

Abstract
In this paper, we describe ongoing efforts to solve challenges to using
straw for bioenergy and bioproducts. Among these, silica in straw forms a
low-melting eutectic with potassium, causing slag deposits, and chlorides
cause corrosion beneath the deposits. Straw consists principally of stems,
leaves, sheaths, nodes, awns, and chaff. Leaves and sheaths are higher in
silica, while chaff, leaves, and nodes are the primary sources of fines. Our
approach to reducing silica is to selectively harvest the straw stems using an
in-field physical separation, leaving the remaining components in the field
to build soil organic matter and contribute soil nutrients.

Index Entries: Wheat straw; silica; selective harvest; bioenergy; combus-


tion; whole crop utilization.

Introduction
Agricultural crop residues are a valuable renewable resource from
which to produce biobased products. In 1999, American farmers harvested
53,909,000 acres of wheat (1). The straw from this acreage of wheat repre-
sents greater than 100 million t annually. Currently, some of the straw is
harvested (baled) for use as livestock bedding or low-grade animal feed.
However, these low-grade uses provide only a minimal return. Nationally
only about 3.2% of the economic return on wheat is from straw (1). Produc-
ers, including the National Association of Wheat Growers and the Idaho
Wheat Commission, have long recognized the potential economic and
environmental benefits of producing bioenergy and bioproducts from
excess wheat straw.
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 43 Vol. 105-108,2003


44 Hess et al.

Internodal

Fig. 1. Simplified anatomic structure of a wheat plant ready for harvest, which
needs to be fractionated into the grain (food), stem (bioenergy), and remaining residue
(returned to field).

In a recent National Research Council report (2) on research priorities


for biobased industrial products, development of "equipment and meth-
ods to harvest, store, and fractionate biomass for subsequent conversion
processes" is listed as a key research priority. Another prominent theme of
the report is development of technologies for the effective and economic
utilization of lignocellulosic biomass, particularly waste sources such as
crop residues.
To completely utilize all biomass, it must be recognized that not all
parts of the crop residue are equally valuable. Just as a flour mill only wants
the grain delivered, biofuels and chemical processors only want the high-
yielding cellulose and hemicellulose biomass components delivered. The
cellulose-rich vascular tissues of crop stems contain the greatest amount of
sugars for digestion and conversion to biofuels and chemicals (Fig. 1). The
formerly physiologically active tissues of leaves, sheaths, and awns are
heavily impregnated with process-fouling minerals, and these tissues also
contain higher amounts of other organic components (i.e., protein, lipids,
pigments, pectin, organic acids) than the stems. The more complex molecu-
lar components of these nonstem tissues are more valuable if directed to
other applications (i.e., cattle feed, left in the field as a soil amendment,
direct extraction of specifically engineered and/or naturally occurring
chemicals). Thus, for cost-efficient utilization of straw and other crop resi-
dues, the undesirable components must be removed.
However, the current paradigm for straw utilization includes the
necessity to transport all the components of the straw to the point of utili-
zation; there is no cost-efficient way to remove the undesirable components
before transporting it. This is expensive not only because of the low bulk
density of straw, but also because it brings the less valuable components to
the manufacturer's gate and creates economic and environmentalliabili-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Selective Harvest of Straw Stems 45
ties. We describe a strategy for reducing the amount of undesirable residue
components shipped to centralized biorefineries. This strategy varies some-
what depending on the end use but generally consists of an in-field physi-
cal fractionation that reduces silica content.

Materials and Methods


Wheat Straw
Four varieties of wheat straw were selected for use in this study, includ-
ing Stephens-soft white winter, produced in Rupert, ID; Boundary-hard
red winter, produced in Monteview, ID; Whitebird-soft white spring, pro-
duced in Idaho Falls, ID; and Westbred 936-hard red spring, produced in
Rupert, ID. These varieties were selected from an Idaho Wheat Commission
list of the most popular existing and new wheat varieties developed for the
intermountain and Pacific Northwest wheat-growing regions. The varieties
produced by and available from collaborating growers determined the four
varieties that were finally selected for this work. Grant 4-D Farms produced
and baled large quantities of Westbred 936; this variety was used for all the
large-scale separation studies.
All straw utilized was produced during the year 2000 cropping sea-
son. Twenty large bales of Westbred 936 (1.2 x 2.4 m [4 x 8 ft] bales) were
produced and stored in a stack at the side of the field at Grant 4-D Farms,
and only the protected center bales were used for the studies. Six 0.61 x
1.2 m (2 x 4 ft) bales of Stephens, Boundary, and Whitebird straw were
also collected. Additional Boundary straw was collected in 1.2 x 2.4 m
(4 x 8 ft) bales. To make the straw easier to handle for laboratory studies,
these large bales were rebaled into smaller 0.61 x 1.2 m (2 x 4 ft) bales
containing about 22.7 kg (50 lb) each. The small bales of all varieties were
stored in covered storage.
Analytical Methods
Wheat straw compositions were measured by acid hydrolysis using
the quantitative saccharification technique (3). Carbohydrate concentra-
tions in hydrolysates were measured by high-performance liquid chro-
matography using a Bio-Rad HPX-87P column with distilled water as
eluent, as previously described (4). Ash compositions were determined
by ashing the straw samples to constant weight at 650°C for 18-24 h in a
muffle furnace. Silicon, potassium, and chlorine contents of the ashed
samples were determined by energy-dispersive spectrometry (EDS) (5),
at 10-20 ke V using a Phillips XL30ESEM. For this method, the ashed straw
was placed in a thin film on double-sided tape mounted on a carbon-
coated aluminum disk. The disk was carbon coated to prevent charging.
For each measurement, a region of the surface was scanned for 5-10 min.
Two regions of the disk were scanned to verify homogeneity of the sample.
Standards were prepared by weighing, grinding, mixing, and ashing
reagent-grade silicates, oxides, and chlorides. These were mounted on
Applied Biochemistry and Biotechnology Vol. 105-108,2003
46 Hess et al.

Fig. 2. University of Idaho plot-harvesting machinery used for the mechanical


separation.

the double-sided tape and on the disks and counted under the same con-
ditions as the straw ash. Calibration curves were prepared and used to
adjust the values from the internal quantitative program on the EDS sys-
tem for matrix effects. Wet methods and wavelength-dispersive spec-
trometry were used to verify the results.
Partitioning of Wheat Straw Biomass
Using Plot-Harvesting Equipment
The first phase of research on the mechanical separation was con-
ducted to determine whether the higher-value wheat straw components
(the stems) could be mechanically separated from the rest of the material
using preexisting plot-harvesting equipment. Several small bales of wheat
straw were threshed and separated at the University of Idaho experiment
station in Aberdeen, using their small plot-harvesting equipment. The
equipment used for the tests is shown in Fig. 2 and included a stationary
tined-cylinder thresher (on the left in the foreground) and a small plot
combine harvester (on the right in the background). Each machine was
initially tested separately, while adjusting airflow and cylinder speeds to
separate the straw stems from the remaining fractions. In addition to sepa-
rate testing, the machines were also tested in sequence to determine the
more effective equipment configuration for separating the stem fraction
from the rest of the straw.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Selective Harvest of Straw Stems 47
Table 1
Compositions of Whole Straw
of Several Wheat Varieties Utilized in this Study
Westbred 936 Boundary Stephens
Component (wt %) (wt %) (wt %)
Gluean 40.6 40.4 39.3
Xylan 25.2 25.8 25.1
Galactan 0.4 0.4 0.4
Mannan 2.5 3.2 2.7
Arabinan ND' ND' ND'
Lignina 14.5 18.1 17.8
Ash 9.0 4.1 5.4
% Reeoveryb 92.2 92.0 90.7
aLignin with extractives.
bDifference from 100% owing to unknown uronic acid content and to
recovery errors in the procedure.
eND, none detected.

Table 2
Compositions of the Straw Ash
from the Several Wheat Varieties Used in this Study
Westbred 936 Boundary Stephens
Component (wt %) (wt %) (wt %)
Total Ash 9.0 4.1 5.4

Si02 2.3 2.8 3.2


K 3.5 0.3 0.2
CI 2.3 0.020 <0.020
Other 0.9 1.0 2.0

Results and Discussion


Straw and Ash Compositions
The compositions of three of the wheat variety straws utilized were
measured by the quantitative saccharification method and are given in
Table 1. These compositions are typical for wheat straw, although the
Westbred 936 variety had a higher ash content owing to a significantly
higher potassium concentration than the other two varieties. Estimated
compositions of the plant polymer and ash fractions are given in Table 2.
The ash content was estimated using the energy dispersive spectrometry
capabilities on a scanning electron microscope.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
48 Hess et al.
1
c
o
;:
u
~ 1
.t:.
U
as
CII
.E
.t:.
I/)
00(

~
Fines Nodes Sheath Heads Stems Baled
Straw

Fig. 3. Distribution of ash in hand-separated straw fractions of Westbred 936. Each


bar represents wt% ash for each respective fraction.

c
o
U
~
.t:.
U
as
CII
.E
aI
~
en
~O%~~--~~~~_ _~. . . .
Fines Nodes Sheath Heads Stems Baled
Straw

Fig. 4. Distribution of silica in hand-separated straw fractions of Westbred 936.


Each bar represents wt% silica for each respective fraction.

Field-Scale Partitioning of Wheat Straw Biomass


Samples of Westbred 936 wheat straw were perfectly separated by
hand to provide a "best separation" target, and also by machine using the
plot-harvesting equipment. All fractions were then analyzed for ash con-
tent, which is shown in Fig. 3. The elemental composition of each ash
sample, including silicon (Si), potassium (K), and chlorine (Cl) contents,
was then estimated using a scanning electron microscope with EDS. The
silica contents of the perfectly separated fractions are shown in Fig. 4. The
ash content of the whole straw used in the perfect hand separations was 9.0
wt%, of which 24 % was silica, giving 2.2% silica in the whole straw. The ash
content of the perfectly separated stems was 6.1 wt%, of which 20% was
silica, and thus the silica content of the perfectly separated stems was 1.2
wt%. Therefore, perfect separation reduced ash content by 32% and silica
content by 45%. This is the benchmark to which the mechanical separation
was compared.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Selective Harvest of Straw Stems 49

Fig. 5. Cylinder and concave tines of the tined-cylinder thresher, visible through the
open cover.

Fig. 6. Nodes separated from the stems using the tined-cylinder thresher.

Neither the stationary tined-cylinder thresher nor the small plot com-
bine harvester machine alone adequately separated the stems from the
other material. Using the plot harvesting equipment, the most effective
method for separation of the bulk of the stem fraction from the nodes
(without optimization for stem yield) was to first pass the straw through
the tined-cylinder thresher. The aggressive threshing action of the tines
(see Fig. 5) broke the stems near the nodes, and the more dense nodes
exited through the airflow separation and out the grain discharge. Sepa-
rated nodes are shown in Fig. 6. Each fragment in Fig. 6 also contained a
residual stem segment on either side as a result of the stem breaking at a
small distance from the node. Although this reduced stem yield, the
purpose of these tests was to test in-field methods for separating the
Applied Biochemistry and Biotechnology Vol. 705-70B, 2003
50 Hess et al.

c:
o

~
r.
Ii!QI 1
c:
~ 1
in

i O.O'lb" ' - -
Baled Straw Separated Stems

Fig. 7. Silica reduction achieved for Westbred 936 using the plot-harvesting equipment.

stems. In the eventual system chosen, the separation would be optimized


for maximum yield of stem recovery. The remaining straw fractions were
discharged from the rear of the tined-cylinder thresher onto the ground,
after internal airflow separation. This pile, comprised of two fractions
that we clssified as "stem" and "chaff," was collected and fed into the plot
combine harvester. The plot combine harvester further separated the
stems from the chaff, and the material that exited off the straw walkers
was primarily the desired straw stems. Nearly all of the remaining straw
material was discharged from the sieves just below the straw walkers.
Mechanical separation was performed on straw from a different field/
bale, resulting in slightly different ash and silica contents in the whole
straw. The silica contents of the baled whole straw and the separated stem
fraction are compared in Fig. 7. The ash content of the whole straw used in
the mechanical separations was 11.2 wt%, of which 12% was silica, giving
1.3% silica in the whole straw. The ash content of the mechanically sepa-
rated stems was 8.8 wt%, of which 8.3% was silica, and thus the silica
content of the mechanically separated stems was 0.73 wt%. Therefore, the
mechanical separation reduced ash content by 21% and silica content by
44 %. The results of these tests are very positive and show that our mechani-
cal separation process is capable of reducing the silica content of the straw
by selectively harvesting predominantly straw stems. As discussed earlier,
in order to be economically feasible, this separation must be done in a single
pass across the field while the crop is being harvested.

Conclusions
Mechanical separations of wheat straw stems using plot-harvesting
equipment were as good as perfect fractionation of stems by hand. These
results indicate that existing harvesting equipment can be modified to do
this fractionation if proper adjustments are made to the equipment set-
tings. Mechanical separation of the stems reduced the ash content of the
harvested fraction by 21 % and the silica content by 44%.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Selective Harvest of Straw Stems 51

Acknowledgments
We thank the University of Idaho Aberdeen Research and Extension
Center for assistance with the in-field fractionation. Dr. Judi Steciak (Uni-
versity of Idaho) played a large role in obtaining support and will be
involved in future tasks on this project. This project is administered by the
Idaho Department of Water Resources Energy Division. This work is sup-
ported in part by the US Department of Energy, Assistant Secretary for
Energy Efficiency and Renewable Energy (EE) under DOE Idaho Opera-
tions Office Contract DE-AC07-99ID13727. Additional support for this
work, both in kind and financial, is provided by the Idaho Wheat Com-
mission, Grant 4-D Farms, and Energy Products of Idaho, Inc.

References
1. USDA NASS (2001), US Department of Agriculture, National Agricultural Statistics
Service; (Website: http://www.usda.gov /nass/).
2. National Research Council (NRC) (1999), Biobased Industrial Products: Priorities for
Research and Commercialization, National Academy Press, Washington, DC.
3. Saeman, J. F., Bubl, J. L., and Harris, E. E. (1945), Ind. Eng. Chern. 17,35-37.
4. Thompson, D. N., Chen, H. -c., and Grethlein, H. E. (1992), Bioresour. Technol. 39,
155-163.
5. Shodson, M. J. and Sangster, A. G. (1989), Can. J. Bot. 67,281-287.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0053/$20.00

Application of a Depolymerization Model


for Predicting Thermochemical
Hydrolysis of Hemicellulose

TODD LLOYD AND CHARLES E. WYMAN*

Thayer School of Engineering, Dartmouth College,


8000 Cummings Hall, Hanover, NH 03755,
E-mail: charles.e. wyman@dartmouth.edu

Abstract
Literature data were collected and analyzed to guide selection of conditions
for pretreatment by dilute acid and water-only hemicellulose hydrolysis, and
the severity parameter was used to relate performance of different studies on
a consistent basis and define attractive operating conditions. Experiments were
then run to confirm performance with com stover. Although substantially
better hemicellulose sugar yields are observed when acid is added, costs would
be reduced and processing operations simplified if less acid could be used
while maintaining good yields, and understanding the relationship between
operating conditions and yields would be invaluable to realizing this goal.
However, existing models seldom include the oligomeric intermediates preva-
lent at lower acid levels, and the few studies that include such species do not
account for the distribution of chain lengths during reaction. Therefore, the
polymeric nature of hemicellulose was integrated into a kinetic model often
used to describe the decomposition of synthetic polymers with the assumption
that hemicellulose linkages are randomly broken during hydrolysis. Predic-
tions of monomer yields were generally consistent with our pretreatment data,
data reported in the literature, and predictions of other models, but the model
tended to overpredict oligomer yields. These differences need to be resolved
by gathering additional data and improving the model.
Index Entries: Hemicellulose; hydrolysis; kinetic model; dilute acid;
depolymerization.

Introduction
Ethanol made from cellulosic biomass has the potential to displace a
significant fraction of petroleum in the United States, reducing the depen-
dence on foreign imports and improving the environment. Biologic process-
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 53 Vol. 705-7 08, 2003


54 Lloyd and Wyman

ing routes offer particular promise of reducing costs sufficiently to make


ethanol cost competitive (1). However, biomass must be pretreated if we are
to realize the high yields vital to commercial success by such processes, and
pretreatment is among the most costly steps (2,3). Several pretreatment
approaches have been and are currently being developed with the intent of
reducing overall processing costs (4), and results for many technologies are
reported in the literature. Yet, it is challenging to compare performance of
these options because different feedstocks have been used and testing and
analytical methods are not always the same. Thus, a research project funded
by the US Department of Agriculture (USDA) Initiative for Future Agricul-
tural and Food Systems Program seeks to evaluate leading biomass pretreat-
ment technologies using a common feedstock and standardized methods.
Specifically, the technologies being evaluated include ammonia percolation,
dilute-acid, water-only, ammonia fiber explosion, neutral hot water, and
lime pretreatment. Com stover from Iowa is currently being used as feed-
stock by all of the investigators involved.
Our effort in this project focuses on pretreatment by removal of hemi-
cellulose either with theaddition of acid or in water-only thermochemical
processes. During these operations, hemicellulose is solubilized to mono-
meric and oligomeric saccharides that can degrade to furfural, tars, and other
products (5). It is desirable to minimize acid use, produce highly digestible
cellulose, and maximize the yield of monomers that are most easily fermented
to fuels and chemicals. The approach we applied to achieve these goals began
with an extensive search of the literature data to define favorable dilute-acid
and water-only pretreatment conditions. Several reported pretreatment stud-
ies present data on water-only pretreatment of com stover (6-10) and corn-
cob (5), and others provide data on dilute-acid pretreatment of com stover
(11-13) (M. Tucker, personal communication, 2002), and a combination of
corncob and com stover (14). Next the severity parameter and a newly devel-
oped modified severity parameter that integrates the weight percent acid
concentration were applied to estimate optimal operating conditions from
the literature data gathered over a wide range of times, temperatures, and
acid levels, and experiments were then conducted to verify the predicted
performance. Finally, because oligomers are found to be important atthe low
acid levels targeted and existing analyses do not consider the range of oligo-
mer chain lengths expected, a model used to describe the breakdown of
synthetic polymers was applied to predict the yield of soluble monomeric
and oligomeric sugars and to help guide definition of conditions to reduce
acid use while maintaining high sugar yields.

Materials and Methods


Preparation of Corn Stover
Corn stover collected by BioMass Agri-Products in Harlan, lA, was
supplied by the National Renewable Energy Laboratory (NREL) for all of
our experiments. Samples were drawn from a lot created for all partici-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Hemicellulose Oepolymerization Model 55
pants in the multi-institutional investigation and ground in a Mitts and
Merrill Model lOx 12 rotary knife mill (Harvard, IL) to less than 6-mm
particle size, air-dried to about 5% moisture, and stored at -4°C.
For the water-only steam gun tests, about 10 kg of this material was
placed into a Hastelloy basket and immersed in a circulation tank con-
taining water at 60°C. Water was pumped through the basket for 4 h to
thoroughly wet it, the basket was then removed from the water, and the
contents were allowed to drain. Next, soaked corn stover was transferred
in approx 1-kg batches to a hydraulic press and dewatered to a nominal
moisture content of 50%. Dewatered stover was coned and quartered and
separated into 740-g (dry) batches.
For dilute-acid experiments with small reaction tubes, frozen corn
stover was taken from the lot and ground further to pass through a 2-mm
screen. This material was separated using a model RX-29 Ro-tap with 8-in.
Tyler screens (Soiltest, Chicago, IL) to recover the -590 + 420pm fraction,
and 1% (w /w) H 2S04 (made from a stock 72 ± 0.1 % H2S04 solution) was
added to this material to achieve a solids concentration of about 5%. Excess
liquid was drained after letting this slurry sit overnight at room tempera-
ture, and the moist solids were pressed in a 4-in. stainless steel cylinder
with a 1-in. diameter hydraulic piston to a final resting pressure of the
hydraulic fluid of 1000 psig. The moisture content of this pressed corn
stover was determined to be 35% in a Precision 1800W drying oven (Win-
chester, VA) following NREL LAP 001 (15). A 1% H 2S04 solution was
added back to bring the final solids content to 25%. Because corn stover has
some neutralizing power, a sample was checked for acid content. Seventy-
five grams of water was added to a 1-g sample and agitated for several
minutes. The pH was determined to be 2.68 using a model 8000
VWR Scientific (West Chester, PA) pH meter and combination electrode
(model 511050; Beckman Coulter, Fullerton, CA). This pH corresponds to
an actual acid concentration of 1.02% (assuming total dissociation).
Steam Gun Tests
The large batch water-only tests were performed using the 4-L, ver-
tical steam gun at the NREL (16,17). The insulated and heat-traced vessel
was preheated for several h to ensure that it was completely up to tem-
perature. Prepared corn stover sample was then loaded into the vessel
through a funnel, and the contents were sealed by actuated ball valves on
either end of the 10-cm-diameter pipe. Next, live steam was introduced at
the top and bottom of the vessel, raising the temperature of thermocouple
probes to the target temperature in about 15 s. The temperature was
maintained at a target value by controlling steam pressure in the vessel
with a valve on the line from the boiler. The reaction time was defined as
the period from when steam was introduced to the reactor until the con-
tents were explosively discharged.
A total of 10 steam gun "shots" were performed at various tempera-
tures and a range of times. Three tests were performed at 210°C, one each

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


56 Lloyd and Wyman

for 2, 6,and 18 min. Four tests were performed at 190°C, one each at 7,14,
22, and 74 min. Two tests were performed at 170°C, one each at 27 and 87
min. One test was performed at 150°C for 107 min.
When a run was completed, the discharge valve on the bottom of the
steam gun was opened, and the contents were blown into a 300-L flash
vessel to rapidly bring the temperature to below 100°C and quench the
reaction. Next, the contents were removed, placed in double plastic bags
for storage at 4°C, and shipped in a cooler to Dartmouth College for analy-
sis. After arriving at Dartmouth, the pretreated biomass was pressed to
obtain about 100 mL of liquid hydrolysate, the remaining material
was slurried with tap water in a 19-L poly bucket, the supernatant was
decanted, and fresh water was added. This procedure was repeated until
the pH of the supernatant reached 6.0. Then the solids were filtered and
weighed, and their moisture content was determined before rebagging
and refrigerating them.
Tube Reactor Batch Tests
Batch tube reactors were assembled from 12.5-mm OD Hastelloy (C276)
tubing with a 0.8255-mm wall thickness cut into 10-cm lengths. About 6 g
of the acid-soaked corn stover described earlier was loaded into each reactor
tube using a small spatula and a specially designed plastic funnel and tamped
lightly with a glass rod. The tubes were capped with inexpensive 304 stain-
less steel end caps protected from the acid by inserting machined Teflon
plugs into the tube ends based on the kind suggestion of Professor Y. Y. Lee
of Auburn University. The tubes were immersed in a 22.8 cm id x 35 cm deep
4-kW model SBL-2D fluidized sand bath (Techne, Princeton, NJ) controlled
at the target temperature, held for a specified amount of time, removed from
the sand bath, and immediately immersed into a room temperature water
bath to quench the reaction. Reaction time was determined as the moment of
immersion into the heated sand bath until the moment of quenching. After
cooling, the contents of the tubes were removed and filtered with 100 mL of
deionized water through a medium-porosity fritted glass filter crucible, and
the solids were dried in a vacuum oven at 45°C.
Runs were made at the following temperatures and times: 180°C for
1,2,5,10,20, and 40 min; 160°C for 5,10,20,40, and 80 min; and 140°C for
5, 10,20,40,80, and 120 min. Temperature transients to be expected using
batch tubes during heat-up were analyzed using the method developed by
Stuhler and Wyman (18). This showed that at 160°C it could be expected
that the center-line temperature of a O.5-in. ID tube would be 153°C (.95!1
T + To) after approx 90 s. This simulation suggests that the longer run times
at lower temperatures would not be affected significantly although tran-
sient effects could be greater at 180°C.
Analyses
Dried solids and filtered hydrolysates were analyzed for their mono-
meric sugar content according to NREL LAP-002 (19) and LAP-013 (20)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Hemicellulose Oepolymerization Model 57

protocols, respectively. LAP-014 was applied to quantify soluble oligomers


in the hydrolysates (21).
The corn stover used contained 36.1% glucan, 21.4% xylan, 3.5%
arabinan, 1.8% mannan, 1.6% galactan, 17.2% lignin, 4.0% protein, 3.2%
acetyl, 3.6% uronic acid, and 7.1 % ash.

Development of Kinetic Models for Hemicellulose Hydrolysis


Severity Parameter Models
In the mid-1940s Saeman (22) modeled the saccharification of wood
cellulose during pulping by assuming that the reaction followed first-order
homogeneous kinetics, and this model has become the basis for describing
hemicellulose hydrolysis and subsequent sugar degradation (23). In the
mid-1950s, this model was refined by assuming that hemicellulose was
compqsed of two distinct fractions, one that is relatively easy to hydrolyze
and the other more difficult (24). A few recent articles have refined this
model to include one or two species of oligomers as intermediates in the
reaction's sequence (5):
Hemicellulose (fast)
..... k
f
.....

/' - -
Monomers ko Oligomers kd Degradation

/,ks
Hemicellulose (slow)

Another approach has been to apply severity models that combine


operating conditions such as time and temperature to the following single
expression for water-only hydrolysis (25):
Ro = t . exp([TH - TR ] /14.75) (1)

in which t is reaction time in minutes; THis hydrolysis temperature; and T R


is reference temperature, most often 100°C. When acid is used, a combined
severity parameter, CS, that includes the effect of added acid catalyst dur-
ing organosolv treatments has been applied by Chum et al. (26):
logeS = logRo-pH (2)
When Eq. 1 is substituted into this expression, the following relationship
results:
(3)

Because most of the literature data available on corn stover only


reported weight percent acid addition but not pH, we modified this expres-
sion by assuming that the hydrogen ion concentration is proportional to the
percent acid:
(4)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


58 Lloyd and Wyman

in which A is the acid concentration in weight percent and n is a proportion-


ality constant close to 10. This is consistent with many of the Saeman-based
models applied to hemicellulose hydrolysis. Substituting relationship 4
into 3 and assuming the proportionality constant n = 10 gives a result we
term the modified severity parameter, Mo:
Mo = t . lOA· exp([TH - TR ] /14.75) (5)
Although not identical to the CS defined by Eq. 3, the modified severity
parameter provides a useful tool for correlating a diverse array of literature
data that only provides weight percent acid addition and not pH.

Depolymerization Model
Most kinetic models for hemicellulose hydrolysis do not consider
the presence of oligomers in the reaction sequence at all, and the few that
include such species lump them into one or two compounds that ignore
the range of chain lengths expected as hemicellulose decomposes from
larger chains to smaller ones. However, kinetic models have been devised
to describe the distribution of chain lengths that occur in the decomposi-
tion of plastics (27) and size reduction operations in the grinding of min-
eral ores (28) based on both continuous (29,30) and discrete (31-33)
product distributions. Furthermore, Agarwal et al. (34) applied discrete
depolymerization kinetics to predict hemicellulose and cellulose degra-
dation in alkaline pulping. The discrete depolymerization approach of
Simha (32) was applied here to capture the range of chain lengths that are
expected during pretreatment by hemicellulose hydrolysis.
Consider the breaking of one bond of a polymer composed of n mono-
mer units to form two new molecules:
N n -Nj+Nn _ j (6)
Subsequently, these products can degrade further as follows:
N n _ j - Nk + N n _ j_k (7)
Nj-Nj+N j_i (8)
If we assume that all the bonds linking monomer units have the same
probability of being broken, then the rate of change in concentration of any
j-mer can be expressed by the following differential equation:
dN. n
-dJ =2k h ~ N j -k h (j-1)N j (9)
t j=j+l

in which kh is the hydrolysis rate constant that is now assumed to be the


same regardless of chain length, in which the first term on the right side
is the rate of creation of j-mers from the scission of molecules larger than
a j-mer (note that there are two scission events that result in identical
products) and the second term describes the rate at which existing j-mers
disappear when anyone of the (j-1) bonds present are broken. When this

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hemicellulose Oepolymerization Model 59
expression is extended to the longest polymer chain of length n that can
only be broken but not formed, the rate of change in its concentration is
described by the following expression:
dNn
Tt=-k h(n-1)N n (10)

Integrating Eq. 10 based on the initial condition that at time t =0, N n=Nno,
we obtain the following result:
Nn=N~exp(-kh[n-1]t) (11)
To solve for the concentration of the (n-1)-mer, Eq. 11 is substituted into
Eq. 9 to give
dN _
n1 0
--cit = 2kh N n exp (- kh [n -1] t) - kh (n - 2)N n_1 (12)

The result of integrating this linear first-order differential equation with


the initial condition N n- 1 = a is
N n_1 =2N~(exp[-kh{n-2}tl-exp[-kh{n-l}t]) (13)
Following this procedure for successively smaller j-mers, we can arrive at
a generalized equation to describe the concentration of any j-mer (j ¢ n)
at any time:
Nj=N~(1-a)(j-l)a(2+[n-j-1]a) (14)
with a = 1- e- kht (15)
Figure 1 illustrates the distribution of products predicted by applying
Eqs. 11 and 14 to describe the decomposition of a hypothetical pentamer.
The concentration of oligomers of chain length 5 rapidly drops while the
concentrations of oligomers with 2, 3, and 4 monomer units build up and
then drop off as monomer is formed. Furthermore, because shorter chains
can be formed in more ways than longer chains, the concentration of oligo-
mers with a chain length of 4 is less than that of chain length 3, which is, in
turn, less than the concentration of chain length 2. Ultimately, depolymer-
ization results in monomer being the only remaining species.
An additional consideration for predicting hemicellulose hydrolysis
kinetics is the reaction of monomer to furfural and other degradation prod-
ucts (35). Assuming a single degradation reaction that can be described by
a first-order dependence on monomer concentration, the differential equa-
tion describing the rate of monomer formation and consumption becomes
dN n
-1=2k h LN i -k dN 1 (16)
dt i=2

in which kd is the rate constant for decomposition of the monomer. This


equation can be integrated to obtain the following result:
N = 2kh([n-1][eHhtl-eHil] _ [n_2][e{-2khtl_eHil])
1
n kd-kh kd- 2k h (17)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


60 Lloyd and Wyman
100

80

...CII 5-mer
E 60
0
c:
0
:l!
1/1
1\1 40
-;l!.
0

..--~ ......
, .....
20 ..... , 2-mer
f ~: . . . . ".,3-mer " ..................
J~ -mer '.'-. . ..... - - -
o ~------",~~:~'~-~-~~~~~"~-~-~-~~~~-----~-T--------~-~-F-~-~--~~
Fig. 1. Distribution curves for depolymerization of a hypothetica15-mer containing
5 monomer units assuming random scission and arbitrary rate constant.

We now have expressions that describe the concentrations of each frag-


mentation product expected for hemicellulose hydrolysis at any time.

Results and Discussion


Water-only hydrolysis data for corn stover from Rubio et al. (6),
Tortosa et al. (9), and Schultz et al. (10), and corn cobs from Garrote et al.
(15) were plotted against log severity parameter to provide a basis for
selecting pretreatment conditions. The fraction of potential xylose remain-
ing in the solids is presented in Fig. 2, and the yield of oligomers only and
oligomers plus monomers plotted in Fig. 3. Based on these results, condi-
tions were defined for additional runs with our controlled source of corn
stover, and trends were found that are consistent with the literature data,
as shown in Figs. 2 and 3. The maximum yield of xylose in the hydrolysate
including monomers and oligomers was about 60% and occurred at a log
severity factor of between 3.8 and 4.0. In addition, we note that xylooligo-
mers predominated, accounting for as much as 90% of solubilized species
at lower severities and about 80% at the conditions corresponding to the
maximum yield.
For dilute-acid hemicellulose hydrolysis, the hydrolysate data of Lee
et al. (14) for corncob / cornstover mix and Tucker (personal communica-
tion, 2002) for corn stover were plotted against the log modified severity
parameter to guide our definition of run conditions. Figure 4 shows that the
yields reported in the literature and those obtained in our tests are very
similar. In this case, the maximum yield of monomers and oligomers was

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hemicellulose Depolymerization Model 61
100
D D Schultz at al.
o Tortosa et al.
0
80 o Garrote et al.

(!;
CD D
I/)
c I:;. Rubio at al.
>-
x
I:;.
• • This Work
iii 60
01:;.
E
CD .1:;.
D

(5
a.. D
40
E 0
:J
E t.
.~ <>
~ 20 • 0
00
0~ 01:;.0

0
• 0-
I:;. 0
0

2.5 3 a5 4 4.5 5 5.5

Log Ro

Fig. 2. Percentage of potential xylose remaining in solid residue vs log Ro (Ro + t .


exp[{TH - TR ) /14.75]) for various investigators and data from this study for water-only
hydrolysis.

100
• Garrote et al. Total
D Garrote Ollgomers
• Rubio et al. Total
80 . This Work Total
CD
I/)
c + This Work Oligomer
x>-
iii
..;::::;
60 • •

D
s::::
0-·
$
c
a.. •
E
40
• +
+ •

:::J
D 4'
E

.
.~
~

0

~ 20

:.!!
0 ~ D
D

0
2.5 3 3.5 4 4.5 5 5.5

Log Ro

Fig. 3. Percentage of potential xylose as monomers plus oligomers (total) and as


only oligomers in liquid hydrolysate vs log Ro (Ro =t . exp[TH - TR/14.75]) for various
investigators and data from this study for water-only hydrolysis.

about 90% and occurred at a log modified severity parameter of about 3.8.
However, xylooligomers represented a much lower fraction of the total
solubilized sugars than for the water-only case, with only about 20% of the
total being oligomers at the optimum yield point.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


62 Lloyd and Wyman
100 • Chen et al. Total
o Chen et al. Oligomer

••
Q)
II) & Tucker et at Total
0
>. 80 • Tucker et al. Oligom
><
~ • - This Work Total
x This Work Oli omer
E 80 • • & _
.to
Q)
0
a.. •

E
::s 40
E
·x 0
as
~
20 0 0
~
0 x x
..
0
II

II X X
0 II X
0 II II

2.8 3.3 3.8 4.3 4.8

Log Mo

Fig. 4. Percentage of potential xylose in hydrolysate as monomers plus oligomers


(total) and oligomers only vs long Mo (Mo = t . An . exp[TH - TR )/14.75]) for various
investigators and data from this study for kilute-acid hydrolysi.

Next, model curves based on Eqs. 11,14, and 17 were fitto our steam
gun data in Fig. 5 and to literature data as shown in Fig. 6 for Garrote
et al.' s (5) data for water-only hydrolysis of corncobs. The kinetic constant
for monomer degradation was calculated from the Arrhenius expression
reported by Converse et al. (35). Then, the hydrolysis constant was deter-
mined to minimize the sum of the squares of the differences between
monomer data and model predictions. Use of the monomer for predicting
xylan partitioning is somewhat arbitrary but was chosen because it could
be fit well with an arbitrary hydrolysis rate constant. In addition, oligo-
mers with nine or more monomer units were arbitrarily assumed to
remain in the residual solids, and those of length 2 through 8 as well
as monomers were assumed to be all in the liquid phase. If the cutoff
between soluble and insoluble oligomers is decreased, the oligomer curve
moves closer to the data but never reaches it even at a cutoff degree of
polymerization (DP) of 2. A cutoff above DP-8 tends to increase the diver-
gence between data and model, but only slowly, as the contribution from
higher-chain oligomers diminishes rapidly with increasing DP.
As shown in Figs. 5 and 6, the data and predictions agree reasonably
well initially, but xylooligomers are overestimated and residual xylan
underestimated at later times. This divergence could be explained, at least
in part, by an accelerated decomposition of xylose, but only analysis could
confirm or deny this. Unfortunately, no analyses of degradation products
were available, and the decomposition kinetics of Converse et al. (35) were
used unmodified.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hemicellulose Oepolymerization Model 63
100 - Residual Xylan (Model)
This Work 190°C Steam Gun - .. - Monomers in Hydrolyzate (Model)
90 Water OnlyTests ...... Oligomers in Hydrolyzate (Model)
.. Residual Xylan This Work Data
80 " Monomers in Hydrolyzate This Work Data
• Oligomers in Hydrolyzate This Work Data
G 70
8

'J.
~ 60
iii
50
kh=.017 min-1
40 kcF.042 min- 1
E
::::I
E
'51! 30 •
i
~ 20

10 .---- .. -' .
0
. -- .. ~-

0 10 20 30 40 50 60 70 80 90 100
Time, min

Fig. 5. Comparison of data and depolymerization model predictions for water-only


hydrolysis of corn stover for this work.

Corn Cob Garrote et al.Monomers


.A

160 OC • Garrote et al. Oligomers


100 \
Kh=0.004 " Garrote et al. Res. Xylan
\
CP - - Model Monomers
III
0 Ii, KcFO.0054
_.... Model Oligomers
>. 80 _." -- - ---.. --
\

.
\

.' ....,' . •
>< \
-, Model Residual Xylan
a;
:;::::
c 60 '
.. ..
CP , '

0 '. ,
Il. ,!" , • ".
E 40
::::I
E
';(
.
,
: '~ ,
\\"
C'II
"", ...,
::i 20
0~
! " '.

.•
0
0 50 100 150 200 250 300 350
Time, min
Fig. 6. Comparison of data and depolymerization model predictions for water-only
hydrolysis of corncobs.

Of note, Fig. 5 shows that the maximum experimental yield (monomers


plus oligomers) occurred at about 14 min, corresponding to a log severity
factor of about 3.8, and Fig. 6 shows that the maximum yield occurred at
about 100 min, also corresponding to a log severity factor of about 3.8.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


64 Lloyd and Wyman
100

Com Stover
140°C .. - . • • • • • • • • • • • • • • • • 0.

,,'
80
1% H2SO4 •
GI
III
~.
, •
0
>. .' '. •
>< ,
!

\
...
!i 60
\


S or nomer

...r::: , This Work Oligomer


GI
0 .
\'
:
kh=O.15
kd=O.OO34
.. This Work Residual Xylan
Il. Model Monomer
E
• \ _ .. _ Model Oligomer
:::J 40
\ _ Model Residual Xylan
E
I
I I ,
'>(
as ..... ,
,
~
fI. .. ,,
.
20

,
..
0
• "
• '- . ""-. -__ 0._ . - .- .. - 0.- _._. __ .__ ..
0 5 10 15 20 25 30 35 40 45
Time, min

Fig. 7. Comparison of data and depolymerization model predictions for dilute-acid


hydrolysi of com stover for this work.

The same procedure was applied to develop the depolymerization


model for dilute H 2S04 hydrolysis using some of our batch tube data and
the data of Lee and Chen for a corncob / corn stover mix (14). A comparison
of the models and data are presented in Figs. 7 and 8, respectively. There
is a particularly large misfit between the oligomer and residual xylan data
and model in both figures. This is apparently a consequence of the addi-
tion of acid catalyst. Figure 7 shows that the maximum experimental yield
occurred at about 40 min, which corresponds to a log modified severity
parameter of about 3.8. In Fig. 8 the maximum yield appears to be near, but
beyond, the limit of the data provided by the literature source (50 min). At
50 min, the log modified severity parameter is about 3.7.

Conclusions
The highest yields of total solubilized corn stover xylose in our steam
gun, water-only hydrolysis was 53% at a log severity parameter near 4.0,
consistent with results reported in the literature. With H 2S04 addition, the
modified severity parameter was found to provide a useful means for com-
paring data from different studies, and the highest yield of solubilized hemi-
cellulose, using our batch tube apparatus, was measured to be 89% at a log
modified severity of 3.8, again consistent with literature values. Oligomeric
xylan comprised about 80% of the total soluble monomers and oligomers at
the maximum total yield point without acid present but contributed only
about 20% of the total sugars in solution when 1% H 2S04 was added.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hemicellulose Oepolymerization Model 65
100
Corn CoblCornStover
'" 0.68% H 2 SO 4
~

II)
\
I 140°C •
I/)
0
80 I
I '" •
>-
><
I
I

--
I

:!
I
I •
60
c:: ,.
II;'" • Chen Monomers
II)
, • Chen Oligomers
0
11.
~
I
.'1
. • kh=0.08
kd=0.0022 .. Chen Residual Xylan
_ Model Monomers
E

40
:::J ! \1 ...... Model Oligomers
E . ___ Model Residual Xylan
.. ..
1
";C
cu
:E 20
\
\ .
'"
~
0 '"
0
0 10 20 30 40 50
Time, min

Fig. 8. Comparison of data and depolymerization model predictions for dilute-acid


hydrolysis of corncob / corn stover mixture.

The severity parameter was found to provide a valuable tool to predict


the combination of conditions that maximize hemicellulose yield. Although
not as useful for predicting quantitative results, the ability of the severity
parameter to relate performance at different times, temperatures and acid
levels is very valuable for selecting run conditions for specific kinetic studies.
An important goal of our research is to reduce acid use while maintain-
ing high yields, and oligomers are expected to become more important at
lower acid levels. Thus, predicting oligomer histories will be very useful in
defining promising paths to this end, but available hemicellulose hydrolysis
kinetic models, including the severity parameter, do not describe the time
course distribution of oligomeric species of varying chain lengths. A depo-
lymerization model was found to predict monomer trends well using an
arbitrary hydrolysis rate constant. In the case of uncatalyzed hydrolysis, the
model-data fit was reasonable during the early stages of reaction, but oligo-
mer yields were overestimated and residual xylan was underestimated by
the depolymerization model at later stages. In the case of catalyzed hydroly-
sis, the fit between oligomer and residual xylan and the model was not
particularly good at any point during the reaction. Thus, further refine-
ments are needed for a depolymerization model to be useful as a predictive
tooL For instance, the assumption that all bonds react at equal rates may be
modified to include differences in end bonds; to consider changes in kinetic
constant due to chain heterogeneity; or to account for the "gel" effect
observed in polymer synthesis, in which the accessibility of molecules
changes with chain length. In addition, we plan to focus on improving our
methods of capturing oligomers to ensure that oligomer data accurately
Applied Biochemistry and Biotechnology Vol. 105-108,2003
66 Lloyd and Wyman

reflect their history during hydrolysis and are not affected by heat transfer
or other effects that could influence the profiles. It will also be valuable to
determine the range of oligomer sizes that are released into solution to
determine whether the definition of soluble and insoluble chain lengths we
arbitrarily assigned is reasonable.

Acknowledgments
We gratefully acknowledge support from the USDA Initiative for
Future Agricultural and Food Systems Program through contract no.
00-52104-9663. We also appreciate interactions with our partners in that
project: Y. Y. Lee from Auburn University; Bruce Dale from Michigan State
University; Rick Elander and Robert Torget from the NREL, Michael
Ladisch from Purdue University; and Mark Holtzapple from Texas A&M
University; and the students involved in this work. We are also indebted
to the NREL supported through the Department of Energy's Biofuels Pro-
gram for making its steam gun apparatus available for performing water-
only tests. In particular, we wish to thank Rick Elander for coordinating
the test work and Mel Tucker for performing these experiments. We also
received guidance from Quang Nguyen and Kyung Kim on using the
steam gun equipment and in analytical methods, respectively. Small batch
reactor experiments were performed at Thayer School of Engineering
using equipment purchased by Thayer School of Engineering.

References
1. Wyman, C. E. (1999), Annu. Rev. Energy Environ. 24, 189-226.
2. Wooley, R., Ruth, M., Glassner, D., and Sheehan, J. (1999), Biotechnol. Prog. 15,794-803.
3. Lynd, L. R., Elander, R. T., Wyman, C. E. (1996), Appl. Biochem. Biotechnol. 57/58,
741-76l.
4. Hsu, T. (1996), in Handbook on Bioethanol: Production and Utilization, Wyman, C.E., ed.,
Taylor & Francis, Washington, DC, pp. 183-187.
5. Garrote, G., Dominguez, H., and Parajo, J. C. (2001), Process Biochem. 36,571-578.
6. Rubio, M., Tortosa, J. F., Quesada, J., and Gomez, D. (1998), Biomass Bioenergy 15(6),
483-49l.
7. Carrion, J., Rubio, M., Gomez, D, Miiiana, A, and Soler, A, (1989), in Proceedings
of the 5th International Conference on Biomass for Industry, Grassi, G., Gosse, G., and
Dos Santos, G., eds., Lisbon, Portugal, Elsevier Applied Science, London, England,
UK, pp. 2.45-2.49.
8. Lamptey, J., Robinson, C. W., and Moo-Young, M., (1985), Biotechnol. Lett. 7(7),531-534.
9. Tortosa, J. F., Rubio, M., and Gomez, D. (1995), Afinidad LII457, 181-188.
10. Schultz, T. P., Templeton, M. c., Biermann, C. J., and McGinnis, G. D. (1984), J. Agric.
Food Chern. 32(5),1166-1172.
11. Esteghlaliart, A, Hashimoto, A G., Fenske, J. J., and Penner, M. H. (1997), Bioresour.
Tech. 59, 129-136.
12. Bhandari, N., MacDonald, D. G., and Bakhshi, N. N. (1984), Biotechnol. Bioeng. 26,
320-327.
13. Torget,R, Walter,P., Himmel,M., and Grohmann, K. (1991),Appl. Biochem. Biotechnol.
28/29, 75-86.
14. Lee, Y. Y., Chen, R, and Iyer, P. (1994), Annual Report for NREL Subcontract no.
XAW-3-13441-01, National Renewable Energy Labortory, Golden, co.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hemicellulose Depolymerization Model 67
15. Ehrman, T. (1994), Standard Method for Determination of Total Solids in Biomass,
Laboratory Analytical Procdure No. 001, National Renewable Energy Labortory,
Golden,CO.
16. Tucker, M. P., Mitri, R. K., Eddy, F. P., Nguyen, Q. A, Gedvilas, L. M., and Webb, J.
D. (2000), Appl. Biochem. Biotechnol. 84-86, 39-50.
17. Nguyen,Q.A, Tucker,M.P., Boynton, B. L.,Keller,F. A,and Schell, D.J. (1998),Appl.
Biochem. Biotechnol. 70-72, 77-87.
18. Stuhler, S. L. and Wyman, C. E. (2003), Appl. Biochem. Biotechnol. in press.
19. Ruiz, R. and Ehrman, T., (1996), Determination of Carbohydrates in Biomass by High
Performance Liquid Chromatography, Laboratory Analytical Procedure No. 002,
National Renewable Energy Labortory, Goldon, CO.
20. Ruiz, R. and Ehrman, T. (1996), HPLC Analysis of Liquid Fractions of Process Samples for
Monomeric Sugars and Cellobiose, Laboratory Analytical Procedure No. 013, National
Renewable Energy Labortory, Golden, CO.
21. Ruiz, R. and Ehrman, T., (1996), Dilute Acid Hydrolysis Procedure for Determination of
Total Sugars in the Liquid Fraction of Process Samples, Laboratory Analytical Procdure
No. 014, National Renewable Energy Labortory, Golden, CO.
22. Saeman, J. F., (1945), Ind. Eng. Chem. 37,43-52.
23. Jacobsen, S. E. and Wyman, C. E. (2000), Appl. Biochem. Biotechnol. 84-86,81-95.
24. Kobayashi, T. and Sakai, Y. (1956), Bull. Agric. Chem. Soc. Japan 20(1), 1-7.
25. Overend, R. P. and Chomet, E. (1987), Philos. Trans. Soc. Lond. A321,523-536.
26. Chum, H. L., Johnson, D. K., Black, S. K., and Overend, R. P. (1990), Appl. Biochem.
Biotechnol. 24/25, 1-14.
27. McCoy, B. J. and Madras, G. (2001), Chem. Eng. Sci. 56,2831-2836.
28. Austin, L., Shoji, K., Bhatia, V., Jindal, V., and Savage, K. (1976), Ind. Eng. Process Des.
Dev. 15(1), 187-195.
29. Wang, M., Smith, J. M., and McCoy, B. J. (1995),AIChE J. 41(6), 1521-1532.
30. McCoy, B. J. and Wang, M. (1994), Chem. Eng. Sci. 49(22),3773-3785.
31. McGrady, E. D. and Ziff, R. M. (1988), AIChE J. 34(12), 2073-2076.
32. Simha, R. (1941), J. Appl. Physiol. 12,569-578.
33. Jellinek, H. H. G. and White, G. (1950), J. Polymer Sci. 6(6),745-756.
34. Agarwal, N., McKean, W. T., and Gustafson, R. R. (1992), Appita 45(3), 165-169.
35. Converse, A a., Kwarteng, I. K., Grethlein, H. E., and Ooshima, H. (1989), Appl.
Biochem. Biotechnol. 20/21, 63-78.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0069/$20.00

Dilute-Sulfuric Acid Pretreatment


of Corn Stover in Pilot-Scale Reactor
Investigation of Yields, Kinetics,
and Enzymatic Oigestibilities of Solids

DANIEL J. SCHELL, * JODY FARMER, MILLIE NEWMAN,


AND JAMES D. McMILLAN
National Bioenergy Center, National Renewable Energy Laboratory,
1617 Cole Boulevard, Golden, CO 80401, E-mail: dan_schel/@nrel.gov

Abstract
Corn stover is a domestic feedstock that has potential to produce signifi-
cant quantities of fuel ethanol and other bioenergy and biobased products.
However, comprehensive yield and carbon mass balance information and
validated kinetic models for dilute-sulfuric acid (H 2S04 ) pretreatment of
corn stover have not been available. This has hindered the estimation of
process economics and also limited the ability to perform technoeconomic
modeling to guide research. To better characterize pretreatment and assess
its kinetics, we pretreated corn stover in a continuous 1 tf d reactor. Corn
stover was pretreated at 20% (w fw) solids concentration over a range of
conditions encompassing residence times 6f 3-12 min, temperatures of 165-
195°C, and H 2S04 concentrations of 0.5-1.4% (w /w). Xylanconversion yield
and carbon mass balance data were collected at each run condition. Perfor-
mance results were used to estimate kinetic model parameters assuming
biphasic hemicellulose hydrolysis and a hydrolysis mechanism incorpo-
rating formation of intermediate xylo-oligomers. In addition, some of the
pretreated solids were tested in a simultaneous saccharification and fer-
mentation (SSF) process to measure the reactivity of their cellulose compo-
nent to enzymatic digestion by cellulase enzymes. Monomeric xylose yields
of 69-71% and total xylose yields (monomers and oligomers) of 70-77%
were achieved with performance level depending on pretreatment sever-
ity. Cellulose conversion yields in SSF of 80-87% were obtained for some of
the most digestible pretreated solids.
Index Entries: Pretreatment; dilute-sulfuric acid; enzymatic conversion;
corn stover; xylan conversion; kinetics; pilot scale.

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 69 Vol. 105-108, 2003
70 Schell et al.
Introduction

The potential of biomass-derived ethanol for use as a transportation


fuel has been previously documented (1,2). One promising technology for
conversion of biomass to ethanol is an enzyme-based process utilizing a
pretreatment process to enhance the enzymatic conversion of cellulose to
soluble glucose. A variety of pretreatment processes utilizing mechanical,
thermomechanical, and thermochemical processing have been developed
to alter the structural and chemical composition of biomass to improve
enzymatic conversion (3,4). One pretreatment that has been extensively
explored is a high temperature, dilute-sulfuric acid (H2S04) process, which
also effectively hydrolyzes the hemicellulosic portion of the biomass to
soluble sugars.
Extensive work has been done on dilute-H 2S04 pretreatment of a
variety of feedstocks (3), however, relatively little work has been done
using corn stover. Corn stover, which includes the leaves, stalks, and cobs
of the corn plant, may be available in quantities that could support signifi-
cant production of ethanol and other bioenergy and biobased products.
Several groups have investigated dilute-H2S04 pretreatment of corn sto-
ver at low solids concentrations in batch, bench-scale reactors (5,6), but
little information on xylan hydrolysis is reported. The primary focus of
previous work has been on obtaining good enzymatic conversion of the
pretreated solids.
More recently, Esteghlalian et al. (7) performed dilute-H2S04 pre-
treatments at 10% solids concentration in a Parr reactor at conditions of
140-1BO°C, 0.6-1.2% (w /w) H 2S04, and estimated residence times span-
ning 1-60 min. They modeled biphasic xylan hydrolysis and determined
the rate constants using Arrhenius-type expressions with preexponential
factors dependent on the effective acid concentration (Le., acid concentra-
tion after neutralization by ash contained in the feedstock) to determine
the rate constants. The model-predicted monomeric yields of 80-90%
could be achieved at temperatures of 170-180°C with effective acid con-
centrations >1.0%.
Chen et al. (8) determined the kinetic parameters for dilute-H2S04
xylan hydrolysis modeled with biphasic xylan hydrolysis and formation of
intermediate xylo-oligomers. Pretreatments were performed in a Parr reac-
tor at 6% solids concentration using temperature conditions of 120-1S0°C,
acid concentrations of 0.44-1.9%, and residence times of 2-90 min. These
investigators calculated that maximum xylose yields were about 89%
regardless of reaction temperature.
The goal of the present work was to investigate dilute-H2S04 pretreat-
ment of corn stover in a pilot-scale (1 dry t of biomass/ d) continuous reac-
tor at high solids concentrations (20% [w fwD. We wished to work at
conditions and with equipment that would generate more commercially
relevant results than have been reported previously. The results produced
during our study were xylan yields with carbon mass balance closure data
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Dilute-H2S04 Pretreatment of Corn Stover 71

and enzymatic cellulose digestibilities at a variety of pretreatment condi-


tions. In addition, rate parameters were determined from the data for a
kinetic model based on biphasic xylan hydrolysis that incorporates inter-
mediate xylo-oligomer formation.
Calculating carbon mass balance across the pretreatment process is
a useful technique for assessing the accuracy of performance measure-
ments. Obtaining good carbon mass balance closure indicates internal
data consistency and provides a higher level of confidence in the accuracy
of the underlying data. This is increasingly important as technology
moves toward commercialization, because confidence in performance
data is essential for engineering companies to commit to guaranteeing
process performance. If carbon mass balance closures near 100% cannot
be obtained, then process stoichiometry or the accuracy of measuring one
or more of the carbon-containing process streams is suspect.

Materials and Methods


Corn Stover
Corn stover was obtained directly from Biomass AgriProducts
(Harlan, IA). The tub-ground material was approx 9 mo old (harvested in
fall 1999) when received at National Renewable Energy Laboratory (NREL)
(summer 2000) and was then used over a period of 9 mo. As received, corn
stover was washed by mixing with water in an agitated 6000-L conical tank
to remove dirt and other foreign matter. The washed corn stover was then
dried to approx 10% moisture before use. The composition of the washed
corn stover (on a dry basis) from an average of five randomly selected
samples from the lot was 37.1 % cellulose, 19.8% xylan, 2.5% arabinan, 1.6%
galactan, 1.4% mannan, 20.7 % lignin (acid soluble and insoluble), 7.8%
protein, 5.2% ash, and 2.4% acetate.

Pretreatment System
Dilute-H2S04 pretreatments were conducted in a continuous pilot-
scale reactor operating at a feed rate of approx 32 kg (dry basis)/h. The
process flow diagram of the pretreatment system is shown in Fig. 1 and
Schell et a1. (9) describe in more detail the pilot plant and the data acquisi-
tion and control system. Biomass is conveyed to the pretreatment system
from a feed hopper via a weigh belt and belt conveyor. The continuous
pretreatment system consists of acid supply tanks; a biomass mixer; a high-
temperature, high-pressure reactor system; and a flash tank. The pretreat-
ment reactor system is a vertical pulp digester supplied by Sunds
Defibrator, (now Metso Paper USA, Norcross, GA) and includes the reactor
and material feed (plug feeder) and discharge (reciprocating popet valves)
systems. The reactor is steam heated and can operate at temperatures of
150-200°C, and residence times of 3-20 min can be achieved by controlling
feedstock level in the reactor. A gamma-ray level sensor measures the level
of biomass in the pretreatment reactor. The acid delivery system consists of
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
72 Schell et al.

Sulfuric Acid
Water - - - ,

Pump

Pug MiU Mixer


Belt Conveyor CrossFccdcr

VeDtStte~s +-~------~~+-~

Pretreated CorD
Stover Slurry

Fig. 1. Diagram of pilot-scale dilute-H2S04 pretreatment system.

two fiberglass-reinforced plastic tanks (only feeding from one tank at a


time) and associated pumps. Acid is diluted to 5-12% (w /w) in the acid
tank, and a calibrated conductivity probe measures acid concentration.
Continuous operation of the pretreatment system begins at the feed
hopper. Washed and dried corn stover in totes is dumped into the feed
hopper. Feedstock is continuously metered from the hopper onto a weigh
belt that controls feed rate. A signal from the weigh belt provides feedback
control to the screw on the bottom of the hopper that controls weigh belt
loading. The feed rate is also used to control other additions of material
(e.g., acid and water). The corn stover then exits the weigh belt onto a belt
conveyor that delivers it to the pug mill mixer for mixing with dilute acid
and water. Water is added as needed to adjust the solids concentration in
the pretreatment reactor. The wetted feedstock is screw conveyed to a plug
feeder that compresses the material into an impermeable plug that is then
forced into the pretreatment reactor. Liquid pressed out of the material
by the plug feeder is pumped into the pretreatment reactor to maintain
desired acid and water concentrations. The corn stover enters the side of the
reactor and is conveyed by twin screws to the top of the reactor, where it
flows over a weir and enters the main body of the reactor. There is no
mixing in the main reactor body and the material essentially moves by plug
gravity flow to the reactor discharge port at the bottom of the reactor. A
rotating scraper on the bottom of the reactor facilitates movement of mate-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Dilute-H2S04 Pretreatment of Corn Stover 73

rial to the discharge port. Two reciprocating popet valves operating as a


pressure lock direct the pretreatment material into the flash tank. The reac-
tor is vented to remove non-condensables and this stream is sent to a con-
denser. The flash tank is a conical screw mixer, and vapor exiting from the
top of the tank is also sent to a condenser. The noncondensable fraction
from both of these streams is sent to a scrubber. The pretreated feedstock
exits the bottom of the flash tank.
Pretreatment Experimental Protocol
Pretreatments were performed by first establishing steady-state con-
ditions in the pretreatment reactor, i.e., ensuring that the feed rates ofbio-
mass, acid, water, and steam were all stable for at least 10-15 min after a
change in operating set points. After steady-state conditions were estab-
lished, the reactor was emptied and then allowed to fill to establish a level
vs residence time calibration. The reactor residence time was then set to the
desired value and the system was again allowed to come to steady state.
After approximately three residence times, the flash tank was reemptied
and pretreated material was collected over a period of three (or more)
residence times. During this collection period, the corn stover was sampled
before entering the pretreatment reactor, and its moisture content was
measured using a Denver Instruments (Arvada, CO) Model IR-I00 Infra-
red Moisture Balance. Additionally, samples of the vent streams from the
reactor and the flash tank were condensed and collected. At the end of the
collection period, the flash tank was emptied and a sample of the pretreated
slurry was obtained. The data acquisition and control system collected data
on flow rates, pressures, temperatures, and level every 30 s. These data
were averaged and used to define operating conditions and calculate yield
and mass balance results.
A total of 41 pretreatment runs were completed, including some
replicates, as summarized in Table 1. Conditions were varied from tem-
peratures of 165-183°C, residence times of 3-12 min, and acid concentra-
tions in the liquid phase of 0.5-1.41 % (w /w). All runs were conducted
using a 20% total solids concentration in the pretreatment reactor on an
unhydrolyzed feed solids basis. Five of the runs (nos. 37-41) were carried
out at conditions outside of these ranges and are discussed separately.
Table 1 shows the nominal acid concentration, which is different from the
effective acid concentration. Variability in the effective acid concentra-
tion may occur because the neutralizing capacity of the stover may not be
constant within a batch of material. The final pH of the post-pretreatment
liquor was used in this work rather than the effective acid concentration
because pH inherently accounts for the feedstock's neutralizing capacity.
Chemical Analysis
The composition of raw and pretreated corn stover was measured using
methods reported by NREL for determining biomass carbohydrates (10),
acid insoluble lignin (11), acid soluble lignin (12), ash (13), and acetate con-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
)..
:g
[ Table 1
e;, Pretreatment Run Conditions and Enzymatic Cellulose and Xylan Conversion Results

n
::r-
(1)
Pretreatment Conditions Xylan conversion
3
iii· Acid Cellulose Monomeric Total Furfural Unconverted Mass
q-
'<: Run Temperature Time concentration Final conversion xylose xylose yield xylan closure
til
:J no. (0C) (min) (% [w/w]) pH CSF (%) yield (%) yield (%) (%) (%) (%)
Q..
e;,
0· 1 165 8.1 1.37 1.28 1.55 79 64 67 14 10 92
Ii!"
n 2 165 8.1 1.37 1.28 1.55 63 67 19 11 97
::r-
:J 3 165 6.0 1.37 1.14 1.57 63 57 65 9 15 89
0
~ 4 165 8.2 0.71 1.85 0.99 35 55 7 27 90
'<: 5 165 8.1 1.35 1.17 1.66 70 71 14 11 95
6 165 8.0 1.39 1.19 1.64 82 71 71 13 11 95
7 165 6.4 0.71 1.84 0.90 31 53 10 34 97
8 165 10.1 1.37 1.16 1.77 59 61 13 8 82
9 165 8.0 1.41 1.05 1.78 83 66 63 15 13 90
10 166 10.2 0.71 1.74 1.20 44 59 16 24 99
11 166 8.0 1.37 1.14 1.69 62 63 21 9 93
12 169 3.9 0.71 1.52 1.12 56 65 7 22 94
13 170 8.3 0.74 2.10 0.89 33 54 6 40 100
14 170 6.1 0.74 2.05 0.81 27 57 7 39 103
15 171 5.1 0.73 2.22 0.57 17 43 9 34 86
16 174 5.8 1.37 1.18 1.78 63 67 18 9 94
17 174 2.9 1.37 1.27 1.38 63 68 6 11 85
18 175 3.8 1.38 1.29 1.49 66 72 7 13 92
19 175 2.9 1.37 1.17 1.49 64 70 12 15 97
20 175 8.1 0.94 1.52 1.59 81 59 61 17 7 85
21 175 8.1 0.95 1.76 1.36 71 50 63 17 23 103
22 175 4.2 0.94 1.71 1.13 70 49 67 8 18 94
-~~
c
- 23 175 5.0 0.94 1.63 1.28 78 59 70 10 21 101
.?> 24 175 6.1 0.95 1.60 1.41 67 62 69 17 12 98
N
c
8
)..
:g Table 1 (Con't)
~
0.. Pretreatment Run Conditions and Enzymatic Cellulose and Xylan Conversion Results
O:l
..,o·
~
Pretreatment Conditions Xylan conversion
Ib
3 Acid Cellulose Monomeric Total Furfural Unconverted Mass
tn·
~ Run Temperature Time concentration Final conversion xylose xylose yield xylan closure
~
::. no. (0C) (min) (% [w/w)) pH CSF (%) yield (%) yield (%) (%) (%) (%)25
0..
O:l 179 8.1 0.97 1.40 1.83 52 56 11 9 77

..,fD 26 179 4.7 1.03 1.78 1.22 50 53 24 19 97
~
::. 27 179 6.2 1.16 1.50 1.63 87 68 70 16 14 100
0
28 180 8.1 0.80 1.42 1.83 55 60 16 10 86
~
'<: 29 180 4.0 1.18 1.47 1.48 66 70 10 11 91
30 180 10.1 0.80 1.48 1.87 78 54 60 14 9 83
31 180 12.2 0.80 1.47 1.96 80 55 58 19 9 86
32 180 10.1 0.99 1.33 2.02 81 54 55 17 7 79
33 180 5.1 1.18 1.56 1.50 56 63 13 9 86
34 180 3.1 1.19 1.49 1.34 66 71 11 13 95
35 180 6.0 0.99 1.37 1.76 54 58 14 11 82
36 180 5.8 1.27 1.39 1.74 66 63 31 7 101
37 181 4.1 1.46 1.52 65 65 27 13 105
38 183 5.3 1.49 1.69 84 67 66 27 13 106
39 190 1.39 1.08 67 70 19 10 99
40 191 1.17 1.43 76 65 77 5 18 99
41 195 1.35 1.13 68 61 19 7 87
Average" 165.4 8.28 1.38 1.19 1.65 81.3 66.1 67.0 15.8 10.9 93.7
SD 0.1 0.06 0.02 0.09 0.09 2.3 3.7 3.5 3.1 1.3 2.6

c "Averages and SDs for runs I, 2, 5, 6,9, and 11 all were targeted for the same run conditions.
bPretreatment reactor operated at zero level in a mode that achieves a short but unknown residence time estimated at 45-75 s.
-Y'~
c
- cCorn stover preimpregnated with acid before pretreatment; actual acid concentration in the pretreatment reactor was not controlled but
~
targeted to achieve an acid concentration equivalent to 1.2% (w /w).
c
'"
c....,
76 Schell et al.
tent (14). Protein content was calculated as 6.25 times the nitrogen content,
which was measured using a Perkin-Elmer Series 2400 CHNS/O Elemental
Analyzer (Norwalk, CT).
Concentrations of soluble components in the hydrolysate liquors
(monomeric and oligomeric sugars, acetic acid, 5-hydroxymethanol fur-
fural, and furfural) were measured using techniques previously reported
(9). Total xylose is defined as the sum of monomeric and oligomeric
xylose. Measurements of hydrolysate liquor pH were made the following
day with the liquor at ambient temperature (approx 25°C). The amount of
insoluble solids in the hydrolysate was measured using a technique (pro-
cedure A) reported by NREL (15).
Enzymatic Digestibility
Enzymatic digestibility or cellulose conversion was determined by
hydrolyzing the cellulose contained in washed pretreated solids using a
simultaneous saccharification and fermentation (SSF) process. The washed
solids were produced by the aforementioned procedure for measuring in-
soluble solids. SSF was conducted in a laboratory-shaking incubator (150
rpm) at a working volume of 100 mL in 250-mL baffled flasks with water
traps. The washed solids were loaded to a level of 6% (w Iw) cellulose (cel-
lulose measured as discussed above) and Iogen (Ottawa, Canada) cellulase
enzyme (lot no. BRC 191095) was added to a level of 15 filter paper units
(FPU)/g of cellulose. The medium consisted of yeast extract (1% [w/v]),
peptone (2% [w Iv]), and citrate buffer (0.05 M). The initial pH was adjusted
to 5.2 using NaOH, and then the culture was inoculated with the yeast, Sac-
charomyces cerevisiae DsA, to achieve an initial optical density (at 600 nm) of
0.5. The flask was maintained at 320C for 7 days and sampled daily. Ethanol
concentration was measured by high-performance liquid chromatography
using a Bio-rad (Richmond, CA) HPX-87H column operating at 65°C with a
0.01 N H 2S04 solution mobile phase and a refractive index detector. Ethanol
yield was calculated based on the theoretical ethanol concentration from
cellulose after subtracting the ethanol added from the inoculum. The theo-
retical ethanol yield number is assumed to be a conservative estimate of
cellulose conversion but is also a number that represents realistic conversion
results, since the SSF test is done at conditions roughly similar to those that
might be encountered in an actual process (i.e., 6% cellulose concentration,
32°C, and a low enzyme loading of 15 FPU I g of cellulose).
Xylan Conversion Yields and Mass Balance Calculations
Calculation of xylan conversion yields (i.e., monomeric xylose, oligo-
meric xylose, furfural, and unconverted xylan) and carbon mass balances
followed a previously reported methodology (16) and relies on accurate
measurements of flow rates and component concentrations in the inlet and
outlet streams. We measured flow rates for corn stover (weigh belt), acid,
and water to the pug mill mixer (Magnetic flowmeter and Coriolis mass
flowmeter, respectively), and the pretreatment-reactor vent stream

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Dilute-H2 S04 Pretreatment of Corn Stover 77
(Coriolis mass flowmeter). The flash-tank vapor flow rate was calculated
from an energy balance around the pretreatment reactor and flash tank.
The hydrolysate slurry flow rate was calculated from an overall mass
balance. Component concentrations were measured for corn stover (mois-
ture and composition), vapor vent streams (furfural and acetic acid), and
the hydrolysate (fraction insoluble solids, solids composition, liquor com-
position). Components yields based on the amount of xylan contained in
the corn stover and xylan mass balance closure are reported as the mass of
output xylan products (monomeric xylose, oligomeric xylose, furfural,
and unconverted xylan) over the mass of input xylan.
Kinetics of Hemicellulose Hydrolysis
Xylan hydrolysis kinetics was modeled as biphasic hemicellulose
hydrolysis with a hydrolysis mechanism incorporating formation of inter-
mediate xylo-oligomers (8) as follows:
Fast-reacting xylan
kl
'lII
Xylose oligomers k3 Xylose k4 Decomposition
7' ~ ~
k2
Slow-reacting xylan

The kinetic rate parameters, ki' were modeled using Eq. I, which
applies the Arrhenius relationship and general acid-base catalysis:
ki=(k~ + k7[H+] + k~H[OH-]) exp( ~~i) (1)
Since all pretreatments were conducted at very low pH «2.25), the
hydroxyl ion term was ignored and rewriting the hydrogen concentrations
in terms of the pH gives
k i=(k~ + k7[ lO-pH)) exp( ~~i) (2)
Using pH is more appropriate than using the effective acid concentra-
tion (i.e., acid concentration after neutralization by ash), which could effec-
tively be zero if there is insufficient acid.
The model contains 13 parameters (4 each of kt, kr, Ei , and the fraction
of fast-reacting xylan) that must be determined. A genetic algorithm
(Evolver, an Excel add-in from Palisade, Newfield, NY) adjusted the 13
parameters to minimize an error term (E):
E = X + 0 + Xl (3)
in which X, 0, and Xl are the sum of the square of the errors between the
measured yield and the predicted yield for monomeric and oligomeric
xylose, and unconverted xylan, respectively. When the measured total
xylose concentration was less than the measured monomeric xylose con-
centration, probably owing to measurement error, the value for oligomeric
xylose yield was set to zero; this occurred in runs 9, 36, and 38.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
78 Schell et al.

Results
Table 1 summarizes run conditions and key results from each of the
41 runs. It presents the pretreatment run conditions (temperature, time,
and acid concentration), the final pH of the hydrolysate liquors, the com-
bined severity factor (CSF), cellulose conversion determined by SSF test-
ing, and xylan conversion results. The CSF is defined as (17,18)

r-1001) - pH
CSF = 10glO(t x exp r 14.75 (4)

in which t and T are pretreatment residence time (min) and temperature


(0C), respectively. The CSF provides a method for consolidating the effects
of pretreatment temperature, residence time, and acid concentration into a
single parameter, which can be useful for analyzing results.
Replicate runs at the same targeted operating conditions were per-
formed to assess reproducibility. At the bottom of Table I, averages and SDs
are given for six runs (1,2,5, 6, 9, and 11) that were all targeted for the same
pretreatment conditions (165°C, 8 min, 1.4% [w /w] acid concentration). The
runs were performed on four different days spanning a period of 6 mo with
two runs performed on the same day for two of the different dates. Although
there is little variability in the operating conditions, there is more variability
in the yields, cellulose conversion, and carbon mass balance results. One SD
was approx 5% of the average value for many of these results, except for
furfural and unconverted xylan yields, for which the value of the SD was
much higher (up to 20%). This information is useful for accessing the overall
reproducibility of results from our pretreatment system.
As noted in Table I, runs 37 and 38 were performed with acid pre-
impregnated corn stover using an impregnation device previously
described (18). After acid soaking, the stover was dried to approx 10%
moisture. This material was fed to the pretreatment reactor but no addi-
tional acid was added at the pug mill mixer. The acid-impregnated stover
was used to test the hypothesis that mass transfer of acid into the feedstock
may limit performance. These runs were performed at conditions close to
those of runs 29 and achieved similar xylan conversion yields, which sug-
gests that acid mass transfer was not limiting performance.
Runs 39-41 were performed without a detectable level in the pretreat-
ment reactor in a "flowthrough" mode that achieves a short but unknown
residence time estimated to be 45-75 s. This operating mode was achieved by
cycling the reciprocating popet valves at a high frequency so that no solids
built up within the reactor. The short residence time required us to conduct
pretreatments at higher temperatures (190-195°C) than were used in the
other experiments in order to achieve high yields. In general, the cellulose
and xylan conversion results obtained under these conditions are similar to
those achieved in the lower-temperature pretreatments. However, a particu-
larly high total xylose yield of 77% was achieved in run 40, which was per-
formed at 191°C and 1.17% (w /w) acid. This result is intriguing, but

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Dilute-H2S04 Pretreatment of Corn Stover 79
80~----------------------------------------------~

70
o
o
60
o o o '"
o
--
~ 50
l!
CD o
'"
'"
'"
>= 40
!
~30

20

I ",Monomeric I
10 o Total

O+-----~----~----~----~----~----._----._--~

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1


Combined Severity Factor

Fig. 2. Monomeric and total xylose yields of pretreated com stover as function of CSF.

additional work needs to be performed to confirm that better performance


can be obtained at this higher temperature and shorter residence time.

Xylan Conversion
Figure 2 presents monomeric and total xylose yields as a function of
the CSF. The results show peaks in monomeric xylose yield at CSF ( 1.6-1. 7
and total xylose yield at CSF ( 1.4-1.5, with both yields reaching highs of 70
to 71%. At lower severities, total xylose yields were much greater than
monomeric xylose yields, but the two values were nearly equal above a CSF
of 1.6. In addition, above a CSF of 1.7, both monomeric and total xylose
yields declined, presumably because the harsher pretreatment conditions
at these higher severities degraded more of the xylose to furfural and other
degradation products.
The fractionation of xylan into monomeric and oligomeric xylose,
furfural, and unconverted xylan, as well as the overall xylan mass balance
is shown in bar chart form in Fig. 3, which plots run numbers in order of
increasing pretreatment severity. In addition to the trends just discussed,
Fig. 3 shows the predicted trend of increasing furfural and decreasing
xylan with increasing pretreatment severity. The top of the bar shows the
overall mass balance closure for xylan, which is usually in the range of 90-
100%, except at the higher pretreatment severities, where mass balance
begins to drop below 90%. We believe that other unmeasured degradation
products are being produced under these conditions that are not included
in the mass balance calculation.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
80 Schell et al.

120

100

......
'ii
u 80 -
;:
f

-
0
CD
.c 60
e
~
'ij
:; 40

20

0
~~~~~~~~~~~N~~~~~~~-N~~~C~~~~~c~~~~gM~

Run.
Monomeric Xylose • Oligomeric Xylose .Xylan 0 Fur1ural I

Fig. 3. Yields of monomeric and oligomer xylose, furfural, and unconverted xylan
plotted by run number in order of increasing severity. The height of each bar repre-
sents the total xylan mass balance closure.

95

90
A

-
--
~ 85
c
0 A
A
A A
A
... 80
A
ii A A
~
6 75
0
:
0 70 A
:2 A
"i
0 65

60

55
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1
Combined Severity Factor

Fig. 4. Enzymatic digestibility of pretreated and washed corn stover cellulose as


function of CSF.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Dilute-H2S04 Pretreatment of Corn Stover 81
Table 2
Estimated kinetic parameters

kt (min-I) 2.61 1025


X 1.00 X 1015 1.03 X 1028 1.00 X 1015
kr (min-I. M-l) 8.54 X 1029 4.59 X 1029 1.00 X 1030 1.00 X 1015
E; (kcal/ g mol) 57.4 59.9 58.6 33.4

80 ~------------------------------------------~

70


&

• • .. .
• ..
• Monomeric Xylose
• Oligomeric Xylose

.. Unconverted Xylan

10 20 30 40 50 60 70 80
experimental Yield <%)

Fig 5. Predicted (kinetic model) vs measured values of monomeric and oligomeric


xylose and unconverted xylan.

Cellulose Conversion
Figure 4 shows cellulose conversion or enzymatic digestibility of the
pretreated solids plotted against the CSF. In general, there appears to be a
trend of increasing cellulose digestibility with increasing pretreatment
severity, particularly at CSFs <1.6. At CSFs >1.6, digestibilities >80% were
routinely obtained, and the highest value obtained was 87%. The typical
composition (dry basis) of well-pretreated corn stover was 55-60% cellu-
lose, 3-7% residual xylan, and 27-31 % lignin.
Pretreatment Kinetics
Kinetic parameters were estimated using data from runs 1-38 and are
shown in Table 2. The ability of the kinetic model to fit the xylan hydrolysis
data is presented in Fig. 5, which shows that the model predicts xylan
hydrolysis performance across the data set fairly well. The best fit esti-
mated value of the fast-reacting xylan fraction was 0.72, slightly higher
than previous estimates (7, 8).
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
82 Schell et al.
90~------~----------------------------~--------~
85
--
<ft 80
:; 75
'i
>= 70
~ 65
~60
§ 55
~II 50
:::& 45
1+-------- Experimental data range
40
35+-------4--------r-------r-------.----~~------~

160 165 170 175 180 185 190


Temperature (Oe)

Fig. 6. Predicted maximum monomeric and total xylose yields as function of tem-
perature and pH when maximizing based on monomeric xylose yield.

The kinetic model was used to predict both maximum monomeric and
total xylose yields as a function of temperature and pH (at 1.0, 1.5, and 2.0)
by adjusting residence time (i.e., calculating the residence time that maxi-
mizes xylose yield at a fixed temperature and pH). The results using mono-
meric xylose yield as the criterion to maximize are shown in Fig. 6. The
vertical dotted lines at temperatures of 165 and 183°C show the range inves-
tigated in the experiments, thus demonstrating that the graph
includes some extrapolation beyond the range within which the model para-
meters were estimated. The graph highlights several interesting results.
First, maximum yields increase with increasing temperature and decreas-
ing pH. Second, maximum total xylose yields are always greater than
monomeric xylose yields, but the differences become small (0.5-2.0 percent-
age points) at low pH and high temperature. Third, the model predicts that
the time required to achieve maximum xylose yield is always shorter for
total xylose yield than for monomeric xylose yield (not shown).
Figure 7 presents the maximum monomeric and total xylose yields
when either maximum monomeric xylose or maximum total xylose is
used as the maximization criterion. The results are presented as a func-
tion of temperature at pHs of 1.0 and 2.0. For clarity, the lines of mono-
meric and total xylose yield when monomeric xylose is the maximization
criterion are not shown at pH 1.0, since these lines lie very close to (almost
on top of) the lines based on maximizing total xylose yield. Figure 7 high-
lights the following results: At pH 1.0, the maximum yields are similar
regardless of which criterion is used to maximize yield. At pH 2.0, maxi-
mizing on total xylose yield gives higher overall total xylose yields than
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Dilute-H2S04 Pretreatment of Corn Stover 83
90~------~---------------------------r---------.

85
l80
'tI 75
! 70
R65
~60
§ 55
~1\ 50
:245
40 .--~------- Experimental data range

~~------~-------.--------'-------~--~~-r------~
160 165 170 175 180 185 190
Temperatura (Oe)
Maximizing monomeric xylose yield MaxImizing total xyiose yield
------- Totel xylose - - 0 Total xyiose
_ _ Monomeric xylose _ Monomeric xylose

Fig. 7. Predicted maximum monomeric and total xylose yields as function of tem-
perature and pH when maximizing based on either monomeric or total xylose yield.

when maximizing on monomeric xylose yield but produces significantly


lower yields of monomeric xylose than when maximizing based on mo-
nomeric xylose yield.

Discussion
Our highest xylose yields obtained with corn stover (70-77%) are sig-
nificantly lower than the best results reported by Esteghlalian et a1. (7) and
Chen et a1. (8) (85-90%). However, our results were obtained at pilot scale
using significantly higher solids concentrations (20% compared with 10%).
Besides the obvious differences in dilute-acid pretreatment reactor systems
(small-scale batch agitated systems compared with a continuous unagitated
pilot-scale system), the higher solids concentration also produces higher
sugar concentrations in the hydrolysate that may be affecting yields
because the reaction rates (k3 and k4 ) depend on xylose concentration. We
believe one or both of these factors contribute to the differences observed
between our results and those previously reported.
Performance variability and scatter in the data are apparent from the
results presented in Figs. 2-5 and the previous discussion of reproducibil-
ity. Several factors possibly contribute to this variability, including uncer-
tainties in the residence time calibration, temperature nonuniformities
within the reactor, changes in feedstock acid-neutralizing capacity that
affect the final measured acid concentration, as well as normal measure-
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
84 Schell et al.
ment and control errors associated with operating continuous pilot-scale
equipment. The high degree of scatter in the performance results reflects a
combination of these factors. Kinetic modeling was used to develop an
overall representation of performance trends.
The most important predictions from the kinetic modeling are that
low pHs are required to achieve the highest xylose yields and that yields
improve at higher temperatures (although shorter residence times are
required). Esteghlalian et a1. (7) reported similar kinetic behavior, although
other investigators have reported that temperature has little effect on maxi-
mum xylose yields (8,19). Differences in reactor systems, solids concentra-
tions, and accounting for acid neutralization, as well as measurement
uncertainties may explain some of the discrepancies. The highest total
xylose yield, 77% achieved in run 40, was obtained using a higher tempera-
ture (190°C). However, this was only a single run and additional work is
needed to confirm this finding. Although, it is outside the scope of this
article, we are also seeing significant corn stover compositional variabil-
ity, which may also be affecting both pretreatment and enzymatic hy-
drolysis kinetics (unpublished results).
Demonstrating effective pretreatment technology at pilot scale using
economically attractive conditions is required to move lignocellulosic con-
version technology toward commercialization. The work reported here, in
which performance data were obtained with component mass balance clo-
sures near 100%, represents an initial effort to generate accurate and reliable
pilot-scale performance data that can be used to more rigorously analyze
process economics than have been performed in the past (20). Future work
will examine the impact that compositional variability of corn stover has on
performance and will apply measurement uncertainty analysis (21) to assess
the accuracy of pretreatment yield and mass balance results.

Acknowledgments
We gratefully acknowledge the technical assistance of Bob Lyons
and Bryan Rohrback with pilot plant operation, Raymond Ruiz, David
Templeton, Amie Sluiter, and Bonnie Hames with chemical analysis of
process samples, and David Templeton with procurement of the corn
stover. The Biochemical Conversion Element of the Office of Fuels Devel-
opment of the US Department of Energy funded this work.

References
1. Lugar, R. and Woolsey, R. (1999), Foreign Affairs 78,88-102.
2. Lynd, L., Cushman, J., Nichols, R., and Wyman, C. (1991), Science 251,1318-1323.
3. Hsu, T. (1996), in Handbook on Bioethanol: Production and Utilization, Wyman, c., ed.,
Taylor & Francis, Washington, DC, pp. 179-212.
4. Chang, V. and Holtzapple, M. (2000), Appl. Biochem. Biotechnol. 84-86,5-37.
5. Knappert, D., Grethlein, H., and Converse, A. (1980), Biotechnol. Bioeng. 22,1449-1463.
6. Torget, R., Walter, P., Himmel,M., and Grohmann,K. (1991),Appl. Biochem. Biotechnol.
28/29, 75-86.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Dilute-H2 S04 Pretreatment of Corn Stover 85
7. Esteghlalian, A, Hashimoto, A, Fenske, J., and Penner, M. (1997), Bioresour. Technol.
59,129-136.
8. Chen, R., Lee, Y., and Torget, R. (1996), Appl. Biochem. Biotechnol. 57/58,133-146.
9. Schell, D., Riley, c., Dowe, N., Farmer, J., Ibsen, K., Ruth, M., et al. (2003), Bioresour.
Technol., submitted.
10. Ruiz, R. and Ehrman, T. (1996), NREL Analytical Procedure, No. 002, National
Renewable Energy Laboratory, Golden, co.
11. Templeton, D. and Ehrman, T. (1995), NREL Analytical Procedure, No. 003, Renew-
able Energy Laboratory, Golden, CO.
12. Ehrman, T. (1996), NREL Analytical Procedure, No. 004, National Renewable Energy
Laboratory, Golden, CO.
13. Ehrman, T. (1994), NREL Analytical Procedure, No. 005, National Renewable Energy
Laboratory, Golden, co.
14. Ruiz, R. and Ehrman, T. (1996), NREL Analytical Procedure, No. 017, National
Renewable Energy Laboratory, Golden, co.
15. Eddy, F., Okafor, J., and Roberson, C. (1996), NREL Analytical Procedure, No. 018,
Procedure A, National Renewable Energy Laboratory, Golden, co.
16. Hatzis, c., Riley, c., and Philippidis, G. (1996), Appl. Biochem. Biotechnol. 57/58,
443-459.
17. Chum, H., Johnson, D., Black, S., and Overend, R. (1990), Appl. Biochem. Biotechnol.
24125,1-14.
18. Nguyen, Q., Tucker, M., Keller, F., and Eddy, F. (2000), Appl. Biochem. Biotechnol.
84-86,561-576.
19. Wright, J. and d' Agincourt, C. (1984), SERIITR-231-2074, Solar Energy Research
Institute, Golden, co.
20. Wooley, R., Ruth, M., Glassner, D., and Sheehan, J. (1999), Biotechnol. Prog. 15,
794-803.
21. Schell, D., Saez,J., Hamilton,J., Tholudur, A, and McMillan, J. (2002), Appl. Biochem.
Biotechnol. 98-100, 509-523.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 / 0087 / $20.00

Hydrothermal Pretreatment Conditions


to Enhance Ethanol Production
from Poplar Biomass

MARIA JOSE NEGRO, PALOMA MANZANARES,


IGNACIO BALLESTEROS, JOSE MIGUEL OLIVA,
ARACELI CABANAS, AND MERCEDES BALLESTEROS*
DER-CIEMAT, Avda. Complutense 22,
28040 Madrid, Spain, E-mail: m.ballesteros@ciemat.es

Abstract
Pretreatment has been recognized as a key step in enzyme-based conver-
sion processes of lignocellulose biomass to ethanol. The aim of this study is to
evaluate two hydrothermal pretreatments (steam explosion and liquid hot
water) to enhance ethanol production from poplar (Populus nigra) biomass by
a simultaneous saccharification and fermentation (SSF) process. The compo-
sition of liquid and solid fractions obtained after pretreatment, enzymatic
digestibility, and ethanol production of poplar biomass pretreated at differ-
ent experimental conditions was analyzed. The best results were obtained in
steam explosion pretreatment at 210°C and 4 min, taking into account cellu-
lose recovery above 95%, enzymatic hydrolysis yield of about 60%, SSF yield
of 60% of theoretical, and 41 % xylose recovery in the liquid fraction. Large
particles can be used for poplar biomass in both pretreatments, since no sig-
nificant effect of particle size on enzymatic hydrolysis and SSF was obtained.
Index Entries: Poplar; ethanol; pretreatment; steam explosion; liquid hot
water.

Introduction
Lignocellulose biomass represents a vast resource that could be trans-
formed into fuel ethanol and can be produced from crops specifically
grown to yield biomass for energy purposes. Among the fast growing
species cultivated in short cycles for biomass production, Populus nigra,
used in the present work as feedstock, is considered a promising energy
crop for central and south Europe owing to its high yield and drought-
resistant features (1).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 87 Vol. 105-108, 2003


88 Negro et al.

The structure of lignocellulosic biomass is complex and consists of


three main fractions (hemicellulose, cellulose, and lignin) that must be
separately processed to achieve efficient conversion of carbohydrates to
ethanol.
Among the different existing technologies for ethanol production,
enzyme-based cellulosic biomass conversion processes offer the potential to
achieve competitive prices in the longer term (2). The relevance of pretreat-
ment in such processes has been recognized for a long time (3). The purpose
of the pretreatment is to increase reactivity and accessibility of cellulose to
enzymatic hydrolysis by removing lignin and hemicellulose and increasing
the porosity of the materials. Pretreatment must meet the following require-
ments: improve the formation of sugars or the ability to subsequently form
sugars by enzymatic hydrolysis, avoid the degradation or loss of carbohy-
drates, avoid the formation of inhibitory byproducts to the subsequent hy-
drolysis and fermentation processes, and be cost-effective (2).
Hydrothermal pretreatment refers to the use of water as liquid or
vapor or both and heat to pretreat biomass. Autohydrolysis steam explosion
has been proved to be an adequate method for pretreating lignocellulosic
biomass, which efficiently increases enzymatic hydrolysis of hardwood and
agricultural residues; however, it is less effective for softwoods. The process
causes hemicellulose degradation and lignin transformation owing to high
temperatures, thus increasing the potential of cellulose hydrolysis (4). How-
ever, it is often found to generate inhibiting hydrolysates (5,6). The factors
that affect steam explosion pretreatment are residence time, temperature,
chip size, and moisture content (7). Optimal hemicellulose solubilization
and hydrolysis can be achieved by either high temperature and short resi-
dence time (270°C, 1 min) or lower temperature and longer residence time
(190°C, 10 min) (7). Recent studies indicate that low temperature and long
residence time are more favorable (8).
Liquid hot water pretreatment, in which biomass is exposed to pres-
surized hot water appears to have the potential to generate reactive fiber
(9-11), recover most of the pentosans (12), and produce hydrolysate that
results in little or no inhibition of solubilized sugar fermentation (13,14).
When using hydrothermal pretreatments to fractionate hardwood and
agricultural bypro ducts, acetic acid is formed by the hydrolysis of acetyl
groups in hemicellulose. High temperature stimulates the breakdown of
the released sugars during pretreatment, generating decomposition prod-
ucts, such as furan derivatives: furfural is formed from pentose sugars and
5-hydroxymethylfurfural (HMF) from hexose sugars. The furan deriva-
tives can, in turn, be degraded even further. HMF can be further broken
down to form equimolecular amounts of levulinic and formic acids (15),
while furfural can be degraded to formic acid. The furan derivatives and
soluble phenolic compounds from lignin will react further to form some
polymeric material. The formation of furfural and HMF is undesirable
because they represent a loss of fermentable sugars and may also inhibit
fermentation.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Poplar Pretreatment 89
In the present study different pretreatment conditions (particle size,
time, and temperature) of steam explosion and liquid hot water were
assayed to enhance ethanol production from poplar biomass. Pretreat-
ment effectiveness was evaluated in terms of hemicellulose-derived sugar
recovery in the liquid fraction, cellulose recovery in the solid fraction,
enzymatic hydrolysis yield, and simultaneous saccharification and fer-
mentation (SSF) yield of cellulose to ethanol.

Materials and Methods


Feed Materials
Chipped biomass (5% moisture) from P. nigra was supplied by The
Renewable Energies Development Centre at Lubia (Soria, Spain). On aver-
age, poplar biomass contained (dry wt basis) 43.5% glucose, 17.4% xylose,
2.5% galactose, 1.7% arabinose, 2.8% mannose,26.2% acid-insoluble lignin,
1.8% ash, and 4.1 % others. The chips were milled using a laboratory ham-
mer mill (Restsch). Milled material was further separated into two different
fractions (2-5 and 12-15 mm) using a portable sieve shaker. Both fractions
were submitted to hydrothermal pretreatment assays.
Pretreatment
Steam explosion pretreatment was performed by applying Masonite
technology in a small batch plant described in a previous work (16,17),
equipped with a 2-L reaction vessel designed to reach a maximum oper-
ating pressure of 4.12 MPa. The reactor was filled with 100 g of feed stock
per batch and was then directly heated with saturated steam to the
desired temperature. After the explosion, the material was recovered in
a cyclone, and the wet material was cooled to about 40°C and filtered for
solid recovery. Poplar biomass was treated at 190 and 210°C and 4- and
8-min reaction time.
Liquid hot water pretreatment was performed in a laboratory-scale
stirred autoclave (model EZE-Seal; Autoclave Engineers, Erie, PA). The
stainless-steel Hastelloy-C reactor has a total volume of 500 mL, with an
electric heater and magnetic agitation. Cooling water was circulated
through a serpentine coil to cool the reactor content at the end of each run.
The amount of dry feedstock loaded corresponds to 40 g, and water was
added at 1/10 (w Iv) dry solid/liquid ratio. The working vessel was 400
mL. The reactor content was initially at room temperature. The heating
ra te was between 2.9 and 3.4°C / min (18). Pretreatment was considered to
be completed when the target temperature was reached, which took 48,56,
60, 63, and 66 min for 180, 210, 220, 230, and 240°C, respectively. After
pretreatment the heater was turned off, the reactor was removed from the
heating jacket, and cooling water was charged through the serpentine coil.
The content of the reactor cooled down to 130°C in about 2 min. The reactor
was kept sealed, and the slurry was agitated until the reactor was cooled
to about 40°C, and sampled.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
90 Negro et at.

After completion of pretreatment assays, the wet material was filtered


and washed with water to recover the solid fraction. This solid fraction was
analyzed for xylans, glucans, and acid-insoluble lignin content. The carbo-
hydrate, furfural, HMF, acetic acid, formic acid, and levulinic acid content
of the undiluted filtrate was also analyzed.
All pretreatment assays were performed in duplicate. To evaluate the
effect of steam explosion parameters (temperature, time, and chip size), a
23 factorial design was applied. In liquid hot water pretreatment, tempera-
ture and chip size parameters were considered. A statistical analysis of
variance was performed on data from each pretreatment using the statis-
tical software program StatGraphics® Plus (Manugistics, Rockville, MD).

Microorganism and Growth Conditions


Kluyveromyces marxianus CECT 10875, a thermotolerant mutant yeast
strain obtained in our laboratory (19), was used in SSF experiments. Active
cultures for inoculation were prepared by growing the organism on a
rotary shaker at 150 rpm for 16 h at 42°C in a growth medium (initial pH =
5.5) containing: 5 giL of yeast extract (Difco), 2 giL of NH4Cl, 1 giL of
KH2P04 , 0.3 giL of MgS04 ·7 H 20, and 30 giL of glucose.

Enzymatic Hydrolysis and SSF


The washed solid fractions after both pretreatments were used as
substrates for enzymatic hydrolysis and SSF experiments. Enzymatic
hydrolysis was performed in 100-mL Erlenmeyer flasks, each containing
25mL of 0.1 M sodium acetate buffer (pH4.8) at 10 % (w Iv) dry pretreated
substrate loading at 50°C for 72 h. Enzyme loading of 15 filter paper
units I g of dry pretreated substrate of Celluclast 1.5L and 12.6 IV I g of
dry pretreated substrate of ~-glucosidase Novozyme 188 was employed.
Enzymes were a gift from Novo Nordisk (Bagsvaerd, Denmark).
SSF experiments were carried out in 100-mL Erlenmeyer flasks, each
containing 25 mL of fermentation medium (growth medium described
earlier), which were agitated at 150 rpm. Glucose was substituted by 10%
(w Iv) dry pretreated substrate concentration. The cellulolytic complex was
also added as in enzymatic hydrolysis assays. In the SSF experiments,
flasks were inocula ted with 4% (v I v) yeast cultures (corresponding to a cell
addition of 0.25 giL), and assays were conducted at 42°C for 72 h. The pH
was not controlled during enzymatic hydrolysis and SSF experiments.

Analytical Methods
The carbohydrate content of the liquid fraction after both pretreat-
ments was determined by performing a mild acid hydrolysis (3%[v Iv]
H 2S04,120°C, and 30 min) and measuring xylose, arabinose, galactose
and mannose concentration by high-performance liquid chromatogra-
phy (HPLC) in a 1081B Hewlett Packard (HP) liquid chromatograph with
refractive index detector. The following HPLC conditions were used: an

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Poplar Pretreatment 91
Aminex HPX-87P column (300 x 7.6 mm) (Bio-Rad, Hercules, CA), a tem-
perature of 85°C, and water as eluent at 0.6 mLI min. HPLC was also used
to analyze the liquid fraction for acetic, formic, and levulinic acid. Fur-
fural and HMF content was analyzed by HPLC in an 1100 HP liquid
chromatograph, equipped with a 1040A diode-array detector. The sepa-
ration was performed with a Bio-Rad Aminex HPX-87H stainless steel
(300 x 7.6 mm) column. The gradient elution was 82% H 2S04 (5 rnM)
and 18% acetonitrile. Column temperature was 55°C and flow rate was
0.3 mL/min.
Ethanol was measured by gas chromatography, using an HP 5890
Series II apparatus with a flame ionization detector and a column of
Carbowax 20 M (2 m x 1.125 in.) at 85°C. Injector and detector temperature
was 150°C.
National Renewable Energy Laboratory standard methods (20,21)
were used to determine cellulose, xylans, and acid-insoluble lignin content
in both raw and pretreated solid material.

Results
Steam Explosion and Liquid Hot Water Pretreatments
Poplar biomass milled to different particle sizes (2-5 and 12-15 mm)
was subjected to two hydrothermal pretreatments: liquid hot water at 180,
210,220,230, and 240°C; and steam explosion at 190 and 210°C and 4- and
8-min residence time. The goal of our study was to evaluate pretreatment
conditions (particle size, time, and temperature) on hemicellulose recovery
in the liquid fraction, cellulose recovery in the solid fraction, cellulose sus-
ceptibility to enzymatic hydrolysis, and SSF yield of cellulose to ethanol.
The chemical compositions of solid and liquid fractions after liquid
hot water and steam explosion pretreatments are given in Tables 1 and 2,
respectively. Data represent the averages of duplicate pretreatments.
Solid recovery (expressed as solids remaining after pretreatment
divided by 100 g of dry raw material) was affected by pretreatment condi-
tions. Solid recoveries ranged from 88.7 to 60%, and 66 to 59% for liquid hot
water and steam explosion pretreatments, respectively. For liquid hot
water, higher solubilization was obtained at harsher conditions, and no
significant dependence of chip size over the range studied was observed.
A similar pattern of solid recovery decreasing at increasing steaming tem-
peratures was found for steam explosion treatment, again showing little
effect of particle sizes.
Cellulose contents (gl 100 g of pretreated material) in the washed solid
fractions for both pretreatments and all conditions assayed increased in
relation to untreated material (39.5% cellulose). After pretreatment, cellu-
lose content ranged from 41 to 58% and from 56 to 59% for liquid hot water
and steam explosion pretreatment, respectively. The maximum cellulose
content in solid fraction for LWH pretreatment (57.5%) was obtained at
220°C and large particle size (12-15mm).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
)..

~
ib'
Q..

o·tJ:l Table 1
n
:;,- Chemical Composition of Solid and Liquid Fraction
(1)
3 After Liquid Hot Water Poplar Pretreatment
in'
q-
'<: 180°C 210°C 220°C 230°C 240°C
:J
'"
Q..
Particle sizes (mm) 2-5 12-15 2-5 12-15 2-5 12-15 2-5 12-15 2-5 12-15
o·tJ:l
fD
9- Solid recovery (g/100 g dry wt) (%) 88.7 88.2 69.5 70.7 64.5 64.5 64.5 63.4 60.1 60.0
:J
0
Solid fraction (%, [w/w])
~
'<: Ash 1.2 0.9 1.0 1.1 0.8 1.1 1.3 0.9 1.6 0.8
Cellulose 42.2 40.7 52.7 51.8 54.0 57.5 53.3 55.9 52.1 57.4
Hemicellulose 17.8 14.7 7.6 7.2 3.5 3.2 0.8 1.0 0.3 0.6
1.0 Lignin 25.1 28.4 34.3 30.5 37.3 32.1 38.8 36.6 40.8 36.4
I'-.)

Liquid fraction (g/lOO g poplar biomass)


Glucose 1.2 1.6 1.9 2.1 1.4 1.9 1.0 1.1 0.8 0.8
Xylose 1.6 1.6 10.8 8.1 3.5 3.1 0.3 0.3 0.4 0.1
Galactose 0.6 0.6 1.0 0.8 0.4 0.3 0.1 0.1 0.1 0.1
Arabinose 0.8 0.7 0.5 0.2 0.2 0.1 0.1 0.1 0.1 0.1
Mannose 0.6 0.5 1.8 1.8 1.4 1.4 0.3 0.4 0.1 0.2
Acetic acid 0.21 0.31 1.53 1.90 3.33 3.48 3.86 4.15 4.40 4.30
Formic acid 0.13 0.15 0.36 0.34 0.40 0.46 0.50 0.55 0.59 0.61
Levulinic acid 0.00 0.01 0.02 0.01 0.03 0.04 0.03 0.05 0.03 0.06
~
:- 2-Furfural 0.01 0.03 0.46 0.73 1.86 2.24 1.68 2.38 2.44 2.52
c
-Y' 5-HMF 0.01 0.01 0.08 0.17 0.37 0.53 0.50 0.93 1.16 1.29
pH 4.2 4.2 3.9 3.7 3.5 3.5 3.4 3.4 3.5 3.5
c
-
$0
N
C
C
~
)..
"0
"0
~
Q..
Table 2
OJ Chemical Composition of Solid and Liquid Fraction

n
::)-
After Steam Explosion Poplar Pretreatment
<ll
3
<n. 190°C 210°C
~
OJ
::J 4 min 8 min 4 min 8 min
Q..
OJ
0· Particle sizes (mm) 2-5 12-15 2-5 12-15 2-5 12-15 2-5 12-15
(0
n::)-
::J Solid recovery (gfl00 g dry wt) (%) 66 66 65 64 60 62 59 61
0
c2
'<: Solid fraction (%, [w fwD
Ash 0.9 0.8 0.9 0.9 1.2 1.1 1.1 1.1
Cellulose 57.2 57.5 57.2 59.3 57.7 58.6 55.9 59.3
<..0 Hemicellulose 8.2 8.5 7.9 8.2 5.2 4.4 4.3 3.8
w Lignin 33.1 30.2 31.7 31.8 35.3 35.5 38.0 36.3
Liquid fraction (gfl00 g poplar biomass)
Glucose 1.5 1.6 1.4 1.7 1.4 1.8 1.5 1.9
Xylose 8.6 6.7 7.7 7.3 7.1 6.4 5.7 4.8
Galactose 1.1 0.9 1.0 1.0 0.8 0.7 0.8 0.7
Arabinose 0.8 0.5 0.6 0.6 0.5 0.5 0.5 0.4
Mannose 1.2 1.0 1.1 1.1 1.1 1.2 1.1 1.1
Acetic acid 1.05 0.92 1.21 1.32 2.03 2.24 2.49 2.89
Formic acid 0.35 0.33 0.25 0.35 0.43 0.46 0.43 0.53
2-Furfural 0.16 0.15 0.27 0.23 0.49 0.62 0.79 0.94
Cl
5-HMF 0.03 0.03 0.04 0.05 0.08 0.12 0.12 0.22
pH 4.1 4.0 4.0 4.0 3.9 3.8 3.8 3.7
-Y'~
-Cl
,Co

Cl
Cl
""
W
94 Negro et al.

Hemicellulose is clearly the major constituent extracted during pre-


treatments. In liquid hot water, at temperatures above 220°C, hemicellulosic
polymer was practically dissolved from the solid (<1 % hemicellulose remain-
ing in pretreated material). Experiments performed with steam explosion at
mild pretreatment conditions (190°C temperature, 4- and 8-minresidence time)
showed that a significant amount of hemicellulose (>8%) remained unhy-
drolyzed in the solid fraction. When pretreatment temperature was increased
to 210°C, higher hemicellulose solubilization was obtained.
The liquid fraction of liquid hot water and steam explosion pretreated
poplar biomass consisted of a mixture of hydrolyzed sugars and degrada-
tion products (e.g., carboxylic acids and furans). In liquid hot water, total
hydrolyzed monomeric sugars ranged from 1.3 to 16.0 g/100 g of poplar
biomass, carboxylic acids from 0.34 to 5.02 g/100 g of poplar biomass, and
furans from 0.02 to 3.81 g/100 g of poplar biomass, depending on the pre-
treatment conditions (Table 1). The formation of carboxylic acids was de-
tected in the liquid fraction for all pretreatment conditions. Maximum
xylose content in the liquid fraction was obtained at 210°C and small par-
ticle size (10.8 g/lOO g of poplar biomass), corresponding to 62% of the
xylose content in the raw material. At these conditions, liquid chromato-
graphy analysis indicated that >90% of the solubilized hemicellulose ap-
peared as oligosaccharides. At higher temperatures increased degradation
of hemicellulose sugars, regardless of particle size, was obtained. Pretreat-
ment temperatures above 220°C resulted in a complete degradation of
hemicellulose-derived sugars in the filtrate.
In the liquid fraction obtained after steam explosion experiments, the
amount of total hydrolyzed sugars ranged from 8.9 to 13.2 g/100 g of pop-
lar biomass, carboxylic acids from 1.25 to 3.42 g/IOO g of poplar biomass,
and furans from 0.19 to 1.16 g/lOO g of poplar biomass (Table 2). Levulinic
acid was not detected in any steam explosion pretreatment conditions.
Increased hemicellulose degradation at higher temperature and residence
time was found. Particle size affected xylose content in the liquid fraction
obtained after steam explosion, resulting in a slightly increased concentra-
tion at small particle size (p < 0.05).
With liquid hot water, pH reached a final value between 4.2 and 3.4.
A strong correlation between final pretreatment temperature and final
pH was not observed, although aliphatic acid content increased at higher
temperatures. Thus, the constituents extracted from poplar during pre-
treatment apparently buffered the pH at 3.4. In experiments with steam-
exploded biomass, pH values were in the range of 4.1-3.7.
To facilitate a comparison of results, Fig. 1 shows hemicellulose-derived
sugars and glucose recovery in solid and liquid fractions for both pretreat-
ments at the different conditions tested. In liquid hot water, total hemicellu-
lose recovery, calculated as the sum of recovered hemicellulose-derived
sugars in the liquid fraction and hemicellulose content in the solid fraction,
ranged from 88% at 180°C to 1.5% at 240°C (Fig. 1A). Total glucose recoveries
above 95% were obtained at 180, 210 and 220°C.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Poplar Pretreatment 95

A 100~--~~----------------------------------~

80

40 -+--WfIi3---i'&IFI--

2·5 mm 12·15 mm 2-5 mm12·15 mm 2-5 mm12·15 mm 2-5 mm 12·15 mm 2-5 mm 12·15 mm
180·C 210·C 220·C 230·C 240·C
Pretreatment conditions

B Yield
100.-~~~,,--~~--~--------------~-----,

80

60

40

20

o 2·5 mm 12·15 mm 2·5 mm 12·15 mm 2-5 mm 12·15 mm 2-5 mm 12·15 mm


4min. 8min. 4 min. 8min.
190·C 210·C
Pretreatment conditions

Fig.l. Total sugar recovery yield at different pretreatment conditions. (A) Liquid hot
water; (B) steam explosion. Yield is expressed as sugars in the solid or liquid fraction
divided by potential sugars in the raw material. Glucose recovery in solid ( ~ ) and
liquid ( D ) fractions and hemicellulose-derived sugars recovery in solid ( § ) and
( • ) liquid fractions.

Steam explosion pretreatment at 190°C (4 min, 2-5 mm) provided total


hemicellulose recoveries close to 72%, while at 210°C (4 min, 12-15 mm)
only 40% of hemicellulose was recovered (Fig. 1B). In relation to total glu-
cose recovery, at 190°C close to 100% recovery value was obtained. At
210°C more cellulose degradation (5-15%) occurred, and this effect was
smaller at large chip size. A significant interaction effect between tempera-
ture and chip size variables was found on total glucose recovery (p < 0.05).
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
96 Negro et al.

Enzymatic Hydrolysis and SSF


To establish the suitability of the tested hydrothermal pretreatments
to enhance the susceptibility of cellulose from poplar biomass to enzymatic
attack and conversion to ethanol, enzymatic hydrolysis and SSF tests were
performed.
Figure 2 illustrates enzymatic hydrolysis and SSF yields after liquid
hot water and steam explosion pretreatment at different conditions. Enzy-
matic hydrolysis yield is expressed as a percentage of glucose produced in
the hydrolysis divided by potential glucose in the pretreated material. SSF
yields are presented in percentage of theoretical yield. The theoretical SSF
yield is calculated by assuming that all the potential glucose in the pre-
treated material is available for fermentation at a fermentation yield of
0.51 g of ethanol! g of glucose.
In the case of poplar biomass subjected to liquid hot water pretreat-
ment (Fig. 2A), enzymatic hydrolysis yield was markedly dependent on
pretreatment temperature, increasing from low values of 10% at 180°C, to
maximum yields of 68-70% (41-44 g of glucose/100 g of pretreated sub-
strate) at the highest temperatures of 230-240°C. No significant differ-
ences (p > 0.05) in enzymatic hydrolysis yield were found at different
particle sizes.
Pretreatment temperature and residence time showed significant main
effects (p < 0.05) on enzymatic hydrolysis yield in steam-exploded materials,
but no significant differences were found with particle size (Fig. 2B). Steam-
exploded poplar biomass at a mild temperature of 190°C showed the lowest
enzymatic hydrolysis yield (40%, equivalent to 26 g of glucose/100 g of
pretreated material). Higher enzymatic hydrolysis yields of 60% (equivalent
to 40 g of glucose/100 g of pretreated material) were obtained at 210°C.
SSF yields in experiments using liquid hot water pretreated materials
as substrate increased as temperature rose, reaching a maximum value of
60% of theoretical yield, corresponding to a final ethanol concentration of
17 gil at 240°C. SSF yield in steam explosion pretreated materials showed
a profile similar to that of enzymatic hydrolysis yield in relation to the
pretreatment parameters studied. At the lowest temperature of 190°C,
values of about 40% were obtained, while at the highest temperature of
210°C, SSF yield increased up to 60%. Ethanol concentration after 72 h
fermentation was about 20 gil at the highest temperature. SSF yields
obtained with pretreated biomass using a steaming temperature of 210°C
reached a value similar to that obtained for liquid hot water at 240°C, about
60% of theoretical yield.

Discussion
We assessed the effectiveness of liquid hot water and steam explo-
sion pretreatment of poplar at two different particle sizes, under several
temperature and time conditions, by measuring hemicellulose and cellu-
lose recovery, fiber susceptibility to cellulase attack, and SSF performance.
Applied Biochemistry and Biotechnology Vol. 705-108, 2003
Poplar Pretreatment 97

A Yield %
100

80

60

40

20

o ~'----r--~----r---~--'----r--~----r---~~
2·5 mm 12·15 mm 2-5 mm 12·15 mm 2·5 mm 12·15 mm 2-5 mm 12·15 mm 2·5 mm 12·15 mm
180 ·C 210·C 220 ·C 230 ·C 24O ·C
Pretreatment conditions

B Yield %
100

80

60

40

20

0
2-5mm 12'l~mm Hmm 12-15mm Hmm 12·15mm Hmm 12·15mm 2-5mm 12·15mm
4 min. 8ml n . 4mln . 8 min. 4mln .
210·C 220·C
Pretreatment conditions

Fig. 2. Enzymatic hydrolysis yield ( IR ) and SSF yield ( • ) after (A) Liquid hot
wa ter and (B) steam exp losion at different pretrea tmen t condi tions. Hydrolysis yield
is expressed as glucose obtained in the enzymatic hydrolysis/potential glucose. SSF
yield is expressed as percentage of theoretical yield.

In general, at the conditions tested, both hydrothermal pretreatments


produced high hemicellulose solubilization from insoluble fiber. Results
obtained in liquid hot water pretreatment are somewhat lower than those
reported by Van Walsun et al. (9). They reported extraction of 98-100% of
the hemicellulose in other lignocellulosic materials, bagasse and aspen,
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
98 Negro et al.

pretreated at 220°C. However, these results were obtained in a flow-


through reactor, which makes comparison difficult. The fraction of hemi-
cellulose that is solubilized depends on the type of biomass and is
associated with structure and chemical composition. Similar results of
high hemicellulose solubilization were obtained by Weil et a1. (22) when
yellow poplar sawdust was pretreated by pressure cooking in water at
220-260°C.
Although hemicellulose is clearly the major constituent extracted in
both liquid hot water and steam explosion pretreatments, unfortunately,
good hemicellulose solubilization does not always correspond to elevated
hemicellulose-derived sugar recoveries in the liquid fraction. Belkacemi
et a1. (23) proposed that hemicellulose solubilization from biomass occurred
in two parallel reactions, for fast and slow solubilization. The different rate
for solubilization of hemicellulose could explain the high degradation of
soluble hemicellulose-derived sugars, at the same time that a considerable
amount of hemicellulose still remains in the fiber.
Liquid hot water pretreatment at l80 0 e scarcely affected the carbohy-
drate composition of poplar biomass. At 21O oe, a relatively high amount of
hemicellulose was solubilized (76%), of which 77% was recoverable in the
liquid fraction. Experiments carried out at a higher temperature of 220 0 e
produced higher hemicellulose solubilization (90%), of which only 25%
was recovered in the liquid fraction, showing high sugar degradation at
this condition. If we consider that in liquid hot water reactor experiments
there is only a 4-min difference in attaining 210 and 220 o e, the elevated
degradation occurred at 220 0 e can be mostly explained by the temperature
effect and not to heat-up time.
Degradation products quantified in the liquid fraction can only par-
tially explain hemicellulose losses in pretreatment. It is also possible that
hemicellulose and other compounds were lost through volatilization of
degradation products (e.g., furfural) and recondensation reactions (14).
Our finding that more material was lost in experiments run at higher liquid
hot water temperature supports this hypothesis. The overall mass balance
at l80 0 e indicates that <5% of the initial material is unaccounted for in
measured products in solid or liquid fractions, while at 240 0 e close to 25%
of the initial material is not recovered (data not shown). Reflecting the
greater sugar degradation that occurs at elevated temperatures, furfural
formation increased from 0.01 g/lOO g of poplar at l80 0 e to 2.5 g/lOO g of
poplar at 240°C. The amounts of furfural and HMF formed are similar to
those obtained by other researchers (11) investigating liquid hot water pre-
treatment of com fiber.
According to preliminary studies (unpublished data), at 210 0 e (the
most favorable liquid hot water conditions for hemicellulose recovery),
the degradation compounds quantified (1.53 giL of acetic acid, 0.36 giL
of formic acid, 0.46 giL of furfural, and 0.08 giL of HMF) in the liquid
fraction (pH 3.9) do not significantly inhibit growth and fermentation of
K. marxianus. However, the concentration of degradation compounds

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Poplar Pretreatment 99
found in the liquid fraction at liquid hot water temperatures above 210°C
(3.86 giL of acetic acid, 0.50 giL of formic acid, 1.68 giL of furfural, and
0.50 giL of HMF) inhibits K. marxianus growth even in synthetic media.
Severe pretreatment conditions, in which high hemicellulose solubi-
lization is obtained, result in enhanced enzymatic digestibility. Hemicel-
lulose solubilization and enzymatic hydrolysis yield were found to be
positively correlated (p < 0.05) in liquid hot water pretreatment.
Weil et al. (10) reported 77% enzymatic hydrolysis yield using 1%
(w Iv) yellow poplar sawdust pretreated by liquid hot water at 240°C as
substrate. In our run, at the same temperature, and using a substrate
loading of 10% (w Iv) liquid hot water-pretreated poplar, 70% enzymatic
hydrolysis yield was obtained. Considerable degradation of hemicellu-
lose-derived sugars was found at this condition.
SSF yield was also found to depend on pretreatment temperature for
both hydrothermal treatments and was negatively correlated with hemi-
cellulose content in the pretreated solids. The maximum SSF yield of 60%
of theoretical was obtained at a liquid hot water pretreatment tempera-
ture of 240°C, a condition at which total hemicellulose-derived sugar
degradation occurs. For poplar sawdust pretreated by pressure cooking
at 240°C, Weil et al. (10) reported on SSF yield of 55% of theoretical using
similar levels of enzyme loading, a 6% (w Iv) solid concentration, and
Saccharomyces cerevisiae as fermenting yeast. Our results demonstrate that
K. marxianus CECT 10875 is capable of attaining similar SSF yields even
using a higher initial solid concentration of 10% (w Iv).
At the highest temperatures tested (240 and 210°C for liquid hot
water and steam explosion, respectively), higher enzymatic hydrolysis
yield for LWH pretreatment (about 70%) was obtained in comparison
with enzymatic hydrolysis yield in steam explosion pretreatment (60%).
Thus, it was to be expected that better SSF yields in solid pretreated by
liquid hot water would be found. Not only SSF yields are similar in both
cases, but lower ethanol concentration was obtained in liquid hot
water-pretreated solid tests. This indicates that the fermentation step
and not enzymatic hydrolysis in SSF is affected in liquid hot water-
pretreated samples, which is supported by the presence of residual glu-
cose in those SSF broths when the ethanol production comes to an end.
Considering our results, the evaluation of pretreatment effectiveness
should not only take into account the enhancement of cellulose hydrolysis
but the overall process from cellulose to ethanol as well.
The best results were obtained when pretreating poplar biomass by
steam explosion at 210°C and 4 min, taking into account cellulose recovery
above 95%, enzymatic hydrolysis yield of about 60%, SSF yield of 60% of
theoretical, and 41 % xylose recovery in the liquid fraction.
In conclusion, our work shows that water is an effective pretreating
agent that enhances enzymatic hydrolysis and SSF of cellulose to ethanol
of poplar biomass, compared to untreated material. Large (12-15 mm)
particle size can be used for poplar biomass substrate in both pretreatments
Applied Biochemistry and Biotechnology Vol. 105-108,2003
100 Negro et al.

since no significant effect of particle size on enzymatic hydrolysis and SSF


was observed.

References
1. El Bassam, N. (1996), in Renewable Energy: Potential Energy Crops for Europe and the
Mediterranean Region, FAa, Rome, Italy, Rev. Technical Series 46,142-156.
2. Sun, Y. and Cheng, J. (2002), Bioresour. Technol. 83, 1-11.
3. McMillan, J. D. (1994), in Enzymatic Conversion ofBiomass for Fuels Production, Himmel,
M. E., Baker, J. 0., and Overend, R. P., eds., American Chemical Society, Washington,
DC, pp. 292-324.
4. Heitz, M., Capek-Menard, E., Koeberle, P.G., Gagne, J., and Chornet, E. (1991),
Bioresour. Technol. 35,23-32.
5. Mes-Hartree, M. and Saddler, J. N. (1983), Biotechnol. Lett. 5,531-536.
6. Ando, S., Arai, I., Kiyoto, K., and Hanai, S. (1986), J. Ferment. Technol. 64,567-570.
7. Duff, S. J. B. and Murray, W. D. (1996), Bioresour. Technol. 55, 1-33.
8. Wright, J. D. (1998), Chem. Eng. Prog. 84, 62-74.
9. Van Walsum, G. P., Allen, S. G., Spenser, M. J., Laser, M. S., Antal, M. J., and Lynd,
L.R. (1996), Appl. Biochem. Biotechnol. 57-58,157-170.
10. Weil, J., Sarikaya, A, Rau, S. 1., Goetz, J., Ladisch, C. M., Brewer, M., Hendrickson,
R., and Ladisch, M. R. (1997), Appl. Biochem. Biotechnol. 68,21-40.
11. Weil,J. P.,Sarikaya,A, Rau,S. L.,Goetz,J., Ladish,M., Brewer,M., and Hendrickson,
R. (1998), Appl. Biochem. Biotechnol. 73, 1-17.
12. Mok, W. S.-L. and Antal, M. J. (1992), Ind. Eng. Chem. Res. 31, 1157-1161.
13. Laser, M., Schulman D., Allen, S. G., Lichwa J., Antal, M. J., and Lynd, 1. R. (2002),
Bioresour. Technol. 81, 33-44.
14. Allen, S. G., Schulman D., Lichwa, J., and Antal, M.J. (2001), Ind. Eng. Chem. Res. 40,
2934-2941.
15. Ulbricht, R. J., Northum, S.J., and Thomas,J.A (1984),Fund.Appl. Toxicol. 4,843-853.
16. Carrasco, J. E., Martinez, J. M., Negro, M. J., Manero, J., Mazon, P., Saez, F., and
Martin, C. (1989), in Biomass for Energy and Industry, 5th Conference, vol. 2, Grassi, G.,
Gosse, G., and Dos Santos, G., eds., Elsevier, Essex, England, UK, pp. 38-44.
17. Ballesteros, I, Oliva, J. M., Navarro, A A, Gonzalez, A, Carrasco, J., and Ballesteros,
M. (2000), Appl. Biochem. Biotechnol. 84-86,97-110.
18. Ballesteros, I., Oliva, J. M., Negro, M. J., Manzanares, P., and Ballesteros, M. (2002),
Appl. Biochem. Biotechnol. 98-100, 717-732
19. Ballesteros, I., Ballesteros, M., Cabanas, A, Carrasco, J., Martin, c., Negro, M. J., Saez,
F., and Saez, R. (1991), Appl. Biochem. Biotechnol. 28-29,307-315.
20. Ruiz, R. and Ehrman, T. (1996), NREL Chemical Analysis and Testing Laboratory
Analytical Procedure, No. 002., National Renewable Energy Laboratory, Golden, co.
21. Templeton, D. and Ehrman, T. (1995), NREL Chemical Analysis and Testing Labo-
ratory Analytical Procedure, No. 003., National Renewable Energy Laboratory,
Golden,CO.
22. Weil, J., Brewer, M., Hendrickson, R., Sarikaya, A, and Ladish, M. (1998), Appl.
Biochem. Biotechnol. 70-72,99-111.
23. Belkacemi, K.; Abatzoglou, N., Overed, R. P., and Chornet, E. (1991), Ind. Eng. Chem.
Res. 30,2416-2425.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0101/$20.00

Estimation of Temperature Transients


for Biomass Pretreatment
in Tubular Batch Reactors
and Impact on Xylan Hydrolysis Kinetics

SUZANNE L. STUHLER AND CHARLES E. WVMAN*

Thayer School of Engineering, Dartmouth College, 8000 Cummings Hall,


Hanover, NH 03755, E-mail: Charles. Wyman@Oartmouth.edu

Abstract
A combined heat transfer jkinetic model was developed to quantify tem-
perature variations in small tubular batch reactors and estimate the effect
of deviations from isothermal operation on the kinetics of biomass pretreat-
ment. Assuming that heat transfer was dominated by conduction in the
radial direction, a classic parabolic time-dependent partial differential
equation was applied to describe the temperature in the system and
dedimensionalized to provide a single solution for application to all situa-
tions. A dimensionless expression for the reaction kinetics for xylan
hydrolysis was then developed, and a single parameter expressed as the
dimensionless ratio of the first-order rate constant times the tube radius
squared divided by the thermal diffusivity was found to control the reaction
rate. Three different characterizations of the deviation between the concen-
tration profile predicted for isothermal xylan hydrolysis and that based on
the transient temperature were directly related to this dimensionless rate
constant parameter for both catalyzed and uncatalyzed hydrolysis kinetics.
These results were then used to project the relationship between deviations
in yield from isothermal results and the tube radius and reaction time.
Index Entries: Reactor design; heat transfer; kinetics; hydrolysis; pre-
treatment.

Introduction
The use of small-volume tubular batch reactors to study the hydrolysis
kinetics of hemicellulose is common in the literature because of the relative
simplicity and ease of use of this type of reactor system. The reactors can be
easily filled with the desired substrate, sealed, and submerged in a constant-

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 101 Vol. 105-108, 2003


102 Stuhler and Wyman
temperature oil or fluidized sand bath set at the reaction temperature.
Researchers have utilized I-in. 316 SS tubing (1), O.5-in. Hastelloy tubing (2),
glass tube reactors (3,4), and a variety of other similar apparatus to conduct
experimental studies on both acid-catalyzed and uncatalyzed (autohydroly-
sis) hydrolysis. A common feature of these studies is that they are typically
assumed to be carried out under isothermal reaction conditions (5). In other
words, it is assumed that the reactor and its contents instantaneously reach
reaction temperature on immersion in a constant-temperature bath, and the
temperature transients that occur as the reactor is heated from ambient to
reaction temperature are not considered. These temperature transients result
in deviations to the predicted reaction progress using isothermal kinetics (5).
Some researchers have acknowledged these transients and have typically
dealt with them in one of two ways: (1) design a preheating strategy to mini-
mize the temperature transients, or (2) quantify the transients for specific
reaction conditions to obtain a measure of their effect on yields.
Most studies have applied the first method. For example Chen et al. (4)
proposed a two-bath procedure in which the reactors are placed in a bath
set 50°C higher than the reaction temperature for 50 s. Jacobsen (6) also used
a two-bath procedure in which the first bath is heated to 150% of the reac-
tion temperature. Only two published studies have been identified in which
an attempt to quantify the temperature transients is made (5,7). In a study
by Tillman et al. (5), a quantitative guideline is presented to assess the
impact of temperature and chip thickness on total xylose yield for acid
hydrolysis of aspen hemicellulose, and in the work ofJ acobsen and Wyman
(7), a comparison of xylan remaining in 0.5- and 1.0-in.-diameter batch
tubes was presented.
The primary objective of the present study was to develop a compre-
hensive model that can be used with any given set of kinetics to estimate
the deviation from isothermal operation. First a dimensionless heat trans-
fer model was derived for the tubular reactor to describe the temperature
profile as a function of tube radius and time assuming that heat transfer
is dominated by conduction. This model was then coupled to a dimen-
sionless equation to describe xylan hydrolysis, and a dimensionless rate
constant was devised to combine similar systems in a single equation.
From this, a quantitative measure was developed between the controlling
factors for reactor design and the deviation of xylan hydrolysis from iso-
thermal operation.

Development of the Model and Dimensionless Rate Constant


First, the heat transfer portion of the combined model was derived by
applying the unsteady-state equation of energy in cylindrical coordinates
for one-dimensional heat conduction (8):

aT = a. (a 2T + 1 aT) (1)
at ar2 r ar
Applied Biochemistry and Biotechnology Vol. 105-708, 2003
Transients in Hydrolysis Kinetics 103
in which Tis the temperature, t is the time, a is the thermal diffusivity, and
r is the radial position. The thermal diffusivity, a, is given by Eq. 2:

a = -'L (2)
pCp

in which k is the thermal conductivity, p is the density, and Cp is the heat


capacity.
To apply Eq. 1 to a tubular batch reactor system, the reactors were
assumed to be infinitely long cylinders with heat transfer by only conduc-
tion in the radial direction.
One solution of this equation could be applied to multiple applica-
tions by rewriting it in dimensionless form. To this end, three dimension-
less parameters were introduced: f, the dimensionless time; r,
the
dimensionless radius; and T, the dimensionless temperature, defined as
follows (8):
t" = at
(3)
RZ
r "= r (4)
R
* T-T.
T =--' (5)
Tf-T j

in which Ti is the initial reactor temperature, Tfis the desired reaction tem-
perature, and R is the reactor radius. These three equations were substi-
tuted back into Eq. 1 to obtain the dimensionless heat transfer equation:
aT" aZT" 1 aT"
-=--+-- (6)
at" ar"Z r" ar"
This equation was solved numerically in an Excel spreadsheet using
an explicit finite-difference method with 10 radial increments across the
tube radius, as presented in Fig. 1. Figure 1 shows that the tube contents
near the wall heat up very rapidly while the contents near the center take
much longer to reach the desired reaction temperature. Because this equa-
tion is dimensionless, its solution is applicable to any combination of tubu-
lar reactor size (radius), tube contents (thermal diffusivity), initial reactor
temperature, and desired reaction temperature.
A dimensionless approach was also applied to describe xylanhydroly-
sis based on the following first-order model for solids decomposition:
~~ = -kX (7)

in which X is the hemicellulosic xylan in solid form and k is the reaction rate
constant. We then introduced the dimensionless time, f, as previously
defined and the dimensionless xylan remaining in solid form, X:
x" = X (8)
Xo

Applied Biochemistry and Biotechnology Vol. 705-108,2003


104 Stuhler and Wyman

e 1.0 I---=:::::::::~~
=
eG)
a.
O.S
E
!!
II) 0.6
II)
CD
C
~ 0.4
c
CD
E
~ 0.2
~
0 .0 .-..L....._ _ _ _~.__-~---___,_--.,_-
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.S
t' (dimensionless time)

Fig. 1. Dimensionless temperature profile for tubular batch reactors showing tem-
perature vs time for 10 radial intervals.

in which Xo is the initial amount of xylan in the substrate. Now the dimen-
sionless kinetic expression can be written as

(9)

in which 13, the dimensionless first-order rate constant, is

(10)

Equation 9 can then be solved by applying the initial condition that


X = 1.0 when f = 0, to obtain the following analytical expression, which
describes the amount of xylan remaining in solid form as a function of
dimensionless time and 13 only when the temperature is constant:
x' = exp (- j3t') (11)
The dimensionless rate constant, ~, can be written for uncatalyzed
(autohydrolysis) hydrolysis by incorporating the Arrhenius relationship
for the rate constant:
k =k e(-E/RT) (12)
a

in which ko is the preexponential factor, E is the activation energy, and R is


the gas constant. For acid-catalyzed hydrolysis, an additional acid concen-
tration term is added to the Arrhenius expression:
(13)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Transients in Hydrolysis Kinetics 105
in which C is the acid concentration and n is the reaction order. Therefore,
the dimensionless rate constants for uncatalyzed and catalyzed xylan
hydrolysis, respectively, can be defined as follows:
2
R
P uncat
= B..-
a
k e(-EIRn
0
(14)

(15)

First, the isothermal xylan profile (X:o) was calculated as a function of


dimensionless time for a given ~ as:
X:o == f ((,~) in which ~ == f (R, a, ko' E, Tf' C, n) (16)
by applying Eq. 11 for a constant target temperature Tfin conjunction with
either Eq. 14 or 15, depending on whether uncatalyzed or acid-catalyzed
kinetics were being studied. Next, the above kinetic expressions given by
Eqs. 9, 14, and 15 were incorporated into an Excel spreadsheet containing
the transient temperature profiles presented in Fig. 1. The transient tem-
perature xylan profile (X;rans) was calculated by dividing the batch reactor
into 10 rings and calculating the relative area of the rings to determine the
initial amount of xylan in each ring. Equation 11 was then integrated across
each ring using Euler's method to determine the amount of xylan remain-
ing in each ring after each time step. Instead of simply using the desired
reaction temperature in this equation, the average temperature of the inner
and outer radius of each radial increment was taken from the transient
temperature profile and used to calculate the reaction rate constant during
that increment of time. Because the actual dimensional temperature had to
be used in this calculation, the transient xylan profile is dependent on the
initial temperature of the reactor in addition to f and ~:
(17)

The two profiles were plotted and compared to produce a quantitative


measure of deviation between the reaction profiles for isothermal and tran-
sient temperature as discussed in the following section. Figure 2 shows an
example of a comparison of isothermal and transient profiles for a value of
~ = 1.0. As expected, the amount of xylan decreases more slowly for tran-
sient than for isothermal operation because of the delay in increase in tem-
perature from the initial to target value.

Development of Quantitative Measures


of Deviation from Isothermal Operation
To quantitatively describe the difference between the transient and
isothermal xylan profiles, three parameters were investigated: the sum of
squares of the error (SSE)' the % area difference (%AD), and the % instan-
taneous difference (%ID), as follows:

Applied Biochemistry and Biotechnology Vol. 105-108,2003


106 Stuhler and Wyman

0.8

0.6
«
x
0.4

0.2

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
t"

Fig. 2. Comparison of isothermal and transient reaction profiles showing dimen-


sionless xylan remaining in solid form vs dimensionless time.

Sum of Squares of Error


SSE was calculated by summing the squares of the differences between
the predicted amounts of xylan left for transient and the amount predicted
for isothermal operation over a given time span as follows:

(18)
in which
* (*
X trans to + iM *) an d X iso
* (to• + iM .)
are, respectively, the numeric values of the predicted amounts of xylan
remaining for transient and isothermal operation evaluated at a specific
time. Clearly, the magnitude of this error is dependent on the number of
time steps that it is evaluated over, so for the purposes of this article a
constant number of time steps was used.
% Area Difference
%AD was based on the difference between the xylan curves for the
transient and isothermal reactions divided by the integral for isothermal
behavior to give a relative deviation:

* • . .)
(!c t. Xtransdt _ !ct' Xisodt
%AD = 100 0 0
rt' • •
Jo Xisodt
(19)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Transients in Hydrolysis Kinetics 107

Table 1
Experimental Conditions and Parameters for Kinetic Models
Garrote et al. (9) Chen et al. (4)

Pretreatment Autohydrolysis Dilute acid


Substrate Corncobs Corncob / stover mixture
Temperature range (0C) 145-190 120-150
Acid conc. range (wt%) 0.11-1.9
ko 1.46 X 1012 S-l 3.33 x 108 S-l wt%-l
E (kJ/mol) 130 86.2
n 1.0

% Instantaneous Difference
%ID was derived by subtracting the predicted amount of xylan left for
isothermal operation from the predicted value for transient operation at a
specified dimensionless time and dividing by the predicted quantity of
xylan for isothermal reaction according to the following expression:

(20)

This parameter is different from the other two measures of deviation


because it only depends on the instantaneous difference in xylan remaining
and is therefore independent of the time history of the profile.

Results
To determine whether the dimensionless rate constant,~, can be cor-
related with a quantitative measure of deviation, the model was applied
with two published kinetic models. Table 1 presents the kinetic parameters
and experimental conditions that were used for the two sets.
The Excel model was run multiple times for both kinetic models by
varying the reaction temperature, acid concentration (for the acid-cata-
lyzed model), and initial temperature (20 or 100°C) within the reported
range of experimental conditions. The value of ~ along with the three
quantitative measures of deviation were calculated for each trial. Figures
3 and 4 show the deviation parameters SSE and %AD defined at 100%
yield (Le., f at which X:o and X;rans are equal to zero) plotted vs ~ for both
catalyzed and uncatalyzed models for two different initial temperatures:
20 and 100°C. The scales for these figures are different for catalyzed and
uncatalyzed models because the ~ values calculated within the experi-
mental ranges in Table 1 are higher for the uncatalyzed model. From Figs.
3 and 4, one can see that for small values of ~, there is a positive linear
relationship between both deviation quantities and ~. For larger values of

Applied Biochemistry and Biotechnology Vol. 105-108,2003


108 Stuhler and Wyman
A 70

60
T =20·C
50 ~' 1

/
V+
40
w
II)
II)
./
V Vi) = 100·C
V ""
30
V
20 / ./
V"
10 ~ ~
0 ~
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Beta

B 40

35

30 ......-- ~
T
..),.
=20·C"

L+
25
w
~ 20
V

---
15 /
V
---
10 ~Ti=100·C

5 /
o ~~
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Beta

Fig. 3. SSE vs dimensionless rate constant, ~, for (A) uncatalyzed and (B) acid-
catalyzed kinetic models and initial temperatures of 20 and 100°C.

~, this relationship becomes less and less linear. The linear relationship
between %AD and ~ (Fig. 4) holds over a wider range of ~ than the linear
relationship between the SSE and ~ (Fig. 3). For both Figs. 3 and 4, a
comparison of uncatalyzed plots (Figs. 3A and 4A) and catalyzed plots
(Figs. 3B and 4B) shows that there is greater scatter for the catalyzed plots.
This is owing to the additional parameter of acid-concentration that was
included and varied in the acid-catalyzed model to obtain the data points.
The regression lines plotted in Fig. 4 for Ti = 20°C are very similar for
the uncatalyzed and acid-catalyzed models and are given by, respectively:
%ADuncat = 37.5 x ~ (21)

%AD cat = 31.4 x ~ (22)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Transients in Hydrolysis Kinetics 109

A 180
160 /'
V"
140 ./
120 T=20·C "" /"
Q 100 .Lt~
~ 80 f Tj' =100·C

60 / /'"
40 ~/
20 !~ ~
0 ~
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Seta

B 100
90
80 ./
70 T=20·C"
'~
80
Q
50 L
~ ./ TI=100·~"""'-
40
V ........ ~
30
/ ,........ rr
20
10 .J" ,....- ........... fo"""

0 ~
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Seta

Fig. 4. %AD vs dimensionless rate constant, 13, for (A) uncatalyzed and (B) acid-
catalyzed kinetic models and initial temperatures of 20 and 100°C.

These two equations can then be used to produce a set of design curves
for each kinetic model. For example, for uncatalyzed kinetics, an expres-
sion for the reaction temperature as a function of tube radius can be written
at any %AD if the thermal diffusivity and kinetic constants are held con-
stant. Figure 5 shows the design curves at 5,10,20,50, and 100% error for
both Ti = 20°C and Ti = 100°C assuming Garrote kinetics as presented in
Table 1 and a = 0.0016 cm2 / s.
A similar plot can be prepared for the acid-catalyzed kinetics using
Eq. 22; however, the initial temperature and the acid concentration must
be specified for each design curve. Figure 6 shows the design curves at 5,
10,20,50, and 100% area difference for an initial temperature of 20°C and
an acid concentration of 0.5 wt%. To demonstrate the effect of varying the

Applied Biochemistry and Biotechnology Vol. 705-108,2003


110 Stuhler and Wyman

260 TT-------------------------------------,

240
0-
~ 220
E
::I

~D- 200
E
{!!. 180
c:
o
tiIII 160
GI
ex:

14° 1~~~~~~~~~~::::~~~~~~~~
120
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Tube Inner Radius (In)

Fig. 5. %AD as function of tube inner radius and reaction temperature for
uncatalyzed model and initial temperatures of 20°C (thick solid lines) and 100°C (thin
solid lines).

acid concentration and initial temperature on the design curve for acid-
catalyzed kinetics for an error of 5%, the plot in Fig. 7 was generated.
Instead of calculating the %AD over the entire time profile (until X
-+ 0) as presented in Figs. 5-7, a different plot was generated by calculat-
ing the %AD at a specified xylan yield (defined as 100% - X) and plotting
vs ~. This plot for uncatalyzed kinetics and two different initial tempera-
tures is presented in Fig. 8.
A third type of plot can be prepared that only considers the error
between the isothermal and transient profiles at an instantaneous time by
using the %ID and plotting the same parameters shown in Fig. 8. Figure 9
shows this plot for the uncatalyzed model.

Discussion
The dimensionless rate constant that was developed in the present
work is directly related to the tube radius squared and the kinetic rate
constant assuming first-order kinetics for xylan hydrolysis and inversely
related to the thermal diffusivity of the contents of the tube. Therefore, it
can be viewed as a ratio of the reaction rate (k) to the rate of heat conduction
(R2 / a for a tube), and the larger the value of ~, the larger the expected
deviation from isothermal operation. This dimensionless group was used
to determine the relative effect of variation of the variables that comprise
it (radius, kinetic constants, reaction temperature, and thermal diffusivity
of tube contents). However, to establish the validity of this dimensionless

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Transients in Hydrolysis Kinetics 111

270

250

230 \
\
~210 ,.\
! 1\
:J 190
f H\
"-.....
8. 170 ,\ ..........
E

-
\',,,- .......
~ 150
........

-
" .....
c: ..... --....
'"" "-
-- ---
..........
.~

~ 130
:--
"- ...........
--
u ............
I'll
~ 110 ............. ~
f- -
20
- 5001. ~ 100% ,
90 - - 5%-:;
10% ==
70

50
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Tube Inner Radius (In)

Fig. 6. %AD as function of tube inner radius and reaction temperature for catalyzed
model, initial temperature of 20°C, and 0.5 wt% acid concentration.

,
,
190

_ 170
U
!...
f 150

.,
:J
f \ ' l\
8. 130 \ I\.'\
E \
"
"' "'"-
o
~
c: 110
[\. i'....
r--..... '-...
-
o ~

&
~
90
.......
......
r--..... r-- __ __
5%. T =100', C::0.5

......-
.............

70 5%. ~o·, C=2.0 r-- .... 5%, T,-20·. C-O.S


I I I I I I
50 I
0.00 0.25 O,SO 0.75 1.00 1.25 1.50 1.75 2.00
Tube Inner Radius (in)

Fig. 7. %AD as function of tube inner radius and reaction temperature for catalyzed
model, initial temperature of 20 and 100°C, and 0.5 and 2.0 wt% acid concentration
with each line representing the limit for an error of 5%.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


112 Stuhler and Wyman
120~----~--~----------------------,

100

80

40

20

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
B

Fig. 8. Percentage xylan yield as function of %AD and ~ for uncatalyzed model and
initial temperatures of 20°C (thick solid lines) and 100°C (thin solid lines).

700
650
600

8 550
!500
~ 450
Q400
eo

1 350
.&300
j 250
.I 200
'It 150
100
50
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
B

Fig. 9. Percentage xylan yield as a function of %ID and ~ for uncatalyzed model and
initial temperature of 20°C.

constant, a relationship between ~ and quantitative measures of deviation


must be shown. The Excel model and Figs. 3-9 described in the previous
section successfully established that three quantitative parameters of
deviation from isothermal operation could be related to the dimensionless
rate constant, ~,for both acid-catalyzed and uncatalyzed kinetic models. A
quantitative indicator of the difference between the isothermal and tran-
sient xylan profiles generated for specific kinetic parameters can be directly

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Transients in Hydrolysis Kinetics 113

calculated by using plots such as in Figs. 3 and 4. For example, to calculate


the deviation for 1/2-in. OD tubes and 0.5 wt% acid, heated from room
temperature to a reaction temperature of 160°C, ~ would first be calculated
using Eq. 15. After inserting the terms in appropriate units, the resulting
value of ~ is 1.24, and, therefore, from Figs. 3B and 4B, the SSE and %AD can
be found to be approximately 22 and 40%, respectively. Figures 5-7 incor-
porate the linear relationships from Fig. 4 to eliminate the intermediate step
of calculating~. For this example, using Fig. 6 with a 1 /2-in. OD tube (0.21-
in. inner radius) and 160°C gives an area difference of <50%.
From the two methods that were used to calculate a quantitative
deviation based on the time history of the xylan profile, the %AD is the most
useful because it gives a relative area so it does have some physical mean-
ing and has a linear relationship with ~ over a much larger range of values.
The %AD parameter also provides a basis for comparison of different
kinetic studies that have been previously reported in the literature and for
determining when assumptions of isothermal operation appear valid. The
third quantitative parameter, %ID, is useful as a means to estimate the error
in any data point.
It is important to note that this analysis is inherently dependent on
assumptions in both the heat transfer and kinetic model that may not always
be valid. For example, the heat transfer model assumes that heat transfer in
the tubular batch reactors is controlled by conduction and that convection
forces are not important. Although experimental measurements of centerline
temperatures in tubular reactors loaded with low (5%) to high (100%) solids
concentrations follow the model prediction, there may be situations in which
convection forces may impact and speed up heat transfer in tubular batch
reactors. Additionally, the kinetic models are based on first-order kinetics
although many researchers have found biphasic kinetics to best fit their data.
For these reasons, the results of this analysis should be viewed in relative
rather than absolute terms and utilized primarily as a tool to guide tubular
reactor analysis and design and interpret the accuracy of the results.

Conclusions
The results of this analysis indicate that it is very important to validate
the assumption of isothermal operation for both acid-catalyzed and
uncatalyzed hydrolysis in batch reactor tubes. A dimensionless rate con-
stant consisting of the ratio of the reaction rate to the heat conduction rate
was shown to be useful as a tool to quantitatively determine the range of
experimental conditions in which the isothermal assumption is reasonable
and those in which this assumption is questionable. Increasing factors that
speed up the reaction rate, such as reaction temperature and acid concen-
tration, were shown to increase the dimensionless rate constant and lead to
larger errors, while increasing factors that speed up the heat conduction
rate, such as thermal diffusivity or smaller reactor tubes, were shown to
decrease the dimensionless rate constant and the resulting error.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


114 Stuhler and Wyman
An Excel spreadsheet model was created to calculate the traditional
isothermal xylan profile as well as a transient xylan profile based on the
transient temperature profile in the reactor that is expected if conduction is
controlling. Three quantitative measures were developed to describe the
differences between these profiles and were shown to be directly related to
the dimensionless rate constant. Through the use of reactor design curves, it
was demonstrated that experimental studies that utilize tubular reactors
larger than 1/2-in. are dramatically affected by temperature transients in all
but the slowest reaction rate conditions (e.g., low temperatures without
addition of acid). Furthermore, our results indicate that if the model assump-
tions are valid, deviation from isothermal operation is expected to be signifi-
cant in many published experimental studies.
Future work on this project will include the utilization of a wider
variety of kinetic data sets, as well as our own laboratory data, to determine
whether there is a universal relationship between the dimensionless con-
stant and the deviation from isothermal operation. It will also be important
to determine how temperature transients affect further reaction products
such as oligomers, monomers, and degradation products as well as to con-
duct experimental work to verify model predictions of the effects of tem-
perature transients.

Acknowledgments
This research was made possible through the support of the USDA
National Research Initiative Competitive Grants Program (contract 2001-
35504-10041) and the Thayer School of Engineering at Dartmouth College.

References
1. Montane, D., Salvado, J., Farriol, X., and Chornet, E. (1993), Biomass Bioeng. 4-6,
427-437.
2. Torget, R W., Kim, J. 5., and Lee, Y. Y. (2000), Ind. Eng. Chem. Res. 39,2817-2825.
3. Baugh, K. D. and McCarty, P. L. (1988), Biotechnol. Bioeng. 31,50-61.
4. Chen, R, Lee, Y. Y., and Torget, R (1996), Appl. Biochem. Biotechnol. 57/58,133-146.
5. Tillman, L. M., Abaseed, A. E., Lee, Y. Y., and Torget, R (1989), Appl. Biochem.
Biotechnol. 20/21, 107-117.
6. Jacobsen, S.E. (2000), MS thesis, Thayer School of Engineering, Dartmouth College,
Hanover, NH.
7. Jacobsen, S. E. and Wyman, C. E. (2001), Appl. Biochem. Biotechnol. 91-93,377-386.
8. Bird, R. B., Stewart, W. E., and Lightfoot, E. N. (1960), Transport Phenomena, John
Wiley, New York, NY.
9. Garrote, G., Dominguez, H., and Parajo, J. C. (2001), Process Biochem. 36,571-578.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0115/$20.00

Effect of Sulfuric and Phosphoric Acid


Pretreatments on Enzymatic Hydrolysis
of Corn Stover

BVUNG-HWAN UM, M. NAZMUL KARIM, * AND LINDA L. HENK


Department of Chemical Engineering,
Colorado State University,
Fort Collins, CO 80523,
E-mail: karim @engr.colostate.edu

Abstract
The pretreatment of com stover with H 2S04 and H 3P04 was investigated.
Pretreatments were carried out from 30 to 120 min in a batch reactor at 121°C,
with acid concentrations ranging from 0 to 2% (w Iv) at a solid concentration
of 5% (w I v). Pretreated com stover was washed with distilled water until the
filtrate was adjusted to pH 7.0, followed by surfactant swelling of the cellu-
losic fraction in a 0-10% (w Iv) solution of Tween-80 at room temperature for
12 h. The dilute acid treatment proved to be a very effective method in terms
of hemicellulose recovery and cellulose digestibility. Hemicellulose recov-
ery was 62-90%, and enzymatic digestibility of the cellulose that remained
in the solid was >80% with 2% (w Iv) acid. In all cases studied, the perfor-
mance of H 2S04 pretreatment (hemicellulose recovery and cellulose digest-
ibility) was significantly better than obtained with H 3P04 • Enzymatic
hydrolysis was more effective using surfactant than without it, producing
10-20% more sugar. Furthermore, digestibility was investigated as a func-
tion of hemicellulose removaL It was found that digestibility was more di-
rectly related to hemicellulose removal than to delignification.

Index Entries: Com stover; enzymatic hydrolysis; H 3P04; pretreatment;


H 2S04 ,

Introduction
The United States and other industrialized countries of the world are
dependent on imported oil. Oil imports continue to increase, threatening
the strategic security of these countries (1). For instance, according to a June
23,2001 article in Time magazine,the United States imports almost 60% of

"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 115 Vol. 105-108,2003


116 Urn et al.
its current oil supply. Furthermore, the transportation sector in the United
States is particularly dependent on oil, with approx 97% of transportation
energy being derived from petroleum. Few substitutes exist for petroleum
for transportation usage (2).
One of the more promising alternative fuels to replace gasoline is
ethanol. To have a significant impact on our current oil consumption, etha-
nol must be both inexpensive and plentiful (3). Lignocellulosic biomass
such as agricultural residues, wood, and crops are abundant renewable
materials for the production of sugars for subsequent fermentation. Of the
agricultural residues, corn stover is the most abundant, with annual US
production rates of 150,000,000 tlyr (4).
The polysaccharide fraction of agricultural residues can be hydro-
lyzed in reaction media using acids or enzymes as catalysts (5). Complete
hydrolysis of cellulose yields the easily fermentable glucose, allowing
biomass to be a potential renewable energy source (4). The enzymatic
approach to hydrolyzing cellulose to glucose is receiving attention
because enzymes can achieve high yields, since they do not catalyze glu-
cose degradation reactions common in the acid process. However, the
cellulose must be accessible to the enzyme. The accessibility depends on
the reagent and the severity of the pretreatment process. Therefore, effec-
tive pretreatment is an essential prerequisite to improve the rate and
yields of saccharification.
One potential method of converting corn stover to ethanol involves
removing the hemicellulose sugars by an acid process. Then, after washing
the sugars from the solids, the solids are subjected to enzymatic hydrolysis.
Previous results (6) have shown that acid is an effective pretreatment re-
agent for subsequent enzymatic hydrolysis of hardwood. However, high
enzyme loadings are required for high conversion. Hence, the use of en-
zyme should be as low as possible. Toward this aim, it has been reported
that the addition of nonionic surfactants, especially Tween-80, improved
cellulase activity and preserved enzyme for recycling (7, 8).
Knappert et al. (9) pretreated corn stover with acid concentrations
from 0 to 1.2% for 0.22 min at 18D-220°C using a cellulase loading of 40
filter paper units (FPU) I g of substrate at a solids concentration of 2.5%.
Good cellulose conversions (7D-100%) were obtained after 2 d of enzymatic
hydrolysis. Schell et al. (4) pretreated corn stover with a 1.0 wt% solution
of H 2S04 and 14D-180°C steam for 5-20 min using a cellulase loading of
40 FPU I g of substrate at a 2% solids concentration. After 5 d of enzymatic
hydrolysis, the cellulose digestibility was 78%.
Thus, the primary objectives of the present study were to compare
the compositions of hydrolysate liquids and solids after mild pretreat-
ment with H 2S04 and H 3P04s, and to compare the enzymatic digestibility
of the H 2S04 and H 3P04 pretreated solids. In these experiments, the tem-
perature was held constant, and residence time and acid concentrations
were varied. The sugar yields after pretreatment and after enzymatic
hydrolysis were investigated.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


H 2S04 and H 3 P04 in Hydrolysis of Corn Stover 117

Materials and Methods


Materials
Chips of com stover (the residue remaining in the field after com
kernels are harvested from the cob), supplied by Agricultural Research
Development and Education Center (Colorado State University, Fort
Collins, CO), were ground to an average size of 20-60 mesh (0.25-0.84 mm)
using a laboratory knife mill. The milled com stover was used directly in
chemical pretreatment studies. Untreated, milled com stover contained
40.8% glucan and 25.8% xylan. Solid com stover was analyzed for moisture,
sugars, Klason lignin, and ash by National Renewable Energy Laboratory
(NREL) standard procedures (Analytical Procedures #001-005) (10).
Pretreatment
The com stover was prepared for acid pretreatment by presoaking
the particles at a 5% (w Iv) solids concentration at room temperature in a
0-2% (w / v) solution of sulfuric or H 3PO4S, overnight. The presoaked slurry
was then pretreated at 121°C in an autoclave reactor with residence times
ranging from 30 to 120 min. At the end of the pretreatment, the contents
were filtered and the solids washed with distilled water until the filtrate
pH reached 7.0. This was followed by surfactant swelling of the cellulosic
fraction in a 0-10% (w / v) solution of Tween-80 (polyoxyethylene sorbitan
monooleate, CAS# 9005-64-5) at room temperature for 12 h. The swollen
cellulosic fraction was again filtered and washed to neutral pH with dis-
tilled water. Wet samples were stored in plastic Petri dishes at-20°C prior
to enzymatic hydrolysis. Effects on digestibility owing to freezing were
not studied.
Enzyme and Enzymatic Hydrolysis
Enzymatic digestibility measurements were performed on several acid-
treated and control samples (untreated) using NREL standard procedure
(Standard Procedure #009) (10). Commercially produced cellulase and
~-glucosidase (Novo Nordisk, Bagvard, Denmark) were used. Celluclast (80
FPU/mL, 80 mg of protein/mL) and Novozym 188 (792 cellobiase units
(CBU) / mL, 73 mg of protein/ mL) were used for cellulose hydrolysis with
a volume ratio of 5 FPU of celluclast/CBU of Novozym to alleviate end-
product inhibition by cellobiose. The amount of washed solids required to
give 0.2 g of cellulose in 20 mL was added to a lS0-mL flask. The buffer for
the digestion was 0.05 M citrate (pH 4.8), containing 800 llg of tetracycline.
The cellulase enzyme loading was adjusted to 40 FPU / g biomass of cellu-
lose in the flask. The contents of the flask were preheated to 50°C before the
enzyme was added. Then, the flask was placed in a shaker incubator oper-
ating at 50°C and 150 rpm. Using the same method, untreated substrate
control was placed in the incubator. Samples were taken every 24 h of the
hydrolysis and subjected to heat denaturation (boiling water for 15 min) to
inactivate and precipitate the enzyme. These samples were analyzed for

Applied Biochemistry and Biotechnology Vol. 105-108,2003


118 Urn et a/.
Table 1
Composition of Untreated and Pretreated Corn Stover a
Acid Cone. Solid remaining Glucan Xmg Klason lignin
(w/v) (%) (%)b (%)b and ash (%)
Untreated 100.0 40.4 25.8 18.7
No acid (water) 76.3 46.8 26.6 23.1
0.5% H 2S04 57.0 61.1 8.4 29.5
1.0% H 2SO4 54.9 60.3 6.6 29.7
2.0% H 2S04 51.8 61.9 NO 31.1
0.5%H3P04 68.7 51.8 23.3 24.9
1.0% H 3P04 63.9 53.4 17.5 26.0
2.0%H3P04 60.8 55.4 14.3 26.9
apretreatment conditions were as follows: presoaking, 12 h; temperature, 121°C; sul-
furic and phosphoric acid concentrations (w Iv), 0-2%; reaction time, 120 min; solid
content (w Iv), 5%.
bAll sugar contents are based on the original oven-dried untreated biomass and expressed
as glucan, xylan, mannan, and galactan equivalents.
CND=not determined.

glucose content and digestibility using a high-performance liquid chroma-


tography (HPLC) and YSI analyzer.
Analytical Methods
All experimental analyses were done in triplicate following NREL stan-
dard procedures (10). Carbohydrate products were measured by HPLC
using a Bio-Rad Aminex HPX-87H column (conditions: 0.6 mL/min, 65°C,
and 0.005 M H 2S04) and the YSI model 2300 glucose analyzer (YSI, Yellow
Springs, OH). Since the HPX-87H column does not resolve xylose, mannose,
and galactose, the three components were presented as xylose + mannose
+ galactose (Xmg). The quantity of Xmg was calculated on a xylose basis,
because xylose consists of more than 90% of these three components (11).
Results and Discussion
Pretreated Corn Stover Composition
The results of analysis ofthe pretreated corn stover are shown in Tables
1-3. Table 1 provides compositions of treated and untreated corn stover.
The percentages of glucan (approx 61 %), and lignin and ash (approx 30%),
did not change with the increase in H 2S04 concentration. However, the
percentage of Xmg decreased with increasing H 2S04 concentration. With
H 3P04 pretreatment, the percentage of glucan did not remain constant as
the concentration of the acid was increased; it increased with the acid con-
centration. However, there was still a significant amount of Xmg (14.3%)
left with 2% (w Iv) H 3P04 • With both pretreatment methods, the mass frac-
tion of corn stover was significantly reduced.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


H 2S04 and H 3 P04 in Hydrolysis of Corn Stover 119

Table 2
Sugar Recoveries in Pretreated Corn Stover Liquors·
Acid conc. (w Iv) Gluca (%)bn Xmg (%)b

Untreated
No Acid (water) 3.2 12.4
0.5% H 2SO4 6.0 71.3
1.0% H 2S04 6.4 74.8
2.0% H 2S04 6.2 89.5
0.5% H 3P04 4.5 30.2
1.O%H3P04 5.0 48.8
2.0%H3P04 5.7 58.1
"Enzymatic hydrolysis conditions were as follows: 1% (w Iv)
solid concentration; 40 FPU of cellulase/ g of cellulose for 96 h
at SO°c.
bAll sugar contents are based on the original oven-dried
untreated biomass and expressed as glucan, xylan, mannan,
and galactan equivalents.

Table 3
Cellulose Digestibility of Pretreated Corn Stover Solids
Acid conc. (w Iv) Digestibility at 72 h (%)
Untreated 10.8
No Acid (water) 24.8
0.5% H 2S04 55.4
1.0% H 2S04 58.7
2.0% H 2S04 75.6
0.5% H 3P04 42.1
1.0% H 3P04 45.8
2.0%H3P04 56.0

Table 2 provides sugar recoveries from the pretreated liquors. The


percentages of glucan in the liquor using both the pretreatment methods
remained approx the same (approx 5%). However, Xmg recovery was sig-
nificantly higher (approx 90% using 2% H 2S04) for H2S04 pretreatment
thanH3P04 pretreatment (approx58% using 2% H 3P04). This indicates that
at mild concentrations, sugar yield is higher for H 2S04 pretreatment.
Table 3 represents the cellulose digestibility (measured at 72 h) of the
pretreated com stover solids. Again, the digestibility increased with higher
acid concentrations, and the H 2S04 pretreatment method showed higher
digestibility compared with the H 3P04 pretreatment method.
Increasing the severity of the pretreatment conditions, apparently
because a greater fraction of the hemicellulose sheath surrounding cellu-
lose is dissolved, enhances the enzymatic digestibility of the pretreated
substrates. This allows the enzyme to have greater access to the cellulose.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
120 Urn et al.
In general, the more severe pretreatment conditions lead to more disso-
lution of the xylan, and more production of the toxic material for fermen-
tation such as acetic acid (12). At a temperature of 121°C, there was a
decrease in xylan content in solids with increasing acid concentration. In
2% (w Iv) H 2S04 for 120 min, little xylan was left after the pretreatment.
H 2S04 was more effective at dissolving the hemicellulose fraction than
H 3P04 • On the other hand, the more severe conditions may result in the
conversion of more of the cellulose to glucose and subsequent conversion
to 5-hydroxymethylfurfural, which is an inhibitory component for subse-
quent fermentation (13). The desired dilute-acid pretreatment conditions
are those that produce the highest ethanol yields in a subsequent fermen-
tation process that is influenced by the enzymatic hydrolysis of the pre-
treated substrate, xylose yield, and presence of toxic byproducts.

H 2 S04 Pretreatment
Moderate temperature was investigated at low acid concentrations,
which would result in savings in catalyst and acid neutralization costs, as
well as material costs of constructing the reactor. This method would also
reduce the problems and costs associated with handling the gypsum formed
during the neutralization (14).
H 2S04 effectively solubilized the hemicellulosic portion of the com
stover and increased the digestibility of the cellulose that remained in the
solid residues. As shown in Table I, the relative percentage of glucan was
near 62%, and the xylose remaining was approx 1%. Thus, about 90% of the
xylose was recovered with 2 % (w Iv) H 2S04 treated for 120 min (Table 2).
When the H 2S04 concentration and residence time were increased from 0.5
to 2% (w Iv) and from 30 to 120 min respectively, hemicellulose recovery
(=[hemicellulose amount recovered in liquid phase] I [initial hemicellulose
content]) was increased by 30%. However, delignification was only 22%
with 2% (w Iv) acid concentration and a residence time of 120 min.

H 3 P04 Pretreatment
Compared with H 2S04 treatment, H 3P04 treatment had considerably
lower hemicellulose degradation. Compositions of hydrolysate liquors from
various pretreatments are given in Table 2. In addition, both acid pretreat-
ments had approx the same effect on the lignin content of the com stover. At
121°C and 120 min, the hemicellulose recovery increased from 30 to 48%, and
finally to 58% at 0.5, I, and 2% acid concentrations, respectively.
The difference of both acid-treated com stovers seems to be as much
a function of reaction time as acid concentration. At moderate tempera-
tures, H 3P04 pretreatment resulted in a lower digestibility and hemicellu-
lose conversion (20 and 30%, respectively), which suggests that H 3P04 as a
reagent for pretreatment is not very effective. Under relatively low tem-
peratures (121°C) and under similar conditions, the results showed that
H 2S04 treatment is a better method compared with H 3P04 treatment.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


H2S04 and H 3 P04 in Hydrolysis of Corn Stover 121
80~--------------------------------------------,

70

60
~
~
i:c 50
~
is 40
Ql

Ql
en
0
:2 30
Qi
()
~
0

20

10

0
0 25 50 75 100
Incubation Time (Hours)

-+- Untreated -II- No Acid (water) -I:r-1 % Sulfuric Acid -9-1 % Phosphoric Acid

Fig. 1. Cellulose digestibility from enzymatic hydrolysis of corn stover pretreated


for 120 min at 1% (w Iv) acid concentration.

Digestibility as Function of Time


The digestibility shown in Table 3 is the percentage of released glucose
during hydrolysis at 72 h. However, to investigate the overall trend in
enzymatic digestibility, samples were taken every 24 h for 96 h. Figures 1
and 2, respectively, show the cellulose digestibility of 1 and 2% (w Iv) acid-
treated corn stover at the pretreatment reaction time of 120 min. As the
pretreatment reaction time was increased from 30 to 120 min, the cellulose
digestibility increased up to 62 and 18% in 2% (w Iv) acid concentration for
H 2S04 and H 3P04, respectively (data not shown).
Increasing acid concentration from 0.5 to 2% (w Iv) increased the
degree of digestibility from 56 to 80%, and from 42 to 60% for H 2S04 and
H 3P04 , respectively, after pretreatment at 121°C for 120 min. Figures 1 and
2 show the effects of acid concentration on digestibility. The digestibility
increased as the acid concentration and residence time were increased. The
digestibilities from various H 2S04 treatments were higher than those of the
H 3P04 treated samples. When the 24-h digestibility of H 2S04 treatments

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


122 Urn et al.
90

80

70

160
~
;g
III
50
(])
Cl
i:5
3l 40
0
2
Qj
() 30
0~

20

10

0
0 25 50 75 100
Incubation Time (Hours)

-+- Untreated ....-- No Acid (water) -ir- 2 % Sulfuric Acid -e- 2 % Phosphoric Acid
Fig. 2. Cellulose digestibility from enzymatic hydrolysis of corn stover pretreated
for 120 min at 2% (w Iv) acid concentration.

are compared with those of the H 3P04 treatment, it is found that initial
digestibility was affected somewhat by the extent of hemicellulose removal,
resulting in increased initial rate. From Fig. 2 it is also apparent that the
percentage of digestibility (at 96 h, and 2% [w Iv] concentration) is approx
25% higher for the H 2S04 than for the H 3P04 pretreatment process.
Effect of Surfactant on Enzymatic Hydrolysis
Figure 3 shows the effect of different concentrations of Tween-80 on
the enzymatic hydrolysis yield of 2% (w Iv) acid-pretreated corn stover.
Tween-80 clearly aided the enzymatic hydrolysis of the pretreated corn
stover; the average rate and extent of conversion with Tween-80-treated
samples were higher than for Tween-free samples. At the loading level of
0-10% (w Iv) solution of Tween-80, the total sugar yield increased up to
a maximum of 15% when compared to results obtained in the absence of
the additive in the 96-h period. It was observed that the addition of Tween-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


H2S04 and H3 P04 in Hydrolysis of Corn Stover 123
100~------------------------------------~

00+------------------------------------------;
80

I>.
:t:
70

60
£u;
CD
01
is
50
CD
Ul

:2 40
0

CD
()
30
*
20

10

0
0 25 50 75 100
Incubation lime (Hours)

-+- Untreated -11-10 % Tween 80


~ 2.5 % Tween 80 -e- Tween Free (Acid Treated)
Fig. 3. Cellulose digestibility from enzymatic hydrolysis of corn stover pretreated
for 120 min at 2% (w Iv) H 2S04 followed by surfactant swelling at room temperature.

80 increased the sugar yield proportional to the concentration of the sur-


factant, and it was found that Tween-80 promoted the availability of the
reaction that increased the hydrolysis rate, as was reported by Kaar and
Holtzapple (7) and Castanon and Wilke (8).
Digestibility as Function of Hemicellulose Removal
Figure 4 shows 96-h enzymatic digestibility as the function of the
amount of hemicellulose removed by dilute-acid pretreatment of the corn
stover. The data were obtained after treating corn stover with H 2S04 or
H 3P04 withoutthe addition of Tween-80. Here, the points are experimental
raw data and the solid lines represent nonlinear regression model predic-
tion using ExceL
The digestibility increased approximately linearly with increasing
removal of xylan. In the case of H 2S04, hemicellulose removal was about 1.5
times higher than that in the H 3P04 treatment.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


124 Urn et al.
90

80
A

./.4.
70

III

~
:is
~ 50
Q)
Cl

./ // •
is
5l 40
~/
o
"S
CD /./
o 30

/.Y
~
o

20

. /'~
'
10

o
o 25 50 75 100
% Xylan Removed (w/w)

• 30 Minutes III 60 Minutes A 120 Minutes

Fig. 4. Cellulose digestibility as a function of percentage xylan removed. Pretreat-


ment conditions were 0.5-2% H 2S04 for 30-120 min at 121°C. Enzyme hydrolysis
conditions were 1% (w Iv), cellulose loading and 40 FPU I g of cellulose for 96 h at SO°c.

Several studies have reported that hemicellulose (7) or lignin (15,16)


hinders enzyme adsorption on cellulose. The results of our study indicate
that the digestibility may be directly related to hemicellulose removal.

Conclusion
Recovery of xylose using H 2S04 pretreatment of corn stover was 10-
30% greater in comparison with that obtained with H 3P04 pretreatment
under similar comditions. There was a linear relationship between cellu-
lose digestibility and percentage of xylan removed from the solid. An
increase in H 2S04 concentration (1-2% [w/v]) in the pretreatment step
increased cellulose digestibility by approx 25%. This research also con-
firms the previous results in the literature that Tween-SO is an effective
surfactant aid, which increases cellulose digestibility. The addition of

Applied Biochemistry and Biotechnology Vol. 705-708,2003


125

10% Tween-80 increased digestibility of H 2S04 (2% [w / v])-pretreated corn


stover by 15% compared with that of the Tween-free process.

Acknowledgments
We wish to acknowledge Colorado State University Experiment
Station, which partially funded this research. We wish to acknowledge
Dr. James Linden for his valuable comments and Dr. Thomas R. Hanley
for providing travel support for attending the symposium.

References
1. Parajo, J. c., Alonoso, J. L., and Santos, V. (1996), J. Wood Chem. Technol. 16(1),61-78.
2. Wyman, C. E. (1996), Handbook on Bioethanol: Production and Utilization, Taylor &
Francis, Washington, DC.
3. Fan, L. T., Gharpuray, M. M., and Lee, Y. H. (1987), Cellulose Hydrolysis, Springer-
Verlag, Berlin, Germany.
4. Schell, D. L Walter, P. J., and Johnson, D. K. (1992), Appl. Biochem. Biotechnol. 23/35,
659-663.
5. Philippidis, G. P., Smith, T. K., and Wyman, C. E. (1993), Biotechnol. Bioeng. 41,846-853.
6. Kim, S. B., Um, B. H., and Park, S. C. (2001), Appl. Biochem. Biotechnol. 91/93,81-94.
7. Kaar, W. E. and Holtzapple, M. T. (1998), Biotechnol. Bioeng. 59,419-427.
8. Castanon, M. and Wilke, C. R. (1981), Biotechnol. Bioeng. 23, 1365-1372.
9. Knappert, D., Grethlein, H., and Converse, A (1980), Biotechnol. Bioeng. 22, 1449-1463.
10. (1996), NREL Chemical Analysis and Testing Standard Procedure, No. 001-005,009,
National Renewable Energy Laboratory, Golden, CO.
11. Um, B. H. (2002), MS thesis, Colorado State University, Fort Collins, CO.
12. Himmel,M. E., Baker,J. 0., and Overend, R. P. (1994),Enzymatic Conversion of Biomass
for Fuel Production, American Chemical Society, Washington, DC.
13. Hernandez-Soto, A (2002), Inhibitory Effects of Acetic Acid and Furfural on Ethanol
Fermentation by Zymomonas mobilis C25. MS report, Department of Chemical Engi-
neering, Colorado State University, Fort Collins, CO.
14. Tengborg, c., Stenberg, K., Gable, M., Zacchi, G., Larsson, S., Palmqvist, E., and
Hahn-Hagerdal, B. (1998), Appl. Biochem. Biotechnol. 70/72,3-15.
15. Ramos, L. P., Breuil, c., and Saddler, J. N. (1992), Appl. Biochem. Biotechnol. 34/35,
27-47.
16. Mooney, C. A, Mansfield, S. D., Touhy, M. G., and Saddler, J. N. (1998), Bioresour.
Technol. 64, 113-119.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0127/$20.00

Combined Use of H2S04 and S02


Impregnation for Steam Pretreatment
of Spruce in Ethanol Production

JOHANNA SODERSTROM, LINDA PILCHER,


MATS GALBE, AND GUIDO ZACCHI*

Department of Chemical Engineering I,


Lund University, PO Box 124, SE-221 00 Lund, Sweden,
E-mail: guido.zacchi@kat.lth.se

Abstract
Fuel ethanol can be produced from softwood through hydrolysis in an
enzymatic process. Prior to enzymatic hydrolysis of the softwood, pretreat-
ment is necessary. In this study, two-step steam pretreatment employing
dilute H 2S04 impregnation in the first step and S02 impregnation in the
second step, to improve the overall sugar and ethanol yield, was investi-
gated. The first pretreatment step was performed under conditions of low
severity (I80°C, 10 min, 0.5% H 2S04) to optimize the amount of hydrolyzed
hemicellulose. In the second step, the washed solid material from the first
pretreatment step was impregnated with S02 and pretreated under condi-
tions of higher severity to make the cellulose more accessible to enzymatic
attack, as well as to hydrolyze a portion of the cellulose. A wide range of
conditions was used in the second step to determine the most favorable
combination. The temperatures investigated were between 190 and 230°C,
the residence times were 2, 5, and 10 min; and the S02 concentration was
3%. The effect of pretreatment was assessed by both enzymatic hydrolysis
of the solids and by simultaneous saccharification and fermentation (SSF)
of the whole slurry, after the second pretreatment step. For each set of
pretreatment conditions, the liquid fraction was also fermented to deter-
mine any inhibitory effects. Ethanol yield using the SSF configuration
reached 66% of the theoretical value for pretreatment conditions in the
second step of 210°C and 5 min. The sugar yield using the separate hydroly-
sis and fermentation configuration reached 71% for pretreatment condi-
tions of 220°C and 5 min.
Index Entries: Enzymatic hydrolysis; softwood; simultaneous saccharifi-
cation and fermentation; separate hydrolysis and fermentation.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 127 Vol. 105-108,2003


128 Soderstrom et al.
Introduction
Acid-catalyzed steam pretreatment of softwood is an effective way of
increasing the overall yield in the wood-to-ethanol process. This form of
pretreatment both increases the recovery of carbohydrates and enhances
the enzymatic hydrolysis (1). S02 and H 2S04 are two acids that have been
used as catalysts.
It is well known that more severe conditions during steam pretreat-
ment will cause greater degradation of hemicellulose sugars (2-5). How-
ever, a high degree of severity is required to enhance the enzymatic
digestibility of the cellulose fibers, especially in softwood (6). Steam pre-
treatment causes the hemicellulose to be degraded to its monomeric sug-
ars and the structure of cellulose to soften to make it more accessible to
enzymes in the following steps. However, the maximum yields of hemi-
cellulose sugars and glucose from cellulose are not reached at the same
degree of severity in the pretreatment. If a high degree of severity is used
sugars may degrade further to furfural, 5-hydroxymethylfurfural (HMF),
levullinic acid, and formic acid together with other substances. The for-
mation of degradation products reduces the overall yield, and the prod-
ucts may also cause inhibition in downstream process steps. On the other
hand, if a low degree of severity is used, cellulose digestibility will not be
enhanced, which will cause the overall sugar yield to be lower.
The overall yield has been found to be the most important parameter
when evaluating the production cost of bioethanol (1). Since the highest
costs in the process are those of the raw material (2) and the enzymes, it is
important to ensure a high degree of utilization of the carbohydrate com-
ponents in the feedstock to decrease prod uction cost. The idea of a two-step
steam pretreatment process has been proposed in the literature several
times as a means of increasing overall yield (4,6,7-10).
It has been shown that H 2S04 impregnation enhances the recovery of
hemicellulose sugars following pretreatment, while impregnation with
S02 promotes a higher recovery of glucose after enzymatic hydrolysis and
also results in less inhibition in the fermentation step compared with H2S04
(11). This indicates that two-step steam pretreatment has the potential to
result in higher sugar yields. Tengborg et a1. (11) proposed two-step steam
pretreatment in which the first step was performed at low severity with
H 2S04 as the acid catalyst to hydrolyze the hemicellulose. In the second
step, the washed solid material from the first step was impregnated
with S02 and steam pretreated, this time at high severity, to enhance the
enzyme accessibility (11).
In the present study, we investigated a two-step steam pretreatment
process. The conditions in the first pretreatment step were chosen to give
a high recovery of fermentable hemicellulose sugars in the liquid, which
involved the use of H2S04 as the acid catalyst. The solid material in the
slurry was washed with water and then treated again in the second pre-
treatment step. In the second step S02 was used as the acid catalyst to
Applied Biochemistry and Biotechnology Vol. 105-108,2003
H 2S04 and S02 in Steam Pretreatment of Spruce 129
promote cellulose degradation by enzymes in the subsequent step, irre-
spective of whether enzymatic hydrolysis or simultaneous saccharification
and fermentation (SSF) was used. We focused on utilization of hexoses, in
this case mainly mannose and glucose, since they can be fermented by
Saccharomyces cerevisae, the yeast used in this study. The effect of pretreat-
ment was assessed by both separate hydrolysis and fermentation (SHF)
and by simultaneous SSF. The second pretreatment step was optimized
with respect to the overall ethanol yield after SSF and, for SHF, to the
overall yield of fermentable sugars after enzymatic hydrolysis.

Materials and Methods


Procedure
The experimental procedure employed is shown schematically in
Fig. 1. The softwood was impregnated with dilute H 2S04 and then steam
pretreated. The resulting material was separated into a solid residue and
a liquid. The liquid was analyzed regarding sugars and then fermented.
The solid material was washed with water, then impregnated with gas-
eous S02 and steam pretreated in the second pretreatment step. The result-
ing material was evaluated by SSF of the slurry, by enzymatic hydrolysis
of the washed solid material, and by fermentation of the liquid.
The severity factor is often used for the evaluation of steam pretreat-
ment. Although it does not provide an accurate measure of the severity, it
can be used for rough estimates (9). The severity correlation describes the
severity of the pretreatment as a function of treatment time (min) and tem-
perature (OC), in which TreE = 100°C (12):

T- Trefl]
Log Ro =log(t· exp [ 14.75 (1)

Raw Material
Fresh spruce, Picea abies, free from bark, was used. Sawdust was sup-
plied by local sawmills. The composition was determined according to the
Hagglund (13) method and is presented in Table 1. The raw material used
for impregnation with H 2S04 in the first step had a dry matter (DM) content
of 55.5%.
Pretreatment
First Pretreatment Step
The first steam pretreatment step was optimized and performed at the
Mid Sweden University, OrnskOldsvik, in a 250-L batch reactor located in
Rundvik, Sweden (14). The reactor is a Masonite gun with direct steam
injection. The heat-up time is 30 s. The sawdust was impregnated with
dilute H 2S04 (0.5% [w /w] based on the water content of the wood) and
pretreated at 180°C for 10 min. The reaction was quenched by blowing the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
130 Soderstrom et al.
Raw material - spruce

Pretreatment
step 1

Fermentation
T=3Q2C
pH =5.5
yeast: 10 gOM/I Liquid
glucose to 50 fil

Pretreatment
step 2

SSF ........- - r - -J Solid material


T=372C
pH=S.O
enz.: IS FPU/g OM
yeast: S g OM/I
Fermentation Enzymatic
OM:S% T=3Q2C bydrolysis
pH=5.5
NaAcbuffer
yeast: 10 g DM/l
enz.:15 FPU/g DM
glucose to 50 f1l
DM:2%

Fig. 1. Experimental setup used for evaluation of two-step steam pretreatment.

material to a flash tank at atmospheric pressure. The impregnated material


had a DM content of 30%. The pretreated material, with a DM of 13%, was
separated by centrifugation into a solid residue and a liquid. The liquid was
analyzed regarding soluble sugars, and their degradation products. The
composition of the solid material was determined with the Hagglund (13)
method. The solid material was washed thoroughly with water to remove
all soluble substances and the yield and composition of the solid material
were determined (13).
Second Pretreatment Step
The second steam pretreatment step was performed at Lund Univer-
sity in a steam-explosion unit with a 2-L reactor volume (8). The solid
washed material from the first pretreatment step, with a DM content of
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
H 2S04 and 502 in Steam Pretreatment of Spruce 131

Table 1
Composition of Raw Material and Material
After First Pretreatment Step
Raw material First-step material
Composition (% ofDM) (% ofDM)

Gluean 49.9 53.7


Mannan 12.3 2.1
Lignin 28.7 38.4
Xylan 5.3 1.6
Galactan 2.3 0
Arabinan 1.7 0.6

Table 2
Experimental Design of Second Pretreatment Step
Experiment Tempature (0C) Time (min) Log (Ro)
1 190 2 2.95
2 190 5 3.35
3 190 10 3.65
4 200 2 3.25
5 200 5 3.64
6 200 10 3.94
7 210 2 3.54
8 210 5 3.94
9 210 10 4.24
10 220 2 3.83
11 220 5 4.23
12 225 5 4.38
13 230 5 4.53

37%, was reimpregnated with gaseous S02' The material was placed in
plastic bags for a 20-min impregnation at room temperature to reach a
concentration of 3% S02 ([w /w], based on the water content of the wood).
The impregnated material was steam pretreated in the second pretreat-
ment step at various temperatures (190, 200, 210, 220, 225, and 230°C) and
residence times (2, 5, and 10 min) (see Table 2). A portion of the pretreated
material was separated by filtration into a solid residue and a liquid for
evaluation with separate enzymatic hydrolysis and fermentation, and some
was kept intact for evaluation with SSF. The liquid was analyzed with
respect to soluble sugars and to their degradation products. The amount of
insoluble solids in the pretreated material was determined.
In two cases, excessive washing was performed on the material from
the first step to try to improve the S02 absorption in the impregnation prior
to the second pretreatment step. Two pretreatment experiments, nos. 8 and
11, which resulted in the highest overall yields in SSF and SHF, respec-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


132 Soderstrom et al.
tively, were repeated using this more thoroughly washed materiaL The
temperatures used were 210 and 220°C, and a residence time of 5 min was
used in both cases. SSF, enzymatic hydrolysis, and fermentation were per-
formed as described earlier.
The equipment and the procedure for determination of oligosaccha-
rides and for enzymatic hydrolysis, SSF, fermentation, and analysis have
been described in more detail previously (9). The methods are summa-
rized next.
Determination of Oligosaccharides by Acid Hydrolysis
The amount of oligomers in the liquid after the first pretreatment step
was determined by acid hydrolysis. This was performed in two ways: by
autohydrolysis using the acetic acid present in the liquid, and by the addi-
tion of H 2S04 •
Enzymatic Hydrolysis
Enzymatic hydrolysis of the washed solid material was used to assay
the second pretreatment step. A commercial cellulase mixture, Celluclast
1.5L (65 filter paper units (FPV)/g and 17 IV/g of ~-glucosidase) was
used, supplemented with the ~-glucosidase preparation Novozyme 188
(376 IV I g of ~-glucosidase), both kindly donated by Novozymes A/S
(Bagsv~rd, Denmark). Duplicate experiments with the washed solid
material were performed. A DM concentration of 2% (w Iw) was used to
avoid end-product inhibition in determination of the potential sugar
yield. A total of 10 g of DM, 2.32 g of Celluclast, and 0.52 g of Novozym
were immersed in 0.1 mol/L of NaAc buffer to a total mass of 500 g.
Hydrolysis was performed at 40°C and pH 4.8 for 96 h.
Simultaneous Saccharification and Fermentation
To assess the steam pretreatment conditions, SSF ofthe slurry from the
second pretreatment step was used as an alternative method. The slurry
was diluted with water to a final insoluble solids concentration of 5% DM.
The cellulase activity was 15 FPV I g of DM and the ~-glucosidase activity
was 23 IV I g of DM, which was the same as in the enzymatic hydrolysis.
Compressed baker's yeast, S. cerevisiae Gastbolaget AB, Rotebro, Sweden)
was used at an initial concentration of 5 g of DM/L. Antibiotics were added
to prevent infection and the formation of lactic acid to ensure comparable
results. SSF was performed at 37°C and pH 5.0 for 72 h. All experiments
were performed in duplicate, and the average values are presented.
Fermentation
Fermentation of the liquid was performed after the first and the sec-
ond pretreatment steps to investigate the fermentability and the extent of
inhibition. Glucose was added to the liquids to adjust the concentration of
fermentable sugar to 50 giL. A reference solution containing 30 giL of
Applied Biochemistry and Biotechnology Vol. 705-708,2003
H 2S04 and 502 in Steam Pretreatment of Spruce 133
glucose and 20 gil of mannose was also fermented. S. cerevisiae was used
at a concentration of 10 g of DM/L. Fermentation was performed at 30°C
and pH 5.5 for 24 h. Experiments were performed in duplicate.
Analysis
The liquids after the pretreatment steps and all samples from the
acid and the enzymatic hydrolysis, fermentation, and SSF were analyzed
by high-performance liquid chromatography (HPLC) (Shimadzu
LC-I0AT; Shimadzu, Kyoto, Japan) with a refractive index detector
(Shimadzu). Glucose, mannose, arabinose, galactose, and xylose were
separated using an Aminex HPX-87P column (Bio-Rad, Hercules, CA) at
80°C, using water as the eluent, at a flow rate of 0.5 mL/min. Cellobiose,
glucose, arabinose, lactic acid, glycerol, acetic acid, ethanol, HMF, and
furfural were separated on an Aminex HPX-87H column (Bio-Rad) at
65°C using 5 mmol/L H 2S04 as the eluent, at a flow rate of 0.5 mL/min.
All samples were filtered through a 0.20-pm filter before HPLC analysis.
Samples from the enzymatic hydrolysis and the liquid phases after the
pretreatment steps were analyzed on the HPX-87P column. However,
because of interference between ethanol and mannose on that column,
samples from SSF and fermentation were analyzed on the HPX-87H col-
umn. Analysis of glucose in the liquid phase after pretreatment was also
carried out on the HPX-87H column.

Results and Discussion


First Pretreatment Step
The composition of the dry raw material is presented in Table 1. Sixty-
two percent of the dry raw material consisted of glucan and mannan. The
material used has also been used in a previous study, in which the results
from the first pretreatment step have been discussed in detail (10), and,
hence, they are only summarized here. Ninety-three percent of the glucan
was recovered after the first pretreatment step. Eighty-one percent was still
present in the solids, whereas 12% was hydrolyzed and present in the liq-
uid as either oligomeric or monomeric sugars. Thirteen percent of the solu-
bilized glucan was recovered as oligomeric sugar. The recovery of mannan
was 100%. Eighty-eight percent of the mannan was solubilized and found
in the liquid, while the remainder was still in the nonhydrolyzed fibrous
material. Some of the solubilized mannan (12%) was recovered as oligo-
meric sugars.
The liquid contained only small amounts of furfural and HMF: 0.7 and
1.4 gil, respectively. Acetic acid was present at a concentration of 3.7 g/
L. The amount of acetic acid, 1.6 g/100 g of dry raw material, corresponds
well with the degree of acetyl substitution in galactoglucomannan.
Fermentation of the liquid from the first pretreatment step resulted in
a yield of 0.48 g of ethanol! g of sugar, i.e., 94% ofthe theoreticalfermentation
yield (data not shown), which was the same as for the reference solution. The
Applied Biochemistry and Biotechnology Vol. 105-108,2003
134 Soderstrom et al.
productivity of ethanol was about half that of the reference solution after
4 h, but after 24 h the yield was the same as for the reference solution.
Second Pretreatment Step
The second pretreatment step was performed using the washed solid
material from the first pretreatment step. This material contained mainly
glucan (53.7%) and lignin (38.4%). Only small amounts of some of the he mi-
cellulosic sugars were present: mannan (2.1 %), xylan (1.6%), and arabinan
(0.6%) (Table 1). The investigation covered a severity factor range ofLogRo
from 2.95 to 4.53 (Table 2). The second pretreatment step was evaluated
using SSF and enzymatic hydrolysis to determine the ethanol yield and the
glucose yield, respectively.
Although impregnation with 502 in the second pretreatment step was
expected to yield a highly accessible material, it resulted in poor overall
yields of sugar and ethanol. Problems in the reimpregnation with 502 were
observed. Only small amounts were absorbed although 3% 502 was added.
A probable explanation is that H 2S04 remained in the material after the first
step, which prevented 502 absorption.
The mean value of absorbed 502 in all experiments was 1.64% of the
water content, which is about half the amount added. However, variations
in the amount absorbed between 0.83 and 2.33% were observed. This can
affect the success of the pretreatment to a considerable extent, since the
catalytic effect could be deficient owing to the low amount of 502 absorbed.
The total yield of mannose and glucose in the second pretreatment
step-expressed as the sum of monomers in the liquid, and cellulose and
hemicellulose in the solid-varied from 46 to 70 g/lOO g of the solid mate-
rial from the first pretreatment step. This corresponds to a yield of 73-110%
based on the theoretical amount in the solid material after the first pretreat-
ment step. The yield was highest for low-severity conditions and decreased
with increasing severity in the second pretreatment step. It was assumed
that the lignin is not degraded during the steam pretreatment when calcu-
lating the yields after the second pretreatment step. This assumption was
made in determining the amount of carbohydrates in the solid material
after the second pretreatment step.
Most of the remaining mannan from the first step was degraded dur-
ing the second pretreatment step and obtained as mannose. The amount of
glucan that was hydrolyzed and recovered in the liquid as glucose varied
between 1 and 25% of the theoretical amount of glucan in the solid material
from the first step. The amount of glucan that was hydrolyzed to glucose
in the second pretreatment step reached a maximum for a severity factor of
Log Ro = 4.23 (220°C, 5 min) (Fig. 2).
At low severity the recovery of material, taking into account glucan,
mannan, their monomers, byproducts, and lignin, was close to 100%. How-
ever, for material pretreated at high severity the recovery was lower: for the
highest severity not more than 66%. Handling losses cannot justify these
poor mass balances. Some material is lost in the vapor, which was not
Applied Biochemistry and Biotechnology Vol. 105-108,2003
H 2S04 and S02 in Steam Pretreatment of Spruce 135
0.16
~ • Mannose, step 21 ,..,
0.14 [JGlucose, step 2I
0
'ii 0.12

I
E 0.10 n
~ D
0 0
" 0.08
!'
t
D
0.06
S

.--
'II
~ 0.04 0
'D

• ••
0.02

0.00
2.5 3.0
o
• •-
D
• I
3.5
Severity Factor (Log Ro)
4.0
•• •
4.5 5.0

Fig. 2. Yield of monomeric glucose and mannose in liquid after second pretreatment
step as function of severity of that step.

collected. Further studies in which this stream is considered are thus nec-
essary. Losses arising from handling were determined by thoroughly wash-
ing the equipment with water and measuring the amount of solid material
not recovered in the pretreated slurry. The average loss of solid material in
the second pretreatment step was estimated to be 2.4% of the original dry
material by weight.
The liquid after the second pretreatment step contained several
byproducts. At low severity the concentrations of acetic acid, HMF, and
furfural were very low, <0.5 giL. The HMF concentration reached a maxi-
mum of 3 giL for pretreatment at high severity. The furfural concentration
never exceeded 0.75 giL, which was expected, since almost all the pentoses
were recovered as monomeric sugars in the liquid from the first pretreat-
ment step (Fig. 3). Several other substances were detected as unidentified
peaks in the chromatograms but were not quantified. These substances
might be derived from the degradation of sugar and lignin.
All fermentation experiments on the liquid derived from the second
pretreatment step showed good fermentability and no apparent inhibitory
effects. The productivity for the first4h was6 gofethanol/ (L·h), which was
the same as in the reference solution.
Washed Material
Excessive washing of the material from the first step with water prior
to reimpregnation resulted in a higher uptake (the desired 3%) of S02.
However, SSF and enzymatic hydrolysis of this material did not result in
an increase in overall yields compared with the moderately washed mate-
rial exhibiting less S02 absorption.
Applied Biochemistry and Biotechnology Vol. 105-10B, 2003
136 Soderstrom et al.
3.5

DHMF
3.0 - & Furfural L.J

~ 2.5
0
...
~ 20 0
.E
c
R
~ 1.5 L.J

~ 0
8 1.0


-•
~

0.5
0

-- --
*j
0.0
2.5 3.0
!~
-- 3.5 4.0 - ...
4.5 5.0
Severity Factor (Log Rol

Fig. 3. Concentration of potential inhibitors in liquid after second pretreatment step


as function of severity factor of pretreatment.

Enzymatic Hydrolysis
For enzymatic hydrolysis to be successful, the cellulose fibers must be
accessible to the enzymes. More severe pretreatment results in a material
that is more accessible to enzymatic attack. If the material is treated under
very severe conditions, much of the cellulose will be hydrolyzed during the
second pretreatment step without the use of enzymes. When treated at very
severe conditions, the sugars are degraded during pretreatment to form
inhibiting substances, which also cause a loss of substrate.
The sugar yields during the enzymatic hydrolysis step ranged from
12 to 20 g of sugar/100 g of dry raw material, depending on the pretreat-
ment conditions. Figure 4 shows the glucose yield in the various steps.
The highest yield in the enzymatic hydrolysis step, 19 g of glucose/100 g
of dry raw material, was obtained for a severity of Log Ro = 3.94, corre-
sponding to pretreatment conditions of 200°C and 10 min. For pretreat-
ment at a severity above 4 the amount of glucan that was hydrolyzed
in the enzymatic hydrolysis was smaller. This is an effect of the higher
hydrolysis yield and increased degradation during the second pretreat-
ment step, which leaves less material for enzymatic hydrolysis.

Simultaneous Saccharification and Fermentation


The success of SSF is dependent on the hydrolysis of the cellulose as
well as on the fermentation of sugar to ethanol. A material pretreated at
low severity in the second pretreatment step will not yield cellulose fibers
that are accessible to enzymatic attack. However, if the material is treated

Applied Biochemistry and Biotechnology Vol. 105-108,2003


H2 S04 and 502 in Steam Pretreatment of Spruce 137

0.35
• Filtrate step 1
0.30 c FlItrate step 2
• Enzymatic hydrolysis
o Overall 0 0
I'i 025 0
E 0 o§
~
..
~ 0.20

0 • 0 ••

) 0.15
~
1St
!!
31 0.10
~
0 05

0.00
2.50 3 .00 3.50 4.00 4.50 500
Severity FlIClo, (Log Ro)

Fig. 4. Yield of glucose formed in each step as function of severity factor of pre-
treatment.

at high severity, inhibitors may form, which affect the fermentation and
inhibit the yeast.
The yield of ethanol in SSP of the slurry from the second pretreat-
ment step was calculated assuming that no lignin degradation occurred
in the pretreatment. The highest yield of ethanol reached in SSP was 71 %
and was obtained following pretreatment at 230°C for 5 min. However,
the highest overall ethanol yield (Le., including both pretreatment steps
and SSP) was 66%.
Overall Yields
The formation of glucose and mannose occurred in different steps of
the process. Mannose was mainly formed during the first pretreatment
step, with a yield of 88% of the theoretical value. In the second step, 2-8%
of the theoretical amount of mannan was obtained, depending on the pre-
treatment conditions. Thus, the total yield of mannose was 90-96% of the
theoretical.
Glucose was mainly obtained in the second pretreatment step and
during enzymatic hydrolysis. Amaximumof21%ofthetheoreticalamount
of glucose was obtained in the second pretreatment step for pretreatment
conditions of 220°C and 5 min. Considering enzymatic hydrolysis, the high-
est yield of glucose was 37%, obtained after pretreatment at 200°C for 10
min. However, the maximum yield following the second pretreatment step
and enzymatic hydrolysis only reached 49%. In this case, pretreatment was
performed at 220°C for 2 min.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
138 Soderstrom et al.
0.7

••
0

• -
~

0.6 A

0
0.5 • 0 •o •0 A

j
11 0.4 • v
• 0
0

! 0 0
~
_ 0.3
..,
~
0.2

0.1
l <> Overall EtOH yield with SSF I
• Overall EtOH yield with SHF
0.0
2.5 3.0 3.5 4.0 4.5 5.0
Severity Factor (log Ro)

Fig. 5. Overall yields of ethanol following SSF and SHF as function of severity factor
of pretreatment. In SHF the fermentation yield after enzymatic hydrolysis is assumed
to be 90%.

The overall yield of fermentable sugars (Le., following the two pre-
treatment steps and the enzymatic hydrolysis step) was about 70% for a
range of pretreatment conditions with a severity factor of about 4 (see Fig. 4).
The maximum yield of fermentable sugars (0.46 g/ g dry raw material or
71 %) was obtained following second-step pretreatment conditions of 220°C
and 5 min.
The maximum yield of sugars obtained in our study is lower than in
previous studies in which either two-step steam pretreatment with S02
impregnation in both steps (80%) (9) or H 2S04 impregnation in both steps
(77%) (10) was utilized. The same raw material and evaluation methods
were used as in the present study. Nguyen et aL (6) obtained an overall
sugar yield of 82% using two-step steam pretreatment followed by enzy-
matic hydrolysis. However, the cellulase activity used in the present study
was very much lower, 15 FPU / g of DM (25 FPU / g of cellulose), compared
with the 60 FPU/g of cellulose used by Nguyen et aL (6).
Figure 5 shows a comparison of SSF and SHF, in which a yield of 90%
in the fermentation after enzymatic hydrolysis was assumed, which was
the yield reached in the fermentation tests. When the material was steam
pretreated in two steps followed by either SHF or SSF approximately the
same ethanol yield resulted. The highest overall ethanol yield using SSF
was 66%, reached at a severity factor of Log Ro = 3.94 (210°C, 5 min). For
SHF the highest overall ethanol yield was 67% and was obtained for a
severity factor of Log Ro = 4.23 (220°C, 5 min).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


H 2S04 and 502 in Steam Pretreatment of Spruce 139
Previous results from one-step steam pretreatment have shown that
SSF resulted in higher ethanol yields (67%) than SHF (60%) (8,15). The use of
antibiotics in SSF when evaluating the two-step steam pretreatment may
cause a decrease in ethanol yield (9). This may, to some extent, explain why
the SSF configuration was inferior to the SHF configuration in this case, while
the opposite was found when employing one-step steam pretreatment.
Two-step steam pretreatment with H 2S04 impregnation in both steps
resulted in an overall ethanol yield of 65% with the SSF configuration and
69% with the SHF configuration (10). The main reason for the lower yield
following SHF in the present study, compared with the one using H 2S04 for
impregnation in both steps, is a lower sugar yield in the second pretreat-
ment step. When S02 was used for impregnation in both steps, the overall
ethanol yield reached 69% with the SSF configuration and 72% with SHF.
In this case, the higher sugar yield was mainly owing to a higher yield and
higher sugar production in the enzymatic hydrolysis step.

Conclusion
Two-step steam pretreatment of softwood with impregnation by
dilute H 2S04 in the first step and S02 in the second step resulted in lower
sugar yields after enzymatic hydrolysis than did procedures incorporating
impregnation with either S02 or H 2S04 in both steps. The ethanol yield after
SSF was about the same as when H 2S04 was used in both steps, however,
when S02 was used in both steps, the yield was higher. The highest overall
ethanol yield reached with the SSF configuration was 66% of the theoretical
(210°C, 5 min), whereas the highest overall sugar yield with the SHF con-
figuration was 71 % (220°C, 5 min).
This is contrary to what was expected based on previous results from
one-step pretreatment with either H 2S04 or S02 (11). Judging from those
results, a first step using H 2S04 followed by a second step with S02 was
thought to be superior. However, the results from our study show that it is
not possible to predict the optimal conditions for two-step steam pretreat-
ment based on a one-step pretreatment procedure. The combination of
H 2S04 in the first step followed by S02 in the second step is thus not a better
alternative than utilization of H 2S04 or 502in both steps.

Acknowledgments
We are grateful to Dr. Robert Eklund at the Mid Sweden University,
brnskoldsvik, for providing the raw material and performing the first
pretreatment step. We also gratefully acknowledge the Swedish National
Energy Administration for financial support.

References
1. von Sivers, M. and Zacchi, G. (1996), Bioresour. Technol. 56(2/3), 131-140
2. Boussaid, A., Robinson, J., Cai, Y., Gregg, D. J., and Saddler, J. N. (1999), Biotechnol.
Bioeng. 64(3),284-289

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


140 Soderstrom et al.
3. Wu, M. M., Chang, K, Gregg, D. J., Boussaid, A., Beatson, R. P., and Saddler, J. N.
(1999), Appl. Biochem. Biotechnol. 77-79,47-54
4. Heitz, M., Capek-Menard, E., Koeberle, P. G., Gagne, J., Chornet, E., Overend, R. P.,
Taylor, J. D., and Yu, E. (1991), Bioresour. Technol. 35(1) 23-32
5. Nguyen, Q. A., Tucker, M. P., Keller, F. A., Beaty, D. A., Connors, K M., and Eddy,
F. P. (1999), Appl. Biochem. Biotechnol. 77-79,133-142
6. Nguyen, Q. A., Tucker, M. P., Keller, F. A., and Eddy, F. P. (2000), Appl. Biochem.
Biotechnol. 84-86,561-576
7. Nguyen,Q.A., Tucker,M. P., Boynton, B. L.,Keller,F.A.,andSchell,D.J. (1998),Appl.
Biochem. Biotechnol. 70-72, 77-87
8. Stenberg, K, Tengborg, c., Galbe, M., and Zacchi, G. (1998), J. Chem. Technol.
Biotechnol. 71, 299-308
9. Soderstrom, J., Pilcher, 1., Galbe, M., and Zacchi, G. (2002) Appl Biochem. Biotechnol.,
98-100,5-21.
10. Soderstrom, J., Pilcher, 1., Galbe, M., and Zacchi, G. (2003) Biomass Bioenergy, in press.
11. Tengborg, c., Stenberg, K, Galbe, M., Zacchi, G., Larsson, S., Palmqvist, E., and
Hahn-Hagerdal, B. (1998), Appl. Biochem. Biotechnol. 70-72,3-15
12. Overend, R. P. and Chornet, E. (1987), Philos. Trans. R. Soc. London A 321(1561), 523-
536.
13. Hagglund E. (1951), Chemistry of Wood, Chapter 8, Academic Press, New York, NY.
14. Eklund, R. and Petterson, P. O. (2000), Proceedings of ISAF XIII, July 2000, Stockholm
Sweden.
15. Stenberg, K, Bo1l6k, M., Reczey, K, Galbe, M., and Zacchi, G. (2000), Biotechnol.
Bioeng. 68(2), 204-210.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0141/$20.00

Effect of Lignocellulosic Degradation


Compounds from Steam Explosion
Pretreatment on Ethanol Fermentation
by Thermotolerant Yeast
Kluyveromyces marxianus
JOSE MIGUEL OLIVA, FELICIA SAEZ, IGNACIO BALLESTEROS,
ALBERTO GONzALEZ, MARIA JOSE NEGRO,
PALOMA MANZANARES, AND MERCEDES BALLESTEROS·
ClEMAT-Departamento de Energfas Renovables, Av. Complutense, 22,
28040, Madrid, Spain, E-mail: m.ballesteros@ciemat.es

Abstract
The filtrate from steam-pretreated poplar was analyzed to identify degra-
dation compounds. The effect of selected compounds on growth and
ethanolic fermentation of the thermotolerant yeast strain Kluyveromyces
marxianus CECT 10875 was tested. Several fermentations on glucose
medium, containing individual inhibitory compounds found in the hydroly-
sate, were carried out. The degree of inhibition on yeast strain growth and
ethanolic fermentation was determined. At concentrations found in the
prehy-drolysate, none of the individual compounds significantly affected
the fermentation. For all tested compounds, growth was inhibited to a lesser
extent than ethanol production. Lower concentrations of catechol (0.96 giL)
and 4-hydroxybenzaldehyde (1.02 giL) were required to produce the 50%
reduction in cell mass in comparison to other tested compounds.
Index Entries: Ethanol production; Kluyveromyces marxianus; poplar bio-
mass; inhibitors; fermentation.

Introduction
Conversion of lignocellulosic biomass into ethanol as an alternative to
conventional petroleum transportation fuels is currently under extensive
investigation (1,2). The simultaneous saccharification and fermentation (SSF)
process has been suggested as one of the most promising systems because the
continuous removal of the sugars by the microorganism reduces the end-
"Author to whom all correspondence and reprint requests should be addressed.
Applied Biochemistry and Biotechnology 141 Vol. 705-708,2003
142 Oliva et al.
product inhibition of the enzyme complex. Over the past 10 years, our labo-
ratory has selected a thermotolerant strain of Kluyveromyces marxianus
capable of ethanol fermentation of glucose from cellulose in an SSF process
at 42°C with a good ethanol yield (3).
Steam explosion has been proposed as an efficient pretreatment of
lignocellulosic materials and has the advantage of being able to be devel-
oped on a commercial scale (4,5). During steam explosion pretreatment,
some degradation products are formed that may be potential inhibitors
during fermentation of the sugar fraction. After pretreatment, these inhibi-
tors are in the aqueous portion of the pretreatment hydrolysate slurry. In
an environmentally sustainable lignocellulose-to-ethanol process, this
aqueous fraction should be used as fermentation broth to minimize fresh
water requirements and decrease the amount of water waste produced.
Several researchers have investigated the nature of the inhibitors present
in dilute-acid hydrolysates and steam explosion-pretreated biomass (6-9).
Nevertheless, not only does the concentration of the final inhibiting com-
pounds vary greatly with the pretreatment conditions and the raw material
used, but also their effects depend on the nature of the microorganism, as
well as pH and temperature conditions of the fermentation broth.
In the present study, the filtrate from steam-exploded poplar was ana-
lyzed to identify degradation compounds. The effect of selected compounds
on ethanol fermentation of the thermotolerantyeast strain K. marxianus CECT
10875 was tested.
Materials and Methods
Chemicals
All chemicals were obtained from Sigma (St. Louis, MO), except for
4-hydroxy-3,5-dimethylbenzyl alcohol (syringyl alcohol), which was
obtained from Lancaster Synthesis (Morecambe, England).
Preparation of Poplar Hemicellulose Hydrolysate
One hundred grams of poplar biomass was subjected to pretreatment
in a steam explosion pilot unit at 210°C and 4 min, operated by batches and
equipped with a 2-L reaction vessel. The plant description and working
methodology were described previously (10). The pretreated material was
suction filtered, and the filtrate was collected (approx 1 L) and analyzed.
Microorganism and Fermentation Conditions
K. marxianus CECT 10875, a thermotolerant strain obtained in our labo-
ratory, was used in fermentation experiments. Active cultures for inocula-
tion were prepared by growing the organism overnight on a rotary shaker
at 150 rpm and 42°C in a growth medium (initial pH 5.5) containing: 5 giL
of yeast extract, 2 giL of NH4Cl, 1 giL of KH2P04, 3 giL of MgS047H20, and
30 giL of glucose.
Experiments were carried out in 100-mL Erlenmeyer flasks each con-
taining 25 mL of the growth medium under nonsterile conditions and agi-
tated at 150 rpm. Flasks were inoculated at 4% (v Iv).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Degradation Compounds from Sf on Fermentation 143
To study the effect of the concentration of toxic substances on growth
and ethanol production, different amounts of individual inhibitory com-
pounds were added to the fermentation medium. A culture without toxic
compounds was used as control. After 24 h of fermentation, the flasks were
checked for cell growth and ethanol production.
In experiments with organic acids supplementation, the broth was
adjusted to pH 5.5 with 2 N NaOH. In experiments studying the assimila-
tion profile of aldehydes and glucose fermentation, samples were taken
periodically. All experiments were carried out in triplicate.
Analytical Procedures
Vanillin, vanillyl alcohol, syringaldehyde, syringyl alcohol, 4-hydroxy-
benzaldehyde, 4-hydroxybenzyl alcohol, catechol, guaiacol, 4-hydroxyben-
zoic acid, syringic acid, and vanillic acid analyzes were performed on a
high-performance liquid chromatography (HPLC) system. An 1100 Hewlett
Packard liquid chromatograph with an Aminex HPX-87H stainless steel
(300 x 7.6 mm) column (Bio-Rad, Hercules, CA) and a 1040A Photodiode-
Array detector was used. The mobile phase was 82% 5 mM H 2S04 and 18%
acetonitrile at a flow rate of 0.3 mL/min and temperature of 55°C. For
5-hydroxymethylfurfural (HMF), furfural, and furfuryl alcohol analysis, a
reversed-phase (250 x 4 mm) Lichrosorb RP18 column was employed. The
mobile phase was a mixture of buffer solution of 1.25 giL of monobasic
sodium phosphate and 1.25 giL of dibasic sodium phosphate, and metha-
nol in a proportion of 90-10%. A 0.6 mL/ min flow rate and 50°C tempera-
ture were used.
Acetic, levulinic, and formic acid analysis was carried out with an
HPLC system with a refractive index detector. Separation was performed
with an Aminex HPX-87H column (Bio-Rad). The mobile phase was 5 mM
H 2S04, at a flow rate of 0.6 mL/min and temperature of 65°C.
Ethanol was measured by gas chromatography using a Hewlett
Packard 5890 Series II apparatus with a flame ionization detector and a
columm of Carbowax 20M (2 m x 32 mm) at 85°C. Injector and detector
temperatures were 150 and 190°C, respectively.
Cell mass was determined by optical density measurements at 600
nm using a Jasco V-530 UV /VIS spectrophotometer, and gravimetrically
by dry weight as follows: culture liquid (5 mL) was filtered (0.45-llm HA;
Millipore) and the filter washed with water and dried to constant weight
in a microwave oven for 15 min. Measurements of cell growth were done
in triplicate.

Results
Identification of Degradation Compounds in Hydrolysate
The hydrolysate from steam explosion pretreatment of poplar at 210°C
and 4 min residence time was analyzed. Table 1 shows the quantitative
determination of degradation compounds identified. Results are expressed
as milligrams of the compound per liter of filtrate.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
144 Oliva et al.
Table 1
Degradation Products Composition of Liquid Fraction
Obtained After Steam Explosion Pretreatment of Poplar
Compound Concentration (mg/L)a
Acetic acid 2100
Formic acid 430
Levulinic acid NQ
Furfural 490
5-HMF 80
4-Hydroxybenzaldehyde NQ
4-Hydroxybenzoic acid 100
Catechol 30
Guaiacol NQ
Syringaldehyde 50
Syringic acid NQ
Vanillin 14
Vanillic acid NQ
aNQ, not quantified.

Acetic acid (2100 mg/L) and furfural (490 mg/L), from hemicellulose
degradation of hardwood poplar, were the main compounds present in
the hydrolysate. 4-Hydroxybenzoic acid (100 mg/L) and syringaldehyde
(50 mg/L) constituted a large fraction of the lignin-derived compounds in
the hydrolysate.
Effects of Degradation Compounds on Growth and Fermentation
The inhibitory effect of various concentrations of toxic compounds on
growth and ethanol fermentation of K. marxianus CECT 10875 was investi-
gated. Cultures of yeast strain were grown in a glucose-containing medium
and supplemented with varying initial concentrations of degradation com-
pounds identified in the hydrolysate. Dose-response curves for ethanol
production and growth at 24 h were determined for all compounds. Results
of growth and ethanol concentration were expressed as percentage of the
control (one hundred percent of the control growth and ethanol production
is equivalent to 3.4 ± 0.1 giL and 12.3 ± 0.5 giL, respectively). Results for
acetic acid, furfural, 4-hydroxybenzaldehyde, 4-hydroxybenzoic acid, and
catechol are shown in Fig. 1. Experiments using formic and levulinic acids
showed results similar to acetic acid assays, results obtained using 5-HMF
were similar to 5-furfural, and those using syringaldehyde were similar to
4-hydroxybenzaldehyde (data not presented).
In all cultures, growth inhibition was more intensive than ethanol pro-
duction. Dose-response curves for growth in cultures supplemented with
4-hydroxybenzaldehyde and furfural exhibited a sigmoidal pattern, while
catechol followed almost a linear pattern. No total inhibition of both ethanol
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Degradation Compounds from Sf on Fermentation 145

•• GROIIVTH
l00'l'=::~ __

O+-~~~~~-r~~~~~
0.00 025 0.50 0.75 1.00 1.25 1.50 1.75 123 4

4-HV0R0XYBENZALDEHYCE ~) 4-HYDROXYBfNZOICAClD~

GROWTH • GROWIlI
100. ETJ-WolOL 100 - - - ' - - t_ _ _ _ ETHANOL
• "!

l75 ~

i~
15

f2
~~
~
~
I~ ~
II::
25

0.0 0.5 1.0 1.5 20 2!> 3.0 0.0 2.5 1.5 10.0
F~FURAI.·(Wlj ACEllC ACID (Wlj

100 • GROIIVTH

l75

I~
I~
f2

0.25 0.50 0.75 1.00 1.25 1.50 1.75


CATECHOL (Wlj

Fig. 1. Effect of degradation compounds produced in steam explosion pretreatment


of poplar biomass on growth and ethanol production of K. marxianus CECT 10875 after
24 h. SDs are represented by error bars. One hundred percent control equivalent to
growth = 3.45 giL, and ethanol production = 12.3 giL.

production and growth was observed at the highest 4-hydroxybenzoic acid


concentration (4 giL) assayed.
Ethanol production was not affected by the presence of acetic acid at
the concentrations tested. At the highest acetic acid concentration (10 giL),
growth was 53% of the control (one hundred percent of the control growth
Applied Biochemistry and Biotechnology Vol. 105-108,2003
146 Oliva et al.
Table 2
Concentration of Toxic Compounds Required
to Inhibit 50% of Growth and Ethanol Production
of K. marxianus CECT 10875 at 24 h
Inhibitory concentration (giL)
Growth Ethanol
Furfural 2.53 2.60
5-HMF 4.01 4.20
Vanillin 2.55 2.67
Syringaldehyde 2.86 3.50
4-Hydroxybenzaldehyde 1.02 1.24
Catechol 0.96 1.19
Guaiacol 1.43 1.75
4-Hydroxybenzoic acid 3.10 3.86

is equivalent to 3.4 ± 0.1 giL). Dose-response curves for ethanol production


followed a sigmoidal pattern for all compounds except acetic acid (Fig. 1)
and formic and levulinic acids (data not shown).
To facilitate comparison of toxicity for the compounds assayed, the
concentrations of the test agents that caused 50% inhibition of growth and
ethanol production are given in Table 2. Catechol and 4-hydroxybenzalde-
hyde showed the highest inhibitory effect. An initial concentration of
these compounds close to 1 giL caused 50% inhibition of both growth and
ethanol production processes. Furfural was more toxic than 5-HMF, and
a twofold higher concentration of HMF was necessary to reach the 50%
level of inhibition.
Effect of Initial pH on Toxicity of Organic Acids
Considering that the initial pH of the fermentation medium has a
marked effect on toxicity of acids, the influence of the different acids
found in the hydrolysates on ethanol production was examined, indi-
vidually, at two different pHs (Fig. 2). The toxic effect of all acids tested,
at a concentration of 5 giL, decreased with increasing pH. At pH 5.5, the
addition of 5 giL of organic acids had almost no effect on ethanol produc-
tion. However, at pH 4.0, the presence of 5 giL of levulinic, formic, or
vanillic acid, individually, blocked ethanol production (Fig. 2). Less inhi-
bition was observed for 4-hydroxybenzoic and syringic acid (30 and 75%
of the control, respectively).
Assimilation Profiles of Aldehydes
Analysis of culture media extracts supplemented with inhibitors over
the course of fermentations by K. marxianus CECT 10875 showed that acids
and alcohols were not metabolized by the yeast, and that their concentra-
tion remained constant during fermentation (data not shown).
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Degradation Compounds from Sf on Fermentation 147

100 c:=JpH 4
~pH5.5

75
~
d
~ 5.0

;
~

25

Ac Hb v Fe Le Sy

Fig. 2. Effect of organic acids (5 giL concentration) on ethanol production by


K. marxianus CECT 10875 at differentinitial pHs. Ac, acetic acid; Hb,4-hydroxybenzoic
acid; V, vanillic acid; Fo, formic acid; Le, levulinic acid; Sy, syringic acid.

In the case of furfural, 5-HMF and aromatic aldehydes, the microor-


ganism had the ability to assimilate these compounds. In Fig. 3 the assimi-
lation profiles for several aldehydes are shown at two different initial
concentrations.
At a concentration of 1 giL (Fig. 3A), furfural was completely metabo-
lized in the first 4 h of fermentation. 5-HMF, vanillin, and syringaldehyde
were also metabolized by the yeast strain, but their assimilation rates were
slower. The concentration of these compounds decreased during fermen-
tation and they became undetectable after 8 h of fermentation. The assimi-
lation of 4-hydroxybenzaldehyde was significantly slower; thus after 8 h
of fermentation, only 0.1 giL of this compound had been assimilated. After
24 h, however, no 4-hydroxybenzaldehyde remained in the medium.
At an initial aldehyde concentration of 2 giL (Fig. 3B), furfural and
5-HMF were completely assimilated after 8 h incubation, and vanillin and
syringaldehyde were assimilated after 16 h. K. marxianus CECT 10875 only
assimilated 30 % of the 4-hydroxybenzaldehyde after 24 h of incubation.
Effect of Aldehydes on Glucose Fermentation
Fermentation profiles in the presence of 2 giL of furfural, vanillin, and
syringaldehyde, and 1 giL of 4-hydroxybenzaldehyde are presented in
Fig. 4. The initial aldehyde concentration was selected according to the
maximum concentration that allows the yeast to reach a final ethanol con-
centration similar to control (12.3 giL).
As can be seen in Fig. 4A, furfural was completely reduced to furfuryl
alcohol in the first 8 h of fermentation. No ethanol production was observed
when furfural was present in the medium. A final ethanol concentration of
12.1 giL was reached after 20 h of fermentation.
Applied Biochemistry and Biotechnology Vol. 105-708,2003
148 Oliva et al.
A
1.0

0.8 0.8

.. 0.6

0,4

0;04,- .........--,r---.-~~::=;=-1'--.;:.~"1'---,.--IJ:,....,.....,.Jp....j 0.0


o 2 46 820 22 24
T1M~(h)

B
l,a

1;6
:::J'
~
~1,2 1,2

~w 1),8 .....

o
:z
8 iM :0,4

b 2 4 68101214162224
T1ME(h}

Fig. 3. Assimilation profiles for K. marxianus CECT 10875 of (A) 1 giL and (B) 2 giL
of different aldehydes obtained in hydrolysate after steam explosion pretreatment of
poplar. (_) Furfural; (e) 5-Hydroxymethylfurfural; (A) Vanillin; (T) Syringaldehyde;
(.) 4-Hydroxybenzaldehyde.

Fig. 4. (opposite page) Conversion of aldehydes and ethanol production during glu-
cose fermentation by K. marxianus CECT 10875. (A) Furfural, (B) Vanillin, (e)
Syringaldehyde, and (D) 4-Hydroxybenzaldehyde.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


)-
~ 12 I ~ 12
~ 2.0 2,0
t:l.
OV to '\. I ~·10

I')
::,-
!b
1.5
.8
1.5
.-----
3
;;;.
~ ~. ~ i
III 1.0
::. ~ J ~il.O I
t:l.
ov
II. w :>b
o· ~i 1:1
fD
I') ~
" )0.5
0.5
::,-
::. 2 -12
0 •
~
'<: 0.0'
A
0
/ B 0
0 5: 1.0 '1la/E (Ill IS 20 Z5 5 lD 1UIE(h)15 20 25
-+- fliflhl-+- furfur,1alcohCII ......... Elhanol -+- VIII1iIin -+- Vanilylllicahol -+-Elhenol
~
1.0

2.0 12 1.0
/ 12
10 -110
0.& 10'
1.5
II ~
i
./]. i
& g
i 0.8 . s3'
u !
4 •
~
:-
11 ~O.5
0
v,
I
/ 2
r:I
0
,Ccl
- 0.0
c 0 I I ,
:; 25 ~
0 lQ lJlIE (h)15 20 o 5 W ft
~ ~1:
20 25
1":] ·,L
_ _ Elhanol TIME (h)
0
""0 _ _.Symgaldehyde -+- SyiInmI aIcohcil --"'hydfOllY~i(Ie~ 4-hyd~ 1IICOh«-+- Ethanol
...,
150 Oliva et al.
Ethanol production in the presence of 2 giL of vanillin (Fig. 4B) and
syringaldehyde (Fig. 4C) followed a similar pattern. These compounds
were metabolized completely by the microorganism after 16 h of fermen-
tation. As observed with furfural, the reduction of toxic aldehydes to their
corresponding alcohols was necessary for ethanol production to proceed.
Final ethanol concentrations close to 12 giL were obtained in both experi-
ments after 24 h of fermentation.
The results of Fig. 4D show that K. marxianus had the ability to reduce
1 giL of 4-hydroxybenzaldehyde after a lag period of 8 h. The reduction
rate was significantly slower in comparison to the other aldehydes tested,
so the microorganism needed an incubation period of 24 h to metabolize all
the 4-hydroxybenzaldehyde. At this time only 0.76 giL of 4-hydroxybenzyl
alcohol was obtained. A final ethanol concentration of 12 giL was achieved
after 30 h of fermentation.

Discussion
Degradation Compounds in Hydrolysate
All compounds found in the liquid fraction obtained from steam
explosion pretreatment of poplar biomass have been previously identified
in other hardwood hydrolysates (6-9). Acetic acid from hydrolysis 0 fhemi-
cellulose and furfural from degradation of xylose were obtained as a con-
sequence of the high xylan content in hard wood. Vanillin, which is formed
by degradation of guaiacylpropane units of lignin and syringaldehyde,
which are formed in turn by the degradation of syringyl propane units, has
been reported in hydrolysates from other hardwoods such as poplar (6,7)
and red oak (8). 4-Hydroxybenzoic acid constitutes a large fraction of lig-
nin-derived compounds in hydrolysates from the hardwoods poplar (7)
and willow (9).
Effect of Degradation Compounds on Growth and Fermentation
The effect of toxic compounds generated in the pretreatment of ligno-
cellulosic biomass has been examined in different microorganisms such as
prokaryotes (11-15), xylose-fermenting yeasts (16,17), and Saccharomyces
cerevisiae (17-23). However, no studies of the inhibitory effect on growth
and ethanol production of these compounds have been reported for the
thermotolerant yeast K. marxianus.
To interpret the results in the literature, it should be kept in mind that
the solubility of these compounds is limited. Thus, when a high concentra-
tion of a compound is tested, it is possible that, in fact, the microorganism
is exposed to a lower concentration (24).
At the concentrations tested, growth inhibition was observed for all
compounds examined. Cell growth was generally more influenced than
ethanol production for all conditions assayed.
Results from dose-response curves of the different compounds tested
showed that catechol and 4-hydroxybenzoic acid inhibit the growth of the
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Degradation Compounds from Sf on Fermentation 151

yeast strain in a near linear pattern, indicating that the inhibition of growth
is closely related to the initial concentration of these molecules. However,
aldehydes exhibited a sigmoidal inhibition with a shoulder at low concen-
trations. Bacteria and yeasts have been shown to metabolize furans (20-22),
and aromatic aldehydes (15,17-19), although enzymes involved in the
metabolic pathways remain unknown in most cases. In our study the
assimilation of furfural, HMF, vanillin, syringaldehyde, and 4-hydroxy-
benzaldehyde by K. marxianus was demonstrated. The strain exhibited
higher assimilation rates for all aldehydes compared to what has been
reported for other glucose-fermenting microorganisms such as Klebsiella
pneumoniae (15), Saccharomyces cerevisiae, and Zymomonas mobilis (17).
According to Delgenes et al. (17), S. cerevisiae exhibited a lag assimilation
period of 24 and 30 h at an initial concentration of 2 giL of furfural and
vanillin, respectively. K marxianus, however, required only 8 and 16 h,
respectively, to completely assimilate these compounds.
The reduction of furfural to furfuryl alcohol at the beginning of fer-
mentation has been observed by other investigators (21,24), and is gener-
ally believed to be catalyzed by NADH-dependent alcohol dehydrogenase
(16). Furfural is a strong inhibitor of ethanol fermentation of K. marxianus.
A total conversion of furfural to the corresponding alcohol is needed to
start ethanol production. The conversion of furfural to furfuryl alcohol and
furonic acid has been reported for a number of yeasts such as S. cerevisiae
(20-22),Pichia (16), Turolopsis, and Rhodotorula (25). We have found furfuryl
alcohol as the only metabolite from furfural assimilation.
Likewise, vanillin and syringaldehyde were metabolized by
K. marxianus in the fermentation process and totally converted into their
corresponding alcohols. In experiments with 4-hydroxybenzaldehyde,
however, although all the aldehyde was assimilated by the tested strain, it
was only partially transformed to 4-hydroxybenzyl alcohol (76%), and an
unidentified compound was detected by HPLC analysis.
4-Hydroxybenzaldehyde was the most toxic aromatic aldehyde for
K. marxianus. At low concentration (1 giL), the yeast exhibited a lag period
of 8 h, but after prolonged incubation (24 h), this compound was completely
reduced. However, at a higher concentration (2 giL), only 25% of the initial
4-hydroxybenzaldehyde was metabolized after 24 h of incubation.
The conversion of vanillin (15,19) and syringaldehyde (15) to their
corresponding alcohols by microorganisms has been previously observed.
By contrast, Nishikawa et al. (15) found that microbial metabolism of van-
illin and syringaldehyde led to the production of other compounds. In the
metabolism of vanillin, besides vanillyl alcohol, veratrole and several uni-
dentified self-condensation products were detected. Analogously, in the
metabolism of syringaldehyde a multitude of minor products, none domi-
nating, was observed.
Several potential mechanisms for the toxicity of aldehydes have been
suggested (11), including damage from chemical reactivity, direct inhibi-
tion of glycolysis and fermentation, and plasma membrane damage.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


152 Oliva et al.
Zaldivar et al. (11) suggested that the toxic effects on growth and fermen-
tation owing to aromatic aldehydes and HMF appear similar. Previous
studies (26) have suggested that the mechanisms for inhibition by these
compounds were cell damage and direct interference with biologic mem-
branes, which affect their ability to serve as selective barriers and enzyme
matrices. Furfural, however, appears to have a direct effect on either the
glycolytic or fermentative enzymes (11).
It is worthwhile to mention that K. marxianus CECT 10875, a strain
selected for resistance to temperature, also exhibits higher resistance to
aromatic compounds and HMF, but not to furfural, in comparison with
other microorganisms (11). Since there is evidence in the literature that
temperature can change fatty acid composition, the composition of plasma
membrane of the thermotolerant K. marxianus strain could be the basis for
its resistance to aromatic aldehydes and HMF.
The decrease in biomass formation when aliphatic acids were added
to the fermentation medium has been previously reported in S. cerevisiae
(27). Concentrations of acid that inhibited growth by 50% caused only
modest inhibition of ethanol production. However, we did not find in-
creased ethanol yield with the addition of a low concentration of acetic acid
to the fermentation medium. Differences in pH have a pronounced effect
on the toxicity of acids but not aldehydes or alcohols. The increase in or-
ganic acid toxicity at reduced pH is consistent with the mechanism ob-
served for other microorganisms: uncoupling and intercellular ion
accumulation (28).

Conclusion
Growth and alcoholic fermentation of glucose by K. marxianus CECT
10875 was significantly affected by the presence of biomass decomposition
products. The results showed that growth is more affected than ethanol
production. The degree of inhibition caused by the toxic compounds greatly
depended on the nature and concentration of inhibitor. At inhibitor con-
centrations found in the hydrolysate of steam-exploded poplar biomass, no
single inhibitor completely inhibited growth or fermentation.
4-H ydroxybenzaldehyde and catechol were the most powerful inhibi-
tors of growth and ethanol production. Many of the aldehydes were me-
tabolized by the microorganism to their corresponding alcohols, which
resulted in the removal of the toxic compounds and partial recovery of both
growth and ethanol production.
K. marxianus CECT 10875, a strain selected for resistance to tempera-
ture, exhibited resistance to aromatic compounds and HMF. The yeast
strain showed higher aldehyde assimilation rates in comparison with other
fermentation microorganisms.

References
1. Lynd, L.R.,Wyman, C.E. and Gerngross, T.V. (1999), Biotech. Prog. 15,777-793
2. Sun, Y. and Chen, J. (2002), Bioresour. Technol. 83,1-11.

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Degradation Compounds from Sf on Fermentation 153
3. Ballesteros, I., Oliva, J.M., Ballesteros, M. and Carrasco, J. (1993), Appl. Biochem.
Biotechnol. 39/40, 201-211.
4. Mason, W.H. (1929), U.S. patent no. 1,655,618.
5. Higuchi, T. (1988), in Biomass Handbook, Hall, C.W. and Kitani, 0., eds., Gordon &
Breach, New York, NY pp. 470-474.
6. Luo, c., Brink, D. 1., and Blanch, H. W. (2002), Biomass Bioenergy, 22, 125-138.
7. Ando, S., Arai, I, Kiyoto, K, and Hanai, S. (1986), J. Ferment. Technol. 64, 567-570.
8. Tran, A. V. and Chambers, R., P.(1985), Biotechnol. Lett. 7,841-846.
9. Jonsson,1. J., Palmqvist, E., Nilvebrant, N. 0., and Hahn-Hagerdal, B. (1998), Appl.
Microbiol. Biotechnol. 49, 691-697.
10. Carrasco, J.E., Martinez, J. M., Negro, M. J., Manero, J., Mazon, P., Saez, F., and
Martin, C. (1989), in 5th EC Conference on Biomass for Energy and Industry, vol. 2,Grassi,
G., Gosse, G., and Dos Santos, G., eds., Elsevier, Essex, England, UK, pp. 38-44.
11. Zaldivar, J., Martinez, A., and Ingram, 1. 0.(1999), Biotech. Bioeng. 65,24-33.
12. Zaldivar, J., and Ingram, 1., 0.(1999), Biotech. Bioeng. 66,203-210.
13. Zaldivar, J., Martinez, A., and Ingram, 1. 0.(1999), Biotech. Bioeng. 68,524-530.
14. Klinke, H. B., Thomsen, A. B., and Ahring, B. K (2001), Appl. Microbiol. Technol. 57,
631-638.
15. Nishikawa, N. K, Sutcliffe, R., and Saddler, J. N. (1988), Appl. Microbiol. Biotechnol. 27,
549-552.
16. Weigert, 8., Klein, K, Rizzi, M., Lauterbach, c., and Dellweg, H. (1988), Biotechnol.
Lett. 10,895-900.
17. Delgenes, J. P., Moletta, R., and Navarro, J. M.(1996), Enzyme Microbiol. Technol. 19,
220-225.
18. De Wulf, 0., Thornat, P., Gaignage, P., Marlier, M., Paris, A., and Paquot, M. (1986),
Biotechnol. Bioeng. Symp. 17,606-616.
19. Larsson, S., Quintana-Sainz, A., Reimann, A., Nilvebrant, N. 0., and Jonsson, 1.
J.(2000), Appl. Biochem. Biotechnol. 84-86,617-632.
20. Palmqvist, E., Grage, H., Meinander, N. Q., and Hahn-Hagerdal, B. (1999), Biotech.
Bioeng. 63, 46-55.
21. Villa, G. P. (1992), Acta Biotechnol. 12,509-512.
22. Taherzadeh, M., Gustafsson, 1., Niklasson, c., and Liden, G., (2000), J. Biosci. Bioeng.
87,169-174.
23. Diaz de Villegas, M.E, Villa, P., Guerra, M., Rodriguez, E., Redondo, D., and Martinez,
A. (1992), Acta Biotechnol. 12,351-354.
24. Palmqvist, E. (1998), PhD Thesis, Lund University, Sweden.
25. Morimoto, S., Hirashima, T., and Matutani, N. (1969), J. Ferment. Technol. 47,486-490.
26. Heipieper, H. J., Weber, F. J., Sikkema, J., Kewelo, H., and de Bont, J. A. M. (1994),
Trends Biotechnol. 12,409-415.
27. Taherzadeh, M. J., Niclasson, c., and Liden, G. (1997), Chem. Eng. Sci. 52,2653-2659.
28. Russel, J. B. (1992), J. Appl. Bacteriol. 73, 363-370.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0155/$20.00

Enzymatic Hydrolysis
of Ammonia-Treated Rice Straw

BETZABE SULBARAN-DE-FERRER, l MARl ELENA ARISTIGUIETA, l


BRuCE E. DALE,2 ALEXIS FERRER/,l
AND GRACIELA OJEDA-DE-RODRIGUEZ 1

1Lab. de Alimentos, Oept. de Qufmica, Facultad de Ciencias,

Universidad del Zulia, Av. Universidad, Grano de Oro. M6dulo 2,


Maracaibo, Venezuela, E-mail: aferrer1@cantv.net; and
20epartment of Chemical Engineering, Michigan State University,
2527 Engineering Building, East Lansing, MI 48824-1226

Abstract
Rice straw pretreated with liquid anhydrous ammonia was hydrolyzed
with cellulase, cellobiase, and hemicellulase. Ammonia-processing condi-
tions were 1.5 g of NH3/ g of dry matter, 85°C, and several sample moisture
contents. There were four ammonia addition time (min)-processing time
(min) combinations. Sugars produced were analyzed as reducing sugars
(dinitrosalicylic acid method) and by high-performance liquid chromato-
graphy. Monomeric sugars increased from 11 % in the nontreated rice straw
to 61% of theoretical in treated rice straw (79.2% conversion as reducing
sugars). Production of monosaccharides was greater at higher moisture
content and was processing time dependent. Glucose was the monosaccha-
ride produced in greater amounts, 56.0%, followed by xylose, arabinose,
and fructose, with 35.8, 6.6, and 1.4%, respectively.
Index Entries: Sugars; rice straw; ammonia treatment.

Introduction
Rice straw is one of the most abundant agricultural wastes in Califor-
nia and Central Venezuela, accounting for more than 1.5 and 0.9 million
t/yr, respectively (1,2). Brazil, Colombia, Peru, Mexico, and Argentina
also have considerable amounts of rice straw. Recently, rice has been
considered one of the four top crop priorities in Venezuela, which will
make the quantity of straw increase.

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 155 Vol. 105-108, 2003
156 Sulbarfm-de-Ferrer et al.
Rice straw has a high content of cellulose and hemicellulose (about
70%) with energetic values similar to those of corn. Unfortunately, these
carbohydrates have no value either as animal feeds, since they are hardly
digested by ruminants and cannot be digested by single gutted animals
and humans, or as feedstocks for sugars production, because of a very low
conversion if they are not chemically and physically pretreated. On the
other hand, rice straw is generally burned in situ after cropping; however,
several concerns about this practice have arisen in terms of air pollution,
and, therefore, it is necessary to look for other uses for the straw. Several
groups are trying to use rice straw as a feedstock for ethanol production
in California.
Several treatments such as acid-catalyzed steam explosion, acid
hydrolysis, and ammonia freeze-explosion (AFEX) have been used (1) to
increase the susceptibility of rice straw to enzymatic hydrolysis. The first
two treatments prod uced greater sugar yields. On the other hand, an ammo-
nia treatment (pressurization depressurization with ammonia) has been
applied to several lignocellulosic materials, such as dwarf elephant grass,
alfalfa, and florigraze rhizoma peanut (3-5), and has produced high sugar
yields (70-95%). Preliminary work on rice straw showed that high conver-
sions could be attained with ammonia treatments (6). Therefore, it may be
possible to convert the rice straw fibers into sugars by using the pressuriza-
tion depressurization with ammonia treatment and appropriate enzymes.
The sugars produced could be used as feeds and foods, as well as substrates
for fermentation processes to produce alcohols, organic acids and other
chemicals. The objective of the present study was to investigate the effect of
pressurization depressurization with ammonia treatment on the extent of
enzymatic hydrolysis and sugar profile of ammonia-treated rice straw.
Materials and Methods
Raw Material
Rice straw derived from a medium-grain rice crop from the northern
part of California's Central Valley was used.

Ammonia Processing
A laboratory-scale batch ammonia reactor unit consisting of a 4-L reac-
tor with appropriate support equipment was used for the treatment of rice
straw. A 15% moisture (wet weight basis [wwb]) rice straw was hammer
milled to nominally l-in. length and kept under refrigeration until used.
Water (adjusted to experimental level) and liquid anhydrous ammonia were
added to 80-g samples (dry matter [DM]), and the temperature was rapidly
raised to 85°C. After the treatment time, pressure was suddenly released
and the treated samples were collected and allowed to air-dry overnight.
Moisture contents are expressed in wet weight basis.
Ammonia treatment conditions were as follows: 15,35, and 60% mois-
ture content; 1.5 g of NH3/ g of DM; and four ammonia delivery time-dwell

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hydrolysis of Ammonia-Treated Rice Straw 157
Table 1
NDF Solubles and Hemicellulose Contents
of Untreated and Ammonia-Treated Rice Straw·
Sampleb Solubles (%DM) Hemicellulose (%DM)
Untreated 22.15 27.63
4-20-15 28.18 23.11
4-20-35 41.41 12.16
4-06-35 36.54 16.33
4-20-60 47.17 11.39
4-00-35 31.02 18.94
1-05-60 30.22 18.50
4-05-60 32.10 17.27
1-00-60 24.40 24.55
4-00-60 26.00 22.73
• Cellulose contents did not vary with treatment.
bNumbers refer to ammonia delivery time (min), dwell time (min),
and moisture content (%, wwb).

time combinations (1-0,4-0,4-5, and 4-20). Delivery time refers to the time
the required amount of ammonia takes to get into the reactor chamber.
Dwell time refers to the time allowed after the ammonia gets in the reactor
until decompression occurs, when ammonia is removed from the chamber.
Treatments and analyses were carried out in duplicate. Untreated rice straw
was used as a control for all experiments. The experimental conditions are
presented in Table l.
Fiber Fractionation Analysis
Neutral detergent fiber (NDF), acid detergent fiber, and acid deter-
gent lignin were determined in triplicate to estimate cellulose, hemicellu-
lose, and lignin by standard fiber analysis. Solubles were calculated as the
fraction solubilized by the NDF solution and includes sugars, oligosaccha-
rides, pectin, protein, and some minerals (7).
Enzymatic Hydrolysis
Enzymatic hydrolysis was performed with unwashed pretreated
solids using cellulase (Spezyme CP; Genencor, Rochester, NY), cellobiase
(Novozym 188; Novo N ordisk, Franklinton, NC), and hemicellulase (Biofeed
Plus CT; Novo Nordisk), at a solids loading of 5% (w Iv), in 500-mL Erlenm-
eyer flasks containing 100 mL of 0.05 M citrate buffer (pH 4.8) placed at 50°C
at 100 rpm for 48 h in an incubator shaker (Innova 4300; New Brunswick
Scientific, Edison, NJ). Sodium azide was added for preservation (0.15%).
The first experiment was set with enzyme dosages of 2 and 5 IU I g of DM
using the ammonia-treated sample with the lowest NDF to determine the
best dosage and hydrolysis time. Sugar production was measured as reduc-
ing sugars during 72 h with the dinitrosalicylic acid (DNS) method (8).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


158 Sulbaran-de-Ferrer et al.
Enzymatic hydrolysis of the rest of the samples was carried out at optimal
enzyme dosage and hydrolysis time, as described in Results and Discussion.
Sugars were measured both as reducing sugars and as individual sugars by
high-performance liquid chromatography (HPLC) (3) before and after enzy-
matic hydrolysis. Standard sugars for HPLC were sucrose, glucose, xylose,
arabinose, fructose, mannose, and galactose (Sigma, St. Louis, MO).
Soluble sugars and sugar composition of the rice straw fiber were
determined by acid hydrolysis and HPLC (3,9). Cellulose conversion to
sugars with respect to theoretical was determined based on initial glucose
content, whereas hemicellulose conversion was based on initial xylose +
arabinose content. Total sugar conversions were based on total initial
individual sugars determined by acid hydrolysis.
Results and Discussion
Solubles, cellulose, hemicellulose, lignin, and protein contents of
untreated rice straw used were 22.1, 45.2, 27.6, 5.1, and 5%, respectively
(NDF solubles include protein, so compositions total >100%). The ammo-
nia treatment increased solubles while decreasing hemicellulose in the
solid phase owing to solubilization (Table 1). Hemicellulose decreased by
58.8% in the 4-20-60 treatment; lignin decreased from 5.1 to 4.14% (18.8%).
Both factors contributed to enhancing the susceptibility of the fibers to
enzymatic hydrolysis, as occurred with dwarf elephant grass, alfalfa, and
florigraze rhizome peanut (3-5). Figure 1 shows that for solubles contents
lower than approx 32%, there is a linear relationship between reduction in
hemicellulose and increase in solubles. As the ammonia treatment condi-
tions improved, the increase in solubles was greater than the decrease in
hemicellulose, suggesting that other components become soluble.
The unwashed 4-20-60 treatment sample was used to investigate the
optimal enzyme loading and hydrolysis time; results are presented in Fig. 2.
After 24 h, the sugar production for an enzyme loading of 2 IV / g
of DM stopped, whereas for an enzyme loading of 5 IU / g of DM it still
increased by approx 8%. Sugar yield for 5 IU / g was higher than for 2 IV / g
(25.5%). This was similar to alfalfa (4). A grass such as dwarf elephant grass
and a legume such as florigraze rhizome peanut (3,5) required less enzyme
to produce even higher sugar yields in the case of the grass. Based on these
results, an enzyme loading of 5 IU / g ofDMfor a 48-hhydrolysis was selected
to determine the susceptibility of all treated and untreated samples to enzy-
matic hydrolysis. Initial sugar content in untreated rice straw, measured
after a 5-h wetting period, was lOA mg/ g of DM, which was mainly glucose
(90%) and galactose. Vlasenko et al. (1) found a lower value in AFEX-treated
rice straw, and xylose, arabinose, galactose, and mannose were also present,
although in very low amounts. Xylose was not detected in the treated
samples, indicating that hemicellulose was partly solubilized (likely owing
to production of oligomers) but not hydrolyzed to monomers during ammo-
niation, as has occurred with other ammonia-treated materials (3-5). The
acid hydrolysis of the untreated sample yielded (in mg of sugar / g of DM)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hydrolysis of Ammonia-Treated Rice Straw 159
30.-------------------------------~

25
:E
0
0~
20
ai'
en
.Q 15
:3
Qj
u 10
·E
Q)
::I: 5

O+-------------,-----------,-------------l
20 30 40 50
Solubles, % DM

Fig. 1. Relationship between hemicellulose and solubles contents of rice straw


samples.

250~----------------------------------------~

200

ui~
a;O 150
::1 __
0l0l
!/)CD
Ol!/)
.~ 8
(,,):3
:3-
'O0l 100
CDOl
a::E
-+-21U/g
50
_51U/g

0
0 20 40 60 80
Time, h

Fig. 2. Kinetics of enzymatic hydrolysis of rice straw measured as reducing sugars


for 2 and 5 IU / g of Spezyme CPo

376.2 for glucose, 164.6 for xylose, 35.5 for galactose, and 42.4 for arabinose,
making up a total of 618.7 mg/ g of DM.
Table 2 presents the results of the sugar yields after enzymatic
hydrolysis for all the unwashed treated samples and the untreated sample.
Reducing sugars were always higher than individual sugars determined
by HPLC, which indicates that there are dimers as well as oligomers.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


160 Su/baran-de-Ferrer et a/.

Table 2
Reducing (DNS) and Individual Sugars (HPLC)
from Enzymatic Hydrolysis of Untreated and Ammonia-Treated Rice Straw
Sugars (mg/ g DM)
HPLC·
Total DNS reducing
Sampleb G X Gal A F sugars sugars
Untreated 66.3 <0.4 2.8 <0.3 <0.4 69.1 82.4
4-20-15 90.4 63.1 <0.5 6.2 8.8 168.5 220.5
4-20-35 173.8 162.8 32.5 <0.3 8.5 377.6 482.8
4-06-35 152.5 138.9 <0.5 26.5 7.6 325.5 406.8
4-20-60 148.3 141.8 <0.5 27.5 7.1 324.7 450.5
4-00-35 126.0 68.5 1.7 14.2 6.9 217.3 231.7
1-05-60 177.4 102.4 <0.5 18.7 7.7 306.2 387.0
4-05-60 207.8 133.0 <0.5 24.4 5.9 371.1 489.8
1-00-60 150.0 39.9 <0.5 9.7 6.4 206.0 268.6
4-00-60 148.0 77.1 <0.5 16.7 7.9 249.7 302.3
aG, glucose; X, xylose; Gal, galactose; A, arabinose; F, fructose.
bNumbers refer to ammonia delivery time (min), dwell time (min), and moisture content
(%, wwb).

The difference rose to 25% for high sugar conversion treatments. The rela-
tionship between both determinations is linear as shown in Fig. 3. Total
reducing sugars increased from 82.4 mg/ g of DM (13.3% conversion) in the
untreated sample to 482.8 of mg/ g DM in one treated sample (78.0% con-
version), corresponding to a factor of 5.86, which shows the great efficiency
of the treatment. Similarly, a factor of 5.46 for total individual sugars was
found in the same treatment. Such treatment (4-20-35), which corresponds
to a delivery time of 4 min and a dwell time of 20 min for a 35% rice straw
moisture, was similar in sugar yield to a treatment (4-05-60) with the same
delivery time and a shorter dwell time (5 min) but with a higher rice straw
moisture (60%).
Complex interactions have been detected in several ammonia-treated
materials; generally, the severity of the treatment increases as moisture
content of the material increases-in other words, it appears that ammonia
needs water to be effective (3). In addition, since the dissolution is exother-
mal, biomass reaches a higher temperature than the reactor temperature,
therefore increasing the severity. The fact that the sugar yield obtained in
a treatment (1-05-60) with the same loading of ammonia, dwell time, rice
straw moisture content but with a shorter ammonia delivery time (1 min)
was 17.5% smaller (371.1 vs 306.2 mg/g of DM) appears to indicate that
time is needed for ammonia to dissolve and to make contact with the
material. Sugar production is linearly related to hemicellulose content
(reduced by solubilization) by the treatment, as Fig. 4 indicates, which
suggests that it can be used to predict sugar conversions from ammonia-
treated materials.

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


Hydrolysis of Ammonia-Treated Rice Straw 161
400~------------------------------.
~
c
CI
~ 300
~
~
~ 200
.2
CD
E R2 =0.9771
o
§ 100
E
S

~
0+----,r----.-----.----~----._--_4

o 100 200 300 400 500 600


Reducing sugars, mg/g DM

Fig. 3. Relationship between monomeric (HPLC) and reducing sugars produced by


enzymatic hydrolysis of untreated and ammonia-treated rice straw.

~ 450
c 400
CI
-a, 350
E
~ 300
~
:::l
250
III
.2 200
"-
CD
E 150
0
c: 100
0
E
S 50 •
~ 0
10 15 20 25 30
Hemicellulose, %DM

Fig. 4. Relationship between monomeric (HPLC) sugars produced by enzymatic


hydrolysis and hemicellulose content of untreated and ammonia-treated rice straw.

Glucose yield increased from 66.3 mg/ g of DM in the untreated rice


straw to 207.8 mg/ g of DM in the 4-05-60 treatment, or a 3.13 factor increase
(55.2% of theoretical). On the other hand, ammonia treatment allowed other
sugars such as xylose, arabinose, and fructose to appear. In the case of
galactose, it was only separated from hemicellulose with the longest treat-
ment (20 min). It is very interesting that this was true only at 35% moisture.
The ammonia treatment used was considerably very efficient for hemicel-
lulose hydrolysis (from negligible to 157.4 mg/ g of DM), showing a conver-
sion of 76.0% of theoretical. The maximum overall conversion based on

Applied Biochemistry and Biotechnology Vol. 105-10B, 2003


162 Sulbaran-de-Ferrer et al.

total analyzed individual sugars (4-20-35 treatment, 377.6 mg/ g of DM)


was 61 % of theoretical, whereas the one based on reducing sugars (4-05-60
treatment, 489.8 mg/ g of DM) was 79.2%. The conversion of the non treated
sample (69.1 mg/ g of DM) was just 11 % of theoretical.
In previous works on rice straw, Dale et al. (10) obtained 95% conversion
(based on reducing sugars) using an enzyme loading of 80 IU / g of DM on
AFEX-treated rice straw, Vlasenko et al. (1) obtained 48.7% with an enzyme
loading of 20 IU / g of DM, and Kaur et al. (11) achieved a maximum saccha-
rification of 54.3% (cellulose basis) with a 26 filter paper units (FPU) of cel-
lulose/ g of rice straw treated with 4% sodium hydroxide in combination
with 60 min of steam pressure. Therefore, the results obtained in our study
with relatively high sugar conversions obtained with a very low enzyme
loading and a 48-h hydrolysis time look very promising. The conversion that
we obtained is similar to that obtained with an acid process and greater than
that obtained with the Swan (acidified steam explosion) process (1).
The release of monosaccharides was moisture dependent, as can be
inferred when the 4-20-15, 4-20-35, and 4-20-60 treatments, which only
differ in rice straw moisture content, are compared. The sugar yield for the
higher moisture contents doubled that for 15% moisture.
Figures 5 and 6 show the effect of moisture as well as dwell time. For
35% moisture content, the optimal ammonia addition time-processing
time was 4-20 min, while for 60% moisture content it was 4-5 min and
produced the highest sugar yield. It can be seen that at 35% moisture, by
increasing dwell time above 6 min, sugar yield increases slightly. How-
ever, since galactose was released, the sugar obtained at this condition is
the maximum value for this moisture. A combination of high moisture
content (60%) and long processing time (4-20 min) decreased glucose
production, but it did not affect xylose production.
When enzymatic hydrolysis was applied to rice straw subjected to
the optimal ammonia treatment, 4-05-60 (owing to lower dwell time),
although similar to treatment 4-20-35, glucose was the monosaccharide
produced in greater amounts (56.0%), followed by xylose (35.8%), arabi-
nose (6.6%), and fructose (1.4%). Other monosaccharides were insignifi-
cantly produced. The sugars produced are fermentable and can be used
to produce a wide variety of chemicals. In addition, all are digestible by
animals, with only some limitations for pentoses in the case of feeding
single gutted animals, however, the sugar profile is suitable for animal
feeding. Glucose can be converted into fructose for sweetening, as it is
currently done in the corn industry.

Conclusion
Enzymatic hydrolysis of ammonia-treated rice straw produced readily
available material as monosaccharides useful for several purposes. The
sugar yield values that we obtained are among the greatest values found in
the literature. Sugar and glucose yields were moisture and dwell-time

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Hydrolysis of Ammonia-Treated Rice Straw 163

250

200

~ ]50
E::;;:
::EQ
Q/
;;:
100

Fig. 5. Effect of ammonia dwell time on sugars yield from rice straw (35% moisture)
for ammonia delivery time of 4 min. Unt, untreated; G, glucose, X, xylose; Gal, galac-
tose; A, arabinose; F, fructose.

250

200

~
E:=
::gQ
]50

Q/
;;
100

so

Fig. 6. Effect of ammonia dwell time on sugars yield from rice straw (60% moisture)
for ammonia delivery time of 4 min. Unt, untreated; G, glucose, X, xylose; Gal, galac-
tose; A, arabinose; F, fructose.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


164 Sulbaran-de-Ferrer et al.
dependent. Hexoses represent >50% of the sugars. Ammonia treatment
increased solubles and decreased hemicellulose content, which, in turn, are
correlated with the sugar production. The optimal treatment condition was
a 4-min ammonia delivery time, 5-min dwell time, 60% moisture content,
and 1.5 g of ammonia/ g of dry rice straw.

Acknowledgments
We gratefully acknowledge financial support from the Technological
Park of the University of Zulia (Maracaibo, Venezuela), Fonacit (Caracas,
Venezuela), and Fundacite-Zulia (Maracaibo, Venezuela.

References
1. Via senko, E. Y., Ding, H., Labavitch, J. M., and Shoemaker, S. P. (1997), Bioresour.
Technol. 59, 109-119.
2. MPC (2000), Ministerio de Producci6n y Comercio, Venezuela.
3. Ferrer, A., Byers, F. M., Sulbanin-de-Ferrer, B., Dale, B, E., and Aiello, C. (2000), Appl.
Biochem. Biotechnol. 84/86, 163-179.
4. Ferrer, A., Byers, F. M., Sulbaran-de-Ferrer, B., Dale, B, E., and Aiello, C. (2002), Appl.
Biochem. Biotechnol. 98-100, 123-134.
5. Ferrer, A., Byers, F. M., Sulbaran-de-Ferrer, B., Dale, B, E., and Aiello, C. (2002), Appl.
Biochem. Biotechnol. 98-100, 135-146.
6. Sulbaran-de-Ferrer, B., Ferrer, A., Byers, F. M., Dale, B, E., and Aristigueta, M. (1997),
Arch. Latinam. Prod. Anim. 5(Suppl. 1), 112-114.
7. Goering H. K. and Van Soest P. J. (1970), ARS-USDA, Handbook No. 379, The Agri-
cultural Research Service (ARS), Washington, DC.
8. Miller, G. L. (1959), Anal. Chem. 31,426-468.
9. Ehrman T. (1992), NREL Chemical Analysis and Testing Standards Procedure,
No. 002, National Renewable Energy Laboratory, Golden, CO.
10. Dale, B. E., Henk, L. L., and Shiang, M. (1985), Dev. Ind. Microbiol. 26,223-233.
11. Kaur, P. P., Arneja, J. S., and Singh, J. (1998), Bioresour. Technol. 66,267-269.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0165/$20.00

Effects of Temperature and Moisture


on Dilute-Acid Steam Explosion
Pretreatment of Corn Stover
and Cellulase Enzyme Digestibility

MELVIN P. TUCKER; KYOUNG H. KIM,


MILDRED M. NEWMAN, AND QUANG A. NGUYEN
National Renewable Energy Laboratory,
Biotechnology Division for Fuels and Chemicals,
National Bioenergy Center, 1617 Cole Boulevard, Golden, CO 80401,
E-mail: melvin_tucker@nrel.gov

Abstract
Com stover is emerging as a viable feedstock for producing bioethanol
from renewable resources. Dilute-acid pretreatment of com stover can solubi-
lize a significant portion of the hemicellulosic component and enhance the
enzymatic digestibility of the remaining cellulose for fermentation into etha-
noL In this study, dilute H 2SO4 pretreatment of com stover was performed
in a steam explosion reactor at160°C, 180°C, and 190°C, approx 1 wt % H 2S04,
and 70-s to 840-s residence times. The combined severity (Log10 [Rol-pH), an
expression relating pH, temperature, and residence time of pretreatment,
ranged from 1.8 to 2.4. Soluble xylose yields varied from 63 to 77% of theo-
retical from pretreatments of com stover at 160 and 180°C. However, yields
>90% of theoretical were found with dilute-acid pretreatments at 190°C. A
narrower range of higher combined severities was required for pretreatment
to obtain high soluble xylose yields when the moisture content of the acid-
impregnated feedstock was increased from 55 to 63 wt%. Simultaneous sac-
charification and fermentation (SSF) of washed solids from com stover
pretreated at 190°C, using an enzyme loading of 15 filter paper units (FPU) /
g of cellulose, gave ethanol yields in excess of 85%. Similar SSF ethanol yields
were found using washed solid residues from 160 and 180°C pretreatments
at similar combined severities but required a higher enzyme loading of
approx 25 FPU / g of cellulose.
Index Entries: Pretreatment; dilute-acid; acid hydrolysis; com stover;
enzymatic hydrolysis.

"Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 165 Vol. 105-108,2003
166 Tucker et al.

Introduction
Enzymatic utilization of the cellulose in lignocellulosic feedstocks
requires effective pretreatment to make the recalcitrant cellulose more
accessible to cellulase enzymes (1-8). Dilute-acid pretreatment can
improve enzyme accessibility to cellulose in pretreated lignocellulosic
feedstocks and solubilize a significant portion of the hemicellulosic com-
ponent under pretreatmenttemperatures ranging from 140 to 180°C (9,10).
In general, higher pretreatment temperatures and shorter reactor resi-
dence times result in higher soluble xylose recovery yields and enzymatic
cellulose digestibility (11).
A preliminary series of pretreatment experiments based on the litera-
ture was carried out using a 4-L steam explosion reactor at 160 and 180°C,
1% H 2S04, and 2 to 14-min reactor residence times. These conditions estab-
lished baseline total soluble xylose recovery yield for this reactor and could
be readily scaled up to the National Renewable Energy Laboratory's
(NREL's) Pilot Development Unit (PDU) vertical Sunds Hydrolyzer
(12,13). However, the maximum total xylose yield obtained in this series
of experiments was 77% of theoretical value, well below our goal of >85%.
In an attempt to increase soluble xylose recovery yield, the temperature
range was increased to 190°C with 1.0-1.1% (w/w) H 2S04 , and 70-s to
4-min reactor residence time. The combined severity factor (Log lO [RaJ -
pH) (14,15) for these higher temperature pretreatment conditions was in
the 1.9-2.2 range. These higher-temperature conditions were based on
modeling results reported in the literature, which suggest that 90% of the
corn stover xylan can be solubilized in 1 min, using H 2S04 concentrations
>0.9% (w /w) and pretreatment temperatures >190°C (9). In addition, the
effects of dewatering methods and moisture content of input corn stover
on pretreatment xylose yields were also investigated (16).
An important goal of pretreatment is to increase the enzymatic digest-
ibility of cellulose remaining in pretreated biomass residues. Higher-tem-
perature dilute-acid pretreatment has been shown to increase cellulose
digestibility of pretreated residues (8). The present study compares the
digestibility of cellulose remaining in corn stover residues pretreated at
160,180, and 190°C using a simultaneous saccharification and fermentation
(SSF) assay (17). The SSF assay was chosen over the standard enzymatic
saccharification assay because of the lower enzyme loading, reaction tem-
peratures, and removal of enzymatic end-product inhibition.

Materials and Methods


Preparation of Feedstock
Corn stover harvested from the Harlan, Iowa, area (B/MAP, Harlan,
IA) during two separate years (1997 and 1998) was coarsely ground at the
source and delivered to NREL. The particles in the separate ground batches
from each harvest year varied from fines to a fraction (approx 10 wt% of
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Dilute-Acid Steam Explosion of Corn Stover 167
input mass) containing particles as large as 2-6 in. (5-15 cm) long and 0.5-
0.8 in. (1.3-2 cm) thick. Typically, a tote (4 ft [1.2 m] on each side) containing
approx 70 kg of dry com stover (93 wt% solids content) from each harvest
year was washed in the NREL PDU wash tank (Sunds Defibrator, Sundsvall,
Sweden) with approx 1500 L of room-temperature city water for 45 min.
The wash tank consisted of an inverted truncated stainless steel cone (90 in.
[2.3 m] diameter x 80 in. [2 m] high) equipped with an agitator powered by
a 60-hp (45-kW) electric motor. The washed com stover was drained into
a stainless steel wire mesh basket overnight to approx 18 wt% solids con-
tent with a wet solids recovery of approx 276 kg (76% recovery of input dry
solids). The mesh size was approx 0.25 in. (6.4 mm), consisting of 0.125 in.
(3.2 mm) wire with an approximate opening size of 0.125 in. (3.2 mm). The
difference in nonrecoverable biomass lost in the washing step is attribut-
able to fines, dirt, rocks, tramp metal, and other material suspended or
dissolved in the wash water. The washed and drained com stover was air-
dried to approx 45 wt% solid content by spreading on a large rubber-coated
tarpaulin and blowing air across it with large fans for several days. The com
stover was mixed and spread with rakes approx once each hour during the
workday to promote even air-drying.
Acid Impregnation
Acid impregnation was carried out in batches of approx 20 kg in
Hastellof' C-276 wire mesh (100 mesh) baskets using a recirculating bath.
The bath recirculated 1% (w /w) H 2S04 (titrated at 0.98% using NIST trace-
able NaOH solution) at 50°C evenly through the baskets for 3 or 4 h. Each
batch was removed from the hot acid, covered with a plastic bag to mini-
mize evaporation, drained overnight, and mixed by coning and quartering
a minimum of three times. The combined batches were dewatered by press-
ing to approx 50 wt% solids using a hydraulic press (the cylinder mold
measured 25 em in diameter by 30.5 em high) at an internal pressure within
the mold of approx 600 psi. The removal of pressate was assisted byvaeuum
(approx 22 in. Hg). The pressed solid cakes were broken up by hand and
cone/ quarter mixed a minimum of three times prior to weighing the mixed
solids into sealed Ziplock® bags. Each bag to be loaded into the reactor
contained 600 ± 0.1 g of processed solids. Moisture loss during mixing
resulted in an increase in the acid concentration in the liquid within the com
stover from 0.98 to 1.067%. The biomass titration procedure consisted of
taking a random sample of 200 ± 0.1 g of the acid-impregnated, pressed, and
mixed feedstock, leaching the acid into 1000 ± 0.1 g of deionized water with
constant stirring overnight at 45°C. This was followed by titration with
NIST traceable standard NaOH solution G. T. Baker, Phillipsburg, NJ) in
triplicate 50- or 100-mL samples of the leachate.
A second large batch of washed com stover from the 1998 harvest was
soaked with hot 1.2% (w /w) H 2S04 solution as just described using the
recirculating bath. The final acid concentration within the com stover feed-
stock was determined to be 1.0% (w /w) by leaching and titration. The
Applied Biochemistry and Biotechnology Vol. 105-108,2003
168 Tucker et al.
difference in pH is attributable to neutralization by some of the ash present
in the feedstock. Two aliquots of the acid-soaked com stover from this
batch were pressed using the hydraulic press to produce a batch of 37 to 38
wt% solids content and a second batch of 47 wt% solids content.
In addition, two 55-kg batches of washed Harlan com stover (1997
harvest) were acid-impregnated with 0.36% (w /w) H 2S04 at 50°C for 4 h
using the recirculating bath. The acid-impregnated and drained com sto-
ver batches were air-dried to approx 46 wt% solids content. The com stover
was mixed every 30 min and the weight of the drying batch monitored until
the target evaporative weight loss was achieved. This gave an H 2S04 con-
centration inside the com stover particles of 1.1 % as measured by titration
using the leaching method described earlier.
Pretreatment
Pretreatment was carried out using a 4-L steam explosion reactor
(NREL Digester) as described earlier (18). The steam pressure was adjusted
to achieve the desired reaction temperature. Residence times were varied
for each experiment. Each pretreatment experiment used 600 g of input
feedstock. The pretreated solids blown from the reactor for each experi-
ment were collected separately in 55-gal nylon Hotfill® or polyethylene
drum liners ("bags") sealed in a cooled flash tank connected to the NREL
Digester. The bags were removed from the sealed flash tank following
either a single or second "shot," and then sealed and cooled to 4°C over-
night to condense any steam and volatile compounds (Le., furfural). The
pretreated materials were transferred from the sealed bags to sealed plastic
containers, and mixed, and 500-g random samples of each were pressed in
the hydraulic press at 600 psi for 8 min. The liquors (pressates) were ana-
lyzed for monomeric sugars, total sugars, and organic acids according to
established methods using high-performance liquid chromatography
(HPLC) (19). The pressed solid cakes were washed exhaustively by sus-
pending in hot water for 30 min and filtering on a Buchner funnel (five
times with 40°C tap water and once with 60°C deionized water) before
submitting samples for solids compositional analysis by outside laborato-
ries and near infrared (NIR) spectroscopy, and use in SSF enzymatic assays.
The results of the liquor and solids compositional analyses were used in a
mass balance spreadsheet to determine yield.
Cellulose Digestibility of Pretreated Corn Stover Using SSF
Enzymatic digestibility of the extensively washed pretreated solids
was determined using an NREL standard SSF assay (17), that measures the
ethanol yield from the fermentation of digested cellulose from the washed
residues. These assays help eliminate the end-product inhibition com-
monly associated with digestibility assays using enzymes alone. The
glucan value of pretreated washed solids was initially determined by a
rapid NIR method (20) for predicting the cellulose content of each residue
for adjustment of enzyme loading. Since the NIR method was in the early
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Dilute-Acid Steam Explosion of Corn Stover 169
stages of development at this time, the predicted cellulose contents were
later confirmed or adjusted using wet chemical analysis results. Aliquots
of the exhaustively washed pretreated solid residues were placed in 250-
mL baffled Erlenmeyer flasks with water bubble trap stoppers and steril-
ized at 121°C for 30 min. The sterilized washed solids were diluted to a
level of 6 wt% cellulose (approx 10 wt% insoluble solids) by adding sterile
components to give a final level: yeast extract (1 wt%), peptone (2 wt%),
and citrate buffer (50 mM, pH 5.2). Iogen (Ottawa, Canada) cellulase en-
zyme (lot no. BRC 191095) was added to give varying enzyme loading
levels of approx 5, IS, and 25 filter paper units (FPU) / g of cellulose based
on cellulose content in the residues.
A control flask without pretreated solids was used to determine etha-
nol produced from the media and enzyme components alone. The flasks
were inoculated with a cocktail containing both the enzyme level needed
and the yeast Saccharomyces cerevisiae DsA, at an initial optical density of 0.5.
Samples were taken at 7, 24, 48, 72, 120, 144, and 168 h. A YSI model 25
glucose analyzer (Yellow Springs Instruments, Yellow Springs, OH) was
used to monitor glucose concentrations in each timed sample. HPLC using
an Aminex HPX-87H column (Bio-Rad, Richmond, CA), operating at 65°C
and a flow rate of 0.6 mL/ min with 0.005 M H 2S04 as eluent, and a Hewlett-
Packard model 1047 refractive index detector was used to measure ethanol
and byproduct concentrations. Ethanol yields were calculated based on the
theoretical ethanol concentration from cellulose after subtracting out etha-
nol produced from the no-solids control flask containing only media and
enzyme. These SSF assays are expected to closely mimic the results achiev-
able in a larger-scale SSF fermentation with pretreated corn stover residues
at a level of approx 10 wt% solids and lower enzyme loading and fermen-
tation temperatures.

Results and Discussion


As shown in Fig. I, the highest total soluble xylose yields (>90%)
were achieved with experiments performed at 190°C with a combined
severity of approx 1.95. The gravimetric material balances for the pre-
treatment experiments at 190°C were between 94 and 100%. Therefore,
the high xylose yield at 96.3% is not owing to problems with mass balance
closure. Yields >90% were found at 190°C even though washed corn sto-
ver feedstocks from two different harvest years, 1997 and 1998, were uti-
lized in the pretreatment experiments several months apart. A drop in
xylose yield performance was experienced when the temperature was
reduced to either 180 or 160°C, even though similar ranges of pretreat-
ment severities were utilized. Material balances dropped from near 100 to
90% under pretreatment conditions with combined severities >2.2. This
suggests losses owing to production of volatile degradation components
under more severe pretreatment conditions. A maximum soluble xylose
yield of appro x 77% occured at 180°C at a combined severity of approx
2.05. The effects of dewatering methods can be seen in Fig. 1 by comparing
Applied Biochemistry and Biotechnology Vol. 105-108,2003
170 Tucker et al.
100

--
iii
u
;;
95
.160"C.I997.prcsscd

...0
Q)
90
+180°C,l997,prcsscd
X U§)OC,l998,airdricd
Q)
.c 85
e l90"C.l997.airdried
I-

--
.t. I9O"C.I 998.prcsscd
0~
80
~
Q)
>
0
75
u
Q)
II: 70
Q)
I/)
0 65
>.
><
Q) 60
:a;:,
"0 55
en
50
1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
Combined Severity (LoQ,o(Ro)-pH)

Fig. 1. Total soluble xylose yields from pretreatment experiments using acid-
impregnated corn stover at various temperatures, combined severities, harvest years,
and dewatering methods after acid soaking. (.) Pretreatment at 160°C using 1997
feedstock pressed to approx 50 wt% solids; ( • ) pretreatment at 180°C using 1997
feedstock pressed to approx 50 wt%; ( x ) pretreatment at 180°C using 1998 feedstock
partially air-dried to 47wt%; ( • ) pretreatments at 190°C using 1997 feedstock par-
tially air-dried to 46-wt%; (A) pretreatments at 190°C using 1998 feedstock pressed
to approx 47 wt% solids.

the two sets of experiments performed at 180 and 190°C, in which the
soluble xylose yields are comparable between pressed and partially air-
dried acid-impregnated feedstocks.
The effects of increased moisture in the acid-impregnated feedstock
on soluble xylose yields for pretreatments at 190°C are shown in Fig. 2. As
the solids content of the acid-impregnated feedstock decreased, the soluble
xylose yields at lower severities decreased. The effect is similar to decreas-
ing the pretreatment temperature to 180°C. However, as the severity of
pretreatment was increased, the soluble xylose yield increased above 90%
theoretical, then rapidly dropped off as the conditions became more
severe. The peak (combined severity approx 2.13) is rather abrupt, and the
optimal condition would be difficult to maintain in a large-scale pretreat-
ment reactor where a few seconds' change in either direction for residence
time would significantly alter yield. The much broader maximum associ-
ated with the drier, higher-solids-content acid-impregnated feedstock
would allow a large-scale pretreatment reactor to maintain conditions
necessary for maximum yield even though process conditions varied
somewhat during operation. The higher acid loading associated with
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Dilute-Acid Steam Explosion of Corn Stover 171
100.-------------------------------------------------~

--
iii 95
.
u
;:;
CD
0 90
CD
.c
.... 85
tf.
.....
3l
.! 80
>

CD
75
.2
>-
>< 70
.!!!
.Q
:::s
'0 65
en
60
60 70 80 90 100 110 120 130 140 150 160
Pretreatment Time (8)

Fig. 2. Effects of moisture level in acid-impregnated com stover feedstocks on pre-


treatments at190°C. ( • ) Pretreatments using feedstock harvested in 1998 and pressed
to 37-38 wt% solids content; ( • ) pretreatments using feedstocks harvested in 1998 and
pressed to 47 wt% solids content.

wetter input feedstocks (g acid/ g dry biomass) is not beneficial in the


pretreatment, since the higher liquid content slows heating of the biomass
particles as the result of higher heat capacity, resulting in lower soluble
hemicellulose yields. The higher acid loading is less desirable because it
would require more neutralizing base (lime) in the overall process, pro-
ducing more gypsum, and increase downstream separation and disposal
costs. The effects of going to lower solids in the reactor (i.e., 10-25%) are
being investigated.
Table 1 summarizes key pretreatment conditions harvest year time,
temperature, acid concentration, solids content going into the reactor,
and calculated combined severities and key results (pH, solids content,
and total soluble xylose yields) from each pretreatment. The measured
pH values listed were used in the calculation for combined severity (15)
reported in this table.
Figure 3 shows the effects of pretreatment temperature on the SSF
ethanol yields on four exhaustively washed solid residues from pretreat-
ment experiments carried out at 160,180, and 190°C for 8 and 14 min, 2 min
and 90 s, respectively. Essentially, 70-92% of the cellulose was converted to
ethanol. The enzymatic digestibility of com stover residues pretreated at
190° was higher than at 160 or 180°C under these assay conditions using an
enzyme loading of approx 25 FPU / g of cellulose, 32°C, approx 6 wt% cel-
lulose loading, and the yeast S. cerevisiae.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
:l> Table 1
:g
~
Summary of Corn Stover Pretreatment Conditions and Results
Cl..
t:xl
o· Input feedstock Pretreatment conditions After pretreatment
(")
:;)-
Cb Total soluble
3 Solids Acid Solids
i;;' Run Harvest Dewatering content conc. Temperature Time Combined content pH xylose yield
~ no. year method (wt%) (% [w/w]) (0C) (s) severity (wt%) of liquor (% theoretical)
OJ
~
Cl..
t:xl 1 1997 Pressing 50 1.07 160 480 1.84 38.3 1.37 68.5
o· 2 1997 Pressing 50 1.07 160 660 1.98 36.5 1.44 66.2
<ii
(")
:;)- 3 1997 Pressing 50 1.07 160 840 2.08 35.7 1.49 70.2
~
c 4 1997 Pressing 50 1.07 180 120 1.81 27.1 1.56 70.8
~
'<: 5 1997 Pressing 50 1.07 180 180 1.98 37.4 1.87 76.5
6 1997 Pressing 50 1.07 180 240 2.11 36.6 1.90 75.4
7 1997 Pressing 50 1.07 180 360 2.28 32.9 1.93 65.6
'I 8 1997 Pressing 50 1.07 180 480 2.41 33.0 1.95 63.3
I'-.)
9 1997 Drying 46 1.06 190 70 1.86 25.2 1.1 85.4
10 1997 Drying 46 1.06 190 85 1.95 26.9 1.1 92.9
11 1997 Drying 46 1.06 190 90 1.97 25.5 1.0 96.3
12 1997 Drying 46 1.06 190 95 2.00 25.5 1.1 89.1
13 1997 Drying 46 1.06 190 110 2.06 27.6 1.0 79.4
14 1998 Pressing 47 1.0 190 70 1.84 26.9 1.2 91.5
15 1998 Pressing 47 1.0 190 90 1.95 29.1 1.3 95.2
16 1998 Pressing 47 1.0 190 110 2.04 27.1 1.3 94.0
17 1998 Pressing 47 1.0 190 130 2.11 25.5 1.4 89.7
c;: 18 1998 Pressing 38 1.0 190 70 1.86 18.4 1.3 79.0
:-
19 1998 Pressing 38 1.0 190 90 1.97 20.8 1.1 78.9
0
<..n 20 1998 Pressing 38 1.0 190 90 1.97 25.3 1.0 78.7
I
0 21 1998 Pressing 38 1.0 190 110 2.05 23.6 1.1 84.5
,Co
to."
22 1998 Pressing 37 1.0 190 130 2.13 22.3 1.2 93.5
0 23 1998 Pressing 37 1.0 190 150 2.09 29.2 1.1 72.8
0
w
Dilute-Acid Steam Explosion of Corn Stover 173

100

--
(ij
0
90

80
~~-----------------.------ ~~
...
~
Qj

0
Qj 70
~
~
;,.!! 60
l?....
"C
Qj 50
:;:
'0 40
C
C'II
~ 30
W __ I .2m,n.25FPl..g ~ 16O'C. m,n.25FPl g
L&.. 20
en
en
10

0
0 20 40 60 80 100 120 140 160 180
Time (h)

Fig. 3. Effects of pretreatment temperature on SSF ethanol yield from exhaustively


washed pretreated residues. Pretreatment was carried out in a 4-L steam explosion
reactor with approx 1.1 wt% H 2S04 ,

The effects of enzyme loading on initial rates and final extent of reac-
tion for converting cellulose to ethanol from corn stover pretreated at 160°C
are shown in Fig. 4. Pretreatment of corn stover was carried out in a 4-L
steam explosion reactor at 160°C for 14 min, with 1.07% H 2S04 , The SSF
assays were carried out with S. cerevisiae DsA yeast with enzyme loadings
of approx 5,15, and 25 FPU / g of cellulose. Solids were loaded to give final
cellulose content of 6 wt% based on NIR analysis of the 160°C pretreated
residue.
Figure 5 shows the effects of enzyme loading on ethanol yields in SSF
fermentations at 32°C for an exhaustively washed residue resulting from
pretreatment of corn stover at 190°C. The cellulose loading was adjusted to
give 6 wt% based on wet chemical analysis of the residue. Ethanol yield is
based on the total expected from the amount of cellulose loaded into each
fermentation flask.
The preceding results show that >90% soluble xylose recovery yield
and >90% SSF cellulose digestibility can be obtained from corn stover using
dilute-acid steam explosion pretreatment at 1% H 2S04 (before steam addi-
tion), 190°, and short residence times (90-11 0 s). Lowering the pretrea tment
temperature to 160 or 180°C, and increasing the residence times to obtain
a combined severity factor similar to 190° pretreatments, resulted in lower
xylose recovery yield and lower cellulose digestibility.
There are several possible causes for the lower yields at lower pre-
treatment temperatures. First, high temperature and short residence times

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


174 Tucker et al.
lUU

-
'ii
u
:;:
90

e0 80
Q) 70
or.

-
~
#. 60
'tJ
a; 50
>= 40
'0
..
c
as 30
or. _5FPU/g
W 20
LL. ___ 15FPU/g
en
en 10 --.- 25 FPU/ g

0
0 50 100 150 200
Time (h)

Fig. 4. Effects of enzyme loading on initial rates and final extent of reaction for
converting cellulose to ethanol. Pretreatment of corn stover was carried out in a 4-L
steam explosion reactor at 160°C for 14 min with 1.07% H 2S04 •

100

.-.. 90
iij
U 80
:;...
0 70
GI
J::.
I-
60
.....
~
0

"C 50
Gi
:; 40
0c

- 30
CIS
J::.
W
u.. 20
U)
U)
10

0
0 20 40 60 80 100 120 140 160 180
Time (h)

Fig. 5. Effects of enzyme loading on initial rates and final extent of reaction of
converting cellulose to ethanol. Pretreatment of corn stover was carried out in a 4-L
steam explosion reactor at 190°C for 90 s with 1.06% H 2S04 •

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Dilute-Acid Steam Explosion of Corn Stover 175

favor maximum xylose recovery yields because of differences in activation


energy for xylan hydrolysis and xylose degradation reactions. The activa-
tion energy of xylan hydrolysis is slightly higher than that of xylose degra-
dation into intermediate products, thus, raising reaction temperatures
increases the potential xylose yield during the initial reaction time when
xylose concentrations are low. Second, corn stalks contain approx 28%
nodes, which have higher density and lignin content (20% lignin in nodes
compared with 16.7% in whole stalks, 14.6% in leaves, and 13.3% in pith)
than nearby tissues (21). The high-lignin content and dense nodes may
require higher temperature to achieve sufficient breakdown of the lignin-
carbohydrate complex and increase xylose yields.
The yield of enzymatic digestion of cellulose in pretreated biomass
has been reported to increase with removal of xylan. Higher-temperature
pretreatments giving the same extent of xylan removal as lower-tempera-
ture pretreatments have been reported to give higher cellulose digestibility
(1). The lower xylan removal from the 160 and 180°C pretreated samples
was a result of incomplete hydrolysis. Therefore, one would expect lower
cellulose digestibility.
The effect of starting total solids content in the range studied (37-47%)
did not appear to have a significant impact on peak xylose yield at 190°C.
Although the peak xylose yield is about the same as that obtained from
drier material, the residence time for obtaining high xylose yield for wet
corn stover shifts from about 90 to about 130 s (38% solids content) (Fig. 2).
The longer residence time is presumably required because of greater heat
capacity and slower heat transfer throughout the wetter biomass. The
higher-solids feedstocks appear to have a broader range of pretreatment
reactor residence time for achieving maximum xylose yields and thus are
more desirable from a process control viewpoint. Furthermore, lower-sol-
ids feedstocks lead to higher steam requirements (for pretreatment and
product recovery) and possibly larger downstream equipment to process
more dilute streams.
Unlike pretreatment of softwood, pressing with the hydraulic press
before pretreatment did not seem to cause any negative effects (compared
to air-drying) on sugar yield from corn stover. This is probably because of
the spongy and elastic characteristics of corn stover structure (21).
The SSF cellulose digestibility assays (using an enzyme loading of
approx 25 FPU / g of cellulose) of 160,180, and 190°C pretreated corn stover
materials were in the 70, 80, and 90% ranges, respectively. Lowering the
cellulase loading to 15 FPU / g of cellulose lowered the SSF cellulose digest-
ibility to 65 and >85% for washed pretreated residues from 160 and 190°C,
respectively. The SSF conditions of enzyme loading, temperature, and yeast
fermentation are more representative of large-scale process conditions than
those found in the typical enzyme digestibility assays in which strong feed-
back inhibition occurs and high enzyme loadings are typically used.
Acid-catalyzed steam explosion pretreatment of corn stover (pre-
impregnated with 1% H 2S04) was shown to produce digestible residues
Applied Biochemistry and Biotechnology Vol. 105-108,2003
176 Tucker et al.
and solubilize significant amounts of the hemicellulosic fraction. Soluble
xylose yields varied from 63 to 77% of theoretical from pretreatments of
corn stover at 160 and 180°C. However, soluble xylose yields >90% of theo-
retical were found with dilute-acid pretreatments at 190°C. This suggests
that a transition occurs between 180 and 190°C for dilute-acid steam explo-
sion pretreatment of corn stover for both hemicellulose hydrolysis and
effects on the enzymatic digestibity of cellulose remaining in the residue.
A similar transition was found for dilute-acid steam explosion pretreat-
ment of softwood forest thinnings (15).

Conclusion
Acid-catalyzed steam explosion pretreatment of corn stover (preim-
pregnated with 1% H 2S04) produced digestible residues and solubilized
significant amounts of the hemicellulosic fraction. Soluble xylose yields
varied from 63 to 77% of theoretical from pretreatments of corn stover at
160 and 180°C. However, soluble xylose yields >90% of theoretical were
found with dilute-acid pretreatments at 190°C.
The effect of starting total solids content in the range studied (37-47%)
did not have a significant impact on peak xylose yield at 190°C; however,
the peak xylose yield shifted to longer residence times, from about 90 to
about 130 s. The longer residence time is presumably required because of
greater heat capacity and slower heat transfer throughout the wetter bio-
mass. The higher-solids feedstocks had a broader range of pretreatment
reactor residence time for achieving maximum xylose yields and thus are
more desirable from a process control viewpoint.
SSF of washed solids from corn stover pretreated at 190°C, using an
enzyme loading of 15 FPU / g of cellulose, gave ethanol yields in excess of
85%. Similar SSF ethanol yields were found using washed solid residues
from 160 and 180°C pretreatments at similar combined severities but
required a higher enzyme loading of approx 25 FPU / g of cellulose.

Acknowledgment
This work was funded by the US Department of Energy, Office of
Fuels Development.

References
1. Grohmann, K, Himmel, M., Rivard, c., Tucker, M., Baker, J., Torget, R., and Graboski,
M. (1984), Biotechnol. Bioeng. Symp. 14, 139-157.
2. Lin, K W., Ladisch, M. R., Voloch, M., Patterson, J. A., and Noller, C. H. (1985),
Biotechnol. Bioeng. 27,1427-1433.
3. Weimer, P. J., Chou, Y. C. T., Weston, W. M., and Chase, D. B. (1986), Biotechnol.
Bioeng. Symp. 17,5-18.
4. Rolz, c., de Arriola, M. c., Valladares, J., and de Cabrera, S. (1987), Process Biochem.
22,17-23.
5. Torget, R., Werdene, P., Himmel, M., and Grohmann, K, (1990), Appl. Biochem.
Biotechnol. 24125, 115-126.

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Dilute-Acid Steam Explosion of Corn Stover 177
6. Grethlein, H. E. and Converse, A. O. (1991), Bioresour.e Technol. 36, 77-82.
7. Chang, V. S. and Holtzapple, M. T. (2000), Appl. Biochem. Biotech. 84-86,5-37.
8. Torget, R.; Himmel, M. E., and Grohmann, K. (1992), Appl. Biochem. Biotech. 34135,
115-123.
9. Esteghlalian, A., Hashimoto, A. G., Fenske, J. J., and Penner, M. H. (1997), Bioresour.
Technol. 59, 129-136.
10. Schell, D. J., Walter, P. L and Johnson, D. K. (1992), Appl. Biochem. Biotechnol. 34135,
659-665.
11. Torget, R., Walter, P., Himmel, M. E., and Grohmann, K. (1991), Appl. Biochem.
Biotechnol. 28/29,75-86.
12. Tucker, M. P., Farmer,J. D., Keller, F. A., Schell, D. J., and Nguyen,Q. A. (1998), Appl.
Biochem. Biotech. 70-72,25-35.
13. Nguyen, Q. A., Dickow, J. H., Duff, B. W., Farmer, J. D., Glassner, D. A., and Ibsen,
K. N. (1996), Bioresour. Technol. 59, 189-196.
14. Chum, H. L., Johnson, D. K., Black, S. K., and Overend, R P. (1990), Appl. Biochem.
Biotechnol. 24125, 1-14.
15. Nguyen Q. A., Tucker, M. P., Keller, F. A., and Eddy, F. P. (2000), Appl. Biochem.
Biotech. 84-86,561-570.
16. Kim, K. H., Tucker, M. P., and Nguyen, Q. A. (2002), Biotechnol. Prog. 18,489-494.
17. Dowe, N. and McMillan, J. (2001), NREL Analytical Procedure, No. 008, National
Renewable Energy Laboratory, Golden, CO; (Website: http://www.ott.doe.gov /
biofuels/ analyticaCmethods.htm1#LAP-002).
18. Nguyen, Q. A., Tucker, M. P., Boynton, B., Keller, F. A., and Schell, D. J. (1998), Appl.
Biochem. Biotechnol. 70-72,77-87.
19. Ruiz, R and Ehrman, T. (1996), NREL Analytical Procedure, No. 014, National Renew-
able Energy Laboratory, Golden, CO. (Website: http://www.ott.doe.gov /biofuels/
analyticaCmethods.htm1#LAP-014)
20. Hames, B. R, Thomas, S. R, Sluiter, A. D., Roth, C. J., and Templeton, D.W. (2003),
Appl. Biochem. Biotechnol. 105-108,5-16.
21. Hamilton, F. and Leopold, B., eds. (1993), in Pulp and Paper Manufacture, vol. 3, Joint
Textbook Committee of the Paper Industry of the United States and Canada, Atlanta,
p.119.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0179/$20.00

Sugar Monomer and Oligomer Solubility


Data and Predictions for Application to Biomass Hydrolysis

MATTHEW C. GRAY, ALVIN O. CONVERSE,


AND CHARLES E. WYMAN*
Thayer School of Engineering, Dartmouth College,
800 Cumming Road, Hanover, NH 03755,
E-mail: charles.e. wyman@dartmouth.edu

Abstract
Oligomer solubility could potentially play an important role in control-
ling the rates and yields in the thermochemical hydrolysis of hemicellulose
as a pretreatment for subsequent enzymatic conversion of cellulose. How-
ever, limited data or models are available to describe the aqueous solubility
of sugar monomers and oligomers. In this work, we measured the solubili-
ties of sugars common to many biomass feedstocks in the temperature
range of 2S-30°C. Then we reviewed solubility models for sugars from the
open literature. Finally, we applied models to test their ability to describe
this and other data reported in the literature. It was found that the solubil-
ity of sugar monomers was not well described by the ideal solubility law
or other more complex models. However, with an empirical adjustment to
the enthalpy of fusion, the ideal solubility law was able to approximately
predict the solubility of cello-oligomers. Based on these results, solubilities
for low molecular weight xylo-oligomers are predicted to investigate
their possible importance in pretreatment and define further experimental
measurements needed to improve our understanding of sugar and oligo-
mer solubility.
Index Entries: Hydrolysis; oligomers; pretreatment; solubility; sugars.

Introduction
Lignocellulosic biomass has the potential to become a valuable raw
material for the production of fuels and chemicals provided efficient and
economical means are developed to convert them into marketable prod-
ucts. Biological processing offers a particularly promising path to realize
such costs, and impressive improvements have been made (1). However,
further reductions in the cost of pretreatment and biological conversion of

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 179 Vol. 105-108,2003


180 Gray et al.
cellulose are essential to achieve this end. Elucidation of the fundamental
mechanisms and the development of accurate predictive models would aid
in identifying opportunities for significant advancements.
Hemicellulose hydrolysis, in which long chains of hemicellulose are
depolymerized into oligomers and monomers, is often favored for the
preparation of biomass for enzymatic cellulose conversion. This process is
typically modeled as a first-order homogeneous reaction in which the poly-
mer reacts to form monomers directly (2). Such models neglect the true
heterogeneous nature of the biomass / water pretreatment system. Further-
more, they suffer from inaccuracies and do not provide the insight needed
to rationalize the next generation of technology that will be competitive in
the marketplace, or the confidence to support commercial applications now
or with more advanced approaches in the future.
Our group postulates that hemicellulose hydrolysis may be limited
by the rate of mass transfer and solubility of the oligomers released from
the solid. For example, if oligomers are only marginally soluble, oligomers
of long-chain length would be unable to dissolve until those in solution are
reacted to smaller units. This could explain some of the differences in
yields realized in different reactor configurations. Knowledge of the solu-
bility of oligomers would give researchers a tool to compare different
configurations and would explain their theoretical limitations. However,
to test this mechanism, data on the solubility of the five sugars in hemicel-
lulose and their important oligomers are needed.
Some solubility information was found in the literature, but the data
and predictive models were limited in the range of sugars and oligomers
considered and sometimes contradictory. Jackson et al. (3) measured the
solubility of a-(D)-glucose at 0.5-S0 C, but the values differ from data
D

reported by Taylor (4) in the temperature range of 20-65°C. Young (5)


reported more values for the solubility of a-(D)-glucose, glucose monohy-
drate, and ~-D-glucose in the temperature range of -17 to 63°C; his data
agreed with Jackson et al. (3) data. Gabas et al. (6) reported solubilites
for mannose and xylose at 25°C. More recently, Jacobsen (7) measured
solubilities of glucose, xylose, and cellobiose in the temperature range of
25-47°C and found agreement with Taylor's (4) data for glucose and cel-
lobiose. A number of researchers obtained data on the solubility of sugar
mixtures and described their data with quasi-chemical models such as
Universal Quasi-Chemical Model (UNIQUAC), Uniquac Functional-
Group Activities Coefficients Model (UNIFAC), the Flory-Huggins
model, and the Entropic Free-Volume (8-15).
Fewer data are available in the literature for oligomers than for mono-
mers. Taylor (4) performed a systematic study on the solubility of glu-
cose, cellobiose, cellotriose, cellotetraose, and cellopentaose in the
temperature range of 25-65°C and showed that solubilities drop off with
increasing chain length, asone would expect. However, only an empirical
fit is provided to describe his data; he does not present any experimental
solubility data points or experimental SDs.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Sugar Monomer and Oligomer Solubility 181

To develop a solubility and mass transfer model for hemicellulose


hydrolysis, more information is needed on the solubility of monomers and
particularly oligomers released in hydrolysis. Furthermore, it would be
very valuable to be able to predict solubility at elevated temperatures typi-
cal for pretreatment by hemicellulose hydrolysis (1) because it would help
to improve the efficiency of accessing such information and (2) because
sugars would degrade during high-temperature solubility measurements.
Thus, we initially focused on measuring the solubility of monomers present
in biomass (arabinose, galactose, glucose, mannose, and xylose) and of
cellobiose, the only oligomer of interest available at a reasonable cost, to
provide a platform for evaluating leading models to determine how well
they could predict the solubility data. We then applied the models to esti-
mate the solubility of xylo-oligomers to evaluate whether solubility might
play an important role in pretreatment by hemicellulose hydrolysis and to
help define further data needs.

Materials and Methods


Chemicals
(o-)-arabinose, (D+ )-cellobiose, (D+ )-galactose, (D+ )-glucose, (D+)-
mannose, and (D+ )-xylose were purchased from Sigma-Aldrich (St. Louis,
MO). High-performance liquid chromatography (HPLC)-gradewater from
Fisher (Pittsburg, PA) was also used.
Determination of Solubility
Each of the sugars was mixed with deionized water in a 60-mL serum
bottle (Fisher) in a ratio of about 2.5 wt% sugar in excess of the expected
solubility (based on data in the literature or previous experimental data).
The bottles were sealed with 20-mm stoppers (Fisher) and 20-mm tear-off
aluminum crimp seals (Fisher). They were then fixed to a 12-in.-diameter,
4-in.-wide plastic wheel containing a 1 3/ 4-in.-deep groove. Up to nine
bottles were fastened on to each side of the wheel using plastic ties. The
wheel was mounted on a steel structure and connected by a chain to a
Dayton DC gear motor (Niles, IL) operating at 50 rpm. The wheel was then
immersed in a 12 x 24 x 20.5 in. water bath containing an Isotemp 2100
(Fisher) Immersion Circulator providing a temperature stability of ± 0.1 °C
and a pumping rate of 14 L/min.
To take samples, the motor was stopped periodically and the bottles
were removed. After removing the caps, a sample was extracted with a
3-mL Luer-Lock, Beckton Dickinson syringe (Franklin Lakes, NJ). A 25-mm
O.5-~m Millipore filter (Bedford, MA) was then fastened to the end, and the
liquor was pushed out onto a VWR aluminum weigh dish and quickly
weighed on an OHAUS AS120 balance (Pinebrook, NJ) (repeatability of
0.1 mg) and diluted. Concentrations were measured on either of two sys-
tems: a Waters Separations Module 2695 (Milford, MA) using an Aminex
HPX-87P column (Bio-Rad, Hercules, CA) with a Waters 2414 Refractome-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


182 Gray et al.
ter; or a Waters 717 autosampler, Aminex HPX-42A ion-exchange column,
and a Waters 410 refractometer. An unpaired t-test was performed to
determine when the change in concentration with time was undetectable.

Models
Ideal Solubility Model
The simplest method for predicting the solubility of carbohydrates is
to use the ideal solubility law, the thermodynamic derivation of which is
relatively straightforward (16). For a component to be ideal, no solvent can
appear in the solid phase(B) and the activity coefficient must be unity (17).
This implies that the solute does not form a hydrate at the given tempera-
ture, that the affinity between solute molecules is approximately the same
as the affinity between solute and solvent molecules, and that the solute
and solvent have similar molecular volumes (16). With these assumptions,
dissolution is thermodynamically equivalent to melting the solute, and the
change in free energy of dissolution (6.G dis ) is equated to the change in free
energy on melting (6.Gf ) at the dissolution temperature. By assuming that
there is no change in entropy, 6.Gf is equal to the change in enthalpy on
melting (6.Hf ) (16). The resulting expression is as follows:

-b.Hf(Tm-T) b.Cp(Tm-T) p
In(X )=---- + - - - -b.C
-In(Tm)
- (1)
R TmT R T R T
Equation 1 is the first form of the ideal solubility law, in which X is the
mole fraction of solute in solution at saturation, T is the absolute tempera-
ture, Tm is the melting point of the solute, R is the ideal gas constant, and 6.Cp
is the heat capacity difference of the solute between pure solid and a
subcooled liquid at the dissolution temperature. However, because the
solute is thermodynamically stable only as a solid at the dissolution tem-
perature, 6.Cp is difficult, and in some cases impossible, to determine. Thus,
in the two most common forms of the ideal solubility law, an assumption
about 6.Cp is necessary. The first assumption is that it can be set to zero,
leading to (16)

(2)

A third equation is derived from empirical observation that the heat


capacity difference 6.Cp can be better estimated by the entropy of fusion,
6.5f . Since 6.Gf =0 at the melting point and 6.Hf = Tm6.5f =Tm6.Cp' the first
two terms in Eq. 1 can be eliminated, leading to (16)

In(X) = - b.HfIn
RTm
(TTm ) (3)

To apply the ideal solubility laws to oligomers, the enthalpy of fusion


must be known. Unfortunately, a thorough search of the literature reveals
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Sugar Monomer and Oligomer Solubility 183

values for only monomers and a few dimers. Thus, it is necessary to esti-
mate these values for oligomers. The enthalpy of fusion at the melting
point temperature can be estimated by the entropy of fusion from the
relationl::.Hf =Tml::.Sf. Walden (18) observed that l::.Sfisabout 13cal/(K·mol)
(54 J/(K·mol) for a large number of organic compounds. However, the
values for l::.Hfcalculated using Walden's value for l::.Sfare much less than
the experimental values reported in the literature, for the monomers and
dimers for which the fusion enthalpies are tabulated. Thus, l::.Sfs were cal-
culated from the experimentall::.Hfs (see Table 1) and found to vary quite
widely. However, note that the values do not vary appreciably between
monomer and dimer.
With this in mind, values of l::.Sf that should not depend on chain
length were bracketed between 60 and 100 J / (K-mol) and used in the ideal
solubility law. For both the cello-oligomers and the xylo-oligomers, a mini-
mum and maximum solubility was calculated using these two values for
the fusion entropies, and these predictions were compared to experimental
values available in the literature.

UNIQUAC/UNIFAC
Activities are often applied instead of concentrations to compensate
for deviations from ideality in the liquid phase. Several methods have
been devised to predict the activity that incorporate a combinatorial term
(also called an athermal term), y f, that accounts for entropic effects arising
from differences in size between solute and solvent molecules, and/ or a
residual term, Yf, that accounts for energetic interactions such as Coulom-
bic forces and hydrogen bonding that are very temperature dependent
(18). UNIQUACand UNIFACaretwosuchmodels. They differinthe way
in which the residual term is calculated (19). Both methods treat the dif-
ferential heat capacity as a linear function of temperature. UNIQUAC
requires between five and six parameters for each component aside from
water, and UNIFAC requires more parameters, depending on the num-
ber of functional groups.
Peres and Macedo (8) applied UNIQUAC to determine a wide variety
of thermodynamic parameters for binary systems of D-glucose, D-fructose,
and sucrose in water. The same group later compared the UNIQUAC pre-
dictions with those obtained from the Flory-Huggins and entropic free-
volume models (9-11). Gabas and Laguerie (12) applied UNIFAC to
describe the solid-liquid-phase equilibria of the ternary system xylose, man-
nose, and water. Likewise, Abed et al. (13) used UNIFAC to describe the
phase equilibria at saturation of mixtures of water, sucrose, and glucose
along with water, sucrose, and fructose. Catte et al. (14), and Spiliotis and
Tassios (15) each created their own UNIFAC models using data from the
literature. None of these groups examined the solubility of a binary, sugar
monomer, or oligomer/water system over a wide temperature range,
although many of them included data from other researchers.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


184 Gray et al.
Table 1
Experimental Fusion Enthalpies and Calculated Fusion Entropies
with Experimental Melting Points and Fusion Enthalpy Values
Tm Fusion enthalpy Fusion entropy
Sugar (OC) (J/mol) (J/K'mol)

Arabinose 160a 35700- 82


Galactose 170a 43740- 99
Glucose 158a 32220- 75
Mannose 134a 24660- 61
Xylose 157a 31650- 74
Fructose 105a 30420a 80
Ribose 70- 21900- 64
Cellobiose 225 b 31058b 62
Sucrose 165a 40356a 92
"From Roos (22)
bFrom Stanek et al. (23).

Other Models
The Flory-Huggins model has been applied to predict sugar solubili-
ties in some circumstances. It contains only a combinatorial term in the
activity coefficent, assuming that deviations from ideality can be accounted
for completely by the entropy of mixing, and requires knowledge of the
molar volume of solution (20). It is often applied to polymer solutions. The
entropic free-volume model is a refinement of the Flory-Huggins model in
which van der Waals volumes are used instead of molar volumes (10).

Results and Discussion


The experimental solubilities are presented for the monomers arabi-
nose, galactose, mannose, xylose, and glucose and for the dimer cellobiose
at 20,25, and 30°C in Table 2 along with SDs that ranged from 0.06 to 2.73%
of the mean values. These data are reported as mole fractions according to
the convention of Taylor (4). The order of solubilities from most soluble to
least soluble was mannose, xylose, glucose, arabinose, and galactose. The
values also increased significantly with temperature. The solubility of ara-
binose displayed the greatest temperature dependence, increasing between
10 and 17% over a 5°C increment, and that of xylose showed the least
temperature dependence, increasing by about 10% each 5°C increment.
Expressed as mass fractions, the solubilities ranged from 28.22% for galac-
tose at 20°C to 77.75% for mannose at 25°C.
Figure 1 shows the experimental values combined with values from
the literature, where available, and values predicted by various models.
For xylose and mannose, close agreement was found between the values
reported in our work and those of Gabas and Laguerie (12). Glucose is the
only sugar considered herein for which the solubility over a wide range of

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Sugar Monomer and Oligomer Solubility 185

Table 2
Experimental Solubilities as Measured iln Mole Fractions"
Mole Fraction at Saturation

Sugar 20°C 25°C 30°C


Arabinose 0.07400 (0.00110) 0.08160 (0.00060) 0.09530 (0.00110)
Galactose 0.00378 (0.00090) 0.04320 (0.00020) 0.05040 (0.00010)
Glucose 0.08029 (0.00219) 0.09447 (0.00114) 0.11386 (0.00017)
Cellobiose 0.00921 (0.00090) 0.00823 (0.00030) 0.00918 (0.00052)
Mannose 0.22241 (0.00079) 0.25884 (0.00374) NDb
Xylose 0.11758 (0.00127) 0.12953 (0.00117) 0.14149 (0.00008)
·5Ds are given in parentheses.
bNot determined.

temperatures is well documented, but, as mentioned previously, there is


substantial variation in the literature. As shown in Fig. lC, reported solu-
bilities of a-(o}-glucose cluster around two sets of values: those by Jackson
et al. (3), Peres and Macedo (8), and Young (5), vs those by Taylor (4), and
Jacobsen (7). The values obtained in our work agreed with the latter group.
For all the sugars, the ideal solubility law Eqs. 2 and 3 underpredicted
the actual solubility. In addition, the change in solubility with increasing
temperature for the ideal predictions is much less than for the experimen-
tal solubilities. The first equation of the ideal solubility law was not used
because values for !lC p are not tabulated. Equation 3 does a slightly better
job than Eq. 2, a result consistent with the finding of Neau et al. (16),
implying that the differential heat capacity is better approximated by !lSI
rather than simply set to zero. However, note that the ideal solubility law
does predict the correct order of solubilities, from mannose to galactose,
a sequence that corresponds to the order of the melting temperatures.
This result implies that as TIT m is greater, the solubility is greater, consis-
tent with the idea that as the temperature approaches the melting point
of a solute, more and more of the solute can be held in a liquid state (i.e.,
a solubilized state).
UNIQU AC was used to estimate glucose solubilities using parameters
for the group volumes, surface area, and molecular interaction given by
Peres and Macedo (8). UNIQUAC agrees well with the empirical model
reported by Young (5), which is a cubic equation fit to his data. None of the
other UNIQUAC or UNIFAC models were tested. In addition, the Flory-
Huggins and the entropic free-volume models could not be used because
they required data that were not available in the literature.
Experimental solubilities for cellobiose are given in Table 2. The SDs
are much greater than for the monomers, ranging from 3.59 to 9.74% of the
mean value. Because of these large experimental errors, the solubilities at
20,25, and 30°C are indistinguishable. Part of the reason for this error is
Applied Biochemistry and Biotechnology Vol. 105-108,2003
186 Gray et al.
0.12 -.------------------A-,

0.1

c O.OB
o
~
,"
U:0.06 Ideal Solubility Law Eqn. 3 ,. ,.,.
~
~,,/ .,'
o
...--" ,.,
-------
-,,-' ,----"1'
:::E 0 .04

0.02 --
_ ' ' _ ' • - •• Ideal Solubility Law Eqn. 2

O+----.----.----.-----.----.----.----.---~

20 25 30 35 40 45 50 55 60
Temperature ee)

0.06.---------------=0
8
0.05

c 0.04
o
~
U: 0.03
OJ
,.,. "
'0 Ideal Solubility Law Eqn. 3 ,., ,.
~ ...... "
" ....
::E 0.02
_......
0.01 f- - - - - -
f- • - - •. - • •
-= .. - - - ''1' '
Ideal Solubility Law Eqn. 2
o +---------,--------,---------,--------4
20 30 40 50 60
Temperature ee)

0.3
UNIQUAe
c
0.25
Taylor Experimental Fit
c 0.2
0
ti<tI
U:0.15
~
0

-----..-----":.....---
.... .....-- . .., ..
:::E 0.1 Ideal Solubility Law Eqn. 3 _

0.05
------ -...-.
•• _ •• - •• - ••
--- .. ~
Ideal Solubility Law Eqn. 2
0
20 30 40 50 60
Temperature ee)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Sugar Monomer and Oligomer Solubility 187
0.3 .------------------,0=0·
I
~
0.25
+ I

.- . - . . ·r
c:
... 1
....!I
\
0 0.2 Ideal SokbiUty Law Eqn. 3 .... .......
'tl
!!! ....,.,
....
.......


......';

I
IL
.m 0.15 ... .' I
0
:E
_... .".."",. .."". ........ .. " !Ii
0.1
-
.. •• - II
0.05
•• - , - Solaity Low E," 2 I
0
20 30 40 50 60
Tel1ll8rature eC)

0.25 .------------------;E~

0.2

c:
~0.15
~
.m

-
~ 0.1 Ideal Solubilty Law Eqn. 3 ......... ..
~--- .....
_....... . .. """"
0.05 _-- _ •• ""f
-_ ...... -- . - - . ..--. ....... - I
I- • - - •• Ideal Solubility Law Eqn. 2
0+----~---~---_r---_1

20 30 40 50 60
Temperature eC)

Fig. 1. Monomer solubilities vs temperature. (A) Arabinose solubility; (B) galac-


tose solubility; (C) glucose solubility; (D) mannose solubility; (E) xylose solubility.
( <> ) This work; ( 0 ) Gabas (6); ( + ) Jacobsen (7); ( x ) Jackson et a1. (3); ( 0 ) Peres
and Macedo (8).

that the lower solubility of cellobiose makes it more prone to variations


during weighing and analysis with HPLC. However, the values in this
work do agree well with those reported by Taylor (4) and Jacobsen (7), as
shown in Fig. 2. For cellobiose and all the oligomers, Eq. 3 of the ideal
solubility law was used to estimate their solubilities because Eq. 3 was
observed to be more accurate than Eq. 2 for the monomers. Maximum
and minimum values were used based on estimates of the entropies of
fusion as discussed previously. Between 20 and 30 o e, the ideal equation
maximum underpredicts the solubility of cellobiose, but at higher tem-
peratures, the maximum and minimum predictions do bracket the experi-
mental solubilities.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


188 Gray et al.
0.04
Ideal Solubitity Law Eqn. 3 '"
0.035
~/
0.03 ,/
Ideal p~dft!fion (maximum) - -......
g 0.025
.,.. " ,
~u. 0.02 .,..'"
..9!
0
~ 0.015

0.01

0.005

0
20 30 40 50 60
Temperature ("C)

Fig. 2. Cellobiose solubility vs temperature. ( 0 ) This work; ( 0 ) Jacobsen (7).

1.0E+00 , . - - - - - - - - - - - - - - - - - - - - ,

1.0E-01 C5 (max.) C3 (max.)


C4(max.)

1.0E-05 +------,-----,-------,------;
20 30 40 50 60
Tel11l8rature ("C)

Fig. 3. Cello-oligomer solubilities vs temperature.

Figure 3 also shows the actual solubilities of cellotriose, cellotetraose,


and cellopentaose as reported by Taylor (4). For all of these oligomers, the
actual solubility falls within the minimum and maximum limits in the tem-
perature range of 20-60°C. Additionally, the predicted solubilities change
more rapidly with temperature than the experimental solubilities, indicat-
ing that for elevated temperatures, the actual values may be less than both
the predicted minimum and the predicted maximum.

Applied Biochemistry and Biotechnology Vol. 705-108, 2003


Sugar Monomer and Oligomer Solubility 189

The estimated values for the solubility of the xylobiose, xylotriose,


xylotetraose, xylopentaose, and xylohexaose are given in Fig. 4. The solu-
bilities are reported as mole fractions to make them comparable to the other
solubility data and as mass fractions of xylose equivalents. The conversion
is as follows:
. oligo mole fraction X MW
oltgomassfraction = ( l'
o 19O mole fraction X
r
MW) + (1 - 0 19O mole fraction ) X
18 (4)

oligo (as xy[ose) mass fraction = oligo mass fraction X (150 X DP/MW) (5)
MW is the molecular weight of the oligomer and DP is the degree of
polymerization of the oligomer. Figure 4 shows that all of the xylo-oligo-
mers except for the trimer are predicted to have much greater solubilities
than the corresponding cello-oligomer. At room temperature, the solu-
bilities range from between 0.02 and 0.001 for xylobiose to 0.005 to 0.0001
for xylohexaose, corresponding to a range of mass percentages of about
1-20% for both species. This implies that, although they are much less
soluble than the xylose monomer, they are at least moderately soluble at
room temperature. When these predictions are extrapolated to 150°C, the
predicted solubilities are very high. The minimum predicted solubilities
of xylopentaose and xylohexaose at 120°C approach the experimental
solubility of the most soluble monomer, mannose, at room temperature
on the basis of mass fraction.
The next step was to compare these numbers with numbers expected
in an uncatalyzed batch hydrolysis to estimate how solubility might in-
fluence hydrolysis. For example, for biomass containing 25% xylan (on a
dry basis) in a batch system with 20% solids, the maximum oligomer mass
percentage in the liquor would be 6.25% (expressed as xylan equivalents)
at a yield of 100%. Converted to xylose equivalents, this maximum oligo-
mer mass percentage is 7.1 %. Figure 4 shows that at 120°C, the predicted
minimum mass percentages for xylobiose to xylohexaose are all >55% (as
xylose equivalents). Thus, for all oligomers with a degree of polymeriza-
tion <6, the maximum expected oligomer concentration would be much
less than the solubility of anyone of the oligomeric components, and for
temperatures exceeding 120°C, the solubility of any xylo-oligomer of
chain length <6 is not expected to be a limiting factor in batch hydrolysis.
However, higher molecular weight oligomers could still be a factor, but
their solubility could not be predicted because their melting points are
not known.

Conclusion
This work examined the solubilities of several monomers and oligo-
mers expected in biomass hydrolysis. The monomers were very soluble in
the temperature range of 20-30°C, with a mass fraction ranging from 28.22%
for galactose at 20°C to 77.75% for mannose at 25°C. The order of solubilities

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


190 Gray et al.
0.7
A
0.6 X2(m"')~.
O.S
c: X3 (max.) XS (min.) •
0
~0.4
IL
~0.3
::E
0.2

0.1 X6 (min.)

0.0
20 40 60 80 100 120 140
Temperature ("C)

1.0 1----------~i=iiiI"'l"""""J
WO.9 X2 (max.) _~~

..m X3 (max.)
~0.8
·s X4 (max.)
'[0.7
CD
~0.6
>.
~O.S
c:
o
UO.4
~
IL
U) 0.3
:a
::E 0.2
0.1 +---.---.!:::......;~:....-__,_--___.__--__r_--...,......:=..J
B
20 40 60 80 100 120 140
Temperature ("C)

Fig. 4. Xylo-oligomer solubility vs temperature.

from most to least soluble is mannose, xylose, glucose, arabinose and galac-
tose. All of the solubilities were strongly temperature dependent, increasing
in solubility by at least 10% in a SoC increment. The data developed were in
close agreement with the values reported by most other researchers.
Neither Eq. 2 nor Eq. 3 of the ideal solubility law predicted the experi-
mental solubilities closely, although both equations predicted the correct
order of solubilities. UNIQUAC did closely predict some of the experi-
mental values reported in the literature. UNIQUAC was simulated using
the parameters reported by Peres and Macedo (8) and are presented in
Table 3. Not enough information was available to estimate the solubilities
of the other sugars using UNIQUAC, UNIFAC, Flory-Huggins, or the
Entropic Free-Volume models.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Sugar Monomer and Oligomer Solubility 191

Table 3
UNIQUAC Parameters (9)
Glucose fixed parameters
To (reference temperature) 298.15 K
Tm (melting point temperature) 425.15 K
6.H (fusion enthalpy) 32432J mol
6.A (constant term for linear temperature dependent 6.C ) 139.5766 J/mol
6.B (slope term for linear temperature dependent 6.Cp ) p oJ/(mol·K)
Size parameters (dimensionless) Glucose Water
Q; (surface area parameter) 8.1528 0.92
R; (volume parameter) 7.92 1.4

Interaction parameters (dimensionless)


aij (constant term for linear temperature dependent interaction) Glucose Water
Glucose 0 -68.6157
Water 96.5267 0

a;j (slope term for linear temperature dependent interaction) Glucose Water
Glucose 0 -0.069
Water 0.277 0

The solubility of cello-oligomers was also investigated as a first step


toward evaluating tools for predicting their behavior in solution. In the
series cellotriose, cellotetraose, and cellopentaose, solubility drops rapidly
with increasing chain length, and it was possible to bracket the actual solu-
bilities with ideal solubility law predictions over the temperature range of
20-60°C, by approximating the entropy of fusion. This model was then
applied to predict the solubilities of xylobiose, xylotriose, xylotetraose,
xylopentaose, and xylohexaose, as shown in Fig. 4, but no experimental
data on their solubilities were available for comparison. The ideal predic-
tions estimate that the solubility of all the xylo-oligomers, except for the
trimer, are greater than that of the corresponding cello-oligomer, with
values at room temperature ranging from 1 to 20% by mass. At 120°C, the
estimated mass percentages (as xylose equivalents) are all >55%. This im-
plies that in the example conversion scheme discussed previously, the solu-
bility of xylo-oligomers with a chain length of 6 or less is not likely to be a
limiting factor.
Continuing work for this project will include taking experimental
measurements of the solubilities of the xylo-oligomers, including longer-
chain oligomers, and comparing the results to model predictions. We also
plan to include measuring the enthalpies of fusion to test the ideal solu-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


192 Gray et al.
Table 4
Parameters Used in Study Models
Parameter Units
Activity coefficient Dimensionless
Athermal activity coefficient Dimensionless
Differential heat capacity (.6.C ) J/(g·K)
Enthalpy of fusion (.6.Hf ) P J
Entropy of fusion (.6.Sj ) Dimensionless
Free energy of dissolution (.6.Gdis ) J
Ideal gas constant (R) J K-l mol-l
Mass fraction Dimensionless
Mass fraction of xylose equivalents Dimensionless
Melting point temperature (Tm) Dimensionless
Mole Fraction (X) Dimensionless

bility law when applied to oligomers. Finally, the solubility of oligomers


in multiple component systems will be examined, including mixed oligo-
mer solutions; solutions containing xylose; solutions containing other
sugars; solutions containing degradation products; and solutions con-
taining other components present in biomass such as lignin, acetic acid,
uronic acid, and extractives.

Acknowledgments
We wish to thank the Thayer School of Engineering, Dartmouth Col-
lege for the use of its facilities. This work was funded by The National
Science Foundation Division of Bioengineering and Environmental Sys-
tems through contract BES-9985351.

References
1. Lynd, L. R., Wyman, C. E., and Gerngross, T. U. (1999), Biotechnol. Prog. 15,777-793.
2. Saeman, J. F. (1945), Ind. Eng. Chem. 37,42-52.
3. Jackson, R. F., Silsbee, C. G., and Profitt, M. J. (1922), Sci. Papers Bur. Stand. 20,588.
4. Taylor, J. B. (1957), Trans. Faraday Soc. 55, 1198-1203.
5. Young, F. E. (1957), ,. Phys. Chern. 61,616-619.
6. Gabas, N., Carillon, T., and Hiquily, N. (1988), ,. Chern. Eng. Data 33, 128-130.
7. Jacobsen, S. E. (2001), MS Thesis, Dartmouth College, Hanover, NH.
8. Peres, A. M. and Macedo, E. (1996), Fluid Phase Equilibria 123, 71-95.
9. Peres, A. M. and Macedo, E. (1997), Fluid Phase Equilibria 139,47-74.
10. Peres, A. M. and Macedo, E. (1997), Carbohydr. Res. 303,135-151.
11. Macedo, E. A. and Peres, A. M. (2001), Ind. Eng. Chern. Res. 40,4633-4640.
12. Gabas, N. and Laguerie, C. (1993), J. Cryst. Growth 128, 1245-1249.
13. Abed, Y., Gabas, N., Delia, M.L., and Bounahmidi, T. (1992), Fluid Phase Equilibria 73,
175-184.
14. Catte, M., Dussap, c., and Gros, J. (1995), Fluid Phase Equilibria 105, 1-25.
15. Spiliotis, N. and Tassios, D. (2000), Fluid Phase Equilibria 173, 39-55.
16. Neau, S. H., Bhandarkar, S. V., and Hellmuth, E. W. (1997), Pharm. Res. 14,601-605.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Sugar Monomer and Oligomer Solubility 193
17. Grant, J. W. and Higuchi, T. (1990), in Techniques of Chemistry, vol. 21, John Wiley &
Sons, New York, NY, pp. 14-19
18. Grant, J. W. and Higuchi, T. (1990), in Techniques of Chemistry, vol. 21, John Wiley &
Sons, New York, NY, p. 22.
19. Fredenslund, A., Jones, R. L., and Prausnitz, J. M. (1975), AICHEI 21,1086-1098.
20. Danner, R. P. and High, M.S., eds. (1975) in Handbook of Polymer Solution Thermody-
namics, American Institute of Chemical Engineers, New York, NY, pp. 17-18.
21. Roos, Y. (1993), Carbohydr. Res. 238,39-48.
22. Stanek, J., Cerny, M., and Pacak, J. (1965), The Oligosaccharides. Academic, New
York, NY.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0195/$20.00

Influence of Pressure in Ethanol/Water


Pulping of Sugarcane Bagasse

ADILSON R. GON(:ALVES· AND DENISE S. RUZENE

Departamento de Biotecnologia, FAENQUIL, CP 116,


CEP 12.600-970 Lorena, SP, Brazil, E-mail: adilson@debiq.faenquil.br

Abstract
The influence of the pressure in the ethanol/water pulping of sugarcane
bagasse was studied using argon pressure varying from 0.5 to 1.5 MPa. The
reaction volume and activation volume were studied. For the reaction vol-
ume, temperature and time were constant and pressure was varied, and for
the activation volume, temperature was constant and pressure and time were
varied. The degradation of cellulose was not promoted by the pressure with
positive reaction volume (4100 cm3 / mol). On the other hand, degradation of
xylan (polyoses) and lignin was strongly favored by the pressure and reac-
tion volume ranged from -1000 to -3000 cm3 / mol.

Index Entries: Ethanol/ water pulping; sugarcane bagasse; pressure; reac-


tion volume; activation volume.

Introduction
Organosolv is a pulping process that uses organic solvents, and it has
been proposed as an alternative for chemical pulp production (1). An etha-
nol/ water mixture is one of the most promising organosolv delignification
process options because it combines high delignification rates with favor-
able conditions for organic solvent recovery, low cost, and abundance of
ethanol in countries where sugarcane is economically important (2).
We studied the influence of pressure of an inert gas (argon) in the
pulping of sugarcane bagasse. The effect of pressure can be evaluated by
the reaction volume ( V) and activation volume ( vt).
In any reaction
reactants (R) - transition state m- products (P)
Reaction volume, V, is given by V =V p - V R and activation volume,
V*, is defined by V* = v+ - V R •

"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 195 Vol. 105-108,2003


196 Gon~alve5 and Ruzene

v
V' > 0 (pre ure di favor)

o
o 6. V < 0 (pressure favor)

reaction coordinate

Fig. 1. Example of reaction volume and activation volume.

Figure 1 illustrates a hypothetical reaction process in which the vol-


ume of the transition state (shown by the upper shaded ellipse) is higher
than that of the reactants ( V+ > 0) and the volume of the reactants is higher
than that of the products ( V < 0) (3).
The parameters V and V+ are useful in the determination of the
better conditions of reactions performed in a closed system, such as the
production of cellulosic pulps.
The volume changes just defined can be determined by making use of
the fundamental thermodynamic equation bG/bp = V, in which G is the
Gibbs' free energy and p is the pressure. Since G =- RT In K, by combining
these two equations the following relationship is obtained (4,5):

( dInK) =-6.V (1)


ap T RT
Analogously using G +Eq. 2 is obtained:

In k) = - L1 V+
( aapT RT
(2)

in whick K is the equilibrium constant; k is the rate constant; V is the


reaction volume; Vtis the activation volume; p is the pressure (Mpa); Tis
the temperature (K); and R is the gas constant (8.314 J/[mol·K]).
Equations 1 and 2 show that pressure influences the thermodynam-
ics and kinetics of the reaction, and the variation in equilibrium and rate
constants as a function of the pressure are proportional to the reaction
(4,5) and activation volumes (4), respectively. The reaction volume, V, is
related to the product formation from the reagents (thermodynamics),
while the activation volume, V+, is related to the formation of transition
state from the reactants (kinetics).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Effect of Pressure in EthanollVVater Pulping 197
Pressure favors product formation when the product volume is
smaller than that of the reactant (there is an increase in the product density
in comparison with reactant density). On the other hand, pressure disfa-
vors product formation when the product volume is higher than that of the
reactant (there is a decrease in the product density) (4,5)
A recent example of this approach to lignocellulosics is the determina-
tion of V =-817 cm3 /mol for the oxidation of organosolv lignins in acetic
acid (6). This value confirms the high influence of the pressure in the oxi-
dation of lignin. In another study, delignification of sugarcane bagasse
with acetone had a V =-1533 cm3 /mol (7).

Materials and Methods


Pulping
Ethanol/water organosolv pulping of sugarcane bagasse was per-
formed in a 200-mL closed vessel, using a 1:1 ethanol/water mixture (v Iv),
a 1:10 bagasse to solvent ratio (m/v), and cooking time varying from 1 to
3 h. Before heating, the vessel was closed and pressurized with argon (inert
gas) at pressures varying from 0.5 to 1.5 MPa. For the reaction volume
study, temperature and time were constant and pressure was varied; for
the activation volume study, temperature was constant and pressure and
time were varied.
Once the reaction time was complete, the vessel was opened, and the
product was removed, filtered, and washed with 2500 mL of a 1:1 ethanol/
water mixture (v Iv). The obtained pulp was air-dried for yield determina-
tion. The total yield was defined as the percentage relationship between
pulp weight (Wp) and the initial bagasse weight (Wb): Yield(%) =(Wp IWb)
x 100%.
Pulp samples were analyzed for kappa number (measure of the residual
lignin content in pulps) and viscosity by standard methods (8,9). The pulp
composition was determined by acid hydrolysis as described next (10).

Hydrolysis of Pulp
One gram of dry pulp was treated with 10 mL of 72% H 2S04 with
stirring at 45°C for 7 min for the hydrolysis and solubilization of carbohy-
drates. The reaction was interrupted by adding 50 mL of distilled water,
and the mixture was then transferred to a 500-mL Erlenmeyer flask and the
volume completed to 275 mL with distilled water. The flask was auto-
claved for 30 min at 1.05 bar for the complete hydrolysis of carbohydrate
oligomers. The mixture was filtered and the hydrolysate completed to 500
mL with distilled water. A 40-mL sample of the hydrolysate was diluted
to 50 mL, and the pH was adjusted to 2.0 with 2 mol/L of NaOH. After
filtration in a Sep-Pak CIS cartridge to remove aromatic compounds (com-
ing from lignin derivatives), the hydrolysate was analyzed in an Aminex
HPX-87H column (300 x 7.8 mm) (Bio-Rad, Hercules, CA) at 45°C by using
a Shimadzu chromatograph and refraction index detector.
Applied Biochemistry and Biotechnology Vol. 705-108,2003
198 Gonc;alves and Ruzene
The mobile phase was 0.005 mol/L of H 2S04 at 0.6 mL/min. Sugar
concentrations reported as xylan and glucan with respect to the amount of
pulp were determined using calibration curves of pure compounds (10).
The dark solid obtained in the filtration after hydrolysis was oven-dried
and weighed. The obtained mass corresponded to the residual lignin in the
pulp and was reported as a percentage with respect to the pulp weight on
a dry basis. Mass balance is defined as the sum of the percentage of xylan,
glucan, and lignin present in the pulp, with respect to the pulp weight on
a dry basis. These values were also normalized to 100% and expressed in
centesimal form for calculations using Eqs. 3-5.
Results and Discussion
Reaction Volume and Activation Volume
Results for ethanol/water pulping of sugarcane bagasse as a function
of different argon pressures and reaction times are presented in Table 1. A
decrease was observed in the yield when the reaction time increased from
1 to 3 h and the pressure increased from 0.5 to 1.5 MPa. This was also
observed in the results of the viscosity, decreasing 20% when the reaction
time increased from 1 to 3 h and decreasing about 30% when the pressure
increased. Values of kappa number in 1 h decreased with an increase in the
pressure; for the 2-h reaction-time data, kappa number increased with an
increase in pressure, and at 3 h (excluding kappa number at 1.0 MPa),
kappa number also increased with an increase in pressure. Residual lignin
(RsL) follows kappa number, with an average relationship of RsL = 0.2 x
kappa. Viscosity to kappa ratio is a measure of the compromise between
carbohydrate preservation (high viscosity) and delignification (low kappa
number). Increasing pressure favored delignification but carbohydrates
were degraded to some extent.
The chemical composition of the organosolv pulps expressed as
glucan, xylan, and residual lignin is also given in Table 1. Xylan degrada-
tion occurred with an increase in the pressure and reaction time. This deg-
radation was also assessed by the xylan/ glue an content ratio.
The values of glucan, xylan, residual lignin, and total yield were used
for the determination of constants used in calculating the reaction volume
and activation volume.
Calculation of Reaction Volume
Initially, the following equilibrium situation was considered for each
individual component:
component in the pulp ~ component in solution
The total volume of the system was constant (closed vessel) and mass
values were used to determine equilibrium constants:
(K = concentrationcomponent in solution/ concentratio~omponent in pulp)'
Calculations of the reaction volume were made using concentrations of
glucan, xylan, residual lignin, as well as pulp yield.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
),.
:g
i5'
Q..
OJ

n
::,-
til
:3
in' Table 1
q-
'<: Ethanol/Water Pulping of Sugarcane Bagasse
::,
'"Q.. as Function of Argon Pressure and Reaction Time and Composition of Pulps
OJ (% in Dry Basis with Respect to Pulp WeighW

iii
n::,- Mass
::,
Total Viscosity/ Residual Xylan/
0 Time Pressure yield Kappa Viscosity kappa lignin Glucan Xylan glucan Balance
~
'<: (h) (MPa) (%)b number (cP) ratio (%) (%) (%) ratio (%)
1 0.5 51.5 52.7 9.3 0.18 12.0 71.2 9.0 0.13 92.1
2 0.5 49.0 38.7 7.5 0.19 9.3 75.1 7.3 0.10 91.7
<..0
<..0 3 0.5 48.5 43.0 7.6 0.18 8.5 46.3 4.7 0.10 59.5
1 1.0 48.2 50.4 5.4 0.11 9.8 64.7 4.4 0.07 78.9
2 1.0 44.7 45.7 5.3 0.12 10.0 64.1 4.1 0.06 78.2
3 1.0 43.4 37.4 4.7 0.12 8.8 78.1 4.3 0.06 91.1
1 1.5 47.0 43.5 6.5 0.15 9.0 76.1 6.0 0.08 90.9
2 1.5 46.0 48.9 5.4 0.11 10.3 77.0 4.9 0.06 92.2
3 1.5 44.0 50.1 4.8 0.10 9.4 78.1 4.3 0.06 91.8
-Composition of original sugarcane bagasse: 43.7% glucan, 24.4% xylan, 28.0% lignin.
bTotal yield = (pulp weight/bagasse weight) x 100%.
~
~
Cl
,00
N
Cl
::2
)..
"0 Table 2
"0
~
Equilibrium Constants (K) for Glucan, Xylan, Residual Lignin, and Total Yield (Yt total) and Corrected Values
c...
OJ Equilibrium constants Corrected equilibrium constants
0'
~
" Time Pressure K K K K K K K
III
3 (h) (MPa) (glue an) (xylan) (residual lignin) (Yt total) (glucan) (xylan) (residual lignin)
in'
q-
'<: 1 0.5 0.2912 4.0957 3.6983 0.9418 0.1892 3.6932 3.3272
'":Jc... 1 1.0 0.5182 10.137 5.1470 1.0747 0.1979 7.7868 3.8499
OJ
0' 1 1.5 0.3237 7.3754 5.8642 1.1277 0.2033 6.6132 5.2396
~ 2 0.5 0.2866 5.6029 5.3717 1.0408 0.1798 5.0549 4.8428
~
":J 2 1.0 0.6524 11.887 5.4957 1.2371 0.2922 9.0779 4.0796
0
~ 2 1.5 0.3367 9.4785 5.1283 1.1739 0.2324 8.6612 4.6503
'<: 3 0.5 1.1084 9.3613 6.0432 1.0619 0.2545 5.165 3.1907
3 1.0 0.3968 11.656 6.6026 1.3042 0.2725 10.53 5.9259
N
3 1.5 0.3778 11.483 3.0202 1.2727 0.2648 10.46 5.4446
0
0

Table 3
Slopes of Straight Line In K = (- V /RD x p and Linear Correlations (r 2) and Corrected Values at Different Reaction Timesa
Linear correlations Corrected linear correlations
Residual Glucan + Yt Residual Glucan +
Glucan Xylan lignin xylan total Glucan Xylan lignin xylan
1 h, slope 0.1059 0.5888 0.461 0.1556 0.1802 0.0717 0.5826 0.4541 0.1427
1 h, r2 0.0298 0.4092 0.9409 0.1138 0.9325 0.9798 0.5519 0.9592 0.6778
~ 2 h, slope 0.1611 0.5257 -0.046 0.1539 0.1203 0.2567 0.5385 -0.041 0.1860
2 h, r2 0.341 0.464 0.4324 0.0705 0.4613 0.2793 0.685 0.0512 0.3921
a
-~ 3 h, slope -1.076 0.2043 -0.004 -0.6690 0.1811 0.0396 0.7056 0.5344 0.1195
,00
N
3 h, r2 0.7842 0.6955 0.0013 0.7812 0.6506 0.3334 0.7429 0.6336 0.6654
a
8 "Bold numbers (significant correlations) were used to calculate the results of the reaction volume and corrected reaction volume.
Effect of Pressure in EthanollWater Pulping 201

Table 4
Calculated and Corrected Reaction Volumes
Reaction volume Corrected reaction volume
(cm3 /mol) (cm3 /mol)
Time Residual Yt Residual
(h) Glucan Xylan lignin total Glucan Xylan lignin
1 -1757 -687 -273 -1730
2 -2052
3 4102 -779 -690 -2689 -2036

Equation 3 was used to calculate the equilibrium constants for indi-


vidual components, and Eq. 4 was used to calculate total equilibrium
constant:
r·- (p. x Yt)
K.= I I
(3)
I (Pi x Yt)
K = 1- Yt (4)
Yt
in which Ki = equilibrium constant of component i; 'i is the content of com-
ponent i in the original sugarcane bagasse (%), Pi is the content of compo-
nent i in the pulp (%), Yt is the pulp total yield (centesimal form), and K is
the total equilibrium constant.
In Table I, the mass balance after chemical analysis was <100%, owing
mainly to the incomplete hydrolysis of the carbohydrates and material
losses in the gravimetric quantifications. The values obtained by Eq. 3 were
corrected by the mass balance (Pi x Yt divided by balance in the centesi-
mal form).
Table 2 presents the values of the equilibrium constants and corrected
equilibrium constants (K) of glucan, xylan, residual lignin, and pulp yield.
At constant reaction times, the pressure was varied from 0.5 to 1.5 MPa, and
nine equilibrium constants were obtained and used for the calculation of
the reaction volume, according to Eq. 1.
Table 3 shows the best results of the slope and linear correlation
obtained, both before and after correction by the normalized mass bal-
ance. The results of the reaction volume and corrected reaction volume
were calculated using the results in Table 3 shown in bold (significant
correlations) .
The reaction volume for the total yield seems to be constant with an
increase in reaction time. On the other hand, with negative reaction volume
(Table 4) the degradation of the glucan was not favored by an increase of
the pressure (Table 4; 3 h).
The degradation of xylans (polyoses) and lignin was favored strongly
by the pressure and the reaction volume ranged from -1000 to -3000 cm3 /
mol, with the same magnitude of values obtained for the delignification
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
).
:g
[
OJ

(")
:::r-
~<;;.
~
OJ
::;,
Q..
OJ
Table 5
o· Rate Constants (k) at Different Reaction Pressures for Glucan, Xylan,
iii
9- Residual Lignin, and Total Yield (Yt total) and Corrected Values
::;,
o
Rate constants Corrected rate constants
~
k k k k k k k
(glucan) (xylan) (residual lignin) (Yt total) (glucan) (xylan) (residual lignin)
N
o
N 0.5 MPa, slope 0.2452 0.3548 0.2024 0.0300 0.0267 0.1364 -0.0160
0.5 MPa, r2 0.7445 0.9763 0.9217 0.8738 0.6410 0.7993 0.0076
1.0 MPa, slope 0.0417 0.0639 0.1063 0.0524 0.0302 0.1358 0.1782
1.0 MPa, r2 0.2459 0.6456 0.9285 0.9401 0.5689 1.0000 0.8455
1.5 MPa, slope 0.0200 0.1996 0.0110 0.0330 0.0249 0.2045 0.0160
1.5 MPa, r2 0.9194 0.9950 0.0238 0.9612 0.9995 0.9910 0.0556
"Bold numbers (significant correlations) were used to calculate the results of the reaction volume and corrected reaction volume.

-~
~
N
g
W
Effect of Pressure in EthanollWater Pulping 203
Table 6
Calculated Activation Volume
Activation Residual Yt
volume Glucan Xylan lignin total
9549 2193 4912 -360

of sugarcane bagasse with acetone (-1533 cm3 / mol) (7). This fact explains
the decrease in kappa number and viscosity shown in Table 1.
Calculation of Activation Volume
For the calculation of activation volume, the same conditions used in
the calculation of reaction volume were initially considered. A first-order
kinetics was assumed and the rate constant was determined by Eq. 5.
Ln(pjxYt)=-kjxt (5)
in which ki is the rate constant for component i, Pi is the content of the
component i in the pulp (%), Ytis the pulp total yield (centesimalform), and
t is the time (h).
As already mentioned, the obtained values after the application of
Eq. 5 were corrected by the mass balance (Pi x Yt divided by the balance
in the centesimal form). The values of k obtained with the linear correla-
tions are given in Table 5 (original and corrected).
The values of activation volume were calculated using Eq. 2 and the
results of Table 5 (shown in bold, significant correlations).
The value of the activation volume was higher for glucan than for
xylan and lignin (Table 6), showing that degradation of the glucan was
disfavored by pressure. The use of pressure favored first the degradation
of the xylan followed by the lignin; glucan was degraded last.
The final yield was low owing to the delignification and degradation
of the xylans that causes the decrease in viscosity.
The easy degradation of the xylans can be explained by their ramified
and amorphous structure. The cellulose has a linear structure with crystal-
line and amorphous parts, being more difficult to intumesce (to increase
the volume) during the reaction. In an aqueous reaction, xylans tend to
swell more easily than the glucan and xylans tends to degrade, also pro-
moting dissolution of the lignin.

Conclusion
Reaction and activation volume can be calculated for complex systems
such as lignocellulose conversion by using appropriate equations, analyses,
and considerations. For the ethanol/water pulping of sugarcane bagasse,
pressure disfavored the degradation of cellulose and favored lignin disso-
lution. These facts are positive for the system, with kappa reduction. On the
Applied Biochemistry and Biotechnology Vol. 105-708,2003
204 Gom;alves and Ruzene
other hand, xylans were easily degraded by the pressure, decreasing the
viscosity of the obtained pulps. The use of pressure for the conversion in
terms of yield was disfavored, and favored with respect to delignification.

Acknowledgments
We acknowledge financial support from Funda<;ao de Amparo a
Pesquisa do Estado de Sao Paulo and Conselho Nacional do Desenvolvi-
mento Cientlfico e Tecnol6gico.

References
1. Gilarranz, M. A, Oliet, M., Rodrigues, F., and Tijero, J. (1998), Canadian J. Chern. Eng.
76(2), 253-260.
2. Aziz, S. and Sarkanen, K. (1989), TAPPI J. 72, 169-175.
3. Ruzene, D. S. (2001), MS thesis, Faenquil/Debiq, Lorena, Brazil.
4. Stochel, G. and Eldik, R. (1999), Coord. Chern. Rev. 187,329-332,.
5. Asano, T. and Noble, W. J. (1978), Chern. Rev. 78,407-489,.
6. Gonc;alves, A R. and Schuchardt, U. (1999), Appl. Biochern. Biotechnol. 77-79,127-132.
7. Benar, P. and Schuchardt, U. F. (1989), in Brazilian Symposium on the Chemistry of
Lignins and Other Wood Components, vol. 1, Sao Carlos, Brazil, pp. 186-192.
8. TAPPI (1985), TAPPI Standard Methods, T 236 cm-85, Technical Association of Pulp
and Paper Industry, TAPPI, Norcross, GA
9. TAPPI (1982), TAPPI Standard Methods, T 230 om-82, Technical Association of Pulp
and Paper Industry, TAPPI, Norcross, GA
10. ASTM (1956) Standard Test Methods for Lignin in Wood, ASTM Method D 271-48,
American Society for Testing Materials, Philadelphia, PA

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0205/$20.00

Post-Harvest Processing Methods


for Reduction of Silica
and Alkali Metals in Wheat Straw

DAVID N. THOMPSON,* PETER G. SHAW,


AND JEFFREY A. LACEY
Biotechnology Department, Idaho National Engineering
and Environmental Laboratory, PO Box 1625,
Idaho Falls, 10 83415-2203, E-mail: thomdn@inel.gov

Abstract
Silica and alkali metals in wheat straw limit its use for bioenergy and
gasification. Slag deposits occur via the eutectic melting of Si02 with K20,
trapping chlorides at surfaces and causing corrosion. A minimum melting
point of 950°C is desirable, corresponding to an Si02:K20 weight ratio of
about 3:1. Mild chemical treatments were used to reduce Si, K, and Cl, while
varying temperature, concentration, % solids, and time. Dilute acid was more
effective at removing K and Cl, while dilute alkali was more effective for Si.
Reduction of minerals in this manner may prove economical for increasing
utilization of the straw for combustion or gasification.
Index Entries: Bioenergy; combustion; gasification; fluidized bed; silica;
potassium; chloride; slagging.

Introduction
Agricultural crop residues are a valuable renewable resource from
which to produce biobased products. In 1999, American farmers harvested
53,909,000 acres of wheat (1). The straw from this acreage of wheat repre-
sents more than 100,000,000 tons annually. Currently, some of the straw is
harvested (baled) for use as livestock bedding or low-grade animal feed.
However, these low-value uses provide only a minimal return. Nationally,
only about 3.2% of the economic return on wheat is from straw (1). Produc-
ers have long recognized the potential economic and environmental ben-
efits in producing bioenergy and bioproducts from excess wheat straw

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 205 Vol. 105-108, 2003


206 Thompson et al.
residue. However, the silica and alkali metals in wheat straw limit its use
for bioenergy and gasification.
Because of slagging, fluidized-bed combustors and gasifiers cannot
be operated above the ash fusion temperature of the feedstock, and boil-
ers can only be operated well above or well below the ash fusion tempera-
ture (2). Wheat straw contains significant amounts of low-melting ash,
which is comprised of mineral oxides including primarily Si02, K20, and
CaO,withsmalleramountsofS03,MgO,Na20,Fe203,Al203,andTi02(3).
This mineral ash deposits onto furnace or heat-transfer surfaces in two
ways. SZagging is deposition of molten or highly viscous ash and occurs
in the hottest regions of the combustor or gasifier. In contrast, fouling is
deposition by condensation of vaporized ash and occurs in the cooler
regions. Vaporized KCI is then held in close contact with the metal sur-
faces of the combustor or gasifier, causing corrosion. Ash fusibility is an
important factor in determining whether slagging or fouling will occur.
K20 and other oxides form a eutectic mixture with Si02, lowering the ash
fusion temperature (4). K20 is of particular importance in the slagging of
bio-mass ash owing to its high concentration. For comparison, pure Si02
melts at 1703°C (5), while a 4:1 (w /w) Si02/K20 mixture melts at about
1100°C, a 3:1 mixture at about 950°C, and a 2.3:1 mixture at about 850°C
(6) (i.e., lower Si02/K20 ratios lead to lower ash fusion temperatures).
While an efficient operating temperature is about 1100°C, the preferred
minimum is about 950°C, depending on the design (6). Therefore, an
effective treatment to remove minerals from straw, thus red ucing slagging
and corrosion during combustion or gasification, must (1) reduce total
Si02 content; (2) increase Si02/ alkaline oxide ratios, particularly Si02/
K20i and (3) increase the Si02 /KCI ratio.
The cellulose-rich vascular tissues of straw stems contain relatively
higher amounts of organic material and fermentable carbohydrates for
conversion to bioenergy, biofuels, and chemicals. In contrast, the for-
merly physiologically active tissues, including the leaves, sheaths, and
awns, are heavily impregnated with silica in the epidermal layer, and
these tissues also contain higher amounts of noncarbohydrate organic
components (Le., protein,lipids, pigments, pectin, organic acids) than the
stems. For cost-efficient utilization of straw and other crop residues for
bioenergy or gasification, the undesirable mineral components must be
removed. However, the current paradigm for straw utilization includes
the necessity to transport all the components of the straw to the point of
use. There is no cost-efficient way to remove undesirable components
from straw before transportation. This is expensive not only because of
the low bulk density of straw, but also because it brings the less valuable
components to the manufacturer's gate and creates economic and envi-
ronmentalliabilities.
At the Idaho National Engineering and Environmental Laboratory
(lNEEL), we are developing an in-field method to selectively harvest the
stem fraction in order to reduce the harvested silica and mineral loads. In

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Removal of Minerals from Wheat Straw 207
this article, we describe tests conducted at the INEEL to further reduce
silica and alkali mineral loads in the stem fraction to allow its use for
bioenergy production and gasification in fluidized beds and boilers. The
objectives of this study were to reduce Si02 content to minimize total slag
formation; to reduce K20, increasing the ratio of Si02 to alkaline oxides and
increasing the temperature at which the slag forms; to remove chlorides,
minimizing the potential for corrosion owing to the generation of KCI
vapor at metal surfaces beneath slagging and fouling deposits; and to per-
form these reductions in a manner that can be used economically in both
distributed and centralized systems.

Materials and Methods


Wheat Straw
Westbred 936 wheat straw, a hard red spring variety, was obtained
from Grant 4-D Farms (Rupert, ID). All straw utilized in the laboratory
studies was produced during the year 2000 cropping season. Twenty large
bales of Westbred 936 (1.2 x 2.4 m [4 ft x 8 ft] bales) were collected and stored
in a stack at the side of the field, using only the protected center bales for
the studies. To better handle the straw for the laboratory studies, the large
bales were rebaled as needed into smaller 0.61 x 1.2 m (2 x 4 ft) bales con-
taining about 22.7 kg (50 lb) each and placed in covered storage. The straw
was rethreshed before use in the mineral removal studies as described by
Hess et al. (7) to remove the high-silica plant components including the
leaves, sheaths, fines, and nodes. Only the separated straw stems were used
in the laboratory studies, except when comparing removal of minerals and
silica between whole straw and the stem fraction.

Chemical Wash Procedure


Five to 20 g of air-dried, whole chopped straw or mechanically sepa-
rated straw stems were weighed to the nearest 0.1 mg into a tared 500-mL
wide-mouth polypropylene, Teflon®, or polycarbonate bottle. Sufficient
wash solution was weighed into the bottle to achieve the desired % solids
for the experiment (4-16 wt%). Wash solutions included distilled water,
0.1-5.0 wt% H 2S04, and 0.1-1.0 wt% NaOH. The bottles were shaken at a
minimum of 150 rpm for 0.5-24 h at the desired temperature (25, 37, or
50°C). The liquid was then poured off into tared Pyrex beakers. The straw
stems were then washed quickly with about 25 mL of distilled water to
remove as much of the wash solution as possible without further removing
minerals; this liquid was then added to the collected wash liquid. Both the
collected liquid fraction and the treated stems were dried at 90-105°C for
at least 2 d. The dried samples were removed from the oven, cooled, and
their weights recorded. The dried straw samples were finally ground to 60
mesh in a Wiley mill and stored at room temperature for carbohydrate,
lignin, ash, and mineral analyses.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


208 Thompson et al.
Analytical Methods
Ashing
At least 1 gram of dry straw, ground to 60 mesh, was weighed to the
nearest 0.1 mg into a dry, tared ceramic crucible. The residue from the
collected liquid fraction was also dried and weighed. Ashing was done in
a muffle furnace at 550-650°C for 18-24 h. On selected samples, a duplicate
or spike such as Si02 flour or potassium silicate was added to the straw
before ashing to validate silica measurements. Ashing temperatures as high
as 940°C were used to determine weight loss as a function of temperature
and composition changes from the volatilization of KCl.
Mineral Analyses
ENERGy-DISPERSIVE SPECTROMETRY

Mineral analyses were done by energy-dispersive spectrometry


(EDS), modifying existing geologic and metallurgical procedures (8). The
spectra were measured at 10-20 KeV using a Phillips XL30ESEM Scanning
Electron Microscope with a Princeton Gamma Tech Detector (Princeton,
NJ). Compositions of minerals in the ash were determined for the ele-
ments Si, K, Ca, S, Mg, Mn, Ti, Fe, AI, P, Na, Cl, and O. Except for some
carbon that remains after the ashing and some losses of KCl, these ele-
ments should account for >95% of the ash. Ashed straw was placed in a
thin film onto aluminum disks using double-sided tape, taking care to
transfer as much ash as possible onto the taped disks. The disks were then
carbon coated to prevent charging. Two or three separate 1.0-mm2 areas
of each disk were scanned to verify homogeneity, depending on whether
the measured Si values differed by >1% among measurements. Scanning
was done for 5-10 min, with longer times used for samples containing
very small distinguishable peaks.
EDS standards, prepared from reagent-grade silicates, oxides, and
chlorides, and ashed at the same temperatures used for the samples, were
used to prepare calibration curves to correct the internal quantitative EDS
values for matrix effects. In addition, Standard Reference Materials, includ-
ing coal fly ash (SRM #1633 and #2690) and soil (SRM #2710), both from the
National Institute of Standards and Technology (Gaithersburg, MD), were
also used to check the accuracy of the instrument and the quantification
software. Finally, several random straw samples were also analyzed by
inductively coupled plasma (ICP) analysis (9) to validate the EDS results.
Oxide compositions were calculated from the elements based on standard
stoichiometric ratios.
INDUCTIVELY COUPLED PLASMA

Mineral analysis by ICP (9) was done to validate EDS results (> 1 wt%
K and Ca, <1 wt% P, K, S, Mg, Fe) and quantify trace elements «0.1 wt%),
particularly the composition of straw micronutrients in the ash, such as Cu,
Zn, Mn, and B. Straw samples and standard reference material! calibration
standards used for the EDS analyses were shipped to Western Labs (Parma,

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Removal of Minerals from Wheat Straw 209
ID) for ICP analyses. Western Labs measured P, Ca, K, Cu, Zn, B, Fe, and
Mn using standard methods (10).
Carbohydrate and Lignin Analysis
Carbohydrate and lignin compositions of untreated and treated straw
samples were determined by quantitative saccharification using the
method of Saeman et al. (11). Carbohydrate analyses were done by high-
performance liquid chromatography using a Bio-Rad HPX-87P carbohy-
drate column. The acid-insoluble fraction, which contained lignin,
extractives, and acid-insoluble ash (primarily Si02), was ashed at 550-650,
and Klason lignin with extractives was calculated by weight difference.

Results and Discussion


The organic and ash compositions of the untreated baled straw and
untreated mechanically separated straw stem fraction are given in Table l.
Mechanical separation reduced the total ash and Si02 concentrations of the
harvested fraction by 23 and 44%, respectively, thereby reducing overall
slagging potential (7). However, mechanical separation concentrated the
alkali metals relative to Si02 , reducing the Si02 /K20 and Si02 /KCl ratios.
Although reduction of total Si02 is important for overall reduction of
potential slag formation, in terms of modifying the eutectic properties and
thereby raising the ash fusion temperature, removal of Si02 is detrimental
without also removing proportional amounts of K, since it decreases the
Si02 /K20 ratio in the ash.
Washing straw with aqueous liquids removed both organic and min-
eral matter. Most of the removal occurred in the first 4 h of washing. The
ratio of organic losses to inorganic losses was similar with all of the wash
solutions; that is, no particular wash solution was more selective for disso-
lution of organic components or minerals. Rather, the minerals appeared to
be released along with organic material, such that very short or very long
washes released approXimately the same ratio of organic and inorganic
components of the straw.
Three wash solutions-distilled water, dilute H 2S04 , and dilute
NaOH-were used separately to remove minerals from separated stems
and from chopped whole straw. General comparisons of the effects of these
washes on Si02 /K20, Si02 /KCI, and the loss of heating value, at 25, 37, and
50 are shown in Figs. 1-3. Acid washing of straw removed most of the K
by dissolution and gave the highest Si02 /K20 ratios. Some acid washes
removed sufficient K to raise the Si02/K20 ratio above the desired mini-
mum ratio of 3.0. The highest removal of K and Cl from straw stems oc-
curred with a 0.2 wt% acid wash at 50°C (Fig. 1). This acid wash increased
the Si02 /K20 ratio from 0.9 to 3.2 and the Si02 /KCl ratio from 0.5 to 71.3
(Fig. 2), significantly reducing corrosion potential. For this calculation, it
was assumed that all CI was present as KCl. The concentration of Si02 in the
distilled water- and acid-washed stems remained constant even though
other mass was lost. Thus, both water and acid removed some Si02, but its

Applied Biochemistry and Biotechnology Vol. 105-108,2003


210 Thompson et al.
Table 1
Organic and Inorganic Compositions of Whole Westbred 936 Straw
and Mechanically Separated Straw Stem Fraction
Component (wt%)a
Westbred 936 Westbred 936
Component whole straw stem fraction
Glucan 32.1 37.2
Xylan 19.3 19.4
Galactan 1.0 0.9
Mannan 4.5 3.0
Arabinan 2.1 1.6
Lignin with extractives 20.3 18.9
Otherb 9.5 10.3
Ash (wt%)' 11.2 8.7
Si02 2.6 1.3
K20d 1.7 1.2
KCld 5.2 5.2
CaO 0.6 0.2
S03 0.4 0.3
MgO 0.3 0.2
P20 5 0.2 0.1
FeO 0.04 0.05
Na20 0.06 0.04
Mn02 0.005 0.003
CuO 0.0005 0.0005
B20 3 0.005 0.005
Al20 3 0.0002 <0.005
Ti02 <0.005 0.0001
ZnO 0.0007 0.004
Total 100 100
aBased on 100% dry wt of material.
bOther organics are attributed to uronic acid, protein, and so on, and to recovery errors
in carbohydrate analysis technique.
cOxide contents estimated by mass balance from elemental composition as determined
byEDS.
<!Potassium assumed to first combine with available chlorine, and then apportioned as
oxide.

removal was offset by the removal of organic matter, and the concentration
in the washed stems remained the same. Alkaline washing dissolved up to
20% of the Si02 and slightly lowered the final Si02 concentration in the
straw. However, the Si02 /K20 ratio was lowered because a less-than-pro-
portional amount of K was removed.
The effect of wash solution and temperature on loss of heating value
is shown in Fig. 3. For these calculations, all components of the organic
matter were assumed to be of equal heating value per unit weight. In gen-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Removal of Minerals from Wheat Straw 211

3.0

-"
~
2.0 N0

1.0 N
0

0.2 wt"fo Acid


0.1 wt"fo Alkal
01 Water
Initial

Fig. 1. Si02 /K20 ratio after acid, alkali, or water washing of straw stems at various
temperatures. DI, distilled water.

80

60 5
N

40 -

20 "Q
o
0.2 wt%Acld
0.1 wt"fo Alkali
01 Water

Fig. 2. Si02 /KCl ratio after acid, alkali, or water washing of straw stems at various
temperatures. DI, distilled water.

12
10 ';ft.

8 OJ
.....
6 c::

-
4 r
o
III
2

Fig. 3. Loss of heating value as function of acid, alkali, and water washing of straw
stems at various temperatures. DI, distilled water.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


212 Thompson et at.

100

- 80

-
iU
o • Cellulose
I- 60 ~ Hemicellulose
o o Lignin/Extractives
?fl. 40 E:;1 Ash
• Other organics
20 on Weight loss

o
None 25 37 50
Temperature (OC)

Fig. 4. Mass balances with increasing temperature for unwashed and washed straw
stems using distilled water at 4% solids for 4 h.

eral, higher wash temperatures resulted in greater loss of mass in acid


washes, with smaller increases in the base and water washes. Increases in
acid concentration increased the amount of minerals removed but did so at
the expense of heating value since organic matter was also removed; about
80% of the mass removed in the acid washes was organic (not shown).
These losses, amounting to 6-11%, indicate that there are trade-offs
between the cost of the lost heating value and the benefits of reducing
minerals through use of the treatment. These trade-offs will depend on a
number of factors, including boiler / gaSifier design, operating conditions,
feedstock, and cost of the downtime required to remove slag from the heat-
transfer surfaces, and will vary significantly among power generation/
gasification facilities.
The effects of temperature on the organic component mass balances
for the tests in Figs. 1-3 are shown in Figs. 4-6. The mass balances for the
distilled water washes are shown in Fig. 4. There were generally higher
losses of organics with increasing temperature, as expected, since this treat-
ment is analogous to an autohydrolysis at low temperature. There were
only small losses of both cellulose and hemicellulose, with losses of organic
matter principally from the other organics"-uronic acids, proteins, and
II

so on-which were not measured in the compositional analyses. No change


was seen in lignin contents, as was expected. There were few differences in
overall ash removal with increasing temperature, although inspection of
Fig. 2 shows that more KCl was removed at higher temperatures.
The mass balances for the 0.1 wt% NaOH washes in Figs. 1-3 are
shown in Fig. 5. For the alkali washes, there were similar weight losses of
organic matter at all temperatures tested. There were no significant losses
of lignin at any of the temperatures tested. This was somewhat surprising
given the alkaline nature of the washes; evidently, either the temperature

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Removal of Minerals from Wheat Straw 213

100

80

60 • Cellulose
r.a Hemicellulose
o Lignin/Extractives
40 ~ Ash
• Other organics
20 lID Weight loss

o
None 25 37 50
Temperature (OC)

Fig. 5. Mass balances with increasing temperature for unwashed and washed straw
stems using 0.1 wt% NaOH at 4% solids for 4 h.

100

80
so
-
t-
o
~
60

40

riI
o
fJ Ash
Cellulose
Hemicellulose
Lignin/Extractives

• Other organics
20 an Weight loss
o
None 25 37 50
Temperature (OC)

Fig. 6. Mass balances with increasing temperature for unwashed and washed straw
stems using 0.2 wt% H 2S04 at 4% solids for 4 h.

or pH was not high enough to effect significant lignin removaL In the alkali
washes, there were only small losses of hemicellulose and cellulose, while
the overall ash removal was similar at all temperatures tested. Again,
inspection of Fig. 2 shows higher removal of KCl at higher temperatures.
Figure 6 shows the mass balances for the 0.2 wt% acid washes. There were
increased losses of organic matter with increased temperature, with greater
hemicellulose losses occurring in the 25 to 37°C step than in the 37 to 50°C
step. Cellulose removal mirrored the hemicellulose removal with increas-
ing temperature, while higher losses of other organics" occurred in the 37
II

to 50°C step. Ash removal also increased with increasing temperature, and
there were no significant losses of lignin.
The effects of additional parameters, including % solids, acid concen-
tration, and the use of whole straw vs straw stems, on the organic compo-

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


214 Thompson et al.

100

-
80

-
iii • Cellulose
o 60
I- 12 Hemicellulose
o o ligninfExtractives
';Ie 40 &:I Ash
Other organics
20 lID Weight loss

o
None 2 4 16 16
%-Solids (wt%)

Fig. 7. Mass balances with increasing % solids for unwashed and washed straw
stems using distilled water at 25° for 4 or 24 h.

nent mass balances are shown for selected experiments in Figs. 7-9. The
values of 5i02 /K20, 5i02 /KC1, and the percentages of the ash represented
by 5i02, K20, and KCl are given in Table 2 for these experiments. Figure 7
compares a 24-h, 15 wt% solids wash and a 4-h, 16 wt% solids wash of straw
stems. Note that distilled water is a slightly more aggressive solvent than
tap water because it has been demineralized; thus, the distilled water
washes indicate the maximum removals possible using water. Equilibrium
was reached by 4 h in all washes. The effect of % solids (solids loading) in
the distilled water washes is also shown in Fig. 7. Increasing the % solids
resulted in decreased removal of both organic matter and minerals. This
occurred whether the straw remained submerged in the bulk liquid or was
only partly submerged (15% solids). This indicates that the process is solu-
bility limited as long as the straw is submerged and may become liquid-
solid contact limited at solids loadings as high as 15% solids.
It was also observed that the differences between mineral removal at
low and elevated temperatures (37 and 50°C; not shown) were not as great
at higher mass loadings, since solubility limits the amount of material that
can be dissolved, rather than the more aggressive solvent characteristics
secured at the higher temperature. Washing with water removed 11.1 % of
the mass of straw in a 4% fully submerged suspension, while only 6.8%
mass was removed in a 15% partially submerged suspension. The ash con-
centration of unwashed straw was reduced to 4.6% in the 4% solids water
wash, but to only 6.7% in the 15% solids wash. A continuous wash that
lowers the effective "loading" even further would overcome the solubility
limitations and may result in higher mineral losses than experienced in
these batch experiments. However, acidic or alkaline washing in a continu-
ous process may present some difficulty in removing the residual wash
fluid from the straw if residual 504 or Na is not desirable. In the specific case
ofNa, this would definitely not be desirable, since both Na and K contribute
to the eutectic composition and lower the ash fusion temperature. The straw

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Removal of Minerals from Wheat Straw 215
Table 2
Si02 /K20, Si02 /KCI, and Percentage of Ash Represented
by Si02, K20, and KCl for Washes Shown in Figs. 7-9
Si02, K2O,
Condition Si02 /KP Si02 /KCI and KCI (wt%)
4-24 h, 25°, stems, H 20 wash,
vary solids-ratios for Fig. 7
Unwashed stems 0.9 0.3 92.9
2 wt% solids, 4-h wash 1.3 32.8 88.0
4 wt% solids, 4-h wash 1.2 4.7 78.5
15 wt% solids, 24-h wash 2.4 3.3 84.2
16 wt% solids, 4-h wash 2.4 3.3 84.2
4 h, 4% solids, 25°, stems,
acid wash-ratios for Fig. 8
Unwashed stems 0.9 0.3 92.9
Owt%H2S04 1.2 11.7 85.7
0.2 wt% H 2SO4 1.3 >75" 83.9
0.4 wt% H 2SO4 2.4 >75 93.5
1.0 wt% H 2SO4 2.8 >75 90.6
24 h, 10% solids, 50°, whole straw,
acid-ratios for Fig. 9
Unwashed whole straw 1.3 0.5 96.4
Owt%H2S04 1.3 35.6 59.0
0.2 wt% H 2SO4 2.3 13.2 63.0
0.4 wt% H 2SO4 2.6 5.6 86.1
1.0 wt% H 2SO4 NIY' NOb NOb
aCI was not detected by EDS in this sample, so the approximate CI detection limit was
used to calculate the minimum possible Si02 /KCl ratio.
"ND, not determined.

stems absorbed about 3.7 times their weight of water, and chopped whole
straw absorbed 4.2 times its weight. Thus, residual 504 or Na might be
difficult to remove efficiently without significant further water washes.
Mass balances for washing of straw stems at 25°C with dilute acid at
various acid concentrations are shown in Fig. 8, while Fig. 9 shows the same
for whole straw at 50°C. Higher acid concentrations were required to reach
the same SiOz/KzO ratios for whole straw as reached for stems at 50°C (not
shown). Still, Si02 /K20 ratios obtained with whole straw approached the
minimum acceptable level of 3.0. However, the whole straw contains larger
amounts of silica (separated from the stem fraction in the selective harvest),
and therefore has greater slagging potential. Both the mineral and organic
content removed increased with higher acid concentrations. The highest K
removal without excessive loss of organic matter was achieved at 50°C with
0.2 wt% H 2S04 in a 4% straw suspension (see Fig. 2). The SiOz/K20 ratio

Applied Biochemistry and Biotechnology Vol. 105-108,2003


216 Thompson et al.

100

--....
80
(ij
• Cellulose
o 60 ~ Hemicellulose

o
o Lignin/Extractives
~ 40
&1 Ash
o • Other organics
IDl Weight loss
20

o
None 0.0 0.2 0.4 1.0
Acid (wt%)

Fig. 8. Mass balances with increasing acid concentration for unwashed and washed
straw stems using dilute H 2S04 at 25° for 4 h, at 4 % solids.

100

--
ca
o
I-
o
80

60 •
rA
o
Cellulose
Hemicellulose
LigninlExtractives
eft. 40 &1 Ash
• Other organics
20 IDl Weight loss

o
None 0.0 0.2 0.4 1.0
Acid (wt%)

Fig. 9. Mass balances with increasing acid concentration for unwashed and washed
of whole chopped straw using dilute H 2S04 at 50° for 24 h at 10% solids.

achieved in this run was approached with whole straw (Si02 /K20 of 2.8) at
10% solids, which was not achievable with stems alone at 10% solids. This
may be because the whole straw starts at a higher Si02 /K20 ratio than
stems alone, since the mechanical stem separation concentrated the alkali
metals relative to the silica in the stems. In any event, lowering the solids
content would probably produce washed straw with an Si02 /K20 ratio
above the desired minimum of 3.0.
Generally, cellulose and hemicellulose concentrations were not sig-
nificantly affected by dilute-acid washes with acid concentrations up to
0.4%. However, Fig. 8 shows significant loss of hemicellulose in a 1% H 2S04
wash. A 5 % H 2S04 wash (not shown) resulted in complete loss of both
cellulose and hemicellulose. In these experiments, the mass balances indi-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Removal of Minerals from Wheat Straw 217
cated a shift from cellulose and hemicellulose to an acid-insoluble fraction,
generally referred to as Klason lignin in the quantitative saccharification
technique (10). It is believed that significant H 2S04 remained within the
straw lignocellulosic matrix during the 105°C drying step. Thus, as water
was removed, the H 2S04 was concentrated, eventually reacting with the
polysaccharides at the elevated temperature and converting them to acid-
insoluble organic matter. Assuming that the acid-insoluble organic matter
was still combustible, the total heating value content may not be signifi-
cantly affected, since the decomposition products were not lost, but con-
verted to the "lignin" category in Fig. 8. Hence, it is clear that acid
concentrations of 1 wt% and higher should not be used if recovery of intact
carbohydrates is desired.
The effect of size reduction is demonstrated in Fig. 9. Since this
chopped (1.9 cm and smaller fragments) straw had not been physically
separated, the initial mineral (ash) content was higher at 11.2%. Size reduc-
tion increased the contact surface area and the mass that could be fully
immersed. Ash removal on washing was not improved, indicating that
solubility constraints were of greater importance than surface contact. The
chopped whole straw was washed in water and various dilute-H2S04 solu-
tions at 6.7 and 10 wt% solids. The mass lost after washing was 14.1% for
the chopped straw at 6.7% loading. This was comparable to previous tests,
in which 15.3% loss for straw stems at 4.5% loading was observed. The
inorganic mineral concentration was reduced from 11.2 to 6.4% in the
chopped whole straw, and from 8.7 to 4.2% in the unchopped straw stems.
It seems clear from the data presented herein that acid is preferred as
a wash solution for removal of alkali minerals from straw for use in flu-
idized-bed combustors or gasifiers or in boilers. Future work includes
testing acid-washed whole straw and physically separated straw stems in
a fluidized-bed combustor to determine the combustion and slagging
properties of the washed straw. In addition, we would like to test the
effect of adding CaO or MgO after washing, since CaO and MgO can
modify the eutectic composition and thereby increase the ash fusion tem-
perature further (2). While CaO or MgO could be directly added to the
straw without washing to achieve an increase in ash fusion temperature,
washing the straw removes large amounts of K, and thus less CaO and
MgO would be required to significantly alter the ash fusion temperature.
Data from these tests, and estimation of eutectic compositions and ash
fusion temperatures, will help in the application of these separations to
fluidized-bed combustion and gasification of biomass and combustion in
boilers. The acid-washed straw will also be tested for ease of drying and
densification as part of a distributed bioenergy system.
Conclusions
Mechanical separation of straw reduced ash and Si02 concentrations
in straw by removing leaves and sheaths and leaving primarily stems.
However, this concentrated K relative to Si02, reducing the Si02 /K20 and

Applied Biochemistry and Biotechnology Vol. 105-108,2003


218 Thompson et al.
Si02 /KCl ratios, which are important indicators of slagging and corrosion
potential. Chemical washes removed much of the soluble K and up to 11 %
of the organic matter. Dilute acid removed both K and Cl but little SiOz,
while dilute alkali removed up to 20% of the Si02, and lesser amounts of
K or Cl than did acid. Elevated temperatures increased both the mineral
removal rate and with sufficient bulk liquid, the amount removed. The
highest Si02 /K20 ratio achieved was 3.2, obtained using a 4-h, 0.2 wt%
H 2S04 wash at 50°C in a 4% straw stem suspension. Increasing the solids
loading minimized loss of organics but also reduced mineral removal.
Washing improved Si02 /K20 and Si02 /KCl ratios for both physically sepa-
rated straw stems and whole straw.

Acknowledgments
We thank the University of Idaho, Aberdeen Research and Extension
Center, for use of their plot-harvesting equipment for the mechanical sepa-
rations. We also thank Dr. Judi Steciak at the University of Idaho and Dr.
Robert Carrington at INEEL for useful discussions on biomass slagging
and combustion. Finally, we thank Tracy Houghton at INEEL, who per-
formed the quantitative saccharification analyses. This work was supported
by the US Department of Energy through the INEEL Laboratory Directed
Research and Development Program under DOE Idaho Operations Office
Contract DE-AC07-99IDI3727.

References
1. USDA NASS. (2001), United States Department of Agriculture, National Agricul-
tural Statistics Service (Website: http://www.usda.gov /nass/).
2. Stultz, S. C. and Kitto, J. B., eds. (1992), Steam: Its Generation and Use, 40th ed., Babcock
& Wilcox, Barberton, OH.
3. Jenkins, B. M., Baxter, L. L., Miles, T. R Jr., and Miles, T. R (1999), Fuel 78, 17-46.
4. Seggiani, M. (1999), Fuel 78, 1121-1125.
5. Weast, R c., Astle, M. J., and Beyer, W. H., eds. (1983), CRC Handbook of Chemistry and
Physics, 64th ed., CRC, Boca Raton, FL, p. B-135.
6. Steciak, J., (2001) personal communication (unpublished data), Department of
Mechanical Engineering, The University of Idaho, Boise, ID.
7. Hess, J. R, Thompson, D. N., Hoskinson, R L., Shaw, P. G., and Grant, D. R (2003),
Appl. Biochem. Biotechnol. 105-108, 43-52.
8. ASTM E1508. (1998), Standard Guide to Quantitative Analysis by Energy Dispersive Spec-
troscopy, American Society for Testing and Materials, Philadelphia, PA.
9. Gavlak, R G., Horneck, D. A., and Miller, R O. (1994), Plant, Soil, and Water Reference
Methods for the Western Region, Western Region Extension Publication 125, University
of Alaska Cooperative Extension Service, Fairbanks, AK.
10. (1994), Plant, Soil, and Water Reference Methods for the Western Region, Far West Fertil-
izer and Agrichemical Association.
11. Saeman, J. F., Bubl, J. L. and Harris, E. E. (1945), Ind. Eng. Chern. 17(1),35-37.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0219/$20.00

Composition and Ethanol Production


Potential of Cotton Gin Residues

FOSTER A. AGBLEVOR,* SANDRA HATZ, AND JESSICA TRUMBO


Department of Biological Systems Engineering,
Virginia Polytechnic Institute and State University,
Blacksburg, VA 24061, Email: fagblevo@vt.edu

Abstract
Cotton gin residue (CGR) collected from five cotton gins was fractionated
and characterized for summative composition. The major fractions of the
CGR varied widely between cotton gins and consisted of clean lint (5-12%),
hulls (16-48%), seeds (6-24%), motes (16-24%), and leaves (14-30%). The
summative composition varied within and between cotton gins and con-
sisted of ash (7.9-14.6%), acid-insoluble material (18-26%), xylan (4-15%),
and cellulose (20-38%). Overlimed steam-exploded cotton gin waste was
readily fermented to ethanol by Escherichia coli K011. Ethanol yields were
feedstock and severity dependent and ranged from 58 to 92.5% of the theo-
retical yields. The highest ethanol yield was 191 L (50 gal)/t, and the lowest
was 120 L (32 gal)/t.
Index entries: Cotton gin waste; steam explosion; characterization;
summative composition.

Introduction
Raw cotton processing generates cotton gin residue (CGR), which is
composed of immature bolls, cottonseed, hulls, sticks, leaves, and dirt.
This material could potentially be used for ethanol production. The major
advantages of this feedstock over other lignocellulosics include high cot-
ton cellulose content and concentration of large volumes of this material
at cotton-ginning plants. Further, conversion of this feedstock to ethanol
could reduce particulate emission, reduce fire hazards from spontaneous
combustion of CGRs, and also create jobs in rural America.
However, because CGR is an agroindustrial byproduct, its chemical
composition varies considerably because of several factors including sea-
son, harvesting, and processing protocols. Several studies have been con-
ducted on CGRs (1-6), some of which showed that the fuel value of the
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 219 Vol. 105-108,2003


220 Agblevor et al.
wastes varied by machine and type of cotton (5). Other studies showed
that the ash content of gin wastes from six cotton gins in Texas ranged
from 11 to 17% while the carbon content ranged from 41 to 43% (6). A few
studies investigated the summative composition of cotton gin waste, but
the samples were collected from only one gin, and therefore the data did
not show between-gin variations (1,2). The amount of gin residue pro-
duced also varied widely between gins and between crop years (3). Esti-
mates of ethanol yield were based on burr trash composition of 40%
cellulose, 30% hemicellulose, and 25% lignin (3). Although some of the
cited studies addressed between-gin fuel value variability, they did not
address variation in summative composition and its impact on potential
bioconversion product yield. In the present study, we analyzed freshly
discharged and old discharged CGRs from five cotton gins in southeast-
ern Virginia in order to assess the impact of ginning practices on feed-
stock composition and yield of ethanol.
Materials and Methods
Sampling and Fractionation of CGR
CGR samples were collected from five cotton gins in southeastern
Virginia on three consecutive days. The cotton gins are located in Empo-
ria, Franklin, Suffolk, Wakefield, and Windsor. The CGR samples were
collected by taking 5-kg samples around the perimeter of the piles as they
were discharged from the gins. These samples were labeled "fresh dis-
charge." Additionally, material that had been piled by bulldozers for
several weeks after discharge were also collected and labeled "old dis-
charge." All samples were transported by truck on the same day to Vir-
ginia Tech, Blacksburg, VA. Because of the continuous spraying of the
material with water during discharge, the samples were wet and thus
were dried at room temperature to equilibrium moisture content and
stored before chemical analysis. About I-kg grab samples of dried mate-
rial were first hammer milled to pass through a 1-mm screen and then
knife milled (Wiley mill model 4) until all the material passed through a
20-mesh screen. These materials were stored at room temperature until
the time of analysis.
About I-kg dry CGR samples from each of the five gins were shipped
to the USDA Cotton Ginning Laboratory (Stoneville, MS) where they were
fractionated into eight fractions consisting of clean lint, hulls, sticks and
stems, grass, seeds, small leaves, and pin trash. About 150 g of each sample
was fractionated according to Shepherd (7).
Determination of Ash
The moisture content of the samples was determined by an oven-dry
method as in ASTM standard method E 1756-95 (8). The ash content of the
CGR was determined using ASTM standard method E 1755-95 (9). Thus,
1 g of knife-milled CGR was weighed in ceramic crucibles that had been
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Composition and Ethanol Production of CGR 221
preheated and ignited at 575°C. The crucible and the CGR were heated at
575°C in a muffle furnace for 3 h and cooled to room temperature in a
desiccator and weighed. The process was repeated until a constant weight
was achieved. The ash content was calculated on a moisture-free basis for
triplicate samples.

Determination of Extractives
The ethanol extractives content of the CGR was determined using
ASTM standard method E 1690-95(10). TheCGRsamples,10gof-20mesh,
were weighed into dry cellulose extraction thimbles and extracted for 8 h
with 350 mL of 95% ethanol in a Soxhlet extraction apparatus. After extrac-
tion, the samples were cooled to room temperature and the residual solids
were vacuum filtered using a Buchner funnel. The filtrate was added to the
ethanol extract and vacuum evaporated to dryness on a Buchii rotary evapo-
rator at 40°C and 84 kPa. The final product was dried overnight in a vacuum
oven, weighed, and the extractives content calculated on an oven-dry bio-
mass basis.

Determination of Acid-Insoluble Residue


The acid-insoluble residue of the samples was determined using
ASTM standard method E 1721-95 (11). Samples of extractives-free CGR
(0.3 g) were hydrolyzed with 72% H 2S04 for 2 h at 30°C. The hydrolyzed
samples were diluted with deionized water to 4% H 2S04 and autoclaved at
121°C for 1 h using a liquid cycle. The hydrolysates were filtered through
preignited ceramic filtering crucibles. The filtrates were collected and used
as stock samples for carbohydrate analysis. The residues were dried over-
night at 105°C in a laboratory oven and weighed. The oven-dried residues
were ashed as described earlier. The acid-insoluble residue content was
calculated on an oven-dry biomass basis.
Determination of Carbohydrates
Carbohydrate content of the CGR samples was determined as alditol
acetates. The acid-hydrolyzed samples saved from the acid-insoluble resi-
due analysis procedure were used for these analyses. The analytes were
prepared according to ASTM standard method E 1821-96 (12). This
method describes the procedure for converting sugars to alditol acetates.
The derivatized sugar samples were analyzed by gas chromatograph
(Shimadzu GC 14A) using a Restek (Bellefonte, PA) Rtx-225 capillary col-
umn (15 m, 0.25-pm ID, 0.2-pm film thickness). The following conditions
were used for the gas chromatographic analysis: carrier gas: helium; total
gas flow rate: 64 mL/ min; column gas flow rate: 0.6 mL/ min; programmed
column temperature: initial temperature of 190°C for 5 min, heating rate of
lOoC/min, final oven temperature of 210°C, total run time of 20 min; injec-
tion port temperature: 240°C; detector: flame ionization at 220°C; sample
size: 3 pL; split ratio: 33:1.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


222 Agblevor et al.
Steam Explosion Treatment of CGR
Steam explosion of the CGR samples was carried out in a 25-L batch
reactor located at the Recycling Laboratory, Thomas M. Brooks Forest Prod-
ucts Center, Blacksburg, VA. The CGR samples were steam exploded at
two severity parameters (3.5 and 4.5) as defined by Overend and Chornet
(13). The samples were treated in two modes: In the first case, the samples
were soaked with their own weight of water and left overnight in covered
lO-gal buckets before the steam explosion treatment. The second set of
samples was steam exploded as received" without any further addition of
1/

water. About 2 kg of each sample was weighed and loaded into a 25-L batch
steam explosion gun. Saturated steam was admitted into the chamber, and
the biomass temperature was raised to 237°C (this usually took 20 ± 5 s). The
run times for the severities 3.5 and 4.5 were 20 and 200 s, respectively. The
steam explosion chamber was washed with water between runs to recover
any residual fiber trapped in the unit. All samples were bagged and stored
in a cold room until the time of analysis or hydrolysis and fermentation.

Fermentation of Steam-Exploded Cotton Gin Waste


Steam-exploded CGR samples were overlimed and hydrolyzed
according to Jeoh and Agblevor (1). Commercial cellulase enzyme prepa-
rations, Spezyme™ CP (94 IU ImL; Genencor) and Econase™ (105 IU ImL;
New York Enzyme Development), were used to hydrolyze the steam-
exploded samples at 50°C, pH 4.7, and 24 h incubation time. About 0.25 mL
of each enzyme preparation was added to 100 g of steam-exploded sample,
which was slurried in 1000 mL of citrate buffer solution.
Escherichia coli KOll (provided by L. O. Ingram, Department of Micro-
biology and Cell Science, University of Florida), a recombinant organism
with genes (pdc, adhB) from Zymomonas mobilis incorporated into its chro-
mosome for enhanced ethanol production, was used for fermentation. The
wild-type microorganism was E. coli ATCC1l303. For long-term storage,
stock cultures were prepared by adding of 20% glycerol (v Iv) to concen-
trated E. coli KOl1 cultures and stored at -70°C.
The growth medium was prepared according to Asghari et al. (14):
5 giL of yeast extract, 10 giL of tryptone, 5 giL of NaCI, 50 giL of xylose,
and 40 mglL of chloramphenicol. Fresh colonies of E. coli KOll from agar
plate (5 giL of yeast extract, 10 giL of tryptone, 5 giL of NaCI, 20 giL of
xylose, and 15 giL of agarose) were used to inoculate 50 mL of the growth
medium in 250-mL Erlenmeyer flasks. The cultures were grown in a
shaker bath at 35°C and 150 rpm for 18 h. The cells were harvested, cen-
trifuged at 11,OOOg under sterile conditions, and resuspended in 2 mL of
deionized water.
The 2-L flasks containing 65 g of enzyme-hydrolyzed steam-exploded
CGR samples were supplemented with 5 giL of yeast extract and inoculated
with E. coli KOll at an optical density of 0.2. After inoculation, the flasks
were flushed aseptically with nitrogen to establish near-anaerobic condi-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Composition and Ethanol Production of CGR 223
Table 1
Fractional Composition of Cotton Gin Waste from Samples
Collected During Two Ginning Seasons
Clean Stick/ Small Pin
Gin lint Hulls stems Grass Seed Motes leaf trash Total
Name (%) (%) (%) (%) (%) (%) (%) (%) (%)

Franklin 10.4 19.7 7.1 0.4 5.6 19.5 30.3 5.0 98.0
Emporia 5.3 35.6 7.1 0.4 12.7 16.1 21.3 0.6 99.1
Windsor 9.0 16.8 3.6 0.2 6.9 23.9 34.6 1.6 96.6
Suffolk 12.5 15.9 5.4 0.3 24.0 18.6 18.5 2.2 97.4
Wakefield 7.1 48.1 6.1 0.4 7.7 15.6 13.9 0.6 99.5

tions and sealed. The initial pH of the fermentation broth was 6.0. Fermen-
tation was carried out at 35°C and 120 rpm for 72 h. There was no pH control
during fermentation and no antibiotic was added to the fermentation broth.
Samples (1.5 mL) of the culture broth were withdrawn at 24-h intervals and
centrifuged at 11,000g for 10 min. The supernatants were collected and ana-
lyzed by gas chromatography for ethanol and sugar contents.

Results and Discussion


Fractional Composition of CGR
Because CGR is a heterogeneous material, fractionation into various
components will aid in the interpretation of the summative compositional
and fermentation data. The fractional composition of the CGR is given in
Table 1. The sticks/stems and grass fractions were relatively uniform,
which indicated that the harvesting methods were quite similar. Clean lint
content ranged from 5 to 12%, seed from 6 to 24%, and hulls from 16 to 48%.
The other fractions of the CGR varied widely among ginning plants prob-
ably because of the different ginning protocols at each plant. This explana-
tion agrees with the findings of Schacht and LePori (6), who reported
significant variation in the ash, carbon, and energy contents of cotton gin
wastes from six cotton gins in Texas.
Mass Balance
The mass closure for the compositional analysis of the CGR varied
considerably within ginning plants and among plants. Mass closures were
as high as 98% for some samples and as low as 69% for others (Tables 2-6).
The large variation in mass closure could be attributed to the heterogeneity
of the feedstocks, the ginning method, and perhaps the sampling method
as well.
The objective of this study was to determine the carbohydrate con-
tent of the feedstock and thus be able to assess theoretical and practical
ethanol yields. The analytical method was therefore optimized for the
carbohydrate fraction at the expense of other fractions such as lipids,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


224 Agblevor et al.
Table 2
Summative Composition of Cotton Gin Waste from Emporia, VA
(moisture-free whole-biomass basis)
Fresh Fresh Fresh Old
Component discharge discharge discharge discharge
(%) 10-21-00 10-22-00 11-26-00 10-2100-00 .

Arabinan 1.84 ± 0.36 2.46 ± 0.28 1.34 ± 0.10 1.65 ± 0.48


Xylan 8.69 ± 0.76 12.56 ± 0.95 6.08 ± 0.95 9.57 ± 0.95
Mannan 1.13 ± 0.04 1.09 ± 0.10 0.89 ± 0.36 1.31 ± 0.06
Galactan 1.65 ± 0.13 1.50 ± 0.18 1.34 ± 0.37 1.77 ± 0.13
Glucan 36.64 ± 0.21 38.54 ± 3.03 30.72 ± 2.40 38.21 ± 0.98
Total carbohydrates 49.95 56.16 40.37 52.51
Extractives 7.7 ± 0.6 8.6 ± 0.2 8.8 ± 0.3 7.6 ± 1.6
Acid-insoluble residue 26.8 ± 2.1 22.3 ± 1.0 22.2 ± 1.1 26.5 ± 1.5
Ash 7.9 ± 0.1 8.6 ± 0.9 8.8 ± 0.8 12.0 ± 0.3
Grand total 92.35 95.66 80.17 98.61

which could constitute an appreciable fraction of the feedstock especially


if it contained cottonseed. The feedstock may also contain residual pesti-
cide and herbicides, uronic residues, acetyl groups, and other extraneous
materials, which were not amenable to the method of analysis. Thus,
samples that contained large fractions of cottonseed could have low mass
closure because the lipids content was not determined and vice versa.

NonCarbohydrate Components
The extractives, acid-insoluble material, and ash contents of the
samples from the five cotton gins are presented in Tables 2-6. Within each
plant, there was no significant difference between the compositions of the
fresh discharge materials on different days. However, the composition of
the old discharge materials from all the gins was significantly different
(p < 0.05) from those of the fresh discharge samples.
The ash contents of the old discharge feedstocks were 15-70% higher
than those for the fresh discharge samples. The wide range of ash compo-
sition may be due to the differences in the microbial degradation of the old
discharge samples from different plants or the method used to move the
piles from the point of discharge. After the initial discharge of the CGR, the
material was moved away from the discharge area with a bulldozer and
compacted. It is plausible that this operation introduced more dirt into the
CGR, which could account for the increased ash content for all old dis-
charge materials.
Microbial degradation could have contributed to the increased ash
content of the old discharge feedstocks. It has been reported (15) that the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Composition and Ethanol Production of CGR 225
Table 3
Summative Composition of Cotton Gin Waste from Franklin, VA
(moisture-free whole-biomass basis)
Fresh Fresh Fresh Old
Component discharge discharge discharge discharge
(%) 10-21-00 10-22-00 11-25-00 10-210D-OO

Arabinan 2.39 ± 0.57 2.59 ± 0.41 2.75 ± 0.26 2.03 ± 0.17


Xylan 10.46 ± 3.54 10.33 ± 0.72 10.62 ± 0.98 7.76 ± 0.47
Mannan 1.28 ± 0.25 0.73 ± 0.04 1.19 ± 0.34 0.95 ± 0.16
Galactan 1.73 ± 0.17 1.24 ± 0.19 1.48 ± 0.19 1.41 ± 0.14
Glucan 30.27 ± 7.31 35.47 ± 2.21 27.75 ± 2.17 30.08 ± 2.50
Total carbohydrate 46.13 50.36 43.79 42.23
Extractives 5.1 ± 1.0 6.4 ± 1.3 9.6 ± 1.2 8.0 ± 0.3
Acid-insoluble residue 24.1 ± 1.0 26.2 ± 1.5 22.9 ± 1.4 22.0 ± 7.6
Ash 11.6 ± 0.3 11.7 ± 0.3 9.6 ± 0.2 13.3 ± 0.9
Grand total 86.93 94.66 85.89 85.53

Table 4
Summative Composition of Cotton Gin Waste from Windsor, VA
(moisture-free whole-biomass basis)
Fresh Fresh Fresh Old
Component discharge discharge discharge discharge
(%) 10-21-00 10-22-00 11-25-00 10-210D-00
Arabinan 1.75 ± 0.39 1.40 ± 0.38 1.32 ± 0.39 1.26 ± 0.50
Xylan 5.41 ± 0.50 6.20 ± 1.45 4.45 ± 1.14 5.23 ± 1.61
Mannan 1.01 ± 0.23 0.92 ± 0.17 1.01 ± 0.04 0.81 ± 0.22
Galactan 1.57 ± 0.12 1.51 ± 0.14 1.79 ± 0.41 1.28 ± 0.15
Gluean 26.95 ± 1.11 30.50 ± 1.55 29.05 ± 0.82 32.65 ± 1.59
Total carbohydrates 36.69 40.53 37.62 41.23
Extractives 8.2 ± 0.5 7.2 ± 0.3 13.9 ± 1.1 6.4 ± 1.8
Acid-insoluble residue 22.4 ± 0.2 25.2 ± 1.6 21.6 ± 1.1 20.6 ± 0.7
Ash 10.4 ± 0.2 9.9 ± 0.1 8.4 ± 0.7 12.3 ± 0.9
Grand total 77.69 82.83 81.52 80.53

hemicellulose component of biomass is more susceptible to microbial deg-


radation than the cellulose and lignin components. Thus, if the increase in
ash was partly due to microbial degradation, there should be an inverse
correlation between the ash and cellulose, or ash and hemicellulose. A plot
of ash vs hemicellulose and ash vs cellulose did not show any correlation
(data not shown). Thus, the increase in the ash content of the old discharge
samples was probably due to the incorporation of dirt into the feedstock
as the result of pile movement and compaction by the bulldozer.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
226 Agblevor et al.
TableS
Summative Composition of Cotton Gin Waste from Suffolk, VA
(moisture-free whole-biomass basis)
Fresh Fresh Fresh Old
Component discharge discharge discharge discharge
('Yo) 11-24-00 11-25-00 11-26-00 11-2400-00

Arabinan 3.03 ± 0.46 2.25 2.91 1.42 ± 0.48


Xylan 11.80 ± 2.53 12.11 13.22 8.24 ± 1.37
Mannan 1.16 ± 0.14 1.04 2.61 0.38 ± 0.04
Galactan 0.23 ± 0.07 1.56 3.17 0.79 ± 0.00
Glucan 26.31 ± 0.78 21.62 26.08 24.22 ± 1.76
Total carbohydrates 42.53 38.58 47.99 35.05
Extractives 10.8 ± 0.9 11.6 ± 0.3 13.2 ± 0.4 10.1 ± 1.2
Acid-insoluble residue 22.3 ± 0.5 22.5 ± 0.3 18.1 ± 0.7 21.9 ± 1.6
Ash 6.8 ± 0.4 8.3 ± 0.1 8.9 ± 0.1 10.3 ± 0.5
Grand total 82.43 80.98 88.19 77.35

Table 6
Summative Composition of Cotton Gin Waste from Wakefield, VA
(moisture-free whole biomass basis)
Component Fresh discharge Fresh discharge Old discharge
('Yo) 11-25-00 10-26-00 11-2500-00

Arabinan 0.69 ± 0.17 0.65 ± 0.11 0.66 ± 0.13


Xylan 3.02 ± 0.39 4.07 ± 0.47 4.29 ± 0.18
Mannan 2.37 ± 0.99 1.33 ± 0.56 2.79 ± 0.15
Galactan 2.26 ± 0.27 1.75 ± 0.43 3.48 ± 0.29
Glucan 25.63 ± 0.50 25.90 ± 1.05 22.57 ± 2.27
Total carbohydrates 33.97 33.70 33.79
Extractives 9.1 ± 0.9 7.5 ± 0.4 8.4 ± 1.2
Acid-insoluble residue 19.4 ± 0.9 19.7 ± 0.7 22.0 ± 0.9
Ash 8.8 ± 0.2 8.1 ± 0.1 14.6 ± 0.4
Grand total 71.27 69.00 78.79

The most surprising data were the high acid-insoluble material con-
tent of the samples (Tables 2-6). The fraction of cotton gin waste that was
insoluble in 72% H 2S04 was comparable with that found in woody biom-
ass (16,17). The acid-insoluble material from woody biomass is normally
classified as lignin. However, it would be erroneous to classify the mate-
rial from the CGR as lignin. Since CGR is a complex mixture of organic and
inorganic materials, there could be other acid-insoluble material apart
from lignin. A probable source of nonlignin acid-insoluble material is the
cottonseed. A typical cottonseed is composed of 32% hull, 23% protein,

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Composition and Ethanol Production of CGR 227
12% fibers, 20% oil, and 14% carbohydrates (18). Our analysis of the cot-
tonseed from the Emporia gin showed that about 34% was acid-insoluble
material (data not shown). However, the hull, which is mostly lignocellu-
losic, constitutes only 32% of the ginned cottonseed. Thus, it is obvious
that the acid-insoluble material is composed of lignin and other condens-
able compounds. It is known (19) that proteins condense and become
insoluble in concentrated H 2S04 • Thus, one could attribute the high acid-
insoluble material content of the cottonseed to the condensation of some
of the protein on the lignin. Similarly, it could be argued that perhaps the
large fraction of acid-insoluble material in the cotton gin residue is a com-
bination of lignin and condensed proteins from the cottonseed.
The 95% ethanol extractives content of the cotton gin waste was rela-
tively low and was comparable with contents for agroindustrial residues
such as wheat straw and sugarcane bagasse (17).
Carbohydrate Component
Carbohydrates are the most important component of the feedstock
with respect to ethanol production. Carbohydrate analyses of the CGR
are presented in Tables 2-6. It is clear from the data that variation in
carbohydrate composition was more pronounced than variation in the
noncarbohydrate fraction discussed earlier. The total carbohydrate con-
tents ranged from 33 to 56%, which is relatively low compared with that
for woody (16) or herbaceous biomass (17). For woody biomass, the total
carbohydrate content ranges from 67 to 82% for softwoods and 49 to 85%
for temperate-zone hardwoods (16). The total carbohydrate content var-
ied within and among ginning plants.
Mannan, arabinan, and galactan contents were very low, as expected
from an agroindustrial residue. Within any ginning plant, these sugars did
not vary much and neither was there much variation among ginning plants.
Because of their relatively low concentrations, their potential influence on
ethanol yield is expected to be smalL
Glucan and xylan components constituted between 80 and 90% of the
carbohydrates. The surprising result was the low xylan content of the feed-
stock, which was lower than those reported for most agroindustrial resi-
dues and hardwoods (15-35%) (17). Since xylan is a typical plant cell wall
component that derives mostly from the seed coats and stems, the large
variation among ginning plants is probably a reflection of the different
ginning protocols, as shown by the data in Table 1. Feedstocks with rela-
tively high seed and mote fractions had higher xylan content. However,
this explanation needs to be investigated further.
Glucan, which derived from both the cotton fiber and the cell wall of
the lignocellulose, was the largest fraction of the feedstocks. However,
glucan was lower than those reported for wood and other agroindustrial
residues (17). Glucan varied widely among ginning plants (20-38%) com-
pared to within ginning plants (Tables 2-6). The variation in glucan among
ginning plants could be attributed to the ginning methods, because some

Applied Biochemistry and Biotechnology Vol. 105-108,2003


228 Agblevor et al.
Table 7
Ethanol Production from Steam-Exploded Cotton Gin Waste at Two Severities
Severity - 3.5 Severity - 4.5a
Ethanol Ethanol Ethanol Ethanol
yield yield yield yield
Sample (% theoretical) (Lit) (% theoretical) (Lit)
Emporia fresh discharged 78.4 155 17.1 31
Franklin fresh discharged 85.8 152 NO NO
Suffolk fresh discharged 92.5 191 21.7 53
Windsor fresh discharged 81.0 133 19.4 40
Wakefield fresh discharged 58.0 120 24.4 52
aND, not determined.

gins had higher cotton lint content than others (Table 1). Consequently, this
could have influenced the glucan content.
Hydrolysis and Fermentation of Steam-Exploded CGR
When some samples of steam-exploded CGR were enzymatically
hydrolyzed and fermented without overliming there was no cell growth
and no ethanol was produced. However, when the steam-exploded sub-
strate was overlimed and hydrolyzed, the E. coli KOll grew very rapidly
and fermented the substrate to ethanol.
Ethanol yields were determined as a percentage of the theoretical yield.
The ethanol yield reported here is a combination of the fermentation of
xylose and glucose by E. coli KOl1. Although other minor sugars were
present in the steam-exploded substrates, it is not known that E. coli KOll
can ferment these sugars into ethanol.
Ethanol yield was a function of both steam explosion severity and the
source of feedstock (Table 7). At the higher severity (4.5), ethanol yields
were very low and similar for all samples from all the cotton gin plants.
The low yields at the high severity were attributed to inhibitory com-
pounds produced during the steam explosion. Although overliming of
lignocellulosic hydrolysates reduces enzyme and microbial inhibitors, it
does not necessarily restore cell conversion efficiency. It has been reported
(J. D. McMillan, personal communication, 2001; [211) that although cells
may grow in overlimed lignocellulosic hydrolysates, the product yield
may be reduced considerably. In the case of the CGR, the samples that
were steam exploded at a severity parameter of 4.5 and overlimed with
calcium hydroxide supported cell growth, but the ethanol yields were
very low. Ethanol yields were reduced from 92.5 to 21.7% for the Suffolk
gin CGR. Clearly, the choice of severity for CGR treatment is very impor-
tant for CGR fermentation and ethanol production.
The highest ethanol yield was achieved for samples treated at a sever-
ity 3.5 and overlimed with calcium hydroxide. It can be seen that the etha-

Applied Biochemistry and Biotechnology Vol. 105-708,2003


Composition and Ethanol Production of CGR 229

nol yields were considerably higher for all feedstocks at this severity com-
pared to the samples at a severity of 4.5 (Table 7). Interestingly, ethanol
yield was different for feedstocks from different cotton gins. This result is
of practical significance because product yield and therefore profitability
could be considerably affected by the source of feedstock.
Ethanol yield was highest (92.5%) for the feedstock from the Suffolk,
or equivalent to 191 Lit of CGR, while that from the Wakefield cotton gin
was the lowest (58%), or equivalent to120 Lit of CGR. Ethanol yield for
most feedstocks was better than those estimated by Beck and Clements (3)
for an acid hydrolysis process.

Conclusion
Production of ethanol from cotton gin waste appeared to be influenced
by several factors including feedstock origin, steam explosion severity,
sample heterogeneity, feedstock composition, and other unknown factors.
To optimize ethanol yield from this resource, some or all of these factors
have to be addressed. Feedstock composition appeared to vary consider-
ably even for samples taken within a ginning season. These feedstocks need
to be analyzed over a few years to establish an average compositional data
on which to base the expected ethanol yield for practical applications.
High steam explosion severities tended to create more inhibitory com-
pounds and these resulted in low ethanol yields. It appears that a severity
of 3.5 is adequate for ethanol production at a reasonable yield. Although
the process was not optimized, the highest ethanol yield for these feed-
stocks was 191 L (50 gal) I t, considerably higher than the 142.8 L (37.8 gal) I
t estimated by Beck and Clements (3) for an acid hydrolysis process.

Acknowledgments
We thank Eric Johnson and Robert Wright, Virginia Polytechnic Insti-
tute and State University, respectively, for assisting in the data collection and
steam explosion. We acknowledge L. O. Ingram, University of Florida, for
providing samples of E. coli KOl1. G. Virgil, Genencor, supplied Spezyme
CP enzyme for the hydrolysis. New York Enzyme Development supplied the
Econase enzyme for the hydrolysis. We also acknowledge MidAtlantic Cot-
ton Gin, Emporia, VA; Commonwealth Gins, Franklin, VA, and Windsor
VA; Suffolk County Gin, Suffolk, VA; and Wakefield Gin, Wakefield, VA for
the collect cotton gin waste. The US DOE through the Southern States Energy
Board and the Southeastern Regional Biomass Energy Program (contract no.
SSEB-SERBEP-2000-VA2-001) funded this project.

References
1. Jeoh, T. and Agblevor, F. A. (2001), Biomass Bioenergy 21,109-120.
2. Jeoh, T. (1999), MS thesis, Virginia Polytechnic Institute and State University,
Blacksburg, VA.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


230 Agblevor et al.
3. Beck, R. S. and Clements, D. L. (1982), in Proceedings of the Symposium on Cotton Gin
Trash Utilization Alternatives, Texas Agricultural Experiment Service, Lubock, TX,
pp.163-181.
4. Brink, D. L. (1981), in Proccedings of the Symposium Cotton Gin Trash Utilization Alter-
natives, University of California Cooperative Extension, Hanford, CA, pp. 2027.
5. Griffin, A. C. Jr. (1974), Paper no. 74-3038, 1974 Annual Meeting of American Society
of Agricultural Engineers, Oklahoma State University, Stillwater, OK.
6. Schacht, O. B. and LePori, W. A. (1978), Paper no. 78-3544, 1978 Winter Meeting of
American Society of Agricultural Engineers, Chicago, IL.
7. Shepherd, J. V. (1972), Agriculture Handbook, USDA Handbook No. 422, The Agricul-
tural Research Service (ARS), Washington, DC.
8. ASTM (1998), in Annual Book ofASTM Standards, ASTM Method E 1756-95, vol. 11.05,
American Society for Testing and Materials, West Conshohocken, PA.
9. ASTM (1998), in Annual Book ofASTM Standards, ASTM Method E 1755-95, vol. 11.05,
American Society for Testing and Materials, West Conshohocken, PA.
10. ASTM (1998), inAnnual Book ofASTM Standards, ASTM Method E 1690-95., vol. 11.05,
American Society for Testing and Materials, West Conshohocken, PA.
11. ASTM (1998), inAnnual Book ofASTM Standards, ASTM Method E 1721-95, vol. 11.05,
American Society for Testing and Materials, West Conshohocken, PA.
12. ASTM (1998), in Annual Book ofASTM Standards, ASTMMethod E 1821-96, vol. 11.05,
American Society for Testing and Materials, West Conshohocken, PA.
13. Overend, R. P. and Chornet, E. (1987), Philos. Trans. R. Soc. Lond. 321,523-536.
14. Asghari, A., Bothast, R. J., Doran, J. B., and Ingram, L. O. (1996),]. Ind. Microbiol. 16,
42-47.
15. Wiselogel, A. E., Agblevor, F. A., Johnson, D. K., Deutch, S., Fennell, J. A., and
Sanderson,. M. A. (1996), Bioresour. Techno/. 56 (1), 103-110.
16. Fengel, D. and Wegener,G. (1989), Wood, Chemistry, Ultrastructure, Reactions, Chapter
3, Walter de Gruyter, New York, NY.
17. Milne, T. A., Chum, H. L., Agblevor, F., and Johnson, D. K. (1992), Biomass and
Bioenergy 2(1-6), 341-366.
18. Hui, Y. H., ed. (1996), in Bailey's Industrial Oil and Fat Products, Fifth Edition, Edible
Oil and Fat Products: Oils and Oilseeds, Chapter 4, Wiley Interscience, New York, NY,
p.159.
19. Agblevor, F. A., Evans, R. J., and Johnson, K. D. (1994),]. Anal. Appl. Pyrol. 30, 125-144.
20. Buhner, J. (2000), MS thesis, Technical University of Dresden, Dresden, Germany.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0231/$20.00

Wood-Ethanol for Climate


Change Mitigation in Canada

PETER J. GRAHAM,* DAVID J. GREGG, AND JOHN N. SADDLER


Canadian Forest Service, Natural Resources.Canada,
580 Booth St., Ottawa, Ontario, Canada, Kl A OE4;
E-mail: pegraham@nrcan.gc.ca

Abstract
The impetus for this paper is Canada's commitment under the United
Nations Framework Convention on Climate Change to reduce national
greenhouse gas emissions as well as reducing our dependency on fossil fuels.
Wood-based ethanol offers an excellent opportunity for greenhouse gas
mitigation due to market potential, an ability to offset significant emissions
from the transportation sector, a reduction of emissions from CO2-intensive
waste-management systems, and carbon sequestration in afforested planta-
tions. While there are technological and economic barriers to overcome, using
wood-biomass as a source of ethanol can be an economically viable tool for
reducing greenhouse gas levels in the atmosphere. This paper examines the
costs and mitigation potential of the production of ethanol from biomass
supplied from industrial wood waste as well as from trees harvested from
afforested land.
Index Entries: Ethanol; greenhouse gas; afforestation; wood waste; economics.

Introduction
Wood-ethanol's contributions to mitigating climate change include its
use as a substitute for fossil fuels and as an octane-boosting gasoline addi-
tive. Also, the wood -biomass to be used as feedstock for ethanol production
can itself be useful in reducing net emissions of greenhouse gases: affores-
tation increases the size of the terrestrial carbon sink and the use of indus-'
trial wood waste replaces other carbon-dioxide-intensive management
methods such as landfill and incineration. Clearly, the production of etha-
nol from wood-biomass merits serious consideration.
The impetus for this paper comes from two related issues: (1) climate
change and the international commitment to its mitigation and (2) the impor-
tance of the reduction of our dependency on fossil fuels for energy. Reducing

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 231 Vol. 105-108,2003
232 Graham et al.
our dependency on fossil fuels, of which there is an essentially finite supply,
would reduce the impact of sudden and major fluctuations in oil and gas
prices on consumers. The incentive for mitigating climate change through
the reduction of greenhouse gas (GHG) emissions is to avoid the resulting
economic, social, and environmental impacts of potential damages.

Climate Change Mitigation Measures


As mentioned above, ethanol production can positively impact climate
change mitigation through both the forestry sector's carbon sink enhance-
ment and the transportation and energy sectors' emissions reduction. A brief
examination of these measures follows.
Forestry MeasureS': Sequestering Carbon
A terrestrial carbon sink such as a forest effectively captures and dis-
poses of atmospheric carbon dioxide (C02) by converting it, through photo-
synthesis, into carbon. The carbon is stored as biomass in four carbon pools
(1): above-ground biomass; dead organic matter; below-ground biomass;
and soil organic carbon. Above-ground biomass consists of tree stems,
branches, and foliage. Below-ground biomass refers to root biomass; dead
organic matter is mostly leaf litter and woody debris; and soil organic carbon
consists of microbiotic organisms in the soil. The success of forestry measures
to sequester carbon must examine their effects on each of these carbon pools.
Afforestation is defined by the United Nations Framework Convention
on Climate Change as the direct human-induced conversion of land that has
not been forested for a period of at least 50 years to forested land through
planting, seeding, and/or the human-induced promotion of natural seed
sources (Article 3.3 of the Kyoto Protocol). So, through planting fast-growing
trees on marginal farmland, for example, afforestation can increase the size
of the terrestrial carbon sink, which in tum reduces atmospheric GHG levels.
As there is a finite amount of land available, this benefit is relatively short-
lived if the trees are just left standing. However, the opportunity for long-
term reductions lies in the use of the afforested biomass as a source of
wood-biomass for energy conversion. By using carbon-sequestering tree
plantations as feedstock for ethanol production, emissions from the produc-
tion and combustion of fossil fuels can be offset within the transportation and
energy sectors.
Transportation and Energy Sector Measures: Reducing Emissions
Terrestrial sinks keep carbon locked up and out of our atmosphere.
However, in order to reduce the level of atmospheric GHGs in the long
term, we must reduce our emissions of them in the first place. The transpor-
tation sector is the largest source of anthropogenic GHG emissions in
Canada, followed by the industry and energy sectors (2). Emissions from
these sectors can be reduced through a number of actions, including the use
of renewable energy, the displacement of fossil fuel use, and improving the
efficiency of energy consumption.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Wood-Ethanol for Climate Change Mitigation in Canada 233
Wood is a renewable resource and the carbon emitted in its conver-
sion to energy is assumed to be fully sequestered by the next rotation of
trees, resulting in zero net emissions (under current carbon accounting
methodologies). Fossil fuels, on the other hand, are non-renewable and
therefore the amount combusted adds to atmospheric GHG levels. Using
biomass from tree plantations to produce ethanol provides the added
benefit of the one-time gain in carbon uptake from initial afforestation.
Net emissions can be further reduced through the use of wood waste that
would have been incinerated or put into landfills, and through the dis-
placement of fossil fuels that would have been burned to provide the
energy previously. Fossil fuel displacement will also be increased through
technological innovations that will make biomass-fuel conversion more
efficient. Estimates of emission savings from fossil fuel displacement
currently range from 1.7 to 9.0 tonnes of carbon per hectare per year
depending on forest type, discount rates, energy conversion efficiency,
and the particular fossil fuel being displaced (3,4). The establishment of
plantations and the operation of biomass systems will also result in an
increase in employment, with most of the jobs created in rural areas.
In conclusion, GHG mitigation would be realized most effectively
through the combination of sink-enhancement and source-oriented mea-
sures. A system of renewable energy production that would use biomass
from afforested land as feedstock would displace the use of fossil fuels in
energy production while increasing the terrestrial carbon sink. Wood-
ethanol production, which potentially involves all of these elements, is an
excellent step toward mitigating climate change.

Mitigative Potential of Wood-Ethanol


In this section, the ethanol industry's potential to reduce atmospheric
GHG levels will be discussed: its technical challenges, its supply of wood-
biomass, and its markets and incentives.
The production of ethanol from cellulosic materials is increasing inter-
nationally, with production plants being established in the United States,
Canada, and Sweden. Production and use of ethanol, as an alternative trans-
portation fuel or as an octane booster, help reduce GHG emissions from
road vehicles and can promote sustainable development and management
of forests and forest industries through waste minimization. This presents
a two-fold environmental advantage as it has been realized thatstabilisation
of atmospheric GHG concentrations requires both reduction of fossil fuel
consumption and preservation and enhancement of carbon sinks and res-
ervoirs, such as forests (5).
Wood-Ethanol Technologies
The technology for converting grain to ethanol has been around for
some time. The impetus for developing wood-ethanol conversion processes
stems from the availability of low-cost wood residues and, more recently,
the desire to reduce GHG emissions. Ethanol is derived from sugars present

Applied Biochemistry and Biotechnology Vol. 105-108,2003


234 Graham et al.
in lignocellulosic materials such as agricultural, hardwood, and softwood
residues. However, owing to the different structural and chemical com-
positions of these three materials, different processes have been developed
to deal with the different forestry feedstocks. Softwoods have greater lignin
content than hardwoods or agricultural residues, and it is the lignin that
initially hinders the processing of the sugars into ethanol.
There are a number of wood -based ethanol processes, each at different
stages of development. While the technology and configuration of each
process may vary the conversion of wood-biomass to ethanol involves the
following basic steps: pretreatment of feedstock, hydrolysis, fermentation,
and ethanol recovery. A report by McCloy and O'Connor (6) examined the
current technologies with a key requirement being their ability to deal with
softwood residues, the dominant feedstock in Canada.
Iogen Corporation, with the assistance of the Canadian federal govern-
ment and a partnership with Petro-Canada Corporation, has constructed an
industrial-scale demonstration plant in Ottawa, Ontario to convert ligno-
cellulosics to ethanol. Iogen uses an enzymatic process designed initially to
process agricultural and hardwood wastes. A co-product of the process is
lignin, which can then be used as starting material in other processes or as a
fuel to produce steam or electricity.
Acid hydrolysis is a process that has been known for over 100 years,
but was abandoned due to the lower production costs of petrochemical-
derived ethanol. Currently, high capital and operating costs limit the de-
velopment of industrial scale facilities. There are at least three different acid
hydrolysis processes in use or in development (6). Another promising pro-
cess involves the gasification of wood. While it is currently in the early
stages of development, the gasification process may prove to be an effective
and efficient technology. The gasification process has the advantage that
many of its technological components are currently used at industrial scales
in other energy applications. Also, in theory, gasification is able to handle
a wide range of feedstock types, including softwoods.
Wood-Biomass Supply for Ethanol Production
The supply of wood-biomass for bioenergy production relies on two
main sources from the forest sector: afforested land, and industrial wood
waste. (Another potential source not covered in this paper is forestry
residue left over at logging sites and log sorting areas.) The availability
and costs of supply from afforestation and wood waste, as well as
afforestation's productivity, are important factors, as they limit the scale
and economic potential of wood-based bioenergy and its contribution to
GHG reduction.
Supply from Afforestation
The supply of biomass from afforested lands is a function of the amount
of land available, the productivity of that land, the growth and yield of the
species planted, and the costs associated with all of these factors.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Wood-Ethanol for Climate Change Mitigation in Canada 235
The physical availability of suitable land is fairly easy to estimate
given government statistics on land and agriculture. These estimates are
not the primary concern, because it is the cost of land, or economic avail-
ability, which restricts the potential scale of afforestation, not its physical
availability (7-10).
How much wood-biomass can be produced on afforested land
depends on the site productivity, the species planted, and the management
regime used. In an afforestation program, site productivity depends on the
quality of the land that the landowner is willing to lease for tree planting.
Fast-growing species are preferred; the rotations must be short, but with
high enough yields to minimize costs. For instance, if the requirements of
these site-specific management regimes are met, hybrid poplar plantations
on marginal farmland may provide economically viable sources of wood-
biomass for energy production. The distance from the plantations to the
conversion facility and the cost of leasing the land are key factors in the
economics of wood-biomass supply. The cost of transportation can restrict
the economic supply of afforested biomass to within a 50 km radius of a
wood-ethanol production facility (10). The opportunity cost of converting
highly productive farmland to tree plantations result in higher leasing rates
for more productive land.
Supply from Industrial Wood Waste
The wood waste discussed in this paper is limited to wood processing
residues, which mainly consist of sawmill and pulpmill residues. The avail-
ability of such wood waste relates explicitly to uncertain annual harvest
and allowable cut levels. Waste/input ratios also vary by tree species, log
condition, type of processing, and end product. Residue supply is expected
to decrease due to developments in technology and process efficiencies
that tend to reduce the amount of waste from the various processing stages.
In addition to decreased residue production, an increase in competition
from other secondary processing industries, such as medium-density fiber-
board, is expected and will continue to reduce the availability of inexpen-
sive wood residues (11-14).
There are currently millions of tonnes of industrial wood wastes that
are being incinerated or dumped into landfills; these wastes are often
given free-of-charge to those willing to bear the expense of transporting
it. However, with the development of secondary industries that use wood
residues in their production processes, these residues are taking on a
value greater than the cost, to the mill, of disposal. The demand for mill
residues depends on the type of waste (e.g., bark, chips, or sawdust) and
the distance of the residues from the secondary industries. Whitewood
chips and sawdust are the most valuable due to demand from medium-
density fiberboard and other engineered wood product industries. They
are also preferred, in the context of this paper, due to their relative ethanol
conversion rates. A decrease in the availability of wood waste due to more
efficient processing, combined with an increase in the number of second-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
236 Graham et al.
ary wood-processing industries and bioenergy systems, will result in an
increase in the value of the wood waste.
As a result of this increase in the value of wood waste, the demand for
biomass from fast-growing plantations will increase. Thus, although the
economic balance may change to favor one over the other, both afforesta-
tion and wood waste are important sources of the supply of wood-biomass
for bioenergy production. The scale of a commercial wood-ethanol produc-
tion facility will depend on the economics of supply and the capacity of
existing transportation infrastructure in the supply area.
Markets and Incentives for Ethanol Production
The main source of information on markets and incentives for ethanol
production in Canada is the McCloy and O'Connor report (6). In it, the
current and projected ethanol markets are discussed along with a number
of incentive mechanisms that may assist the industry's growth in Canada.
Comparisons to the United States are made due to the current level of
development of the ethanol industry in that country.
Markets
Currently in Canada, about 175 ML of ethanol are being produced
from grain. No industrial wood-ethanol plants are presently operational.
Total ethanol production represents only 0.5% of the 35 GL of gasoline sold
in Canada each year, and only 5% of the potential market if all gasoline was
blended with 10% ethanol. The wholesale price of ethanol is estimated at 40
cents per liter in Alberta and 43 cents in British Columbia, including federal
and provincial tax incentives (6). The wholesale price of gasoline is approx
21 cents per liter.
Ethanol is used by gasoline companies in six provinces and the Yukon
Territory. Some companies, such as Mohawk Oil, use it for its clean-burn-
ing properties to address a particular market niche, while other companies,
such as Petro-Canada, use it for its octane-boosting properties. However,
ethanol must compete with other octane-boosting agents such as MMT
(methylcyclopentadienyl manganese tricarbonyl) and MTBE (methyl ter-
tiary butyl ether), currently used by refiners. In light of the current health
and environmental concerns related to the use of MMT and MTBE (dis-
cussed in the following pages), ethanol presents an attractive and effective
alternative to these additives, as well as contributing to the reduction of
harmful emissions from gasoline combustion. These benefits, combined
with relatively low costs, indicate the opportunity for the ethanol industry
to take a larger share of the fuel additive market and increase its share of the
transportation fuels market as well.
An examination of US ethanol markets provides an indication of the
potential for growth in the Canadian industry given appropriate incentives.
The US currently produces about 6.5 GL/yr of ethanol from 42 pro-
duction facilities. Almost all US gasoline retailers use ethanol for at least
part of the year. Domestic US consumption accounts for approx 5.2 GL.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Wood-Ethanol for Climate Change Mitigation in Canada 237
With annual gasoline consumption of 450 GL, and most of the ethanol used
in 10% blends, ethanol is used in approx 12% of the gasoline sold in the US.
The potential short-term growth in the US ethanol market will
depend mostly on changes in state legislation to favor ethanol use. Also,
an increase in production of E85 vehicles (vehicles that run on 85% ethanol
blends) will increase demand. Studies looking at the role of ethanol in
climate change mitigation scenarios project an annual demand of 36 GL by
2015 (15), or 145 GL by 2020 (16). These projections assume the success of
research and development programs in reducing the cost of ethanol pro-
duction, and a substantial increase in the sales of E85 vehicles. The US
Department of Energy (17) projects the production of cellulose-based etha-
nol to be between 20 and 28 GL by 2010.
Incentives
Health and environmental concerns are leading incentives to reduce
GHG and other noxious gas emissions. Ethanol can play an important part
in these reductions.
As previously mentioned, MMT and MTBE are currently widely used
instead of ethanol as octane-boosting agents in gasoline. The manganese
component of MMT is a known neurotoxin at high exposure levels, but
current research on the effects of low-level exposure is inconclusive. (The
Canadian government imposed a ban on the interprovincial trade ofMMT,
but this was later rescinded due to the threat of legal challenge by manu-
facturers and provincial concerns.) MTBE is a more expensive octane
booster than MMT but has the added benefits of lowering carbon monoxide
(CO) and volatile organic compound emissions (VOCs). However, MTBE
has been linked to groundwater contamination in the US and is a significant
source of GHG emissions, as it is a by-product of the natural gas industry.
The combustion of gasoline releases VOCs, nitrogen oxides (NO x), CO,
and particulate matter (PM). The combination of VOCs and NO x with sun-
light results in the formation of low-level ozone, the main component of
smog. CO is a deadly poison and the inhalation of fine particulate matter
(PM2 .S) is a serious health concern. In 1990, Health Canada and Environment
Canada estimated that 13% of all PM2.5 emissions in Canada were from the
transportation sector. A blend of 10% ethanol in gasoline has the potential to
reduce VOCs by up to 10%, and reduce CO emissions by 8-30%. The use of
ethanol-blended gasoline has the potential to significantly reduce particulate
emissions, although this claim has yet to be fully substantiated with data (6).
If blended at the refinery, as opposed to "splash blending" outside the refin-
ery, ethanol-blended gasoline can reduce NOx emissions as well, thus further
reducing the potential for smog.
A number of incentive mechanisms exist in both Canada and the US
that should lead to an increase in the production of ethanol. First, the US
will be discussed; there, these mechanisms include legislation to stimulate
demand, tax reduction on ethanol-blended gasoline to stimulate produc-
tion, and the provision of guaranteed loans for capital costs to reduce risk.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


238 Graham et al.
Table 1
Summary of Canadian Incentive Programs (6,18,19)

Province Incentive Conditions


Quebec 19.76 cents per liter ethanol Not yet proclaimed (as of 1999)

Ontario 14.7 cents per liter ethanol


Manitoba 2.5 cents per liter of fuel Ethanol produced in
containing a minimum of Manitoba from Canadian
10% ethanol biomass (grain or wood)
Saskatchewan 1.5 cents per liter of fuel Not yet proclaimed (as of 2002)
containing a minimum of
10% ethanol, plus 4.5 cents
per liter of ethanol
infrastructure incentive
Alberta 9.0 cents per liter ethanol
British Columbia 11.0-15.0 cents per liter fuel Applies to blends containing
a minimum 85% ethanol

Legislation requiring minimum oxygen content levels in gasoline has


been passed in a number of states to deal with air quality problems, specifi-
cally to reduce CO levels in the winter and ground-level ozone in the sum-
mer. Minnesota has been very effective at increasing ethanol production in
that state by requiring that almost all gasoline sold in the state contain etha-
noL The US federal government has offered reduced taxes on ethanol-
blended gasoline since the 1970s (prompted by the oil crises), and the
incentive is currently at about 21.6 cents per liter (Cdn). Individual states
have provided additional incentives (up to 3.25 cents/liter [Cdn] in Alaska)
directly to ethanol producers to encourage production in their states. Addi-
tional incentives in the form of special loans, property tax assessments,
income tax credits, and preferential purchasing policies can assist in the
financing of individual projects.
In Canada, there is currently a Federal Excise Tax exemption for etha-
nol produced from biomass (wood or grain). In addition, a National Biom-
ass Ethanol Program administered by the Farm Credit Corporation
provides a line of credit to qualified ethanol manufacturers as a means of
rescheduling their long-term debt, thereby reducing risk related to the
potential for future increases in taxes and feedstock prices or a decrease in
oil prices. However, this program is limited in terms of the ethanol produc-
tion volumes it can cover (6). Five of the provinces have incentive programs
for ethanol production as summarized in Table 1. Two large grain-ethanol
plants have been financed in Ontario and Quebec through the provision of
agreements that guarantee the level of tax incentives for a number of years,
thereby reducing the level of uncertainty regarding future changes to gov-
ernment policies and programs.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Wood-Ethanol for Climate Change Mitigation in Canada 239
The difference between the level of federal incentives appears to be the
main reason why Canada's ethanol industry has not kept pace with the US
As McCloy and O'Connor (6) point out, "the Canadian fuel industry is much
more national in scope than the US industry and prefers single marketing
and product programs across the country" (p. 48). For this reason, provincial
incentives are not sufficiently effective in developing the ethanol industry. A
doubling of federal incentives, combined with financing assistance at the
provincial level, would likely result in a rapid increase in investment (most
likely from the oil companies themselves) in ethanol production.
To conclude this section, then, wood-ethanol's potential to help miti-
gate climate change is considerable. Given that 1 L of ethanol contains 65%
of the energy of 1 L of pure gasoline, and accounting for relative efficiencies,
1 L of 10% ethanol-blended gasoline replaces 0.8 L of gasoline. The lignin
co-product from the wood-ethanol conversion process can be burned to
generate power, thereby replacing power generated from natural gas or
used as a source material for other (perhaps chemical production) pro-
cesses. This combined reduction in the use of gasoline and natural gas
translates to about 2.7 kg CO2 /L of avoided emissions through the produc-
tion of wood-ethanol. By 2020, ethanol production is estimated to reach 525
ML / yr, resulting in 1.4 Tg of avoided CO2 emissions per year (6). However,
the potential GHG reductions from the increased production of ethanol
will depend greatly on the development of additional tax incentives and
the development of the technology.
The adoption of stronger incentive programs, particularly at the fed-
eral government level, would provide the industry with the initial push it
requires to justify the high capital cost and risks associated with wood-
ethanol production. The potential market growth for ethanol is likely to be
maintained due to the general population's level of concern for the envi-
ronment and due to efforts to meet Canada's Kyoto commitment through
reductions in GHG emissions from the transportation sector. Whereas the
potential damage resulting from climate change may be perceived by many
as not critical enough to require immediate action, the effects of decreased
air quality in rapidly growing cities are pushing policy and law makers to
demand cleaner sources of transportation fuel.
Discussion
The use of wood-biomass from afforested lands and industrial wood
waste as a feedstock for ethanol production has the potential to be an eco-
nomically viable tool to reduce greenhouse gas levels in the atmosphere.
Afforestation as a supply of wood-biomass for ethanol production has the
combined benefit of increasing the terrestrial carbon sink and offsetting the
production and combustion of fossil fuels, thereby reducing net atmo-
spheric greenhouse gas levels.
As a forestry measure for mitigating carbon-dioxide emissions
through a change in land use, afforestation has a number of potential ben-
efits. The primary benefit is an increase in the size of the terrestrial carbon
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
240 Graham et al.
sink. The use of afforested biomass for renewable energy has the advantage
of being emissions-neutral under the Kyoto Protocol, and further green-
house gas reductions are realised through the displacement of fossil fuels.
There are currently many millions of tonnes of industrial wood waste
being disposed of by incineration or in landfills. The conversion of this
waste wood into energy products provides an opportunity to avoid these
current carbon-dioxide-intensive waste management practices while off-
setting emissions from the production and consumption of fossil fuels.
Future reductions in the availability of low-cost wood waste will in-
crease the demand for wood-biomass from fast-growing plantations. The
potential for afforestation to meet this demand is limited mainly by the cost
and productivity of marginal farmland and its suitability for growing trees.
Considering the conversion of wood-biomass to energy products,
wood-ethanol stands out for three main reasons. First, its use as a transpor-
tation fuel derived from renewable wood-biomass reduces the net green-
house gas emissions through the reduced consumption of gasoline, the
increase in the terrestrial carbon sink from bioenergy plantations, and the
avoided emissions from the incineration and landfilling of wood waste. Sec-
ond, ethanol has a number of health and environmental benefits, including
improving air quality, in addition to mitigating greenhouse gas emissions.
The third benefit of ethanol production is that, as it gains a share of the
transportation fuel market, the consumer becomes less vulnerable to drastic
changes in oil and gas prices (as experienced in the winter of 2000-2001).
To date, most wood-biomass conversion systems burn the lignin
co-product for process steam and electricity generation to power the con-
version process. If more lucrative markets for the lignin co-product of the
wood -ethanol conversion process can be found (within the chemical ind us-
try, for example), the economics of production would improve. The use of
lignin-based chemicals could replace current petroleum-based chemicals,
thereby further reducing the use of fossil fuels.
For the wood-ethanol industry, there remains considerable risk for
investors in this new technology due to high capital costs, technological
obstacles, and market barriers. In order for the industry to develop, the fed-
eral and provincial governments will have to create incentives beyond those
that currently exist. Government subsidies aimed at overcoming the risks
associated with this new technology can be justified on the basis of carbon
gains. An increase in federal incentives in the form of tax concessions on
ethanol-blended fuel, combined with provincial incentives designed to help
overcome the large capital costs and associated risks, will provide the indus-
try with the necessary boostto get up and running. Government can also help
promote the industry through the development of carbon markets in which
ethanol producers would be allowed to sell carbon offset credits. In the United
States, incentives such as tax benefits, loan guarantees, and regulations have
led to the development of a strong ethanol industry. The market potential in
the US and Canada is very large and could contribute significantly to eco-
nomic growth over the next few decades. Canada can gain from a successful

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Wood-Ethanol for Climate Change Mitigation in Canada 241

ethanol industry not only reduced greenhouse gas emissions, but also an
increase in economic activity, particularly in rural areas. To the consumer,
the debate over the future availability of fossil fuels is not as much of an issue
as the effects of sudden, large jumps in oil and gas prices. Buffering the
demand for gasoline through the addition of ethanol in the transportation
fuel market will reduce the sensitivity of prices at the pump.
As a "no-regrets" option, the production of ethanol from afforestation
and wood waste is economically viable; through both the forestry sector's
carbon sink enhancement and the transportation and energy sectors' emis-
sions reduction, ethanol production can help to mitigate climate change.

Acknowledgments
This paper is based on research that was funded by the Sustainable
Forestry Management Network. The views expressed in this paper are
not necessarily those of the Government of Canada or the Canadian For-
est Service.

References
1. Apps, M. J., Kurz, W. A., Beukema, S. J., and Bhatti, J. S. (1999), Environmental Science
& Policy 2(1), 25-41
2. van Kooten, G. C. and Hauer, G. (2001), Canadian Public Policy 27(3),267-279.
3. Wright, L. L., Cushman, J. H., and Martin, S. A. (1996), in Forests and Global Climate
Change, vol II, Sampson, R.N., and Hair, D., eds., Chapter 8.
4. vanKooten, G. c., Stennes, B., Krcmar-Nozic, E., and van Gorkom, R. (1999), Canadian
Journal of Forestry Research 29(11), 1669-1678.
5. Lashof, D. and Hare, B. (1999), Environmental Science and Policy 2,101-109.
6. McCloy, B. W. and O'Connor, D. V. (1999), Wood-Ethanol Opportunities and Barriers.
Prepared for the Forest Sector Table, National Climate Change Process.
7. van Kooten, G. c., Krcmar-Nozic, E., Stennes, B., and van Gorkom, R. (1999), The
Forestry Chronicle 76(1),165-172.
8. Stennes, B. (2000), Carbon Uptake Strategies in the Western Boreal Forest Region ofCanada:
Economic Considerations. PhD thesis. Department of Forest Resource Management,
University of British Columbia, Vancouver, Canada.
9. Suchanek, P., Shaikh, S. L., and van Kooten, G. C. (2001), Carbon Incentive Mechanisms
and Land-Use Implications for Canadian Agriculture. Sustainable Forest Management
Network Working Paper 2001 (see ref. 6).
10. Graham, P. J., (2001), An Economic Analysis of Fossil Fuel Substitution for Climate
Change Mitigation, MF thesis. Faculty of Forestry, University of British Columbia,
Vancouver, Canada.
11. Canadian Forest Service (CFS) (1999), Canada's Wood Residues: a Profile of Current
Surplus and Regional Concentrations, Draft report prepared for National Climate
Change Process, Forest Sector Table. Canadian Forest Service, Industry, Economics
and Programs branch.
12. Bronson Consulting Group (1999), Demand for Wood Residue for Non Energy Products.
Report prepared for National Climate Change Process, Forest Sector Table.
13. Skog, K. E., Marcin, T. c., and Heath, L. S. (1996), in Forests and Global Climate Change,
vol II, Sampson, R.N., and Hair, D., eds., Chapter 12, pp. 209-215.
14. Rinebolt, D. C. (1996), in Forests and Global Climate Change, vol II, Sampson, R.N., and
Hair, D., eds., Chapter 6, pp. 117-129.
15. Sheehan, J. (1998), The role of bioethanol in global climate change, in Proceedings of the
1998 National Conference on Ethanol Policy and Marketing, Albuquerque, New Mexico.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


242 Graham et al.
16. Lynd, L. R. (1996), Overview and Evaluation of Fuel Ethanol for Cellulosic Biomass: Tech-
nology, Economics, the Environment, and Policy. Annual Review Energy Environment.
17. US Department of Energy (1998), Scenarios of us Carbon Reductions: Potential Impacts
of Energy-Efficiency and Low-Carbon Technologies by 2010 and Beyond.
18. McMillan, D. (2002), Ethanol Blueprint Unveiled, Western Producer, Saskatoon,
Canada, February 28, 2002.
19. Government of Saskatchewan (2002), Greenprint for Ethanol Production to be Developed.
News Release, Regina, Canada, May 24, 2002.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


SESSION 2
Microbial Catalysis and Metabolic Engineering
Session 2

Microbial Catalysis
and Metabolic Engineering
NANCY W. Y. Hol AND FRANCISCO F. ROBERT02

1Laboratory
of Renewable Resources Engineering,
Purdue University, West Lafayette, IN; and
21daho National Engineering and Environmental Laboratory, Idaho Fall, 10

Thirty-seven outstanding papers related to the application of natu-


rally occurring or genetically modified microorganisms as catalysts for the
conversion of sugars from various feedstocks, particularly from cellulosic
biomass, to industrial products were selected for presentation in the 24th
Symposium on Biotechnology for Fuels and Chemicals. As in previous years, for
a variety of reasons, many excellent papers presented at the symposium
were not published in this volume. For the readers who did not attend the
Symposium, the complete program can be viewed at the websites provided
in the Introduction to the Proceedings (this volume). Among the submitted
papers, seven were selected for oral presentation in this session and the rest
were presented in the poster Session. Approximately half of the papers
were related to biofuel, particularly cellulosic ethanol production, and the
remainder dealt with the production of chemicals. The selection of papers
for oral presentation was based on the criteria described below. In general,
papers reporting innovative ideas for cellulosic biomass conversion and
the use of new cutting-edge technologies for the study of microbial cataly-
sis were preferentially selected for oral presentation. In addition, a few
papers regarding recent advances in metabolic engineering of the Saccha-
romyces yeast for fermenting cellulosic sugars to ethanol were also selected
for oral presentation. More papers submitted to this year's Symposium
were related to further improving metabolically engineered Saccharomyces
yeast for cellulosic ethanol production than any other microorganism. Since
the attempt to metabolically engineer the Saccharomyces yeast to ferment
xylose was initiated more than 20 years ago, great progress has been made
on this difficult and seemingly doomed task. Thanks to persistent scien-
tists, genetic engineering of the Saccharomyces yeast for cellulosic ethanol
production was ultimately successful. Further validating this research, the
US Department of Energy recently recommended that the metabolically

"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 245 Vol. 105-108,2003


246 Introduction to Section 2
engineered Saccharomyces yeast is still the choice microorganism for the
production of cellulosic ethanol, owing to its robustness and proven indus-
trial record for fermenting glucose to ethanol.
Other key microbial presentations focused on the possibilities of" con-
solidated bioprocessing" (using one fermentation to perform the complete
hydrolysis to fermentation), further platform biochemicals (e.g., succinic
acid), and novel microbes (marine organisms). Dr. Chatterjee of CoDexis/
Maxygen describe R&D and applications of gene shuffling to create new
and variant pathways. These presentations show the value to exploiting
both traditional microbes, such as yeast, as well as new capabilities.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0247/$20.00

Saccharification of Marine Microalgae


Using Marine Bacteria for Ethanol Production

MITSUFUMI MATSUMOTO/,1,3 HIROKO YOKOUCHI,l


NOBUKAZU SUZUKI/ HIROSHI OHATA,J
AND TADASHI MATSUNAGA1

Department of Biotechnology, Tokyo University of Agriculture


1

and Technology, 2-24-26, Naka-cho, Koganei, Tokyo, japan,


E-mail: mitsufumi-matsumoto@jpower.co.jp;
2Chemical and Insulation Technology Group,
Power and Industrial Systems R&D Center, Toshiba Co. Ltd.,
Tsurumi-ku, Yokohama, 230-0045, japan; and
3New Energy and Technology Development Department
Electric Power Development Co. Ltd., Chuo-ku, 104-8165, Tokyo, japan

Abstract
The saccharification of marine microalgae using amylase from marine
bacteria in saline conditions was investigated. An amylase-producing bacte-
rium, Pseudoalterimonas undina NKMB 0074 was isolated and identified. The
green micro alga NKG 120701 was determined to have the highest concentra-
tion of intracellular carbohydrate and was found from our algal culture
stocks. P. un dina NKMB 0074 was inoculated into suspensions containing
NKG 120701 cells and increasingly reduced suspended sugars with incuba-
tion time. Terrestrial amylase and glucoamylase were inactive in saline sus-
pension. Therefore, marine amylase is necessary in saline conditions for
successful saccharification of marine microalgae.
Index Entries: Saccharification; marine algae; marine bacteria; amylase;
ethanol; biomass.

Introduction
Marine micro algae living in marine environments have been studied
for biologic production of useful material such as fatty acids (1,2), bioactive
compounds (3-5), hydrogen (6,7), and polysaccharides (8) for three
decades (9). Marine microalgae can produce these materials using only
solar conversion and CO2 (10-12).

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 247 Vol. 105-108,2003
248 Matsumoto et al.
Application of high carbohydrate-producing marine microalgae can
generate an alternative biomass resource for ethanol production. So far,
biomass resources for ethanol production commonly have been sugarcane
and sweet corn (13). As other biomass resources, wood chips (14), xylose as
sugar sources (15-17), culled apple juice (18), and raw wheat flour (19) has
been used.
The production of ethanol by fermentation from biomass consisting of
carbohydrates was composed of two processes such as saccharification and
fermentation. Saccharification requires enzymes to hydrolyze carbohy-
drates prior to fermentation. It is primarily performed using terrestrially
derived amylase for ethanol production (20,21). The utilization of marine
biomass would require desalinization if a terrestrially sourced enzyme
were to be used. Therefore, development of a system utilizing amylase
from a marine source would be beneficial. In the present study, we inves-
tigated the saccharification of marine microalgae using amylase derived
from a marine bacteria.

Materials and Methods


Cell Culture
Marine bacteria were maintained in marine YS medium plates con-
taining 10 giL of potato starch and 1 giL of yeast extract in artificial
seawater. The artificial seawater was purchased from Senjyu. Marine
micro algae were maintained in BG-11 medium with 3% NaCI (22) under
a light intensity of 40 JA.E/(m2·s).
Isolation and Screening of Amylase Producing Marine Bacteria
Samples of sand, mud, and sea grass from the Japanese coastline
(Tateyama, Chiba) were collected. Samples were diluted with sterilized
artificial seawater. Aliquots were inoculated on minimum medium plates
(10 giL of starch in artificial seawater). The plates were incubated for
2 d at room temperature (24-26°C). Colonies were picked up and trans-
ferred to marine YS medium. Pure strains of the bacteria were isolated by
repeated plating. The bacteria strain was determined by 16S rDNA
sequence analysis.
Marine bacteria were inoculated at room temperature into L glass tubes
containing 10 mL of marine YS medium (OD66O: 0.05) and shaken at 100 rpm.
Culture suspensions (1 mL) were harvested after 24 and 48 h. The cells were
removed by centrifuging at 8,000g for 5 min, and the supernatants were
measured for glucose using an enzyme-based F-kit (Boehringer).
Amylase Activity of Marine Bacteria
Amylase activity was determined by measuring reduced sugars from
starch. The assay mixture contained 0.7 mL of 1.0% soluble starch in artifi-
cial seawater and 0.3 mL of cell-free culture supernatant. The assay solution
(n =3) was incubated for 5 min at room temperature. Reduced sugars were

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Amylase Activity of Marine Bacteria 249
determined by the dinitrosalicylic acid method using glucose as the stan-
dard (23). One unit of activity is defined as the amount of enzyme per
milliliter that produces 1 !lmol of reduced sugars from starch in 1 min.
Cell growth was monitored by OD660 • Protein production was detected
by the Lowry method (24). The protein profiles in YS medium were ana-
lyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-
PAGE) and stained with silver.
Screening of High Starch-Containing Marine Microalgae
Marine microalgae were cultured in aerated 500-mL hexagonal flasks
with illumination (40 !lE/[m2·s]) at room temperature. The carbohydrate
content from 76 strains was determined using the phenol-sulfuric acid
method. Dry cells of 0.5 mg (n = 3) were used and evaluated as reduced
sugars per gram of dry cell.
Saccharification Using Marine Bacteria and Bioethanol Production
Fresh algal cells of 1.82 g wet wt were centrifuged and resuspended
in 10 mL of marine YS medium without starch. Cells were mechanically
broken using a vortex containing silica beads. The algal residues were
autoclaved at 121°C for 10 min. Marine bacteria (0.14 g wet wt) were inocu-
lated into 10 mL of this solution and incubated with shaking (100 rpm) at
room temperature. The samples (n =3) were harvested at 12 and 24h, and
the reduced sugars in culture suspension was measured by the method
mentioned previously.

Results
Glucose Production from Starch by Marine Bacteria
The bacteria of 191 strains were isolated from the marine environ-
ment. The isolated strain NKMB 0074 showed the highest glucose concen-
tration (Table 1). Most of the tested strains had little glucose concentration.
The glucose concentration of NKMB 0074 after 24 h of incubation was
0.59 giL and reached 1.85 giL after 48 h.
Using 165 rDNA sequence analysis, NKMB0074 was determined to
be in the Pseudoalteronomas genus with 99% homology. The species was
determined by 165 rDNA phylogenic analysis and was identified as
Pseudoalteronomas undina.
Amylase Activity of Marine Bacteria
Amylase activity of P. undina NKBG 0074 was defined in artificial
seawater. After 1 dayofincubation, amylase activity was 51.18 ±4.35 U Img
of protein (Table 2). The a-amylase of Porcine pancreas and glucoamylase of
Rhizopus sp. were used for comparative study. The two enzymes were found
to be inactive in artificial seawater. Marine amylolytic enzymes were pre-
served in saline conditions and could therefore be used for saccharification
of marine algae.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


250 Matsumoto et al.
Table 1
Glucose Concentration in Culture
Supernatant of Marine Bacteria
Glucose concentration (mg/L)a
Strain 1d 2d
NKMBOO06 0.86 698.1
NKMBOO18 ND 61.3
NKMB0019 ND 15.6
NKMB0021 2.6 10.4
NKMB0064 8.6 8.6
NKMB 0066 14.0 10.4
NKMB0069 1.7 19.1
NKMBOO72 ND 65.7
NKMB0074 592.7 1855.9
NKMB0085 ND 25.9
NKMB0087 6.0 494.2
NKMB0091 40.6 16.4
NKMB 0093 605.0 1131.8
NKMB00101 58.8 631.6
NKMB00106 125.3 875.2
NKMB 00130 150.6 94.2
NKMBOO144 ND 186.6
Others: 173 strains ND ND or >8.6
Total strains = 190
aND, not determined

Cell Growth, Glucose Accumulation and Specific Protein Production


The growth of P. un dina NKMB0074, glucose accumulation, and pro-
tein profiles in YS medium was analyzed. Glucose accumulation in the
supernatant started after 18 h of incubation, and cell growth reached
stationary growth after 24 h (Fig. 1). Glucose accumulation started at the
beginning of stationary growth. This result was expected that glucose
consumption of bacteria cell was decreased. Therefore, the glucose con-
centration in suspension was increased. The final glucose concentration
reached 2.3 giL after 48 h of incubation.
Protein concentrations in the supernatant were determined to increase
from 18 h of incubation, the same as the onset of glucose accumulation
(Fig. 2). Total protein concentrations were up to 1 giL at the end of incuba-
tion and were related to glucose accumulation.
The protein profile in YS medium was investigated by SDS-PAGE,
and the production of a specific protein (86 kDa) was observed (Fig. 3).
Furthermore, only the 86-kDa protein increased with incubation time.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Amylase Activity of Marine Bacteria 251
Table 2
Enzyme Activity of P. undina NKMB0074 in Artificial Seawatera
Sample Amylase activity (Db /mg protein)
Culture supernatants 51.18 ± 4.35
a-Amylase C 0.59 ± 0.02
Glucoamylased 0.94 ± 0.04
-Suspension at 1 d of culture was used.
bOne unit of enzyme activity was defined as the amount of enzyme
that released 1 ~mol of reducing sugars/ (min·mL) of culture suspension.
ea-Amylase was obtained from P. pancreas and activity showed
20,500 U / mg of protein.
dGlucoamylase was obtained from Rhizopus sp. and showed 38.1 U / mg
of protein.
2.5 3500

3000 ~
2.0 §
'""'
0
ID 2500 §
ID 'p
Cl 1.5 <:<j

2- 2000 tl
5
-5 ~
~ 1.0 1500 0
0 u
0 1000
11)

'"0
u
0.5 ::l
500 6
0.0 0
0 10 20 30 40 50 60

Time (h)

Fig. 1. Time course of growth of P. undina NKB 0074 and glucose concentration in
medium. ( 0 )Growth; ( <> ) glucose concentration.

3000 125
'""'

----
'a.
~ 2500 8
E 100
=
'-' bI)
0 2000 3
'i 75 =
0
i§ 'p
11) 1500 <:<j
.t:;
~
0 50 =~
0=
u
11) 1000
'"0
u
u
::l 25 .5
500 B
6
0
£
10 20 30 40 50

Time (h)

Fig. 2. Time courses of glucose and protein concentrations in medium when


P. undina NKBG0074 was incubated for 48 h. (0 )Glucose concentration; ( <» protein
concentration.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


252 Matsumoto et al.

96
67

43

30

20

Fig. 3. Extracellular protein analysis by 50S-PAGE of different incubation times.


Lane 1, standard; lane 2,1 h; lane 3,6 h; lane 4,12 h; lane 5, 18 h; lane 6, 24 h; lane 7,
36 h; lane 8, 48h. Twenty micrograms of protein of was applied.

Saccharification of Marine Microalgae Biomass


The reduced sugars from 76 marine micro algae are given in Table 3.
The green alga NKG 121701 strain was found to contain the highest reduced
sugars, totaling >50% of dry cell weight. The other strains were <40%.
Therefore, the green alga NKG 121701 was used for further work.
P. undina NKMB0074 was inoculated into suspensions containing NKG
120701 cells. The reduced sugars in suspension increased with increasing
time of incubation (Table 4). After 24 and 48 h of incubation, reduced sugars
were determined at concentrations of 0.9 ± 0.28 and 2.82 ± 0.62 mg/mL.
After 48 h, there was a 38% reduction in microalgae biomass, correlating to
a 1.82-1.13 g in 10 mL. Of this, 4.1 % of NKG 120701 biomass was converted
to reduced sugars. Saccharification of potato starch and microalgae using
terrestrial amylase and glucoamylase was also tested. However, no saccha-
rification was observed in saline conditions (Table 4).

Discussion
The utilization of renewable resources for energy and chemicals is
expected to increase in the near future. Ethanol can be produced by micro-
bial fermentation from renewable plant sources. Carbohydrates contain-
ing starch and polysaccharides, and lignocellulosic materials containing
cellulose, hemicellulose, and lignin, are the most abundant renewable
organic resources. Much attention has therefore been focused on geneti-
cally engineering strains that can efficiently utilize carbohydrates and
lignocellulosic materials and convert them into useful compounds such as
ethanol (25).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Amylase Activity of Marine Bacteria 253
Table 3
Reduced Sugar Amounts of Marine Microalgae
Reduced sugar amounts Content
Strain (Mg/ g dry wt) (%)

Green alga
NKG 040502 419.7 ± 73.5 41.9
NKG 040502 419.7 ± 73.5 41.9
NKG 042501 425.6 ± 58.0 42.5
NKG 041304 429.7 ± 45.2 42.9
NKG041102 453.2 ± 37.2 45.2
NKG042905b 451.8 ± 103.1 45.1
NKG 121701 531.3 ± 27.0 53.1
NKG 132301 442.9 ± 76.1 44.2
Others: 68 strains >40 >40
Total strains = 76

Table 4
Bioconversion to Reduced Sugars of Marine Microalgae Using Several Amylase
Reduced sugars concentration (mg/mL)a
Enyme 1d 2d
Amylase of P. un dina NKMB 0074 0.90 ± 0.28 2.82 ± 0.62
Amylase of P. pancreas c ND ND
Glucoamylase of Rhizopus sp.c ND ND
aND, not determined
bAlive cells of P. undina NKMB 0074 were inoculated at a concentration of 0.14 g/10 mL.
cEach enzyme was used at a concentration of 1 mg of protein/mL.

Renewable marine resources including macro- and microalgae could


be applied to convert to energy and chemical compounds. Previous work,
reported that as applications of renewable marine resources, methane pro-
duction using marine micro algae biomass (26) and a method utilizing float-
ing ceramic supports for the cultivation of marine microalgae have been
developed to take advantage of this vast resource (27). Application utiliz-
ing a large surface area of an ocean can produce vast amounts of marine
biomass and useful materials.
The microbial production of ethanol from carbohydrates is mainly
dependant on saccharification. Amylase produced from P. undina
NKMB0074 converted the biomass of the marine green alga NKG 121701
to reduced sugars. Terrestrial amylase activity was not observed under
saline conditions, indicating that it cannot be used industrially for the
saccharification of marine microalgae. The inactivity of these enzymes

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


254 Matsumoto et al.
resulted mainly from effects of salinity. Although saccharification of
marine microalgae biomass using amylase of the marine bacterium
P. undina NKMB 0074 was observed, enzyme production of amylase from
P. undina NKMB 0074 was only about 1 giL, and, therefore, the conver-
sion efficiency to reduced sugars was low (4.1 %). To increase conversion
efficiency, the application of other glucoamylases and proteases may pre-
serve activity under the saline conditions needed. Future work will inves-
tigate the optimum conditions required for saccharification using marine
amylase and microalgae for ethanol production.

References
1. Burgess, J. G., Iwamoto, K., Miura, Y., TaKano, H., and Matsunaga, T. (1993), Appl.
Microbiol. Biotechnol. 39,456-459.
2. Miura, Y., Sode, K., Nakamura, N., Matsunaga, N., and Matsunaga, T. (1993), FEMS
Microbiol. Lett. 107,163-168.
3. Wachi, Y., Burgess, J. G., Takahashi, J., Nakamura, N., and Matsunaga, T. (1995), J.
Biotechnol. 2, 210-213.
4. Wake, H., Akasaka, A., Umetsu, H., Ozeki, Y., Shimomura, K., and Matsunaga, T.
(1992), Plant Cell Rep. 11, 62-65.
5. Miura, Y., Sode, K., Nakamura, N., and Matsunaga, T. (1993), J. Mar. Biotechnol. 1,
134-146.
6. Kumazawa, S. and Mitsui A. (1981), Int. J. Hydrogen Energy 6, 339-348.
7. Nandi, R and Sengupta S. (1998), Crit. Rev. Microbiol. 24,61-84.
8. Sudo, H., Burgess, J. G., Takemasa, H., Nakamura, N., and Matsunaga, T. (1995),
Curro Microbiol. 30,219-222.
9. Mitsui, A. (1975), in Proceedings of the 3rd Ineternational Ocean Development Conference.
10. Borowitzka, 1.J. and Borowitzka M.A. (1989), in Industrial Production: Method and
Economics. Cresswell, R c., Rees, T. A. V., and Shah, N., eds., Elsevier Applied Sci-
ence, London, UK, pp. 294-316.
11. Borowitzka, M. A. (1992), J. Appl. Phycol. 4, 267-279.
12. Patterson, G. M. 1. (1996), J. Sci. Ind. Res. 55, 669-684.
13. Mielenz, J. R (2001), Curro Opin. Microbiol. 4,324-329.
14. Razmovski, Rand Pejin D. (1996), Folia Microbiol. 41, 201-207.
15. Joachimsthal, E. and Rogers P. (2000), Appl. Biochem. Biotechnol. 84/86, 343-356.
16. Krishhan, M., Blanco, M., Shattuck, C. K., Nghiem, N. P., and Davison, B. H. (2000),
Appl. Biochem. Biotechnol. 84/86, 525-54l.
17. Lawford, H. and Rousseau, J. (2000), Appl. Biochem. Biotechnol. 84/86,277-293.
18. Sandhu, D. and Joshi V. (1994), Indian J. Exp. Bioi. 32,873-876.
19. Montesinos, T. and Navarro, J.-M. (2000), Enzyme Microbiol. Technol. 27, 362-370.
20. Abouzied, M. M. and Reddy, A. C. (1986), Appl. Environ. Microbiol. 52,1055-1059.
21. Fumihisa, K., Sawada, T., Nakamura, Y., Ohnaga, M., Godliving, M., and Ushiyama,
T. (1998), Appl. Biochem. Biotechnol. 69,177-189.
22. Rippka, R, Deruelles, J., Watherbury, J. B., Herdman, M., and Stanier, R Y. (1979), J.
Gen. Microbiol. 111, 1-6l.
23. Miller, G. 1.(1959), Anal. Chem. 31,426-428.
24. Lowry, 0., Rosebrough, N. J., and Randall, R J. (1951), J. Bioi. Chem. 193,265-275.
25. Aristos, A. and Merja, P. (2000), Curro Opin. Biotechnol. 11,187-198.
26. Matsunaga, T. and Izumida, H. (1984), Biotechnol. Bioeng. Symp. 14,407-418.
27. Matsumoto, M., Yoshida, E., Takeyama, H., and Matsunaga, T.(2000), Appl. Biochem.
Biotech. 84/86,51-57.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0255/$20.00

D-Xylose Transport by Candida succiphila


and Kluyveromyces marxianus

BORIS U. STAMBUK,*,1,2 MARY ANN FRANDEN,l


ARJUN SINGH, l AND MIN ZHANG 1

National Bioenergy Center, National Renewable Energy Laboratory,


1

1617 Cole Boulevard, Golden, CO, 80401; and


20epartamento de Bioqufmica,
Universidade Federal de Santa Catarina,
Florianopolis, SC 88040-900, Brazil, E-mail: bstambuk@mboxl.ufsc.br

Abstract
The kinetics and regulation of D-xylose uptake were investigated in the
efficient pentose fermentor Candida succiphila, and in Kluyveromyces
marxianus, which assimilate but do not ferment pentose sugars. Active high-
affinity (Km - 3.8 roM; V max - 15 nmol![mg·min]) and putative facilitated
diffusion low-affinity (Km -140 roM; V max -130 nmol![mg·min]) transport
activities were found in C. succiphila grown, respectively, on xylose or glu-
cose. K. marxianus showed facilitated diffusion low-affinity (Km - 103 roM;
V max - 190 nmol! [mg· min]) transport activity when grown on xylose under
microaerobic conditions, and both a low-affinity and an active high-affin-
ity (Km - 0.2 roM; V max -10 nmol! [mg·min]) transport activity when grown
on xylose under fully aerobic conditions.
Index Entries: D-Xylose, transport kinetics, fermentation, Candida succiphila,
Kluyveromyces marxianus.

Introduction
A substantial fraction (up to 25%) of the monosaccharides in lignocel-
lulose hydrolysates consists of the pentose sugars D-xylose (5-20%) and
L-arabinose (1-5%). Xylose is second only to glucose in natural abundance,
and although this sugar can be fermented by some species of bacteria, yeast,
and filamentous fungi, the ethanol yields are low. Thus, there has been a
great emphasis in the last two decades on developing an efficient organism
for xylose fermentation through metabolic engineering (1-3). Although some
bacteria (Zymomonas mobilis and Escherichia coli) seem to be the best-perform-
ing biocatalysts for xylose fermentation, the preferred organism for indus-
*Author to whom all correspondence and reprint requests should be addressed.
Applied Biochemistry and Biotechnology 255 Vol. 105-108,2003
256 Stambuk et al.
trial ethanol fermentation processes is the yeast Saccharomyces cerevisiae. Since
wild-type strains of S. cerevisiae do not utilize D-xylose, several laboratories
have attempted to engineer S. cerevisiae for xylose fermentation (4,5). Results
with such genetically engineered yeasts have been encouraging, although
the xylose utilization rates and ethanol productivities are still low compared
to glucose fermentation by this yeast (5).
The first metabolic step in the fermentation of sugars by yeasts is the
uptake through the plasma membrane, and several reports have shown
that transport is the rate-limiting step for fermentation (6-8). Recently,
Eliasson et al. (9) have concluded from studies with chemos tat cultures that
xylose transport limits the xylose flux and metabolism by recombinant
S. cerevisiae cells. Xylose is taken up in S. cerevisiae cells by the glucose
transporters (10-12), which mediate the uptake of xylose by facilitated dif-
fusion with very low affinity (Km > 100 mM). Thus, the transport step would
pose a limitation on the flux, at least at low substrate concentrations. The
specificity of the transporters is also of concern, since glucose inhibits
xylose uptake when these two sugars are present in the fermentation
medium (13,14). Additionally, it is worth noting that xylose reductase (XR),
the first enzyme in the xylose-utilizing pathway, has a low affinity toward
xylose (Km > 50-100 mM), which means that high intracellular concentra-
tions of xylose are necessary for efficient utilization (15,16). Thus, the prop-
erties of the S. cerevisiae transporter(s) suggest the need for the improvement
of this metabolic step by genetic engineering (5,17).
Although hexose transport by yeast has been extensively investigated,
little attention has been given to pentose uptake, including the mechanisms
and the regulation of the transport activity. Here we report studies on the
kinetics and regulation of xylose transport activity in two species of yeast,
Candida succiphila and Kluyveromyces marxianus. C. succiphila is one of the few
yeasts capable of fermenting both D-xylose and L-arabinose (18). Although it
has been reported that some K marxianus strains ferment xylose (19), the
K marxianus (formerly Kfragilis) strain used in the present study assimilates
but does not ferment this sugar.
Materials and Methods
Yeast Strains and Growth Conditions
C. succiphila (NRRL Y-11998) and K marxianus (ATCC 52486) cells
were grown at 30°C in YEP medium (1 % Difco yeast extract, 2% Difco
Bacto Peptone) to which the 2-5% carbon source was added and the pH
adjusted to 5.0. Microaerobic conditions employed 50 mL of medium in
a 125-mL unbaffled Erlenmeyer flask shaken at 100 rpm. Aerobic condi-
tions employed 50 mL of broth in 250-mL baffled Erlenmeyer flasks
shaken at 220 rpm.
Analytical Methods
Growth was followed by turbidity measurements at 600 nm. One
absorbance unit corresponds to approx 0.25 mg (dry wt) of C. succiphila

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


D-Xylose Transport by Yeasts 257
cells/mL, or approx 0.35 mg (dry wt) of K. marxianus cells/mL. Substrate
consumed and products formed were analyzed in the supernatants of
samples of cultures removed periodically after cells were separated by
centrifugation. Xylose, xylitol, ethanol, and acetate were determined by
high-performance liquid chromatography using a Hewlett-Packard (HP)
1090L chromatograph equipped with an HP 1047A refractive index detec-
tor and a Bio-Rad HPX-S7H organic acid column operating at 65°C with a
0.01 N sulfuric acid mobile phase flow rate of 0.6 mL/min (20).

Transport Assays
Cells were harvested in mid-growth phase, centrifuged, washed
twice with cold distilled water, and suspended in water to a cellular den-
sity of about 60 g (dry cell mass) /L. The uptake of D-(1-14C)xylose (55 mCi/
mmol; American Radiolabeled Chemicals) was measured as previously
described (10,21). As a modification, assays were performed with 50 mM
succinate-Tris buffer, pH 5.0, and the uptake was measured during 30-s
periods. Appropriate experiments had shown that uptake of labeled
xylose was linear for at least 1 min. Transport activity is expressed as
nanomoles of xylose transported per milligram (dry cell mass) per minute.
Kinetic parameters were determined as described elsewhere (22,23) using
0.05-900 mM final substrate concentrations.
For assays in which the effect of inhibitors was evaluated, cell suspen-
sions were incubated with the indicated concentration of the inhibitors for
15 min prior to the assay, and 10 mM labeled xylose was used as substrate,
except for K. marxianus cells grown under aerobic conditions in which the
substrate concentration was 1 mM (see Subheading liD-Xylose Transport by
K. marxianus" and Table 2). The following compounds were dissolved in
ethanol: diethylstilbestrol, 2,4-dinitrophenol (DNP), carbonyl-cyanide-m-
chlorophenylhydrazine (CCCP), and dicyc1ohexyl-carbodiimide (DCCD).
Ethanol did not inhibit the transport activity at the concentration used in
the assays «2% [v Iv]). To determine the inhibitory effect of a sugar on the
transport of xylose, an excess of the test sugar was added to the labeled
xylose. All determinations were done at least in duplicate, which did not
differ by more than 15%.

Results
Growth on D-Xylose
C. succiphila and K. marxianus differed in their mode of xylose utiliza-
tion during growth on this carbon source under microaerobic conditions.
C. succiphila showed low rates of growth and xylose consumption but fer-
mented this sugar during growth producing Significant amounts of ethanol
and xylitol (Fig. 1). By comparison, K. marxianus grew and assimilated
xylose faster, but almost no ethanol and only low quantities of xylitol were
observed in the growth medium (Fig. 1), indicating that this yeast diverts
almost all carbon and energy from xylose metabolism into cell growth. Both

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


258 Stambuk et at.

30 A B 100

"0

-
c:
(II
.s::
Ec:
10
-
0
Q)
.... 20 ..0···· .. ······0
0

0_ .... </'
0··· .s::
0_
-~


~
~C)
>..-
x
cD <;) 1 C)
III
10 Qi
°
>..
><
()

0.1
0
0 30 60 90 120 0 30 60
Time (h)

Fig. 1. Growth of C. succiphila (A) or K. marxianus (B) on 2.5% xylose under


micro aerobic conditions. At the indicated time points, (O)cell growth, (e)xylose,
(O)xylitol, and (Ll)ethanol were determined as described in Materials and Methods.

strains produced low amounts of acetate «0.8 giL) at the end of the fer-
mentations and consumed all these products from the medium after the
sugar was exhausted. When grown under aerobic conditions, both yeasts
grew faster than under microaerobic conditions, and higher cellular den-
sities were obtained at the end of the incubations (data not shown). No
ethanol was produced by C. succiphila during aerobic growth on xylose.
D-Xy/ose Transport by C. succiphila
Kinetic analysis showed that xylose-grown cells took up xylose by a
single low-capacity (Vrnax = 15 nmol/[mg·min]) and high-affinity (Km =
3.8 mM) transport system (Fig. 2). The low capacity of this transporter may
explain the low sugar consumption rates observed when these cells are
growing on xylose (Fig. 1). This transport system is an active transporter
since the rate of xylose uptake in xylose-grown cells was significantly inhib-
ited in the presence of protonophores (NaN3, DNP, and eeep) and the H+-
adenosine triphosphatase (ATPase) inhibitors diethylstilbestrol and DeeD
(Table 1), indicating that the eletrochemical H+ gradient across the plasma
membrane is required for uptake of the sugar. Sugar competition studies
indicated that a general monosaccharide transporter probably mediates this
high-affinity system, since both an excess of either hexoses (glucose and
galactose) or pentoses (L-arabinose) significantly inhibited the rate of xylose
uptake (Table 1). Furthermore, control experiments using unlabeled xylose,
under the same conditions of excess sugar as those used for glucose inhibi-
tion, showed that this sugar competed as well as glucose for the uptake of
labeled xylose, indicating that this transport activity may have the same

Applied Biochemistry and Biotechnology Vol. 105-108,2003


o-Xylose Transport by Yeasts 259

150

'c:
·e
'~ 100
(5
E
S
::::.
50

o 1 234

V I [xylose] (nmol mg-1 min-1 mM- 1)

Fig. 2. Eadie-Hofstee plot of xylose transport by C. succiphila. The initial rates of


labeled xylose uptake (0.4-900 mM final concentration) by (0) xylose-grown or (e)
glucose-grown cells were determined as described in Materials and Methods.

Table 1
Effect of Inhibitors on Rate of Xylose Transport by C. succiphila
Relative xylose transport (%)a
Concentration Xylose-grown Glucose-grown
Inhibitor (mM) cells cells
None 100 b 100e
NaN 3 10 5 107
DNP 2.5 2 86
CCCP 2.5 2 62
Diethylstilbestrol 5 31 68
DCCD 5 9 107
Glucose 250 2 2
Galactose 500 3 24
Arabinose 600 11 56
a Determined with 10 mM labeled xylose.
b Rate of xylose transport was 9.0 nmol/(mg·min).
e Rate of xylose transport was 8.5 nmol/(mg·min).

affinity for both sugars. However, further studies would be required to


determine the affinity and/or K; for each sugar.
Glucose-grown cells also transported xylose with a single and very
low-affinity (Km =140 mM) uptake system, as indicated by the linear kinet-
ics of xylose transport observed with these cells (Fig. 2). This transport
activity had an approx 10-fold higher capacity (Vmax =130 nmol! [mg· min])

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


260 Stambuk et al.
than that observed in xylose-grown cells and is probably mediated by a
facilitated diffusion glucose transporter (Table 1). Unlabeled xylose did not
affect the uptake of labeled xylose under the same conditions of excess
sugar as those used for glucose inhibition, indicating that this transport
activity probably has a higher affinity for glucose compared with that
towards xylose.

D-Xylose Transport by K. marxianus


K. marxianus growing on xylose transported this sugar (Fig. 3) with a
single transport system with low affinity (Km = 103 roM) and high capacity
(Vmax = 190 nmol/[mg·min]). This transport activity was not significantly
inhibited by protonophores and H +-ATPase inhibitors (Table 2), indicating
that, most likely, xylose transport is mediated by facilitated diffusion. The
only sugar that strongly inhibited xylose transport was glucose (Table 2).
Unlabeled xylose had almost no effect on the uptake oflabeled xylose «20%
inhibition) under the same conditions of excess sugar as those used for
glucose competition, indicating that this transport activity probably has a
higher affinity for glucose.
The results just described were obtained when the cells were grown
under microaerobic conditions. A very different pattern of xylose transport
was obtained when the cells were grown under aerobic conditions. In this
case, the Eadie-Hofstee plot was nonlinear (Fig. 3), indicating a multicom-
ponent uptake mediated by at least two uptake systems: one transporting
xylose with very high affinity (Km = 0.2 roM) and low capacity (Vmax = 10
nmol/[mg·min]), and another with kinetic properties similar to the trans-
porter found in cells grown under microaerobic conditions (Km = 110 roM;
Vmax = 190 nmol/[mg.min]). Both protonophores and H+-ATPase inhibi-
tors, as well as hexoses and pentoses, significantly reduced the rates of the
high-affinity xylose uptake (measured with 1 roM labeled xylose) by aero-
bically grown cells (Table 2). Indeed, significant inhibition by all com-
pounds tested was still observed even in the presence of a saturating
concentration of the substrate (10 roM) for this high-affinity transport ac-
tivity. Thus, our results indicate that growth on xylose under aerobic con-
ditions induced production of a high-affinity active xylose transporter in K.
marxianus cells.

Discussion
It is well known that efficient conversion of xylose to ethanol by most
xylose-fermenting yeasts requires a limited amount of oxygen (24-26). The
explanation for this finding appears to lie in the specificity of the cofactor
required for XRactivity, with XR enzymes utilizing either NADPH orNADH
as cofactor permitting more efficient fermentation of xylose under limited
oxygen availability (27). However, xylose transport into the cell may also
limit fermentation of this sugar, as is the case with Pichia stipitis grown under
different conditions (28-30). Eadie-Hofstee plots of xylose uptake by several

Applied Biochemistry and Biotechnology Vol. 105-108,2003


o-Xylose Transport by Yeasts 261

150

I~

50

D··.·.
~t
••
O~~J---~--~--~~····-····-····~·~~
h ••••••

o 10 20 30 40

V I [xylose] (nmol mg-1 min-1 mM-1)

Fig. 3. Eadie-Hofstee plot of xylose transport by K. marxianus. The initial rates of


labeled xylose uptake (0.05-600 mM final concentration) by xylose-grown cells under
(0) microaerobic or (e) aerobic conditions were determined as described in Materials
and Methods.

Table 2
Effect of Inhibitors on Rate of Xylose Transport
by Xylose-Grown K. marxianus Cells
Relative xylose transport (%)
Concentration Aerobic Microaerobic
Inhibitor (mM) conditions conditions

None 100· 100h


NaN3 5 4 77
DNP 1.25 3 70
CCCP 1.25 2 68
Diethylstilbestrol 2.5 15 80
DCCD 2.5 18 Nne
Glucose 250 2 2
Galactose 500 9 53
Arabinose 600 6 89
aRate of xylose transport was 9.9 nmol! (mg· min), determined with 1 mM labeled xylose.
Rate of xylose transport was 21.2 nmol! (mg· min), determined with 10 mM substrate
b

concentration.
Not determined.
C

xylose-fermenting yeasts have revealed the presence of at least two kineti-


cally distinct xylose transport systems (17,21,30-33). The low-affinity trans-
port component is usually a facilitated diffusion transport system, while the
high-affinity components are proton symporters that use the proton motive

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


262 Stambuk et al.
force to actively transport the sugar into the cells. While most of these xylose
transport systems are shared with the structural sugar analog glucose, some
transporters seem to be specific for xylose (21,32).
Our data show that the rate of growth and xylose consumption by
C. succiphila and K. marxianus under microaerobic conditions can be explained
by the kinetics of xylose transport: the low capacity (Vmax) of xylose uptake
by C. succiphila determines slow growth and sugar consumption rates, while
K. marxianus cells harbor a high-capacity transport system which allows rapid
growth and sugar depletion from the medium. Neither the affinities (high vs
low) nor the energetics of transport (active vs facilitated diffusion) could be
correlated to the growth or fermentation performance of these yeasts.
Our results also indicate that C. succiphila shares a common character-
istic with the other Candida species (c. shehatae and C. utilis) for which xylose
transport data are available; that is, the kinetics of xylose transport by xylose-
grown cells show the presence of a single high-affinity transporter, while
after growth on glucose they exhibit a low-affinity transport activity, which
is probably owing to the marginal xylose uptake by glucose permeases
(21,32). All other yeast species usually have complex nonlinear kinetics of
transport, harboring both high- and low-affinity xylose permeases during
growth on this sugar. Since none of the yeast high-affinity xylose transport-
ers have been characterized at the molecular level thus far, these Candida
species harboring single xylose transporters could be a good source for
developing strategies for molecular cloning of suitable xylose transporters.
K. marxianus demonstrated facilitated diffusion transport activity with
very low affinity for xylose. This activity is probably mediated by a previ-
ously characterized hexose transporter with very low affinity for xylose (34).
These results and the data recently published on xylose (and glucose) trans-
port by P. stipitis (35), indicate that oxygen availability plays a role in the
regulation of the expression of high-affinity active sugar transporters. In
addition, our results indicate that, in yeast, there are a variety of o-xylose
transporters with different patterns of expression regulated by both the sugar
substrates and growth conditions. Further investigations should unravel the
molecular basis for the regulated expression of pentose transporter gene(s)
in response to oxygen availability by yeasts.

Acknowledgments
We thank Dr. C. Kurtzman (National Center for Agricultural Utiliza-
tion Research, USDA, Peoria, IL) for providing yeast strains. This work was
funded by the Biochemical Conversion Element of the Office of Fuels
Development of the US Department of Energy.

References
1. Bothast, R. J., Nichols, N. N., and Dien, B. S. (1999), Biotechnol. Prog. 15,867-875.
2. Aristidou, A. and Penttila, M. (2000), Curro Opin. Biotechnol. 11,187-198.
3. Zaldivar, J., Nielsen, J., and Olsson, L. (2001), Appl. Microbiol. Biotechnol. 56, 17-34.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


D-Xylose Transport by Yeasts 263
4. Ho, N. W., Chen, Z., Brainard, A P., and Sedlak, M. (1999), Adv. Biochem. Eng.
Biotechnol. 65, 163-205.
5. Hahn-Hagerdal, B., Wahlbom, C. F., Gardonyi, M., van Zyl, W. H., Otero, R. R. c.,
and Jonsson, L. J. (2001), Adv. Biochem. Eng. Biotechnol. 73,53-84.
6. Ye, J., Kruckeberg, A L., Berden, J. A., and van Dam, K. (1999), J. Bacterial. 181,
4673-4675.
7. Zastrow, C. R., Hollatz, c., de Araujo, P. S., and Stambuk, B. u. (2001), J. Ind. Microbial.
Biotechnol. 27, 34-38.
8. Goffini, P., Ferrero, I., and Donnini, C. (2002), J. Bacterial. 184,427-432.
9. Eliasson, A, Christensson, c., Wahlbom, C. F., and Hahn-Hagerdal, B. (2000), Appl.
Environ. Microbial. 66, 3381-3386.
10. Busturia, A and Lagunas, R. (1986), J. Gen. Microbial. 132,379-385.
11. Kotter, P. and Ciriacy, M. (1993), Appl. Microbial. Biotechnol. 38,776-783.
12. Ozcan, S. and Johnston, M. (1999), Microbial. Mol. BioI. Rev. 63,554-569.
13. Meinander, N. Q. and Hahn-Hagerdal, B. (1997), Appl. Environ. Microbial. 63,
1959-1964.
14. Meinander, N. Q., Boels,l., and Hahn-Hagerdal, B. (1997), Bioresour. Technol. 68,79-87.
15. Rizzi, M., Erlemann, P., Bui-Thanh, N. A, and Delleweg, H. (1988), Appl. Microbial.
Biotechnol. 29, 148-154.
16. Petrash, J. M., Murthy, B. S. N., Young, M., Morris, K., Rikimaru, L., Griest, T. A and
Harter, T. (2001), Chem. BioI. Interact. 130-132,673-683.
17. Spencer-Martins, I. (1994), Bioresour. Technol. 50,51-57.
18. Dien, B. S., Kurtzman, C. P., Saha, B. c., and Bothast, R. J. (1996), Appl. Biochem.
Biotechnol. 57-58, 233-242.
19. Margaritis, A and Bajpai, P. (1982), Appl. Environ. Microbial. 44, 1039-1041.
20. Deanda, K., Zhang, M., Eddy, c., and Picataggio, S. (1996), Appl. Environ. Microbial.
62, 4465-4470.
21. Lucas, C. and van Uden, N. (1986), Appl. Microbial. Biotechnol. 23,491-495.
22. Coons, D. M., Boulton, R. B., and Bisson, L. F. (1995), J. Bacterial. 177,3251-3258.
23. Stambuk, B. U. and de Araujo, P. S. (2001), FEMS Yeast Res. 1,73-78.
24. Ligthelm, M. E., Prior, B. A., and du Preez, J. C. (1988), Appl. Microbial. Biotechnol. 28,
63-68.
25. Boynton, B. L. and McMillan, J. D. (1994), Appl. Biochem. Biotechnol. 45-46,509-514.
26. Passoth, V., Zimmermann, M., and Klinner, U. (1996), Appl. Biochem. Biotechnol. 57-
58, 201-212.
27. Bruinenberg, P. M., de Bot, P. H. M., van Dijken, J. P., and Scheffers, W. A (1983),
Appl. Microbial. Biotechnol. 19, 256-260.
28. Ligthelm, M. E., Prior, B. A, du Preez, J. c., and Brandt, V. (1988), Appl. Microbial.
Biotechnol. 28, 293-296.
29. Sreenath, H. K. and Jeffries, T. W. (1997), Appl. Biochem. Biotechnol. 63-65,109-116.
30. Kilian, S. G. and van Uden, N. (1988), Appl. Microbial. Biotechnol. 27,545-549.
31. Does, A L. and Bisson, L. F. (1989), Appl. Environ. Microbial. 55,159-164.
32. Kilian S. G., Prior, B. A, and du Preez, J. C. (1993), World J. Microbial. Biotechnol. 9,
357-360.
33. Nobre, A., Lucas, c., and Leao, C. (1999), Appl. Environ. Microbial. 65,3594-3598.
34. Sa-Correia, I. and van Uden, N. (1983), Biotechnol. Lett. 5,413-418.
35. Weier stall, T., Hollenberg, c., and Boles, E. (1999), Mol. Microbial. 31,871-883.

Applied Biochemistry and Biotech~ology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0265/$20.00

Molecular Characterization of a Gene


for Aldose Reductase (CbXYL 1)
from Candida boidinii and Its Expression
in Saccharomyces cerevisiae

MIN HVUNG KANG, l HAlVING NI, l


AND THOMAS w.
JEFFRIES*,1,2
1Department of Bacteriology, University of Wisconsin-Madison,
1550 Linden Drive, Madison, WI 53706,
E-mail: twjeffri@facstaff.wisc.edu; and 2Forest Products Laboratory,
One Gifford, Pinchot Drive, Madison, WI 53705

Abstract
Candida boidinii produces significant amounts of xylitol from xylose, and
assays of crude homogenates for aldose (xylose) reductase (XYLlp) have
been reported to show relatively high activity with NADH as a cofactor even
though XYLlp purified from this yeast does not have such activity. A gene
coding for XYLlp from C. boidinii (CbXYLl) was isolated by amplifying the
central region using primers to conserved domains and by genome walking.
CbXYLl has an open reading frame of 966 bp encoding 321 amino acids. The
C. boidinii XYLlp is highly similar to other known yeast aldose reductases
and is most closely related to the NAD(P)H-linked XYLlp of Kluyveromyces
lactis. Cell homogenates from C. boidinii and recombinant Saccharomyces
cerevisiae were tested for XYLlp activity to confirm the previously reported
high ratio of NADH:NADPH linked activity. C. boidinii grown under fully
aerobic conditions showed an NADH:NADPH activity ratio of 0.76, which
was similar to that observed with the XYLlp from Pichia stipitis XYLl, but
which is much lower than what was previously reported. Cells grown under
low aeration showed an NADH:NADPH activity ratio of 2.13. Recombinant
S. cerevisiae expressing CbXYLl showed only NADH-linked activity in cell
homogenates. Southern hybridization did not reveal additional bands. These
results imply that a second, unrelated gene for XYLl p is present in C. boidinii.

Index Entries: Candida boidinii; Saccharomyces cerevisiae; aldose reductase;


CbXYLl; xylose reductase; NADH; NADPH; gene cloning; gene expression.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 265 Vol. 105-708, 2003


266 Kang et al.
Introduction
o-Xylose is one of the major components of lignocellulosic biomass.
Alcoholic fermentation of this renewable carbon source is essential if the
overall process of lignocellulose bioconversion is to be economical (1). Since
the discovery that some naturally occurring yeasts cause pentose fermen-
tation (2,3), considerable interest in the study of the xylose catabolic path-
way has arisen (4-6). Yeasts convert xylose to xylulose through sequential
reduction (either NADH or NADPH-linked xylose [aldose] reductase,
XYLlp) and oxidation (NAD-dependent xylitol dehydrogenase, XYL2p)
(7,8). The subsequent phosphorylation of xylulose by o-xylulokinase (9,lD)
allows entry of the sugar phosphate into the pentose phosphate pathway.
Xylose (aldose) reductase (XYLlp) catalyzes the reduction ofxyloseto
xylitol, which is the first step for xylose metabolism in yeasts. Most com-
monly, xylose reductase (XR) is specific for NADPH. However, the main
XR of Pichia stipitis has significant activity with NADH as a cofactor (7). Its
NADH:NADPH activity ratio is 0.7. A second cryptic gene in P. stipitis is
NADPH dependent (11). NADH activity is critical to the anaerobic metabo-
lism of xylose because the second step in the pathway, xylitol dehydroge-
nase (XYL2p), is specific for NADH, and cofactor imbalance can occur
unless some means exists to regenerate NAD+ (12). Most known XYLlp
enzymes favor NADPH over NADH. Interestingly, xylose-fermenting
yeasts display different ways to avoid cofactor imbalances in the cellular
redox system with NADH + H+ being accumulated at the expense of
NADPH + H+ during anaerobic or oxygen limited growth on xylose. They
either express multiple forms of XRs with different coenzyme specificities
as in Pachysloen tannophilus (13) or only one single enzyme with dual cofac-
tor specificity as in Pichia stipitis (7).
In one xylitol-producing yeast, Candida boidinii, the CbXYLlp
NADH:NADPH activity ratio has been reported to range between 2.0 and
5.9, depending on the aeration rate (14). When two separate aldose reduc-
tases with different NAD(P)H activities are induced to different extents,
the ratio can vary. The lower ratio was observed at higher aeration rates,
which is consistent with the higher respiratory function ofNADPH-linked
enzymes under those conditions (15,16). This finding suggests that if one
were to replace or supplement NADPH-linked aldose (xylose) reductase
with theCbXYLlp, aldose (xylose) reductase activity would be less depen-
dent on NADPH produced by the oxidative pentose phosphate cycle and
could be more active with NADH produced through fermentation.
Therefore, in the present study, isolation of the CbXYLl gene was
attempted and CbXYLlp activity was investigated in recombinant
S. cerevisiae that was transformed with CbXYLl.
Materials and Methods
Yeasts, Media, Enzymes, and Chemicals
C. boidiniiNRRL Y-17213 was maintained on agar plates at 4°C (15). S.
cerevisiae L2612 (Mata, trp-, ura-, leu-) (17) was used for transformation and
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
Aldose Reductase from Candida boidinii 267
expression of aldose reductase (AR) gene from C. boidinii. Escherichia coli
DH5a was used for all of DNA manipulations.
E. coli was grown in Luria-Bertani medium. Ampicillin (50 ~gl mL) was
added to the medium when required. Yeast strains were grown in yeast
peptone medium (10 giL of yeast extract, 20 giL ofbacto peptone), or yeast
synthetic (YS) medium containing 6.7 giL of yeast nitrogen base (YNB) with-
out amino acids (Difco, Grayson, GA), plus 20 giL of casamino acid (Difco).
Glucose (20 giL) or xylose (20 giL) was used as a carbon source. Yeast cells
were cultivated at 30°C in 50 mLof medium in a 125-mLErlenmeyer flask.
Restriction enzymes, DNA-modifying enzymes, and other molecular
reagents were obtained from New England Biolabs (Beverly, MA), Promega
(Madison, WI), Stratagene (La Jolla, Ca), Invitrogen (Carlsbad, CA),
Clontech (Palo Alto, CA), Roche (Indianapolis, IN), and Applied Biosystems
(Foster City, CA). Reaction conditions were as recommended by the suppli-
ers. All general chemicals were purchased from Sigma (St. Louis, MO). Prim-
ers for polymerase chain reaction (PCR) amplification and sequencing were
synthesized by Sigma-Genosys (The Woodlands, TX).
Cloning of CbXYL 1 Gene
To isolate the CbXYLl gene, a set of closely related AR proteins from
the yeasts Candida tenuis (18), P. stipitis (19), Candida tropicalis (20), Candida
guilliermondii (21) and Pachysolen tannophilus (22) were identified using
the Blast search program on the NCBI server (see Website: http:/ I
www.ncbi.nlm.nih.gov Iblast/). Conserved regions of the various aldose
(xylose) reductase amino acid sequences were then aligned using
ClustalW with default settings at the EBI Website (see Website: http:/ I
www.ebi.ac. ukl clustalw). Degenerate primers were designed against the
most conserved segments and PCR amplification was used to obtain a
partial CbXYLl gene from C. boidinii genomic DNA. The 5'-terminal
degenerate primer was 5'-CCCGGGATCCGGNTAYAGRTTRTT- YGAY-
3' and the 3'-terminal degenerate primer was 5'-CCCGGGATCCYTGYA-
ARTANGGRTGRTGYTC-3'. PCR amplification was carried out in 50 III
of a reaction mixture containing 500 ng of C. boidinii genomic DNA,
100 ng of each primer, 1X reaction buffer, 1.5 mM MgCl2, 0.2 mM dNTP,
and 2.5 U of AmpliTaq-gold DNA polymerase (Applied Biosystems. It
was then run for 30 cycles, denatured at 95°C for 1 min, annealed at 45°C
for 1 min, and polymerized at 72°C for 2 min. About 465 bp of PCR prod-
ucts was subcloned into cloning vector, pCR2.1-Topo (Invitrogen), and
then sequenced by the University of Wisconsin Biotechnology Center
using a 371 ABI automated sequencer (Perkin Elmer, Foster City, CA).
To obtain the full CbXYLl sequence, a mini-genomic library was con-
structed using the Genome Walker™ kit (Clontech) with the following
modification: After genomic DNA of C. boidinii was digested with Msc I to
generate blunt-ended fragments, adapters (Clontech) were ligated onto the
5'- and 3'-terminals of digested DNA, respectively. PCR amplification was
then attempted with gene-specific primers and with the adapter primers

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


268 Kang et al.
(API and AP2, from the genome walker kit from Clontech). Gene-specific
primers were synthesized on the base of the partial CbXYLl gene. For pri-
mary PCR amplification, 5XRSEQ (5'-TTCTCTTTTAACTAAACCATC-3')
and 3XRSEQ2 (5'-GGATCCATGGAAGAATTGGTTGAAGAA-3')
were used, and then, for secondary PCR amplification, 5XRSEQ2
(5'-CGGGATCCTTTGATCGCTTTGTTAACACC-3') and 3XRSEQ (5'-
GGTTTAGTTAAATCAATT-GGT-3') were used. A single PCR product
from the Msc I minilibrary was obtained, subcloned into pCR2.1-Topo
(Invitrogen), and sequenced.
Restriction enzyme digestion, electrophoresis, DNA ligation, trans-
formation, and DNA preparation from E. coli were performed by using
standard methods (23), as was Southern blot analysis for CbXYLl. Genomic
DNA was digested with Msc I, Pst I, and Xba I. After separation on an
agarose gel, DNA was transferred onto a nylon membrane, hybridized
with probe DNA that was synthesized by random primer extension. The
signal was then developed according to the protocol provided by Roche.
DNASIS software was used for sequence analysis (24). A phylogenetic tree
was generated from the alignments using the neighbor-joining method
(25). By throwing out the sequences in the gap, ambiguous parts of the
alignment were excluded from the phylogenetic analysis.
Expression and Enzyme Assays
DNA fragments (1.4 kb) containing the CbXYLl open reading frame
(ORF) (1.0 kb), upstream region (0.4 kb), and terminator (0.1 kb) were ampli-
fied by using two synthetic primers from C. boidinii genomic DNA.
5'-CTGCAGTGGCCACAATGGCATGGC-TTC-3' was used for the 5'-termi-
nal primer and 5'-CTGCAGATGTATTGAAAAGCGTGATTAATT-3' was
used for the 3'-terminal primer. These primers have Pst I-restriction enzyme
sites on their ends. The amplified CbXYLl gene was subcloned into the Pst I
site of a yeast shuttle vector, pRS424 (26), and then pRS424-CbXYLl was
introduced into S. cerevisiae L2612 (Mata, trp-, ura-, leu-) (17) by using a Yeast
EZ-transformation kit (BIO 101, Vista, CA). Transformants were selected on
YS medium containing 20 giL of glucose (YSD) and 20 giL of agar.
Wild-type and recombinant S. cerevisiae (S4a) were grown on Y5-2%
xylose (YSX) and 2% glucose medium for 2 d at 30°C and 200 rpm. Culture
medium was then collected to analyze the accumulation of xylitol by using
high-performance liquid therapy (HPLC) (HP, Wilmington, DE) with an
ION300 column (Interaction Chromatography, San Jose, CA).
Enzyme activity for CbXYL1p expressed in S. cerevisiae was assayed
by measuring the decrease of absorbance of NADH or NADPH at A 340 (27)
using a photodiode array spectrophotometer (Hewlett Packard). The 1-mL
reaction mixture contained 0.1 M Na phosphate buffer (pH 6.5), 0.4 M D-
xylose, various concentration of NAD(P)H, and cell homogenate at 30°C.
Reactions were started by addition of NAD(P)H. The initial reaction rate
was determined as a function of NAD(P)H concentration and expressed as
units I milligram of protein. One unit of enzyme activity refers to 1 lAffiolof

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Aldose Reductase from Candida boidinii 269
NAD(P)H consumed/min. All rates were corrected for the appropriate
blank readings to account for the nonspecific oxidation of NAD(P)H such
as autooxidation mediated by aldoses (28). Affinity of the AR for other
aldoses also was assayed. The amounts of total proteins in crude extracts
were determined by bicinchonic acid assay (Pierce, Rockford, IL).

Effect of Aeration on NADH- and NADPH-/Linked CbXYL 1P Activities


C. boidinii was cultured in YSX (6.7 giL ofYNB, 2% casamino acids, and
2% xylose). Recombinant S. cerevisiae (S4a) was cultured in YSX plus 2%
glucose. For aerobic conditions, cells were cultured in 2 mL of YSX or YSX
plus 2% glucose, overnight, and used to inoculate into 50 mL of medium in
125-mL flasks on the same media at 30°C. For low oxygen transfer conditions,
the cell suspensions were adjusted to an initial cell density of 1.25 mg/mL
and cultured at 50 rpm. For high oxygen transfer conditions, cell suspensions
were adjusted to an initial cell density of 0.125 mg/mL and cultured at 200
rpm. After 18 h, cells were harvested and resuspended in 0.1 M Na-phos-
phate buffer (pH 6.5). Vigorous mixing with glass beads disrupted the cells,
and the homogenates were clarified by centrifuging at 10,OOOg for 10 min.
Supernatant solutions were then assayed.

Results
Cloning ofCbXYL 1
We successfully amplified a DNA fragment of approx 400 bp from
genomic DNA of C. boidinii using degenerate primers designed against con-
served regions GYRLFD and EHHPYLQ found in xylose (aldose) reduc-
tases (18-22) (Fig. I), and the amplicon sequence was identified as a partial
gene for CbXYLl (data not shown). Gene-specific primers were designed
from this fragment. The 5'-upstream, ORF and 3'-downstream sequences for
CbXYLl were obtained by genome walking and deposited into GeneBank
(accession no. AF451326). This sequence consists of 387 bp 5'-upstream, an
ORF of 966 bp that encode putative 321 amino acids with a calculated mo-
lecular weight of 36 kDa, and 52 bp 3'-downstream. The calculated molecu-
lar weight is in good agreement with the protein size of 36 kDa mentioned
in a previous review (29). Sequence analysis showed a TATAAA and three
CAAT boxes located -54, and -120, -159 and -201, respectively, in the
5'-upstream region. As shown in Fig. 2, a phylogenetic analysis showed that
the CbXYL1p is significantly different from three Candida XYL1 proteins
and fromP. stipitis XYL1p, and it is most closely related to the XRof K.lactis
and S. cerevisiae, with which it shows 63 and 62 % identity and 78 and 76%
similarity, respectively. IPKS, which is the coenzyme-binding motif, was
highly conserved in the C. boidinii protein in other yeast aldose (xylose)
reductases, and in mammalian aldo-keto reductase family enzymes (30).
Southern blot analysis was performed with the CbXYLl ORF as a
molecular probe to determine whether other related sequences were
present in the C. boidinii genome. As shown in Fig. 3A, only a single band

Applied Biochemistry and Biotechnology Vol. 105-108,2003


270 Kang etal.
c. boidinii 29 CAET IYEAI KVGYRLFDGAMDY 50
Conserved sequence GYRLFD
C. tenuis 31 AGEQVYQAlKAGYRLFDGAEDY 52
P. stipitis 27 CSEQIYRAIKTGYRLFDGAEDY 48
C. tropicalis 33 AADQIYNAIKTGYRLFDGAEDY 54
C. guilliermondii 26 CADTIYNAIKVGYRLFDGAEDY 47
P. tannophilus 27 AADMVYAAlKEGYRLFDCACDY 48

c. boidinii 180 GCKIRPAVLEI EHHPYLVQPRL 201


Conserved sequence EBBPYLQ
C. tenuis 182 GAT I KPAVLQVEHHPYLQQPKL 203
P. stipitis 178 GAT I KPSVLQVEHHPYLQQPRL 199
C. tropicalis 184 GAT I KPAVLQI EHHPYLQQPKL 205
C. guilliermondii 177 SAKIKPAVLQIEHHPYLQQPRL 198
P. tannophilus 178 AARIKPASLQIEHHPYLQQNKL 199

Fig. 1 Alignment of amino acid sequences between CbXYLlp and known XRs or
ARs. Amino acid sequences were aligned by using FASTA (see Website: http:/ /
www.ncbi.nlm.nih.gov). Highly conserved sequences in bold were used to design
degenerate primers .

. - - - - - - Candida tropicalis

. - - - - - - - Pichia stipitis

, . - - - - Candida tenuis
'----\
' - - - - Candida shehatae

, - - - - - - - - - Candida guilliermondii

. - - - - - - - - - - - - Pachyso/en tannophilus
, - - - - - - - - - - Candida boidinii

, . - - - - - - - - Kluyveromyces lactis
'------I
' - - - - - - - - - - Saccharomyces cerevisiae
0.1

Fig. 2. Phylogenetic tree of XRs or ARs found in yeasts. Amino acid sequences were
identified by BLAST, and the resulting protein sequences were then aligned. A phy-
logenetic tree was generated from the aligned regions after excluding sequences cre-
ating gaps.

hybridized to the probe under low-stringency conditions, which implies


that only a single-copy gene of CbXYLl exists in C. boidinii.

Expression and Activity


To express CbXYLl in S. cerevisiae, a DNA fragment of 1405 bp was
amplified and cloned from genomic DNA of C. boidinii by using the 5'-
primer (5'-CTGCAGTGGCCACAATGGCATGGCTTC-3') and the 3'-
primer (5'-CTGCAGATGTATTGAAAAGCGTGATTAATT-3') for the full
sequences. This fragment was then subcloned into the pRS424 vector (26)

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Aldose Reductase from Candida boidinii 271

A 1 2 3 B

F1
>10k~

5~ Amp'
Pcbxyl1

2.5kb

Fig. 3. Genomic Southern blot analysis of CbXYLl and construction map for expres-
sion in S. cerevisiae. (A) Genomic DNA from C. boindinii was digested with Msc I, Pst
I, and Xba I. After running on an agarose gel, DNA was transferred onto a nylon
membrane, hybridized with probe DNA that was synthesized by random primer ex-
tension. The signal was then developed according to a protocol given by Roche. Ar-
rows indicate the approx size of hybridized DNA fragments. (B) Construction map of
pRS424-CbXYLl for expression in S. cerevisiae. TheCbXYLl gene was constructed under
the control of its own promoter. PRS424-CbXYLl was introduced into S. cerevisiae
L2612 (Mata, trp-, ura-, leu-) (12). Pcbxyll and CbXYLl indicate promoter and coding
region, respectively.

Table 1
Specific Activity of CbXYLlp Expressed in S. cerevisiae"
NADH (U/mg) NADPH (U/mg)
Control CbXYLlp Control CbXYLlp
D-Arabinose 0.009 0.010 0.009 0.017
L- Arabinose 0.019 0.032 0.020 0.494
D- Galactose 0.006 0.006 0.012 0.094
D- Glucose 0.008 0.008 0.008 0.047
D- Ribose 0.007 0.007 0.015 0.159
D- Xylose 0.009 0.016 0.020. 0.417

aAR activity was not detected in S. cerevisiae L2612 (wild-type).

(Fig. 3B). After selection for growth on YSD medium, the transformant was
confirmed by PCR amplification (data not shown) and named S4u. To iden-
tify CbXYLlp enzyme activity, S4u was cultured on YS-2% glucose and
2% xylose medium. After culturing for 2 days, xylitol production was ana-
lyzed by HPLC. The transformant produced significant (5.8 g) xylitol from
20 g of xylose whereas the wild-type strain produced only trace amounts
(0.71 g). Glucose was not detected because it was consumed completely for
cell growth within 2 d. This experiment demonstrated that CbXYLl could
be expressed under control of its own promoter in S. cerevisiae.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


~
[
til
c'
9-
~
<;;'
~
~
::l
Q..
til
c' Table 2
[ Effect of Aeration on CbXYLlp Activity in C. boidinii and S, cerevisiae (S4a) a
::l
C
C. boidinii S, cerevisiae (S4a)
~
~

50 rpm 200 rpm 50 rpm 200 rpm


I\..) NADH NADPH NAOH NAOPH NAOH NAOPH NAOH NAOPH
'J
I\..)
V max (U/mg) 0.19 0.09 0.2 0.26 NO 0.7 NO 0.22
Km (JAM) 3.38 7.36 2.57 5.39 NO 3.08 NO 3.66
NADH/NADPH
ratio for specific activity 2.13 0.76 NA NA
aND, not detected; NA, not available.

~
:-

~-.
c
~
"->
§
Aldose Reductase from Candida boidinii 273
To investigate the relative affinities of CbXYL1p for NADPH and
NADH and the substrate specificity, an enzyme assay of the CbXYL1p
was performed with several carbohydrates: D-arabinose, L-arabinose,
D-galactose, D-glucose, D-ribose, and D-xylose. As shown in Table 1,
CbXYL1p had higher activity with S-carbon than with 6-carbon sugars,
and it also had much higher affinity for NADPH than for NADH. The
results were in good agreement with a previous report on the activity of
the same enzyme purified from C. boidinii (31). CbXYL1p showed even
higher activity for L-arabinose than D-xylose, thereby demonstrating that
the cloned gene coded for an NADPH-linked AR.

Effect of Aeration on AR Activity Against NAD(P)H


When we cultivated C. boidinii at two different agitation levels, we
noted a change in the relative XYL1 p activities of the crude cell homogenates
against NADH and NADPH. As the aeration rate was increased, the activ-
ity ratio for NADH/NADPH decreased from 2.13 to 0.76. However, when
CbXYLl was expressed in S. cerevisiae under the same aeration rates and
tested for activity with NADH and NADPH, cellhomogenates only showed
the activity for NADPH (Table 2). When the endogenous S. cerevisiae AR
activity was tested as control, no significant activity was detected (data not
shown). A decrease in the NADH/NADPH ratio with increasing aeration
was in good agreement with a previous report (15).

Discussion
The XYL1 gene coding for XYL1 p was isolated from C. boidinii through
PCR ~mplification and by genome walking. The CbXYLl clone was identi-
fied as XYL1 p by comparing its implied protein sequence with other aldose
(xylose) reductases. The catalytic tetrad of XYL1p, Tyr48/ Asp43/Lys77 /
His110 (32-34) is completely conserved as TyrSO / Asp4S /Lys79 /Hisl12 in
CbXYLlp. According to these reports (32-34), TyrSO is proposed to be the
proton donor of the aldehyde reaction by CbXYL1p. The residue pair
Asp4S/Lys79 is expected to interact with the hydroxy group of TyrSO.
Hisl12 may facilitate proton donation by TyrSO. Therefore, CbXYL1p has
the conserved amino acids for AR function. Jez et al. (32) have reviewed the
cofactor-binding domain and residues. These consensus sequence and
amino acids were well conserved in CbXYL1p.
When this gene was expressed in S. cerevisiae under its native promoter,
xylitol accumulation was observed in the medium. This result showed that
the CbXYLl promoter is recognized in S. cerevisiae. When the cell homoge-
nate from recombinant S. cerevisiae S4a. was used for enzyme assay AR activ-
ity was detected, but no AR activity was detected in homogenates from the
nonrecombinant parent. XR from P. stipitis can be expressed in S. cerevisiae
with its native promoter (19,35). Other aldose (xylose) reductases were also
successfully expressed in E. coli (18,19) or Pichia pastoris (21).
CbXYL1p could reduce several aldose sugars when it was expressed
in S. cerevisiae. It showed higher activity for S-carbon than 6-carbon sug-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


274 Kang et al.
ars. This CbXYLlp showed even higher activity for L-arabinose than for
D-xylose. However, CbXYLlp showed cofactor affinity only for NADPH.
XR from P. stipitis with dual activity for NAD(P)H can also assimilate
other aldoses (36). Even though crude cell homogenates were used for
enzyme assays rather than purified activities, AR activity was detected
only in recombinant cells, so the cofactor and substrate activities are in-
dicative of the recombinant protein.
Cell homogenates from C. boidinii cells grown under different aeration
rates showed different affinities with NADH and NADPH. As the aeration
rate was increased, the NADH:NADPH affinity ratio decreased. This was
in good agreement with the previous report in which NADH/NADPH
activity ratio for CbXYLlp varied when xylitol production was examined
at various aeration rates (15).
We infer from these findings that other aldose (xylose) reductases
with significant activity for NADH are induced in C. boidinii under oxygen-
limited conditions. Previous reports have shown that AR activities with
different affinities for NADH and NADPH could exist in the cells. Some
yeasts possess aldose (xylose) reductases that are specific for NADPH
(21,37). Other yeasts have enzymes with dual activity for NADH and
NADPH (7,38,39), and still other organisms have multiple forms (40,41)
that show different activity against NADH and NADPH, respectively.
However, genomic Southern blot analysis in the present work revealed
only a single gene for AR in C. boidinii. The cloned CbXYLl also had highly
specific activity for NADPH when expressed in S. cerevisiae. The NADH-
linked enzyme activity in cell homogenates may be due to other cellular
enzymes capable of utilizing NADH in C. boidinii (21). Because the con-
sumption of NADH at high aeration would be greater than at low aeration,
it could cause the NADH/NADPH ratio to decrease. Therefore, we con-
clude that the CbXYLl of C. boidinii is a single-copy gene for AR that is
specifically linked to NADPH.
We also infer from these results that at least two NADH- and NADPH-
linked reductases are present in C. boidinii. The gene that we have cloned
only shows activity for NADPH-such as the NADPH-linked XYLlp of
K. [act is, which was most closely related to CbXYLlp. The second gene is
not closely related to CbXYLl because no additional bands were observed
in Southern hybridization.

Acknowledgments
Y.-S. Jin and J. Laplaza provided helpful comments and discussion.
We acknowledge financial support from the National Renewable Energy
Laboratory under subcontract no. ZCG-9-29009-01 and from Iogen.

References
1. Aristidou, A. and Penttila, M. (2000), Curro Opin. Biotechnol. 11,187-198.
2. Maleszka, R. and Schneider, H. (1982), Can. ]. Microbiol. 28(3),360-363.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Aldose Reductase from Candida boidinii 275
3. Slininger, P. J., Bothast, R J., van Cawenberge, J. E., and Kurtzman, C. P. (1982),
Biotechnol. Bioeng. 24, 371-384.
4. Skoog, K. and Hiihn-Hagerdal, B. (1988), Enzyme Microb. Technol. 10, 66-80.
5. Schneider, H. (1989), CRC Crit. Rev. Biotechnol. 9, 1-40.
6. Mishra, P. and Singh, A (1993), Adv. Appl. Microbiol. 39,91-152.
7. Verduyn, c., van Kleef, R, Frank, J., Schneider, H., van Dijken, J. P., and Scheffers,
W. A (1985), Biochem. J. 226,669-677.
8. Bolen, P., L. and Detroy, R W. (1985), Biotechnol. Bioeng. 27(3), 302-307.
9. Jin, Y. S., Jones, S., Shi, N.-Q., and Jeffries, T. W. (2002), Appl. Environ. Microbiol. 68,
1232-1239.
10. Richard, P., Tiovari, M. H., and Penttila, M. (2000), FEMS Microbiol. Lett. 190,39-43.
11. Hagedorn, J. and Ciriacy, M. (1989), Curro Genet. 16,27-33.
12. Bruinenberg, P. M., de Bot, P. H. M., van Dijken, J. P., and Sheffers, W. A (1984), Appl.
Microbiol. Biotechnol. 19, 256-260.
13. Verduyn, c., Frank, J., van Dijken, J. P., and Scheffers, W., A (1985), FEMS Microbiol.
Lett. 30,313-317.
14. Vandeska, E., Kuzmanova S., and Jeffries, T. W. (1995), J. Ferment. Bioeng. 80,513-516.
15. Jeffries, T. W. and Shi, N.-Q. (1999), Adv. Biochem. Eng. Biotechnol. 65, 117-16l.
16. Flores, c.-L., Rodriguez, c., Petit, T., and Gancedo, C. (2000), FEMS Microbiol. Rev. 24,
507-529.
17. Cho, K. M., Yoo, Y. J., and Kang, H. S. (1999), Enzyme Microb. Technol. 25,23-30.
18. Hacker, B., Habenicht, A., Kiess, M., and Mattes, R (1999), J. BioI. Chem. 380,
1395-1403.
19. Amore, R, Kotter, P., Kuster, c., Ciriacy, M., and Hollenberg, C. P. (1991), Gene 109,
89-97.
20. Yokoyama, S., Kinoshita, Y., Suzuki, T., Kawai, K., Horitsu, H., and Takamizawa, K.
J. (1995), J. Ferment. Bioeng. 80, 603-605.
21. Handumrongkul, c., Ma, D. P., and Silva J. L. (1998), Appl. Microbiol. Biotechnol. 49,
399-404.
22. Bolen, P. L., Hayman, G. T., and Shepherd, H. S. (1996), Yeast 12, 1367-1375.
23. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989), Molecular Cloning: A Laboratory
Manual, 2nd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
24. Mandler, J. and Ludwig, S. (1990), SUBSIS: DNASIS/PROSIS subdirectory manage-
ment. Camp. Appl. Biosci. 6,291-292.
25. Saitou, N. and Nei, M. (1987), Mol. BioI. Evol. 4,406-425.
26. Christianson, T. W., Sikorski, R S., Dante, M., Shero, J. H., and Hieter, P. (1992), Gene
110,119-122.
27. Bruinenbur, P. M., van Dijken, J. P., and Alexander-Scheffers, W. (1983), J. Gen.
Microbiol. 129,965-97l.
28. Wolff, S. P. and Crabbe, M. J. C. (1985), Biochem. J. 226,625-630.
29. Lee, H. (1998), Yeast 14, 977-984.
30. Kostrzynska, M., Sopher, C. R, and Lee, H. (1998), FEMS Microbiol. Lett. 159,107-112.
31. Bolen, P. L. and McCracken, D. A (1990), J. Ferment. Bioeng. 69,211-214.
32. Jez, J. M., Bennett, M. J., Schlegel, B. P., Lewis, M., and Penning, T. M. (1997), Biochem.
J. 326, 625-636.
33. Bohren, K. M., Grimshaw, C. E., Lai, C. J., Harrison, D. H., Ringe, D., Petsko, G. A,
and Gabbay, K. H. (1994), Biochemistry 33, 2021-2032.
34. Tarle,I., Borhani, D. W., Wilson, D. K.,Quiocho,F.A,andPetrash,J. M. (1993),J. BioI.
Chem. 268,25,687-25,693.
35. Takuma, S., Nakashima, N., Tantirungkij, M., Kinoshita, S., Okada, H., Seki, T., and
Yoshida, T. (1991), Appl. Biochem. Biotechnol. 28/29,327-340.
36. Shi, N.-Q., Prahl, K., Hendrick, J., Cruz, J., Lu, P., Cho, J.-Y., Jones, S., and Jeffries, T.
(2000), Appl. Biochem. Biotechnol. 84/86,201-216.
37. Billard, P., Menart, S., Fleer, R, and Bolotin-Fukuhara, M. (1995), Gene 162, 93-97.
38. Neuhauser, W., Haltrich, D., Kulbe, K. D., and Nidetzky, B. (1997), Biochem. J. 326,
683-692.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


276 Kang et al.
39. Ho, N. W. Y., Lin, F. P., Huang, S., Andrew, P. c., and Tsao, G. T. (1990), Enz. Microb.
Technol.12,33-39.
40. Bolen,P. L., Bietz,J.A.,andDetroy,R. W. (1985), Biotechnol. Bioeng. Symp.15,129-148.
41. Mayr, P., Bruggler, K., Kulbe, K. D., and Nidetzky, B. (2000), J. Chromatogr. B 737,
195-202.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 / 0277 / $20.00

Changing Flux of Xylose Metabolites


by Altering Expression of Xylose Reductase
and Xylitol Dehydrogenase
in Recombinant Saccharomyces cerevisiae

YONG-SU JINl AND THOMAS W. JEFFRIES*,1-3


Department of Food Science, University of Wisconsin,
1

1605 Linden Drive, Madison, WI 53706;


2Department of Bacteriology, University of Wisconsin,
1550 Linden Drive, Madison, WI 53706; and
3USDA Forest Service, Forest Products Laboratory,
One Gifford Pinchot Drive, Madison, WI 53705,
E-mail: twjeffri@facstaff. wisc.edu

Abstract
We changed the fluxes of xylose metabolites in recombinant
Saccharomyces cerevisiae by manipulating expression of Pichia stipitis genes
(XYLl and XYL2) coding for xylose reductase (XR) and xylitol dehydroge-
nase (XDH), respectively. XYLl copy number was kept constant by inte-
grating it into the chromosome. Copy numbers of XYL2 were varied either
by integrating XYL2 into the chromosome or by transforming cells with
XYL2 in a multicopy vector. Genes in all three constructs were under con-
trol of the strong constitutive glyceraldehyde-3-phosphate dehydrogenase
promoter. Enzymatic activity of XR and XDH in the recombinant strains
increased with the copy number of XYLl and XYL2. XR activity was not
detected in the parent but was present at a nearly constant level in all of the
transformants. XDH activity increased 12-fold when XYL2 was on a
multicopy vector compared with when it was present in an integrated single
copy. Product formation during xylose fermentation was affected by XDH
activity and by aeration in recombinant S. cerevisiae. Higher XDH activity
and more aeration resulted in less xylitol and more xylulose accumulation
during xylose fermentation. Secretion of xylulose by strains with multicopy
XYL2 and elevated XDH supports the hypothesis that D-xylulokinase limits
metabolic flux in recombinant S. cerevisiae.

Index Entries: Metabolic flux; metabolic engineering; xylose; xylose


reductase; xylitol dehydrogenase.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 277 Vol. 105-108,2003


278 }in and Jeffries

Introduction
Xylose-fermentation in lignocellulose hydrolysate is indispensable
for economic conversion of biomass to fuels and chemicals (1). Saccharomy-
ces cerevisiae is widely used for the industrial fermentation of glucose to
ethanol because it has the capacity for high specific ethanol production
rates (2,3) and will tolerate high ethanol concentrations (4). However,
native strains of S. cerevisiae do not use xylose as a carbon source for growth
(5-7). In contrast to S. cerevisiae, other yeasts use xylose very well. One of
the best xylose-fermenting yeasts, Pichia stipitis, has been studied exten-
sively for the regulatory and physiologic properties that enable it to fer-
ment xylose (8), and it has served as the source of genes for engineering
xylose metabolism in S. cerevisiae. The XYLl and XYL2 genes coding for
xylose reductase (XR) and xylitol dehydrogenase (XDH) from P. stipitis
have been expressed in S. cerevisiae (9-12). The resulting transformants can
grow on xylose aerobically but only produce significant amounts of etha-
nolin low yield under oxygen-limited conditions (13). This results in xylitol
accumulation in the medium as a byproduct, which is a major obstacle in
obtaining high ethanol production from xylose. XR from P. stipitis has a
high affinity for NADPH even though it can use NADH as a cofactor (14).
However, P. stipitis XDH only uses NAD+ as a cofactor (15). This difference
results in cofactor imbalance in the cytosol during xylose metabolism.
Bruinenberg et aL (16) hypothesized that xylitol accumulates during
the metabolism of xylose because yeasts cannot oxidize NADH to NAD+
efficiently through respiration under oxygen-limited conditions, and
because the cytosolic NAD+ /NADH ratio is unfavorable to the XDH reac-
tion. Walfridsson et al. (11) reported that the ratio of XR and XDH activity
is important in reducing xylitol formation in the fermentation of glucose/
xylose mixtures. They found that a strain with a lower ratio (0.06) of XR/
XDH activity accumulated less xylitol and produced more ethanol than a
strain with a higher ratio (17.5). However, they also obtained very low
xylose consumption rates and did not report the effect of aeration (11). This
prompted us to investigate whether or not increasing XDH activity can
reduce xylitol accumulation during metabolism of xylose alone under aero-
bic and oxygen-limited conditions. To alter XDH activity, we introduced
the XYL2 gene into the XYLl integrated S. cerevisiae either by integration
into chromosome or by expression from a multicopy plasmid. To deter-
mine whether we could further alter the ratio of reduced and oxidized
intermediate metabolites formed by the engineered cells, we also studied
the effect of aeration.

Materials and Methods


Microbial Strains and Plasm ids
The microbial strains and plasmidsused are listed in Table 1 (12,17,18).
We routinely used Escherichia coli DH5a (F- recAl endAl hsdR17 [rK- mK+]
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylose Metabolite Flux in Saccharomyces 279
Table 1
Plasmids and Microbial Strains Used
~ame Description Reference
Plasmids
pY2XR TRP1,2!lm origin, 12
GAPDHp-PsXYLl-GAPDHt
pY2XDH TRP1,2 !lm origin, 12
GAPDHp-PsXYL2-GAPDHt
pRS305 Yeast integration vector, URA3 20
pRS306 Yeast integration vector, LEU2 20
pYS10 URA3-GAPDHp-PsXYLl-GAPDHt This study
pYS20 LEU2-GAPDHp-PsXYL2-GAPDHt This study
Strain
S. cerevisiae L2612 MATaleu2-3 Ieu2-112 ura3-52 17
trp1-298 can1 cyn1 gaI+
S. cerevisiae FPL-YS10 Isogenic of L2612 This study
except for Ieu2::LEU2-PsXYLl
S. cerevisiae FPL-YS1020 Isogenic of L2612 This study
except for Ieu2::LEU2-PsXYLl,
ura3::URA3-PsXYL2
S. cerevisiae FPL-YS1022 FPL-YS10 (pYXDH) This study
P. stipitis FPL-UC7 ura3-3,~RRL Y-21448 18

supE44 thi-l gyrA relAl)(Gibco-BRL, Gaithersburg, MD) for gene cloning


and manipulation.
Enzymes, Primers and Chemicals
Restriction enzymes, DNA-modifying enzymes, and other molecular
reagents were obtained from New England Biolabs (Beverly, MA), Promega
(Madison, WI), Stratagene (La Jolla, CA), and Roche (Indianapolis, IN).
Reaction conditions were as recommended by the suppliers. All general
chemicals were purchased from Sigma (St. Louis, MO).

Media and Culture Conditions


Yeast and bacterial strains were stored in 15% glycerol at -70°C.
Yeast strains were routinely cultivated at 30°C in YP medium (10 giL of
yeast extract, 20 giL of Bacto peptone) with either 20 giL of glucose
(YPD), 20 giL of xylose (YPX-2%), or 40 giL of xylose (YPX-4%). YPD or
YPX plus 20 giL of agar was used for plates. To select for yeast trans-
formants using the URA3, TRP1, or LEU2 selectable markers, we used
Applied Biochemistry and Biotechnology Vol. 105-108,2003
280 Jin and Jeffries
6.7 giL of yeast nitrogen base (YNB) without amino acids plus 20 giL of
glucose, 20 giL of agar, and a mixture of appropriate nucleotides and
amino acids. To prevent loss of p Y2XDH during inoculum development
on glucose, we used yeast synthetic complete (YSC) medium, consisting
of6.7 giL ofYNB plus 20 glLof glucose supplemented with all necessary
amino acids and nucleotides except tryptophan. For fermentation,
recombinant strains were cultivated in YPX-2% or YPX-4% with xylose as
the sole sugar source, which maintained the xylose metabolism genes.
Aeration conditions were modulated by the choice of flask type or agita-
tion speeds in a shaker. We cultivated cells in 50 mL of medium in a
125-mL Erlenmeyer flask shaken at 100 or 200 rpm or in a 125-mL baffled
flask shaken at 200 rpm The oxygen transfer rates were determined by the
sulfite method (19). All cultures were incubated at 30°C.
Plasmid Construction
The plasmids used are summarized in Table 1. To construct an integra-
tion vector, p Y2XR (12) was digested with Hind III. The 2.3-kbp Hind 111-
Hind III fragment, which contains PsXYLl between the S. cerevisiae GAPDH
promoter and terminator, was inserted into the Hind III site of pRS305 (20).
The resulting plasmid was named p YS10. In a similar manner, the 2.8-kbp
Hind III-Hind III fragment, having PsXYL2 between the GAPDH promoter
and terminator, was isolated from p Y2XDH (12). The vector p YS20 was
constructed by inserting the 2.8-kbp Hind III-Hind IIlfragment into the
Hind III site of pRS306 (20).
Yeast Transformation
A yeast EZ-Transformation kit (BIO 101, Vista, CA) or Alkali-Cation
Yeast Kit (BIO 101) was used for yeast transformations. Integration vectors
were linearized with an appropriate enzyme prior to transformation.
Transformants were selected on YSC medium (21) containing 2% glucose
with 20 giL of agar plus appropriate nutritional supplements.
Preparation of Crude Extract and Enzyme Assays
Yeast cells were grown to midlog phase at 30°C in YSC with glucose
supplemented with appropriate nucleotides and amino acids. Cells were
harvested by centrifuging at 3000g for 5 min. The cell pellet was washed
and suspended in the buffer (50 mM phosphate buffer, pH 6.5). The sus-
pended cells were mixed with glass beads (Sigma) and vortexed at maxi-
mum speed for 5 min at 4°C. Vortexing was repeated two or three times
with intermittent cooling on ice. Cell debris was removed by centrifuging
at 13,000g for 10 min. XR (EC 1.1.1.21) activity was measured in a reaction
mixture with the following composition: 50 mM phosphate buffer, pH
6.0; 100 mM xylose; and 0.4 mM NADPH (14). XDH (EC 1.1.1.9) activity
was measured in a reaction mixture containing 50 mM Tris-HCl buffer,
pH 8.5; 4 mM NAD+, 5 mM MgCI2, and 100 mM xylitol (15). We used a
photo diode array spectrophotometer (Hewlett Packard, Wilmington, DE)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Xylose Metabolite Flux in Saccharomyces 281

to monitor the rate of NADPH oxidation and NAD+ reduction in the


reaction by absorbance at 340 nm. One unit of enzyme activity is defined
as the amount of enzyme that catalyzes 1 !lmol of substrate/min at 30°C.
Protein concentration was determined by the bicinchoninic acid method
(Pierce, Rockford, IL).
Analytical Methods
Glucose, xylose, xylitol, xylulose, and ethanol concentrations were
determined by high-performance liquid chromatography (Hewlett
Packard) with an ION 300 column (Interaction Chromatography, San Jose,
CA). Cell growth was monitored by optical density at 600 nm (OD600)' One
unit at 600 nm was equivalent to 0.167 g of dry cell/L.

Results and Discussion


Construction of Recombinant S. cerevisiae Strains
Having Different Levels of XDH Activity
The vector pYS10 containing PsXYLl under control of S. cerevisiae
GAPDHp was cut with StuI within URA3 and integrated into the chromo-
some of S. cerevisiae L2612 by homologous recombination. We observed a
significant increase in XR activity after transformation of p YS10. Crude
extract from FPL-YS10 showed XR activity of 0.367 V/mg. whereas cell
homogenate from L2612 did not show measurable XR activity (Table 2). We
then introduced PsXYL2 under the control of GAPDHp into FPL-YS10 by
two methods. First, we integrated the linearized p YS20 containing PsXYL2
into the S. cerevisiae chromosome atthe LEU210cus to construct FPL-YS1020.
Second, we transformed FPL-YSlO with the multi-copy vector pY2XDH,
which resulted in FPL-YS1022. As expected, specific activity of XR was
almost the same in both strains, but XDH activity increased 12-fold (0.625
vs 0.051 U / mg) in FPL-YS1022 compared with FPL-YS1020 (Table 2). Even
though we used GAPDHp for both the XYLl and XYL2 expression cassettes,
XR activity was seven times higher than XDH. XR activity remained essen-
tially constant even when XYL2 was highly expressed from the multicopy
(2!l) vector in FPL-YS1022.
Xylose Fermentation by Recombinant YSI020 and YSI022
Growth on xylose provides selective pressure to maintain genes cod-
ing for xylose assimilation (12,22). The parental strain L2612 and FPL-
YSlO could not grow on YPX-4%. Therefore, we used YP medium with 40
giL of xylose for xylose fermentation with the recombinant strains FPL-
YS1020 and FPL-YS1022. However, in order to prevent loss of the p Y2XDH
plasmid from FPL-YS1022 during preculture with glucose, YNB medium
supplemented with appropriate amino acids and nucleotides was used
for incoculum preparation. To change the oxygen transfer rate, fermenta-
tion was performed with 50 mL of medium in a 125-mL Erlenmeyer flask
or a 125-mL baffled flask, each of which was shaken at 200 rpm. Both
Applied Biochemistry and Biotechnology Vol. 105-108,2003
282 Jin and Jeffries
Table 2
In Vitro Enzyme Activities in Crude Extract
from Cell Cultures of Yeast Strains
Specific enzyme activity (U / mg)a
Strains XR XOH Xylulokinase
L2612 NO NO NO
YS10 0.367 NO NO
YS1020 0.354 0.051 NO
YS1022 0.357 0.625 NO
P. stipitis UC7 0.881 0.129 0.529
"Data are the average of at least two experiments and SDs were <10 %
for all assays. ND, not detected.

recombinant strains consumed and grew on xylose under both condi-


tions. Xylitol and xylulose accumulated in the media as byproducts. An
insignificant amount of ethanol «1 giL) was detected at the end of fer-
mentation in both cultures (Fig. 1). Interestingly, the pattern of byproduct
formation differed with the copy number of PsXYL2 gene and with the
oxygen transfer rate (Fig. 2). The FPL-YS1022 strain, containing PsXYL2
in a multicopy vector, produced more xylulose and less xylitol as com-
pared to the FPL-YS1020 strain that has PsXYL2 integrated into the chro-
mosome. The changes in XDH activity clearly affected metabolic fluxes
and resulted in altered byproduct formation patterns. The higher XDH
activity resulted in more xylulose and less xylitol accumulation in recom-
binant S. cerevisiae (Fig. 1).
Accumulation of xylulose in the medium has not been reported pre-
viously even though there have been many studies of XYLl and XYL2
expression in S. cerevisiae (9-12). This could be owing to product analysis
methods used in previous studies or to a difference among host strains.
Xylitol and xylulose accumulation were also affected by aeration. The
FPL-YS1022 strain accumulated more xylulose when cultivated in a
baffled flask than when cultivated in an Erlenmeyer flask (3.4 vs 2.6 giL
of xylulose), whereas the FPL-YS 1020 strain accumula ted almost the same
amount of xylulose in both flasks (Fig. 2). This suggested that XDH activ-
ity was limiting in FPL-YS1020 because increased aeration did not
enhance xylulose accumulation. However, FPL-YS1022, which showed
higher XDH activity, accumulated less xylitol with increased aeration.
The increase in xylulose accumulation with increased expression of XYL2
in FPL-YS1022 provided additional evidence that D-xylulokinase limits
xylose metabolism in S. cerevisiae (22). The most likely limitation is low
expression of endogenous S. cerevisiae xylulokinase (ScXKS1) (23). Alter-
natively, S. cerevisiae could be limited by other enzymes in the pentose
phosphate pathway (24).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylose Metabolite Flux in Saccharomyces 283

A B
40 8 40 8

35 7~ 35 7~
'0 '0
~ 30 6 [;l ~ 30 6 [;l
.... 0;9
'0 25 - .:l
:>.., 25 5r.l
~ 0: ~,., 11
.., "~ .., 20 4 ..
1,.,
~ 20 ~
4

."
0:
15 3 ] ,., .."
0:

!i
15 3 .!

~ ~ ~

:J
10 ~10
~

5
1l
~

5 ~j 1l
U OU
0 0
0 30 60 90 120 150 180 30 60 90 120 150 180
Tbne (bour) Time (hour)

Fig. 1. Xylose-fermentation profiles by recombinant S. cerevisiae expressing PsXYLl


and PsXYL2: (A) FPL-YSI020; (8) FPL-YSI022. Cells were cultured in YP medium with
40 giL of xylose at an agitation rate of 200 rpm. (£) xylose; (e) cell mass; (L,) xylitol;
(+) xylulose; (_) ethanol

A B
16

120 168
11me (hour)

Fig. 2. Comparison of (A) xylulose and (8) xylitol production during xylose fermen-
tation by recombinant S. cerevisiae. Strains and culture conditions: (a) FPL-YS1020 in
an Erlenmeyer flask at 200 rpm; (b) FPL-YSI020 in a baffled flask at 200 rpm; (c) FPL-
YS1022 in an Erlenmeyer flask at 200 rpm; (d) FPL-YSI022 in a baffled flask at 200 rpm.

Metabolic Flux Distributions in FPL- YSI 020 and FPL- YSI 022 Strains
Metabolic fluxes of the xylose assimilation steps were calculated from
the xylose consumption rates, and the xylitol and xylulose accumulation
rates (Table 3), during xylose fermentation by two of the recombinant
S. cerevisiae strains (FPL-YSI020 and FPL-YS1022). As shown in Fig. 3, alter-
ing levels of XDH activity significantly altered metabolic flux distributions
in xylose assimilation. Higher XDH activity in FPL-YSI022 decreased
xylitol accumulation 20% and increased xylulose accumulation 54% com-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
284 Jin and Jeffries
Table 3
Specific Rates During Xylose Fermentation by Recombinant S. cerevisiae
Specific rate (mM [g cell h]-I) "

Oxygen transfer Xylose Xylulose Xylitol


Strain rate (mM 02/h ) uptake production production
YS1020 4.3 1.24 ± 0.27 0.22 ± 0.03 0.31 ± 0.02
YS1022 4.3 1.78 ± 0.26 0.34 ± 0.03 0.24 ± 0.01
YS1020 5.3 1.09 ± 0.06 0.22 ± 0.03 0.26 ± 0.02
YS1022 5.3 1.70 ± 0.28 0.37 ± 0.03 0.21 ± 0.01
aData represent the average ± SD of three replicates. Specific rates shown here were
initial rates (from 0 to 24 h) of batch fermentations in YP medium with 40 giL of xylose.

A Xylose
B Xylose

24.9 13.7
15.4±0.61 75. 1 12.6 ± 0.61 86 •3
46.2± 13.5 76.9 ± 2.2

17.6
11.0 ± 1.7
X I I
Y U ose
19.1
17.0 ± 1.7
XYIUIose

157.5
35.7 ± 15.19 167.2
59.9 ± 0.4

Xylulose-5-phosphate Xylulose-5-phosphate

Fig. 3. Metabolic flux distributions in recombinant (A) S. cerevisiae FPL-YSI020 and


(B) FPL-YS1022 during xylose fermentation. Fluxes were calculated from initial xylose
consumption rates, and xylitol and xylulose accumulation rates of three independent
batch fermentations in YP medium with 40 giL of xylose at an oxygen transfer rate of
4.3 mM 02/h. Fluxes are represented as the average ± SD in micromoles per gram of
cell times hour. Bold figures correspond to the normalized fluxes with respect to xylose
uptake flux.

pared with FPL-YSI020. Although there was more than a 12-fold increase
in XDH activity in FPL-YSI022 compared with FPL-YSI020, the change in
metabolic flux was not proportional to the increase in enzymatic activity.
This clearly shows that not only enzyme activities but also physiologic
states of cell are responsible for the control of metabolic flux in the xylose
assimilation pathway.

Conclusion
During xylose fermentation by recombinant S. cerevisiae expressing
XYLl and XYL2 from P. stipitis, metabolic flux partitioning from xylitol
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylose Metabolite Flux in Saccharomyces 285

to xylulose depends on aeration and XDH activity. Increased aeration


resulted in less xylitol accumulation and more xylulose accumulation. We
hypothesize that the cytosolic NAD+ /NADH ratio becomes more favorable
for the XDH reaction under aerobic conditions. An increase in XDH activity
could reduce xylitol formation. However, the metabolic flux from xylitol to
xylulose does not increase proportionately with XDH activity.

Acknowledgments
We thank Prof. Jin-Ho Seo at Seoul National University for kindly
providing the S. cerevisiae L2612 and plasmids pY2XR and pY2XDH. We
also thank Prof. Culbertson at UW-Madison for generous access to his
strains and plasmids.

References
1. Hinman, N. D., Wright, J. D., Hoagland, W., and Wyman, C. E. (1989), Appl. Biochem.
Biotech. 20/21,391-401.
2. Maiorella, B. 1., Blanch, H. W., and Wilke, C. R. (1984), Biotechnol. Bioeng. 26, 1003-1025.
3. Jeffries, T. W. (1985), Trends Biotechnol. 3,208-212.
4. Casey, G. P. and Ingledew, W. M. M. (1986), Crit. Rev. Microbiol. 13,219-280.
5. Chiang, L.-c., Gong, c.-S., Chem, L.-F., and Tsao,G. T. (1981),Appl. Environ. Microbiol.
42,284-289.
6. Senac, T. and Hahn-Hagerdal, B. (1990), Appl. Envinron. Microbiol. 56, 120-126.
7. Wang, P. P. and Schneider, H. (1980), Can. J. Microbiol. 26, 1165-1168.
B. Jeffries, T. W. and Shi, N. Q. (1999), Adv. Biochem. Eng. Biotechnol. 65, 117-161.
9. Kotter, P., Amore, R., Hollenberg, C. P., and Ciriacy, M. (1990), Curro Genet. 18,493-500.
10. Tantirungkij, M., Nakashima, N., Seki, T., and Yoshida, T. (1993), J. Ferment. Bioeng.
75,83-88.
11. Walfridsson, M., Anderlund, M., Bao, X., and Hahn-Hagerdal, B. (1997), Appl.
Microbiol. Biotechnol. 48, 218-224.
12. Jin, Y. S., Lee, T. H., Choi, Y. D., Ryu, Y. W., and Seo, J. H. (2000), J. Microbiol. Biotechnol.
10, 564-567.
13. Tantirungkij, M., Izuishi, T., Seki, T., and Yoshida, T. (1994),Appl. Microbiol. Biotechnol.
41,8-12.
14. Verduyn, c., Van Kleef, R., Frank, J., Schreuder, H., Van Dijken, J. P., and Scheffers,
W. A (1985), Biochem. J. 226,669-677.
15. Rizzi, M., Harwart, K., Erlemann, P., Buithanh, N. A., and Dellweg, H. (1989),
J. Ferment. Bioeng. 67, 20-24.
16. Bruinenberg, P. M., de Bot, P. H. M., van Dijken,J. P., and Scheffers, W. A (1983), Eur.
J. Appl. Microbiol. Biotechnol. 18,287-292.
17. Cho, K. M., Yoo, Y. J., and Kang, H. S. (1999), Enzyme Microb. Technol. 25,23-30.
lB. Lu, P., Davis, B. P., Hendrick, J., and Jeffries, T. W. (1998), Appl. Microbiol. Biotechnol.
49,141-146.
19. Cooper, C. M., Fernstrom, G. A, and Miller, S. A (1944), Ind. Eng. Chem. 36,504-509.
20. Sikorski, R. S. and Hieter, P. (1989), Genetics 122, 19-27.
21. Rose, M. D., Winston, F., and Hieter, P. (1990), Methods in Yeast Genetics: A Laboratory
Course Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New
York, NY.
22. Lachke, A and Jeffries, T. W. (1986), Enzyme Microb. Technol. 8,353-359.
23. Ho, N. W. Y., Chen, Z. D., and Brainard, A P. (1998), Appl. Environ. Microbiol. 64,
1852-1859.
24. Walfridsson, M., Hallborn, J., Penttila, M., Keranen, S., and Hahn-Hagerdal, B. (1995),
Appl. Environ. Microbiol. 61,4184-4190.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0287/$20.00

Effect of Media on Spore Yield and Thermal


Resistance of Bacillus stearothermophilus

THEREZA CHRISTINA VESSONI PENNA, *


IRENE A. MACHOSHVILI, AND MARINA ISHII
Department of Biochemical and Pharmaceutical Technology,
School of Pharmaceutical Science, University of Sao Paulo,
Rua Antonio de Macedo Soares, 452 04607-000,
Sao Paulo/SP, Brazil, E-mail: tcvpenna@usp.br

Abstract
The interference of eight components in the yield of sporulation and ther-
mal resistance to moist heat (121°C) of Bacillus stearothermophilus spores sus-
pended in 0.02 M calcium acetate solution and inoculated on paper strips
previously treated with calcium acetate/calcium hydroxide was studied.
The spore yield of 1.0 x 108 / mL was developed at 62°C in 17 media containing
different concentrations of D-glucose, sodium chloride, L-glutamic acid, yeast
extract, peptone, manganese sulfate, potassium phosphate, and ammonium
phosphate. The combined effects of yeast extract, peptone, and glucose con-
tributed positively to the spore yield and to the stability of the thermal resis-
tance of both spores in suspension and on strips.
Index Entries: Bacillus stearothermophilus; thermoresistance; sporulation;
bioindicator.

Introduction
The length of the exponential growth phase determines the final cell
population of Bacillus stearothermophilus. Symmetrical cell division depends
on favorable exogenous conditions such as temperature, humidity, nutri-
ents, and aeration. The sporulation of B. stearothermophilus is characterized
by a multiphase, morphologic synchronized process that occurs at the onset
of the stationary phase (1). During the maturation of the spore, thermo-
resistance characteristics develop. Although the complex morphologic
changes that occur during sporulation result from regulated changes in
gene expression, the thermal resistance of mature spores can be manipu-
lated in vitro (2-4). Spores in suspension, when treated with sodium, potas-
sium, magnesium, and manganese cations, are more resistant than spores
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 287 Vol. 105-108, 2003


~
[
0;)

9-
(1)
3
tn·
~
OJ Table 1
::J
Cl.. Media Used for Sporulation of B. stearothermophilus ATCC 7953 to Evaluate
0;)
o· Interference of Eight Components in Thermal Resistance of Developed Spores
~
::J-
::J Media concentrations (% w Iv)
o
~
'<: Components 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
D( +)Glucose 0.25 0.25 0.018 0.25 0.018 0.018 0.018 0.25 0.018 0.018 0.25 0.018 0.25 0.018 0.25 0.018 0.25
N L-Glutamic acid 0.04 0.04 0.04 0.04 0.04 0.04
0:>
0:> Yeast extract 0.4 0.4 0.4 0.05 0.05 0.4 0.4 0.4 0.4 0.05 0.05 0.05 0.05 0.4 0.4 0.4 0.4
Peptone 0.5 0.3 0.3 0.3 0.5 0.5 0.3 0.5 0.5 0.3 0.3 0.5 0.5 0.3 0.3 0.5 0.5
NaCl 0.001 0.001 0.001 0.001 0.001 0.001 1.0 1.0 1.0
MnS04 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001
NH4H 2P0 4 0.035 0.035 0.035
KH2 P04 0.05 0.05 0.05
Agar 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
pH 9.84 10.02 9.30 9.50 9.50 9.05 9.63 9.10 9.4 9.4 9.6 9.6 9.6 9.6 9.6 9.6 9.6
"The final pH of the medium was adjusted to 9.7 ± 0.1 with Ca(OH)2' L-Glutamic acid was added to the medium following autoclaving at 121°C
for 30 min of the other components dissolved in water.
~

,§5
-~
to."
a
8
Media and Spore Yield ofB. sterothermophilus 289
in the hydrogenionic form. The hydrogen form of the spores can be con-
verted in a calcium-resistant form by treatment with calcium acetate (CaAc)
at alkaline pH (2). Acid conditions of the heating menstruum may cause
partial demineralization of the spores (3). On the other hand, the mineral-
ization provides a reduction in the content of water in the protoplast of the
spore, provoking an increase in thermal resistance (4). When inoculated on
strips previously treated with 0.02 M CaAc, the B. stearothermophilus ATCC
7953 spores displayed greater stability (5).
Mineralization of the cortex depends either on the supply of the sporu-
lation medium or on the suspension of harvested spores (6). Thermal resis-
tance is associated with the peptidoglycan present in the spore cortex. At
low pH, the acid protonates the peptidoglycan, decreasing thermal resis-
tance and the viability of spores. Therefore, the pH of the culture medium
influences spore yields, which at pH 8.7 were higher than those obtained in
media at pH values between 7.0 and 5.5 (6).
Spores of B. stearothermophilus are mainly used as a bioindicator (BI) to
monitor and ensure the reproducibility of moist heat processes (7,8).
Bioindicator's are commercially available in suspension form, for impreg-
nation in the load unit; inoculated on paper carriers; or in a self-contained
ampoule where spores are suspended in a recovery medium. Strips
of paper, which are inexpensive and small, are easily placed in strategic
positions in order to verify the homogeneity of the sterilant distribution in
the chamber and in the unit of the load.
The purpose of the present study was to evaluate the interference of
eight components in the yield of sporulation and thermal resistance to
moist heat (121°C) of B. stearothermophilus spores in a suspension of 0.02 M
CaAc and on strips previously treated with CaAcl calcium hydroxide
(Ca[OH]2) (pH 11.0) (9). The principal goal was to verify which component,
individually or in association, showed the greatest influence on viability
and thermal resistance.

Materials and Methods


Culture Media
Sterile techniques were carried out in a class 100 flow cabinet and the
same lots of component (Sigma, St. Louis, MO) and culture media (Difco)
were used: Plate count agar (PCA), (0.5% tryptone, 0.25% yeast extract,
0.1 % dextrose, and 1.5% agar at pH 7.0 ± 0.2), Trypticase soy broth (TSB),
(1.7% tryptone, 0.3% soytone, 0.25% dextrose, 0.5% NaCl, and 0.25% dipo-
tassium phosphate at pH 7.3 ± 0.2) and Trypticase soy agar (TSA), 1.5%
tryptone, 0.5% soytone, 0.5% NaCl, and 1.5% agar at pH 7.3 ± 0.2).
Spore Production
B. stearothermophilus ATCC 7953 spore suspension cultures were
developed in 17 media (Table 1) containing the respective components and
concentrations (% [w Iv]): D-glucose (0.018-0.25%), NaCI (0.001-1.0%),L-glu-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


290 Penna et al.
tamic acid (0.04%), yeast extract (0.05--0.4%), peptone (O.3-O.5%),manganese
sulfate (0.001 %), potassium phosphate (0.05%), and ammonium phosphate
(0.05%). After24hat62°C, smoothmorphotype variant single colonies (small
and transparent) were harvested from PCA tube slants, resuspended in 0.1 %
peptone salt fluid, and each 5-mL suspension was heat shocked at 120°C for
5 min. Each 200-mL medium (Roux flask) was inoculated with 10 mL of
activated suspension. After 6 d at 62°C saturated humidity, growth was
harvested in 0.02 M CaAc, and the suspension was adjusted to pH 9.7 with
0.14% Ca(OH)2 and centrifuged (1935g for 30 min) four times. The spore
pellets were suspended in chilled 0.02 M CaAc (pH 9.7), and stored at 4°C (8).
The viability of 107-108 colony-forming units(CFU)/mL (average CFU
spores/mL) of heat-shocked spores in suspension (97.8°C for 30 min) was
estimated through pour plates in TSA at 62°C for 48 h, from a minimum of
ten plates.

Preparation of Bioindicators Formed


by Spores Impregnated on Strips (6)
Sterile paper strips (250 g/m2, 7 mm wide, 0.2 mm deep, 20 mm
long) were immersed in 0.02 M CaAc (pH 11/ 0.14% Ca[OH]2) for 12 h to
a final pH of 9.0 ± 0.1, dried at 45°C for 24 h and stored at -18°C. The
monolayer disposed strips were individually inoculated with 0.1 mL of
B. stearothermophilus homogenized suspension (pH 9.7). The impregnated
strips (bioindicators) were dried at 45°C for 24 h and stored at -18°C. Strip
spore quantification was performed by blending 10 strips in 100 mL of
0.1 % peptone' salt fluid. Standard pour plate enumeration in TSA (62°C
for 48 h) was used for initial heat shocked (97.8°C for 30 min) spore-per-
strip density determination.

Thermal Treatments
Thermal treatments at 121°C were performed through the serum bottle
technique apparatus (5,9). The homogenized spore suspensions (5 mL) were
transferred into 20-mL glass serum bottles (Wheaton SB205A; 60 x 25 mm).
Ten B. stearothermophilus spore strips, previously humidified by instanta-
neous contact with sterile distilled water, were transferred to 20-mL serum
bottles and placed between 10 Durham tubes (30 x 3 mm). The distance
between neighboring parallel tubes was 4.0 ± 0.5 mm, and each Durham
tube was filled with water to two thirds its total capacity. The bottles were
closed with a sterile flange rubber stopper and then sealed with an alumi-
num seal.
Air was removed (5 min) by suction (General Electric vacuum pump,
1725 rpm, 1/6 Hp), with a 22-gage stainless steel needle (30 x 7 mm) through
the rubber stopper of the sealed bottles. A thermocouple sensor end, inside
a 316 stainless steel needle (1.5 x 100 mm), was placed in the geometric
center of the bottles, using the hole through the rubber stopper left by the
pump needle. Triplicate sets of bottles were heated in a thermostatically

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Media and Spore Yield ofB. sterothermophilus 291

controlled oil (Dow Coming® silicone 200/220 CS, P = 0.948 g/cm3) bath
(OTB-7A, Haven Automation) at different time intervals at 121.0 ± O.l°C
and were cooled and held in an ice-water bath. The experimentally deter-
mined lag correction of 7 min was used. The thermocouples, type J
(2x32AWG), overwrapped with Teflon, were equilibrated at 121.0 ± O.l°C
and attached to multipoint recording equipment (lOPE therm 400-CE 12;
recording rate: 30 s). Survivors on the strips were quantified for spore
viability and expressed by decimal logarithms of the average colony-form-
ing units per strip, from at least 10 plates for each time heating condition
and system used.

Determination of Decimal Reduction Times


The decimal reduction times (D value, the time in minutes to kill 90%
of the initial spore population at 121°C) were determined from the nega-
tive reciprocal of slopes (b l ) of the regression lines, using the linear por-
tions of the survivor curves (ioglo survivors vs time). The angular
coefficient standard error SE (b l ) and the multiple determination coeffi-
cient (R2) were also calculated. The estimated value of the D value stan-
dard error SE(D) was obtained from the root of the variance (V[D])I/2,
which was calculated as follows:
V(D) = ([1/b I 2 ]2 x [SE{bl }]2)
The upper (D - mean + SE[D]) and lower (D - mean - SE[D]) D values
were also expressed.

Results and Discussion


Seventeen different media were used to examine, simultaneously,
the influence of cations (Ca++, Mg++, NH/, K+, Mn++, Na+) and the concen-
tration of D-glucose, peptone, and yeast extract on the formation of
B. stearothermophilus spores and their thermal resistance. The initial popu-
lation of the activated spores varied from 1.0 x 107 to s.o x 108 spores/mL
of 0.02 M CaAc (Table 2). The pH of the CaAc suspension varied from 7.32
to 7.80 after 1 mo storage at 6 ± 1°C. Following readjustment with Ca(OHh,
the pH was equilibrated between 9.30 and 10.02, where it remained stable.
For an overkill in moist heat sterilization process (total time =12 d), Bls
containing C!:S x lOS spores of B. stearothermophilus / carrier should present
a D I2I ,C value of about 1.9 min (7). Therefore, the thermal resistance of the
bioindicators's formed by spores in suspension and by spores on strips can
be classified (9S% confidence), according to D I2I ,C values, into three groups:
group 1 (D I2I ,C < 2.0 min); group 2 (2.0min :s Dl2l ,c :s 2.46 min); and group
3 (DI2I ,C > 2.46 min).
In group 1, spores developed in media 4 and 8 showed a very close
average Dl2l ,c for both Bls in suspension (1.76 and 1.71 min) and on strips
(1.82 and 1.87 min). In medium 4, the maximum glucose concentration was
incorporated with the minimum concentrations of yeast extract, peptone,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


).
"0
"0
i%"'
Q.. Table 2
Co
o· Decimal Reduction Times (D values, min)
":J-
<1>
at 121°C for Spores Developed in 1-17 Media a
:3
tn· Spores in suspension Spores on strips
~
Adjusted pH Initial D l2l ,c values (min) D 121 ,C values (min)
:J
'"Q..
Co Medium Before After (spores/mL) Mean Lower Upper Mean Lower Upper

c;;
":J-
:J
1 7.81 9.84 5.0 x 108 2.41 2.32 2.51 2.0 1.89 2.11
0 2 6.68 10.0 3.0 X 108 1.54 1.49 1.60 1.94 1.86 2.02
~
'<: 3 9.30 2.0 x 107 2.66 2.54 2.78 1.90 1.82 2.00
4 9.50 3.0 x 108 1.76 1.71 1.82 1.82 1.74 1.91
5 9.50 2.0 x 108 2.71 2.64 2.77 2.34 2.22 2.47
N 9.05 2.0 x 108 2.09 2.04 2.15 1.57 1.51 1.63
<..0 6
N 7 9.63 2.0 x 108 2.41 2.32 2.50 2.47 2.40 2.53
8 9.10 5.0 x 107 1.71 1.65 1.77 1.87 1.82 1.93
9 9.40 1.0 x 107 2.00 1.94 2.07 1.83 1.74 1.92
10 8.06 9.60 8.9 x 107 1.86 1.80 1.92 2.26 2.18 2.33
11 7.32 9.60 5.1 x 107 1.84 1.76 1.92 2.48 2.37 2.59
12 7.80 9.60 6.0 x 107 2.84 2.73 2.94 2.32 2.19 2.44
13 7.46 9.60 5.0 x 107 2.36 2.27 2.44 2.26 2.18 2.34
14 7.27 9.60 3.4 x 107 2.15 2.07 2.23 2.79 2.59 2.99
15 7.54 9.60 9.5 x 106 2.23 2.10 2.36 2.69 2.54 2.84
~ 16 7.46 9.60 1.4 x 107 2.76 2.62 2.88 3.46 3.28 3.63
:- 2.25 2.53
17 7.60 9.60 1.0 x 108 2.86 2.68 3.04 2.39
0
-1" aD 121 ,C values (min), decimal reduction times; D value, the time in minutes to kill 90% of the initial spore population at 121°C.
-0
,Co
to..>
0
0
w
Media and Spore Yield ofB. sterothermophilus 293
and NaCl. By contrast, in medium 8, NaCI at a maximum concentration
showed a deleterious effect on the BIs thermoresistance, even in the pres-
ence of glucose, peptone, and yeast extract also at maximum concentration.
In medium 2, glucose and yeast extract were present in maximum concen-
trations and peptone and NaCl in minimum concentrations, and the D
value of 1.54 min for BIs in suspension was the lowest obtained, although
the D value was 1.94 min for BIs on the strips. With the D values of 1.86 and
1.84 min for the spores in suspension, respectively from media 10 and II,
the minimum concentration of yeast extract in the absence of NaCl, was
sufficient to maintain the thermal resistance of bioindicators stable on
the strips (2.26 and 2.48 min, respectively). The spores on the strips from
medium 7 presented the lowest D value of 1.57 min, with the maximum
concentrations of yeast extract and peptone and minimum concentrations
of glucose and NaCl. The negative interference of the maximum concentra-
tions of glucose and NaCl in the thermoresistant properties of the spores
confirmed the data obtained for the spores developed in 32 media prepared
by 11 components (9). The presence of ammonium phosphate, yeast, and
peptone, all at maximum concentrations, did not annul but merely partially
canceled the negative effect of NaCI and glucose on the Dl21 c values.
0

Group 2 comprised the majority of the BIs in suspension and on


strips. In this group, it was observed that the deleterious effect of glucose
at maximum concentration (medium 1) was partially neutralized in the
presence of yeast and/ or peptone. In media 7 and 9, even with glucose at
a minimum concentration, the presence of N aCI at a maximum concentra-
tion annulled the positive effects of yeast and/ or peptone. The presence
of the other salts, ammonium phosphate, potassium phosphate, and
manganese sulfate revealed no influence on the thermoresistance of the
BIs. This fact was confirmed for the spores from media 13-15.
In group 3, the BIs that showed Dl21 C values >2.46 min for the spores
0

in suspension were developed in media 3, 5, 12, 16, and 17, with Dl21 c 0

values equal to 2.66, 2.71, 2.84, 2.76, and 2.86 min, respectively, and for the
spores on strips from the media 15 and 16, respectively, with D12l"c values
equal to 2.69 and 3.46 min. Glucose was present at the minimum concentra-
tion, with the exception of media 15 and 17. Yeast extract was present at the
maximum concentration in media 3, 15, 16, and 17; peptone was incorpo-
rated at the maximum concentration in media 5,12,16, and 17. Both com-
ponents were simultaneously incorporated at their respective maximum
concentrations in media 16 and 17, where the presence of glucose at the
maximum concentration in media 17 interfered only slightly in the
thermoresistance of the spores on the strip. Medium 16 showed the best
conditions for maturing the spores, providing the nutrients required to
confer maximum thermal stability to spores developed over a period of 6
d at 62°C, with saturated humidity.
Peptone at a maximum concentration is an excellent source of mineral
salts and amino acids at concentrations required for the formation of
mature spores and the maintenance of thermal resistance. Glucose is a

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


294 Penna et al.

source of carbon that supplies the energy required for the logarithmic phase
of development before the stationary growth phase is reached and then
forms the spores. Yeast is an excellent source of B complex vitamins essen-
tial for growth and sporulation.
The components peptone and yeast at maximum concentrations and
glucose at minimum concentration provided good viability, about 1 x 108
spores/mL, with a D value at 121°C> 1.5 min for both BIs (the spores in
suspension and on strips), matching international standards.

Acknowledgments
This study was made possible by financial support provided by Bra-
zilian Committees for Scientific Technology Research (Conselho Nacional
de Desenvolvimento Cientifico e Tecnol6gico and Funda~ao de Amparo a
Pesquisa do Estado de Sao Paulo).

References
1. Foster, S. J. (1994), J. Appl. Bacteriol. Symp. Suppl. 76,255-39S.
2. Reich, R R, Whitbourne, J. E., and McDaniel, A W. (1979), J. Parenter. Drug Assoc. 33,
228-234.
3. Marquis, R E. , Sin, J., Shin, and S. Y. (1994), J. Bacteriol. 76, (Suppl.) 405-48S.
4. Beaman, T. C. and Gerhardt, P. (1986), Appl. Environ. Microbiol. 52,1242-1246.
5. Vessoni Penna, T. c., Machoshvili, 1. A, and Taqueda, M. H. S. (1996), J. Pharm. Sci.
Technol. 50,1-11.
6. Vessoni Penna, T. c., Machoshvili, 1. A, and Taqueda, M. E. S. (1998), J. Parenter. Sci.
Technol., 52, 198-208.
7. United States Ph.armacopeia, 23rd and 24th eds. (1995 and 2000), United States
Pharmacopeial Convention, Rockville, MD.
8. Graham, C. S. and Boris, C. A. (1993), In Sterilization Technology: A Practical Guide for
Manufacturers and Users of Health Care Products, Morrissey, R F. and Phillips, G. B.
eds., Van Nostrand, Reinhold, NY, pp. 36-69.
9. Vessoni Penna, T. c., Machoshvili, 1. A, Taqueda, M. E. S., and Ishii, M. (2000), J.
Parenter. Sci. Technol., 54, 398-412.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 /0295 /$20.00

Cassava Flour Wastewater as a Substrate


for Biosurfactant Production

MARCIA NnSCHKE* AND GLAUCIA M. PASTORE


Laborat6rio de Bioqufmica, Departanento de Ciencia de Alinentos,
Faculdade de Engenharia de Alimentos,
UN/CAMP, Rua Monteiro, Lobato, 80 Cx. Postal 6121,
CEP 13083-970, Campinas, SP, Brazil, E-mail: nitschke@bol.com.br

Abstract
Five cassava flour wastewater (manipueira) preparations were tested as
culture media for biosurfactant production by a wild-type Bacillus sp. isolate.
No-solids (F), no-solids diluted (F /2), natural (I), natural diluted (II2), and
decanted (IPS) were the tested manipueira media. The microorganism was
able to grow and to produce biosurfactant on all manipueira preparations.
The media whose solids were removed (F and F /2) showed better results
than preparations with the presence of solids (I, II2, and IPS). No-solids
medium (F) showed a surface tension of 26,59 mN / m and reciprocal of criti-
cal micelle concentration of over 100 and was selected as a potential substrate
for biosurfactant production.

Index Entries: Biosurfactant; Bacillus sp.; cassava flour wastewater.

Introduction
Surface-active molecules produced by microorganisms, also called
biosurfactants, are of great interest as substitutes for chemically derived
surfactants mainly because of their low toxicity and biodegradability 0,2).
Biosurfactants find applications in the cosmetic, pharmaceutical, and food
industries as emulsifiers, humectants, dispersants, and detergents (3,4).
Moreover, they are ecologically safe and suited for environmental applica-
tions such as bioremediation, dispersion of oil spills, and waste treatment
(5,6). Although interest in biosurfactants is increasing, these compounds
are not competitive with synthetic surfactants owing to high production
costs (7). The use of alternative substrates (substitutes for conventional
media), usually renewable resources, could be explored since substrates
account for 50% of final product costs (8).

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 295 Vol. 105-108,2003


296 Nitschke and Pastore
Selecting an alternative resource involves finding a residue with the
right balance of nutrients to support optimal growth and production (8,9).
Agroindustrial residues with high contents of carbohydrates or lipids meet
the requirements for use as substrates for biosurfactant production. Peat
hydrolysate, olive oil mill effluent, lactic whey, and potato process efflu-
ents are some examples of low-cost substrates that have been reported for
biosurfactant production (10-14).
One of the biggest problems faced by cassava flour factories in Brazil
is the disposal of manipueira, a liquid waste generated in large amounts
(300 Lit of processed roots) (15). The disposal of manipueira causes envi-
ronmental problems owing to its high organic load, although this waste
is composed of carbohydrates, nitrogen, minerals, and trace elements and
therefore has potential as a substrate for the biotechnological processes.
In the present study, we investigated the production of biosurfactant by
a Bacillus sp. isolate using different manipueira preparations as substrates.

Materials and Methods


Microorganisms
A wild-type Bacillus sp. LBSa isolate selected for its capacity to decrease
surface tension on manipueira medium (16) was maintained on nutrient agar
(Difco, Detroit, MI) slants at 4°C.

Preparation of Substrate
Manipueira from the manufacture of cassava flour was collected at Sao
Paulo State and stored at -18°C until needed. The composition of manipueira
waste utilized is given in Table 1. Five different media were prepared from
the original waste and autoc1aved at 121°C for 15 min:
1. Natural manipueira medium (I) with the presence of insoluble solids
(6.2% [w Iv] total solids) was dispersed and distributed in Erlenm-
eyer flasks. The initial pH was 5.9.
2. Naturally diluted manipueira (1/2) was natural manipueira diluted 1:2
with distilled water (3.8% [w Iv] total solids). The initial pH was 5.8.
3. Decanted manipueira (IPS) was natural manipueira waste maintained
for 1 h at room temperature to permit the decantation of solids. The
upper liquid was gently removed and distributed in flasks (2.3%
[w Iv] total solids). The initial pH was 5.8.
4. No solids manipueira (F) was prepared by heating the waste to boil-
ing point to remove all the solids. After cooling, the substrate was
centrifuged at SOOOg for 20 min in a centrifuge (model J2-21;
Beckman). The supernatant was distributed in flasks. The initial pH
was 5.S.
5. No solids diluted manipueira (F 12) was no solids manipueira diluted
1:2 with distilled water. The initial pH was 5.9.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Manipueira in Biosurfactant Production 297
Table 1
Composition, of Cassava
Flour Wastewater (Manipueira)
Parameter Value
Total solids (giL) 62.0
Total carbohydrates (giL) 41.5
Reducing sugars (giL) 18.2
No reducing sugars (giL) 23.2
Total nitrogen (giL) 2.1
Phosphorus (mg/L) 244.5
Potassium (mg/L) 3472.6
Calcium (mg/L) 292.5
Magnesium (mg/L) 519.0
Sulfur (mg/L) 154.0
Iron (mg/L) 7.8
Zinc (mg/L) 2.8
Manganese (mg/L) 1.7
Copper (mg/L) 1.0
pH 5.8

Inoculum and Culture Conditions


The bacterial isolate was streaked in a nutrient agar slant and incu-
bated for 24 h at 30°C. A loop of culture was inoculated in 20 mL of nutrient
broth (Difco) in a 50-mL Erlenmeyer flask and incubated in a rotary shaker
(model G25R; New Brunswick) for 7 h, at 150 rpm and 30°C. An aliquot
of 1 mL of inoculum was transferred to 15 mL of manipueira medium in a
50-mL Erlenmeyer flask and incubated at 30°C and 150 rpm in a rotary
shaker (model G25R; New Brunswick). Samples were collected at set time
intervals and submitted for analysis. The experiments were conducted in
three independent replicates.
Analytical Measurements
Samples were submitted to serial dilutions and viable counts were
performed by spread plate technique. Total carbohydrates were estimated
using the phenol-sulfuric assay (17).
Culture samples were centrifuged at 8000g for 20 min for cell removal.
The supernatant was submitted to surface tension critical micelle dilution
(CMD) and reciprocal of critical micelle concentration (CMC-l) analysis.
Surface tension was determined with a Kriiss Processor Tensiometer
(model K12 T) using the plate method. For CMD measurements, the cell-
free broth was diluted 10 (CMD-l) and 100 times (CMJ)-2) with distilled
water, and the surface tension of these dilutions was determined. The
maximum SD allowed for surface tension measurements was 0.20.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


298 Nitschke and Pastore
Table 2
Comparison of Data Obtained from Different Manipueira Media

Total Viable Surface Crude


Time carbohydrates count tension CMD-l CMD-2 biosurfactant
Medium (h) (g/L) (ufc/mL) (mN/m)a (mN/m)b (mN/m)b (g/L)

I 0 43.57 1.0 x 107 44.04 54.08 66.49


24 24.65 1.8 x 109 31.57 34.56 48.07 1.23
48 15.95 1.36 x 109 29.86 31.99 41.04 1.84
72 9.58 7.0 x 108 29.14 29.66 37.97 1.69
1/2 0 27.93 1.0 x 107 46.57 57.06 67.43
24 11.74 1.68 x 109 29.16 31.05 44.09 0.90
48 6.44 9.8 x 108 29.37 31.22 41.31 1.00
72 5.50 1.47 x 109 28.13 29.07 36.86 1.36
F 0 30.24 1.0 x 10 7 49.47 54.94 59.45
24 8.22 1.48 x 109 26.57 27.24 34.05 2.34
48 4.99 9.3 x 108 26.87 27.20 32.42 3.17
72 4.60 1.32 x 109 26.59 26.92 32.06 2.74
F/2 0 18.24 1.0 x 107 51.31 57.42 64.16
24 4.25 4.8 x 108 26.47 27.46 36.38 1.05
48 4.04 7.1 x 108 26.67 27.43 35.89 1.29
72 3.44 1.15 x 109 26.63 27.45 35.85 1.32
IPS 0 44.36 1.0 x 107 45.24 51.41 59.54
24 15.63 2.25 x 109 29.89 31.48 40.45 1.33
48 8.92 6.6 x 108 27.9 28.8 36.7 1.81
72 8.38 1.48 x 109 27.51 27.99 34.74 2.23
aMaximal SD of 0.20.
bCMD-l = CMD lOX; CMD-2 = CMD 100X.

The critical micelle concentration (CMC) is a useful index for evaluat-


ing surface activity and is defined as the surfactant concentration at which
an abrupt increase in surface tension is observed (18). The CMC-l is the
dilution factor required to reach the CMC and was determined by measur-
ing the surface tension of serial diluted broth samples as described by
Sheppard and Mulligan (10 ).
Biosurfactant was isolated from cell-free broth by precipitation after
adjusting broth pH to 2.0 using 6 N HCI and keeping it at 4°C overnight.
The precipitate thus obtained was pelleted at 8000g for 20 min, dried, and
weighed.

Results
The data obtained during cultivation of Bacillus sp. LB5a in tested
cassava wastewater media are presented in Table 2. The microorganism
was able to grow and produce biosurfactant in all tested media. Growth
patterns were similar; after 24 h cells reached stationary phase. Total car-

Applied Biochemistry and Biotechnology Vol. 105-708,2003


Manipueira in Biosurfactant Production 299
120

100

80
':"t)
::!E 60
t)

40

20 112 _IPS-o-I

0
0 20 40 60 80 100
time(h)

Fig. 1. Time course of CMC-l on manipueira media.

bohydrates consumption was about 80% for all media tested, with most of
the carbohydrates (>50%) being consumed during the first 24 h of fermen-
tation. No-solids media (F and F /2), which had a lower carbohydrate con-
tent, demonstrated the lowest surface tension results, whereas the high
carbohydrate content of media I and IPS did not significantly increase cell
growth or surface tension. No-solids media showed a surface tension of
26.57 mN /m after 24 h, while on solid-content media (I, IPS, and 1/2) the
lowest surface tension attained was 27.5 mN/m after 72 h (IPS).
CMD-l and CMD-2 data were helpful as an indirect indication of
surfactant concentration on culture media (19). The results obtained from
medium F after 72 h showed a CMD-l of 26.92 mN / m and a CMD-2 of 32.06
mN/m, whereas for F/2, although the surface tension was similar, the
values of27.43mN/m (CMD-l) and 35.82 mN/m (CMD-2) suggested that
surface activity was higher on F medium. In fact, the crude biosurfactant
concentration obtained on manipueira F corroborates this statement.
More than 55% of crude biosurfactant was produced during the first
24 h of cultivation, and surface tension reached minimum levels in approx
48 h. After 48 h, biosurfactant production was considerably reduced and
even a decrease in biosurfactant concentration was observed for media I
and F. The decrease in surface activity also occurred at minor rates and
could be observed by CMD data.
The CMC-l obtained for the tested media is shown in Fig. l.
Manipueira media with solids contents reached approximately the same
Applied Biochemistry and Biotechnology Vol. 105-708,2003
300 Nitschke and Pastore
values of CMC-l after 96 h while no-solids media showed the highest
values. Time course evaluation indicated that F medium had the highest
surface activity and at approx 15 h, the CMC-l was about 80, reaching 105
after 72 h. After 96 h, a CMC-l of 114 was attained; therefore, culture broth
should be diluted 114 times to reach 0.87% (v Iv) CMC.

Discussion
The carbohydrates present in manipueira were consumed, and the
levels of iron and manganese, important factors for biosurfactant produc-
tion by Bacillus (20,21), suggested thatthe manipueira medium is a suitable
substrate for supporting cell growth and biosurfactant accumulation by
Bacillus sp. LB5a.
Biosurfactant production by Bacillus sp. LB5a on manipueira media
seemed to be growth associated because the bulk of product accumula-
tion (>55%) occurs during the first 24 h when cell growth reached the
maximum level.
The CMC-l is a direct indication of surface activity in the broth; the
higher the CMC-1 the greater the surface activity. This is, therefore, a more
important measurement than the actual quantity of surfactant if the sur-
factant characteristics vary from cultural conditions. From CMC-l data it
is clear that medium F showed the highest surface activity of all the tested
media, and although medium F12 also showed good results, the dilution
of manipueira F waste was not necessary.
The results suggest that the presence of solids in manipueira waste is
inversely related to broth surface tension. Thompson et al (13), using a
Bacilllus subtilis strain, also observed that a high-solids potato process efflu-
ent showed lower biosurfactant production than a low-solids effluent. Few
differences in media appear to influence the characteristics of the surface-
active compounds produced, as reflected by the different CMC-l curves.
The composition of culture medium was previously described as one of
critical importance for determining product yield and biosurfactant prop-
erties (10).
The medium F prepared from manipueira offers promise as a substrate
for biosurfactant production by the isolate Bacillus sp. LB5a. The tested
solids contents media demonstrated high surface tension and low CMC-1
values and are not recommended for biosurfactant production.

Conclusion
This initial study indicates that manipueira could be used as an alter-
native substrate for biosurfactant production and the no-solids medium
preparation (F) presented the highest potential as a culture medium. The
utilization of cassava flour wastewater could reduce environmental prob-
lems for processing industries while decreasing the estimated cost of
biosurfactant production.
Applied Biochemistry and Biotechnology Vol. 705-108, 2003
Manipueira in Biosurfactant Production 301

Acknowledgments
We wish to thank Plaza S.A for the cassava flour wastewater dona-
tion and Comissao de Aperfei<;oamento de Pessoal de Ensino Superior
(CAPES) and Funda<;ao de Amparo aPesquisa do Estado de Sao Paulo for
financial support.

References
1. Banat, I. M., Makkar, R. S., and Cameotra, S. S. (2000), Appl. Microbiol. Biotechnol. 53,
495-508.
2. Banat, I. M. (2000), Biofutur 198, 44-47.
3. Desai, J. D. and Banat, I. M. (1997), Microbiol. Mol. Rev. 61,47- 64.
4. Rosenberg, E. and Ron, E. Z. (1999), Appl. Microbiol. Biotechnol. 52, 154-162.
5. Banat, I. M. (1995), Bioresour. Technol. 51, 1-12.
6. Banat, I. M. (1995), Acta Biotechnol. 15,251-267.
7. Cameotra, S. S. and Makkar, R. S. (1998), Appl. Microbiol. Biotechnol. 50,520-529.
B. Makkar, R. S. and Cameotra, S. S. (1999), J. Surf Oet. 2,237-241
9. Mercade, M. E. and Manresa, M. A. (1994), JAOCS 71, 61-64.
10. Sheppard, J. D. and Mulligan, C. (1987), Appl. Microbiol. Biotechnol. 27, 110-116.
11. Mercede, M. E., Manresa, M. A., Robert, M., Epsuny, M. J., Andres, c., and Guinea,
J. (1993), Bioresour. Technol. 43, 1-6.
12. Koch, A. K., Reiser, J., and Kappeli, O. (1988), Biotechnology 6, 1335-1339.
13. Thompson, D. N., Fox, S. L., and Bala, G. A. (2000), Appl. Biochem. Biotechnol. 84186,
917-929.
14. Fox, S. L. and Bala, G. A. (2000), Bioresour. Technol. 75,235-240.
15. Cereda, C. P. (1994), Industrializariio da mandioca no Brasil, Pauliceia, Sao Paulo, Brazil.
16. Nitschke, M., Ferraz, c., and Pastore, G. M. (2001), in Proceedings of XXI Congresso
Brasileiro de Microbiologia, Sociedade Brasileira de Microbiologia (SBM), Foz do
Igua<;:u, Brazil, pp. 311-312.
17. Daniels, L., Hanson, R. S., and Phillips, J. A. (1994), in Methods for General and Molecu-
lar Bacteriology, Gerhardt, P., Murray, R. G. E., Wood, W. A., and Krieg, N.R., eds.,
American Society for Microbiology, Washington, DC, pp. 518-519.
lB. Gerson, D. F. and Zajic, J. E. (1979), Process Biochem. 14,20-29.
19. Makkar, R. S. and Cameotra, S. S. (1997), JAOCS 74, 887-889.
20. Cooper, D. G., Macdonald, C. R., Duff, S. J. B., and Kosaric, N. (1981), Appl. Environ.
Microbiol. 42,408-412.
21. Sheppard, J. D. and Cooper, D. G. (1991), Appl. Microbiol. Biotechnol. 35,72-76.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0303/$20.00

A New Oxygen Sensitivity and Its Potential


Application in Photosynthetic H2 Production

JAMES W. LEE* AND ELIAS GREENBAUM


Chemical Sciences Division, Oak Ridge National Laboratory,
Oak Ridge, TN 37831-6194, E-mail: Leejw@ORNL.gov

Abstract
We have discovered a new competitive pathway for O 2sensitivity in algal
H2 production that is distinct from the O 2 sensitivity of hydrogenase per se.
This O 2sensitivity is apparently linked to the photosynthetic H2 production
pathway that is coupled to proton translocation across the thylakoid mem-
brane. Addition of the proton uncoupler carbonyl cyanide-p-trifluoro-
methoxy-phenylhydrazone eliminates this mode of O 2 inhibition on H2
photoevolution. This newly discovered inhibition is most likely owing to
background O 2that apparently serves as a terminal electron acceptor in com-
petition with the H2 production pathway for photosynthetically generated
electrons from water splitting. This 02-sensitive H2 production electron trans-
port pathway was inhibited by 3[3,4-dichlorophenyl]1,1-dimethylurea. Our
experiments demonstrated that this new pathway is more sensitive to O 2
than the traditionally known O 2sensitivity of hydrogenase. This discovery
provides new insight into the mechanism of O 2inactivation of hydrogenase
and may contribute to the development of a more-efficient and robust system
for photosynthetic H2 production.
Index Entries: Oxygen sensitivity; H2 production; photosynthetic H2 pro-
duction; H2 production pathways; hydrogenase.

Introduction
Algal photosynthetic hydrogen (H2) production by light-activated
water splitting is a potentially clean energy resource. However, compared
to our knowledge of the pathway of atmospheric CO2 reduction, and in
spite of the potential importance of the hydrogen-producing reaction, rela-
tively little is known concerning the mechanistic pathway of electron flow
in hydrogen-producing algae. In green algae, such as Chlamydomonas
reinhardtii, photoevolution of H2 and O2occurs in the same cell, where the
photosynthetically produced O 2can inhibit the production of H2 (1). There-

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 303 Vol. 105-108, 2003
304 Lee and Greenbaum

fore, the application of green algae for H2 production must address the
problem of O 2sensitivity. Historically, this 02-sensitive phenomenon was
generally interpreted as the direct O 2inhibition of hydrogenase activity (2).
We report here that the classic interpretation of O 2 sensitivity should be
revised. In recent experiments characterizing O 2tolerance in H 2-producing
wild-type C. reinhardtii, we observed a new O 2 sensitivity that is clearly
distinct from that of classic O2inhibition of hydrogenase. The O 2sensitivity
indicates that there is a competitive electron transport pathway that can
redirect electrons from the hydrogenase-catalyzed H2 production pathway
to O 2. That is to say, suppression of H2 evolution in the presence oflow-Ievel
background concentrations of O 2is owing to the drain of reducing equiva-
lents away from the hydrogenase pathway and toward the reduction of O 2.
Our experiments demonstrated that the competitive pathway mechanism
is more sensitive to O 2than the classic O2sensitivity of hydrogenase and can
be suppressed by proton uncoupler carbonyl cyanide-p-trifluoro-methoxy-
phenylhydrazone (FCCP).

Materials and Methods


First evidence of the new O 2sensitivity was obtained from H2 produc-
tion assays in C. reinhardtii wild-type strain 137c. The assays were con-
ducted using a laboratory-built dual-reactor flow detection system (3). For
each assay, 35 mL of algal sample (3 !-tg ChI/mL) was placed and sealed in
each of the two water-jacketed reactors and held at 20°C with a tempera-
ture-controlled water bath. The algal sample was purged with a helium
flow (50 mL/min) through the liquid reaction medium. The helium flow
served two purposes: (1) to remove O2 from the algal sample to establish
and maintain anaerobic conditions that are necessary for induction of the
algal hydrogenase activity and production of H 2, and (2) to carry the
photoproduced H2 gas product to the hydrogen sensors. After induction
of hydrogenase and establishment of steady-state photo evolution of H2
under the helium atmosphere (which requires about 8 h), a primary stan-
dard of 1000 ppm of O 2in helium replaced the pure helium at the same flow
rate (50 mL/min) to characterize the oxygen sensitivity of photo evolution
ofH2. Actinic illumination of 100 !-tEl (m2·s) for the H 2photoevolution assay
was provided by an electronically controlled light emitting diode (LED)
light source at a wavelength of 670 nm. As illustrated in Fig. I, introduction
of 0.1000% (1000 ppm) of O 2 dramatically reduced the rate of algal H2
photoevolution. The steady-state H2 production rate in the presence of
0.1000% O 2was 0.33 !-tmol of H 2·mg of Chl-1·h-1, which is only about 2.8%
of the full steady-state rate (12 !-tmol H 2·mg of Chl-1·h-1) before the introduc-
tion of the 0.1000% O 2. In the past, this type of H2 production decay was
commonly interpreted as the inhibition of hydrogenase activity by O 2. Our
results prove that this classic interpretation of oxygen sensitivity on algal
H2 production is not consistent with the data. According to the classic inter-
pretation, the decrease in H2 production after the introduction of 0.1000%

Applied Biochemistry and Biotechnology Vol. 105-708,2003


New O2 Sensitivity in Algal H2 Production 305

I I I I

20 - ••••.... 137c H2 Production Rate -


- % LED Intensity

15 -J 01000%0, -

.....,
l- 1.
s:::. I,

:c
'I
I.
I,

() ,,, I'
- ,,
I ::'\ -
~ 10 'I
I;
"N I'
Ii
:r::
~ ..
II i1
:i
i

_
::t 5 - \ I -
i I
i\ \

--
\
\.""........- ....._........ ...... ...........
o-
,

I I I I
95 100 105 110 115 120
Time (hr)

Fig. 1. Observation of a new O2 sensitivity to algal H2 production in C. reinhardtii.

O 2 is owing to O 2inhibition of hydrogenase per se: that is, loss of hydroge-


nase activity is the limiting factor for the rate of H2 photoevolution. If this
interpretation were correct, after a brief dark period in the presence of
0.1000% O 2, one would expect the rate of H2 photo evolution to be no higher
than the inhibited rate (0.33 !lmol of H 2·mg of Chl-1·h-1) preceding the dark
interval. However, the data are quite different from the classic expectation.
As shown in Fig. I, there was a surge of H2 photo evolution after a 2-h dark
period in the continuous presence of 0.1000% 02' The peak rate of H2
photoevolution was about 15 !lmol of H 2·mg of Chl-1·h-1, which is about
45 times higher than the classically predicted rate (0.33 !lmol of H 2·mg of
Chl-1·h-1) and well outside the experimental error of the measurement. This
assay has now been repeated six times, and all results were consistent with
the observation presented in Fig. 1.
This observation clearly indicated that hydrogenase activity was not
the limiting factor for H2 photo evolution at this level of O 2, There must be
an alternative electron transport pathway that takes the photo gene rated
electrons away from ferredoxin (Fd) to O 2, The observed reduction of H2
production after the introduction of 0.1000% O 2can be explained by such
a pathway that competes for electrons with the Fd/hydrogenase-catalyzed
H2 production pathway. This is a significant discovery since it fundamen-
tally redefines the meaning of "oxygen tolerance" in algal H2 production.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


306 Lee and Greenbaum
~~---------------------------------r~U
- - LED Intensity
18 18
- - H2 Production
16 16
14 14

B 12 12 ~
r10 10 I
~8 8 .5
C

I 6

4
6

4
~

2 2
0 0

20 22 24 26 28 30 32 34
11me (h)

Fig. 2. Stimulation of photosynthetic H2 production of C. reinhardtii 137c following


addition of proton uncoupler FCCP in background atmosphere of 1000 ppm of O2,
Addition of 5 fAM FCCP produced a dramatic increase in H2 production followed by
a slow decay. The slow decay was owing to a side effect of FCCP known as ADRY, in
which FCCP gradually inhibits PSII activity. This experimental result indicates that
use of a polypeptide proton channel that does not have the ADRY effect could enhance
H2 production by eliminating the problems of both the proton gradient accumulation
and the newly discovered alternative O2 sensivity.

Studies with the chemical inhibitor 3[3,4-dichlorophenyl]l,l-dimeth-


ylurea (DCMU) and the proton uncoupler FCCPyielded additional evidence
for the new O 2 sensitivity. FCCP is a proton uncoupler that can dissipate the
proton gradient across the thylakoid membrane in algal cells. As illustrated
in Fig. 2, in the presence of 1000 ppm of O 2 after the induction of the hydro-
genase enzyme, the steady-state photoevolution of H2 around the time of
20 h was slightly less than 1 Ilffiol of H 2·mg of Chl-1·h-1 (d. Fig. 1 at approx
110 h). After a brief dark period (from 20:20 to 22:20), a burst of H2
photoevolution appeared, followed by an oscillation in the decay curve. Since
both the actinic intensity and the background O 2 concentration (1000 ppm)
remained the same, this H2 production oscillation also indicated that the
decay in the rate of H2 photo-evolution resulted not from O2inhibition of the
hydrogenase enzyme per se, but from a competitive kinetic effect of O 2 on
electron transport that is related to the H2 production process. The addition
of 5 !lM FCCP produced a dramatic removal of O 2 inhibition on H2
photoevolution. The rate of H2 production rose to about 16 !lmol of H 2·mg of
Chl-1·h-1. This FCCP-stimulated H2 production is clearly photo-dependent.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


New O2 Sensitivity in Algal H2 Production 307
As soon as the actinic light was turned off, H2 production stopped. The data
(Fig. 2) also demonstrated that FCCP-enhanced photoevolution ofH2can last
for more than 4 h, although at a decreasing rate. The decrease in rate of H2
photoproduction is owing to a secondary effect of FCCP known in the pho-
tosynthesis research literature as the acceleration of the deactivation reac-
tions of the water-splitting system Y (ADRY) effect, in which FCCP gradually
inhibits photosystem II (P5II) activity by deactivating the photosynthetic
water-splitting complex in the 52 and 53 states (4). However, FCCP does not
have any known effect on hydrogenase per se. Therefore, the observed stimu-
lation ofH2photoevolution by FCCP in the presence of 1000 ppm of O 2clearly
demonstrated that the newly discovered 02-sensitive electron transport path-
way requires the presence of a proton gradient (or ATP) to operate.
DCMU is a chemical inhibitor that binds atthe QB site ofP5II and blocks
transport of electrons acquired from P5II water splitting to photosystem 1.
The experimental data (not shown) demonstrated that the addition of
DCMU inhibited the burst of H2 photoevolution after a dark period in the
presence of 1000 ppm of background O 2. This result indicated that> 90% of
the electrons that are used in the photoproduction of H2 are derived from
P5II water splitting. Therefore, water is the main source of electrons for the
H2 burst after the dark period in the presence of 1000 ppm of O 2. Organic
reserves such as starch are thus not the main source of electrons in this mode
of H2 production.
The new O2 sensitivity was further characterized using a series of O 2
concentrations: 10,100,300,1000,5000, and 10,000 ppm of O2in helium. The
experimental results showed that the introduction of 100 ppm of O 2had no
significant effect on the steady-state rate of H2 photo evolution in
C. reinhardtii wild-type strain 137c. However, the addition of 300 ppm of O 2
began to show an effect. Figure 3 plots the percentage of steady-state H2
production rate vs background O 2concentrations in the wild-type 137c. The
O 2concentration that gave 50% inhibition of H2 photoevolution was about
500 ppm. When the O 2concentration was raised to 5000 ppm, the inhibition
on H 2production was dramatic and the rate of H2 photo-evolution decreased
to nearly zero (Figs. 3 and 4). However, the hydrogenase in the algal cells
was still active even after the continued presence of 5000 ppm of O2for more
than 10 h. When the actinic was turned on again after hour 198, a small peak
of H2 photoproduction was observed. As illustrated in the expanded scale
(Fig. 4B), this H2 photoproduction peak was clearly above the background
noise and/ or dark H2 signal, indicating the presence of active hydrogenase
in the algal cells. Therefore, hydrogenase in the wild-type cells can tolerate
up to 5000 ppm of O 2. The newly discovered O2sensitivity is about 10 times
more sensitive to O 2than that of the hydrogenase.

Results and Discussion


The new O 2 sensitivity is apparently linked to the photosynthetic H2
production pathway that is coupled with proton translocation across the

Applied Biochemistry and Biotechnology Vol. 105-108,2003


308 Lee and Greenbaum

100

!
!c 80

~ 60
J
~
.c 40
0..
'I,N
....0 20
'#.

10 100 1000 10000


[OJ ppm

Fig. 3. Effect of background O2 concentrations on steady-state H2 photoevolution.

thylakoid membrane. The addition of the proton uncoupler FCCP can elimi-
nate this mode of O2inhibition on H2 photoevolution. This O2inhibition on
H2 production is most likely owing to the competitive uptake of reducing
equivalents by background O2, with the H2 production pathway for photo-
synthetically generated electrons from water splitting. The 02-sensitive H2
production pathway can be inhibited by DCMU. Our experiments demon-
strated that the competitive pathway is more sensitive to O 2than the classic
O2sensitivity of hydrogenase. These findings redefine the meaning of" oxy-
gen tolerance" in algal H2 production. As discussed, this O2 sensitivity
apparently represents a new pathway in the photosynthetic H2 production
that is coupled with proton translocation across the thylakoid membrane.
As illustrated in Fig. SA, the site for the reduction of O2 could be at the
RuBisco enzyme, which can serve as an RuDP (also known as RuBP) car-
boxylase and/ or an RuDP oxygenase in the Calvin cycle. Under conditions
for H2 photoevolution in which CO2 is not present and ATP is abundant
owing to associated photophosphorylation, the Calvin cycle enzymes are
fully activated and RuBisco could act as a strong oxygenase. This hypoth-
esis can explain how FCCP mitigates O2 inhibition of H2 photoevolution,
since operation of the Calvin cycle requires formation of ATP using the
proton gradient across the thylakoid membrane. Another possible site for
O2interaction could be at Fd, which, according to the classic Mehler reac-
tion, can serve as an electron donor to O2, Additional experimental studies
with different chemical inhibitors and genetic mutants are under way to
elucidate this 02-sensitive H2 production electron transport pathway.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


New O2 Sensitivity in Algal H2 Production 309

A 111
• , I
18
, l5OO0 ppm 02 - - H2 Production
J; 16 "" 18
:c -%LED
oat 14 ~- 14
E
~ 12 · ~ 12
i- 10 b
J!
· i- 10'i
c
I. 8 ·
0
c - .s!8
0
ts 6 · t- 6~
:::J
'#.
l
r-
4 4
N
J: 2· t- 2


\. 0
• , I
188 188 190 192 194 196 198 200 202
Time (h)

B
.....
1.0
It I I - - H2 Production
1.0

-'=: -0/0 LED


:co 0.8 t- 0.8

fi 0.8 · t- O.Bb
........
!
5000 ppm °2 Wi
c
!c
~c
o
0.4 -0.4 w0

~"
:g

\
..J

-02 '#.
:::J

e 0.2
'tJ

Q.
N
J:
0.0 · '-- 0.0

186 188

190 192 194 196
I
198 200
I I
202
Time (h)
Fig. 4 (A) The introduction of 5000 ppm of O2 dramatically inhibits H2 production,
but the hydrogenase remains active. (B) Expanded vertical scale of (A), showing clear
peak of H2 photoevolution after 10 h of continued presence of 5000 ppm of 02'

Applied Biochemistry and Biotechnology Vol. 105-108,2003


~ A H2t
~
Q..
t\ DARK
OJ ,
o· 1000 ppm ,,~-~ .....
n , '
::J-

3;:;;. H,t /
°2,
I
" :t ~...cot:..."'" 'C'Oln . etc
2H" I
~ ..n(>,'l ___ RuBiSQO
OJ "'~~....,,. I
:J ~';;
Q.. .., ~I c~=::;r .
OJ .... .21 RuDP
o· CAlVIN
(ii ~I CYCLE
g. &1
:J
o jl
a% .cl
'<:
~I
1
w
,
o
LHC

~
:-

Fig. 5. (A) The newly discovered O 2 sensitivity is likely owing to the background O 2 (at about WOO-ppm levels) acting as a terminal sink, in
~ competition with the Fd/hydrogenase H2 production pathway, for photosynthetically generated electrons. Illustrated by the blue-dotted
a
,Co arrows is a speculated mechanism of how the background O 2 might interact with the photosynthetic H2 production pathway at the point(s) of
a RuBisco and/or Fd.
'"
::;; (Continued)
:g B
~ 02 ~H2
' DARK
Q..
\:Xl
o· H3C H (CHzO)" . etc.
9- Hzf • ell:. •,
3tn· ~
, /
~ Co, ~ ~ _,-r >~
:;,
'" CALVIN "
Q..
\:Xl y~_ ~' : CYClE,'
o· '\ I
~ DNA NAOP'
I HAD ,'. ' ~ ... --
:;,- PH +H" ...
:;,
o Hydroget*e Prornoler
'._n~
_ r I
,- '
,,
~ If

w LHC

H+

H2 Production Under Anaerobic Conditions

Fig. 5. (Continued) (B) Development of efficient algal H2 production system by construction and transformation of vector that contains
hydrogenase promoter and a piece of synthetic DNA for the polypeptide proton channel. The transformed alga could grow normally using
~ ambient air CO2 under aerobic conditions without the polypeptide proton channel, which could be expressed only with the induction of the
~ hydrogenase under anaerobic conditions when its function is needed for enhanced H2 production. This diagram illustrates a predicted mecha-
a nism of how photosynthetic H2 production could be enhanced through the action of the proposed polypeptide proton channel by dissipation
,00
of the static proton gradient across the thylakoid membrane so that the photosynthetic electron transport from PSII water splitting to the Fdj
~

8v" hydrogenase H2 production pathway can become more efficient, and by inactivation of the Calvin cycle so that both the background O 2 and CO2
could be prevented from acting as a competitive terminal electron sink.
312 Lee and Greenbaum

Potential Application for Enhanced


Photosynthetic H2 Production
The discovery of alternative Oz sensitivity also provides a new oppor-
tunity to develop a more-efficient and robust system for photosynthetic Hz
production. The experimental data with FCCP (Fig. 2) indicate that use of
a polypeptide proton channel that does not deactivate the oxidizing
equivalents of PSII could enhance Hz production by eliminating the prob-
lems of both the proton-gradient accumulation (5) and the newly discov-
ered alternative Oz sensitivity. Therefore, we propose to create a "designer"
photosynthetic organism for production of Hz by genetic insertion of pro-
grammable proton channels in thylakoid membranes. The genetic inser-
tion of programmable thylakoid-membrane proton channels can be
achieved by transformation of a host alga with a genetic vector that con-
tains a hydrogenase promoter-linked CF 1 suppressor or membrane
polypeptide proton-channel gene. The envisioned "super" alga that can
be created in this way should be able to perform autotrophic photosynthe-
sis using ambient air COz as the carbon source and grow normally under
aerobic conditions such as in an open pond. When the algal culture is
grown and ready for Hz production, the CF1 suppressor or proton-channel
gene will then be expressed simultaneously with the induction of the
hydrogenase enzyme under anaerobic conditions. The expression of the
proton-channel gene should produce polypeptide proton channels in
the thylakoid membrane, thus dissipating the proton gradient across the
thylakoid membrane without ATP formation (Fig. SB). The expression
of the CF1 suppressor should create CFo, which may act as a free proton
channel without the eF1 cap, thus similarly dissipating the proton gradi-
ent across the thylakoid membrane without ATP formation. The free pro-
ton-conductive CFo or polypeptide proton channels in the thylakoid
membrane could provide two advantages for Hz photoevolution: (1) accu-
mulation of a proton gradient that impedes photosynthetic electron trans-
port from water to Fd/hydrogenase could be prevented, and (2) the newly
discovered Oz-sensitive pathway that competes with the Hz production
pathway for photosynthetically generated electrons could be eliminated.
Therefore, the coexpression of the polypeptide proton channel (or CF1
suppressor) and hydrogenase genes will make this alga a more-efficient
system for production of Hz by photosynthetic water splitting under anaero-
bic conditions (Fig. SB). This organism contains normal mitochondria,
which can use reducing power (NADH) from organic reserves (and/or
exogenous acetate) to power the cell immediately after returning to aerobic
conditions. Therefore, when the algal cell is returned to aerobic conditions
after its use under anaerobic conditions for photoevolution of Hz and Oz,
the cell will stop generating free CFo (or polypeptide proton channels) in
thylakoid membranes and restore its normal photoautotrophic capability
by synthesizing functional thylakoids. Consequently, it should be possible
to use this type of genetically transformed organism for repeated cycles of

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


New O2 Sensitivity in Algal H2 Production 313
photoautotrophic culture growth under normal aerobic conditions and
efficient production of H2 and O2by photosynthetic water splitting under
anaerobic conditions.

Acknowledgments
We thank C. A. Sanders, B. Forbes, and B. Kusiak for culture media
preparation; B. Mathis and A. Jones for secretarial support; M. K. Savage for
editorial assistance; and C. D. King and V. W. Purdue for technical illustra-
tions. We also thank Drs. M. Seibert and M. L. Ghirardi for their 02-tolerant
mutants and informative discussions. This research was supported by the
US Department of Energy Hydrogen Program, the DOE Office of Science
Young Scientist Award (to J. W. Lee), and the Office of Basic Energy Sci-
ences. Oak Ridge National Laboratory is managed by UT-Battelle, LLC, for
the US Department of Energy under contract DE-ACOS-000R2272S.

References
1. Greenbaum, E. and Lee, J. W. (1998) in BioHydrogen, Zaborsky, O. R., ed., Plenum
Press, New York, NY, pp. 235-24l.
2. Ghirardi, M. L., Togasaki, R. K., and Seibert, M. (1997) Appl. Biochem. Biotechnol., 63-65,
141-15l.
3. Lee, J. W., Blankinship, S. L., and Greenbaum, E. (1995), Appl. Biochem. Biotechnol.,
51/52,379-385.
4. Samuilov, V. D., Barsky, E. L., and Kitashov, A. V. (1995) in Photosynthesis: from Light
to Biosphere, vol. II, Mathis, P., ed., Kluwer Academic Publishers, The Netherlands,
pp.267-270.
5. Lee, J. W. and Greenbaum, E. (1997), Fuels and Chemicals from Biomass, ACS Sympo-
sium Series 666, Saha, B. C. and Woodward, J., eds., American Chemical Society,
Washington, DC, pp. 209-222.

Applied Biochemistry and Biotechnology Vol. 105-708,2003


SESSION 3
Bioprocessing Research
Session 3

Bioprocessing Research

MARK A. EITEMAN1 AND AMIT VASAVADA 2


1 Universityof Georgia, Athens, GA; and
20iversa Corporation, San Diego, CA

The production of fuels and chemicals from diverse biological


resources such as crops and waste by-products requires the integration of
numerous processing steps. Not only can an incremental improvement
in one step improve an overall process, but new developments and
approaches in processing technology can also dramatically alter the pro-
cessing sequence and result in a step-change in process performance.
Bioprocessing ultimately involves conversion, and a wide range of opera-
tions such as chemical reaction, hydrolysis, fractionation, mixing, separa-
tion, and fermentation are required to process a raw and complex material
into a desired, refined fuel or chemical product. This session thus natu-
rally represented a wide range of research areas and included over 70
contributions.
The dominant theme of the presentations was hydrolysis, and of par-
ticular interest was dilute acid hydrolysis and its optimization. Of the
numerous substrate materials studied, which included rice waste, bagasse,
paper sludge, wheat straw, and solid waste, there was a particular interest
in corn stover, and contributions described rheological properties, hydroly-
sis, and simultaneous saccharification and fermentation of this materiaL
Numerous products were also considered including ethanol, lactic acid,
butanol, fumaric acid, arachidonic acid, and erythritoL
Another theme of this year's session was reactor configuration. Sev-
eral contributions highlighted the benefit of multistage reaction configura-
tions, in which independent stages improved the overall operation. For
example, two-stage fermentation might include two separate reactors with
different microorganisms and conditions, each completing one portion in
the complete conversion of the raw material into the product.
Tools continue to be developed to quantify and understand biopro-
cesses, including neural network models of complex bioprocesses as well
as molecular probes to understand responses at the level of gene expres-
sion. Several contributions dealt with environmental concerns, such as odor
abatement and solubilization of inorganic phosphates.
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 317 Vol. 105-108, 2003


318 Introduction to Section 3

All aspects of biotechnology for the production of fuels and chemi-


cals rely on bioprocessing. Based on the new developments presented in
this session and the volume of contributions and interest in this area, we
are optimistic for continued progress and improvements in bioprocessing.

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0319/$20.00

Optimization of S02-Catalyzed Steam


Pretreatment of Corn Fiber
for Ethanol Production

RENATA BURA,l RODNEY J. BOTHAST/


SHAWN D. MANSFIELD,l AND JOHN N. SADDLER,*,l
1Forest Products Biotechnology, Department of Wood Science,
University of British Columbia, 2424 Main Mall, Vancouver, Be, V6T 1Z4,
Canada, E-mail: saddler@interchg.ubc.ca; and
2Fermentation Biochemistry, National Center for Agricultural Utilization
Research, USDA, ARS, 1815 N. University Street, Peoria, IL 61604

Abstract
A batch reactor was employed to steam explode corn fiber at various
degrees of severity to evaluate the potential of using this feedstock as part of
an enzymatically mediated cellulose-to-ethanol process. Severity was con-
trolled by altering temperature (150-230°C), residence time (1-9 min), and
S02concentration(O-6% [w /w] dry matter). The effects of varying the differ-
ent parameters were assessed by response surface modeling. The results
indicated that maximum sugar yields (hemicellulose-derived water soluble,
and cellulose-derived following enzymatic hydrolysis) were recovered from
corn fiber pretreated at 190°C for 5 minutes after exposure to 3% S02' Sequen-
tial S02-catalyzed steam explosion and enzymatic hydrolysis resulted in a
conversion efficiency of 81% of the combined original hemicellulose and
cellulose in the corn fiber to monomeric sugars. An additional posthydrolysis
step performed on water soluble hemicellulose stream increased the concen-
tration of sugars available for fermentation by 10%, resulting in the high
conversion efficiency of 91 %. Saccharomyces cerevisiae was able to ferment the
resultant corn fiber hydrolysates, perhydrolysate, and liquid fraction from
the posthydrolysis steps to 89,94, and 85% of theoretical ethanol conversion,
respectively. It was apparent that all of the parameters investigated during
the steam explosion pretreatment had a significant effect on sugar recovery,
inhibitory formation, enzymatic conversion efficiency, and fermentation
capacity of the yeast.

Index Entries: Corn fiber; steam pretreatment; enzymatic hydrolysis;


posthydrolysis; fermentation; ethanol.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 319 Vol. 105-108, 2003


320 Bura et al.
Introduction

Corn fiber is a very attractive lignocellulosic substrate for the biocon-


version to ethanol process because it has a high carbohydrate content, it is
present in large quantities in most ethanol- and starch-processing plants,
and it has few competing uses. Corn fiber is a mixture of corn hulls and
residual starch not extracted during the wet-milling process and comprises
up to 11 % of the dry weight of the corn kernel (1). It has been predicted that
the utilization of the corn fiber fraction in the conversion process could
potentially increase the overall ethanol yield by an additional 10% (2).
For efficient production of ethanol from lignocellulosic biomass by
sugar hydrolysis and fermentation, the lignocellulosic substrate should
be pretreated to more effectively recover the hemicellulose and, concur-
rently, to make the cellulose more accessible to enzymatic hydrolysis. To
be fully sufficient, the pretreatment should provide effective recovery of
the majority of the hemicellulose-derived sugars, lignin, and cellulosic
component that can be readily hydrolyzed by enzymes and, later on,
fermented. Several pretreatment options for corn fiber have been evalu-
ated for their efficacy in solubilizing, fractionating, hydrolyzing, and
separating these major compositional constituents. Dilute-sulfuric acid
hydrolysis of corn fiber at 140-160°C and pH 1.5-2.0 for 10-30 min fol-
lowed by partial neutralization and enzymatic saccharification of polysac-
charides gave a yield of 81-87% of total carbohydrates as free sugars (2).
Although this process showed that enzymatic hydrolysis achieved a high
conversion of all polysaccharides in the corn fiber, it was also apparent
that significant inhibitor formation occurred at all pretreatment condi-
tions tested between 140 and 160°C (2). Other pretreatments used with
corn fiber (ammonia fiber explosion [AFEX] treatment and hot water),
although effectively preparing cellulose for enzymatic hydrolysis, did
not completely hydrolyze the xylan (3,4). Although many corn fiber pre-
treatment methods have been proposed, none have been proven to be
fully effective. Previously, we analyzed S02-catalyzed steam explosion
of corn fiber in the temperature range of 170-200°C, time of 1.5-5 min,
and S02 concentration of 0-6% (5). Although, the results indicated that
maximum sugar yield (soluble and following enzymatic hydrolysis) was
recovered from corn fiber pretreated at 190°C for 5 min with 6% S02' the
effects of varying the different parameters were not assessed by response
surface modeling. We also did not analyze oligomeric sugars, or the
potential fermentability of water soluble fractions obtained from each of
the pretreatment, posthydrolysis, and enzymatic hydrolysis.
The purpose of the present study was to investigate the influence of
steam explosion pretreatment parameters (time, temperature, and pH) on
the subsequent enzymatic saccharification and fermentation of corn fiber
and, as a result, to establish optimum pretreatment conditions for the feed-
stock by using standard response surface technique. This statistical model
involved fitting an empirical model to the experimental data and identify-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 321

1 Com fibre 1

1
Pretreatment
Temperature (150-230°C), Time
(1-9 min), SO:z (0-6%)

1
...--__--11 Filtration ll--__--.
I
Prehydrolyste
1 1 Soli1ds

I Analysis of sugars, HMF


and furfurals J
Fermentation Sugar analysis
S.cerevisiae (6WL), time 6 r--
hours, pH adjusted by
O.5M NaOH to 6 Hydrolysis
2% wlw, NaAc buffer, pH=4.8,
CBU: FPU 4:1, time 72 hours,
Analysis of sugar,
ethanol, HMF and
fufurals Sugar analysis

Post-hydrolysis Fermentation
3% H2SO4, 1 hours, 120°C S.cerevisiae (6WL), time 4 hours,
pH adjusted by O.5M NaOH to 6
Sugar analysis

Fermentation
S.cerevisiae (6WL), time 24
hours, pH a4justed by O.5M
1 Analysis of sugar, ethanol, HMF and fufurals
NaOHto6

l Analysis of sugar, ethanol, HMF and furfurals .1


Fig. 1. Experimental procedure to S02-catalyzed steam explosion of corn fiber and
consequent posthydrolysis, fermentation, and enzymatic hydrolysis.

ing the optimal residence time, temperature, and S02-impregnation level


in the pretreatment stage.

Materials and Methods


Substrate Pretreatment
The experimental procedure used in this study is shown schematically
in Fig. 1. Com fiber (60.0% moisture content) was obtained from the National
Center for Agricultural Utilization Research (Peoria, IL) and stored at -20°C
until use. Samples of 300 g (dry wt) were impregnated overnight with anhy-
drous S02 in plastic bags. The uptake of S021 expressed as a percentage of the

Applied Biochemistry and Biotechnology Vol. 105-108,2003


322 Bura et at.
Table 1
Conditions for S02-Catalyzed Steam Pretreatment
of Com Fiber and Shot Yields Expressed as Percentage
of Original Oven Dried Com Fiber After Steam Explosion
Severity of Temperature Time S02 Shot yield
Experiment pretreatment (log Ro) (0C) (min) (%) (%)

1 2.06 170 1 3 97
2 2.17 150 5 3 85
3 2.76 170 5 0 89
4 2.76 170 5 6 92
5 3.02 170 9 3 89
6 3.35 190 5 3 87"
7 3.35 190 5 3 91-
8 3.35 190 5 3 86"
9 3.58 210 2.2 1 89
10 3.58 210 2.2 5 86
11 4.13 210 7.8 1 79
12 4.13 210 7.8 5 81
13 4.53 230 5 3 74
14 4.35 190 5 0 91
15 3.35 190 5 6 86
•Average shot yield: 88 ± 2.6 %.

oven-dried com fiber, was measured by weighing the com fiber before and
after S02 addition. The samples were then loaded, in 50-g batches, into a
preheated 2-L Stake Tech II batch reactor (Stake Tech-Norvall, Ontario,
Canada) and exploded at different severities (temperature ranging from 150-
230°C; time from 1.5 to 5 min; 502 concentration from 0 to 6% weight/oven-
dried weight of fiber) (Table 1). The severity of the steam explosion
pretreatment was represented by the severity factor as defined by Overend
and Chornet (6). This severity factor (Ro) combines the effects of time and
temperature, as follows:
(1)
in which t is the residence time (min), Tr is the reaction temperature (0C),
and Tb is a reference temperature (100°C).
Following steam explosion, the concentration of sugars in the water
soluble fraction was quantified by high-performance liquid chromatogra-
phy (HPLC), while the water-insoluble fraction (water-washed) was col-
lected and adjusted to2% (w /w) dry matter content and subsequently used
for enzymatic hydrolysis. Shot yields (%) were determined by dividing the
dry weight of the steam-exploded sample by the dry weight of the
nonpretreated sample (300 g) (Table 1).

Applied Biochemistry and Biotechnology Vol. 705-108,2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 323

Characterization of Substrate
The chemical composition of the original starting material and the
steam-exploded solids was determined using a modified Klason lignin
method derived from the TAPPI Standard method T222 om-98 (7). Briefly,
0.2 g of sample was incubated at 20°C with 3 mL of 72% H 2SO4 for 2 h with
mixing every 10 min. The reaction was then diluted with 112 mL of deionized
water (final acid concentration of 4% H 2S04) and transferred to a serum
bottle. The solution was subjected to autoclaving at 121°C for 1 h and filtered
through a medium coarseness sintered-glass filter for gravimetric determi-
nation of acid-insoluble lignin. Each experiment was run in triplicate. The
concentration of sugars in both the filtrate and the stream pretreatment
hydolyzate, as well as inhibitors such as 5-hydroxymethylfurfurals (HMFs),
were determined by HPLC analysis. The HPLC system (Dionex DX-500;
Dionex, Sunnyvale, CA) was equipped with an ion-exchange PAl (Dionex)
column, a pulsed amperometric detector with a gold electrode, and a Spectra
AS 3500 autoinjector (Spectra-Physics, Oroville, CA). Prior to injection,
samples were filtered through 0.45-mm HV filters (Millipore, Bedford, MA)
and a volume of 20 ilL was loaded. The column was equilibrated with 250
mM NaOH and eluted with deionized water at a flow rate of 1.0 mL/min.
Ethanol and furfural concentration were determined using a Hewlett-
Packard 5890 gas chromatograph equipped with a 6890 autoinjector, and a
flame ionization detector. Components were separated using a 30-m
Stabilwax-DA column supplemented with a 5-m deactivated guard col-
umn (Restek).

Posthydrolysis
Duplicate samples containing 27 mL of the water soluble fraction were
posthydrolyzed after adding concentrated H 2S04 to achieve a final concen-
tration of 3% acid. Posthydrolysis was performed by heating the solution
at 121°C for 1 h in an autoclave. A batch of sugar standards was also auto-
claved under the same conditions, to estimate any hydrolysis loss. Sugar
concentrations were detected by HPLC as described previously.

Enzymes
A complete Trichoderma reesei cellulase (Celluclast®, Novo Nordisk,
Franklinton, NC) was used in combination with a commercial ~-glucosi­
d~se (Novozym 188®, Novo Nordisk) for cellulose degradation, while
gluco-amylase (200 V /mL) and a-amylase (300 V /mL) (Sigma, St. Louis,
MO) were used to ensure complete hydrolysis of the starch. The Celluclast
preparation contained 49 mg of protein/mL as measured by the Bio-Rad
protein assay (Bio-Rad, Hercules, CA) and had the following hydrolytic
activities: 80 filter paper units (FPV)/mL (filter paper activity), 52 IV /mL
of carboxymethylcellulase (CMCase),226 IV /mL of xylanase, and 50 IV /
mL of ~-glucosidase. The protein content and activities of Novozym 188
were as follows: 44 mg/ mL, 792 cellobiose units (CBV) / mL, 5 FPV / mL,

Applied Biochemistry and Biotechnology Vol. 705-708,2003


324 Bura et al.
34 IU I mL of CMCase, 94 IU I mL of xylanase, and 500 IU I mL of ~-glucosi­
dase. The enzyme activities were measured as described by Ghose (8).

Enzymatic Hydrolysis
Water-washed pretreated solids obtained during steam explosion
were hydrolyzed at a 2% (w Iv) solid concentration in 50 mM sodium
acetate buffer (pH 4.8) at 45°C with continuous agitation (200 rpm) for
72 h. Each flask was inoculated with enzyme based on the amount of FPU
of cellulase and CBU (Novozyme 188) added at a CBU:FPU ratio of 4:1,
with excess glucoamylase and a-amylase as described by Grohmann and
Bothast (2). Aliquots of 200 ilL were aseptically removed at different reac-
tion intervals and boiled for 5 min to inactivate the enzymes. The sugar
concentration was then determined by HPLC. Each experiment was run in
duplicate. The fermentability of sugars obtained after enzymatic hydroly-
sis (liquid fraction) was tested using Saccharomyces cerevisiae.

Fermentation Microorganisms
A spent sulfite liquor-adapted strain of S. cerevisiae was generously
provided by Tembec (Quebec, Canada), and used for all fermentations. The
yeast was maintained at 4°C on YPD medium (2% glucose, 1% yeast extract,
2% peptone, and 1.8% agar). For each fermentation, S. cerevisiae was
pregrown in 500 mL of YP medium (1% glucose, 1% yeast extract, and 1%
peptone) at 30°C for 3 d, and then harvested after 24 h, and resuspended in
fresh YP medium. After extensive washing, the inoculum cell concentra-
tion was adjusted with sterile deionized water to provide the final cell
concentration of 6 giL (oven-dried cell weight).
Fermentation of Corn Fiber Hydrolysate
Fermentation of liquid sugar fractions (prehydrolysate, post-hydro-
lyzed prehydrolysate, and liquid fraction obtained during enzymatic
hydrolysis) was conducted in 50-mL beakers by preadjusting liquid broth
to pH 6.0 with 0.5 M sodium hydroxide and supplementation with 0.125
mL of 2 M (NH4hP04• Each 50-mL beaker contained 20 mL of the water
soluble sugar source and 1 mL of S. cerevisiae inoculum (6 giL of oven-dried
cell weight). The fermentation vessels were maintained at 30°C with
continuous agitation (200 rpm). Sugars, ethanol, HMF, and furfurals were
determined periodically from the aliquot culture samples during the course
of the fermentation. The relative ethanol yield, YEtOw'Y:tbHl was defined as
the ratio of the ethanol yield of the filtrate and the reference fermentation.
Each experiment was run in duplicate.
Response Surface Methodology
A response surface methodology was used to study the effects of tem-
perature, time, and S02 concentration on hemicellulose recovery, enzy-
matic digestibility, and fermentability of corn fiber after steam explosion

Applied Biochemistry and Biotechnology Vol. 105-108,2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 325
Table 2
Fitted Coefficients in Eq. 1 for Empirical Models and Corresponding R2 Values
Monomer sugar recovery Sugar Prehydrolysate
Experiment in prehydrolysate hydrolysability fermentability
Intercept 26.47 80.47 94.33
C1 2.88 1.59 0
C2 5.06 1.65 -2.38
C3 2.51 2.91 0
C4 -1.55 -3.61 0
Cs 0 0 0.13
C6 0 0 0
C7 0 0 -2.56
Cs -3.03 -{j.79 -5.83
C9 0 -1.41 -2.94
R2 0.99 0.99 0.97

(9). A full three-level factorial design consisted of 13 sets of experimental


conditions, including the center-point experiment, which was chosen based
on the results of our previous finding (5), and was performed in triplicate
(Table 1). In addition, to test the influence of S02 impregnation during
pretreatment, two additional steam explosion conditions were included.
Statistical analysis was performed by fitting Eq. 2 to experimental response
data (Y) by multiple linear regression, using Microsoft Excel:
Y = Intercept + C1 x T + C2 X q + C3 x 5 + C4 X Tq +C5 x
T5 + C6 x q5 + C7 x T2 + Cs x q2 + C9 X 52 (2)
in which T-time (min), q is temperature (OC), and 5 is 502 (% [w fwD.
Terms that were statistically insignificant (at the 95% confidence level)
were eliminated from Eq. 2 to give an empirical model for each dependent
variable in terms of time, temperature, and 502concentration. The coef-
ficient of determination and R2 are reported in Table 2 to indicate the
adequacy of fit for each empirical model. Means and variances were cal-
culated for three repeats of the center-point conditions.

Results and Discussion


Shot Yield
As mentioned in the Introduction, it is important that pretreatment
conditions that ensure good hydrolysis of the water-insoluble cellulose
stream but that are not so severe that the overall yield of recovered carbo-
hydrate is unacceptably low are defined. When we first assessed a range of
conditions for pretreating corn fiber it was apparent that the severity had

Applied Biochemistry and Biotechnology Vol. 105-108,2003


326 Bura et al.
a substantial effect on total recovery of solids after steam explosion with the
shot yield ranging from 74 to 97% (Table 1). Clearly, increased pretreatment
severity reduced the yield recovery of the original com fiber. The best
yields were obtained when com fiber was treated at lower temperatures
(ISO-170°C), while lowest shot yield was obtained at 235°C. At the same
temperature (170 and 210°C), longer reaction times led to lower shot yield.
It was apparent that the more severe steam explosion conditions signifi-
cantly decreased the shot yield for the recovery of solid material. It was
probable that the reduction in the amount of fibrous material with increas-
ing severity could be attributed to the solubilization of the hemicellulose,
since similar results have been reported during S02 -catalyzed steam explo-
sion of Pinus radiata (10), Pseudotsuga menziezii (11), and mixture of Picea
abies and Pinus silvestris (12).
Total Sugar Yield
Monomer Sugar Yield
Once we had established conditions that provided optimum recov-
ery of the hemicellulose and cellulose fractions, we next wished to assess
the nature and concentration of sugars in the water-soluble, hemicellu-
lose-rich fraction. When comparing the chemical composition of the origi-
nal com fiber (Table 3), it was apparent that following steam explosion, a
high percentage of all monomeric hemicellulose-derived sugars was
recovered at temperatures ranging from 170 to 210°C (Table 4). As the
pretreatment severity increased, the concentration of monomers in the
prehydrolysate increased. At each of the pretreatment temperatures used
(170, 190, and 210°C), increased pretreatment times and 502 concentra-
tions yielded higher recovery of monomeric hemicellulose-derived sugars
(Table 4 and Fig. 2A,B). Maximum recoveries of the hemicellulose-derived
monomers (34%) were obtained at temperatures between 190 and 220°C
and pretreatment times of 6-8 min (Fig. 2A). Although temperature and
time appeared to be the main factors influencing recovery optimization,
interaction between temperature and 502 levels also appeared to influence
hemicellulose recovery. Maximum recovery of the hemicellulose-derived
monomers (37%) was obtained at temperatures between 190 and 220°C
and S02 concentrations between 4 and 6% (Fig. 2B). Increasing S02 concen-
tration increased monomer sugar yield but reached a plateau at roughly
3%. The impregnation of additional S02 prior to pretreatment did not
seem to significantly improve monomer sugar recovery, although it did
result in a decrease in the concentration of total sugars. Similar findings
were previously reported by Clark and Mackie (10), in which the effect of
502 loading on total sugar recovery of steam-exploded Pinus radiata was
significant up to approx 3% (w /w) dry matter. These findings imply that
the observed decrease in prehydrolysate oligomer recovery with increased
502 concentration results from the lower pH, which facilitates the com-
plete depolymerization of oligomeric hemicellulose.

Applied Biochemistry and Biotechnology Vol. 705-70B, 2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 327
Table 3
Composition of Original Corn Fiber (% wt)
Total Klason
Ara Gal Glu Xyl Man sugars lignin Ash
12.11 2.70 42.75 18.03 0.58 76.17 8.72 0.65
±0.31 ±0.16 ± 1.08 ±0.56 ±0.03 ±2.08 ±0.21 ±0.02

Table 4
Yield of Hemicellulose-Derived Sugars (Excluding Glucose)
in Prehydrolysate After Steam Explosion (Monomers)
and Steam Explosion and Posthydrolysis (Oligomers and Total Sugars)
Expressed as GI100 g of Hemicellulose in Original Corn Fiber
Treatment Monomers (%) Oligomers (%) Total sugars (%)
170°C, 1 min, 3% 6.1 10.4 16.5
150°C, 5 min, 3% 5.7 7.1 12.8
170°C, 5 min, 0% 10.9 18.6 29.5
170°C, 5 min, 6% 18.5 24.5 43.0
170°C, 9 min, 3% 23.8 31.0 54.8
190°C, 5 min, 3% 26.5% ± 1.0 23.8% ± 1.6 50.3% ± 0.8
210°C, 2.2 min, 1% 21.5 19.2 40.8
210°C, 2.2 min, 5% 27.6 6.3 33.9
210°C, 7.8 min, 1% 22.5 12.5 35.0
210°C, 7.8 min, 5% 24.1 5.9 30.0
230°C, 5 min, 3% 23.0 4.1 27.1
190°C, 5 min, 0% 15.4 30.6 46.0
190°C, 5 min, 6% 27.8 19.2 46.9

Oligomer and Total Sugar Yield


It was apparent that posthydrolysis of the recovered prehydrolysate
using 3% H 2S04 (120°C, 1 h) effectively depolymerized the solubilized oli-
gomeric hemicellulose-derived sugars. The percentage of oligomers
present in the prehydrolysate ranged from 4 to 31% of the total soluble
sugars (Table 4). Time and S02 concentration had different effects on oligo-
mer and total sugar recovery, depending on the pretreatment temperature.
At the lower temperatures (170°C), longer times and higher S02 concentra-
tions led to better oligomer and total sugar recoveries. Conversely, at higher
temperatures (210°C), longer times and higher S02 concentrations led to
decreased oligomer and total sugar recoveries. Similar effects were reported
by San Martin et al. (13) in which the water soluble fraction obtained from
steam-exploded P. radiata at 220°C and 30 s contained 50% oligomers, while
the samples steam exploded at the same temperature for 2 min contained
only 15% oligomers. However, there was no statistically significant corre-
lation among time, temperature, and S02 concentration regarding oligo-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


328 Bura et at.

Sugar 1
n1Covery(%1

Sugar
recovery (%1

Sugar
recovery (%1

-
Temperawre reI
nme(mln)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 329
mer recovery. This can be attributed to the degradation of sugars during
posthydrolysis, as shown previously by 5hevchenko et al. (14).
There was an interaction between temperature and time in terms of
total sugar recoveries, which was illustrated in the response surface con-
tour (Fig. 2C). Maximum recoveries of total hemicellulose sugars (64%)
were obtained at a temperature ranging from 160 to 190°C and times from
6-8 min. Previous research has indicated that hemicellulose-derived sug-
ars are less amenable to inhibitory byproduct generation at temperatures
lower than or near 200°C for softwood and hardwood residues (10,15).
Low concentrations of monomeric hemicellulose-derived sugars and high
concentrations of oligomers have also been observed during the pretreat-
ment of corn fiber by hot water and steam fractionation (4) or by ammonia
fiber explosion system (3,16). It is likely tha t the prod uction of oligosaccha-
rides and consequently the incomplete hydrolysis of xylans is partially
owing to their structure. As observed by Hespell et al. (16), during enzy-
matic hydrolysis of AFEX-treated corn fiber, starch and cellulose compo-
nents were converted solely to glucose. However, oligo saccharides
represented 30-40% of the xylan degradation products. In the case of hot
water and steam fractionation, the majority of the solubilized pentosans
generated after both pretreatments existed as oligomers (4). Corn fiber
xylan is poorly degraded and structural analyzes suggest that> 70% of the
xylose backbone residues have one or more arabinose, 4-0-methylglu-
couronic acid, or other side chains. As a result, there are few unsubstituted
regions in corn fiber xylan (17).
Enzymatic Hydrolysis
Once we had established conditions that would allow maximum
recovery of the hemicellulose sugars (while it was hoped minimizing
sugar decomposition products), we next had to assess whether the result-
ant water-insoluble cellulosic fraction was readily hydrolyzed. Recov-
ered pretreated, water-washed corn fiber solids were subjected to
enzymatic hydrolysis for 72 h with a combination of cellulases and amy-
lases supplemented with an excess of ~-glucosidase. The results indicated
that the hydrolysability of the solids improved as pretreatment condi-
tions became more severe, reaching a maximum at 190°C,5 min, and 3%
502, as was previously shown (5). At higher severities (3.6,4.1, and 4.5) a
significant decrease in total sugar yield was observed (Table 5). This trend
had previously been observed in softwood residues, in which optimal
enzymatic hydrolysis of the cellulosic component occurred at higher pre-
treatment severities (18). The sequential steam explosion pretreatment

Fig. 2. (opposite page) Response surface for monomeric hemicellulose recovery in


prehydrolysate with respect to (A) temperature and time, (B) temperature and 502
concentration, and (C) total hemicellulose recovery in prehydrolysate regarding tem-
perature and time.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


330 Bura et al.
Table 5
Percentage of Monomeric Sugars in Water Soluble Fraction
of Steam Exploded Corn Fiber Following
Different Pretreatment Conditions,
and Hydrolysates Following Enzymatic Hydrolysis
Monomers Monomers after enzyme
Treatment (%) hydrolysis (%)
170°C, 1 min, 3% 2.3 57.4
150°C, 5 min, 3% 2.1 46.8
170°C, 5 min, 0% 4.0 57.5
170°C, 5 min, 6% 7.0 65.5
170°C,9 min, 3% 9.4 71.1
190°C, 5 min, 3% 11.4 ± 0.5 69.1% ± 2.1
210°C, 2.2 min, 1% 9.3 62.9
210°C, 2.2 min, 5% 13.6 61.2
210°C, 7.8 min, 1% 13.3 47.1
210°C, 7.8 min, 5% 22.2 53.0
230°C, 5 min, 3% 16.2 41.6
190°C, 5 min, 0% 5.9 70.7
190°C, 5 min, 6% 12.8 61.4

followed by enzymatic hydrolysis of com fiber in the current study showed


an 81 % conversion of all original polysaccharides to monomeric sugars,
which is comparable with that of the work of Grohmann and Bothast (2),
who successfully converted 85% of the total available carbohydrate using
dilute-acid treatment in combination with enzyme hydrolysis. The lower
yield in the current study is owing to higher concentrations of oligomers
(xylose) present in the prehydrolysate after steam explosion. According to
the model, the optimum in yield of fermentable sugars after pretreatment
and enzymatic hydrolysis was obtained at 170-200°C and 4-7 min (Fig. 3A).
S02 concentrations also influenced the recovery of monomeric sugars after
pretreatment and hydrolysis. Maximum recovery of monomeric sugars
(85%) was obtained at temperatures between 185 and 205°C and S02 concen-
trations between 3 and 6% (Fig. 3B). In addition, the influence ofS02impreg-
nation at a given temperature on the monomer recovery after pretreatment
and enzymatic hydrolysis was evident at 190°C, 5 min at S02 concentrations
ranging from 0 to 6% (Table 4). Similar to the findings by Clark et al. (19),
the enzymatic digestibility of the water-insoluble com fiber and monomer
sugar yield were substantially improved over the situation in which no S02
impregnation was employed. However, above 3% S02 concentration, there
were no further observable improvements.
Fermentation of Corn Fiber Hydrolyzate
The water soluble fractions obtained from each of the pretreatment,
posthydrolysis, and enzymatic hydrolysis steps were next assessed for their

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


S02-Catalyzed Steam Pretreatment of Corn Fiber 331

Sugar
recovsry ( ..)

Tempel1lture ("C)

Sugar
recovery (..)

Tem peratu r. ( ·Cl

Fig. 3. Response surface for monomeric sugar recovery after pretreatment and
enzymatic hydrolysis regarding (A) temperature and time, and (B) temperature and
502 concentration.

efficiency of fermentation to ethanol, without employing any detoxifica-


tion steps. The average ethanol yield in the fermentation of sugars present
in the liquid fractions obtained after pretreatment, posthydrolysis, and
enzymatic hydrolysis were determined (Table 6). The relative ethanol yield,
YEtOIIY~~~)H was defined as the ratio of the ethanol yield of the filtrate and the
reference fermentation. There was a clear relationship between the
fermentability of the prehydrolysate obtained after steam explosion and
severity. As expected, after 6 h, only the hexose sugars glucose and man-
nose, liberated from the corn fiber in the prehydrolysate, were effectively

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


332 Bura et at.

Table 6
Relative Ethanol Yield for Liquid Fraction Obtained After Pretreatment
(Prehydrolysate) (6 h), Enzymatic Hydrolysis (2 h), Posthydrolysis (24 h),
and Concentration of HMF and Furfurals in Prehydrolysate (Original Sugar %)
Enzymatic
Prehydrolysate hydrolysis Posthydrolysis
HMF Furfurals YEtOH/yref
EtOH
Y
EtOH
/yref
EtOH
Y
EtOH
/yref
EtOH
Treatment (%) (%) (%) (%) (%)
170°C, 1 min, 3% 0.01 0.57 79 85 80
150°C, 5 min, 3% 0.01 0.18 85 100 83
170°C, 5 min, 0% 0.03 0.34 79 83 81
170°C, 5 min, 6% 0.07 0.87 82 81 89
170°C, 9 min, 3% 0.12 1.10 85 100 83
190°C, 5 min, 3% 0.24 ± 0.01 1.17 ± 0.07 94 ± 1 89 ± 1 85 ± 4
210°C, 2.2 min, 1% 0.29 0.97 79 81 86
210°C, 2.2 min, 5% 0.45 1.03 81 87 84
210°C, 7.8 min, 1% 1.07 1.05 66 92 84
210°C, 7.8 min, 5% 1.47 1.38 69 95 82
230°C, 5 min, 3% 1.73 1.93 69 96 81
190°C, 5 min, 0% 0.23 1.01 91 86 80
190°C, 5 min, 6% 0.26 1.20 95 89 89

consumed by S. cerevisiae. Conversions to ethanol increased with increased


severity of the pretreatment, reaching a maximum at 190°C, 5 min, and 3%
S02' demonstrating a yield of 94% of theoretical (Table 6). The model pre-
dicted a range of optimum pretreatment conditions regarding time, tem-
perature, and S02 concentrations (Fig. 4 A,B) and was in agreement with
experimental data. According to the model, the optimum relative ethanol
yield (96%) of fermentable sugars after pretreatment was obtained at 181-
193°C, 4-5.5 min, and S02 concentrations from 2.2 to 3.6% (Figs. 4A,B).
However, at severities higher than 3.4, a distinct decrease in fermentability
was observed. It is well established that as steam explosion severity in-
creases, so does the degradation of monomeric sugars, by dehydration and
condensation reactions (20). As shown in Table 6, the amount of HMF and
furfurals present increased with the severity of pretreatment conditions.
However, the low concentration of inhibitors generated at low tempera-
tures (lower than 190°C) and the poor fermentability of prehydrolysate
obtained at these conditions suggests that other byproducts may be acting
as inhibitors. It has been shown that sugar degradation products such as
HMF and furfurals are less inhibitory than lignin-derived compounds such
as vanillin or syringylaldehyde on ethanol fermentation by S. cerevisiae (18).
There was no clear relationship between the fermentability of the liq-
uid fractions obtained after enzymatic hydrolysis and posthydrolysis and

Applied Biochemistry and Biotechnology Vol. 105-108,2003


SOl-Catalyzed Steam Pretreatment of Corn Fiber 333

Ethanol
yield (""J

Tempenlture rCJ ~ ~
.
B

Ethanol
yield (""J

Fig. 4. Response surface for relative ethanol yield for liquid fraction obtained after
pretreatment respect to (A) temperature and time, and (B) temperature and 502
concentration.

pretreatment severity (Table 6). Relative ethanol yields were high, ranging
from 81 to 100%, and 80 to 89% for sugars obtained after enzymatic
hydrolysis and posthydrolysis, respectively. The relatively high ethanol
yields obtained during 2 h of fermentation of the liquid fraction collected
after enzymatic hydrolysis were comparable with the results obtained by
Allen et al. (4). After simultaneous saccharification and fermentation of
pretreated corn fiber by hot water and steam explosion, 86 and 90% conver-
sion of glucan to ethanol by S. cerevisiae was observed, respectively.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


334 Bura et al.

Posthydrolysis, using the H 2S04 catalyzed aqueous solution obtained from


S02-catalyzed steam explosion of com fiber depolymerized the soluble
oligomeric hemicelluloses and released most of the available carbohydrates
in a fermentable form. At optimal pretreatment conditions, posthydrolysis
increased the concentration of sugars available for fermentation by 10%.
These results were comparable to work done by Schevchenko et al. (14),
who reported 100% conversion rates for hexoses present in the post-
hydrolysates of S02-catalyzed steam exploded Douglas-fir.

Conclusion
Each of the parameters investigated during steam pretreatment by
response surface modeling (temperature, time, and S02 concentration)
had an effect on the hemicellulose monomer recovery, its subsequent
fermentability, and enzymatic hydrolysis. Although the wide range of
pretreatment conditions investigated precluded an exact determination
of optimal conditions, the predicted optimal pretreatment conditions
were in agreement with the optimum conditions observed for maximum
sugar yield obtained after pretreatment and enzymatic hydrolysis, and
fermentability of the monomeric sugars. The results show that there is a
need for S02 impregnation of com fiber prior to steam explosion.
A two-stage treatment for com fiber processing, comprising S02-cata-
lyzed steam explosion and hydrolysis by a mixture of cellulolytic and
amylolytic enzymes, proved to be a very effective method for converting
the available polysaccharides in the residue to monomeric sugars, as evi-
denced by the 81 % conversion observed without the need for a detoxifica-
tion step. Maximum sugar yields (soluble and following enzymatic
hydrolysis) were recovered from com fiber pretreated at 190°C for 5 min
with 3% S02. Posthydrolysis of the aqueous solution from S02-catalyzed
steam explosion effectively depolymerized the soluble oligomeric hemicel-
lulose and increased the concentration of carbohydrates in a fermentable
form by 10% at optimum pretreatment conditions. Subsequently,
S. cerevisiae was able to convert the resultant com fiber hydrolysates,
prehydrolysate and liquid fraction from posthydrolysis to ethanol effi-
ciently, respectively yielding 89, 94, and 85% of theoretical conversion.

Acknowledements
We thank Dr. A. Boussaid and D. J. Gregg for assistance in the steam
explosion experiments.

References
1. Anderson, R. A. and Watson, S. A. (1982), in Handbook of Processing and Utilization in
Agriculture, vol. 2, Wolff, I. A. ed., CRC, Boca Rat~n, FL, pp. 31-61.
2. Grohmann, K. and Bothast, R. J. (1996), Process Biochem. 32,405-415.
3. Moniruzzaman, M., Dale, B. E., Hespell, R. B., and Bothast, R. J. (1997), Appl. Biochem.
Biotechnol. 67, 113-126.

Applied Biochemistry and Biotechnology Vol. 705-108, 2003


SO[Catalyzed Steam Pretreatment of Corn Fiber 335
4. Allen, S. G., Schulman D., Lichwa, J., and Antal, M. J. (2001), Ind. Eng. Chem. Res. 40,
2934-294l.
5. Bura, R., Mansfield S. D., Saddler J. N., and Bothast R. J (2002), Appl. Biochem. Biotechol.
98-100, 59-72.
6. Overend, R. P. and Chornet, E. (1987), Phil. Trans. R. Soc. Lond. 321, 523-536.
7. TAPPI, Technical Association ofthe Pulp and Paper Industry (1998), TAPPI Standard
Methods, T-222 om-98, TAPPI Press, Atlanta, GA.
B. Ghose, T. K. (1987), Pure Appl. Chem. 59,257-268.
9. Box,G. E. P., Hunter, W. G., and HunterJ. S. (1978),inStatisticsfor Experimenters,John
Wiley & Sons, New York, NY, pp. 510-539.
10. Clark, T. A. and Mackie, K. L. (1987), J. Wood Chem. Techno/. 7,373-403.
11. Boussaid, A., Jarvis J., Gregg, D. J., and Saddler, J. N. (1997), in The Third Biomass
Conference of the Americas, Overend, R. P. and Chornet, E., eds., Montreal, Canada,
Elsevier Science, pp.878-880.
12. Stenberg, K., Tengborg, Ch., Galbe, M., and Zacchi, G. (1998), J. Chem. Technol.
Biotechnol. 71, 299-308.
13. San Martin, R., Perez, c., and Briones, R. (1995), Bioresour. Technol. 53,217-223.
14. Shevchenko, S. M., Chang, K., Robinson, J., and Saddler, J. N. (2000), Bioresour.
Biotechnol. 72,207-21l.
15. Excoffier, G., Toussaint, B., and Vignon, M. R. (1991), Biotechnol. Bioeng. 38,1308-1317.
16. HespeU, R. B., O'Bryan, P.J., Moniruzzaman, M., and Bothast R. J. (1997), Appl.
Biochem. Biotechnol. 62, 87-97.
17. Montgomery, R. and Smith, J. (1970), J. Am. Chem. Soc. 79,695-699.
lB. Delgenes, J. P., Moletta, R., and Navarro J. M. (1996), Enzyme Microb. Technol. 19,
220-225.
19. Clark, T. A., Mackie, K. L., Dare, P., H. and McDonald, A. G. (1989), J. Wood Chem.
Technol. 9, 135-166.
20. Schwald, W., Smaridge, T., Chan, M., Breuil, c., and Saddler, J. N. (1987), in Enzyme
Systems for Lignocellulose Degradation, Coughlan, M. P. ed., Elsevier, New York, NY,
pp.231-242.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 / 0337 / $20.00

A Comprehensive Kinetic Model


for Dilute-Acid Hydrolysis of Cellulose

QIAN XIANG, l JUN SEOK KIM/ AND Y. Y. LEE*,l

1Department of Chemical Engineering,


230 Ross Hall, Auburn University, Auburn, AL 36849,
E-mail: yylee@eng.auburn.edu; and
2Deptment of Chemical Engineering, Kyonggi University,
Paldalku, Ewuidong, Suwon, Korea

Abstract
Acid-catalyzed hydrolysis is controlled not only by temperature and acid
concentration but also by the physical state of the cellulose. Under low tem-
perature and acid condition the cellulose structure stays in stable crystalline
form. Therefore, the prevailing reaction mode is endwise hydrolysis. Glu-
cose then becomes the main sugar product. However, when temperature
and/ or acid concentration is raised to a certain level, the cellulose structure
becomes unstable by breakage of hydrogen bonding, the primary force that
holds the cellulose chains. Once the crystalline structure of the cellulose is
disrupted, acid molecules can penetrate into the inner layers of the cellulose
chains. In support of this hypothesis, we have experimentally verified that a
substantial amount of oligomers is formed as reaction intermediates under
extremely low-acid and high-temperature conditions. We also found that the
breakage of hydrogen bonds occurs rather abruptly in response to tempera-
ture. One such condition is 210°C, 0.07% H 2S04 • Glucose, once it is formed in
the hydrolysate, interacts with acid-soluble lignin, forming a lignin-carbo-
hydrate complex. This occurs concurrently with other reactions involving
glucose such as decomposition and reversion. On the basis of these findings,
a comprehensive kinetic model is proposed. This model is in full compliance
with our recent experimental data obtained under a broad range of reaction
conditions.

Index Entries: Acid hydrolysis; cellulose; kinetics; model; hydrogen bond;


oligomer; lignin.

Introduction
The origin of the organized kinetic study of cellulose hydrolysis dates
back to that of the work of Saeman (1). This kinetic model describes dilute-
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 337 Vol. 105-108,2003


338 Xiang et al.
acid hydrolysis as two pseudo-homogeneous consecutive first-order
reactions:

Cellulose ~ Glucose !..L. Degradation products


Hydrolysis of glycosidic bonds of soluble reactants in homogeneous
phase has indeed been verified to follow first-order reaction (2,3). The
reaction rate constants are expressed in the form of an Arrhenius equation
with an additional term indicating the effect of acid:
kl = kiO x Am; x e(E/RT)
in which kiQ is the preexponential factor, A is the acid concentration, miis the
acid exponential, and Ei is the activation energy.
This simple kinetic model was established on the basis of data obtained
under a limited range of reaction conditions: the conventional dilute-acid
hydrolysis conditions of 180-200°C, 0.2-2% acid. The question is, can this
model be extrapolated beyond the aforementioned region, and if so, how far?
This question arises because the reaction is heterogeneous and factors other
than chemical reaction may influence the overall process. Limitation of this
model is also seen in the case of partial hydrolysis with highly dilute acids,
yielding so-called hydrocellulose, a product with reduced degree of poly-
merization (DP) but higher crystallinity (4).
The rate of heterogeneous hydrolysis of celluloses is one to two orders
of magnitude lower than those of homogeneous hydrolysis of model com-
pounds. Heterogeneous hydrolysis has been found to be influenced by
such physical factors as the degree of crystallinity, the state of swelling, and
mechanical disintegration (5,6).
Significant advances have been made to the original model of Saeman
(1). The presence of amorphous cellulose and reversion reactions of glucose
were implemented in a model suggested by Conner et al. (7):

Easily hydrolyzed
celluose Anhydrosugars
~ /11/
Glucose ~ Degradation products

Resistant cellulose
/I ~t
Dissacharides ~
" t
Glucosides

Another important development in the kinetics was the finding of


parasitic pathways. Using thermogravimetric analysis, differential scan-
ning calorimetry, and diffuse reflectance infrared, Bouchard et al. (8)
detected some significant changes in the chemical structure of unhy-
drolyzed cellulose. Mok et al. (9) confirms this finding. For these studies,
the hydrolysis conditions were within the 200-240°C temperature ranges

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Model for Dilute-Acid Hydrolysis of Cellulose 339
and at low sulfuric acid concentrations (0.2-1.0% [w fwD. Both studies
reported that a portion of the cellulose turns into nonhydrolyzable oligo-
mers and, thus, cannot be hydrolyzed to glucose on further treatment. This
was cited as the major reason that glucose yields above 60-65% are not
attainable. The reaction pattern of the parasitic pathway is as follows:

Modified cellulose
insoluble oligomers

Cellulose Glucose ~ Further reactions

Soluble
Oligo saccharides

Cellulose and hemicellulose experiments have both reported evidence


of oligomer intermediates. Kim and Lee (10) reported improvements in
yields following secondary hydrolysis. Abatzoglou et al. (11) observed the
presence of oligomers during hydrolysis of a-cellulose in a cascade reactor
under conditions within 200-240°C and at low sulfuric acid concentrations
(0.2-1.0% [w fwD. Significant amounts of oligoderivatives were observed
in the hydrolysate of the early stages of hydrolysis, and conversion of oli-
gomers to glucose was two to three times faster than hydrolysis of cellulose
to soluble oligomers.
Our laboratory, in cooperative research with the National Renewable
Energy Laboratory (NREL), explored the use of a bed-shrinking flow-
through (BSFT) reactor (12) under extremely low-acid and high-tempera-
ture conditions. We have obtained experimental data in which glucose
yields surpassed 80%. This certainly contradicts the current kinetic models.
We conducted an experimental investigation seeking a plausible explana-
tion for this contradiction. From the results, we propose a comprehensive
kinetic model for cellulose acid hydrolysis. This model recognizes the het-
erogeneous nature of the reaction and offers a potential application for a
broader range of reaction conditions.

Materials and Methods


Sawdust, Feedstock, and Chemicals
Yellow poplar sawdust was provided by NREL. It was milled and
stored frozen at NREL before being sent to our laboratory. It was further
milled to pass through a 2-mm screen. The chemical composition of the
yellow poplar was 42.8 wt% glue an, 14.8 wt% xylan, 2.5 wt% mannan,
0.9 wt% galactan, 0.5% arabinan, 24.3 wt% klason lignin, 2.8 wt% acid-
soluble lignin (ASL), and 0.7 wt% ash, on an oven-dried basis. Sugarcane
bagasse feedstock was provided by BCl (Jennings, LA). The chemical

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


340 Xiang et al.
composition of bagasse, which was milled to pass through a O.B-mm
screen, was determined to be 37.4% glucan, 17.9% xylan, and 21.7% klason
lignin. .
Chemicals were of chemical grade, purchased from Sigma-Aldrich.
The chemical composition of a-cellulose (P. no. C-B002) (20 micro) was
analyzed to be 92.2% glucan, 3.4% xylan, and 3.2% mannan on a dry basis.

Experimental Apparatus and Procedures


Hydrolysis of pretreated yellow poplar was performed in both batch
and BSFT reactors. For dilute-acid-catalyzed hydrolysis at high-tempera-
ture conditions, experiments were performed using sealed, tubular batch
reactors. The reactors (19.0 cm3 of internal volume) were constructed of
Hastelloy C-276 tubing (1/2-in. id x 6-in. lenght) capped with Swagelok
end fittings. Approximately 0.6 g of a-cellulose and 12 g of 0.07% H 2S04
solution were loaded into the reactor. The solid/liquid ratio was main-
tained at the same level of 1/20 in all batch operations. The reactors were
first submerged into the preheating bath set at 30°C above the desired
reaction temperature for rapid preheating. The reactors were then quickly
transferred into another sand bath set at the desired reaction temperature
after 1.5 min. The reactor temperature was monitored by a thermocouple
inserted into the reactor. After the desired reaction time, each reactor was
taken out of the sand bath and quenched in a cold-water bath. The experi-
mental parameters were the time variation of the solid composition and the
monomeric sugar concentrations in liquid. For batch hydrolysis under high
concentration of acid (4% H 2S04) at 120°C, experiments were conducted
in an autoclave using Pyrex glass bottles with lS0-mL volume capacity.
Hydrolysis experiments were also conducted in a semi-batch flow-through
reactor and BSFT reactor. Details of the flow-through kinetic experiments
are described elsewhere (12).
a-Cellulose was treated by concentrated H 2S04 for the study of the
effect of cellulose structure on hydrolysis. Three grams of cellulose was
added to SO-mL solutions with varying H 2S04 concentrations. The treat-
ment was conducted at 25°C for 4 h. After the pretreatment, distillated
water was added to dilute the acid. The reprecipitated cellulose was fil-
tered and washed to neutral. It was then stored in a wet condition with 50-
70% moisture content for further experiments.

Analytical Methods
Solid samples from experiments were analyzed for glucan content
according to the NREL analysis procedures (13). Oligomeric sugars in the
hydrolysate liquor were converted into monomers using 4% (w /w) H 2S04
hydrolysis at 120°C for 60 min. Sugars and other compounds were deter-
mined by an high-performance liquid chromatography (HPLC) using a
Bio-Rad Aminex HPX-87P column. Oligomer samples were also analyzed
by HPLC using a Bio-Rad Aminex HPX-42A column.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Model for Dilute-Acid Hydrolysis of Cellulose 341
Table 1
Observed Glucose Yields from Various Reactor Operations
Maximum yield (%)
Feedstock Condition Reactor 250°C 220°C 235°C
Yellow poplar BSFT 87.5 90.3 90.8
Nonshrinking
flow-through 64.4
a-cellulose Batch 40.2
BSFT 80.6
Sugarcane bagasse 0.07% H 2S04 Untreated 52.6
and BSFT reactor 50% Delignified 61.3
85% Delignified 76.5

Results and Discussion


Observed Glucose Yields from 85FT Reactors
The unique characteristic of the BSFT reactor is that a spring is placed
inside the reactor to press the biomass during the hydrolysis reaction. The
BSFT reactor, which was originally developed at NREL, keeps a dense bed
condition and minimizes the residence time of the liquid. During the acid
hydrolysis process, degradation of glucose occurs concurrently. Short resi-
dence time in the BSFT reactor reduces such degradation, resulting in high
glucose yields. For the present work, we modified the BSFT reactor by
adding a quick quenching heat exchanger at its outlet. With this modifica-
tion, we were able to eliminate posthydrolysis degradation, raising the
accuracy in the kinetic experiments. The superior performance of the BSFT
reactor is reaffirmed when it is operated at extremely low-acid and high-
temperature conditions. Table 1 summarizes the glucose yields obtained
in our work. With yellow poplar feedstocks, the highest yield obtained
using the BSFT reactor was 90%, occurring at 205-235°C and 0.07% H 2S04 •
This result agrees closely with the work at NREL (14). However, with use
of a normal nonshrinking flow-through reactor, the glucose yield was only
64.4%, which is in line with the prediction from the conventional hydroly-
sis kinetic theory. For a-cellulose, the maximum yield in a batch reactor
was about 40.2% under 0.07% H 2S04 and 205°C. The yield was greatly
improved to 80.6% when the BSFT reactor was used. A number of
BSFT experiments were also conducted with sugarcane baggase. Among
the notable points of these experiments is that hydrolysis of sugarcane
bagasse was strongly affected by the amount of lignin present in the feed-
stock. For hydrolysis of untreated bagasse, an average yield of 52.6%
was obtained under 0.07% H 2S04 and 220°C using the BSFT reactor. For
bagasse pretreated with 1% H202 solution at 170°C, about 50% of the

Applied Biochemistry and Biotechnology Vol. 105-108,2003


342 Xiang et al.

8 r---------------------~==~~~~~==~--~
_____ Raw Glueo •• wi %
Raw C.llublolt wi %

.2(ii 1
Total Clueo •• wi %
Clueo.. Oflgomtr wi %
"3
E
:::I
6
U
U

'c:" 5
'e
-
It>

~ 4
c:
.2
'5 3
Etil
oc: 2

'"
U
::J
(5 1

o 5 10 15 20 25 30 35 40 45 50 55 60 65 10 75 80 85
Time (min)

Fig. 1. a-Cellulose hydrolysis in BSFT reactor at 205°C and 0.07% H 2S04 ,

lignin was removed. Using this substrate, the glucose yield was improved
to 61.3%. The substrate was also pretreated with 10% aqueous ammonia to
remove 85% lignin. When this substrate was applied, the yield was further
improved to 76.5%. The level of glucose yield observed in our work and its
dependency on pretreatment and reactor type cannot be explained by
existing hydrolysis kinetic theory.
Presence of Oligomers in Hydrolysate
Figure 1 shows the hydrolysis profile for a-cellulose in a BSFT reactor
at 205°C and 0.07% H 2S04 , The observed yield of total soluble sugars
(glucose, cellobiose, and oligomers) is 80.6%. The yield is substantially
higher than those observed in our previous experiments using batch
reactors and percolation reactors. One notable feature of this run is the
presence of oligomers that were not observed in our previous experi-
ments. We believe this is owing to the unique characteristics of the BSFT
reactor and the quick quenching of the hydrolysates, which prevented
further reactions of intermediates. A sample HPLC chromatogram is
shown in Fig. 2. Glucose and cellobiose contents were obtained from
HPLC chromatograms of raw hydrolysate samples. The same liquid
samples were put through a secondary hydrolysis that was carried out
with 4% H 2S04 at 121°C for 1 h. The glucose yield was calculated from the
total glucose value after secondary hydrolysis. The yields of this magni-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Model for Dilute-Acid Hydrolysis of Cellulose 343

,I Glucose

/'

Cellobiose J

t
J
I
Fig. 2. Evidence of oligomer presence in HPLC chromatogram.

tude and the presence of oligomers in the hydrolysate cannot be explained


by published cellulose hydrolysis kinetic models. To validate these experi-
mental results, selected samples from these experiments were sent to
NREL for further analysis on sugar contents and oligomers. A sample
chromatogram with high resolution of oligomers is shown in Fig. 3.
The chromatogram shows that glucose and short-chain oligomers are the
major products under the hydrolysis conditions applied for a-cellulose.
The amount of oligomers existing in the hydrolysates decreased with the
increase in degree of polymerization.
Acid hydrolysis of crystalline cellulose in biomass is a heterogeneous
reaction. According to the kinetic model of 5aeman (1), glucose monomer
is the only sugar product in the hydrolysate. The results obtained from
B5FT reactor experiments (Figs. 1-3) obviously disprove it since we found
a significant amount of oligomers as well as glucose monomer. The oligo-
mers are produced either through a parallel pathway or as intermediate
products before glucose. The facts that glucose is the predominant compo-
nent among the three different species of sugar products (glucose, cello-
biose, and higher oligomers) and that the amount of oligomer decrease
with increasing DP indicate that the serial (intermediate) pathway is
unlikely. With this understanding, we modified the hydrolysis kinetic
model by inserting a parallel pathway for release of oligomers.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


344 Xiang et al.

I
./
~

I

. •..
o
•-

,
I
.......01'---:._ _ _ _ _ .1. -

Fig. 3. HPLC chromatogram with high resolution of oligomers (courtesy of Dr.


Raymond Ruiz, NREL).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Model for Dilute-Acid Hydrolysis of Cellulose 345

r-------,
H+
I r, . .t ,
Glu
... .
GIU-GIU" Glu-GIU;Glu
.ol i ... ,
H+

-=G1u;..Gl u-Glu",Glu-G1 u""Glu-:-G1 g",..Glu-Glu;GIU:::··


Cellulose
C bains ::::GI u:GIlC.GID=GI tr.:Ghl::·GlltGl U::GIu-:::Gl u::::G1u-

Fig. 4. Illustration of crystalline cellulose structure and acid hydrolysis mechanism.

Association of Glucose/Oligomer with Cellulose into Kinetic Model


We are also investigating on the heterogeneous nature of cellulose
hydrolysis and the role of hydrogen bonding on the kinetics of cellulose
hydrolysis in a parallel research in our laboratory (15). From this study
we find that the supramolecular structures of cellulose must be taken into
consideration in the mechanism of cellulose hydrolysis. Since cellulose
supramolecular structures are built on various types of hydrogen bonding
between OH groups, we now propose a hypothesis that the acid hydrolysis
of cellulose is influenced by hydrogen bonding within the macrostructure
of crystalline cellulose. Figure 4 illustrates the hydrogen bonds within crys-
talline cellulose and the association between glucose/oligomers and the
cellulose. Hydrogen bonding is believed to be the major factor that controls
the structure and the physical property of cellulose (16-18). There are sev-
erallevels of hydrogen bonding in the cellulose structure: intramolecular
hydrogen bonding, intermolecular hydrogen bonding, and hydrogen
bonding between the surface of cellulose and water/medium molecules
(16). Formation of intramolecular hydrogen bonds has been proven to con-
tribute directly to certain physical properties of cellulose such as crystallin-
ity and solubility in various solvents having different polarities. It also
affects the relative reactivity of the cellulose (18). In the overall molecular
structure, glucose (C 6H 120 6) is quite similar to xylose (CSHlOOS). However,
hydrolysis of cellulose is about one to two orders of magnitude lower than
xylan. By comparing the ring structures of both glucose and xylose pyra-
nose form, the additional CH20H group connected to the cyclic structure
of glucopyranose (the stable structure form for glucose) provides much
higher probability for the formation of intra- hydrogen bonding and inter-
hydrogen bonding than the cyclic structure of xylopyranose (the stable
structure form for xylose). The strength of the hydrogen bonds in cellulose
makes it stable to the attack by acidic catalytic elements in aqueous solu-
tion. The hydrogen bonds impede the access of the catalyst to the active site

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


346 Xiang et al.

1
a-Ce1Iu1ose

a. -Cellulose Dissolvt!d Into


65% H,SO. &\ Repreclpltllted

0.1 +------.--.-------.--.------.--.------.--.-------.-'------1
o 10 20 30 40 50 60 70 ~ 00 100
Time (min)

Fig. 5. Comparison of hydrolysis rates of starch, and treated and untreated cellulose
(120°C, 4% H 2S04),

and the release of glucose. During the catalytic acid-hydrolysis process


shown in Fig. 4, the hydronium ions in the liquid randomly attack the
glucosidic bonds and consequently lower the stability of hydrogen bonds.
When both glucosidic bonds and hydrogen bonds are cleaved, glucose and
oligomers are readily released to bulk liquid.
The observed kinetics of acid hydrolysis follows a first-order reaction
and the temperature effect conforms to an Arrhenius equation. However,
we find that the Arrhenius equation is applicable only within a limited range
of temperature and acid concentration. The kinetics of cellulose hydrolysis
strongly depends on the physical state of cellulose. Particularly notable is
the drastic difference in hydrolysis rate between crystalline cellulose and
the dissolved and reprecipitated cellulose (Fig. 5). Cellulose is totally dis-
solved in 65% H 2S04 at room temperature. The cellulose is reprecipitated
when it is diluted with water. The reprecipitated cellulose is hydrolyzed at
a rate two orders of magnitude higher than that of crystalline cellulose and
at about the same rate as cornstarch (Fig. 5). Sasaki et. a1. (19,20) reported a
similar case. They discovered that cellulose is rapidly dissolved and depo-
lymerized in supercritical water at 300-320°C without any acid. The cellu-
lose hydrolysis rate under these conditions is again far above the prediction
by extrapolation of the Arrhenius equation based on the data over
260-280°C. The investigators reported a one- order-of-magnitude jump of
hydrolysis rate when the temperature was raised from a sub critical tem-
perature to near or above critical temperature (-300°C). We observed simi-
lar phenomenon in the hydrolysis of pretreated yellow poplar carried out at
high-temperature and extremely low-acid conditions. In Fig. 6, an abrupt

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Model for Dilute-Acid Hydrolysis of Cellulose 347

~
Cl
c
'2
...
'(ij
E ,. 1.:-........'. -.
"
a:
1iu
estimated treadllne for
:::I
a hydrolysis at 225 °c

Actual Experlmentlll data


for hydrolysis at 226°C

'0 20 ,5

Time (min)

Fig. 6. Influence of temperature on hydrolysis rate. Kinetic profiles from hydrolysis


of pretreated yellow poplar in batch reactor with 0.07% H 2S04 are shown.

jump of reaction rate is observed between 210 and 225°C. The actual mea-
sured rate constant at 225°C is 0.137, which is 2.3 times larger than the value
extrapolated by Arrhenius equation based on the data at lS0-200°C.
We have also observed similar behavior with hydrolysis of a-cellulose
in that a discontinuity of reaction behavior occurred at a certain tempera-
ture (15). This was owing to a temperature-induced disruption of cellulose
physical structure. On devastation of cellulose supramolecular structure,
catalytic hydrolysis becomes much more rapid since the catalytic group can
gain much easier access to cellulose without having to penetrate into the
crystalline barrier. Furthermore, the hydrolysis products, including glu-
cose and oligomers, are readily released into liquid owing to weakened
hydrogen bonding. This finding partly explains why the cellulose hydroly-
sis rate is drastically improved and high glucose yields are obtained under
high-temperature and extremely low-acid conditions.
For the sake of discussion, we introduce a reaction index applicable for
the hydrogen bonds, called the instability index (10)' It is an index similar
to the severity factor in that it involves two parameters: temperature and
acid concentration. The ionic strength of the solution may also be included
into the definition of the instability index. The exact mathematical expres-
sion of 10 is yet to be determined. It is an entity that increases when acid
concentration and/ or temperature increases.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


348 Xiang et al.
Ea

Zone or Un tabl
H-Bonds

Tran itiona!
Zon of Stable Zone
H-Bonds

Fig. 7. Activation energy vs instability index of hydrogen bonds.

Because of the direct involvement of hydrogen bond in the hydrolysis


reaction, its instability affects the kinetics, especially the activation energy
or the temperature dependency (Fig. 7). For convenience, the hydrolysis
conditions are divided into three different zones according to different
instability index. Each zone has different activation energy for the acid
hydrolysis reaction. At low instability index region, which means low tem-
perature or low acid concentration, the hydrogen bonds are relatively stable
and the hydrolysis rate constant has low activation energy. According to
our experimental data, temperatures under 210°C and <0.07% H 2S04 fall
under the stable zone. Above this zone, hydrogen bond becomes progres-
sively unstable. Thus, the experimental data taken in the stable zone cannot
be extrapolated to the unstable zone, and vice versa.
We have observed oligomers in the hydrolysates. The distribution of
oligomers with varying DP is determined by reaction conditions and reac-
tor configurations. According to its definition, the 10 determines the state of
hydrogen bonds, and thus determines how easily the crystalline structure
is broken up and products are released into liquid medium. The higher-DP
oligomers have a higher number of hydrogen bonds to the residual cellu-
lose; thus, it is more difficult to be released into the liquid medium than the
low-DP oligomers. Obviously, glucose is released most easily. At lower 10 ,
hydrogen bonds are so stable that only glucose monomer is released. The
long-chain oligomers stay on the surface of the cellulose because of the
strong binding force of hydrogen bonds until they are further hydrolyzed
to monomer. We believe that is the reason why glucose is the predominant
product under low 10 conditions. This would explain why the amount of
oligomer decreases with DP in the hydrolysate (Fig. 3). On the other hand,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Model for Dilute-Acid Hydrolysis of Cellulose 349

4-Pure Glucose Solution


-+-Glucose In Pre-hydrolys te liquor

~ 0.8

0.7

Time (min)

Fig. 8. Effect of lignin on glucose decomposition at 200°C and 0.1 % H 2S04 ,

at high 10 conditions such as concentrated H 2S04 or above 210°C, hydrogen


bonds become unstable and eventually disrupted. Only then, are the oligo-
mers directly released. When this happens, the high-DP oligomers are less
likely to be released than the low-DP ones because high DP induces a high
number of hydrogen bonds, and thus high binding force.
Glucose-Lignin Interaction
In several of our experiments, we have observed that the disappear-
ance of glucose during acid hydrolysis of lignocellulosic biomass was faster
than that of model prediction. We speculate that glucose may recombine
with ASL in hydrolysates; the data in Table 1 support this hypothesis.
Glucose yield from hydrolysis of sugarcane bagasse increases substan-
tially when the lignin content in the feedstock is reduced by pretreatment.
We also observed that glucose decomposed faster when the medium was
supplemented with ASL (Fig. 8). From these observations, we conclude
that there exists an additional pathway for glucose decomposition caused
by the glucose-lignin interaction, most likely recondensation between glu-
cose and ASLs. With part of lignin being solubilized into medium at high
temperature, the presence of protons causes the formation of intermediate
carbonium ions that have a high affinity for nucleophilic reaction moiety.
This in turn causes lignin fragmentation or lignin condensation reactions,
depending on the type of the active nucleophile (21). Thus, the active sites

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


350 Xiang et at.

ICrystuJlin2 Celbdose I
Surface Long
" ChtUn OligOntel'$

Surface Glucose I
Short ChtUn Oligomel'$

Degradation + Soluble Lignin


Products .....- + --~~. LCC

t jr---T""r~'" (Lignin CarlJohydrate Complex)

Glucosides
r-------- . .
________
I
IL
LOlqUl°d JI II
II
_______________________________________ J

Fig. 9. Proposed comprehensive kinetic model for acid hydrolysis of lignocellulose.

on ASL will lead to the condensation reaction between ASL and glucose as
follows:
ASL + Glucose - ASL - Glucose
However, we have yet to positively identify the lignin-carbohydrate
complex in the hydrolysates. Further investigation on this topic is currently
under way in our laboratory.
Proposed Hydrolysis Kinetic Model
On the basis of our experimental findings, we propose a comprehen-
sive kinetic model as described in Fig. 9. This model accounts for factors
pertaining to the heterogeneous nature of the reaction, the hydrogen-bond-
ing theory, and posthydrolysis reaction, specifically glucose-ASL interac-
tion. In the solid phase during the hydrolysis process, the strength of
hydrogen bonds controls the release of either glucose or oligomers to the
liquid. Oligomers are released at high 10 conditions, whereas glucose is
released at low instability conditions. Within the high 10 reaction zone,
oligomers are released into the liquid and further hydrolyzed to glucose in
the acidic liquid medium. In the liquid phase, glucose takes an additional
pathway owing to interaction with ASL at high-temperature conditions.
This occurs concurrently with other reactions involving glucose such as
decomposition and reversion.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Model for Dilute-Acid Hydrolysis of Cellulose 351

Conclusion
The conventional kinetic model of two sequential reactions is inad-
equate as a general model since it does not account for the heterogeneous
nature of the reaction. The heterogeneity of the reaction creates physical
resistances that are far greater than the reaction resistance. The primary
factor controlling the physical resistance is the state of the hydrogen bond-
ing in cellulose molecular structure.
The state of hydrogen bonding affects the kinetic behavior. If the
hydrolysis reaction is carried out under a condition that severely disrupts
hydrogen bonding (high 10 condition), glucose and its oligomers are
formed. Since oligomers are more stable than monomer against decompo-
sition, adoption of such a reaction condition may increase sugar yield.
Quick quenching of the reactor effluent is an efficient way to raise the
glucose yield, especially in the high 10 zone. Glucose-lignin recondensation
occurs during acid hydrolysis of cellulose and is an important factor caus-
ing loss of sugar.
A comprehensive kinetic model is proposed implementing additional
factors into the conventional model: a parallel oligomer pathway, associa-
tion of glucose/oligomers with cellulose through hydrogen bonding, and
interaction of glucose with ASL. The proposed kinetic model provides valid
explanations for our recent laboratory data that contradict the conven-
tional acid hydrolysis model.

Acknowledgments
We wish to thank Rebert Torget of NREL and Par O. Pettersson of Mid
Sweden University for their technical insights and helpful suggestions, and
Dr. Raymond Ruiz of NREL for his contribution to the oligomer analysis by
HPLC. This research was sponsored by the US Department of Energy
through financial agreement DE-FC36-01GOII072, and by NREL through
subcontract ACO-1-31003-01.

References
1. Saeman, J. F. (1945), Ind. Eng. Chem. 37,43-52.
2. Springer, E. L. (1966), Tappi 49,102-106.
3. Daruwalla, E. H. and Shet, R. T. (1962), Text. Res. J. 32, 942-954.
4. Nelson, M. L. (1960), J. Polym. Sci. 43,351-371.
5. Philipp, B., Jacopian, V., Loth, F., Hirte, W., and Schulz, G. (1979), in Hydrolysis of
Cellulose: Mechanisms of Enzymatic and Acid Catalysis, Advances in Chemistry Series,
vol. 181, Brown, R. D., Jr. and Jurasek, L., eds., American Chemical Society, Washing-
ton, DC, pp. 127-143.
6. Millett, M. A., Effland, M. J., and Caulfield, D. F. (1979), in Hydrolysis of Cellulose:
Mechanisms of Enzymatic and Acid Catalysis, Advances in Chemistry Series, vol. 181,
Brown, R. D., Jr. and Jurasek, L., eds., American Chemical Society, Washington, DC,
pp.71-89.
7. Conner, A. H., Wood, B. F., Hill, C. G., and Harris, J. F. (1986), in Cellulose: Structure,
Modification and Hydrolysis, Young, R. A. and Rowell, R. M., eds.,J. Wiley & Sons, New
York, NY, pp. 281-296.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


352 Xiang et al.
B. Bouchard, J., Abatzoglou, N., Chornet, E., and Overend, R. P. (1989), Wood Sci. Technol.
23, 343-355.
9. Mok, W. S., Antal, M. J., and Varhegyi, G. (1992), I&EC 31, 94-100.
10. Kim, S. B. and Lee, Y. Y. (1987), Biotech. Bioeng. Symp. 17,71-84.
11. Abatzoglou, N., Bouchard, J., Chornet, E., and Overend, R. P. (1986), Can. J. Chem.
Eng. 64, 781-786.
12. Torget, R. W., Kim, J. S., and Lee, Y. Y. (2000), Ind. Eng. Chem. Res. 39,2817-2825.
13. Ehrman, T., Ruiz, R., and Templeton, D. (1995), NREL Chemical Analysis & Testing
Standard Procedure, LAP No. 002-05, No. 10, National Renewable Energy Labora-
tory, Boulder, CO.
14. Torget, R. W., Nagel, N., Jennings, E., Ibsen K., and Elander, R. (1999), presented at
21st Symposium on Biotechnology for Fuels and Chemicals, Fort Collins, CO.
15. Xiang, Q., Pettersson, P., Torget, R. W., and Lee, Y. Y. (2002), presented at 24th Sym-
posium on Biotechnology for Fuels and Chemicals, Gatlinburg, TN.
16. Lazareva, T. G. and Vashuk, V. (1996), Int. J. Mod. Phys. 10(23/24),2983.
17. Heiner, P. A. and Teleman, O. (1997), Langmuir 13(3), 511-518.
lB. Knodo, T. J. (1997), Polymer Sci. 35, 717-723.
19. Sasaki, M., Kabyemela, B., Adschiri, T., Malahuan, R., Hirose, S., Takeda, N., and
Arai, K. (1997), presented at 4th International Symposium on Supercritical Fluids, Sendai,
Japan.
20. Sasaki, M., Fang, Z., Fukushima, Y., Adschiri, T., and Arai, K. (2000), Ind. Eng. Chem.
Res. 39,2883-2890.
21. Fengel, D. and Wegener, G. (1984), in Wood Chemistry, Ultrastructure, Reactions, Walter
de Gruyter, Berlin, pp. 279-281.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0353/$20.00

Effect of Agitation on Removal of Acetic Acid


from Pretreated Hydrolysate
by Activated Carbon

SARAH A. PRIDDY AND THOMAS R. HANLEY*


Department of Chemical Engineering, University of Louisville,
Louisville, KY 40292, E-mail: tom.hanley@louisville.edu

Abstract
The effect of agitation on the adsorption of acetic acid by activated carbon
was tested utilizing an external mass transfer-diffusion model. Simulated
pretreated biomass was contacted with activated carbon under prescribed
conditions of temperature and agitation. Adsorption isotherm studies are
presented as well as batch kinetic rate studies. Use of these data enabled the
determination of isotherm constants, an external mass transfer coefficient,
and an effective diffusivity for each agitation rate studied. The external film
coefficient results ranged from 33.62 !-lm/ s to a complete absence of external
mass transfer resistance, and the diffusivity results ranged from 0.8625 to
10.70 !-lm2 / s. The optimum combination of no external film resistance, and
highest diffusivity, 10.70 !-lm2/s, occurred at 250 rpm and 25°C. The results
of these models and the experimental parameters suggested an efficacious
method and conditions for the removal of this undesirable chemical.
Index Entries: Adsorption; activated carbon; external mass transfer; effec-
tive diffusivity; detoxification.

Introduction
Biomass sources such as softwoods and corn stover have been touted
as viable solutions for the production of fuel ethanol. However, because
these sources do not contain significant quantities of sugar in their natural
state, they must be treated in a manner in which the lignocellulosic material
is reduced to fermentable sugars. This is accomplished by any number of
pretreatment processes such as dilute-acid hydrolysis (1). During the pre-
treatment of most biomass systems, several components resultant from
sugar and lignin degradation processes may be formed, including furfural,
hydroxymethylfurfural, and acetic acid. The quantity of the individual
chemicals can vary with biomass source and may range from parts per
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 353 Vol. 105-708, 2003


354 Priddy and Hanley
million to grams per liter. While most of these chemicals are never pro-
duced in great quantity, they impact the fermentation capability of the
ethanologenic organisms, especially bacteria (2). The impact of these
inhibitory chemicals on fermentation yield has also been studied. In
the case of the ethanologenic bacterium Zymomonas mobilis, concentrations
of acetic acid found in pretreated softwoods may reach 10 giL, completely
inhibiting fermentative ability (3).
To combat the problems associated with these inhibitory chemicals,
several detoxification schemes have been investigated including treatment
with alkali, sulfites, and ion exchange (4-7). While these methods were able
to remove some of the inhibitory components, little attention was paid to
the potential of activated carbon adsorption. Activated carbon adsorptions
provide process advantages such as ease of use and scale-up ability. They
are regenerated easily with steam, and the stripped components may be
collected and sold. In addition, powdered activated carbons are much less
costly than ion-exchange resins. Activated carbon processes have been in
use for more than 200 yr and are quite commonly used in the alcoholic
beverage industries (8). Activated carbons are produced by a wide variety
of substrates and may be found in either granular or powered forms for use
in both continuous and batch applications (9).
In the present study, an activated carbon hydrolysate detoxification
step was investigated. Following biomass pretreatment, the resulting
hydrolysate was processed through a separation step. In this step, the
hydrolysate substrate was contacted with an activated carbon under spe-
cific conditions of temperature, carbon concentration, composition, and agi-
tation. To avoid issues associated with biomass inconsistency, acetic acid
was added synthetically at predetermined levels. Through isotherm and
kinetic rate testing, key parameters were determined and applied to a math-
ematical model. This model allowed determination of the external mass
transfer coefficient and the effective diffusivity. These mass transfer para-
meters then permitted evaluation of the carbon treatment process (10).

Materials and Methods


Isotherm and Activated Carbon Kinetic Rate
For isotherm testing and activated carbon kinetic rate testing, activated
carbon manufacturers were contacted and asked to supply samples of car-
bon that may effectively remove acetic acid from a water-based medium.
Two carbon manufacturers, Calgon Carbon (Pittsburgh, PA) and Carbon
Resources (Oceanside, CA), responded and supplied samples. Each sample
was deemed by its respective manufacturer to be effective under the previ-
ously described process conditions. Calgon Carbon supplied two products:
BL pulverized and PWA powdered. Carbon Resources provided four car-
bon products of varying origin. CRWCA, 2SAP, CR325BP, and CRCNSP-60.
A summary of commercially available activated carbons, including those
tested, is given in Table 1.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Agitation Effect of Acetic Acid Removal 355
Table 1
Classifications of Commercially Available Carbons
Carbon source Commercial example
Bituminous coal Calgon BL and PWA, Carbon Resources CR325BP
Coconut shell PICA USA G216, Carbon Resources CRCNSP-60
Wood Darco KB, Carbon Resources CRCWA and 2SAP
Lignite Darco S

An Innova Model 4230 benchtop refrigerated incubator I shaker from


New Brunswick Scientific (Edison, NJ) was used for the adsorption iso-
therm experiments. The incubator has stainless steel shelves and an Erlen-
meyer flask platform capable of holding twenty-five 250-mL flasks.
Microprocessor controls maintain speeds ranging between 25 and 400 rpm
and temperatures from 20°C below ambient temperature to 80°C.
For each carbon and condition tested, seven 250-mL Erlenmeyer flasks
were cleaned and dried. One hundred milliliters of the isotherm medium
consisting of 10 giL of acetic acid in deionized water was added to each
flask. The predetermined activated carbon amount was then added to each
flask: 10,20,40, 60, 80, or 100 giL. The flasks were covered with Parafilm
to prevent evaporation. The flasks and two different controls (medium
without carbon and water with carbon) were placed in the refrigerated
incubator I shaker set at a controlled agitation rate and temperature
(200 rpm, 15°C). The adsorption isotherm study was allowed to proceed for
24 h, at which point the flasks were removed from the shaker. The carbon
was allowed to settle for a short period of time and the medium was then
filtered with a 0.2-!!m syringe filter to remove the remaining carbon fines.
The isotherm samples were then titrated with standard alkali to determine
acetic acid content.
Adsorption Kinetics
A 4-L reaction kettle was used for the adsorption kinetic testing. The
reactor consisted of a glass housing and one stainless steel R100 impeller
held on a central stainless steel agitation shaft. A Lightnin mixer with digi-
tal-rotations-per-minute indicator provides the agitation. Samples were
withdrawn through a port in the vessel cover. A diagram of this apparatus
is provided in Fig. l.
With the impeller operating at the desired rate and the vessel at the
desired temperature, the selected activated carbon was carefully intro-
duced into the medium through a port in the reaction vessel cover at time
zero. Samples were removed at predetermined time intervals (2 min)
though the same port and were immediately filtered with a 0.2-!!m syringe
filter to remove the activated carbon and to stop the adsorption process.
The samples were then titrated with standard alkali to determine the acetic
acid concentration. These steps were repeated for each tested condition.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


356 Priddy and Hanley

Agitator Drive and Shaft ----.

Carbon Introduction Port--~---, ~-- Sampling Port

~.--- Glass Kettle Lid

Flange----t;=~~====1.t=====:::=~

Glass 4L Kettle---+---+ 40cm


RIOO Impeller -----1--~

IDcm

I4cm

Fig. 1. Diagram of reaction kettle.'

Model
To describe the adsorption rate data, a mathematical model must be
developed. This model incorporates a method for determining the two
major factors affecting adsorption: external mass transfer and diffusivity.
These two factors play an important role in the design and scale-up of
adsorption equipment.
To supply information for the model, data are collected at predeter-
mined time intervals during the kinetic portion of experimentation as pre-
viously described. The initial fraction of the plot obtained is then interpreted
using a four-step mechanism. In this mechanism, only the initial data, <10
min, are used owing to the presence of a linear isotherm during this time
period. The four-step mechanism is as follows: (1) mass transfer of solute
through bulk solution, (2) mass transfer from bulk solute to particle sur-
face, (3) intraparticle diffusion, and (4) adsorption on an interior site. If a
solution is well mixed, no concentration gradients will exist within the bulk
solution and the bulk diffusion term may be ignored. It can also be assumed
that adsorption on an interior site is rapid with respect to the remaining two
steps (11). In the initial phases of adsorption, external mass transfer is the
controlling mechanism. As time progresses or as agitation rates increase,
external mass transfer plays a smaller role and intraparticle diffusion

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Agitation Effect of Acetic Acid Removal 357
begins to dominate the adsorption process. To calculate the external mass
transfer coefficient the following equations are used (11):
dCb - 3kjm(C b- Cs )
(1)
dt Rps(l-e)

(2)

Eq. 1 is the result of a differential mass balance on the carbon particle,


solved in Eq. 2 for the limiting case of a linear isotherm
_(C m C
KC s -
bO - b)

A linear isotherm has been validated for this system for contact times of
<10 min.
The effective diffusivity is then determined. Although diffusivity
terms include bulk, surface, and intraparticle diffusivities, these indi-
vidual parameters are difficult to determine. By calculating the effective
diffusivity, all of the individual diffusivities are lumped into a single
parameter that may be determined by observation of bulk fluid proper-
ties. A shell balance on the system for constant density and diffusivity is
found in Bird et al. (12). With no reaction, assuming no convection, and
diffusion occurring only radially in the carbon particle, this equation
reduces to Fick's law of diffusion, Eq. 3. The differential mass balance for
this system can be seen in Eq. 4. Equation 4 has the following initial con-
dition: at t = 0, Cb = C bo and C = 0 for 0 s r s R:

dC =D l~(r2 ac) (3)


dt e r2 ar ar

(4)

For an external mass transfer-controlled adsorption, Eq. 3 is subject to


the following conditions:
Initial condition:
t=O C=OforOsrsR

Boundary conditions:

t>O a;:
aC I r=O=O and Dea;:
ac I r=R=kj(Cb-Cs)

Equation 5 allows determination of effective diffusivity (13):


Cb 1+Bexp(-3l1+BJ-r)
C bo 1+B (5)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


358 Priddy and Hanley
For an intra particle diffusion-controlled adsorption, Eq. 3 is subject to
the following conditions:
Initial condition:
t=O C=OforOsrsR
Boundary conditions

t>O BC I r=O=O
a;: and C=Cbatr=R

This equation may be solved by a generalized Sturm-Liouville Inte-


gral transform, resulting in Eq. 6 (14):

~ = _B_ + 6B ~ exp (- s~'t)


(6)
Cbo B+l n=1(9+9B+BZS~)

in which the Eigenvalues are determined by Eq. 7:

tans n
3s n 2 (7)
3 + BSn
In all cases, the following assumptions are made: linear isotherm re-
gion, well-mixed bulk solution, sample times close to to, initially solute-free
carbon pores, spherical carbon particles, and constant effective diffusivity.
From these equations it can be seen that a plot of

In (CC b
bo
1)
1 + mK
vs t is linear with a slope of

_(1 +mK) 3mk f


mK Rp.(l- e)

thereby allowing kf to be calculated. In addition, since t and 't are related


through the dimensionless variable, a plot oh vs t should yield a straight line
with a slope equal to DJR2. The effective diffusivity may then be dete rmined.

Results and Discussion


Results of the adsorption isotherm experiments are presented in
Fig. 2. The data resulting from these experiments were fitted to a Freundlich
isotherm model. Using this isotherm model, three items must be investigated
to properly evaluate the activated carbons: isotherm slope, isotherm inter-
cept, and final achievable substrate concentration. By directly comparing
isotherms, the equilibrium effects of the activated carbon are revealed.
For the case of acetic acid adsorption, a smaller slope indicates a decreased
dependence on concentration and is therefore considered favorable. The
highest possible intercept is also desired. This value indicates the capacity of
the activated carbon. Finally, the lowest residual acetic acid concentration is

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Agitation Effect of Acetic Acid Removal 359

Fig. 2. Adsorption isotherm curve comparison for acetic acid at 15°C.

desired. When Calgon carbons and Carbon Resources carbons were com-
pared on the adsorption of acetic acid, two of the tested carbons, Calgon BL
and Carbon Resources 2SAP, performed the best. Each of these carbons had
isotherm slopes at least 1.5 times smaller than the nearest competing carbon
and isotherm intercepts 1.36 times larger than the other carbons. Both of
these carbons were also capable of removing >67% of the initial acetic acid
from solution at the 15°C reaction temperature. However, one aspect of
activated carbon usage-cost-eliminated the Carbon Resources carbon.
Therefore, utilizing the criteria mentioned previously (isotherm slope, iso-
therm intercept, and final substrate concentration) and cost factors, a single
activated carbon was selected for further testing: Calgon Carbon BL at a
usage rate of 80 giL. Although not presented in graphic form here, iso-
therms for the adsorption of glucose were also performed. Calgon Carbon
BL adsorbed the least amount of glucose from solution, approaching nearly
20% of the initial concentration. This amount of sugar is lost only in the pore
volume of the carbon and is not actually adsorbed to the carbon surface.
Even this level of sugar loss is considered unacceptable, future process
modifications may decrease this loss.
Following the selection of an activated carbon, adsorption kinetic tests
were performed using the 4-L reaction kettle shown in Fig. 1. The curves
resulting from these tests as shown in Fig. 3. Two important pieces of infor-
mation may be gleaned from this plot. First, the curves help determine the

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


360 Priddy and Hanley
10 ._------------------------------------,
50 rpm
~ 100 rpm
c· 9
o 150 rpm
', c
::I
l 8
200 rpm

.=
'0
250 rpm

'v
-<...
',c
...
...-<'"
o
.:
o
:c 5
e
Eu
g 4
U

3 +---------.--------.---------.------~
o 50 150 200
Time, min

Fig. 3. Acetic acid concentration as function of time and rpm at 15°C using Calgon
Carbon BL.

time required to reach equilibrium. Calgon Carbon suggested that the time
required to reach equilibrium for this system would be >24 h. However,
from these curves it can be seen that local equilibrium concentrations were
reached in <30 min for all agitation rates and in as little as 20 min for the
highest agitation rates. These fast equilibrium times may be attributed to
the affinity of this carbon for acetic acid. Figure 3 also indicates final local
equilibrium concentrations. Note that after a great amount of time passed
(>48 h), the same equilibrium was achieved, and that the equilibrium
recorded was a local, not a true, equilibrium. Clearly, the 50-rpm agitation
rate does not perform well, achieving half of the equilibrium concentration
of the remaining rates. This can be attributed to incomplete mixing of the
carbon particles since it was observed that a quantity of carbon remained
at the bottom of the reaction vessel at the end of the kinetic experiment.
Equilibrium concentrations also were observed to decrease with agitation
rate. This result was not originally expected but indicates that a controlling
mass transfer resistance exists for the lower agitation rates and shows that
the degree of agitation does indeed have an effect on the adsorption of
acetic acid by activated carbon.
Having completed the kinetic studies, we applied the developed
model to the collected data. The results of the model application are given
in Fig. 4, describing the external mass transfer coefficient, and in Fig. 5,
describing the effective diffusivity. The calculated values for the external

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Agitation Effect of Acetic Acid Removal 361

0,------------------------------------,
-0.5

-1

-1.5
~
~ -2
Ei
+

-.a
.-!
"r' -2.5
"";;'
U

.~~
-3
U
;:::;
-3.5
• 100 rpm
-4 • 150 rpm
x 200 rpm
-4.5
:I: 250 rpm
-5
0 2 4 6 8 10 12
Time,min

Fig. 4. Dimensionless external film coefficient term as function of time and rpm at
15°C using Calgon Carbon BL.

1.2
• 50 rpm
• 100 rpm
1 • 150 rpm
x 200 rpm
GI 0.8 :Ie 250 rpm

!<1.1
!B

1
0.6

GI

~ 0.4

0.2

0
0 2 4 6 8 10 12
Time, min

Fig. 5. Dimensonless time as function of time and rpm at 15°C using Calgon Car-
bon BL.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


362 Priddy and Hanley
Table 2
Summary of Calculated Mass
Transfer Parameters at 15°C
rpm
50 3.577 X 10-8 8.625 x 10-13
100 4.996 X 10-8 1.275 X 10-12
150 2.056 X 10-7 5.925 X 10-12
200 2.642 X 10-7 9.170 X 10-12
250 NO' 1.049 X 10-11
"Not determined.

mass transfer coefficient and the effective diffusivity are summarized in


Table 2.
The overall trend that can be deduced from Fig. 4 is that the external
film coefficient increases with increasing agitation rate. The difference
between the 50- and 250-rpm agitation rate is nearly 640% at 15°e. Note
that these studies uncovered an interesting occurrence in the mathemati-
cal model used to calculate the external mass transfer resistance: as con-
centrations began to approach equilibrium, one of the terms in the
governing equation tended toward zero. A logarithm of this value then
becomes undefined. This situation is encountered early in the 250-rpm
agitation rate for 15°C, because at this rate equilibrium concentrations are
approached in under the 10-minute process time and indicate the disap-
pearance of any external mass transfer resistance. The end effect of these
external mass transfer resistances is reflected in the effect of agitation on
kinetic rate. As agitation increases, the external mass transfer coefficient
also increases, indicating a decrease in resistance. This decrease in mass
transfer resistance leads to the lower equilibrium concentrations seen in
the 200- and 250-rpm agitat~on rates.
Based on the results of the external mass transfer coefficient experi-
ments, both external mass transfer and intraparticle diffusion resistance
modes are in effect at different periods during experimentation. When
an external mass transfer coefficient can be reliably calculated, the model
equations governing diffusion limited by external mass transfer are
employed. This is the case for 50, 100, ISO, and 200 rpm at 15°e. When
an external mass transfer coefficient cannot be calculated because of the
elimination of the external mass transfer resistance, intraparticle diffusion
limits. This occurred in the 250 rpm, 15°C case. The overall trend that can
be deduced from Fig. 5 is that diffusivity increases with increasing agita-
tion rate. The difference between the 50 and the 250-rpm rate is nearly
1300% at15°e.
Overall, the process conditions of 15°C and 250 rpm using the Calgon
Carbon BL provided the highest level of acetic acid adsorption with the
highest effective diffusivity, the fastest mixing time, and in an environment

Applied Biochemistry and Biotechnology Vol. 105-10B,2003


Agitation Effect of Acetic Acid Removal 363
free from external mass transfer resistance. All of the phases of experimen-
tation can be combined to show that the final equilibrium concentration of
the 2S0-rpm kinetic test was reflected in the final equilibrium concentra-
tions found during adsorption isotherm testing. It can also be concluded
that the linear models presented previously performed well in predicting
the external mass transfer coefficient and the effective diffusivity. Both
Figs. 4 and S show a high degree of correlation with the data and the applied
linear model. However, all results found during our study are limited to
the reactor size used during testing, and the data collected are not easily
extrapolated to other systems. Scale-up issues for the pretreatment detoxi-
fication system were not investigated and are the subject of future research.
Clearly, if this process were to be implemented, mixing and reactor char-
acteristics would need to be studied.
Finally, the possibility of a combination detoxification and fermenta-
tion was not explored. Activated carbon, when added directly to the fer-
mentation broth, could adsorb the detrimental acetic acid. This method
could also reduce the amount of sugar removed from the system since the
glucose was lost in the pore volume of the carbon. Glucose utilization by the
fermenting organism would allow sugar trapped in this pore volume to
diffuse back into the bulk solution for consumption. Additionally, while the
temperature selected for this study was lSoC, a 2SOC study was also per-
formed but not presented here. Based on recommendations from the carbon
manufacturers, the favorable results seen here suggest that an investigation
into higher temperatures may be beneficial. Carbon treatment at the higher
temperatures would reduce energy requirements for cooling the hydroly-
sate for detoxification and then heating it again for fermentation.

Conclusions
The results of the adsorption kinetic experiments are valid for 10 giL
acetic acid solutions, agitated by a Rushton turbine impeller at SO, 100, 150,
200, and 2S0 rpm in a 4-L glass reaction vessel, and contacted with 80 giL
of the powdered activated carbon Calgon BL at lS°C. Equilibrium was
achieved 24 times faster than the time recommended by the manufacturer
of the activated carbon for this tested system. For lS0, 200, and 2S0 rpm,
>62% of the initial acetic acid was adsorbed. The presented model describ-
ing a linear approach to the external film resistance is supported with
coefficients of correlation ranging from 0.93 to 0.99, and those describing a
linear approach to the effective diffusivity are supported with coefficients
of correlation ranging from 0.84 to 0.99.

Acknowledgments
We are grateful to Judy Grady of Jim Beam Brands and Tom Effler of
Brown-Forman for their help in the sample analysis. This work was funded
by the US Department of Energy, Office of Fuels Development with the
National Renewable Energy Laboratory subcontract XCO-1-31016-01.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


364 Priddy and Hanley
Nomenclature
B capacity ratio = 1I mK
Cb bulk reservoir concentration (giL)
Cbo = initial solute concentration (giL)
De = effective diffusivity (m2 /s)
kf = external mass transfer resistance (m/s)
K = linear isotherm constant
m = carbon dosage (giL)
R = radius of a particle (m)
t = time (s)
E = porosity
Ps = adsorbent solid density (kg/m3)
T = dimensionless time parameter = DJ R2

References
1. Nguyen, Q. A, Tucker, M. P., Keller, F. A, Beaty, D. A., Connors, K M., and Eddy,
F. P. (1999), Appl. Biochem. Biotechnol. 77-79, 133-142.
2. Larsson, S., Palmqvist, E., Hahn-Hagerdal, B., Terrborg, c., Stanberg, K, Zacchi, G.,
and Nilvebrant, N. O. (1999), Enzyme Microbiol. Technol. 24, 151-159.
3. Ranatunga, T. D., Jervis, J., Helm, R. F., McMillan, J. D., and Hatzis, C. (1997), Appl.
Biochem. Biotechnol. 67,185-198.
4. Larsson, S., Reimann, A, Nilvebrant, N. 0., and Jonsson, L. J. (1999), Appl. Biochem.
Biotechnol. 77-79,91-103.
5. Lee, W. G., Lee,J. S.,Shin,C. S.,Park, S. C.,Chang,H. N., and Chaik, Y.K. (1999),Appl.
Biochem. Biotechnol. 77-79,547-559.
6. Rivard, C. J., Engel, R. E., Hayard, T. K, Nagle, N. J., Hatzis, c., and Philippidis, G.
P. (1996), Appl. Biochem. Biotechnol. 57-58,183-191.
7. Jonsson, L. J., Palmqvist, E., Nilvebrant, N. 0., and Hahn-Hagerdal, B. (1998), Appl.
Microbiol. Biotechnol. 49, 691-697.
8. Hassler, J. W. (1974), Purification with Activated Carbon, Chemical Publishing Com-
pany, New York, NY.
9. Morresi, A. C. and Cheremisinoff, P. N. (1978), in Carbon Adsorption Handbook,
Cheremisinoff, P. Nand Ellerbusch, F., eds., Ann Arbor Science, Ann Arbor, MI,
pp.I-54.
10. Mathews, A P. and Weber, W. J. (1977), AIChE Symp. Ser. 16673,91-98.
11. McKay, G. (1983), J. Chem. Technol. Biotechnol. 33A,205-218.
12. Bird, R. B., Stewart, W. E., and Lighfoot, E. N. (1960), Transport Phenomena, John Wiley
& Sons, New York, NY.
13. Suzuki, M. (1990), Adsorption Engineering, Elsevier Science, New York, NY.
14. Crank, J. (1956), The Mathematics of Diffusion, Oxford University Press, London, UK

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0365/$20.00

Enzymatic Digestibility
of Used Newspaper Treated with Aqueous
Ammonia-Hydrogen Peroxide Solution

SUNG RAE KIM* AND NAM Kvu MOON


Division of Applied Chemical Engineering and ERI,
Gyeongsang National University, Jinju 660-701 Korea,
E-mail: sb_kim@nongae.gsnu.ac.kr

Abstract
Wastepaper constitutes approximately half of municipal solid waste,
making it a potential source of bioenergy. Newspaper was pretreated with
an ammonia-hydrogen peroxide (H20 2) mixture in a shaking bath from room
temperature to 80 a C, and then its enzymatic digestibility was measured. A
significant amount of ink was removed from the newspaper slurry by the
reciprocating movement of the shaking bath. In addition, the ammonia-H 20 2
significantly swelled the substrate, thereby greatly increasing its susceptibil-
ity to enzymatic digestion. After pretreating the newspaper with conditions
of 40 a C, 3 h, 130 strokes/min, and 4 wt% ammonia-2 wt% H 20 2, the enzy-
matic digestibility was almost 90% of theoretical, or about 25% higher than
that of untreated substrate. Digestibility was also investigated as a function
of ammonia concentration, H 20 2 concentration, shaking speed, pretreatment
temperature, and time.
Index Entries: Pretreatment; newspaper; ammonia; hydrogen peroxide;
enzymatic digestibility.

Introduction
Lignocellulosic materials have considerable promise as a source of
future energy. Wastepaper is a plentiful and low-cost feedstock for making
bioethanol (1). Typically, wastepaper constitutes half of municipal solid
waste, and newspaper alone 14% of the waste (2). In the past, most of this
material was used only once and then was landfilled or incinerated. Even
recycled wastepaper can be used only two to three times before the fibers
become unacceptably short.
Lignocellulosic biomass generally resists enzymatic hydrolysis; thus,
effective pretreatment is an essential prerequisite to enhance the digestibil-
*Author to whom all correspondence and reprint requests should be addressed.
Applied Biochemistry and Biotechnology 365 Vol. 105-108,2003
366 Kim and Moon
ity of lignocellulosic residues. Of a number of studies on the pretreatment
of lignocellulosic materials, there have been limited studies on wastepaper
pretreatment using electron beam irradiation (3), carbon dioxide (4), steam
explosion (5), ammonia fiber explosion (6), and ammonia-hydrogen perox-
ide (H20 2) in a percolation reactor system (7). The pretreatment methods
used in most of these studies, however, were the same as those used in
woody and herbaceous materials. Because paper has already had consid-
erable chemical and/or physical treatment, wastepaper does not require
the extensive pretreatment developed for woody and herbaceous materi-
als. Another difference between wastepaper and raw lignocellulosic mate-
rial is ink and fillers added in the paper-making process. These chemicals
could potentially interfere with the enzymatic hydrolysis of wastepaper.
H 20 2 in aqueous ammonia solution is very effective for pretreating
lignocellulosic materials (8,9). It also has been used to pretreat wastepa-
per (6,7). However, the reported methods appear to be uneconomical
because of high chemical costs. To increase enzymatic digestibility with-
out severe pretreatment, a pretreatment process was devised that can
remove ink and swell the substrate easily on a shaking bath. In our pre-
vious study (10), this process proved to be extremely effective at remov-
ing ink from newspaper. The main purpose of the present study was to
investigate parameters that affect enzymatic digestibility when treating
newspaper in an ammonia-H20 2 mixture on a shaking bath. The pretreat-
ment temperatures used were <80 a C, which are very mild compared to
conventional pretreatment conditions.

Material and Methods


Newspaper and Chemicals
A mixture of four newspapers issued in Korea was used as substrate.
Newspaper was cut into approx 1 x 1 cm pieces. The moisture of this
paper was 7.9% with the following composition: 61.3 wt% glucan, 9.8 wt%
xylan + mannan + galactan (XMG), 12.0 wt% klason lignin, 5.7 wt% ash,
and 11.3 wt% unaccounted for components. Commercial cellulase and
j3-glucosidase (Novo Nordisk, Bagvard, Denmark) were used. A mixture
of Celluclast (80 IV or international filter paper units [IFPV]/mL) and
Novozym 188 (792 cellobiase units [CBV]/mL) was used with a ratio of
4 IV of Celluclast/CBV Novozym to alleviate end-product inhibition by
cellobiose.
Pretreatment
Pretreatment was performed on a reciprocating shaking water bath.
Ten grams of substrate was added to a 1-L autoclave bottle with 200 g of
ammonia or ammonia-H20 2 solution. The concentration of each compo-
nent was expressed as wt% based on the total amount of ammonia-H20 2
solution. Then the bottle was placed for 1-3 h on a shaker operating at 20-
80 a C and 70-130 strokes/min. After pretreatment, the wet solid was

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Treatment of Used Newspaper with Ammonia-H2 0 2 367
washed with deionized water and separated into two portions. One was
oven dried at 105°C overnight to measure weight loss and was further
subjected to composition analysis. The other was used in the enzymatic
digestibility test.
Digestibility Test
Enzymatic digestibility of pretreated substrate was performed accord-
ing to National Renewable Energy Laboratory (NREL) standard procedure
no. 009 (11). The amount of solid required to give 0.5 g of glucan in 50 mL
was added to a 250-mL flask. The buffer solution was 0.05 M citrate, pH 4.8,
and the cellulase enzyme loading was 60 IFPU / g of glucan. The contents of
the flask were preheated to 50°C before the enzyme was added. The flask
was placed on a shaking bath operating at 50°C and 90 strokes/ min. Using
the same method, untreated substrate was placed in the bath as a control.
Samples were taken periodically and analyzed for glucose using high-per-
formance liquid chromatography (HPLC). The glucose content after 72 h of
hydrolysis was used to calculate enzymatic digestibility.
Analytical Methods
The solid biomass sample was analy~ed for moisture, sugars, klason
lignin, and ash by NREL standard procedures (no. 001-005) (11). Sugars
were measured by HPLC (Thermo Separation Products) using a Bio-Rad
HPX-87C column (conditions; 0.6 mL/min, 85°C, water). Because this col-
umn does not resolve xylose, mannose, and galactose, the combined value
of XMG is used in this article.

Results and Discussion


Newspaper is mostly derived from softwood and exhibits low enzy-
matic digestibility because of its high lignin content and dense structure.
Additionally, chemicals such as fillers, ink, and other additives make it
difficult to hydrolyze enzymatically. Our previous study (10) investigated
factors that affect newspaper hydrolysis, such as ash content, substrate
size, and ink. We found that ink had a Significant effect on digestibility,
whereas ash content and substrate size had a very small effect. The basic
components of most water-based newsprint inks-constituting 1 to 2 wt%
of newspaper-are carbon black pigment; acrylic resin binder; and various
additives such as defoamers, plasticizers, and lubricants (13). Most conven-
tional de-inking of wastepaper is done with alkali and other chemicals.
Alkali has two purposes: (1) to remove rosin sizing from the paper and (2)
to saponify the ink vehicle and release the pigment in the ink (14). The ink
particles released from the fiber surface are removed from the slurry by
either washing or flotation.
Ammonia is a proven alkaline reagent for pretreating lignocellulosic
materials. Previous studies (8,9) showed that ammonia alone was not effec-
tive in pretreating lignocellulosic materials; thus, H 20 2 was added to the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
368 Kim and Moon
Table 1
Effect of H 20 2 and Ammonia Concentration
on the Composition of Solid Residue of Newspaper"
Solids Glucan XMG Klason Ash
Pretreatment remaining (%) (%) (%) lignin(%) (%)
Untreated 100.0 61.3 9.8 12.0 5.7
4 wt% Ammonia 95.4 54.7 8.7 11.5 5.1
4 wt% Ammonia 94.7 55.5 8.1 11.5 4.9
+2wt%H20 2
8 wt% Ammonia 95.0 53.9 8.6 11.2 5.0
8 wt% Ammonia 93.3 55.3 8.1 11.5 5.0
+2 wt%H20 2
"All sugar contents are based on the original oven-dried untreated biomass and ex-
pressed as glucan, xylan, mannan, and galactan equivalents. Pretreatment condition: 3 h,
40°C, 130 strokes/min; mass ratio of solid to liquid = 1/20.

ammonia solution as an oxidant. H 20 2 in alkaline solution promotes rapid


oxidative depolymerization of lignin. In addition, H 20 2 can aid de-inking
operations (14).
The appearance of the ammonia- H 20 2-treated sample in the present
experiment was very different from the ammonia-treated sample. The
volume of ammonia- H 20 2-treated sample was about 1.5 times bigger than
that of the ammonia-treated sample. Furthermore, a significant amount of
dark-colored ink components, which separated from cellulose fibers, was
observed in the upper portion of the bottle. In the case of the ammonia-
treated sample, only a very small amount of ink components was observed.
Therefore, H 20 2 is very effective in swelling fibers and removing ink com-
ponents from fibers.
Table 1 shows the effect of H 20 2 and ammonia concentration on the
composition of the solid residue of newspaper. The percentage of solid
remaining was 93.3-95.4%, which indicates that only a small portion of
newspaper was solubilized by H 20 2 and/ or ammonia concentration. The
percentage of glucan, XMG, klason lignin, and ash were almost the same,
regardless of the different pretreatment conditions. However, the digest-
ibility for each condition was significantly different (Fig. 1). The digest-
ibilities of substrates treated with 4 and 8 wt% ammonia were the same as
the untreated ones, which was almost 25% less than those treated with
ammonia-H20 2• (Note that here, untreated sample means a substrate that
is soaked in water for 3 h at room temperature.) Such an increase in digest-
ibility, despite no significant change in composition after pretreatment,
probably depends on ink removal and substrate swelling, not the removal
of individual component. However, in hardwood, digestibility usually
depends on removing hemicellulose and/ or lignin (8).
Another consideration in this pretreatment is the physical shock pro-
vided by the shaking bath. In a conventional de-inking process, ink compo-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Treatment of Used Newspaper with Ammonia-H2 0 2 369
100.-----------------------------------------,

___ untreated
-6- 4 wt% ammonia
---6.
4 wt% ammonla+2 wt% HP2
___ 8 wt% ammonia

o 20 40 60 80
Time (h)

Fig. 1. Effect of H 20 2 in ammonia solution on enzymatic digestibility (pretreatment


conditions: 3h, 40°C, 130 strokes/min).

nents must be removed from fibers by either washing or flotation; other-


wise ink is left in the space between fibers and interferes with enzymatic
hydrolysis. Figure 2 shows the effect of shaking speed on digestibility. As
the speed increased, digestibility increased. This can be explained from
experimental observation; as the speed increased, a greater amount of ink
components was observed in the upper portion of the bottle and the sub-
strate was more swollen. This means that the physical shock provided by
the shaking bath gives an effect equivalent to flotation and increases the
contact between substrate and ammonia- either. Experiments above 130
strokes/min were not done because of equipment limitations.
Figure3 shows the effect of ammonia concentration on enzymatic di-
gestibility at 2 wt% H 20 2• The digestibilities of substrates treated with >4
wt% were almost the same, so 4 wt% ammonia concentration was enough
for pretreatment.
H 20 2 is almost 10 times more expensive than ammonia, so the H 20 2
concentration was decreased to 0.5 and 1 wt%. As shown in Fig. 4, the
digestibility decreased as the H 20 2 concentration decreased; therefore, the
H 20 2 concentration increased to 3 wt%, but the digestibility was marginally
increased. The optimum H 20 2 concentration seems to be 2 wt%.
The effect of pretreatment temperature was investigated using 4 wt%
ammonia and a mixture of 4 wt% ammonia and 2 wt% H 20 2 (Fig. 5). When
only ammonia was used, digestibility increased from 68% to 78% as tem-
perature increased from 20 to 80 a C. But when the substrate was treated
with ammonia-H20 2, the digestibility only slightly increased as the tem-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


370 Kim and Moon
l00r-------------------------------------------~

80

____ untreated
-0- 70 strokeslmin
-T- 100 strokeslmin
20
--v- 130 strokeslmin

o 24 48 72
Time (h)

Fig. 2. Effect of shaking speed on enzymatic digestibility (pretreatment conditions:


3h,40°C,4 wt% ammonia + H 20 2).

100~--------------------------------------------~

80 ---~ ~ ....,::;:~~
.
................................ .
................................


·······0·······
untreated
2 wt% ammonia
---T"-- 4 wt% ammonia
_ .. -v-.. - 6 wt% ammonia
20
- .... - 8 wt% ammonia
_.-0-.- 10 wt% ammonia

o 24 48 72
Time (h)

Fig. 3. Effect of ammonia concentration on enzymatic digestibility at 2 wt% H 20 2


concentration (pretreatment conditions: 3h, 40°C, 130 strokes/min).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Treatment of Used Newspaper with Ammonia-H2 0 2 371
100r------------------------------------------,

80


·······0·······
untreated

---T---
_ . -v--.. -
20
- .... -
o 24 48 72

Time (hour)

Fig. 4. Effect of H 20 2 concentration in 4 wt% ammonia solution on enzymatic digest-


ibility (pretreatment conditions: 3h, 40°C, 130 strokes/min).

100

90

80

;e
~
70

60

~
:c 50
~
:g,40
is
30

20 ..... 4 wt% ammonia


-0- 4 wt% ammonia+2 wt%HP2
10

0
20 40 60 80
Temperature(°C)

Fig. 5. Effect of pretreatment solution on enzymatic digestibility (pretreatment con-


ditions: 3h, 40°C, 130 strokes/min).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


372 Kim and Moon
100~------------------------------------------~

___ untreated
-0- 1h
--T- 2h
-V- 3h

48 72

Time (h)

Fig. 6. Effect of pretreatment time on enzymatic digestibility (pretreatment condi-


tions: 4 wt% ammonia + 2 wt% H 20 2 , 40°C, 130 strokes/min).

perature increased from 20 to 40°C. Further increases in temperature did


not affect digestibility.
Figure 6 shows the effect of pretreatmenttime on digestibility. As time
increased, digestibility increased. As can be seen at least 2 h were required
to pretreat newspaper.

Acknowledgments
We gratefully acknowledge the financial support provided for this
work by the Ministry of Commerce, Industry and Energy, Korea.

References
1. Wyman, C. E. (1994), Bioresour. Technol. 50,3-16.
2. Scott, C. D., Davison, B. H., Scott, T. c., Woodward, J., Dees, c., and Rothrock, D. S.
(1994), Appl. Biochem. Biotechnol. 45/46, 641-653.
3. Khan, A W., Labrie, J., and McKeown, J. (1987), Radiat. Phys. Chem. 29, 117-120.
4. Zheng, Y., Lin, H., and Tsao, G. T. (1998), Biotechnol. Prog. 14,890-896.
5. Cuna, D., Ricci, E., Alfani, F., Cantarella, M., D'Ercole, 1., Gallifuoco, A, Spera, A,
Zimbardi, F., and Viggiano, D. (1998), in Proceedings of International Conference of
Biomass for Energy and Industry, Wiirzburg, Germany, pp. 560-563.
6. Holtzapple, M. T., Lundeen, J. E., Sturgis, R., Lewis, J. E., and Dale, B. E. (1992), Appl.
Biochem. Biotechnol. 34135, 5-21.
7. Kim, J. S., Lee, Y. Y., and Park, S. C. (2000), Appl. Biochem. Biotechnol. 84-86, 129-139.
8. Kim, S. B., Urn, B. H., and Park, S. C. (2001), Appl. Biochem. Biotechnol. 91-93,81-94.
9. Kim, S. B. and Lee, Y. Y. (1996), Appl. Biochem. Biotechnol. 57/58,147-156.
10. Moon, N. K. and Kim, S. B. (2001), Korean J. Biotechnol. Bioeng. 16,446-451.

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Treatment of Used Newspaper with Ammonia-H2 0 2 373
11. (1996), NREL Chemical Analysis and Testing Standard Procedure, No. 009, National
Renewable Energy Laboratory, Golden, CO.
12. (1996), Chemical Analysis and Testing Standard Procedures, No. 001-005, National
Renewable Energy Laboratory, Golden, CO.
13. Fernandez, E. O. and K. T. Hodgson (1996), J. Pulp Paper Sci. 22(11),452-456.
14. Felton, A. J. (1980), in Pulp and Paper: Chemistry and Chemical Technology, Casey, J. P.,
ed., Wiley, pp. 568-594.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0375/$20.00

Continuous Production of Butanol


from Starch-Based Packing Peanuts

THADDEUS C. EZEJI, MARISA GROBERG,


NASIR QURESHI, * AND HANS P. BLASCHEK
University of Illinois, Biotechnology and Bioengineering Group,
Department of Food Science and Human Nutrition,
1207 W. Gregory Drive, Urbana, IL 61801, E-mail: nqureshi@uiuc.edu

Abstract
Acetone, butanol, ethanol (ABE, or solvents) were produced from
starch-based packing peanuts in batch and continuous reactors. In a batch
reactor, 18.9 giL of total ABE was produced from 80 giL packing peanuts
in 110 h of fermentation. The initial and final starch concentrations were
69.6 and 11.1 giL, respectively. In this fermentation, ABE yield and pro-
ductivity of 0.32 and 0.17 g/(L·h) were obtained, respectively. Compared
to the batch fermentation, continuous fermentation of 40 giL of starch-
based packing peanuts in P2 medium resulted in a maximum solvent pro-
duction of 8.4 giL at a dilution rate of 0.033 h-1• This resulted in a
productivity of 0.27 gl (L·h). However, the reactor was not stable and fer-
mentation deteriorated with time. Continuous fermentation of 35 giL of
starch solution resulted in a similar performance. These studies were per-
formed in a vertical column reactor using Clostridium beijerinckii BA101 and
P2 medium. It is anticipated that prolonged exposure of culture to
acrylamide, which is formed during boiling I autoclaving of starch, affects
the fermentation negatively.
Index Entries: Starch-based packing peanuts; continuous fermentation;
butanol; Clostridium beijerinckii BAlOl; reactor; dilution rate.

Introduction

As a result of increasing oil prices (two to four times higher than in


July-August 2000), and constant instability in the oilsupplyregion(Middle
East), various bioconversion programs have been initiated to produce
biochemicals and bioenergy in the United States. To hasten research and
development and find a solution to this crisis, President WilliamJ. Clinton
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 375 Vol. 105-108,2003


376 Ezeji et al.
issued Executive Order 13134, entitled "Developing and Promoting
Biobased Products and Bioenergy" (64 FR 44639, August 16, 1999). A pri-
mary goal of the order is to triple the nation's use of biobased products
and bioenergy by 2010, generating as much as $20 billion annually in new
income for farmers and rural communities.
As a result of these initiatives, we intensified our research on butanol
fermentation. Butanol, which is an excellent biofuel, has numerous other
applications in food, plastics, and chemical industries (1) and has been
studied by numerous investigators (2-6). In a recent economic study, we
found that fermentation substrate is one of the most important factors
that influence the price of butanol (7). For that reason, waste substrates
must be used for conversion to fuels and chemicals such as butanol.
Starch-based waste packing peanuts, which were developed to biode-
grade after use, are one such substrate and can be used for this bioconver-
sion process. Successful bioconversion of such wastes would not only
convert these waste substrates to useful chemicals, but also solves waste
disposal and environmental problems faced by the relevant industries.
Clostridium beijerinckii BA101, which was developed a few years ago, is
capable of efficient starch hydrolysis in addition to superior production
of solvents.
Fermentation studies using such wastes can be performed in either
batch or continuous reactors. Since continuous reactors offer comparatively
high reactor productivities, we chose to produce acetone, butanol, ethanol
(ABE, or solvents) in these reactors. Hence, our objective was to produce
ABE from starch-based packing peanuts in continuous reactors and com-
pare reactor productivity with that of batch reactors.

Materials and Methods


Microorganism
A stock culture of C. beijerinckii BA101 was maintained as a spore
suspension in distilled water at 4°C. Spores were heat shocked in cooked
meat medium (Difco, Detroit, MI) containing 30 giL of glucose at 80°C for
10 min. The culture was found to be growing actively within 16-18 h. This
was followed by transferring 5 mL of actively growing culture to 100 mL of
tryptone-glucose-yeast extract (TGY) medium (8). Cells were grown
anaerobically for 16-18 h at 34°C before being transferred into solvent pro-
duction medium containing starch and I or packing peanuts.
Preparation of Medium
Continuous butanol production was studied from starch or starch-
based packing peanuts (average volume of 67.57 cm3 / g, starch content of
88.4 wt%, average diameter of 1.54 cm, length of 2 to 3 cm; Storopack,
Cleveland, OH). To prepare the medium for continuous fermentation, 35-
40 giL of starch or packing peanuts and 1 giL of yeast extract (Difco) were
dissolved in distilled water followed by sterilizing at 121°C for 15 min. The
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Production of Butanol from Waste Substrate 377

medium was cooled by sweeping 02-free N2 gas across the surface. On


cooling to room temperature, filter-sterilized P2 medium stock solutions
(9) were added (10 mL of each of these solutions to 970 mL of autoclaved
medium containing substrate and yeast extract):
1. Buffer:50g/LofKH2P04,50g/LofK2HP04,220g/Lofammonium
acetate.
2. Vitamin: 0.1 giL of paraaminobenzoic acid, 0.1 giL of thiamin,
0.001 giL biotin.
3. Mineral: 20 giL of MgS04·7H20, 19/L of MnS04·H20, 19/L of
FeS04·7H20, 1 giL of NaCl.
The stock solutions were filter sterilized through a 0.2-llm-pore-size
polyethersulfone membrane filter unit (Nalgene, Rochester, NY). The
medium was kept at room temperature and stirred at <100 rpm to avoid
any sedimentation. The medium was kept in an airtight sealed bottle pro-
vided with an inlet and outlet for 02-free N2 gas to keep it anaerobic. The
gas was swept across the surface of the medium at all times. Starch or
packing peanut concentration was 35-40 giL to keep feed viscosity low
and to be able to pump the feed through the pump tube.
For batch fermentation, the medium was prepared in a 500 to 1000-mL
beaker followed by transferring to 150-mL screw-capped bottles (125 mL of
medium). The medium contained starch or packing peanuts at 60-80 giL,
yeast extract at 1 giL, and was autoclaved in bottles. On cooling, the stock
solutions were added to the bottles followed by transferring them to an
anaerobic chamber (Coy, Ann Arbor, MI) for 24 h for anaerobiosis. This was
followed by inoculation with actively growing TGY cell suspension. Fer-
mentation proceeded at 35°C until complete and 1-mL samples were col-
lected intermittently. The samples, from both the bottles and the reactor,
were centrifuged at 14,000 rpm in Eppendorf centrifuge tubes before inject-
ing the supernatant into the gas chromatograph for solvent analysis.
Note that pH was not controlled in any of the fermentations. However,
in the case of batch fermentation (starch or packing peanuts), it was ad-
justed automatically to approx 6.8 by the addition of buffer solution. Such
a high pH favors quick growth and acid production, which lowers pH to
approx 5.0 and triggers solvent production. In addition to temperature and
pH control, maintaining anaerobic conditions in the incubator is a prereq-
uisite to a successful butanol fermentation using C. beijerinckii BAlOl.
Continuous Reactor
The reactor was composed of a 312-mL total volume jacketed poly-
acrylic column (192 x 46 mm) and was used in vertical mode. The reactor
was sterilized with 30% (v Iv) ethanol solution for 48-72 h, after which it
was drained and washed thoroughly with sterilized deionized water.
02-free N2 gas was passed through the column overnight to ensure anaero-
bic conditions inside the reactor. Twenty milliliters of actively growing
C. beijerinckii BA101 cells from a TGYbottle was inoculated into the reactor,

Applied Biochemistry and Biotechnology Vol. 705-708,2003


378 Ezeji et al.

and the reactor was filled with the starch/packing peanut-based fresh P2
medium described earlier. Cell growth was allowed inside the reactor for
6 h, after which the P2 medium was continuously fed using a peristaltic
pump (Cole-Parmer, Veron Hills, IL) and silicone tubing. The reactor was
fed at the bottom, thus getting effluent at the top. Since it was a free-cell
reactor, feed rate was kept low for solventogenesis (Dilution rate [D, h-1]
< !! [specific growth rate constant, h-1] and v [specific rate of solvent produc-
tion, h-l]). Water at 35°C was circulated through the jacket of the column to
control temperature using a circulating water bath (Polystat; Cole-Parmer).
Analytical Procedures
Because of the opaque nature of starch and of peanut solutions, cell
concentration measurement by optical density was not possible. ABE and
acids (acetic and butyric) were measured using a 6890 Hewlett-Packard
gas chromatograph (Hewlett-Packard, Avondale, PA) equipped with a
flame ionization detector and 6 ft x 2 mm glass column (10% CW-20M,
0.01% H 3P04, support 80/100 Chromosorb WAW). Batch fermentation
productivity was calculated as total ABE concentration (g/L) divided by
fermentation time (h). Fermentation time was defined as the time period
when a maximum ABE concentration was reached. ABE yield, which
does not have a unit, was calculated as total ABE produced (g) divided by
total carbohydrate (as starch) utilized (g). In the continuous fermentation,
productivity was calculated as ABE concentration (g/L) multiplied by
dilution rate (h-l). Dilution rate is defined as feed flow rate (mL/h) per
reactor volume (mL).
Glucose concentration was determined using a hexokinase and glu-
cose-6-phosphate dehydrogenase (Sigma, St. Louis, MO) coupled enzy-
matic assay. To measure glucose, the fermentation broth was centrifuged
(microfuge centrifuge) at 16,OOOg for 3 min at 4°C. A portion of the super-
natant (10 !!L) was mixed with glucose (HK 20) reagent (1.0 mL) and incu-
bated at room temperature for 5 min. Standard solutions of anhydrous
D-glucose containing 1-5 mg/mL of glucose in distilled water were pre-
pared. Ten microliters of each of the standard solutions was mixed with
glucose (HK 20) reagent (1.0 mL) and incubated at room temperature for
5 min. A blank (deionized water) (10 !!L) was incubated with the reagent
and used for zero adjustment of the spectrophotometer. After 5 min, the
absorbance was measured at 340 nm using a Beckman Du 640 spectropho-
tometer, and the glucose content in the sample was computed by least
squares linear regression using a standard curve.
Starch concentration of the sample was determined using a modified
method of Holm et al. (10). A sample (250 mg) was suspended in distilled
water (15 mL) in a 50-mL beaker. Heat-stable a-amylase (100 !!L) (Sigma)
was added and mixed gently with a magnetic stirrer. The beaker was
placed in a boiling water bath for 30 min with mixing every 5 min. The
suspension was allowed to cool under continuous agitation on a magnetic
stirrer and was transferred to a 25-mL volumetric flask followed by filling
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Production of Butanol from Waste Substrate 379
it with water to the volume. One milliliter of the solution was transferred
to a test tube followed by the addition of 2.9 mL of 0.1 M sodium acetate
buffer (pH 4.75) and 100 t-tL of amyloglucosidase (Sigma). The mixture was
incubated for 60 min at 55°C with careful mixing every 5 min. The sample
was then transferred to a 50-mL volumetric flask and filled to volume with
distilled water. Ten microliters of the solution was assayed for glucose
according to the hexokinase and glucose-6-phosphate dehydrogenase
assay method as described earlier.
% starch = (mg glucose x 25 a x 50 a x 0.9 b x 100)/( Sample weight [250 mg])
in which a is the dilution factor and b is the correction glucose ~ glucan.
At the time of these studies, it was not possible to analyze starch-based
packing peanuts for all of their constituent components.

Results and Discussion


To evaluate fermentability of packing peanuts, a batch fermentation
was run with 80 giL of packing peanuts using C. beijerinckii BA101 in P2
medium. The initial and final concentrations of starch were 69.6 and 11.1
giL, respectively (11). This resulted in utilization of 58.5 giL of starch and
a solvent yield of 0.32. During 110 h of fermentation, 18.9 giL of ABE (losses
notincluded)wasproduced,resultinginaproductivityofO.17g/(L·h). The
concentration of acids was low at 0.2 giL at 110 h. The results proved that
packing peanuts were successfully utilized by C. beijerinckii BA101 to pro-
duce ABE. However, productivity was low at 0.17 g/(L·h).
To improve reactor productivity, a continuous reactor was run at a
flow rate of 10.2 mLlh (dilution rate of 0.033 h-l). For the first 121 h, fer-
mentation was vigorous but then slowed down. After a period of 48 h of
continuous operation, 6.7 giL of total ABE was obtained in the effluent,
and it continued to increase to 8.4 giL at 83 h (Table 1). At 121 h, ABE
concentration was 8.3 giL, with a maximum productivity of 0.27 gl (L·h).
After 121 h of operation, the fermentation started deteriorating constantly,
as shown in the Table 1. Fermentation was continued for 246 h, and at that
time 2.3 giL of total ABE and 5.7 giL of total acids were present in the
effluent. At that stage, significant amount of sediments had accumulated
in the bottom part of the reactor (up to 6.5 cm high). It was thought that
accumulated sediments containing undisclosed chemicals present in the
packing peanuts had affected the fermentation negatively. A second reac-
tor run under identical conditions produced similar results.
To check if fermentation efficiency was reduced owing to suspension,
reactor liquid was drained into a separating funnel and sediments were
separated. In addition, the volume of sediments was madeup to 312 mL
(equal to the liquid volume in the reactor) by adding distilled water. To this
18.7 g of glucose (60 giL) and 0.3 g of yeast extract (1 giL) were added
followed by autoclaving. On cooling, P2 medium stock solutions were
added. The mixture was fermented in 150-mL screw-capped bottles inocu-
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
380 Ezeji et a/.

Table 1
Continuous Production of ABE
from Starch-Based Packing Peanuts Using C. beijerinckii BAlOl
Products (giL) ABE
Time Acetic Butyric Total productivity
(h) Acetone Butanol Ethanol acid acid ABE (g/(L·h])
24 0.9 2.8 0.1 2.7 1.1 3.8 0.12
48 1.7 4.9 0.1 2.3 1.0 6.7 0.22
72 1.6 6.4 0.1 1.6 1.1 8.1 0.26
83 1.7 6.6 0.1 1.3 0.9 8.4 0.27
102 1.4 5.7 0.1 1.8 0.8 7.2 0.24
121 1.7 6.5 0.1 1.4 1.3 8.3 0.27
143 1.0 4.2 0.1 1.9 1.3 5.3 0.17
151 1.1 4.7 0.1 1.8 2.2 5.9 0.19
169 1.0 2.8 0.1 3.0 1.5 3.9 0.13
175 0.7 2.1 0.1 1.9 2.1 2.9 0.09
196 0.7 2.2 0.1 2.2 2.3 3.0 0.09
216 0.7 2.1 0.1 3.7 3.3 2.9 0.09
225 0.8 2.2 0.1 3.5 2.2 3.1 0.10
241 0.6 1.8 0.0 3.5 2.2 2.4 0.08
246 0.6 1.7 0.0 3.5 2.2 2.3 0.08

lated with freshly grown inoculum. In 96 h of fermentation, 14.5 giL of total


ABE and 5.2 giL of total acids were produced, suggesting that the sedi-
ments were not toxic to the fermentation in the reactor (Fig. 1).
To further strengthen these findings, a continuous reactor was oper-
ated with 35 giL of starch solution in feed and P2 medium ingredients. The
continuous reactor was operated under identical conditions (dilution rate of
0.033 h-l). The performance of the reactor was similar to that of the previous
reactors fed with packing peanuts solution. At 46 h of continuous operation,
8.1 giL of total ABE and 2.1 giL of total acids were produced (Table 2). For
103 h the reactor was productive and produced up to 8.5 giL of ABE, result-
ing in a productivity of 0.28 gl (L·h). Although this productivity is higher
than achieved in the batch reactor, the reactor was not steady. At 226 h,
3.7 giL of ABE and 5.5 giL of acids were present in the reactor effluent. In
this experiment, a similar type of sediment was also observed at the bottom
ofthe reactor. However, the previous batch experiment suggested that sedi-
ments were not toxic to the culture.
A recent study reported that formation of acrylamide results from
boiling of the starch solution (personal communication). It is well known
that acrylamide inhibits microbial cell growth and activity. Hence, it was
possible that acrylamide produced during boiling and autoc1aving of
the starch-based packing peanutslstarch solution was inhibitory to
C. beijerinckii BA101 cells. Possibly, prolonged exposure inhibited the cul-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Production of Butanol from Waste Substrate 381

20 A

-
15~~---+--~--+-~r-~
:J'
.9 • Acetone
~ 10 1---+---1:-........-+--.",-::=--+__--1 • Butanol

e
:l • Ethanol
* TotalABE
a.

0_:.......J.J,..............................___=.l...----l
o 20 40 60 80 100 120
Fermentation Time [h]

20
B
15

o Acetic acid
o Butyric acid

~
A Total acids

\.. ....
5 ..(]

no. ~

oV
~

o 20 40 60 80 100 120
Fermentation Time [h]

Fig. 1. Effect of packing peanut sediments on butanol fermentation using


C. beijerinckii BAlOI and P2 medium. (A) Solvents; (B) acids.

ture. At this stage, we conclude that batch fermentation of packing pea-


nuts was successful. However, continuous fermentation of starch/pack-
ing peanuts, which always had a high initial productivity, deteriorated
after approx 120 h of operation.

Acknowledgments
We gratefully acknowledge the kind gift of the packing peanuts from
Storopack, Cleveland, OH. This work was supported by grants from Illi-
nois Corn Marketing Board and Illinois Council on Food and Agricultural
Research (CFAR IDA CF 01E-35-1).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
382 Ezeji et a/.

Table 2
Continuous Production of ABE from 35 giL
of Starch Using C. beijerinckii BA101

Products (g/L) ABE


Time Acetic Butyric Total productivity Glucose
(h) Acetone Butanol Ethanol acid acid ABE (g/(L.h]) (g/L)"

24 0.8 2.4 0.1 2.5 1.4 3.3 0.11 1.26


31 0.9 2.9 0.1 2.2 1.3 3.9 0.13 0.89
46 2.5 5.4 0.2 1.7 0.4 8.1 0.26 0.03
51 2.5 5.7 0.2 1.2 0.3 8.4 0.27 0.07
70 2.3 5.7 0.2 1.6 0.8 8.2 0.27 0.04
77 2.4 5.5 0.2 1.4 0.5 8.1 0.26 0.04
91 2.4 5.9 0.2 1.6 0.5 8.5 0.28 0.05
103 2.3 5.8 0.2 2.1 0.7 8.3 0.27 0.05
114 2.1 5.5 0.1 2.8 0.9 7.7 0.25 0.08
124 1.8 5.0 0.1 2.3 0.8 6.9 0.23 0.06
139 1.8 4.9 0.1 2.9 1.1 6.8 0.22 0.07
150 1.6 4.5 0.2 2.9 1.1 6.3 0.21 0.21
163 1.5 4.1 0.1 3.1 1.3 5.7 0.19 0.20
170 0.8 2.7 0.1 3.5 1.1 3.6 0.12 0.62
188 1.0 2.6 0.1 3.5 1.5 3.7 0.12 0.38
196 1.0 2.6 0.1 3.8 1.5 3.7 0.12 0.31
211 1.0 2.6 0.1 3.6 1.6 3.7 0.12 0.21
226 0.8 2.8 0.1 3.7 1.8 3.7 0.12 0.04
aln reactor effluent.

References
1. Qureshi, N. and Blaschek, H. P. (2001), J. Ind. Microbiol. Biotechnol. 27,292-297.
2. Davison, B. H. and Scott, C. D. (1988), Appl. Biochem. Biotechnol. 18, 19-34.
3. Davison, B. H. and Thompson, J. E. (1993), Appl. Biochem. Biotechnol. 39,415-425.
4. Friedl, A., Qureshi, N., and Maddox, I. S. (1991), Biotechnol. Bioeng. 38,518-527.
5. Ennis, B. M., Maddox, I. S., and Schoutens, G. H. (1986), NZJ Dairy Sci. Technol. 21,
99-109.
6. Chen, C. K. and Blaschek, H. P. (1999), Appl. Microbiol. Biotechnol. 52,170-173.
7. Qureshi, N. and Blaschek, H. P. (2000), Trans IChemE 78(Part C), 139-144.
8. Formanek, J., Mackie, R., and Blaschek, H. P. (1997), Appl. Environ. Microbiol. 63,
2306-2310.
9. Qureshi, N. and Blaschek, H. P. (1999), Biomass Bioenergy 17,175-184.
10. Holm, J., Bjorck, I., Drews, A., and Asp, N.-G. (1986), Starch/Starke 38, 224-226.
11. Jesse, T. W., Ezeji, T. c., Qureshi, N., and Blaschek, H. P. (2002) J. Ind. Microbiol.
Biotechnol. 29,117-123.

Applied Biochemistry and Biotechriology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0383/$20.00

Measurement of Rheological Properties


of Corn Stover Suspensions

NATALIA V. PIMENOVA AND THOMAS R. HANLEY*


Department of Chemical Engineering, University of Louisville,
Louisville, KY 40292, E-mail: tom.hanley@louisville.edu

Abstract
Corn stover is currently being evaluated as a feedstock for ethanol pro-
duction. The corn stover suspensions fed to reactors typically range between
10 and 40% solids. To simulate and design bioreactors for processing highly
loaded corn stover suspensions, the rheologic properties of the suspension
must be measured. In systems with suspended solids, rheologic measure-
ments are difficult to perform owing to settling in the measurement devices.
In this study, viscosities of corn stover suspensions were measured using a
helical ribbon impeller viscometer. A calibration procedure is required for
the impeller method in order to obtain the shear rate constant, k, which is
dependent on the geometry of the measurement system. The corn stover
suspensions are described using a power law flow model.
Index Entries: Corn stover; rheological properties; helical impeller; cone-
and-plate impeller; power law parameters.

Introduction
Production of fuel ethanol from renewable lignocellulosic material
(bioethanol) has the potential to reduce world dependence on petroleum
while decreasing net emissions of CO2, the principal greenhouse gas.
Lignocellulosic biomass includes hardwoods, herbaceous crops, agricul-
tural residues (i.e., com stover), and wastepaper and other fractions of
municipal solid waste. These materials are primarily cellulose, hemicellu-
lose, and lignin (1-3).
The lignin-hemicellulose network of biomass retards cellulose bio-
degradation by cellulolytic enzymes. To remove the protecting shield of
lignin-hemicellulose and make the cellulose more readily available for
enzymatic hydrolysis, biomass must be pretreated (4).
Thermochemical treatment (e.g., with steam and dilute H2S04) is a
popular pretreatment process. This treatment opens the lignocellulose pore
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 383 Vol. 105-108,2003


384 Pimenova and Hanley

structure and increases the susceptibility of biomass to enzymatic attack


(1). This pretreatment step effectively hydrolyzes liquor typically rich in
pentose sugars and produces a cellulose-rich solid with greater porosity
and improved enzymatic digestibility (1).
Stirred tanks are usually used for the thermochemical pretreatment.
To simulate flow of corn stover slurries in stirred tanks, the rheologic prop-
erties of these suspensions must be known. The corn stover slurries in
stirred tank reactors typically range from 10 to 40 % solids (5).
In systems with suspended solids, rheologic measurements are dif-
ficult to perform owing to settling in the measurement devices. Conven-
tional methods for measuring rheologic properties (cone-and-plate,
concentric cylinder, and rotating bob viscometers) do not produce accu-
rate and reliable data for some solid suspensions. Common problems
encountered with conventional instruments include phase separation
near the vicinity of the bob, particle settling, destruction of particles in the
vicinity of the rotating device, and blockage of narrow gaps by large
particles (6,7).
To avoid the apparent complications with absolute rheologic mea-
surement techniques, a number of investigators (8,9) have used relative
measurement systems to make rheologic measurements. The major differ-
ence between the relative and absolute measurement techniques is that the
fluid mechanics in the relative systems are complex. The constitutive equa-
tions needed to find the fundamental rheologic variables cannot be readily
solved. Relative measurement systems require the use of Newtonian and
non-Newtonian calibration fluids with known properties to relate torque
and rotational speed to the shear rate and shear stress (10).
Research on the impeller method using the helical ribbon impeller is
well documented (6,7). The impeller method is often employed to measure
the rheology of suspensions. Previous researchers assumed that the effec-
tive shear rate of such a device is related to the impeller speed by a fluid-
independent constant, but there is evidence that this is not true for all
impellers (7,11). It has been suggested that a properly designed helical
ribbon impeller might be more appropriate for this technique.
The principal goal of the present investigation was to study the rheol-
ogy of corn stover slurries. The objectives were to determine the effective-
ness of the impeller method in measuring the rheologic properties of corn
stover slurries, to determine the power law parameters of suspensions with
different concentrations of corn stover fiber using the impeller method, and
to compare the power law parameters of the corn stover slurries with those
of filamentous suspensions reported in the literature.

Data Analysis for Impeller Viscometer Technique


The complex flow field created by the impeller does not allow direct
calculation of the shear rate (7,10). It is assumed that the dimensionless
power number (PNo) is inversely proportional to the impeller Reynolds
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Rheological Properties of Corn Stover 385
number (Rei) for Newtonian fluids in a laminar flow regime in which the Rei
is <10:

PNo = c / Rei = 2rcf1s for Re < 10 (1)


pN Di

pND~ (2)
Rei=-~-

in which k and c are empirically determined constants, M is the torque, Di


is the diameter of the helical impeller, and p is the density of the using fluid.
For a given impeller, the torque is directly proportional to the impeller
speed and the apparent viscosity:
cD 3
M=_l ~N (3)
2rc
If the torque is measured as a function of the impeller speed for a
known viscosity Newtonian fluid, the constant, c, can be determined. The
apparent viscosity for a non-Newtonian fluid can then be determined from
measurements of the impeller torque as a function of impeller speed from
Eq. 3 (6).
Replacing the viscosity, /!, in the Rei with the apparent viscosity of the
non-Newtonian fluid, 'YJa' at the average shear rate, and solving Eq. 2 for the
apparent viscosity, produces
1'] = 2rcM
a cND~ (4)
I

The average shear rate in the measuring vessel, Yavg' is assumed to be


proportional to the impeller speed, N, and independent of the rheology of
the fluid in the vessel. The shear rate constant, k, is used as a fluid-indepen-
dent constant:
Yavg =kN (5)
If this approach is valid, the shear rate constant can be determined
from experimental measurements of torque vs impeller speed for non-
Newtonian fluids of known properties (12). .
The apparent viscosity is determined from Eq. 4. The value of shear
rate that corresponds to this viscosity is obtained from the known-vise os-
ity-vs-shear-rate rheogram for the non-Newtonian fluids generated using
the cone-and-plate method. The value of k is determined from Eq. 5.

Materials and Methods


Equipment
Two Brookfield rheometers with full-scale spring torques of 7178
and 57,496 dyne-em were used: a digital RV -DV III cone-and-plate instru-
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
386 Pimenova and Hanley

Fig. 1. Helical impeller.

ment and a digital HB-DV III, respectively. The uncertainty specified by


the manufacturer for these devices is 1% of the full-scale range. Therefore,
no data were taken unless the torque displayed was >5% of the maximum
value for a given instrument. A cone/plate attachment (RV-DV III vis-
cometer) was used to characterize the rheology of the Newtonian and
non-Newtonian calibration fluids used (Brookfield spindle cp-42). The
temperature was maintained at 25.0 ± 0.1 °C using a circulating water bath
for all tests.
The helical impeller used was fashioned from nylon using selective
laser sintering technology (see Fig. 1). The impeller had a diameter of 0.04
m and a pitch of 0.02 m and featured two helices: an ascending outer flight
and a descending inner flight. The length of the impeller was 0.055 m, and
it was located at an off-bottom clearance of 0.025 m.
The use of Newtonian and non-Newtonian calibration fluids allows
determination of constants relating the measured torque and speed to vis-
cosity and shear rate. Silicone oil and glycerol (Newtonian) with viscosities
of 1.024 and 0.912 Pa(s, respectively, were used to determine the impeller
constant, c, while xanthan and guar gum solutions (non-Newtonian) were
used to determine the shear rate constant, k.
Applied Biochemistry and Biotechnology Vol. 105-1 DB, 2003
Rheological Properties of Corn Stover 387
0.8 , - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

0.6

E
~ 0.4
<Il
:l
~
~
0.2

0
0 2 3 4 5 6 7
Rotational Speed (rpm

Fig. 2. Torque-speed relationship for Newtonian silicone oil.

Corn Stover Suspensions


The corn stover suspensions used were composed of various concen-
tration of corn stover particles (average fiber length = 120 !lm) suspended
in water. Corn stover liquor was used to prepare corn stover suspensions.
Cooking corn stover in a 1.4% H 2S04 solution and then dewatering the
resultant slurry produced corn stover liquor. The liquor contained
byproducts of the pretreatment process including lignocellulosic sugars,
H 2S04, and acetic acid.
Rheologic Measurements
Rheologic measurements were performed at 25°C with the cone-and-
plate and the helical impeller viscometers. Impeller viscometer measure-
ments were performed in a 1000-mL beaker with a diameter of 0.115 m. A
liquid height of 0.115 m was used for all tests.
For the impeller ribbon viscometer technique, the power number of an
impeller is inversely proportional to the Rei (Eq. 1). As the impeller's rota-
tional speed increases, the flow will gradually change from laminar to tur-
bulent, passing through a transition region. Parameter c can be obtained
from the calibration fluids. If the same value for c is assumed to apply to a
non-Newtonian fluid, then Eq. 4 can be used to calculate the apparent
viscosity of that fluid. The range of the impeller method is determined by
the minimum and maximum torques that can be measured (9).

Results
From the Newtonian calibration fluids measurements, the value for
the constant, c, was determined to be 135. Figure 2 shows the typical
example of the torque-impeller speed curves in the case of silicone oil.
Figure 3 shows the relationship between the impeller constant, c, and the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
388 Pimenova and Hanley

100 r-----------------------------------------------~

155

U 145

135 c Glycerol

• Silicon Oil DMPS-1M

o 2 3 4 5 6 7 8 9
Re
Fig. 3. Relationship between c and Re for glycerol and silicone oil DMPS-IM using
helical impeller.

Re for silicone oil and glycerol. The deviation in the value of c between Re
from 1 to 10 was <5%.
The value of the shear rate constant, k, was determined for solutions
of xanthan gum and guar gum with concentrations of 0.5,1.0, and 1.5%.
Figure 4 compares the helical impeller and cone-and-plate data for the 0.5
and 1.0% xanthan gum solutions.
The average value for k determined for each solution is shown in
Table 1 along with the power law indices calculated from the cone-and-
plate and impeller viscometer data for each fluid. The shear rate constant
k obtained from the cone-and -plate and helicalimpellers was 10.9. A simi-
lar value of k was obtained for 1.0 and 1.5% of xanthan and guar gum
solutions. The same value of k was reported for this type of impeller in
earlier investigations (6,11).
The results of the rheologic measurements for the corn stover suspen-
sions are presented in Fig. 5. Corn stover suspensions of 5, 10,20, and 30%
were used. The viscosity of the suspension increased as the fiber loading
increased, as expected. The power law parameters (the consistency index
number, n; the power law parameters, ~l) for the various fiber suspensions
are presented in Table 2 and may be compared with the results of Dronawat
et aL (11), who conducted tests with the Solka-Floc fiber of similar length
(215 !lm). The parameters are independent of the method of measuring
rheologic data and dependent on the fluid.
Applied Biochemistry and Biotechnology Vol. 105-708,2003
Rheological Properties of Corn Stover 389
A 100
I ~ Cone-and-P1ate I
I. I Ielical Impelle~
<0 0
0

-v0 6
--0
.t.
... ~
0.01
100 10000
Shear Rate ( lis)

B 10
J 0 Conc-and-P1atc
l. Helical Impelle

<>
~

-~
. ... ~

0.001
1 100 10000

Shear Rate (lis)

Fig. 4. Shear rate vs viscosity for (A) 0.5 % and (B) 1.0 % xanthan gum.

Table 1
Values of Shear Rate Constant for Different Fluids and 1000-mL Vessel
n
Solution k Cone-and-plate impeller Helical impeller
XanthanGum
0.5% 10.8 ± 1.2 0.37 0.39
1.0% 10.1 ± 1.7 0.26 0.23
1.5% 10.8 ± 0.6 0.19 0.19
Guar gum
0.5% 11.6 ± 1.3 0.45 0.44
1.0% 11.2 ± 0.7 0.31 0.29
1.5% 11.0 ± 1.0 0.19 0.18

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


390 Pimenova and Hanley
10000

. .. .5%
.10% f- c-
• •
D

..
100 .. " .tl:lO%
.. 30% t:- ~

~
.tl

~
.. ... ..
.tl
.tl
.tl .tl

- - . • ...
0.01
10
'"' - 100
Shear Rate. lis

Fig. 5. Rheology of com stover suspensions.

Discussion
Calibration Procedure
The basic assumption of the impeller viscometer approach is that the
shear rate constant is independent of the rheologic properties of the fluid.
It allows the helical impeller viscometer to be calibrated for homogenous
non-Newtonian fluids, which are difficult to analyze by conventional
rheologic instruments. Xanthan and guar gum solutions were chosen as
non-Newtonian calibration fluids, because their rheologic behavior at low
shear rates is similar to that of yield stress fluids. The calibration results for
guar and xanthan gum solutions ranging in concentration from 0.5 to 1.5%
produced a single value of k = 10.8 sufficient to represent all the data.
The power law indices obtained using the helical impeller compare
well with the cone-and-plate viscometer results, as can be seen in Tables 1
and 2. The difference in the value of k may be explained by the fact that in
the vicinity of the low-shear Newtonian transition, the viscosity is relatively
insensitive to the shear rate, which is not desirable for determining k.

Corn Stover Suspensions


The viscosity of the corn stover suspensions was determined for con-
centrations up to 30%. The helical impeller method was ineffective at corn
stover concentrations >32%.
The power law parameters for the corn stover suspensions could not
be located in previous studies. Dronawat et al. (11) studied a similar system
Applied Biochemistry and Biotechnology Vol. 105-708,2003
Rheological Properties of Corn Stover 391
Table 2
Power Law Parameters for Non-Newtonian Solutions Used
Solution Kp/ (Pa·s) n R2
Helical impeller
Xanthangum
0.5% 2.09 0.29 0.97
1.0% 6.02 0.23 0.98
1.5% 13.56 0.23 0.99
Guar gum
0.5% 0.78 0.44 0.99
1.0% 4.77 0.29 0.99
1.5% 16.32 0.18 0.98
Corn stover suspension
5% 0.05 0.91 0.90
10% 1.87 0.08 0.99
20% 71.82 0.06 0.99
30% 1684.50 0.05 0.99
Cone-and-plate
Xanthangum
0.5% 2.07 0.29 0.98
1.0% 6.01 0.22 0.98
1.5% 13.55 0.22 0.99
Guar gum
0.5% 0.69 0.42 0.99
1.0% 4.68 0.31 0.99
1.5% 17.00 0.18 0.98

using filamentous particles (Solka-Floc with a fiber length of 215 f.lm in a


0.5% xanthan gum water solution). Comparison of the power-law param-
eters indicates that the suspensions studied herein were less viscous and
less shear thinning than the suspensions Dronawat et al. (11) studied. The
difference can be attributed to the factthat our study used waterrather than
a 0.5% xanthan gum solution. In both studies the increase in the consistency
index as the power law index increased was moderate.

Conclusions
The parameter c is a linear function of Re. The linear regression analy-
sis performed on the data collected for glycerol and silicone oil yielded
regression coefficients ranging from 0.98 to 0.998. Furthermore, the shear
rate constant is independent of the rheologic properties of the fluid. The
percentage difference between the highest and lowest values of shear rate

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


392 Pimenova and Hanley

constant calculated for the xanthanand guar gum was 10% (10.9 ± 1.1). The
impeller method can accurately and reliably measure the rheologic prop-
erties of filamentous suspensions, based on the results obtained from the
corn stover suspension experiments.

Acknowledgment
Funding for this project was provided by the National Renewable
Energy Laboratory (subcontract no. XCO-1-31016-01).

References
1. McMillan, J. D. (1997), Renewable Energy 10(2/3), 295-302.
2. Wenzl, H. F. J. (1996), The Chemical Technology of Wood, Academic, New York, NY.
3. Hayn, M., Steiner, W., Klinger, R, Steinmueller, H., Sinner, M., and Esterbauer, H.
(1993), in Bioconversion of Forest and Agricultural Plant Residues, Saddler, J. N., ed.,
CAB, Wallingford, UK, pp. 33-72.
4. Esteghlalian, A., Hashimoto, A. G., Fenske, J. J., and Penner, M. (1997), Bioresour.
Technol. 59,129-136.
5. Ranatunga, T. D., Jervis, J., Helm, R. F., McMillan, J. D., and Wooley, R J. (2000),
Enzyme Microb. Technol. 27,240-247.
6. Svihla, C. K., Dronawat, S. N., and Hanley, T. R (1995), Appl. Biochem. Biotechnol. 51152,
355-366.
7. Allen, D. G. and Robinson, C. W. (1990), Chem. Eng. Sci. 45(1),37-48.
8. Kemblowski,1. and Kristiansen, B. (1986), Biotechnol. Bioeng. 28, 1474-1483.
9. Metz,B.,Kossen,N. W.F.,and VanSuijdam,J.C. (1979), Adv. Biochem. Eng. 11,103-155.
10. Charles, M. (1978), Adv. Biochem. Eng. 8, 1-62.
11. Dronawat, S. N., Rieth, T. c., Svihla, C. K., and Hanley, T. R (1996) in Proceedings of the
5th World Congress of Chemical Engineering, vol. I, AIChE, New York, NY, pp. 629-633.
12. Svihla, C. K., Dronawat, S. N., Donnely, J. A., Rieth, T. c., and Hanley, T. R (1997)
Appl. Biochem Biotechnol. 63/65, 375-385.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0393/$20.00

Effect of Bacillus circulans D1


Thermostable Xylanase on Biobleaching
of Eucalyptus Kraft Pulp

DANIELA A. BOCCHINI, VALQUIRIA B. DAMIANO,


ELENI GOMES, AND ROBERTO DA SILVA*

Oepartmento de Quimica de Geociencias,


Laborat6rio de Bioqufmica dos Processos e Microbiologia Aplicada,
IBILCE-UNESP, Rua Cristovao Colombo 2265,
Sao Jose do Rio Preto, SP, CEP 15054-000, Brazil,
E-mail: dasilva@qeg.ibilce.unesp.br

Abstract
The alkalophilic Bacillus circulans 01 was isolated from decayed wood.
It produced high levels of extracellular cellulase-free xylanase. The enzyme
was thermally stable up to 60°C, with an optimal hydrolysis temperature
of 70°C. It was stable over a wide pH range (5.5-10.5), with an optimum
pH at 5.5 and 80% of its activity at pH 9.0. This cellulase-free xylanase
preparation was used to biobleach kraft pulp. Enzymatic treatment of kraft
pulp decreased chlorine dioxide use by 23 and 37% to obtain the same
kappa number (K number) and brightness, respectively. Separation on
Sephadex G-50 isolated three fractions with xylanase activity with distinct
molecular weights.
Index Entries: Bacillus circulans; biobleaching; kraft pulp; thermophilic;
xylanase.

Introduction
After cellulose, hemicelluloses are the most abundant organic materi-
als found on Earth and are the main polysaccharides of plant cell walls.
They are closely associated with cellulose in plant tissues and account for
40-45% of the dry weight of hardwood and softwood. Xylan is the main
constituent of hemicellulose from wood, and it accounts for >90% of the
hemicellulose in kraft pulp from hardwood, and about 50% of hemicellu-
loses in softwood pulps (1). There is evidence that xylan binds to lignin by

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 393 Vol. 105-108,2003


394 Bocchini et at.
covalent bounds, and that hydrogen bonds and Van der Wall's forces bind
xylan to the cellulose chain (2).
In recent years, there has been increasing interest in applying
xylanolytic enzymes, mainly endoxylanases, to the pulp bleaching process.
After cooking wood chips under alkaline conditions (kraft process), about
90% of lignin is removed and about 10% remains, encrusting the cellulose
and hemicellulose fractions. Traditional bleaching using chlorine-based
compounds effectively removes residual lignin, but it is very harmful to the
environment. Many researchers are investigating environmentally clean
bleaching processes using chemical or biotechnological approaches (3,4).
The efficacy of xylanase in biobleaching has been demonstrated in many
scientific reports. These reports conclude that xylanase depolymerizes
xylan, opening the polymeric cellulose-hemicellulose-lignin matrix, releas-
ing lignin-bound fragments, and facilitating the chemical removal of lignin
during bleaching (4). In addition, these enzymes improve paper qualities,
such as brightness and viscosity (5-9). Over the last decade, a number
of microbial enzymes have been assessed for potential application in the
paper and pulp industry (3-30). Bleaching occurs under high temperature
and alkalinity and the enzymes must tolerate these conditions.
Bacillus circulans Dl, an alkalophilic and thermophilic bacterial
strain, was isolated in our laboratory, from decayed wood. It shows high
xylanase production. The crude xylanase produced by this microorgan-
ism is thermostable and cellulase free. Our goal was to employ the crude
xylanase produced by B. circulans Dl as a booster in chlorine-based
biobleaching of Eucalyptus kraft pulp, and to analyze its effectiveness on
reducing chlorine consumption.

Materials and Methods


Media and Screening of Microorganisms
Nutrient medium was prepared as described by Horikoshi (31),
containing xylan (10 giL), beef extract (10 giL), peptone (10 giL), NaCl
(10 giL), and KH2P04 (1 giL), and Na2C03 (5 giL) (separately sterilized).
For the solid nutrient medium, 15 giL of agar was added. To screen for
xylanolytic and alkalophilic microorganisms, the nutrient medium was
used with dried corn straw strips, as a carbon source replacing xylan. The
corn straw strips were placed vertically into tubes containing nutrient
medium. Samples of soil, agricultural wastes, and decayed wood were
collected (approx 0.5 g of each sample) and put into flasks containing
1.5 mL of nutrient medium. In the laboratory, these samples were trans-
ferred to tubes containing screening medium and incubated at 50°C
for 120 h. After incubation, 1.0 mL of the contents from the tubes, with
degraded straws trips, was diluted in 5.0 mL of sterile distilled water and
one loop was inoculated on plates containing solid nutrient medium. The
plates were maintained at 50°C, and then the developed colonies were
transferred to slant tubes with solid nutrient medium.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
B. circulans 01 and Biobleaching of Kraft Pulp 395
Microorganism and Growth Conditions
B. circulans D1 was isolated from decaying wood and was identified
by Fundacao Andre Tosello, Campinas, SP, Brazil. For enzyme production,
the B. circulans D1 was grown in alkaline medium (pH 9.0) except using 2%
com straw, powdered to a diameter of 1.0 mm, as the carbon source. Twenty
milliliters of the medium, in 125-mL Erlenmeyer flasks, was inoculated
with 107 cells/mL and incubated at 45°C with shaking at 150 rpm. After 48
h, the bacteria were harvested by centrifuging at 12,OOOg at 4°C for 20 min.
The cell-free solution was used as crude enzyme solution.

Determination of the enzymatic activity


Enzyme activity was expressed in international units (IU). Xylanase
(l,4-~-D-xylan xylanhydrolase, EC 3.2.1.8) activity, carboxymethyl cellu-
lase (endo-1,4-~-D-glucan glucanohydrolase, EC 3.2.1.4), and Avicelase
(exo-1,4-~-D-cellobiohydrolase, EC 3.2.1.91) were assayed by incubating
0.1 mL of appropriately diluted enzyme with 0.9 mL of a solution contain-
ing 0.5% of the respective substrate (xylan [Birchwood; Sigma, St. Louis,
MO], carboxymethyl cellulose [C5678; Sigma], and avicel [Merck]) inO.1 M
acetate buffer, pH 5.5. After incubating at 60°C for 10 min, the reducing
substances released were assayed by dinitrosalicylic acid as described
by Miller (13). Controls were prepared with enzyme added after boiling.
The definition of 1 IU of activity toward the substrate just mentioned was
1 f..tmol of xylose or glucose equivalent released/ min under the stated assay
conditions, by using a xylose or glucose standard curve.

Effect of Temperature and pH on Xylanase Activity and Stability


Xylanase activity was measured between 45 and 80°e. Thermal sta-
bility (without substrate) was also estimated by maintaining the enzyme
for 1 h at the same temperatures. After cooling, the residual activities
were estimated under the standard condition. The effect of pH on xylanase
activity was tested from 3.5 to 10.5. pH stability (without substrate) was
estimated by maintaining the enzymes for 24 h in the same pH range and
then determining the residual activities under the standard condition.

Fractionation of Xylanase
The crude enzyme was precipitated by adding ethanol to make a 70%
(v Iv) solution; the precipitate was collected by centrifugation. The concen-
trated crude enzyme (12 mL), containing a total of 601 IU, was applied to
a Sephadex G-50 column that had been equilibrated with 20 mM acetate
buffer, pH 5.0, and was eluted with the same buffer.

Products of Xylan Hydrolysis


Birchwood xylan hydrolysates were examined by descending paper
chromatography, as described by Trevelyn et aL (32).
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
396 Bocchini et at.
Enzymatic Treatment of Pulp
Eucalyptus kraft pulp was courteously donated by Champion Paper
and Cellulose from Mogi Gua<;u, SP, Brazil. It had an initial K number
(milliliters of 0.1 N potassium permanganate consumed by 1 g of dry wood
chemical pulp under specific conditions) of 13.85 and brightness of 37.0
ISO. The unbleached pulp was washed twice with distilled water to remove
residual brown liquor. Crude enzyme was applied to the pulp at a concen-
tration of 13.5 IV / g of dried pulp. The enzymatic treatment was made in
plastic bags at a slurry concentration of 10%. The pH of this mixture was
adjusted to 5.5 by adding glacial acetic acid, and then the mixture was
homogenized and the plastic bags were maintained at 60°C for 4 h. Then,
the plastic bags were cooled in cold water for a few minutes, and the pulp
was washed twice with distilled water and filtered in a Buchner funnel.

Chemical Bleaching
Pulps, treated and untreated with enzyme, were bleached with chlo-
rine dioxide (CI02) followed by extraction with NaOH(E) (DE bleaching
stages). The amount of Cl02 used in the bleaching was determined by cal-
culating the concentration of active chlorine on the aqueous solution of
Cl02 and varied around the kappa (K) factor recommended (K factor is a
standard quantity, determined experimentally, used in studies comparing
bleaching processes and represents the percentage of active chlorine di-
vided by the K number of the pulp: K number x K factor =% of Cl02), which
under typical industrial conditions is 0.26. The samples contained 10.0 g of
oven-dried pulp and Cl021 at a 10% slurry concentration. Each sample was
homogenized, maintained in a bath at 60°C for exactly 30 min, cooled, and
then washed with distilled water in a Buchner funnel. The alkaline extrac-
tion was made with pulp at 10% concentration, at 70°C for 1 hand NaOH
final concentration of 1.6%. Afterward, the samples were washed with
distilled water and filtered in a Buchner funnel.

Pulp Properties
Pulp properties were investigated according to the Standard Methods
of the Technical Association of the Pulp and Paper Industry (TAPPI Stan-
dard Methods). For determination of K number, the microKappa method-
ology was used, by reacting pulp samples with acidified potassium
permanganate, as described in TAPPI protocol T-236 OM-85. After
delignification, the pulp was pressed and transformed in dried sheets, at
room temperature, for viscosity and brightness experiments. Viscosity was
evaluated by dissolving pulps in cupriethylenodiamine and by measuring
the viscosity with an Ostwald viscosimeter, as described in TAPPI protocol
T-230 OM-82. The brightness of the paper sheets was measured by reflec-
tance at 457 nm with an Elrepho 2000 instrument, according to TAPPI pro-
tocol T-452 OM-87.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
B. circulans 01 and Biobleaching of Kraft Pulp 397
Table 1
Characteristics of Treated and Untreated Kraft Pulp
Untreated pulp Pulp treated with xylanase

K Brightness Viscosity K Brightness Viscosity


number (ISO) (cP) number (ISO) (cP)

Before bleaching 13.85 37.00 60 12.80 41.00 61.5


After bleaching 2.70 56.50 32 1.70 60.00 35
Pulps were submitted to bleaching under typical conditions: K factor of 0.26 (0.36 g of
CI02 /10 g dry wt pulp) and 1.6% NaOH extraction.

Results and Discussion


Properties of Crude Xylanase
The crude enzyme solution was analyzed for cellulase activity, but it
was free of this enzyme. The optimum pH and temperature for xylanase
activity were 5.5 and 70°C, respectively. The thermal stability of the crude
enzyme was measured, in the absence of substrate, from 20 to 80°C for 1 h.
This enzyme loses activity when incubated above 60°C. The crude xylanase
was stable in the pH range of 5.5.-10.5 after incubating in different buffers
for 24 h. These values agree with those reported for xylanases produced by
other Bacillus strains (23-25).

Effect of Enzymatic Treatment and Chemical Bleaching on Pulp


Table 1 shows the results from enzymatically treating and bleaching
with Cl02• The initial k number of the pulp was 13.85, and it was reduced
after enzymatic treatment to 12.80. In addition to this reduction in k num-
ber, pulp brightness increased by about 4 U.
The pulps, treated and untreated with xylanase, were submitted to DE
bleaching sequence, with varying Cl02 concentrations. For xylanase treat-
ment, investigators have used only one CI02 concentration, but the chlo-
rine concentration used after enzyme treatment may influence the results.
To analyze this point, bleaching was performed with different CI02 con-
centrations. As shown in Fig. I, the K number of the pulp treated with
xylanase decreased quickly (from 3.5 to 1.7) until it reached the value of
1.70 IV (0.36 g of Cl02), and then it slowed down, tending to stabilize at
about 1.0-1.5 IU. From this result we can suppose that there is a Cl02 con-
centration at which the majority of the delignification reaction occurs and
that an additional increase in chlorine concentration will not significantly
decrease K number. This reaction occurs probably because there are few
xylan-lignin associations at this stage of K number. In our study, according
to Table 1 and Fig. I, at a K factor of 0.26 (0.36 g of Cl02), the range of
delignification was 80 (13.5-2.70) and 88% (13.5-1.7) for untreated pulp
and enzyme-treated pulp, respectively. This is importantin order to reduce
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
398 Bocchini et al.
5

Q) 4
.0
E
::l
c:: 3
to
a.
a.
~ 2

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55

CI02 (gl10 oven dried pulp)

Fig. 1. K Number of kraft pulp (-e-) treated and (-0-) untreated withxylanase
from B. circulans D1. Data are mean of two replicates.

the bleaching process cost and discharge of chlorinated organic com-


pounds. The enzymatic treatment of pulp resulted in significant chlorine
savings. If we consider, e.g., the K number of 2.7, this number was already
obtained with a K factor of 0.17 (0.23 gofCI02 /10 gofdried pulp) in the pulp
treated with enzyme. To reach this value in the untreated pulp, it was
necessary to employ a K factor of 0.22 (0.3 g of Cl02 /10 g of dried pulp). The
enzyme decreased by 23% the amount of Cl02 used for pulp de lignification.
Many scientific articles have reported a decrease in the active chlorine rate
in relation to the K number, varying from 20 to 40% (12).
Enzymatic treatment of the pulp increased brightness by approx 4 IU
compared to untreated pulp (Fig. 2). The results indicate that for a given
brightness, e.g., 56.5, a K factor of 0.25 (0.35 g of CI02 /10 g of dried pulp)
would be needed for the untreated pulp, compared with 0.16 (0.22 g of
Cl02 /10 g of dried pulp) in the pulp treated with xylanase. This is 37% less
CI02 used to reach the same brightness. Using xylanase from Streptomyces
thermoviolaceus in the biobleaching of kraft pulp, Garg et al. (22) observed
a 30-35% saving in the chlorine required to obtain pulp brightness, compa-
rable with the control. Using xylanase of B. circulans AB 16 in biobleaching
of eucalyptus pulp, Dhillon et al. (8) reduced chlorine consumption by 20%
to achieve the same final pulp brightness compared with the control.
The pulp treated with enzyme had more viscosity than the untreated
pulp. This was likely owing to the selective enzymatic hydrolysis of low
molecular weight pulp xylans that probably contributed to a smaller aver-
age viscosity (8,20). This parameter indicates thatthe crude enzyme did not
disrupt fiber, and thus loss of paper properties (8,10). A slight increase in
viscosity was also obtained by Dhillon et al. (8) using xylanase from
B. circulans AB 16, and by Buchert et al. (21) using xylanase from Tricho-
derma reesei. Viscosity reduction is not desirable because this property is
related to the degree of cellulose polymerization and to paper strength.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
B. circulans 01 and Biobleaching of Kraft Pulp 399

_60
oC/)
- g j56

-c:
..c:
Q)

.g> 52
ID

48+-~-.~-''-~~~~~--r-~.

0.2 0.25 0.3 0.35 0.4 0.45 0.5


. CI02 (gl10 oven dried pulp)

Fig. 2. Brightness of kraft pulp (-0-) treated and (-e-) untreated with xylanase
from B. circulans 01. Data are mean of seven replicates.

3 0.35

--
::J 2.5 0.3
E
::J 0.25
2 E
~ c:
0.2
~al 1.5
0
co
N
Q) 0.15 f/)
f/) .0
al <C
c: 0.1
al
>.
X 0.5
0.05

30 40 50 60 70 80 90 100 110
Fraction number

Fig. 3. Fractioning of B. circulans 01 xylanase on Sephadex G-50. (-e-) Xylanase


activity; (-0-) 280-nm absorbance.

Fractions of Xylanases on Crude Enzymatic Extract


The crude precipitated enzyme was applied to a Sephadex G-50
columm to separate it from pigments and other enzymes present in the
crude enzymatic extract. The column separated the crude enzyme into three
xylanase fractions (Fig. 3) designated xylanases I, II, and III, representing
distinct molecular weights of enzymes. Silva et al. (26) cited gel filtration as
an initial step of xylanase purification. Khanna et al. (27) also used Sephadex
G-50, and elution of the xylanase resulted in one peak of xylanase activity.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
400 Bocchini et al.
Multiplicity in xylanases has been shown in other Bacillus species (28).
Our result on xylanase partial purification is in disagreement with that
observed by Nakamura et al. (28), in which B. circulans AB 16 displayed
two isoforms of xylanases after one multistage purification procedure.
Additional assays on B. circulans D1 xylanase purification are being per-
formed in order to obtain more accurate results. Xylanases I-III were
concentrated with a Centricon concentrator with a 10-kDa cutoff and then
characterized regarding optimum activity at different pH and tempera-
ture values and to assess thermal and pH stability. Xylanases I-III showed
higher activity at pH near 5.5, 6.5-7.0, and 5.5-6.5, respectively. The
optimal temperatures for xylan hydrolysis were 65°C for xylanase I and
III and about 65-70°C for xylanase II. At 80°C, >67% of the activity of
xylanase III was retained, whereas at the same temperature, the remain-
ing activities of xylanases I and II were about 37 and 44%, respectively.
Products of Xylan Hydrolysis
When using the crude enzyme and the enzymatic fractions collected
after chromatography on Sephadex G-50, the main products of xylan
hydrolysis identified by paper chromatography (data not shown) were
xylotetraose and xylotriose. Xylose was not observed as a hydrolysis prod-
uct, which suggests the presence of only endoxylanases in the enzymatic
extract produced by B. circulans D1. Our results are in good agreement with
Nakamura et al. (28), Kang et al. (29), and Breccia et al. (30), who have cited
similar results.

Conclusion
The extracellular cellulase-free xylanase produced by alkalophilic
B. circulans D1 exhibited desirable properties such as activity at elevated
temperature, and alkali and thermal stability, which are advantageous
for application in the pulp and paper industry. Application of xylanase to
the kraft pulp significantly reduced the requirement of oxidizing CI02 in
the bleaching process. The crude xylanase displayed action of endoxylanase
and was separated in three fractions of distinct molecular weight.

Acknowledgments
We gratefully thank Champion Paper and Cellulose Uda for provid-
ing the pulp samples, Dr. Jorge L. Colodette for the analytical tests of pulp
samples, and Funda~ao de Amparo aPesquisa do Estado de Sao Paulo for
financial support.

References
1. Erikson,O.,Goring,D.A.I.,and Lindgren,B.O. (1980), Wood Sci. Technol.14,267-269.
2. Ferreira Filho, E. X. (1994), Braz. ,. Med. BioI. Res. 27, 1093-1109.
3. Tavares, V.B., Gomes, E., lembo, T., Bocchini, D. A., and Da Silva, R. (1997), in Fifth
Brazilian Symposium on the Chemistry ofLignins and other Wood Components, pp. 367-371.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


B. circulans 01 and Biobleaching of Kraft Pulp 401
4. Chen C. c., Adolphson, R, Dean, J. F. D., Eriksson, K. E., Adams, M. W. W., and
Westpheling, J. (1997), Enzyme Microb. Technol. 20,39-45.
5. Tolan, J. J. and Canovas, R V. (1992), Pulp Paper Can. 93,39-42.
6. Viikari, L., Buchert, J., Tenkanen, M., and Kantelinen, A (1995), Bioresour. Technol. 50,
65-72.
7. Beg,Q. K.,Bhushan, B., Kapoor,M., and Hoondal,G. S. (2000),Enzy me Microb. Technol.
27, 459-466.
8. Dhillon, A, Gupta, J. K., Jauhari, B. M., and Khanna, S. (2000), Bioresour. Technol. 73,
273-277.
9. Antonopoulos, V. T., Hernadez, M., Arias, M. E., Mavrakos, E., and Ball, AS. (2001),
Appl. Microbiol. Biotechnol. 57, 92-97.
10. Da Silva, R Yim, D. K. Park, Y. K. (1994) J. Ferment. Bioeng. 77,109-111.
11. Kulkarni, N., Abhay, S., and Rao, M. (1999), FEMS Microbiol. Rev. 23,411-456.
12. Tolan, J. J. and Canovas, R V. (1992), Pulp Paper Can. 93,39-42.
13. Wong, K. K. L, Tan, L. U. L, and Sadler, J. N. (1988), Microbiol. Rev. 52,305-317.
14. Viikari, L., Ranua, M., Kantelinen, A, Sundquist, J., and Linko, M. (eds.) (1986),
Proceedings of the 3rd International Conference on Biotechnology Pulp and Paper Industry,
Stockholm, Sweden, pp. 67-69.
15. Eligir, G., Sykes, M., and Jeffries, T. W. (1995), Enzyme Microb. Technol. 17,954-959.
16. Patel, R N., Grabski, A c., and Jeffries, T. W. (1993), Appl. Microbiol. Biotechnol. 39,
405-412.
17. Tavares, V. B., Gomes, E., and Da Silva, R J. (1997), Brazilian Soc. Microbiol. 28, 179-182.
18. Senior, D. J., Meyers, P. R, Miller, D., Sutclije, R, Tan, L., and Saddler, J. N. (1988),
Biotechnol. Lett. 10, 907-912.
19. Yang, V. W., Zhaung, Z., Elegir, G., and Jeffries, T. W. J. (1995), Indust. Microb. 15,
434-441.
20. Paice, M. G., Bernier, R, Jr. and Jurasek, L. (1988), Biotechnol. Bioeng. 32,235-239.
21. Buchert, J., Ranua, M., Kantelinen, A, and Viikari, L. (1992), Appl. Microb. Biotechnol.
37,825-829.
22. Garg, A P., Roberts, J. c., and McCarthy, A. J. (1998), Enzyme Microb. Technol. 22,
594-598.
23. Harris, G. W., Pickersgill, R W., Connerton, I., Debeire, P., Touzel, J. P., Breton, c.,
and Perez, S.(1997), Proteins: Struct. Funct. Genet. 29, 77-86.
24. L6pez, c., Blanco, A, and Pastor, F. I. J. (1998), Biotechnol. Lett. 20(3),243-246.
25. Bataillon, M., Cardinali, A P. N., and Duchiron, F. (1998), Biotechnol. Lett. 20(11),
1067-1071.
26. Silva, C. H. c., PuIs, J., Sousa, M. V., and Filho, E. X. F. (1999), J. Braz. Soc. Microbial.
30(2),114-119.
27. Khanna, P., Sundari, S. S., and Kumar, N. J. (1995), World J. Microbiol. Biotechnol. 11,
242,243.
28. Nakamura, S., Wakahayashi, K., Nakai, R, Aono, R, and Horikoshi, K. (1993), Appl.
Environ. Microbiol. 59,2311-2316.
29. Kang, M. K., Maeng, P. J., and Rhee, Y. H. (1996), Appl. Environ. Microbiol. 62(9),
3480-3482.
30. Breccia, J. D., Sifieriz, F., Baigori, M. D., Castro, G. R, and Kaul, R H. (1998), Enzyme
Microb. Technol. 22,42-49.
31. Horikoshi, K. (1973), Agric. Bioi. Chem. 35(11), 1783-1791.
32. Trevelyn, W. E., Procter, D. P., and Harrison, J. G. (1950), Nature 166, 444, 445.
33. Miller, G. L. (1959), Anal. Chem. 31,426-428.
34. Dhillon, A, Gupta, J. K., Khanna, S. (2000), Process Biochem. 35,849-856.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0403/$20.00

Fungi Allergens Produced


by Solid-State Fermentation Process
Optimization and Allergen Characterization

SALAH D. M. HASAN,l WALDEREZ GAMBALE,2


RICARDO L. ZOLLNER/ AND MARIA H. A. SANTANA*,l
1School of Chemical Engineering, State University of Campinas,
PO Box 6066, 13083-970, Campinas-SP, Brazil,
E-mail: lena@feq.unicamp.br;
2Laboratory of Mycology, Biomedical Science Institute,
University of Sao Paulo, 05508-900, Sao Paulo, Brazil; and
3School of Medical Sciences, State University of Campinas,
PO Box 6111, 13083-970, Campinas-SP, Brazil

Abstract
Allergenic extracts were produced from Drechslera (Helminthosporium)
monoceras biomass cultured by solid-state fermentation using wheat bran as
the substrate. The main fermentation variables were selected by statistical
design, and the optimized biomass yield (1.43 mg/[g of dry substrate· d])
was obtained at pH 9.5 and 45.8% moisture. The allergenic extracts were
produced from crude extract by protein precipitation and polyphenol
removal. Proteins in the range of 16-160 kDa were identified in the extracts.
Their reactions in patients were characterized by in vivo cutaneous tests
(positive in 40% of the atopic patients) and by dot-blotting assays.
Index Entries: Allergenic extract; Drechslera (Helminthosporium) monoceras;
solid-state fermentation; statistical experimental design; wheat bran; pro-
teins; biomass.

Introduction

Allergy-related diseases may be considered a worldwide public health


problem. Asthma and rhinitis are the most frequent clinical manifestations.
Allergenic extracts are of fundamental importance to the diagnosis and
therapy of allergies. The spores and mycelium fragments of molds in the
atmosphere are considered to be one of the most potent allergenic agents.

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 403 Vol. 105-108,2003
404 Hansan et al.

Their allergenic compounds are usually proteins that induce in humans the
formation of IgE isotype antibodies when inhaled, swallowed, or injected
(1). About 340 genera of molds associated with respiratory allergies are
listed in the literature. Alternaria, Cladosporium, Aspergillus, and Penicillium
are the most extensively studied genera, and their extracts have been char-
acterized and standardized (2).
Previous studies showed intense allergic reactions to Drechslera
(Helminthosporium) monoceras extracts in asthmatic patients, as measured
by cutaneous tests (2,3). These extracts were obtained from the fungi bio-
mass cultured by liquid fermentation. The main antigens were identified as
14.4,36, and 60 kDa proteins, and these extracts were successfully used in
allergy diagnosis (2,3).
D. monoceras is a saprophyte fungi that is found in the soil and is
associated with pathogenicity in plants such as maize, oats, wheat, sugar-
cane, and grasses. This characteristic suggests that solid-state fermentation
using agricultural residues as substrates could be a suitable process for
saccharification and fermentation, owing to the similarity to its natural
habitat. In addition, it could be a less expensive process to produce aller-
genic extracts on a large scale. Solid-state fermentation can be a powerful
process for microorganism growth using agroindustrial residues, and it is
more advantageous in many ways than liquid fermentation, especially
when yeasts or filamentous fungi are used (4). In recent years, several pro-
cesses that utilize agroindustrial residues as raw materials for the produc-
tion of bulk chemicals and value-added products, such as ethanol,
single-cell protein, edible mushrooms, enzymes, organic acids, amino ac-
ids, and biologically active secondary metabolites, have been reported (5).
This article describes a statistical process optimization for production
of D. monoceras biomass cultured by solid-state fermentation and presents
the characterization of the allergenic extracts obtained.

Materials and Methods


Microorganism and Inoculum
The strain ICBUSP K-1-16, CBS 15426 of D. monoceras used was obtained
from the culture collection of the Laboratory of Microbiology, Biomedical
Science Institute, University of Sao Paulo, Brazil, and maintained at 25°C on
potato dextrose agar submerged in mineral oil. The subculture was made in
a 1000-mL Erlenmeyer flask containing slant agar with 4% wheat bran and
nutrients in accordance with modification of the Czapeck broth (CB) made
by Yunginger et al. (6) (2 g of NaN03 , 1 g of K2HP04, 0.5 g of MgS04' 0.5 g
of KCI, 0.01 g of FeS04, 15 g of maltose, 15 g of dextrose, 10 g of tryptone,
1000 mL of distilled water, pH of solution: 6.8-7.0). The inoculum was
obtained from 14-d-old colonies through suspension with sterile water of
the biomass from the agar surface. Biomass concentration in the inoculum
was estimated by absorbance measurements at 550 nm using a standard
curve relating absorbance and dry biomass weight.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Fungi Allergens Produced by Solid-State Fermentation 405
Kinetic Characterization
Solid-state fermentation was carried out in Raimbault-type fixed-bed
columns (7), and the experimental setup was adapted from Moraes (8).
Wheat bran was used as the substrate and enriched with nutrients in accor-
dance with the CB composition. The mixture was sterilized for 15 min at
121°C and adjusted to a final pH and moisture content. The medium was
inoculated with a predetermined cell mass concentration and distributed
among 12 columns (35 mm diameter x 200 mm height). Each column was
individually aerated, and the set of columns was submerged in a thermo-
static bath at 25°C for temperature control. Fermentation took 12 d, with the
removal of a column as a sample every 24 h.
The kinetic behavior of fermentation was characterized by biomass
production, sugar depletion, water activity, and moisture contents. Biom-
ass concentration was determined by protein dosage using the Bradford
method (9), and the total reducing sugars by dinitrosalicylic acid reagent
(10). Both analyses were conducted with the crude extract obtained from
the fermented solid. Yield was calculated at 7 d of fermentation and
expressed as a ratio of milligrams of protein produced per gram of dry
substrate per day (mg/[g of dry substrate· d]). The efficiency of glucose
conversion in the biomass was estimated by the yield coefficient, YXIS (mg
of protein/mg of glucose).
Crude and Allergenic Extracts
Crude extracts were obtained after drying the fermented samples at
37°C for 72 h. Extraction was achieved by the addition of deionized water
to solids at a 1:15 (w Iv) ratio, adjustment of pH to 9.0, and incubation in a
shaker at 30°C for 16 h, according to Saraiva (11). The suspension was
filtered and centrifuged at 10,OOOg for 10 min.
The allergenic extracts were prepared from the supernatant crude
extract by protein precipitation and removal of salts, polyphenols, and
other interfering compounds. Proteins were precipitated using ethanol
(70% final concentration) or saturated ammonium sulfate and shaken for 1
hat O°c. After removal of the supernatant the precipitate was dissolved in
20 mM phosphate buffer (pH 7.0) and centrifuged for more efficient sepa-
ration of insoluble solids. The protein solution was desalted using Sephadex
G-25 M columns. The eluted proteins were then dialyzed (3500 mol wt
cutoff membrane) in 20 mM phosphate buffer (pH 7.0) for 48 h with three
changes of the buffer to remove polyphenols and other interfering com-
pounds. Polyphenols were determined using a spectrophotometric method
and a catechin standard curve, as described by Price and Butler (12).
Optimization Strategies
Optimal operating conditions for biomass production by solid-state
fermentation were determined using statistical experimental design and
the surface response analysis methodology (13). Initial moisture (M) and

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


406 Hansan et al.
pH of medium, inoculum concentration (Co), substrate particle size (dp),
and airflow rate (F) were the variables studied. The experimental response
was the biomass yield obtained after 7 d of fermentation and expressed in
terms of total protein per gram of dry substrate (mg/[g of substrate· d]).
First, the influence of the variables on biomass production was evalu-
ated using a two-level fractional factorial design, 25-2• The variables were
tested at the levels -I, 0, and +1, which correspond to pH 7.0, 9.0 and 11.0,
M = 40, 50, and 60%; F = 1,2, and 3 L/h; dp = 0.35, 0.59, and 0.84 mm; and
Co =0.2, 0.4, and 0.6 gil, respectively. Eight runs were performed accord-
ing to the 25-2 design (levels -1 and +1), and three runs were added to the
central point (level 0).
The strategy for optimization of the most significant variables included
an initial central composite design, an evaluation of the maximum slope
trajectory, and a second central composite design of the experiments. The
software STATISTICA (version 5.5) was used for calculating the effects of
variables, statistical models, and analysis of variance (ANOVA).
Characterization of Allergenic Extracts
Proteins of high and low molecular mass were identified in the aller-
genic extracts by soldium dodecyl sulfate polyacrylamide gel electro-
phoresis (SDS-PAGE) (14-16), using 10 and 15% polyacrylamide gels,
respectively. Protein concentration was determined using the Bradford
method (9). Antigenicity of extracts was evaluated by cutaneous prick
tests, and the most reactive human sera were collected for dot-blotting
analysis.

Results and Discussion


Effects of Process Variables
The statistical effects estimated for the five variables studied are shown
in a Pareto chart (Fig. 1). It was observed that the most Significant effects on
biomass yield were obtained for the pH and M variables with a confidence
level of 90%.
Optimization of Variables
Initial Central Composite Design
An initial central composite design of experiments was carried out for
the main variables pH and M. The variables were tested at the levels _21/2,
-1,0, +1, and +21/2, which correspond to pH 5.0,5.6, 7.0,8.4, and 9.0, and 30,
33,40,47, and 50% for M, respectively. Four runs were performed accord-
ing to the 22 design (levels -1 and +1), three runs to the central point (level
0), and four runs to the star design (levels _21/2 and +21/2) (13). The real
values of pH and M were calculated from the previously defined coded
values (Xl and X2), according to Eq. 1 and 2.
Xl = (pH -7.0)/1.4 (1)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Fungi Allergens Produced by Solid-State Fermentation 407

MS Pure Error=O.0193
Yield
p=0.1

pH -6.10784
M ~===;~~~=+~--3-,2-5-75-l--~~~=-~~
t=====;::=:~=+;--'
pHxF
~::::;::::==::;-I
dp I==::;;::::=:::;--.J
F
1==::;::::;---1
Co
!---....I
pHxdp

1.0 2.0 3.0 4.0 5.0 6.0 7.0


Estimated Effect

Fig. 1. Pareto chart of effects (at 90% of confidence level) for two-level fractional
factorial design (2 5•2).

Xz = (M - 40)/7 (2)
The biomass yield results obtained for the initial central composite
design varied from 0.7 (corresponding to the experiment at Xl = _2l/Z,
Xz = 0) to 1.56 mgl (g of dry substrate· d) (corresponding to the experi-
ment at Xl = +1, Xz = +1). The values used in the experiments for F, dp, and
Co were 2 Llh, 0.59 mm, and 0.4 giL, respectively. The experimental data
for yield as a function of the coded variables pH and M (Xl and Xz) could
be better fitted by a linear model (Eq. 3), indicating that yield tends to
increase with pH and M. The linear model was validated by ANOV A with
a confidence level of 95%, and the correlation coefficient value obtained
(R2 = 0.977) was considered satisfactory for this kind of experiment:
Yield = 1.187 + 0.21·Xl + 0.08·Xz + 0.04,Xl ,X2 (3)

Maximum Slope Trajectory


Making constant values for yield, expressions for Xl against X2 were
obtained from Eq. 3, which represent straight lines (with the same slope),
for each value of yield. From the central point (Xl = 0, X2 = 0) and going
perpendicularly to straight lines, the maximum slope trajectory was
obtained (13). In algebraic terms, this trajectory can be determined from the
ratio of Xzto Xl in Eq. 3, which corresponds to 0.38 in this case. Attributing
different values to X2 and Xli according to this ratio, five experiments were
carried out with pHs ranging from 7.8 to 11.2 and M from 41.6 to 48%. The
highest yield (1.47 mg/[g of dry substrate· d]) was achieved for the experi-
ment corresponding to pH 9.5 and M = 44.8%.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
408 Hansan et al.

:;::
-
co
-c 1.8
en
-c 1.4
C)

--
C
.Ci)
1.0
o 0.6
c .. 0.2

-
C)
E -0.2
f'o~
~
~~
+ I"'"~ ~
~~
~Co 1
~

Fig. 2. Response surface for yield obtained from Eq. 6 for real values of pH and M.

Second Central Composite Design


Based on the results obtained from evaluation of the maximum slope
trajectory, a second central composite design was developed for determi-
nation of the optimal region. The variables were tested at the levels _21/2,
-1,0, +1, and +21 /2, which correspond to pH 7.5, 8.1, 9.5,10.9, and 11.5, and
35.1,38.0,45.0,52.0, and 54.9 % for M, respectively. A total of 11 runs were
performed, similarly to the initial central composite design. The coded
values were calculated from Eqs. 4 and 5. The experimental results
obtained for yield varied from 0.74 (corresponding to the experiment with
Xl = -1, X2 = -1) to 1.46 mg/(g of dry substrate· d) (corresponding to the
experiment with Xl =0, X2 =0). The experimental data could be fitted with
a nonlinear model (Eq. 6) with the coded values of pH and M:
Xl = (pH - 9.5)/1.4 (4)

X2 = (M - 45)/7 (5)

Yield = 1.43 - 0.274·X12 + 0.057,X2 - 0.241,X22 (6)


The model was validated by ANOVA with a confidence level of 95%
and experimentally. The correlation coefficient value obtained (R2 =0.954)
was considered satisfactory for this kind of experiment. The results were
reproducible under controlled conditions within experimental error for
solid state fermentation.
Figure 2 shows the surface response for yield in terms of real values of
pH and M (calculated from Eq. 4 and 5). Thus, an optimum region was
determined with a maximum point of yield (1.43 mg/[g of dry substrate ·dD

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Fungi Allergens Produced by Solid-State Fermentation 409

A 1.2
12
1.0
"U
.,
fl! 0
>" 0.8 10
S'
0:- 3'
Ii 0.6

-
8 -3
3: ca
0.4 6 ca
0.2

0.0
4 -
0..
(n

o 50 100 150 200 250 300

B 58
200
180 /~
/"--'" 56

/ s::
160
en
~
ui
140
120
ti 54
--
0
i'
c
CiJ

-
100
J!: 52 '#.
80 +",--
60 +-+-+--+
--+-+-+-+ 50
40
0 50 100 150 200 250 300
Fennentation time (h)
Fig. 3. Time course of solid-state fermentation. (A) Water activity (-D-), polyphe-
nols (mM, -T-), YXIS (mg protein/mg glucose, -_-, and protein (mg/ g of dry
substrate [gds], -e-); (B) total reducing sugars (TRS) (mg glucose/gds, -+-),
reducing sugars (RS) (mg glucose/gds, -+-), and moisture (%,-6-).

obtained at pH 9.5 and 45.8% M, which corresponds to coded values of


X1= 0 and X2 = 0.116.
Fermentation Kinetics
Figure 3A, B shows the time course profiles for protein, polyphenols,
water activity, conversion efficiency (Y x / s), moisture, and sugars for the
fermentation carried out under the optimized conditions. The protein
profile shows the fungal growth. It may be considered that the log phase
starts after 48 h and the conidial phase appears after 192 h. Sugars were
consumed during fermentation but were not totally depleted. No signifi-
cant changes were obtained for water activity and moisture during fer-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


410 Hansan et al.

A MM tFO 48h 96b 144h 192b 240h 288b

kDa
94.0
67.0
43.0

30.0

20.0

14.4

B MM tA> 48b 96b 144b 192b 240h

kDa
212
170
116
76

53

Fig. 4. 50S-PAGE of allergenic extracts, revealed by silver strain. (A) Fifteen percent
polyacrylamide gel; (B) 10% polyacrylamide gel. MM, molecular marker; tf' fermenta-
tion time (h).

mentation, as expected. Polyphenol concentration increased with time in


a behavior similar to that of protein concentration. It was also observed
that YXIS decreased during fermentation. These results indicate an oxygen
limitation in the system, and, therefore, if maintenance requirements were
negligible, the main reduction in the yield parameter is owing to anaero-
bic conversion. In addition, the increase in polyphenols during fermenta-
tion indicates that polyphenols are probably products of anaerobic
conversion or maintenance metabolism.
Characterization of Allergenic Extract
Figure 4 shows the SDS-PAGE of extracts obtained with 0, 48, 96, 144,
192,240, and 288 h fermentation times. The lower molecular mass proteins

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Fungi Allergens Produced by Solid-State Fermentation 411

PI P2 P3 P4 PS P6 P? P8 c

EtOH,96h
dilution 1:4

SAm,240 h

SAm,96 h

EtOH, 240 h

EtOH, 0 h

Fig. 5. Dog-blotting analysis of allergenic extracts of D. onoceras (extracts obtained


by precipitation with EtOH or ammonium sulfate [SAm] at 0, 96, and 240 h of fermen-
tation ) for sera of patients (PI-P8) who had positive reactions in skin tests. C, control.

(Fig. 4A) were identified in an approximate range of 16.1-76.4 kDa. The


highest molecular mass proteins (Fig. 4B) were in the range of 103.3-157.3
kDa. In both cases, the complete protein spectrum appears after 96 h of
fermentation. The pool of proteins obtained in this work includes the mo-
lecular masses 14.4,36.0, and 60 kDa, identified by Menezes et a1. (2) as
being the most allergenic fractions produced by liquid fermentation.
Cutaneous reactivity tests (prick tests) using the allergenic extracts
were carried out in 33 atopic patients with asthma symptoms. Medium
reactions were observed in 40% of the cases. High reactivity was observed
in eight patients, and their sera were collected for dot-blotting analysis.
Figure 5 shows the dot-blotting analysis of allergenic extracts of D.
monoceras for sera of the patients with higher reactivities (P1-P8). Water
was used as the control in the experiments. These results showed good
agreement with prick tests, and it can be observed that in comparison to the
control, patients P1-P6 showed higher reactivity to allergenic extracts.

Conclusion

The production of D. monoceras biomass by solid-state fermentation is


possible and the allergenic extracts obtained were reactive in cutaneous
tests and dot blotting. The optimized process conditions are useful for
further scale-up studies. The kinetic behavior of biomass growth suggests
Applied Biochemistry and Biotechnology Vol. 105-108,2003
412 Hansan et al.
oxygen limitations, and the oxygen supply to the system and mass transfer
are important aspects to be studied in further work. The production of aller-
genic extracts under controlled conditions is reproducible, and the stan-
dardized extracts are valuable for the diagnosis and therapy of allergies.

References
1. Basomba, A (1982), in Purificaci6n y Estandardizaci6n de Alergenos, Departamento
Alergia de Abello, Madrid, Spain, pp. 91-98.
2. Menezes, E. A, Gambale, W., Macedo, M.S., Abdalla, D. s. P., Paula, C. R., and Croce,
J. (1995), Mycopathology 131, 75-81.
3. Mohovic, J., Gambale, W., and Croce, J. (1998), Allergol. Immunopathol. 16,397-402.
4. Pandey, A, Selvakumar, P., Soccol, C. R, Soccol, V. T., Krieger, N., and Fontana, J. D.
(1999), Appl. Biochem. Biotechnol. 81,35-52.
5. Pandey, A, Benjamin, S., Soccol, C. R., Nigam, P., Krieger, N., and Soccol, V. T. (1999),
Biotechnol. Appl. Biochem. 29, 119-131.
6. Yunginger, J. W., Jones, R T., Nesheim, M. E., and Geller, M. (1980), J. Allergy Clin.
Immunol. 66, 138-147.
7. Raimbault, M. and Alazard, D. (1980), Eur. J. Appl. Microbiol. 9, 199-209.
8. Moraes, R O. (1999), MS thesis, FEQ/UNICAMP, Sao Paulo, Brazil.
9. Bradford, M. M. (1976), Anal. Biochem. 72,248-254.
10. Miller, G. L. (1959), Anal. Chem. 31,426-428.
11. Saraiva, c.P. (2000), MS thesis, FEQ/UNICAMP, Sao Paulo, Brazil.
12. Price, M. L. and Butler, L. G. (1977), J. Agric. Food Chem. 25, 1268-1273.
13. Barros-Neto, B., Scarminio, I.S. and Bruns, RE. (1996), Planejamento e Otimiza~iio de
Experimentos, 2nd ed., Unicamp, Campinas, Brazil.
14. Laemmli, U. K. (1970), Nature 227,680-685.
15. Bio-Rad. (1998), Mini Protean® Electrophoresis Cell. Instruction Manual, Bio-Rad,
Hercules, CA
16. Morrissey, J. H. (1981), Anal. Biochem. 117,307-310.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0413 / $20.00

Hybrid Model for an Enzymatic Reactor


Hydrolysis of Cheese Whey Proteins by Alcalase
Immobilized in Agarose Gel Particles

Ruy SOUSA JR., MARIAM M. RESENDE,


RAQUEL L. C. GIORDANO, AND ROBERTO C. GIORDANO*
Departamento de Engenharia Qufmica,
Universidade Federal de Sao Carlos, Via Washington Luiz,
km 235, CP 676, Sao Carlos/SP, Brazil 13565-905,
E-mail address:roberto@deq.ufscar.br

Abstract
Cheese whey proteolysis, carried out by immobilized enzymes, can either
change or evidence functional properties of the produced pep tides, increas-
ing the potential applications of this byproduct of the dairy industry. Opti-
mization and scale-up of the enzymatic reactor relies on its mathematical
model-a set of mass balance equations, with reaction rates usually given by
Michaelis-Menten-like kinetics; no information about the distribution of
peptides' molecular sizes is supplied. In this article, a hybrid model of a batch
enzymatic reactor is presented, consisting of differential mass balances
coupled to a "neural-kinetic model," which provides the molecular weight
distributions of the resulting peptides.
Index Entries: Cheese whey proteolysis; enzymatic reactor; hybrid model;
mass balance equations; artificial neural networks.

Introduction
Reduction in the discharge of liquid protein residues generated in the
food industry process is a relevant concern (1). An interesting possible
solution to the problem is the conversion of those residues into market
products. Milky whey, arising from cheese manufacture, was considered,
for a long time, a product to be discharged. However, its high biologic
oxygen demand (35,000 mg/L) (2) and the associated treatment cost have
turned it into a byproduct of the food industry.
Whey and its products have been increasingly used in a great number
of applications such as for the production of creamer for foaming bever-
ages, edible food films, and milk and salt substitutes. On the other hand,
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 413 Vol. 105-108,2003


414 Sousa et al.
whey protein hydrolysis, carried out by enzymes immobilized in an inert
support, can either change or evidence functional properties of the pro-
duced peptides, increasing the potential applications of cheese whey. Whey
protein hydrolysates can be used as a protein source for individuals with
a reduced capacity for digestion, as a food supplement for phenylketonuria
patients, in compositions of bioactive peptides free from bitterness, and so
on (3).
Usually, the mathematical model of the enzymatic reactor where
whey hydrolysis takes place comprises a set of mass balance equations
with reaction rates given by Michaelis-Menten-like kinetics (4-6). Fol-
lowing this approach, substrate concentrations are expressed in terms of
unspecific variables such as the number of hydrolyzable peptide bonds in
the substrate. The drawback of this approach is that it does not supply any
information about the peptides' molecular weight distribution along the
reaction course. Only after time-consuming, off-line analyses can one
determine the detailed product composition.
In this article, a hybrid model of a batch enzymatic reactor is pre-
sented, consisting of differential mass balances coupled to a "neural-
kinetic model," which can provide the molecular weight distributions of
the resulting peptides. In this way, a more detailed description of the state
of the system is achieved.
Feedforward multilayer Perceptron (MLP), (7) neural networks (NNs)
are trained against empirical data (8). For network training, cheese whey
was hydrolyzed by alcalase®, multipunctually immobilized in agarose gel
particles, at different pH values. A laboratory-scale batch reactor, with pH
control, was used. Samples were periodically withdrawn and analyzed via
high-performance liquid chromatography (HPLC). The trained artificial
NNs were directly coupled to mass balance equations.

Materials and Methods


Four batch hydrolysis assays were carried out, at different pHs, in a
50-mL jacketed vessel with pH (Metrohm® 718 Stat Titrino) and tempera-
ture (thermostatic bath Neslab®) control. The operation temperature was
50°C. Reaction pHs were 7.0, 8.0, 9.0 and 10.0. For each experiment, 30 mL
of cheese whey at a nominal concentration of 60 gil (according to the
Kjeldhal method [91) and 0.3 g of agarose gel containing 13.4 UBAEE/ggel
were used. One UBAEE corresponds to the quantity of alcalase that hydro-
lyzes 1 ~mol of benzoil arginine ester etilic/min at pH 8.0 and 25°C.
During each assay, free-of-enzyme samples containing 300 ~L of hy-
drolyzed whey were taken. These samples, after properly diluted (15X)
were analyzed via size-exclusion chromatography (9) (with 0.25 M NaCI
in 0.02 M phosphate buffer, pH 7.2; flow rate of 4.17 x 10-9 m 3 /s and
detection at 214 nm).
The results of this chromatography analysis were used for training
feedforward MLP NNs. For network learning, the "back-propagation"
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Hybrid Model for Enzymatic Reactor 415
Table 1
HPLC Standards for Evaluation of Molecular Weight Distribution of Peptides
Standard MW Retention
no. Identification (Daltons) (min)
1 Bovine serum albumin 67,000 30.0
(Sigma, St. Louis, MO)
2 13-Lactoglobulin 18,000 33.5
(Sigma)
3 Insulin 5000 44.2
(Biobras, Montes Claros, Brazil)
4 Angiotensin II (Asp-Arg-Val-Tyr-Ile-His-Pro-Phe) 1047 65.0
(Sigma)
5 Leucine enkephalin (Tyr-Gly-Gly-Phe-Leu) 556 74.1
(Sigma)

2.4
:::r A MW67000 Da
g 2.0
T

MW18000 Da
MW5000 Da
c::: • MW 1047 Da
0
;; 1.6
• MW556Da
Unearfits
-=CDc:::
CJ 1.2
c:::
0
CJ
en 0.8
en
as
E 0.4

0.0
0.0 6.0x107 1.2x1OS 1.8x1OS
peak area (!J.V*min )

Fig. 1. Calibration curves: mass concentration of standard vs HPLC peak area.

algorithm was used (10). Finally, the classic Runge-Kutta method solved
the differential mass balances (11).

Results
Some calibration procedures were necessary for obtaining the pep-
tides' molecular weight distribution for each sample. Initially, five stan-
dards were injected into the column (Table 1). Then a calibration curve
mass concentration (giL) vs peak area (~V x min) was built for each stan-
dard (Fig. 1).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
416 Sousa et al.
hidden
layer

m1 --+ --+ , 1
m2 --+ --+ '2
m3 --+ --+ '3
m4 --+ --+ '4
m5 --+ --+ '5
biBs --+

Fig. 2. Feedforward MLP NN.

The linear coefficient of the curves in Fig. 1 was assumed constant and
equal to 0.03 giL «0 giL). The slopes, in turn, could be properly fitted as
a function of molecular weight (MW, Daltons) (Eq. 1):
Slope = -9.14 x 10-9 + 5.03 x 1O-9 1og 1o (MW) (1)
Molecular weights could be expressed as a function of the retention
time (retention, min) inside the column, as presented in Table 1:
loglO (MW) = 5.40 - 0.04 retention (2)
By combining Eqs. 1 and 2, a general equation was obtained that
relates mass concentration (Cone, giL) to the area in the chromatogram
(area, !A.V x min), as a function of retention time (min):
Cone = 0.03 + (-9.14 x 10-9 + 5.03 x 10-9[5.40 - 0.04 x retention]) x area (3)
Through HPLC analysis, it was possible to obtain the peptides' con-
centrations, for each sample, within five predefined ranges of interest
(MWI :s 650 Daltons, 650 Daltons < MW2 :s 1050Da, 1050 Daltons < MW3 <
4150 Daltons, 4150 Daltons :s MW4 < 14,000 Daltons, 14,000 Daltons :s MWs
:s 67,000 Daltons).
Considering each of the five molecular weight ranges as a pseudo-
component, feedforward MLP NNs were trained. The NNs were capable
of mapping the mass concentrations m1 (MWI :s 650 Daltons), m2 (650
Daltons < MW2 :s 1050 Daltons), m3 (1050 Daltons < MW3 < 4150 Daltons),
m4 (4150 Daltons s MW4 < 14,000 Daltons) and ms (14,000 Daltons s MWs
:s 67,000 Daltons) into an output vector constituted by the reaction rates
'2t
of the respective MW ranges, '1' '3' '4, and 's'
The feedforward NN (Fig. 2) comprises interconnected layers of pro-
cessing units (neurons). Processing units in adjacent layers are joined by
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Hybrid Model for Enzymatic Reactor 417
Table 2
Mass Concentrations of Each Range
of Peptides' MW (pH 10.0) for Input to NN
m1 (giL) m2 (g/L) m3 (g/L) m4 (g/L) ms(g/L)
54.22 0.00 0.00 0.00 0.00
47.44 3.43 1.86 0.88 0.62
34.96 10.46 5.82 1.92 1.06
26.63 13.51 8.17 2.64 3.27
24.17 12.54 8.98 2.97 5.56
20.92" 12.41 10.02 3.38 7.50
19.05 12.31 11.06 3.82 7.98
8.20 14.02 14.72 5.41 11.86
5.22 14.17 15.52 6.13 13.19
4.87 12.64 16.37 7.01 13.33
3.77 11.77 16.33 7.60 14.75
2.94 11.32 16.68 8.25 15.03
2.47 10.56 16.57 8.50 16.12
aOne validation datum point.

weighed connections (wji ). Each unit sums the input weighed signal (WjiX i )
and an offset term (a bias wjb ).
" .. x. + W'b
net.J =i=1
~w (4)
JI I J

A nonlinear transfer function (in our case, a sigmoid) evaluates the


node output (Yj) using the value of netj , obtained from Eq. 4:

(5)

The experimental reaction rates ri were obtained after the direct dif-
ferentiation of mi x time by the proper "calculus tool" of the software
Microcal Origin®. Tables 2 and 3 show typical data (with experimental and
predicted rates). The dispersion ofNN learning can be assessed by exam-
ining Fig. 3A. Figure 3B shows the dispersion of a validation test using
extra data. The numbers of neurons in the hidden layer are 40 (pH 7.0),
45 (pH 8.0), 48 (pH 9.0) and 48 (pH 10.0).
The mass balance equation for the batch stirred-tank reactor (BSTR) is
as follows:
d(m i v)
---=r.pV (6)
dt I

in which m i is mass concentration within range i, V is reactor volume


(admitted constant), t is time, rj is reaction rate supplied by the neural-
kinetic model (within range i), and pis UBAEE/V.
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
).
:g
if
Q..
OJ
0'
,..,
::r-
(])
3 Table 3
tn'
~ Enzymatic Reaction rates (pH 10.0) for Output of NN
til
::J
Q..
OJ
r1 r2 r3 r4 rs
0' (g/UBAEE·min) (g/UBAEE·min) (g/UBAEE·min) (g/UBAEE·min) (g/UBAEE·min)
<b
,..,
::r-
::J Experimental NN Experimental NN Experimental NN Experimental NN Experimental NN
0
0-
Otl
'<: -0.0759 -0.0759 0.0383 0.0383 0.0208 0.0208 0.0098 0.0098 0.0069 0.0069
-0.0554 -0.0554 0.0290 0.0290 0.0160 0.0160 0.0064 0.0064 0.0041 0.0041
-0.0252 -0.0253 0.0127 0.0127 0.0077 0.0077 0.0021 0.0021 0.0027 0.0027
~
-0.0093 -0.0093 0.0023 0.0022 0.0027 0.0027 0.0009 0.0009 0.0035 0.0035
co -0.0038 -0.0038 -0.0007 -0.0007 0.0012 0.0012 0.0005 0.0005 0.0028 0.0028
-0.0033 -0.0026 -0.0002 -0.0001 0.0013 0.0011 0.0005 0.0005 0.0016 0.0010
-0.0026 -0.0026 0.0002 0.0002 0.0011 0.0011 0.0005 0.0005 0.0008 0.0008
-0.0020 -0.0020 0.0003 0.0003 0.0007 0.0006 0.0003 0.0003 0.0008 0.0008
-0.0005 -0.0005 -0.0002 -0.0001 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002
-0.0001 -0.0002 -0.0002 -0.0002 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
-0.0001 -0.0001 -0.0001 -0.0001 0.0000 0.0000 0.0001 0.0001 0.0001 0.0001
-0.0001 -0.0001 -0.0001 -0.0001 0.0000 0.0000 0.0001 0.0001 0.0001 0.0001
0.0000 -0.0001 -0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0001 0.0001
~
:-

~
a
,Cx:l
N
§
Hybrid Model for Enzymatic Reactor 419
o.
A

-0.03 ·(102 -0.01 0.00 0.01 0.02 -0.015 -0.010 -0.005 0.000 0.005 0.010

14000 Da ¢ MW < 67000 Do


4150 Da<= MW<14000 Do
1050 Da< MW< 4150 00
650 Da< MW<= 1050 Do
MW<= 650 III

-0.06 -0.04 -0.02 0.00 0.02 0.04 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04
experimenlal reaction rates experimental reaction rates
(g J UBAEE min) (g J UBAEE min)

B
0.01

~
I:
,gu -0.01 14000 Da <= MW < 67000 Da [l
4150 Da<= MW< 14000Dao
m 1050Da< MW< 4150Da~
z 650 Da< MW<= 10500a'>.7
z /~J' MW<= 6500ao
"
-0.03V+-"----.,----r-----.----r---.-----,
-0.030 -0.015 0.000 0.015
experimental reaction rates (9 I USAEE min)

Fig. 3. (A) Dispersion of NN learning; (B) dispersion in a validation test (pH = 9.0,
T= 50°C).

By making use of the classic Runge-Kutta method, the system of


ordinary differential equations could be properly solved, providing the
hydrolysate composition inside the jacketed BSTR along time. Initial con-
ditions were as follows:
m1(0) = 63.55 giL for pH 7.0 assay, 57.78 giL for pH 8.0 assay,
57.01 giL for pH 9.0 assay, and 54.22 giL for pH 10.0 assay; and
m2(0) = m3(0) = miO) = ms(O) = 0 giL for all assays.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
420 Sousa et al.
70 14000 Da <= MW < 67000 Da::J experimental--model
4150 Da <= MW < 14000 Da 0 experimental-------model
1050 Da < MW < 4150 Da 6 experimental ----- - model
60 650 Da < MW <= 1050 Da '\l exper~mental------model
....
::J \ MW <= 650 Da 0 expenmental------model
S 50
I::
0 o

-
\u
~ 40 u
I ::
Q)
u 30 o
I:: n
8
:g 20
o
CIS ~
::::il
10

0
~~~,,:':;=:~;:"==!
o 50 100 150 200 250 300
time (min)

Fig. 4. Distribution of peptides along time (pH = 7.0, T = SODC).

14000 Da <= MW < 67000 Da::J experimental-----model


60 4150 Da <= MW < 14000 Da 0 experimental-------model
1050 Da < MW < 4150 Da 6 experimental--- ----model
:J' 50 650 Da < MW <= 1050 Da'\l experimental-------model
S MW <= 650 Da 0 experimental-----model
I::
40

-
.Q
~
I ::
~ 30
I::
8
Ul 20
gj
::::il
10

o
o 50 100 150 200 250 300
time (min)

Fig. 5. Distribution of peptides along time (pH = 8.0, T = SODC).

Note that m1(0) is not exactly the same for all experiments. The average
of the obtained values is 58.14 giL. This is very close to the nominal cheese
whey concentration determined through the Kjeldhal method. Figures
4-7 show the model predictions compared to experimental data, for each
of the experimental assays.
Applied Biochemistry and Biotechnology Vol. 705-70B, 2003
Hybrid Model for Enzymatic Reactor 421

60 14000 Da <= MW < 67000 Da:J experimental----model

-..
4150 Da <= MW < 14000 Da 0 experimental-------model
1050 Da < MW < 4150 Da 8 experimental- -- -- -- model
:J 50 650 Da < MW <= 1050 Da v experimental------model
S MW <= 650 Da <> experimental--------model
r::

-
0
40
~
r::
~ 30
r::
8
I
::!!
20

10

o 50 100 150 200 250 300


time (min)

Fig. 6. Distribution of peptides along time (pH = 9.0, T = SO°C).

60
14000 Da <= MW < 67000 Da:J experimental--model
4150 Da <= MW < 14000 Da 0 experimental-------model

-
50 1050 Da < MW < 4150 Da 8 experimental-model
:J 650 Da < MW <= 1050 Da v experimental-------model
S MW <= 650 Da <> experimental----model
r:: 40
0

-
',1:1
~
r:: 30
CD
0
r::
8 20
Ul
Ul
III
::::iii
10

o 50 100 150 200 250 300


time (min)

Fig. 7. Distribution of peptides along time (pH = 10.0, T = SO°C).

Discussion
Using HPLC and an appropriate calibration procedure, it was pos-
sible to quantify the distribution of peptides' molecular weights along
time for different experiments of cheese whey hydrolysis. Feedforward
MLP NNs can map accurately the reaction rates as a function of peptides'
molecular weight distribution, at different pHs. For intermediary pH e.g.,
9.5, it is possible to use direct interpolation.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
422 Sousa et al.
It was possible to observe that the model predictions for molecular
weight distribution of peptides are quite accurate. In this work, the trained
NNs were used for simulating the BSTR itself. They can be used for mod-
eling any other kind of reactor utilized in whey enzymatic hydrolysis. In a
different work (12), by coupling a reactor modeling presented by Giordano
et al. (13) and the neural-kinetic model presented here, we built up a hybrid
model in order to represent whey proteolysis in a continuous vortex flow
reactor. Combining mathematical models with artificial NNs seems to be
an important trend in bioprocess modeling (14-16).

Acknowledgments
We thank Funda~ao de Amparo a Pesquisa do Estado de Sao Paulo,
Conselho Nacional de Desenvolvimento Cientilico e Tecnol6gico, Programa
de Apoio ao Desenvolvimento Cientifico e Tecnol6gico/Conselho Nacional
de Desenvolvimento Cientilico e Tecnol6gico, and Cooperativa de Lactidnios
Sao Carlos (Brazil).

References
1. Viotto, W. H. (1993), PhD thesis, Unicamp, Campinas, Brazil.
2. Demetrakakes, P. (1997), Food Processing 58, 75-79.
3. Mann, E. (2000), Dairy Ind. Int. December, 13-14.
4. Segel, I. H. (1975), Enzyme Kinetics, a Wiley-Interscience, New York, NY.
5. Svendsen, I. (1976), Carlsberg Res. Commun. 41,237-291.
6. Adler-Nissen, J. (1986), Enzymic Hydrolysis ofFood Proteins, Elsevier Applied Science,
London, England and New York, NY.
7. Scarselli, F. and Tsoi, A. C. (1998), Neural Networks 11, 15-37.
8. Medler, D. A. (1998), Neural Comput. Surv. 1,61-101.
9. Silvestre, M. P. C. (1997), Food Chem. 60,263-271.
10. Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986), Nature 323, 533-536.
11. Press, W. H., Teukolsky, S. A., Vetterling, W. T., and Flannery, B. P. (1996), Numerical
Recipes in Fortran 90: The Art of Parallel Scientific Computing, Cambridge University
Press, Cambridge, NY.
12. Resende, M. M., Sousa, R., Jr, Tardioli, P. W., Giordano, R. L. c., and Giordano, R. C.
(2002), submitted.
13. Giordano, R. c., Giordano, R. L. c., Prazeres, D. M. F., and Cooney, C. L. (2000), Chem.
Eng. Sci. 55,3611-3626.
14. Liibbert, A. and Simutis, R. (1994), Trends Biotechnol. 12,304-311.
15. Azevedo, S. F., Dahm, B., and Oliveira, F. R. (1997), Comput. Chem. Eng. 21, S751-S756.
16. James, S., Legge, R., and Budman, H. (2002), J. Process Control 12, 113-121.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0423/$20.00

Preliminary Investigation of Fungal


Bioprocessing of Wheat Straw for Production
of Straw-Thermoplastic Composites

DAVID N. THOMPSON,* TRACY P. HOUGHTON,


JEFFREY A. LACEY, PETER G. SHAW, AND J. RICHARD HESS
Biotechnology Department,
Idaho National Engineering and Environmental Laboratory,
PO Box 1625, Idaho Falls, 1083415-2203, E-mail: thomdn@inel.gov

Abstract
Straw utilization for composites is limited by poor resin and polymer
penetration, and excessive resin consumption owing to the straw cuticle,
fines, and lignin-hemicellulose matrix. White-rot fungi degrade these com-
ponents of straw and could, therefore, potentially be used to improve resin
penetration and resin binding without the use of physical or chemical pre-
treatments. Although long treatment times and large footprints the limit use
of fungal treatments on a large scale, distributed fungal pretreatments could
alleviate land requirements. In this article, we present progress toward the
development of a passive fungal straw upgrading system utilizing white-
rot fungi.
Index Entries: Fungal upgrading; white-rot fungi; wheat straw; Pleurotus
ostreatus; straw composite.

Introduction
The principal barriers to straw utilization for straw-thermoplastic
composites are resin penetration and resin consumption. Resin penetration
is limited by the physical barrier presented by the cuticle and underlying
epidermis on the external surface of the straw, and by the lignin-hemicel-
lulose matrix in the internal vascular layer. Resin consumption is increased
because the resin does not bind well to the cuticle, and because fines, cre-
ated when the nodes and leaves are ground, have high surface areas and
require much more resin. Since straw thermoplastics can contain as much
as 50-70 wt% straw, resin binding and performance are of the utmost
importance, as has been shown in wood-plastic composites (1,2).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 423 Vol. 105-708,2003


424 Thompson et al.

The physical barriers to resin penetration are the same physical barri-
ers that limit access of cellulase enzymes to the cellulose fibers when pro-
ducing glucose from straw cellulose for fermentation to ethanol or for
production of fuels and chemicals. Much of the research on removal of the
lignin-hemicellulose barrier to date has been on conversion of the cellulose
to glucose, since glucose can easily be fermented to ethanol using common
yeasts (3). Dilute-acid hydrolysis of the cellulose to glucose lowers ferment-
able carbohydrate yields due to thermal decomposition of xylose to fur-
fural and glucose to hydroxymethylfurfural (4). Thus, much work has been
done on the use of cellulase enzymes, since they are specific for cellulose,
form only glucose, and the hydrolysis is performed at mild temperatures.
However, cellulases are relatively large enzymes and cannot fit through
most of the spaces in the intact vascular layer of straw (5,6). Thus, physical,
chemical, and thermal pretreatments are employed to increase the penetra-
tionofcellulases into this lignocellulose matrix (3,5,7). Manypretreatments
have been developed, including acids (8,9), alkalis (10,11), organosolvents
(12), steam explosion (10,13), and physical treatments. Although effective,
the pretreatments are costly, negatively affecting the economics of utiliza-
tion. Pretreatments with white-rot fungi, which have been shown to com-
pletely degrade lignocellulose, increase glucose yields without significant
capital or energy-intensive steps (14). The principal drawbacks to central-
ized white-rot fungal pretreatments are that the process footprints are large
and that treatment times are often 8 wk or longer, much too long for use at
large industrial facilities (14).
Since lignin and hemicellulose limit resin penetration just as they
limit cellulase penetration into the matrix, these pretreatments would be
expected to improve resin penetration as well. Indeed, steam explosion
has been investigated to remove lignin and hemicellulose to allow better
resin penetration and adhesion (15). While significant improvement in
interfacial adhesion was seen in steam exploded broom fibers, the exten-
sive physical damage to the fibers imparted by the steam-explosion pro-
cess eliminated any mechanical property improvements. Of course, steam
explosion pretreatment, whether for better resin penetration or for better
penetration of cellulases, requires significant capital equipment and
energy input. Thus, an inexpensive, low-capital, low-energy-input treat-
ment, such as a fungal pretreatment, that removes the cuticle, lignin, and
hemicellulose would be ideal. Physical removal of straw components that
form fines (16) would also help to reduce resin consumption. Combining
physical removal of fines (16) and fungal treatment into a distributed
process that could be done on-site at a small scale would minimize the
land area required as well as capital and energy inputs.
White-rot fungi remove lignin using extracellular peroxidase
enzymes, attacking the lignocellulose matrix by growing into the cell
walls, where they secrete extracellular cellulases, hemicellulases, and per-
oxidases (17). Ligninolytic enzymes are produced in secondary metabo-
lism under conditions of carbon or nitrogen deprivation (17). The
Applied Biochemistry and Biotechnology Vol. 105-708,2003
Fungal Bioprocessing of Wheat Straw 425
degraded lignin is not used as a growth substrate but is removed to open
up the matrix to cellulases and hemicellulases so that over time near-
complete degradation is possible (18). While cellulose and hemicellulose
are the principal growth substrates for white-rot fungi (17), some white-
rot fungi, including several Pleurotus species, attack straw lignin and
hemicellulose without much cellulose removal (19,20). There is also evi-
dence of degradation of the cuticle during degradation by white-rot fungi
(21). Once through the cuticle, the fungi possess the necessary cellulases
and hemicellulases to degrade the epidermal layer, allowing access to the
vascular layer from the outer surface of the residue. Thus, in a single
degradation step, white-rot fungi could potentially upgrade the straw to
a more desirable feedstock for straw-thermoplastic composites. That is,
the upgraded straw product should have a higher cellulose content, and
partial degradation of the vascular hemicellulose and lignin, the cuticle
layer, and the epidermis should allow better penetration by resins used
in thermoplastic extrusion processes.
This article describes preliminary results from ongoing work at the
Idaho National Engineering and Environmental Laboratory (INEEL) to
bracket the process parameters necessary to reproducibly operate a pas-
sive, distributed, fungi-based straw "bioupgrading" system. These data
will be used by INEEL, together with Washington State University, to
devise and build a pilot-scale fungal bioupgrading system suitable for
use in both centralized and distributed systems. Included in this work are
the preliminary results of ongoing exploratory tests conducted at INEEL
to bracket the "best" conditions of inoculum and moisture for fungal
upgrading of the straw. Although preliminary, the results show that by
limiting nitrogen and providing sufficient inoculum, it is possible to
operate a selective fungal degradation system without prior sterilization
of the wheat straw. The pilot-scale design, as well as testing of the fungal-
treated straw in composite materials, is being implemented at Washing-
ton State University.

Materials and Methods


Wheat Straw
Wheat straw (Westbred 936), a hard red spring variety, was obtained
from Grant 4-D Farms (Rupert, ID). All the straw utilized was produced
during the year 2000 cropping season. Twenty large bales of Westbred 936
(1.2 x 2.4 m [4 x 8 ft] bales) were produced and stored in a stack at the side
of the field at Grant 4-D Farms. Only the protected center bales from the
interior of the stack were used for the studies. This was done to minimize
the effects of exogenous nitrogen sources (i.e., bales touching the soil) and
water (i.e., bales on the outer periphery of the stack), which were important
variables because we intended to limit nitrogen and vary water in the fun-
gal degradation tests. To better handle the straw for the laboratory studies,
the large bales were rebaled as needed into smaller 0.61 x 1.2 m (2 x 4 ft)
Applied Biochemistry and Biotechnology Vol. 105-108,2003
426 Thompson et al.
Table 1
Composition of Westbred 936 Straw Stem Fraction
Used in Fungal Treatment Studies
Composition
(wt%)a

Glucan 37.2
Xylan 22.1
Galactan 1.2
Mannan 3.0
Arabinan 1.6
Lignin with extractives 18.9
Ash 10.1
Total b 89.7
aBased on 100% dry wt of material.
bRemaining fraction attributed to unknown contents of uronic acids,
protein, and so on and to recovery errors in analysis techniques.

bales containing about 22.7 kg (50 lb) each and placed in covered storage.
To remove the plant components that are the sources of high-surface-area
fines (leaves, sheaths, nodes, and fines), the straw was rethreshed before
use as described by Hess et a1. (16). Only the separated straw stems were
used in the laboratory studies. The composition of the untreated straw stem
fraction, determined as described under Compositional Analysis, is shown
in Table 1.
Cultures and Maintenance
Pleurotus ostreatus NRRL 2366 was chosen for use in the fungal degra-
dation tests based on its ability to selectively degrade the noncellulose com-
ponents of wheat straw (22,23). It was obtained from the Northern Regional
Research Laboratory (NRRL) (Peoria, IL). Stock cultures were maintained
on agar slants at room temperature prepared at 20 giL of YM agar (Difco,
Detroit, MI) and containing the following trace minerals- 0.02 giL of
FeS04 ·7H20, 0.004 giL of CuS04 ·5H20, 0.002 giL of ZnS04 ·7H20, 0.002 gl
L of MnS04 ·H20, and 0.001 giL of ammonium molybdate tetrahydrate-
and were subcultured every 2 wk. Stock mycelial inocula were produced as
follows. Fungal mycelia were transferred from the maintenance slants to
100 mL of 20 giL YM broth (Difco) using a sterile loop and grown in agitated
culture for 2 to 3 d at room temperature and 180 rpm. This culture was
transferred to a sterile Fernbach flask containing 1 L of 20 giL YM broth with
trace minerals as just described, and agitated for 4 d at room temperature
and 180 rpm. The fungal pellets were harvested by light centrifugation (380g)
in sterile centrifuge bottles, transferred to sterile bottles with sufficient spent
medium to submerge the pellets, and stored at 4°C until use, typically 2 to
3 wk or less.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Fungal Bioprocessing of Wheat Straw 427
Nitrogen Consumption Experiments
Since it would be uneconomical to sterilize the straw for use in distrib-
uted systems, unsterilized straw is preferred, and thus the inoculated
P. ostreatus must be able to outcompete or overtake the indigenous micro-
bial population. White-rot fungi are late colonizers in nature and dominate
under conditions of nitrogen limitation (24). Thus, minimizing the amount
of nitrogen carried over to the straw in the inoculum could allow the inocu-
lated fungi to more quickly overtake the indigenous microbes, minimizing
or eliminating the need for sterilization. The nitrogen consumption tests
were designed to estimate the initial nitrogen level required to produce a
suitable amount of biomass for inoculating the straw while minimizing the
amount of nitrogen remaining in the culture medium.
In these experiments, approx 500 mL of wetfungal pellets of P. ostreatus
were transferred to a sterile blender, and 500 mL of medium was added.
Media tested included YM broth with trace minerals (as described earlier),
and a nitrogen-limited medium containing 20 giL of glucose, 1.0-3.5 giL
of yeast extract, and trace minerals (as described earlier). The mixture was
blended for 2 min, producing a slurry of finely chopped mycelial frag-
ments. The optical density (OD) at 550 nm was determined for dilutions of
this slurry using a standard UV IVIS spectrophotometer. The undiluted
slurry was then inoculated to 1.0 OD into fresh nitrogen-limited medium
in sterile shake flasks and incubated for 14 d at 30°C, 135 rpm. Replicate
flasks were sacrificed periodically, and the fungal pellets were washed
with distilled water and separated from the liquid by centrifuging for 10
min at 26,892g. Fungal biomass was measured gravimetrically after drying
for 48 h at 105°C. Total Kjeldahl nitrogen (TKN) was measured as previ-
ously described (25).
Column Inoculum Preparation
The most suitable conditions for minimization of nitrogen remaining
in the culture and production of biomass in the inoculum were used
(see Results). The yeast extract concentration used in the nitrogen-limited
medium for column inoculum production was 3.0 giL. The column inocu-
lum cultures were cultured and harvested in a manner identical to that
used for the nitrogen consumption experiments just described, except that
the cultures were harvested between d 5 and 7, and the mycelial pellets
were homogenized in the spent medium, without centrifugation.
Column Experiments
Because we intended to standardize the inoculum, it was desirable
to know the exact amount of fungal inoculum added to the straw. In
addition, it was desirable to minimize heterogeneity in the treated straw
by initially evenly distributing the fungus over the entire surface of the
straw. To accomplish this, P. ostreatus was added to the straw by spraying
a liquid suspension of homogenized mycelia onto the straw with mixing.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
428 Thompson et al.
Assuming the availability of a viable stabilized homogenized mycelial
inoculum, this would also be a more cost-effective method for inoculating
straw piles in the field than using a tumbler to mix solid inoculum into
untreated straw. However, because spraying is not a commonly used
inoculation method, there was no obvious initial range of fungal inocu-
lum to test. Since 10 wt% of solid inoculum is often used in soil bio-
remediation to inoculate with white-rot fungi using the solid inoculation
method (Dr. S. D. Aust, December 2001, personal communication, Utah
State University), we arbitrarily assumed that the solid inoculum con-
tained 1% of its weight as dry fungal biomass and chose the range of 0-
1 wt% dry fungal biomass for the initial column experiments.
The fungal pellets in the inoculum cultures, produced immediately
before inoculation of the straw columns, were transferred with the spent
medium to a sterile blender and blended for 2 min. The OD at 550 nm was
determined for dilutions of this slurry and the concentration of biomass
was estimated from a previously measured calibration. The undiluted
slurry was transferred to a sterile standard hand-pump garden sprayer for
addition to the straw stems. No extraordinary measures were taken beyond
this point to maintain sterility, except for using initially sterile equipment.
Air-dried straw stems (150 g dry wt) were spread onto a clean, dry
tray in a thin layer, and the inoculum was sprayed onto the stems, with
frequent mixing of both the inoculum and the stems. Enough inoculum
was added to reach the desired initial level of fungal inoculum in the
stems. Periodically during addition of inoculum, a fan was used to blow
air across the tray of inoculated straw to evaporate excess water, with
frequent mixing of the straw. After the desired amount of inoculum was
added, additional sterile distilled water was sprayed onto the straw as
needed to reach the desired initial moisture content for the particular
experiment. A separate sample of the well-mixed inoculum slurry was
then added to a tared bottle and dried at 105°C to determine the actual
biomass concentration of the slurry. The initial moisture range to be tested
was chosen based on empirical evidence from compost biofiltration and
soil bioremediation using white-rot fungi. First, a gravimetric moisture
range of up to 70% (41 % on a wet basis) has worked very well in compost
biofiltration experiments degrading volatile organic compounds (VOCs)
at the INEEL (26,27). Second, it has been shown that white-rot fungi seem
to grow and compete best with indigenous microbes in soil at moisture
levels at or below 0.5 g of H 20/ g dry soil (28). Thus, since it was desired
to have the inoculated fungus successfully compete with the indigenous
microbes, the initial gravimetric moisture range was chosen to bracket
these values, atOAO-0.70 g of H 20/ g of dry stems (40-70% ona dry basis).
The inoculated straw was added to clean, initially sterile glass col-
umns fabricated from glass process pipe. The airtight columns were com-
prised a 12-in. section of 3in. id Pyrex process pipe with 2in. id reducers at
each end (ACE Glass, Vineland, NJ), and Teflon® end caps. The columns
were prepared in triplicate with approx 50 g dry wt of inoculated stems in

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Fungal Bioprocessing of Wheat Straw 429
each column. The loaded columns were supplied with humidified oil-free
instrument air at 15.5 psig and a flow rate sufficient to turn over the air in
the system once per day. Approximately 2.5 g dry wt of straw was sampled
from the top and bottom of each column initially and approximately every
3 to 4 wk thereafter. The samples were dried at 105°C overnight and ground
to 60 mesh in a Wiley mill for compositional analysis.

Compositional Analysis
Carbohydrate and lignin compositions of untreated and treated straw
samples were determined by quantitative saccharification using the
method of Saeman et al. (29). Two aliquots of each sample were analyzed
per column by quantitative saccharification for each set of three replicate
columns at each condition, for a total of 12 independent measurements of
each composition. Carbohydrate analyses were done by high-performance
liquid chromatography using a Bio-Rad HPX-87P carbohydrate column,
and lignin was calculated by weight difference as Klason lignin with extrac-
tives and ash, as previously described (6).

Determination of Dominance of Inoculated Fungus


Pleurotus species have been shown to be somewhat selective for hemi-
cellulose and lignin rather than cellulose (22,30), including P. ostreatus
(22,23). Because a mixed culture of indigenous microbes can degrade the
most easily accessible fractions of both cellulose and hemicellulose, the
indigenous microbes would not be expected to show significant selectiv-
ity for specific polysaccharides. Thus, the mass balances of cellulose and
hemicellulose could potentially be used to determine whether the inocu-
lated P. ostreatus was dominant, by comparing the relative amounts of
degradation of cellulose and hemicellulose. Thus, a column that showed
similar degradation of both cellulose and hemicellulose (cellulose/hemi-
cellulose ratio similar to that for undegraded straw) was said to be domi-
nated by indigenous microbes. Conversely, a column that showed
significantly greater hemicellulose degradation than cellulose degrada-
tion (significantly increased ratio of cellulose/hemicellulose) was said to
be dominated by the inoculated P. ostreatus. The performance measures
chosen for the inoculum competition column tests were the relative
changes in cellulose and hemicellulose after degradation (expressed
as the cellulose/hemicellulose ratio), as well as overall hemicellulose
removal. Thus, cellulose, hemicellulose, lignin/ extractives, and ash were
measured, and the mass balances from d 0 to the sample date were com-
pared. It is true that lignin removal could be used separately or together
with cellulose/hemicellulose ratio as an indication of P. ostreatus activity.
However, routine measurement of extractives would be necessary to dis-
cern changes in lignin and extractives. Since the cellulose/hemicellulose
ratio was believed to be a good indicator of the relative activity of the
inoculated white rot fungus vs the indigenous microbes, it was decided
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
430 Thompson et al.
C;; 600r-;:::===:::::;--------,
.5. _ Stationary 1--______________---1

I~ -0- Agitated
400.l-~===~--~~:::::::::t

..a
iii

200+-------~~------------------~

~
~
c 0 D'I=::::::::;:...--r-----,~___r-....,..-__r-_I
o 2 4 6 8 10 12 14
Time (days)

Fig. 1. Comparison of P. ostreatus biomass production in stationary and agitated


cultures. This test used the YM broth/trace minerals medium.

not to measure the extractives for treated straw samples until the final
conditions were chosen. Thus, lignin is expressed as "lignin with extrac-
tives" in this article. The extractives will be determined by soxhlet extrac-
tion for the final conditions chosen.

Results and Discussion


Inoculum Production and Nutrient Carryover
Since it would be uneconomical to sterilize the straw for use in dis-
tributed systems in the existing infrastructure, and also on such a large
scale, it is imperative that the system treat only unsterilized straw. The
inoculum preparation tests were done to estimate the initial nitrogen level
required to produce a suitable amount of biomass for inoculating the
straw while minimizing the amount of nitrogen remaining in the culture
medium. These tests were necessary because the slow-growing white-rot
fungi are late colonizers in nature and express their ligninolytic systems
under conditions of nitrogen limitation (24). Thus, minimizing the amount
of nitrogen that carries over to the straw in the fungal inoculum could
allow the inoculated fungus to more quickly overtake the microbes al-
ready present in the straw, and thereby minimize or eliminate the need for
sterilization of the straw.
Experiments were performed initially using the YM broth/ trace min-
erals medium, with inoculum amounts ranging from 0.1 to 5.0 OD. Tests
were run to compare the production of fungal biomass in stationary vs
agitated cultures (see Fig. 1). As expected, biomass production in the agi-
tated cultures far exceeded that in the stationary cultures, by more than
four times. The TKN of the cultures was monitored over the course of the
experiments, but it did not change significantly even when the cultures
entered secondary metabolism (data not shown). This indicated that the
cultures were not nitrogen limited.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Fungal Bioprocessing of Wheat Straw 431

..... 1.0g/L -0- 2.5 giL


400 -0- 1.5 giL ...... 3.0 giL
..... 2.0 giL -&- 3.5 giL

-
:J 300
Cl
§.
z 200
~
I-

100

0
0 2 4 6 8 10 12 14
Time (days)

Fig. 2. Effect of yeast extract addition (which contains about 10 wt% nitrogen) on
minimum nitrogen level attainable in inoculum production cultures.

Cultures were then tested in the nitrogen-limited medium, inoculat-


ing homogenized mycelia to 1.0 00. The minimum nitrogen concentration
observed in the cultures was about 80-100 ppm of TKN (Fig. 2). The first
sample was taken on d 5 to ensure that a sample would be obtained before
the minimum nitrogen concentration was reached (onset of stationary
phase), because the original agitated YM broth-grown cultures entered
secondary metabolism on or about d 7-10, depending on the amount of
inoculum (not shown). As shown in Fig. 2, the minimum nitrogen levels
were observed in the first sample on day 5, after which the TKN of the
culture fluid increased. The increase in medium nitrogen indicates either
exportation of nitrogen from the cells to the medium (as extracellular
enzymes), or significant cell death. The shift to earlier onset of stationary
phase is consistent with the switch to a more severe nutrient limitation in
the new medium. Biomass production did not increase significantly going
from 3.0 to 3.5 giL of yeast extract in the culture, and thus it was decided
to produce inoculum for the column tests using 3.0 giL of yeast extract in
the nitrogen-limited medium. Use of this medium resulted in the produc-
tion of 5300-6700 mglL of P. ostreatus mycelia in 5 d (data not shown).
Column Tests
Exploratory experiments were performed to estimate the minimum
amount of inoculum necessary to overtake the indigenous microbes in the
straw in 12 wk or less. For these tests, the amount of fungal inoculum was
varied from 0 to 11 mg of homogenized P. ostreatus mycelial g of air-dried
straw stems. The moisture content used for these experiments was 60-70%
on a dry basis (0.6--0.7 g of H 20 I g dry stems), based on prior experience at
INEEL using compost biofilters to degrade VOCs (26,27). A comparison of
the degradation patterns for each amount of inoculum is shown in Fig. 3.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
432 Thompson et al.

40

30

wt% 20

10

o
o
1.6
mg p 5.1
. OS/reatus 19 8 .2 10.9
stems

Fig. 3. Straw stem compositions at d 0 and 84 as function of fungal inoculum amounts


from 0 to 10.9 mg of P. ostreatus/ g stems.

Few differences were seen between the control (0 mg of P. ostreatus/g)


and inoculated columns containing <5.1 mg/ g. Above 5.1 mg/ g, relatively
less cellulose was degraded with increasing amount of inoculum. At 10.9
mg/ g, little cellulose was degraded, suggesting dominance of the cultures
by P. ostreatus. Lignin/ extractives contents increased, suggesting that lignin
was not degraded; however, it is likely that the extractives increased and
lignin decreased, as has been shown previously (20).
The next tests performed in the columns were designed to bracket
the moisture range necessary for the fungal degradation to proceed. Since
hemicellulose is generally the most easily degraded fraction of the stems,
hemicellulose degradation was used as the performance measure for these
tests. The effect of 40, 55, and 70% moisture on the degradation of
hemicellulose at various times over 12 wk of treatment is shown in Fig. 4
for both the uninoculated control and for the 10.9 mg/ g columns. Mois-
ture had little effect on straw stem degradation in control columns below
70% (0.7 g of H 20/g dry stems) (see Fig. 4A). In columns inoculated with
P. ostreatus, there was a slight effect of moisture below 70% (Fig. 4B), but
only after 60 d of culture. Although this effect was not significant at 55%
moisture, there was a linear trend of increased hemicellulose removal at
84 d, going from 40 to 70% moisture (Fig. 4B). In all experiments, includ-
ing those using >5.1 mg/ g of inoculum, the first 3 wk of degradation gave
nearly identical results, indicating that even though dominance by
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Fungal Bioprocessing of Wheat Straw 433

!•
30
• • •A
:2
~ ~

20
'ii
u
r-
"e ""0 days

::c 10 ~22days
-t:r56 days
r-
jfe ...... 84 days
i r-
0
30
B
• • •
!• --:
:2 20
'i
u
"e ""0 days
-

::c 10 ~22days
-t:r 56 days
-
~ ...... 84 days -
0
40 50 60 70
% Moisture (9 H20 I 9 dry straw)

Fig. 4. Effect of moisture content on degradation of hemicellulose at various times:


(A) control (0 mg P. ostreatus/g stems); (B) 10.9 mg P. ostreatus/g stems.

P. ostreatus was observed at 84 d using higher amounts of inoculum, the


initial 3 wk of the culture was still dominated by the indigenous organ-
isms. This may be due to nitrogen levels being too high during this period,
or to a significant lag time in P. ostreatus growth during its acclimation to
straw stems as a substrate for growth. In any event, it seems clear that
even higher moisture contents will give increased degradation of hemi-
cellulose in the column tests; hence, we have begun testing gravimetric
moisture contents from 90 to 150%.
The last set of column tests was designed to determine the amount of
inoculum necessary to obtain suitable feedstock for production of straw
thermoplastic composites, while obtaining this composition in 12 wk or
less. It is expected that removal of hemicellulose will allow better resin
penetration and adhesion, since similar results have been obtained using
steam-exploded broom fibers (15). The fiber damage imparted by steam
explosion would not be expected to be present in straw stems treated with
fungi, especially by a fungus that does not degrade significant cellulose
(22). However, for this to occur the inoculated P. ostreatus must overtake the
indigenous microbes. Since P. ostreatus is selective for hemicellulose and
lignin degradation vs cellulose degradation, a clear signal that the inocu-
lated P. ostreatus has overtaken the indigenous organisms would be a sig-
nificant increase in the cellulose/hemicellulose ratio. Therefore, the
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
434 Thompson et al.
Table 2
Cellulose/Hemicellulose Ratios at 0 and 84 d
for the 70%-Moisture Column Treatments·
Organism(s) Removal
Inoculum Cellulose/ assumed of hemicellulose
(mg P. ostreatus/g stems) hemicellulose dominant in 84 d (%)

Day 0 1.33 ± 0.08


0 1.43 ± 0.05 Indigenous 13.7
1.6 1.49 ± 0.05 Indigenous 15.1
5.1 1.53 ± 0.09 Indigenous 21.7
8.2 1.46 ± 0.09 Indigenous 14.1
10.9 1.75 ± 0.11 P.ostreatus 24.8
aA significantly increased cellulose/hemicellulose ratio vs the d 0 ratio indicates selec-
tive degradation of hemicellulose vs cellulose, and thus degradation by P. ostreatus that is
relatively greater than that by indigenous organisms.

laboratory performance measures for these experiments were set as


increased removal of hemicellulose, together with an increased ratio of
cellulose /hemicellulose. The composite material properties will dictate the
final amount of inoculum and treatment time chosen.
The cellulose/hemicellulose ratios and percentage of hemicellulose
removal data at 0 and 84 d for the 70% moisture column treatments are
given in Table 2. In only the 10.9 mg/ g case was the cellulose/hemicellu-
lose ratio increased significantly. This indicates that the amount of
P. ostreatus inoculum should be set at 10.9 mg/ g or greater. Thus, the range
of inoculum amount for the next experiments will be 11-100 mg/g or
higher. Nearly 60% removal of hemicellulose in 12 wk has been shown in
the literature using Pleurotus species, albeit in initially-sterile straw (20).
At 10.9 mg/ g of inoculum, we achieved a nearly 25% reduction in hemi-
cellulose with little loss of cellulose. Thus, increased amounts of inoculum
over 10.9 mg/ g are expected to increase the amount of degradation in the
same time period; experiments utilizing 21, 31, and 41 mg of P. ostreatus/g
of straw stems are under way at the INEEL.

Conclusions
A minimum of 80-100 ppm of nitrogen was carried over to the straw
stems in the mycelial inoculum. Above inoculum levels of 5.1 mg of
P. ostreatus/g dry stems, the inoculated fungus has overtaken the indig-
enous microbes in nonsterile stems by 3 to 4 wk into the treatment, at the
conditions tested. Moisture content was shown to have little effect on
degradation below about 70% gravimetric moisture content. An inocu-
lum amount of 10.9 mg of P. ostreatus / g stems was shown to be sufficient
to effect the desired degradation trends, but not in the desired time frame
of 12 wk or less. Higher amounts of inoculum or inoculum that is
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Fungal Bioprocessing of Wheat Straw 435
pre acclimated to straw may be required for this to occur. Future work
includes testing up to 150% moisture and 100 mg/ g of fungal inoculum.

Acknowledgments
We thank the University of Idaho, Aberdeen Research and Extension
Center for use of its plot-harvesting equipment for the mechanical straw
stem separation; Dr. Stephen Aust and Paul Swaner at Utah State Univer-
sity for maintaining and supplying the fungal cultures for inoculum pro-
duction; and Karen Miller, Neal Yancey, and Jeff Parry at the INEEL. This
project is administered by the Idaho Department of Water Resources En-
ergy Division. This work is supported in part by the US Department of
Energy, Assistant Secretary for Energy Efficiency and Renewable Energy
(EE) under DOE Idaho Operations Office Contract DE-AC07-99ID13727.
Additional support is provided by the Idaho Wheat Commission, Grant 4-
D Farms, and Energy Products of Idaho.

References
1. Wolcott, M. P. and Englund, K. (1999), in Proceedings of the 33rd International Particle-
board/Composite Materials Symposium, Wolcott, M. P., Tichy, R. J., and Bender, D. A.,
eds., Washington State University, Pullman, WA, pp. 103-111.
2. Sanadi, A. R., Caulfield, D. F., and Jacobson, R. E. (1997), in Paper and Composites from
Agro-Based Resources, Rowell, R. M., Young, R. A., and Rowell, J. K., eds., CRC, Boca
Raton, FL, pp. 377-402.
3. Marsden, W. L. and Gray, P. P. (1986), Crit. Rev. Biotechnol. 3(3),235-276.
4. Converse, A. 0., Kwarteng, I. K., Grethlein, H. E., and Ooshima, H. (1989), Appl.
Biochem. Biotechnol. 20/21, 63-78.
5. Cowling, E. B. and Kirk, T. K. (1976), Biotechnol. Bioeng. Symp. 6,95-123.
6. Thompson, D. N., Chen, H.-C., and Grethlein, H.E. (1992), Bioresour. Technol. 39,
155-163.
7. Fan, L. T., Lee, Y.-H., and Gharpuray, M. M. (1982), Adv. Biochem. Eng. 23, 157-187.
B. Knappert, D., Grethlein, H., and Converse, A. (1980), Biotechnol. Bioeng. 22, 1449-1463.
9. Goldstein, I. S., Pereira, H., Pittman, J. L., Strause, B. A., and Scaringelli, F. P. (1983),
Biotechnol. Bioeng. Symp. 13, 17-25.
10. Playne, M. J. (1984), Biotechnol. Bioeng. 26,426-433.
11. Weimer, P. J., Chou, Y.-C.T., Weston, W. M., and Chase, D. B. (1986), Biotechnol.
Bioeng. Symp. 17, 5-18.
12. Avgerinos, G. C. and Wang, D.1. C. (1983), Biotechnol. Bioeng. 25,67-83.
13. Taylor, J. D. (1981), in Energy from Biomass, 1st E. C. Conference, Palz, W., Chartier,
P., and Hall, D.O., eds., Applied Science, London, UK, pp. 330-336.
14. Hatakka, A. I. (1983), Eur. J. Microbiol. Biotechnol. 18,350-357.
15. Avella, M., Casale, L., Dell'erba, R., Focher, B., Martuscelli, E., and Marzetti, A. (1998),
J. Appl. Polym. Sci. 68, 1077-1089.
16. Hess, J. R., Thompson, D. N., Hoskinson, R. L., Shaw, P. G., and Grant, D. R., (2003),
Physical Separation of Straw Stem Components to Reduce Silica, in Applied Biochem-
istry and Biotechnology, vol. 105-108, Humana Press, Totowa, NJ, pp. 43-52.
17. Boominathan, K. and Reddy, C. A. (1992), in Handbook of Applied Mycology, vol. 4,
Arora, D. K., Elander, R. P., and Mukerji, K. G., eds., Marcel-Dekker, New York, NY,
pp. 763-822.
lB. Blanchette, R. A., Abad, A. R., Farrell, R. L., and Leathers, T. D. (1989), Appl. Environ.
Microbiol. 55,1457-1465.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


436 Thompson et al.
19. Valmaseda, M., Almendros, G., and Martinez, A T. (1990), Appl. Microbiol. Biotechnol.
33,481--484.
20. Moyson, E. and Verachtert, H. (1991), Appl. Microbiol. Biotechnol. 36,421--424.
21. Gamble, G. R, Akin, D. E., Makkar, H. P. S., and Becker, K. (1996), Appl. Environ.
Microbiol. 62, 3600-3604.
22. Hadar, Y., Kerem, Z., and Gorodecki, B. (1993), J. Biotechnol. 30,133-139.
23. Lindfelser, L. A, Detroy, R W., Ramstack, J. M., and Worden, K. A. (1979), Dev. Ind.
Microbiol. 20, 541-55l.
24. Kirk, T. K. and Farrell, R L. (1987), Annu. Rev. Microbiol. 41,465-505.
25. American Public Health Association. (1989), in Standard Methods for the Examination
of Water and Wastewater, 17th ed., Clesceri, L. S., Greenberg, A E., Trussell, R R, and
Franson, M. A H., eds., American Public Health Association, Washington, DC,
pp.4-144--4-147.
26. Cherry, R S. and Thompson, D. N. (1997), Biotechnol. Bioeng. 56(3),330-339.
27. Kastner, J. R, Thompson, D. N., and Cherry, R S. (1999), Enzyme. Microb. Technol.
24(1/2),104-110.
28. Stahl, J. D. and Aust, S. D. (1998), Rev. Toxicol. 2, 189-194.
29. Saeman, J. F., Bubl, J. L., and Harris, E. E. (1945), Ind. Eng. Chern. 17(1),35-37.
30. Martinez, AT., Camarero, S., Guillen, F., Gutierrez, A, Munoz, c., Varela, E.,
Martinez, M. J., Barrasa, J. M., Ruel, K., and Pelayo, J. M. (1994), FEMS Microbiol. Rev.
13,265-274.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0437/$20.00

Simulation of Aerated Lagoon


Using Artificial Neural Networks
and Multivariate Regression Techniques

KARLA PATRICIA OLlVEIRA-EsQUERRE/,l ALINE C. DA COSTA, 1


Roy EDWARD BRUNS/ AND MILTON MORl 1
lOPQ/FEQ/UNICAMP, PO Box 6066, 13081-970 Campinas,
SP, Brazil, E-mail: karla@feq.unicamp.br; and
2IQ/UNICAMP, PO Box 6154, 13083-970 Campinas, SP, Brazil

Abstract
The aim of this study was to develop an empirical model that provides
accurate predictions of the biochemical oxygen demand of the output
stream from the aerated lagoon at International Paper of Brazil, one of the
major pulp and paper plants in Brazil. Predictive models were calculated
from functional link neural networks (FLNNs), multiple linear regression,
principal components regression, and partial least-squares regression
(PLSR). Improvement in FLNN modeling capability was observed when
the data were preprocessed using the PLSR technique. PLSR also proved to
be a powerful linear regression technique for this problem, which presents
operational data limitations.
Index Entries: Biochemical oxygen demand; functional link neural net-
works; partial least squares; principal components regression; multiple lin-
ear regression.

Introduction
Industrial and municipal wastewater are major sources of contamina-
tion of aquatic biota, accounting for several thousand types of chemicals
released into the environment. Thus, the importance of implementing
efficient monitoring and control techniques for wastewater treatment sys-
tems is well known. Operational control of biologic wastewater treatment
plants is often complicated because of variations in raw wastewater com-
positions, strengths, and flow rates, owing to the changing and complex
nature of the treatment process (1). Moreover, the lack of suitable process
variable measurements limits the effective control of effluent quality (2,3).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 437 Vol. 105-108,2003


438 Oliveira-Esquerre et al.
The modeling traditionally used in bioprocesses is based on balance
equations combined with rate equations for microbial growth, substrate
consumption, and products formation. These methods have proven to
be inefficient to represent the nonlinear, time-variant, and complex inter-
nal workings of the wastewater treatment process (1,3-5).
Multivariate linear regression techniques may be used to approximate
complex relationships over small regions of the predictor variables (6). This
approach is based on the assumption that the underlying nonlinear rela-
tionship can be locally approximated by a linear model. Partial least squares
regression (PLSR), or projection to latent structures regression, has been
shown to be a powerful linear regression technique for problems in which
the data are noisy and highly correlated and in which there are only a
limited number of observations (6). PLSR provides the potential to summa-
rize the underlying structures of the process as linear combinations of the
original variables (latent variables [LVs]) and to relate the predictor and
response variables by means of linear regression models among the LVs.
Recently, some studies using artificial neural networks (ANNs) in
modeling biologic wastewater treatment processes have been published,
providing an alternative approach (1,3,4,7-12). In contrast to PLSR model-
ing, the main well-known drawbacks of neural network training are that it
requires a large quantity of experimental data and its nonlinear approxima-
tion function can cause local minima problems. Moreover, the interpreta-
tion of such models is often difficult (13).
There are no methods proposed to define the best structure to be used
for a given case. A structure that has been little explored in bioprocesses is
the functional link neural network (FLNN) (13). In this network, a nonlin-
ear functional transformation or expansion of the network inputs is ini-
tially performed and the resulting terms are combined linearly; that is, the
estimation of network weights is linear. The structure obtained yields rapid
convergence and good nonlinear approximation capability for a variety of
functional approximation and inverse system modeling applications
(13,14).
Recent studies indicate that consideration of multivariate statistical
principles in the neural network model building process may improve
modeling performance (15). Baffi et a1. (6), Kanjilal (16), and Kompany-
Zared (17) suggest the use of principal component (PC) variables, obtained
by an orthogonal variable transformation, for pruning ANNs and improv-
ing their nonlinear mapping capabilities. PLSR LV can be used for the same
purposes. Both techniques have been shown to be appropriate tools for
extracting relevant information from correlated data. The use of principal
component analysis (PCA) has been shown to be an efficient preprocessing
technique for input data of ANNs (9,17-19). Baffi et a1. (6) and Holcomb and
Morari (19) explored the combination of PLSR and ANN applied to nonlin-
ear chemical processes.
The main aim of the present study was to develop an estimation model
that provides accurate predictions of the biochemical oxygen demand
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Simulation of Aerated Lagoon Using ANNs 439
(BOD) of the output stream of an aerated lagoon at International Paper of
Brazil, one of the major pulp and paper plants in Brazil. There is a 5-d delay
in the determination of BOD, and when this is added to the 3-d hydraulic
residence time, it is often too late to make proper adjustments in the waste-
water treatment process.
In this work, the predictive models are calculated from FLNN and
multivariate regression techniques (multiple linear regression [MLR], prin-
cipal components regression [PCR] and PLSR). The potential improvement
in FLNN modeling capability is also evaluated when PCs and LVs are used
as the model's inputs.
The results show that FLNNs, MLR, nor PCR techniques are satisfactory
when used individually for the aerated lagoon modeling and simulation.
The best prediction performance is achieved when the data are preprocessed
using PLSR before they are fed to an FLNN obtained by a first-order polyno-
mial expansion and the logistic sigmoid activation as a transfer function.
Similar results are obtained using a linear PLSR technique, showing that it
can be appropriately used for modeling nonlinear systems.

Materials and Methods


The process wastewater was routed for primary treatment followed
by biologic treatment in an aerated lagoon. To construct the predictive
model, 10 variables of the aerated lagoon and 2 of the milling process were
chosen using engineering judgment regarding which ones had an impor-
tant effect on BOD prediction. Variations within sampling, including flow
rate (FLOW), chemical oxygen demand (COD), suspended solids (S.5.),
nitrate (N.N.) and ammonia (N.AM.) concentrations, conductivity
(COND.), color (COLOR), pH, temperature (T.), rainfall (RAIN), and pulp
(PULP) and paper (PAPER) production were evaluated. Their basic statis-
tics are provided in Table l.
The average hydraulic residence time in the aerated lagoon was used
to establish a data input/output relationship. The original database,
obtained from the plant control system and from the laboratory, has much
unusable information that must be eliminated, so the 4-yr daily records
available for investigation were reduced to 74 data points, each having 12
input variables and 1 output variable (BOD).
MLR, peR, and PLSR
MLR models very often involve large numbers of predictor variables
and their consequent collinearity or correlation provokes problems on re-
gression. PCR and PLSR models perform an orthogonal transformation of
data to form a set with reduced dimensionality; thus, there is no correlation
between the transformed variables (PCs) used in the predictive regression,
and the collinearity problem is solved. However, PCR is a two-step method
and therefore there are the risks that useful predictive information ends up
in discarded PCs and that some noise remains in the pes used for regres-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


440 O/iveira-Esquerre et a/.

Table 1
Basic statistical Descriptors for Selected Variables
Parameter Average Minimum Maximum SO Skewness Kurtosis

COD 624.19 297 881 99.84 -0.46 1.67


FLOW 61,103.81 35,156 73,693 5506.87 -1.74 6.15
S.5. 143.80 16 464 77.13 1.23 3.06
N.N. 1.55 0.568 5.99 0.93 2.47 8.52
N.AM. 2.70 0.03 9.2 1.52 1.43 3.88
CONDo 1672.66 1036 3360 403.56 1.81 4.52
COLOR 442.69 111 911 159.65 0.44 1.06
pH 7.28 6.28 10.4 0.79 2.36 5.84
T. 45.85 29 50 3.20 -2.39 9.87
RAIN 4.17 0 40.7 9.47 2.49 5.35
PULP 894.24 0 1079.67 182.06 -3.26 12.15
PAPER 1060.31 685.3 1231.9 98.00 -1.66 4.19
BOD 83.23 20 155 26.44 0.35 0.30

sion (20,21). Even though PLSR is closely related to the PCR method, PLSR
is designed to maximize the prediction rather than fit the input data; that
is, this method optimizes LVs in order to maximize the proportion of vari-
ance of their description of the prediction variables. Hense, the risk of los-
ing predictive information is discarded or minimized. Mardia et a1. (22) and
Draper and Smith (23) provide detailed information on MLR and PCR.
Geladi and Kowalski (21), Wold et a1. (24), and Wold and Kowalski (25)
provide details of the PLSR methodology.
FLNN
A conventional multilayer neural network contains one or more stages
of neural processing between inputs and outputs. These stages, known as
hidden layers, add to the complexity and cost of neural training (13,26).
When the number of inputs to the model and the number of records avail-
able for training becomes extremely large, the training procedure for this
neural network architecture become increasingly more time-consuming
(27), whereas when the records available are too small, there is a problem
of overfitting (9).
The FLNN is a neural network with no hidden layers, trained using
supervised learning. The input is "enhanced" by generating additional
terms via some transformation rule, such as a polynomial expansion. The
idea is to increase the dimensionality of the feature-space without requir-
ing any additional information. These enhanced values are passed through
to a summation node (the output), which transforms these weighted values
via some nonlinear activation function. The use of the activation function
is a modification of the structure of the FLNNs proposed by Henrique (28).
It increases the non-linear approximation ability of the network, while
estimation of the parameters remains a linear problem (26). In addition,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Simulation of Aerated Lagoon Using ANNs 441
Henrique (28) proposed the use of an orthogonal estimator (29) to calculate
the network weights and to eliminate nonlinear significant nodes during
the training of the network (30).
As can be seen, FLNN offers a remedy for time-consuming training
and overfitting problems. It trains faster than multilayer networks because
each interaction of the training procedure is linear and consequently takes
less time. Its linear mapping requires a smaller data set for training and also
avoids local minima problems.
Theoretical foundations of FLNN can be found in Costa et al. (3D),
Hornik et al. (31), and Pao (32). The merits of the network output transfor-
mation proposed by Henrique (28) can be found in Costa et al. (30,33),
Harada et al. (34), and Henriques et al. (26).

Data Analysis by MLR, peR, PLSR, and FLNN Modeling


Using a random selection method, 80% of all data records was assigned
to multivariate model construction and FLNN training, while the remaining
20% was relegated to the validation set.
Stepwise regression was used to remove and add variables from the
MLR model for the purpose of identifying a useful subset of predictors,
reducing its dimensionality and collinearity (23,35,36). a. = 0.10 (90% con-
fidence level) and a. = 0.15 (85% confidence level) were considered for
adding and removing variables, respectively.
Unfortunately, there is no universal, automatic criterion for the esti-
mation of the number of PCs or LVs, and discrepancies among different
criteria are common (37). In the present work, the Todeschine (38) criterion
was used. Todeschine (38) proposed a new correlation index (K) as a mea-
sure of quantity of correlation in the data set, as well as linear (KL) and
nonlinear (KP) functions for estimating the maximum and minimum num-
ber of significant PCs to be used in the PCR model, respectively.
The optimal number of LVs is identified considering the increase in
predicted accumulated variance (AV) that must be at least 2% for adding
an LV (39).
A polynomial expansion up to the third degree was performed on the
FLNN inputs to generate nonlinear monomials and the orthogonal least-
squares estimator proposed by Billings et al. (29) was used to calculate the
network weights and to eliminate the monomials that are not significant in
explaining the output variance. This reduces the size and complexity of the
neural network and avoids overfitting the data (26,30).
According to the modification of the structure of the FLNNs proposed
by Henrique (28), the network output is transformed by an invertible con-
tinuously differentiable nonlinear function. This activation function is cho-
sen by trial and error to maximize the training performance of the FLNN.
The three functions tested were

y = f(T} =12 log (11-T


+ T) (1)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


442 Oliveira-Esquerre et al.

y = f(T) = t log ( 1 ~ T ) (2)

y = f(T) =1 log
2
(0.5I-T+ T) (3)

in which, Eq. 1 is a hyperbolic tangent sigmoid activation transfer function


(Tansig), Eq. 2 is a logistic sigmoid activation transfer function (Logsig),
and Eq. 3 is a centered logistic sigmoid activation transfer function (Logsic).
Finally, PCs and LVs were used as FLNN inputs and the prediction
performance of the obtained models were compared.
p Values for the adjusted coefficient of multiple determination (R2
adj.) and the normality test of the residual were calculated in order to
verify the significance of the multivariate regression and neural network
models, as well as to determine whether additional terms in the regres-
sion model would be useful. Both tests assumed a 95% confidence level.
The F test (40) was used as an indicator of adequacy of model with differ-
ent numbers of parameters. Root mean square error (RMSE) was calcu-
lated for modeling and validation results. Linear correlation index (R2)
values were calculated for the validation set.
In addition to regression graphs, three additional measures were used
for the purpose of identifying outliers, or unusual observations that can
have a significant influence on the regression: leverages, Cook's distance,
and DFITS index (35).
The same data sets were used for the training and validation of both
neural network and multivariate regression models.
The Minitab, Statistica and Unscrambler computer programs were
used for statistical analysis, data pretreatment; and MLR, PCR, and PLSR
modeling. A Matlab computer program, developed by Henrique (28), was
used for the training and validation of the FLNN.

Results
The data were auto scaled to mean zero and unit variance prior to
analysis. Table 2 provides the eigenvalues and the variance percentages
(accounted for and cumulative) corresponding to the PCs for the predic-
tor variable. PC modeling does not involve the predicted variable space.
Table 2 also gives the variance percentages (accounted for and cumula-
tive) corresponding to the LVs in both predictor and predicted variable
space. Loadings were normalized to 1 and scores to their corresponding
eigenvalues.
By taking a close look at the variance of the LVs and PCs, it can be
verified that for any particular number of LVs or PCs, the LVs always
describe more predicted variable variance and less predictor variable vari-
ance than the PCs.
Applying the Todeschine (38) criterion to the data matrix, we obtained
K =0.3747, which suggests a moderate degree of correlation, KL =8, corre-
sponding to 92.6% of AV and KP = 5 (75.7% AV) for PCA. This measure
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Simulation of Aerated Lagoon Using ANNs 443
Table 2
Statistical descriptors for PCs and LVs·

EV (1) AV (1) EV (1) AV (1) EV (2) AV (2)


PC Eigenvalue (%) (%) LVs (%) (%) (%) (%)
1 3.6410 30.3 30.3 1 21.50 21.50 42.36 42.36
2 1.7805 14.8 45.1 2 20.01 41.51 12.99 55.35
3 1.3065 10.9 56.0 3 9.15 50.66 7.79 63.14
4 1.2502 10.4 66.4 4 7.51 58.17 4.43 67.57
5 1.1102 9.3 75.7 5 5.04 63.21 2.64 70.21
6 0.9454 7.9 83.6 6 7.08 70.29 0.49 70.70
7 0.5821 4.9 88.5 7 4.44 74.73 0.29 70.99
8 0.4977 4.1 92.6 8 6.22 80.95 0.13 71.12
9 0.3779 3.2 95.8 9 4.08 85.03 0.06 71.18
10 0.2579 2.1 97.9 10 8.16 93.19 0.01 71.19
11 0.1688 1.4 99.3 11 3.23 96.42 0 71.19
12 0.0819 0.7 100.0 12 3.58 100.00 0 71.19
"Explained variance (EV) and A V are given as predictor variable space (1) and predicted
variable space (2).

reduced the number of PCs to five, which exhibited eigenvalues> 1. This


result agrees with the criterion suggested by Mardia et al. (22) and Morales
et al. (41), for which PCs whose eigenvalues are <1 ought to be excluded.
Considering the increase in the predicted AV, the optimal number of
LVs was identified as five (which corresponds to 63.21 and 70.21% of the
predictor and predicted AV, respectively). Only a 0.5% of increase in AV
was reached with the inclusion of the sixth LV.
Figure 1 depicts the loading and score plots corresponding to the
first two PCs (left) and PCs 3 and 4 (right). From the loading plots it can
be deduced that the first PC (PC1) reflects an inverse correlation between
CONDo and FLOW, a~d T. and PULP (loadings -0.42, 0.41, 0.43, and 0.47,
respectively). The second PC (PC2) is associated with COD and suspended
solids 5.5. (loadings -0.51 and -0.58, respectively). The third PC (PC3)
reflects a negative correlation between N.N. and COLOR (loadings -0.56
and 0.49, respectively). The fourth PC (PC4) is related to COLOR, RAIN),
pH, N.N., and N.AM. data (loadings 0.48, 0.38, 0.38, -0.43, and -0.40,
respectively). As can be seen, PC1, PC3, and PC4 are difficult to interpret.
PC2 may be accounted for by a microbiologic factor, and the fifth PC
(PC5) is exclusively related to PULP (loading 0.78) (not shown). There is
a relatively spread-out distribution among the PCs, which is quite com-
mon when working with environmental data (41).
As expected, there is no indication that the PCs are correlated in the
score plots. However, some samples exhibit more anomalous scores than
the rest, so they can be indicated as possible outliers.
Consistent with the stepwise results, the COD, COND., T. and PULP,
and PAPER parameters appear to be the most significant in the first five
Applied Biochemistry and Biotechnology Vol. 105-108,2003
444 Oliveira-Esquerre et al.
0.4,..-------:------...,

i·. ·
0.6
N.~. ~.AM'RAIN FLOW COLOR
0.2
: PAPER
• 0.4
N.N.
• ~N i pH •
0.0 -····-···-····-··-·-··--·--····-···+··-··-·Clioii--··-·--.. ---.--.--.. i •
PAPER .CO~. ~OW
..
0.2
N C~ND. 1 '1lLP T.. i
~ -0.2 rfH i • ~ 0.0 .-...- ..--..- ----..- ..-.~.-..- ..-.. --.-.-------
T. COD • i

~l
-0.4 -0.2
COD

-0.6
s.s.
• •
-0.4

-0.4 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
PC1 PC3
4,..--------~~-~
4
3 •

.... . ..
3
2 • • •
• i !C •
~ 0
. . . _. . __.._._. _._:. . _. . _. __._. __...__..._. - . -.. -~~--!. -
Q. -1
~
.i ..
-2 ~
• i
-3
~+-~~~~~--,..-~~~~
.. • I
-10 -8 ~ -2 2 o -3 -2 -1 0 2 3
PC1 PC3

Fig. 1. Loading (top) and score plots (bottom), corresponding to first four PCs.

LVs. As expected, since the COD is the parameter most correlated with
BOD, the first LV is exclusively related to the COD (loading 0.70) and COD
is the first parameter included in the MLR by the stepwise method. The
multivariate regression models were then constructed and their prediction
performance results are given in Table 3. As expected, MLR, PCR, and
PLSR models present the same performance results when there is no pa-
rameter exclusion. The p values for the adjusted coefficient of multiple
determination (R2 adj.) are zero for all models, indicating their high statis-
tical significance. In spite of this, the p values for the normality test of the
residuals and the F test for model adequacy analysis indicate that predic-
tive information has been lost by exclusion of the last seven PCs. It had not
occurred when the stepwise method was used for identifying the most
significant PCs for the predictive model (eighth and ninth have been ex-
cluded). Only the stepwise-MLR model residuals did not pass in the nor-
mality test. From the modeling and validation statistical results, it can be
seen that best prediction performance was achieved using the shortest
PLSRR model.
The FLNN models were then generated. The number of monomials
generated is a function of the number of inputs and the degree of the poly-
nomial expansion, in this case 12 and 3, respectively. Then, the number of
monomials generated is 455. Using the orthogonal least-squares estimator,

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Simulation of Aerated Lagoon Using ANNs 445
Table 3
Statistical Results of Multivariate Regression Models

Modeling Validation

Model R2 adj. RMSE p Value a F testb R2 RMSE

MLR 0.637 0.063 0.108} 0.619 0.070


0.98
Stepwise-MLR 0.636 0.068 0.003 0.656 0.069
PCR 0.637 0.063 0.108} 26.61c 0.619 0.070
PCR (1-5 PCs) 0.320 0.092 0.786 0.379 0.095
PCR (1-7,10-12 PCs) 0.638 0.064 0.181 7.60d 0.615 0.070
PLSR 0.637 0.063 0.108} 0.619 0.070
0.22
PLSR (1-5 LV) 0.674 0.064 0.250 0.666 0.067
'p Value for the normality test of the residuals.
bTest indicated by Barros et al. (40). F O.o5, 7,46 = 2.22
'Between PCR and PCR (1-5PC) models and d between PCR and PCR (1-7, 10-12 PC)
models.
Table 4
Statistical Results of FLNN, PCA-FLNN, and PLSR-FLNN Models

Modeling Validation

Model R2 adj. RMSE p Value a F test b R2 RMSE

FLNN 0.647 0.062 0.181 } 0.555 0.096


0.76
Stepwise-FLNN 0.659 0.066 0.003 0.585 0.098
PCA-FLNN 0.647 0.062 0.181 } 0.602 0.083
29.34 c
PCA-FLNN (1-5 PCs) 0.198 0.094 0.575 0.350 0.096
PCA-FLNN
(1-7,10-12 PCs) 0.629 0.064 0.679 1.19d 0.604 0.086
PLSR-FLNN 0.647 0.062 0.181 } 0.648 0.076
0.31
PLSR-FLNN (1-5 LVs) 0.679 0.063 0.297 0.717 0.069
'p Value for the normality test of the residuals.
bTest indicated by Barros et al. (40). F O.o5, 7, 46 = 2.22.
'Between PCA-FLNN and PCA-FLNN (1-5 PC) models and Fo 05, 2, 46 = 3.26.
dBetween PCA-FLNN and PCA-FLNN (1-7, 10-12 PC) models.

nonsignificant monomials were eliminated and the weights are estimated.


The best neural network prediction results are given in Table 4.
By comparing the FLNN approach with the approaches proposed
herein, PC A-FLNN and PLSR-FLNN, it can be seen thatthe best prediction
performance was achieved when the data were preprocessed using PLSR
before they were fed to an FLNN using a first-order polynomial expansion
and the logsig activation function. Considering the validation set results,
PLSR-FLNN modeling performance was significantly improved when
only the most significant LVs were used as input. However, this pruning
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
446 Oliveira-Esquerre et al.

.
0.8 r;:::::=========;:------/,,~ 0.8 Tr======;:-----"~,,,7'1
Regression Regression /
0.7 ------ 95% PI

. . .
."",,"". 0.7 ------ 95% PI

.......... . .
.. .,,/.
: •

.."...
,,;.,. . .
# .......... . ........
0.6
.........." 0.6
"0 "0

.......... .. ........
:c~ 0.5 .. ,"
/;"
....
.' --: ....... . ,,/
~ 0.5 " "...... /"
,,,,,

.
~ ,............ ....... . CD
.....
...
c. 0.4
o
I
:. /~••........ ,"
Q. 0.4
o
/'
=
....... . """
,.. . .
oco oco ......~ ......
0.3

0.2 JA-",i "


, ........ 0.3

0.2
.........
~..........
,,'

O~ O~ OA O~ OB 0.7 0.8 0.2 0.3 0.4 0.5 0.6 0.7 0.8


BOD measured BOD measured

Fig. 2. Relation between predicted vs measured BOD (solid lines). Upper and lower
dashed lines indicate the 95% prediction estimation interval (according to the PLSR
and PLSR-FLNN models, without the last seven pes).

does not help FLNN and PCA-FLNN prediction results for the validation
data set.
Comparison of multivariate regression and neural network results,
reveals that PLSR and PLSR-FLNN models gave similar prediction re~ults
when the last seven LVs were excluded. It really shows that the linear PLSR
technique can be used to approximate complex relationships over the
regions of the predictor variables available for the considered process. How-
ever, for the validation set, PLSR-FLNN gave slightly better results.
Statistical results also show that the best modeling structures-i.e.,
PLSR and PLSR-FLNN without the last seven Pes-although still far from
the perfect, do capture prediction information and are able of estimating
outputs within a 95% prediction band (see Fig. 2). This is particularly impor-
tant when one takes into account the complexity of the wastewater treat-
ment system and the large quantity of missing data in the modeling set.
Figure 3 presents a graphic representation of the measured and pre-
dicted BOD data for the modeling and validation data sets using PLSR and
PLSR-FLNN models without the last seven PCs. The modeling features
were well reproduced by both modeling structures, whereas the validation
features were slightly better reproduced by the PLSR-FLNN model.
Multivariate regression and the neural network have also provided
robust models. Their model parameters do not change very much when
the samples identified as possible outliers were taken from the training
data set.

Discussion
In recent years, ANNs have become extremely popular for predic-
tion and forecasting in a number of areas, including water resources,
bioprocesses, and environmental science. The FLNN appeared as a good

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Simulation of Aerated Lagoon Using ANNs 447

modeling validation
- - Measured values
0,8 set set
., '0" Predicted values (PLS)
- ..... - Predicted values (PLS-FLNN)

__ 0,6
~
c
0::1
O~
COl!
:e 0,4
.!!

0,2

o 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Sequential number of data

Fig. 3. Measured and predicted BOD, using the PLSR and PLSR-FLNN models
(without last seven PCs).

option to remedy the time-consuming training and overfitting problems


of the classic feedforward back-propagation networks.
In the present work, some modifications were used in the functional
link network structures, such as the use of an invertible activation function,
which has been proved to increase the nonlinear approximation capability.
As in PCR and MLR models, the simple structure of FLNNs led to an
unsatisfactory performance for the simulation and prediction of BOD for
the data set treated here. Then, in addition to the method used to elimi-
nate nonsignificant monomials during the network training, PCA and
PLSR were used as input pretreatment techniques for pruning the FLNN
structure. The combined use of these techniques and FLNNs provided
prediction results that have statistical parameters significantly superior
to those obtained using the simple FLNN structure.
Best prediction results were given using the PLSR and FLNN struc-
ture obtained by using a first-order polynomial expansion and the logistic
sigmoid activation as a transfer function, when the last seven latent vari-
ables were excluded. The PLSR technique helped FLNN mapping by
its orthogonal transformation of variables and reduction of the system's
dimensionality.
Our work also shows the advantage of linear PLSR in its ability to
represent highly nonlinear relationships, even for a system that presents
operational data limitations (e.g., imprecision associated with measured
variables, a limited range of variables, a large number of missing values).
In spite of the nonlinear nature of the process, linear PLSR gave similar
prediction results for the modeling data set and slightly inferior ones for
thevalidation set when compared with the PLSR-FLNN approach.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
448 Oliveira-Esquerre et al.

Acknowledgments
We are grateful to professor Dr. Fernando Von Zuben, Vitaly F.
Rodriguez Esquerre, Roberto C. Colacioppo, and Celeste M. Diaz C6nsul for
the many stimulating discussions held during the development of this work;
and to International Paper of Brazil for providing the data set. We also thank
Funda~ao de Amparo aPesquisa do Estado de Sao Paulo (Proc. N. 99 110257-
0) for their financial support.

Nomenclature
BOD = outlet wastewater BOD (mg/L)
COD = inlet wastewater COD (mg/L)
COLOR = color (ppm or mg/L)
CONDo = conductivity (!-lS/cm at 20°C)
f = transfer function proposed by Henrique (28)
FLOW = inlet flow rate (m3 I d)
h = polynomial expansion
N = number of monomials
N.AM. = inlet ammonia concentration (mg/L)
N.N. = inlet nitrate concentration (mg/L)
PAPER = paper production (tl d)
PULP = pulp production (tl d)
RAIN = rainfall (mLI d)
5.5. = inlet suspended solids (mg/L)
T. = temperature (OC)
x = input matrix
y = predicted output
w ij = neural network weights of i input and j monomial
References
1. Hamoda, M. F., AI-Ghusain, I. A., and Hassan, A. H. (1999), Water Sci. Technol. 40,
55-65.
2. Harremoes, P., Capodaglio, A. G., Hellstrom, B. G., Henze, M., Jensen, K. N.,
Lynggaaard-Jensen, A., Otterpohl, R., and Soeborg, H. (1993), Water Sci. Technol. 27,
71-115.
3. Lee, D. S. and Park, J. M. (1999), J. Biotechnol. 75,229-239.
4. Cote, M., Grandijean, B. P. A., Lessard, P., and Yhibault, J. (1995), Water Res. 29,
995-1004.
5. Steyer, J. P., Rolland, D., Bouvier, J. c., and Moletta, R. (1997), Water Sci. Technol. 36,
209-217.
6. Baffi, G., Martin, E. B., and Morris, A. J. (1999), Comp. Chern. Eng. 23, 1293-1307.
7. Gontarski, C. A., Rodrigues, P. R., Mori, M., and Prenem, L. F. (2000), Comp. Chern.
Eng. 24, 1719-1723.
8. Hack, M. and Kohne, M. (1996), Water Sci. Technol. 33,101-115.
9. Oliveira-Esquerre, K. P., Mori, M., and Bruns, R. E. (2002), Braz. J. Chern. Eng. 19,
365-370.
10. Pu, H., Hung, Y. (1995), Environ. Manage. Health 6, 16-27.
11. Wilcox, S. J., Hawkes, D. L., Hawkes, F. R., and Guwy, A. J. (1995), Water Res. 29,
1465-1470.

Applied Biochemistry and Biotechnology Vol. 105-10B,2003


Simulation of Aerated Lagoon Using ANNs 449
12. Zhao, H., Hao, O. I., Fellow, A. S. C E., McAvoy, T. J., and Chang, C H. (1997),
J. Environ. Eng. 123,311-319.
13. Chen, S. and Billings, S. A. (1992), Int. J. Control 56, 319-346.
14. Alkulaibi, A. and Soraghan, J. J. (1997), Signal Processing 62,101-109.
15. Maier, H. R. and Dandy, G. C (2000), Environ. Modelling Software 15,101-124.
16. Kanjilal, P. P. (1995), IEEE Trans. Neural Networks 6,1061-1070.
17. Kompany-Zared, M. (1999), Talanta 48, 283-292.
18. Cancilla, D. A. and Fang, X. (1996), J. Great Lagoons Res. 22,241-253.
19. Holcomb, T. R. and Morari, M. (1992), Compo Chem. Eng. 16,393-411.
20. Despagne, F. and Massart, D. L. (1998), Analyst 123, 157-178.
21. Geladi, P. and Kowalski, B. R. (1986), Anal. Chim. Acta 185, 1-17.
22. Mardia, K. V., Kent, J. T., and Bibby, J. M. (1979), Multivariate Analysis, Academic,
London, UK.
23. Draper, N. R. and Smith, H. (1981), Applied Regression Analysis,2nd ed., Wiley, New
York, NY. .
24. Wold, S., Martens, H., and Russwurm, H. (1983), Food Research and Data Analysis,
Applied Science Publishers, London, UK.
25. Wold, S. and Kowalski, B. (1984), Chemometrics: Mathematics and Statistics in Chemis-
try, Reidel, Dordrecht, The Netherlands.
26. Henriques, A. W. S., Costa, A. C, Alves, T. L. M, and Lima, E. L. (1999), Braz. J. Chem.
Eng. 16, 171-177.
27. Cass, R. and Radl, B. (1996), Control Eng. Pract. 4, 1579-1584.
28. Henrique, H. M. (1999), PhD thesis, PEQ/COPPE/UFRJ, Rio de Janeiro, RJ, Brazil.
29. Billings, S. A., Chen, S., and Korenberg, M. J. (1989), Int. J. Control 49, 2157-2189.
30. Costa, A. C, Henriques, A. S. W., Alves, T. L. M., Maciel Filho, R., Lima, E. L. (1999),
Braz. J. Chem. Eng. 16, 53-63.
31. Hornik, K., Stinchcombe, M., and White, H. (1989), Neural Networks 2, 359-366.
32. Pao, Y. H. (1989), Adaptative Pattern Recognition and Neural Networks , Addison-Wesley,
Reading, MA.
33. Costa, A. C, Alves, T. L. M., Henriques, A. W. S., Maciel Filho, R., and Lima, E. L.
(1998), Compo Chem. Eng. 22(Suppl.), S859-S862.
34. Harada, L. H., da Costa, A. C, and Maciel Filho, R. (2002), Appl. Biochem. Biotechnol.,
98-100,1009-1023.
35. Montgomery, D. C,Peck,E.A. (1992), Introduction to Linear Regression Analysis, Wiley,
New York, NY.
36. Sjostrom, M. and Wold, S. (1983), Anal. Chim. Acta 150, 61-70.
37. Ortiz-Estarelles, 0., Martin-Biosca, Y., Medina-Hernandez, M. J., 5agrado, 5., and
Bonet-Domingo, E. (2001), Chem. Intel. Lab. Syst. 56,93-103.
38. Todeschine, R. (1997), Anal. Chim. Acta 348, 419-430.
39. Eigenvector Research (1998) MATLAB PLS Toolbox help. Eigenvector Research Inc.,
Manson, WA
40. Barros, B. N., Scarminio, I. S., and Bruns, R. E. (1995), Planejamento e Otimiza(ao de
Experimentos, UNICAMP Press, Campinas, SP, Brazil.
41. Morales, M. M., Marti, P., Llopis, A., Campos, L., and Sagrado, S. (1999), Anal. Chim.
Acta 394,109-117.

.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0451/$20.00

Simplistic Modeling Approach


to Heterogeneous Dilute-Acid Hydrolysis
of Cellulose Microcrystallites
PAR o. PETTERSSON,*,1 ROBERT W. TORGET/ ROBERT EKLUND/
QIAN XIANG,4 Y. Y. LEE,4 AND GUIDO ZACCHI 5

Mid Sweden University, 891 18 6rnskoldsvik, Sweden,


1
E-mail: par.pettersson@mh.se;
2National Renewable Energy Laboratory, Golden, CO;
3Umea University, 90187 Umea, Sweden;4Auburn University, Auburn, AL;
and 5 Lund University, 22100 Lund, Sweden

Abstract
The classic kinetic model for cellulose hydrolysis is often referred to as
pseudo-homogeneous, a term revealing the insight that the process is actually
heterogeneous. During the past 10-15 yr, the shortcomings of this model
have been demonstrated in various studies and the interest in the heteroge-
neous aspects has increased. The present work presents a simplistic model in
which the intrinsic, heterogeneous hydrolysis and transport rates are coupled
by the assumption of a constant glucosidic surface concentration. The mecha-
nisms affecting these two rates are largely unknown, but the model serves as
a guideline for further exploration of the process.
Index Entries: Dilute-acid hydrolysis; kinetic model; heterogeneous
model.

Introduction
The model presented here deals with glucan hydrolysis alone, ignor-
ing sugar degradation. The model is simplistic, and its validity is discussed
following its presentation.

Model and Simulation Procedure


In this heterogeneous model, the total (T) surface concentration of
glucopyranose rings is assumed constant. It is further assumed that these
glucopyranoses are parts of either glucan (G) or sugar (5):
CT = Cc + Cs (kg/m2) (1)
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 451 Vol. 105-108,2003


452 Pettersson et al.

By sugar we mean all mono-, di-, or oligomers that are solvable. The
mass change owing to hydration is ignored for simplicity. The hydrolysis
step does not leave the sugar solubilized, but attached to the surface. Once
the sugar is solubilized (i.e., released from the surface), an underlying
glucan unit is revealed, thereby keeping the total surface concentration Cr
constant. The two processes of hydrolysis (h) and transport (t) (...solubiliza-
tion) are assumed to be first order:
rh = k h · CG (kg/lm 2s]) (2)
r t =k t · Cs (kg/lm 2s]) (3)
in which kh and kt are rate constants. We use the term transport for the entire
escape of the hydrolyzed sugar from the domain of the solid surface. Note
that the rate of transport is assumed independent of the sugar concentra-
tion in the bulk. Since the sugar is a mix of mono-, di-, and oligomers, the
rate constants are lumped entities, averaging a set of hydrolysis and trans-
port mechanisms.
The surface concentration of sugar (C s) is then determined from the
rate:
dCs
dt = rh - rt = k h • (C r - Cs)- k t · Cs (4)

With the initial condition Cs(t) = 0, Cs(t) becomes


k C
Cs(t) = ~ (1- e-(kh+kt)t) (5)
kh +kt
If we assume that the experimental data on remaining glucan include
also the sugar at the surface (which is hydrolyzed but not solubilized), we
are interested in the simulated amount of all material that has not been
transported away. This amount (mass) m is given by Eq. 6:

dm =-A'r =-A.k .C {t)=-A.k khC T (l_e-(kh+kt)t) (6)


dt t t S t kh + k t

The total solids area A is dependent on the conversion. Assuming that


the particles are long microcrystallites of size 3 x 3 x 100 nm (corresponding
to 6 x 6 strands of glucan), we can consider the length to be constant in
comparison to the crystallite width r, so that
A{r) = 4Lr (m 2) (7)
(8)
in which L is the total length of all particles. The assumption of a constant
particle length is not merely a mathematical convenience but is largely
supported by experiments (1). We can now describe the area dependence
on the remaining mass m:

(9)

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Simplistic Modeling Approach 453
in which p is the density, and the subscript zero denotes initial values.
Inserting Eq. 9 into Eq. 6, and dividing both sides by mo, we arrive at

d (m/mo) = _ Q ~. cp (t) (1O)


dt V mo
in which
4C k k
Q = __T -hkkt (overall rate constant, S-l) (11)
rop h + t

(12)
Parameter values used in our simulations are p = 2000 kg/m3, CT =
10--6 kg/m2, ro = 3 x 10-9 m. Some values of kh and kt are given in the next
section. MATLAB's ode-solver ode45 was used to simulate m{t) / mo, which
has a real-value solution only up to a critical time tc with m(tc) = O. If <p{t)
were equal to 1 (it actually starts off at 0 and approaches 1 exponentially),
the analytical solution would be given by Eq. 13, and tc would then be
equal to 2/ Q. Since <p{t) is not constantly 1, tc is somewhat bigger than this
value. Numerically, there is a breakdown around m{t) = O.Olmo.

(13)

This numerical limitation does not limit the use of this model, since we
have plenty of reasons to mistrust the model at high conversion anyway.

Discussion
The model invariably yields glucan profiles with a certain curvature,
but the overall rate is determined by the parameter Q (Eq. 11). In Fig. 1, two
simulations are shown, in which a 12-fold difference in the value of kt yields
different overall rates. Figure 1 also shows two experimental data sets, one
for batch and one for percolation. These data were presented earlier (2), as
were the experimental procedures (3) under which they were produced. As
can be seen in Fig. 1, the simulations coincide, to some extent, with the
experimental data. This is just a simple example to show the fundamental
applicability of the model. The observed difference between two reactors
could, in this example, be qualitatively explained by a hypothesized differ-
ence in transport efficiency.
In fact, the notational distinction between hydrolysis and transport
may be short on physical significance, since the two processes are linked in
a more complicated way than the model describes. Hydrolysis itself infers
steric changes that alter the structure of the surrounding solvation shell,
which is the beginning of solvation.
The two so-called rate constants kh and kt are hardly constant, and we
therefore refer to them hereafter as rate parameters. Not only are tempera-
ture, pH, and flow likely to influence the rate parameters, but conversion
Applied Biochemistry and Biotechnology Vol. 705-108,2003
454 Pettersson et at.
% Remaining Glucan
10'~~~---.--------,--------,--------,

40

20 ..,..,

1%~------~5~----~7.10~------~15~------=20
Time (min)

Fig. 1. Measured hydrolysis profiles for pretreated yellow poplar in (A) batch and
(T) percolation at 225°C, 0.07% (w /w) (2). The lines are simulations, in which kh =
4.93 X 10-3 in both cases, but kt differs 12-fold (1.97 x 10-3) for the upper profile, (23.68
x 10-3 for the lower).

could have a great impact as well. A dependency on conversion would


influence the curvature of the profiles, whereas the other three factors
only affect the overall rate. Quantitative-and even qualitative-descrip-
tions of these dependencies are hard to come by with current experimen-
tal procedures, since we cannot observe one rate parameter in isolation
from the other.
For all these unknowns, we still have not addressed the complicated
chemistry involved in lignocellulose hydrolysis. Bouchard et al. (1) and
Mok et al. (4) claimed the destruction of some 30% of the glucan, in a way
that glucose is never formed. Although these chemical pathways have never
been elucidated, it can be assumed that there is more to glucan chemistry
than the simple production of sugar. Given the presence of lignin and other
compounds, the picture is further obscured. It must therefore be stressed
that the complexity of the process is greater than both the model and our
understanding. However, much of the complexity can be implicitly ac-
counted for by the two rate parameters.
Xiang et al. (5) assumed that at low severity (especially low tempera-
ture), the microcrystallites are hydrolyzed at the ends, and the product is
glucose. The strong cellulosic structure remains intact. At higher severity,
this structure is weakened, and entire glucan chains are dissolved. This
idea touches a very important issue: the observed homogeneous kinetics
of a presumably heterogeneous process. In the present model, this divide

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Simplistic Modeling Approach 455
is represented by the exponent of min Eq. 10. With an exponent of 0.5
(represented by the square root sign in Eq. 10), the process is modeled as
entirely heterogeneous, whereas it becomes entirely homogeneous with
an exponent of 1. Any intermediate value is plausible, although only the
two extremes are easily interpreted. This exponent is probably heavily
dependent on conversion and other factors.
This model only concerns the glucan hydrolysis. What happens to the
dissolved sugars is a crucial question, and the reader is referred to the
works from the laboratory of Dr. Y. Y. Lee (6,7).

Conclusions
There is little reason to assume that heterogenous dilute-acid hydroly-
sis of cellulose microcrystallites can be adequately described by a simple
model with only a few parameters. There is a multitude of interdependent
mechanisms, and a comprehensive model is far out of reach. However,
insights can be gained by exploring new modelling concepts.
This work has presented a simplistic model in which the intrinsic,
heterogeneous hydrolysis rate and the heterogeneous transport rate are
coupled by the assumption of a constant glucosidic surface concentra-
tion. The mechanisms affecting the hydrolysis and transport rates are
largely unknown, but the model serves as a guideline for further explor-
ing the process.

Acknowledgments
We gratefully acknowledge the financial support of the Swedish
Energy Agency and the county of Vasternorrland.

References
1. Bouchard,J.,Abatzoglou,N.,Chornet,E.,andOverend,RP.(1989), Wood Sci. Technol.
23,343-355.
2. Torget, R W., Kim J. 5., and Lee Y. Y. (2000), Ind. Eng. Chem. Res. 39,2817-2825.
3. KimJ. 5., Lee Y. Y., and Torget R W. (2001), Appl. Biochem. Biotechnol. 91-93,331-340.
4. Mok, W. S. 1., Antal, M. J., and Varhegyi, G. (1992), Ind. Eng. Chem. Res. 31,94-100.
5. Xiang, Q., Kim, J. 5., and Lee Y. Y. (2003), Appl. Biochem. Biotechnol. 105-108,337-352.
6. Xiang, Q. and Lee, Y.Y. (2001), presented at the 23rd Symposium on Biotechnology for
Fuels and Chemicals, Breckenridge, CO.
7. Lee, Y.Y., Xiang, Q., Zhu, Y., Kim, J.S., and Kim, S.B. (2001), Annual final report to
DOE no. DE-FC36-00G010592, US Department of Energy, Golden Field Office,
Golden,CO.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0457/$20.00

Cellulosic Fuel Ethanol


Alternative Fermentation Process Designs
with Wild-Type and Recombinant Zymomonas mobilis

HUGH G. LAWFORD* AND JOYCE D. ROUSSEAU


Bio-Engineering Laboratory, Department of Biochemistry,
University of Toronto, Toronto, Ontario, Canada MS5 1A8,
E-mail: hugh.lawford@utoronto.ca

Abstract
Iogen (Canada) is a major manufacturer of industrial cellulase and
hemicellulase enzymes for the textile, pulp and paper, and poultry feed indus-
tries. Iogen has recently constructed a 40 t/ d biomass-to-ethanol demonstra-
tion plant adjacent to its enzyme production facility. The integration of enzyme
and ethanol plants results in significant reduction in production costs and
offers an alternative use for the sugars generated during biomass conversion.
Iogen has partnered with the University of Toronto to test the fermentation
performance characteristics of metabolically engineered Zymomonas mobilis
created at the National Renewable Energy Laboratory. This study focused on
strain AXIal, a xylose- and arabinose-fermenting stable genomic integrant
that lacks the selection marker gene for antibiotic resistance. The "Iogen Pro-
cess" for biomass depolymerization consists of a dilute-sulpfuric acid--cata-
lyzed steam explosion, followed by enzymatic hydrolysis. This work examined
two process design options for fermentation, first, continuous cofermentation
of Cs and C6 sugars by Zm AXIal, and second, separate continuous fermenta-
tions of prehydrolysate by Zm AXIal and cellulose hydrolysate by either wild-
type Z. mobilis ZM4 or an industrial yeast commonly used in the production of
fuel ethanol from corn. Iogen uses a proprietary process for conditioning the
prehydrolysate to reduce the level of inhibitory acetic acid to at least 2.5 giL.
The pH was controlled at 5.5 and 5.0 for Zymomonas and yeast fermentations,
respectively. Neither 2.5 giL of acetic acid nor the presence of pentose sugars
(C 6 :Cs =2:1) appreciably affected the high-performance glucose fermentation
of wild-type Z. mobilis ZM4. By contrast, 2.5 giL of acetic acid significantly
reduced the rate of pentose fermentation by strain AXIal. For single-stage
continuous fermentation of pure sugar synthetic cellulose hydrolysate
(60 giL of glucose), wild-type Zymomonas exhibited a four-fold higher volu-
metric productivity compared with industrial yeast. Low levels of acetic
acid stimulated yeast ethanol productivity. The glucose-to-ethanol conver-
sion efficiency for Zm and yeast was 96 and 84%, respectively.
*Author to whom all correspondence and reprint requests should be addressed. (Present
address: RRG, Mordale, ON Canada NOC 1HO.)

Applied Biochemistry and Biotechnology 457 Vol. 105-108, 2003


458 Lawford and Rousseau

Index Entries: Genomic integration; recombinant Zymomonas AXIOI;


Zymomonas mobilis; arabinose; xylose; ethanol; prehydrolysate; biomass
hydrolysate; acetic acid; yeast.

Introduction
Lignocellulosic feedstocks, in the form of either waste materials or
designated energy crops, offer an opportunity to greatly expand the capac-
ity of the fuel ethanol industry. Lignocellulose is recalcitrant to enzymatic
digestion by cellulase unless it has been "pretreated" to remove the hemi-
cellulose and lignin components. The hemicellulose that comprises 15-25%
of the lignocellulosic feedstock is easily hydrolyzed by dilute-acid hydroly-
sis to its monomeric sugars, the pentose (5-carbon) sugars xylose and ara-
binose; and, to a smaller extent, the hexose sugars mannose and galactose.
The amount of xylose produced is one-third to one-half the amount of
glucose produced from the saccharification of lignocellulosic material.
Hence, fermentation of the pentose sugars represents an opportunity for
major improvement in ethanol yield. Economic analyses have suggested
that, to be well positioned in the competitive liquid fuels market, cellulosic
ethanol must be produced by the rapid and efficient conversion of all the
major sugar components of the hydrolyzed cellulosic feedstock (1). Mod-
ern recombinant DNA technology has been successfully used to create
many different microbial biocatalysts-both yeast and bacteria-that are
capable of fermenting the constituent monosaccharides to ethanol.
Because the fermentation unit operation in the biomass-to-ethanol
process is located downstream from the feedstock pretreatment and
hydrolysis operations, the fermentation biocatalyst is impacted by both the
type of feedstock and the process flow configuration of the process with
respect to the distribution of the process streams from the pretreatment
(hemicellulose hydrolysis or prehydrolysis) and cellulose digestion opera-
tions (saccharifying stage reactor). From a bioengineering perspective, the
biocatalyst must conform to the performance characteristics demanded by
a particular process with respect to both feedstock and overall process
design. The feedstock affects the composition of the prehydrolysate with
respect to the type and amount of the different C 6 and Cs sugars as well as
other potentially inhibitory substances such as acetic acid (HAc).
Iogen (Ottawa, Canada) is a major manufacturer of industrial enzymes.
Iogen primarily produces cellulase and hemicellulase enzymes for the tex-
tiles, pulp and paper, and poultry feed industries. Iogen has recently built
a 40 tf d biomass-to-ethanol demonstration plant adjacent to its enzyme
production facility (2). The location of the ethanol demonstration plant
offers the advantages that the enzyme can be used without the expenses of
stabilization and preservation, and that the process sugars can be used for
enzyme production.
Although Saccharomyces yeast currently enjoys a monopoly as the fer-
mentation process biocatalyst in the fuel ethanol industry, it is not the only
ethanol-producing microorganism. By virtue of its demonstrated superior

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Cellulosic Ethanol with Zymomonas 459

Acetic acid D
......1 - - - Conditioning

60g/L Glul 30g/L XyI/3.Sg/L Ara


< 2.Sg/LHAc

CS/C6 CSTR Fermentor


Fig. 1. Iogen biomass-to-ethanol process as proposed in 1999 (11). The original
diagram has been modified to include a hydrolysate conditioning operation primarily
for the reduction in level of acetic acid. CSTR, continuous-flow stirred-tank reactor.

fermentation performance characteristics, the bacterium Zymomonas mobilis


(Zm) offers an opportunity for process improvement with respect to both
conversion efficiency (yield) and productivity (3). It has the potential to
revolutionize the fuel ethanol industry. Although Zm is not used commer-
cially (for fermentation trials at industrial scale, see ref 4; for pilot-scale
trials see review by Doelle et a1. [5]), laboratory- and pilot-scale operations
indicate that it can generate near-theoretical maximum yields from diverse
feedstocks including cellulosics. Iogen has partnered with the University
of Toronto to assess the fermentation performance of genotypic variants of
the bacterium Z. mobilis (6,7). Wild-type Zm and Saccharomyces ferment
hexoses but cannot ferment pentoses. Our work with Iogen is a continua-
tion of our ongoing collaboration with the National Renewable Energy
Laboratory (NREL) directed at assessing the physiologic and biochemical
characteristcs of NREL's patented, pentose-fermenting, recombinant
Z ymomonas cultures (8-10).
The "Iogen Process", as it was originally proposed in 1999 (11), is
schematically represented in Fig. 1. Biomass depolymerization consists
of "pretreatment" by a dilute-sulfuric acid-catalyzed steam explosion at
200-250 D C followed by enzymatic hydrolysis using on-site-generated cel-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


460 Lawford and Rousseau

Feedstock

--J
_ _-'~~ Pretreatment Cellulase Cellulase
enzyme
Digester production

Hydrolysate 80-\ OOgiL Glu

Acetic acid
Conditioning

<2.5g/L HAc

C6 Fermentor

Cs Fermentor

Ethanol

Fig. 2. Revised Iogen process flow diagram. The process involves the separate
hydrolysis and continuous fermentation of Cs and C6 components of lignocellulosic
biomass for the production of fuel ethanol.

lulase enzymes. In this continuous-flow separate hydrolysis and fermenta-


tion (SHF) process, the total hydrolysate coming from the digesters con-
tained about 6% (w Iv) glucose, 3% xylose, and 0.35% arabinose, with little
sugar oligomers (6). The lignin was removed prior to fermentation (Fig. 1).
Biomass hydrolysates contain acetic acid by virtue of the presence of the
acetylated pentosans in hemicellulose (12). The pH-dependent
inhibitory effect of acetic acid on ethanologenic biocatalysts is well docu-
mented (for a review see ref. 13). We have studied the effect of acetic acid
on both wild-type (14,15) and recombinant Z. mobilis (16). Since acetic acid
represents a major limiting factor for high-performance pentose fermenta-
tion, the Iogen process incorporates a proprietary hydrolysate condition-
ing stage that reduces the acetic acid level to <0.25% (w Iv) prior to
fermentation (c. Nicholson, February 2000, personal communication).
Figure 2 illustrates an alternate process that is currently being contem-
plated by Iogen in its biomass-to-ethanol demonstration plant. Although
the overall process remains a continuous-flow SHF process, the pentose-
rich prehydrolysate is fermented separately (Fig. 2). Adjustments to mass
loadings and other operational parameters result in separate Cs and C6
streams of the approximate composition shown in Fig. 2.
Given these two process design options, the purpose of the present
study was to examine the fermentation performance characteristics of both
recombinant and wild-type Z. mobilis cultures in pH-controlled continuous

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Cellulosic Ethanol with Zymomonas 461

fermentations using synthetic media formulated to mimic Iogen's various


process streams. NREL's patented recombinant Z. mobilis strain AX101 is a
genomic integrant that ferments both xylose and arabinose, and because it
lacks the selection marker gene for Tc resistance (17,18), it has a more accept-
able regulatory approval status (7). Another objective of this work was to
compare the C6 fermentation performance of the well-known wild-type
Zymomonas strain ZM4 (3,19) with that of an industrial Saccharomyces cerevisiae
yeast commonly used in the production of fuel ethanol from com (20).

Materials and Methods


Organisms
Recombinant Z. mobilis strain AX101 (derived from Zm ATCC 39676)
(17) was obtained from M. Zhang (NREL, Golden, CO). Stock cultures were
stored in glycerol at -70°C and pre cultures were prepared as described
previously (7). Wild-type Z. mobilis strain ZM4 was obtained from the
American Type Culture Collection (ATCC) (Rockville, MD) as ATCC 3182l.
Allyeast™ is an industrial strain of S. cerevisiae and was a gift from Alltech
(Nicholasville, KY).

Fermentation Media
The synthetic biomass hydrolysate media contained 3 giL of Difco
yeast extract (Difco Laboratories, Detroit, MI), 0.8 giL ofNH4Cl, and "Zymo
salts" as described previously (6). The amounts of D-glucose, D-xylose,
L-arabinose (Sigma, St Louis, MO), and HAc that were added to the differ-
ent fermentation media were variable. The media and stock sugar solutions
were autoclaved separately.

Preparation of Inoculum
A 1-mL aliquot of a glycerol-preserved AXIOI culture was removed
from cold storage (freezer) and transferred to about 200 mL of RM medium
(10 giL of yeast extract and 2 giL of KH 2P04) containing about 10 giL of
xylose, 10 giL of arabinose, and 30 of giL glucose in loosely capped 250-mL
Erlenmeyer flasks and grown in a waterbath shaker overnight at 30°C. This
preseed was subcultured into inoculation flasks containing the synthetic
hydrolysate medium with 30 giL of glucose, 10 giL of xylose, and 10 giL
of arabinose and grown in a water bath shaker overnight at 30°C . This
overnight culture was used as inoculum at a level of approx 10% (v Iv). The
initial optical density (l-cm light path at 600 nm) was in the range of 0.25-
0.5, corresponding to 70-140 mg of dry cell mass/L. Inocula for both wild-
type Zymomonas and the Alltech yeast culture were prepared by overnight
growth in RM medium containing 60 giL of glucose.
Fermentation Equipment
pH-controlled continuous fermentations were conducted with either
NBS C30 BioFIo chemos tats or 2-L NBS Bioflo 2000 bioreactors. The work-

Applied Biochemistry and Biotechnology Vol. 705-108,2003


462 Lawford and Rousseau
ing volumes of these chemostats were about 350 and 1500 mL,
respectively. Flow from the medium reservoir to the batch culture in the
chemostat vessel was initiated when the level of fermentable sugar had
decreased to about 3-5 giL. For strain AXI0l, the initial dilution rate was
0.025-0.03/h; for wild-type Zm, approx O.l/h; and for the yeast culture,
0.05/h. Steady state was assumed only after a minimum of 3 vol had
exchanged and when samples of effluent taken on successive days gave
similar values for cell mass, sugar, and ethanol concentrations. The pH
was monitored using a sterilizable combination pH electrode (Broadley
James, Irvine, CA). The standard pH control set point was 5.5 for
Zymomonas and 5.0 for yeast. The pH was kept constant by automatic
titration with 4N KOH. Temperature was controlled at 30°C using a cir-
culating water bath, and the agitation was moderate.
Analytical Procedures and Growth and Fermentation Parameters
Growth was measured turbidometrically at 600 nm (l-cm light path).
In all cases, the blank cuvet contained distilled water. Dry cell mass was
determined by microfiltration of an aliquot of culture followed by wash-
ing and drying of the filter to constant weight under an infrared heat
lamp. Fermentation media and cell-free spent media were composition-
ally analyzed by high-performance liquid chromatography as described
previously (10). The "process" ethanol yield was calculated as the mass
of ethanol produced per mass of fermentable sugar in the medium (Le.,
glucose and xylose). The maximum dilution rate (Dmax) was achieved
when the steady-state level of unfermented sugar in the chemostat efflu-
ent was 15% of the total amount of fermentable sugar in the medium. The
volumetric productivity was calculated as the product of Dmax and the
steady-state ethanol concentration at Dmax.

Results and Discussion


The major portion of our previous work with NREL's metabolically
engineered pentose fermenting Z. mobilis contemplated a consolidated
process design with simultaneous hydrolysis and cofermentation of the Cs
and C6 sugars in a single flow through bioreactor (9). By contrast, the bio-
mass-to-ethanol process originally proposed by Iogen (11) involved sepa-
rate hydrolysis of the cellulose and subsequent continuous fermentation of
the combined prehydrolysate and cellulose hydrolysate in the same vessel
(Fig. 1). One of the principle differences in these two designs is the concen-
tration of glucose in the fermentation medium. The composition of the
hydrolysate coming from the Iogen digester was approx 60 giL of o-glu-
cose, 30 giL of D-xylose, and 3.5 giL of L-arabinose. The continuous fer-
mentation experiments employed a yeast extract-based pure sugar
synthetic biomass hydrolysate formulated to mimic the sugar composition
of the Iogen hydrolysate. Iogen specified a target of 85% utilization of the
xylose (equivalent to 5 giL in the fermentor effluent) and a pH set point no

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Cellulosic Ethanol with Zymomonas 463
50~--------------------------------------~
rec Zm AXIOI 60g/L Glul30g/L Xyl/3.5g/L Ara (-/+ 2.5g/L HAc)
45
11 Ethanol
40
~
II> 35
~
i 30 11 No acetic acid
§ 25 j. + 2.5g/L acetic acid
u
3 20
rn
;;.,

] 15
rn

10

O+---~----~--~----~--~----~--~--~
0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Steady-State Dilution Rate (lib)

Fig. 3. Steady-state levels of xylose, arabinose and ethanol as function of dilution


rate in continuous fermentations of pure sugar synthetic hydrolysates by integrated
Zm strain AXI01. The yeast extract-based medium (see Materials and Methods) con-
tained 60 giL of glucose, 30 giL of xylose, and 3.5 giL of arabinose. The pH was
controlled at 5.5. When acetic acid was added, the concentration was 2.5 giL. No
glucose was detected in the effluent over the course of the experiment. Fermentation
parameters are summarized in Table 1.

higher than 5.5. Using integrated 2m strain AXlOl, the maximum volumet-
ric productivity in the absence of acetic acid was 3.3 gl (L·h) (Table 1). For
85% xylose utilization, Dmax was 0.078 Ih and the ethanol concentration was
42 giL (Fig. 3), representing a process yield of 0.45 gig or 88% overall
conversion efficiency (Table 1). The pH was 5.5 and the temperature was
30°C. Although the Iogen process incorporates a hydrolysate conditioning
stage, not all the acetic acid is removed. Figure 3 shows the sensitivity of
strain AXlOl to inhibition by acetic acid. With the estimated upper level of
2.5 giL of HAc in the synthetic biomass hydrolysate feed, the maximum
volumetric productivity decreased to 1.7 gl (L·h) (Dmax was 0.04/h), but the
overall conversion efficiency remained at the 88% level (Table 1). Previous
experimentation has shown that the fermentation performance can be
slightly improved by elevating the pH set point from 5.5 to 6.0 (7). In a study
reported at last year's symposium, Mohagheghi et a1. (18) compared C61Cs
cofermentation results obtained with strain AX101 with those reported by
Toon et a1. (21) for recombinant Saccharomyces yeast strains and noted that
AXlOl achieved significantly higher process yields.

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


).
:g
~
Cl..
0:>
o· Table 1
"<t>::>- Summary of Steady-State Fermentation Performance Parameters for Zm AX101, Wild-Type ZM4,
3
;;;.
q- and Commercial Yeast in Continuous Fermentations Using Synthetic Biomass Hydrolysates
'<:
III
:;, Feed composition HAc EtOH Productivity Process yield
Cl.. Dmax
0:> Biocatalyst (g/L) (g/L) (g/L) (l/h) (g/[L.hD (g/g)

iii
":;,::>- Process design option no. 1
0 Stage 1 (synthetic total hydrolysate)
~
'<:
Rec ZM AX101 60G/30X/3.5A 0 42.0 0.078 3.3 0.45
Rec ZM AX101 60G/30X/3.5A 2.5 42.5 0.040 1.7 0.45
ZM4 60G/30X/3.5A 2.5 27.0 0.380 10.3 0.29 h
.j:::..
0"> Process design option no. 2
.j:::..
C s CSTR (synthetic hemicellulose hydrolysate)
Rec ZM AX101 5G/30X/3.5A 0 16.8 0.060 1.0 0.44
Rec ZM AX101 10G/30X/3.5A 2.5 20.0 0.040 0.8 0.46
C 6 CSTR (synthetic cellulose hydrolysate)
ZM4 60Glu 0 29.4 0.395 11.6 0.49
ZM4 100 Glu 0 49.1 0.235 11.5 0.49
S. cerevisiae 60Glu 0 27.0 0.100 2.6 0.43
S. cerevisiae 60Glu 2.5 27.4 0.125 3.4 0.43
az. mobilis ZM4 = ATCC 31821; G, glucose; X, xylose; A, arabinose. For Zm fermentations, the pH was controlled at 5.5; for yeast fermentations,
~ the pH was 5.0.
bWild-type Zm culture cannot utilize pentose sugars.
-~
,§i
N
a
a
w
Cellulosic Ethanol with Zymomonas 465
30

Ethanol
,-., 25
~
'-'
Medium: 60gIL Glu/30gIL Xyl/3.5g/L Ara
III (+ 2.5gIL acetic acid) pH 5.5
= 20
=
~ [J

.l:I
=
GI
l:!
U= 15 [J

....
GI [J

S
r;LJ
;;., 10
"$=
r;LJ Note: xylose and arabinose are not fermented
5
Glucose

0
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
Steady-state Dilution Rate (lib)
Fig. 4. Steady-state levels of glucose, cell mass, and ethanol as function of dilution
rate in continuous fermentation of synthetic biomass hydrolysate by wild-type Zm
strain ZM4 (ATCC 31821). The yeast extract-based medium contained 60 giL of glu-
cose, 30 giL of xylose, 3.5 giL of arabinose, and 2.5 giL of acetic acid. The pH was
controlled at 5.5. For 95% utilization of the glucose (effluent contains 3 giL), the Dmax
=O.4/h. Fermentation parameters are summarized in Table 1.

Remaining within the constraints of the process design outlined in


Fig. 1, an alternative fermentation option that we explored involved using
wild-type Zymomonas strain ZM4 (ATCC 31821). The continuous fermen-
tation of total hydrolysate (with 2.5 giL of HAc) by strain ZM4 is shown in
Fig. 4. Although the wild-type Zm is unable to ferment either xylose or
arabinose, there are advantages to considering its deployment. First, acetic
acid is not as inhibitory for glucose fermentation by Z. mobilis (14,15); sec-
ond, although the ethanol concentration is lower (27 giL), there is a marked
improvement in productivity (10.3 g/[L·h]) because the Drnax is 0.38/h
(Table 1). This design would have merit if the Cs component could be used
for another purpose (e.g., in the enzyme plant) within the context of the
proposed biorefinery or sugar platform" for multiple products. The down-
II

side to this option is the potential for fouling of the distillation columns by
the unfermented Cs sugars during ethanol recovery.
Over the course of our work with rogen, the overall process design
evolved from a single fermentation process to a double fermentation process
with separate continuous fermentations of the pentose-rich hemicellulose
hydrolysate ("prehydrolysate") and the cellulose hydrolysate (see Fig. 2).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


466 Lawford and Rousseau
30.------------------------------------,
30gIL Xy1/5-10gIL Glu/3.5gIL Ara
Il. No acetic acid
.. + 2.5gIL HAc

Xylose
inose
O+-------~--~--~----~~------~~~
0.020 0.030 0.040 0.050 0.060
Steady-state Dilution Rate (lib)

Fig. 5. Steady-state levels of sugars and ethanol as function of dilution rate continu-
ous fermentations of pure sugar synthetic prehydrolysates by integrated 2m strain
AXI01. The yeast extract-based medium (seeMaterials and Methods) contained 30
giL of xylose, 5 giL of glucose, and 3.5 giL of arabinose. The pH was controlled at
5.5. When acetic acid was added, the concentration was 2.5 giL and the glucose
concentration was 10 giL. Fermentation parameters are summarized in Table 1.

The conditioned biomass prehydrolysate consisted primarily of xylose at


a concentration in the range of 3-4% (w Iv) and with glucose and arabi-
nose being minor components. This would be fed to the Cs continuous-
flow fermentor (Fig. 2). Cellulose would be hydrolyzed using cellulase
and, depending on the solids loading of the digester, the level of glucose
in the feed to the C6 continuous-flow fermentor could be in the range of
60-100 giL (Fig. 2). The advantage of this type of design is that it allows
for optimization by employing different biocatalysts that are best suited
for either Cs or C6 fermentation.
Pure sugar synthetic media were used in experiments designed to
assess the performance of rec Zm AX101 in the prehydrolysate fermenta-
tion and either wild-type Zymomonas (ZM4) or Alltech's S. cerevisiae yeast
in the fermentation of the cellulose hydrolysate. For the Cs fermentation,
the feed contained 30 giL of xylose,S or 10 giL of glucose, and 3.5 giL of
arabinose (with or without 2.5 giL of HAc) at pH 5.5 and 30°C (Fig. 5). The
maximum volumetric productivity for the Cs fermentor in the absence of
acetic acid was 1 g/(L·h) (Table 1). For 85% xylose utilization, Dmax was
0.06/h and the overall conversion efficiency (based on sugar input) was
90% (Table 1). With 2.5 giL of HAc in the synthetic biomass hydrolysate
feed (1% glu), the maximum volumetric productivity decreased to 0.8
gl (L·h) (Dmax was 0.04/h); however, the overall conversion efficiency
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Cellulosic Ethanol with Zymomonas 467
60~--------------------------------------------,
Dmax based on 95% glucose utilization


55

~ 50
J ...
Ethanol

45 60g/L Glucose
... 40
CI

~
• I OOg/L Glucose


35
Ethanol
I 30 Ethanol
~
Cl 25

j 20

1 15
10

0
0.0 0.1 0.2 0.3 0.4 0.5
Steady-State DUution Rate (lib)

Fig. 6. Steady-state levels of glucose and ethanol as function of dilution rate in


continuous fermentations of pure sugar synthetic cellulose hydrolysates. The biocata-
lyst was either wild-type Z. mobilis 31821 or an industrial strain of S. cerevisiae yeast.
The yeast extract-based medium (see Materials and Methods) contained either 60 giL
or 100 giL of glucose. The pH was controlled at 5.5 for Zm cultures and at 5.0 for yeast
fermentations. When acetic acid was added (dashed lines for yeast), the concentration
was 2.5 giL. Lines indicate dilution rate at 85% sugar utilization (see Materials and
Methods). Fermentation parameters are summarized in Table 1.

remained at the 90% level (Fig. 5 and Table 1). For glucose fermentation in
the C6 fermentor, at a sugar loading of 60 giL, the wild-type Zm by far
outperformed the industrial yeast in terms of both conversion efficiency
(96% for Zm vs 84% for Alltech yeast) and productivity (11.2 g/(L·h) for
Zm vs 2.6 for the yeast) (Fig. 6 and Table 1). For strain ZM4 and Allyeast
the respective Dmax values were 0.385/handO.1/h (Table 1). A particularly
interesting observation was the seemingly stimulatory effect of 2.5 giL of
acetic acid (pH 5.0) on the productivity of the yeast fermentation (Fig. 6
and Table 1). The stimulatory effect of low levels of HAc on yeast ethanol
productivity has been reported by others (22,23). At a sugar loading of 10%
glucose, Dmax for Zymomonas decreased to 0.235/h (Fig. 6), but the produc-
tivityremained at 11.5 g/(Loh) (Table 1). In the present work, replacing the
more traditional yeast biocatalyst in the C6 fermentor with wild-type
Zymomonas resulted in a significant increase in yield and more than a
fourfold improvement in productivity (Table 1). It was concluded that, for
this type of SHF process (Le., with separate Cs and C6 fermentation), the
biocatalyst of choice would be NREL's metabolically engineered stable

Applied Biochemistry and Biotechnology Vol. 105-10B, 2003


468 Lawford and Rousseau

integrant Zm AXlOI for prehydrolysate fermentation and nontransformed


Z. mobilis for fermentation of the cellulose hydrolysate. This work sup-
ports our previous claims regarding the superior fermentation character-
istics of Zymomonas in glucose-based fermentations (24,25).

Acknowledgment
We are grateful to Dr. Min Zhang (NREL) for providing the integrated
recombinant Zm strain AXIOL We also thank Drs. Ali Mohagheghi (NREL)
and Jeffrey Tolan (rogen) for helpful advice and Drs. Pearse Lyons and David
Kelsall (Alltech) for the sample of industrial yeast. This work was funded
jointly by rogen (Ottawa, Canada) and the Biochemical Conversion Element
of the Office of Fuels Development of the US Department of Energy (NREL
subcontract ZDH-9-29009-02).

References
1. Hinman, N. D., Wright, J. D., Hoagland, W., and Wyman, C E. (1989), Appl. Biochem.
Biotechnol. 20/21, 391-401.
2. Foody, B. F. and Tolan, J.5. (2001), in 23rd Symposium on Biotechnology for Fuels and
Chemicals, Finkelstein, M. and Davison, B., eds., Breckenridge, Breckenridge, CO,
Abstract no. 6-05.
3. Rogers, P. L., Lee, K. J., Skotnicki, M. L. and Tribe, D. E. (1982), Adv. Biochem. Eng. 23,
37-84.
4. Bringer, S., Sahm, H., and Swyzen, W. (1984), Biotechnol. Bioeng. Symp. 14,311-319.
5. Doelle, H. W., Kirk, L., Crittenden, R, Toh, H., and Doelle, M. (1993), Crit. Rev.
Biotechnol. 13,57-98.
6. Lawford, H. G., Rousseau, J. D., and Tolan, J.5. (2001), Appl. Biochem. Biotechnol. 91-
93,133-146.
7. Lawford, H. G. and Rousseau, J. D. (2002), Appl. Biochem.Biotechnol. 98-100,429-448.
8. Lawford, H. G. and Rousseau, J. D. (2000), Appl. Biochem. Biotechnol. 84-86,277-294.
9. Lawford, H. G., Rousseau, J. D., Mohagheghi, A., and McMillan, J. D. (2000), Appl.
Biochem. Biotechnol. 84-86, 295-310.
10. Lawford, H. G. and Rousseau, J. D. (2001), Appl. Biochem. Biotechnol. 91-93,117-131.
11. Tolan, J. S. (1999), in The Alcohol Textbook, Jacques, K. A., Lyons, T. P., Kelsall, D. R,
eds., Nottingham University Press, Nottingham, UK, pp. 117-128.
12. Timell, T. E. (1964), Adv. Carbohydr. Chern. 9,247-302.
13. McMillan, JD.(1994), in Enzymatic Conversion of Biomass for Fuels Production, ACS
Symposium Series 566, Himmel, M. E., Baker, J. 0., and Overend, R A., eds., Ameri-
can Chemical Society, Washington, DC, pp. 411-437.
14. Lawford, H. G. and Rousseau, J. D. (1994), Appl. Biochem. Biotechnol. 45/46,437-448 .
15. Lawford, H. G. and Rousseau, J. D. (1993), Appl. Biochem. Biotechnol. 39/40,687-699.
16. Lawford, H. G., Rousseau, J. D., and McMillan, J. D. (1997), Appl. Biochem. Biotechnol.
63-65, 269-286.
17. Zhang, M., Chou, Y-C, Picataggio, S. K., and Finkelstein, M. (1998), US patent no.
5,843,760.
18. Mohagheghi, A., Evans, K., Chou, Y.-C and Zhang,M. (2002), Appl. Biochem.
Biotechnol. 98-100, 885-898.
19. Roger, P.L. and Tribe, D. E. (1984), US patent no. 4,443,544.
20. Ingledew, W.M. (1999), in The Alcohol Textbook, ACS Symposium Series 566, Jacques,
K. A., Lyons, T. P., Kelsall, D. R, eds., Nottingham University Press, Nottingham,
UK, pp. 49-87.
21. Toon, S., Philippi dis, G. P., Ho, N. W. Y., Chen, Z. D., Brainard, A., Lumpkin, R E.,
and Riley, C J. (1997), Appl. Biochem. Biotechnol. 63-65,243-255.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Cellulosic Ethanol with Zymomonas 469
22. Maiorella, B., Blanch, H. W., and Wolke, C. R. (1983), Biotechnol. Bioeng. 25, 103-11l.
23. Vega, J. L., Claussen, E. c., and Gaddy, J. L (1987), Biotechnol. Bioeng. 29,429-435.
24. Lawford, H. G. (1988), in VIII International Symposium on Alcohol Fuels, New Energy
and Industrial Technology Development Organization, Tokyo, Japan, pp. 21-27.
25. Beavan, M., Zawadzki, B., Droniuk, R., Fein, J., and Lawford, H. G. (1989), Appl.
Biochem. Biotechnol. 20/21,319-326.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0471 /$20.00

Effects of Pressure Pulsation on Oxygen


Transfer Rate Measured by Sulfite Method

WEI-CHO HUANG,* CHENG S. GONG, AND GEORGE T. TSAO

School of Chemical Engineering, Purdue University,


West Lafayette, IN 47907-1283; E-mail: huangwc@che.eng.ohio-state.edu

Abstract
Pressure pulsation (PP) was investigated for its effects on oxygen transfer
rate (OTR) measured by sulfite oxidation. By manipulating airflow rate, 0.41-
1.2 vvm, and a control valve in a 4-L bioreactor, the frequency of PP was varied
at different gas pressures3-15 psig. A mathematical model of OTR was built
and compared to experimental data. OTR was also examined at constant gas
pressure, 4.5-15.0 psig. The results indicate a good agreement between mea-
surement and model prediction. OTR above 9 psig during PP showed signifi-
cant enhancement at 25°C. This proves that PP not only affects the elevation of
DO level, but also increases the interfacial area and mass transfer coefficient.

Index Entries: Oxygen transfer; pressure pulsation; sulfite oxidation.

Introduction
Oxygen limitation is a long-term problem in aerobic fermentation or
cell culture, because oxygen has low solubility in water. Dissolved oxygen
(DO) concentration is generally kept as high as possible by increasing the
oxygen transfer rate (OTR) to support aerobic processes. High agitation
and/ or airflow rate is the most commonly used strategy to improve OTR
(1,2). There are also some studies on the effects of DO (3) and periodic
changes in pressure on the bioprocess (4-6). In recent years, the techniques
of pressure pulsation (PP) (7) and oscillating dissolved oxygen tension (8)
have been applied to biologic systems, to enhance OTR or DO level in
fermenting liquid to produce more desired product (metabolite).
Our previous report showed increased productivity and yield of glyc-
erol in highly aerobic yeast fermentation by about 6.8 and 26%, respectively
(7), owing to enhanced OTR resulting from PP. The OTR was not affected
by gas flow rate above the loading point. To evaluate quantitatively the
effect of PP on OTR, a sulfite oxidation method was applied to measure
OTR at different peak pressure, Pm! and cycle, t 2 , (reciprocal of frequency)
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 471 Vol. 105-108, 2003


472 Huang et al.
of PP, respectively. A mathema tical model for oxygen transfer was derived
from the equation of OTR with the resistance of liquid film as the rate-
limiting step. The model includes considerations of various physical phe-
nomena involved in oxygen transfer. By maintaining gas at a different
constant pressure, the measured OTR was compared with that calculated
from the modeL
Materials and Methods
Pressure Pulsation
PP is achieved by placing an on-off valve at the gas outlet of an
enclosed reactor (7). When the exit valve is closed (t = 0), the pressure in
the vessel will build up owing to the accumulation of inflow; when the
exit valve is open (t =t 1), the vessel depressurizes to atmospheric pressure
at the end of each cycle (t = t2). This pressurization and depressurization,
repeated periodically, is called PP.
A 6-L New Brunswick glass, stirred bioreactor was used in this study
involving sulfite oxidation. By changing the aeration rate from 0.41 to 1.2
vvm, but fixing each cycle (tl = 24 s, t2 = 30 s), one can bring pressure in the
enclosed fermentor to pulsate at different Pms (mean pressure) between 3
and 15 psig. By changing different cycles t2 between 15 and 50 s, but keep-
ing the aeration rate from 0.63 to 1.3 vvm, and also tl / t2 =4/5, one can bring
different frequencies of PP at the same Pm! 9 psig. By adjusting inlet flow
and holding outlet flow between 0.93 and 1.1 vvm, one can achieve the
condition of different fixed pressures, from 4.5 to 15 psig. Further details
are discussed next.
Oxygen Transfer Rate
The liquid film, surrounding the gas bubbles, is usually the rate-lim-
iting step for oxygen transfer from gas bubbles to cells owing to low solu-
bility of oxygen in aqueous solutions (9). The rate of oxygen transfer (10)
from the gas to liquid phase is given by
dC L (. )
OTR=Tt=kLa C -C L

in which kL is the oxygen transfer coefficient, a is the gas-liquid interfacial


area, C* is the saturated DO concentration, and CL is the actual DO concen-
tration in the broth.

Measurement of Sulfite Oxidation


In the presence of Cu2+ as a catalyst, sulfite is oxidized to sulfate in a
zero-order reaction (11). Because of the rapid reaction, the actual DO con-
centration in the broth approaches zero. The OTR is the same as the rate of
oxygen consumption, which is half the rate of sulfate formation.
In a 6-L fermentor, 4 L of 0.2 mol/L Na 2S03 and 6.7 mL of 0.63 mol/L
CuS04 solutions were added. Under different peak pressures, cycles of PP,
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Effects of PP on OTR by Sulfite Method 473
Table 1
Effect of Different PPs on OTR Measured by Sulfite Oxidation"
Pm b (psig) Aeration rate (vvm) OTRe SE
No 1.0 2.5E-03 1.4E-04
3 0.41 2.4E-03 2.5E-04
6 0.65 2.5E-03 1.4E-04
9 0.88 2.8E-03 5.4E-05
12 1.1 3.IE-03 2.5E-04
15 1.2 3.2E-03 2.4E-04
"Same cycle of PP was kept at tl = 24 sand t2 = 30 s.
b pm = peak pressure of PP.
cOrR (M/min) was measured at 25°C with a working volume of 4 L.

or constant pressures, described before, OTR was studied at 25°C. A 5-mL


sample of sulfite solution taken from the fermentor at different intervals was
mixed with 5 mL of 0.2 N KI-12 solution. Excess 12 was titrated with the 0.1
mol/L standard thiosulfate (Na2S20 3) solution. The rate of the molar change
of Na2S20 3 solution used for the titration of two consecutive samples equals
the fourth of the OTR.

Results
Effect of Different PPs on OTR
OTR with different PPs (same cycle, t2 = 30 s) was measured three
times for each set of experimental conditions at 25°C. Table 1 shows that a
higher peak pressure of PP can reach a faster OTR. OTR with 15 psig of PP
was about 35% faster than that with 3 psig ofPP. OTR without PP was about
5% faster than that with 3 psig of PP and was close to that with 6 psig of PP.
Since kL is related to airflow rate below loading point (12), OrR with an
aeration rate below 0.65 vvm or 3 psig will be a function of aeration rate.
OTR decreased by lowering gas flow rate has a greater effect than that
increased by higher Pm. Nevertheless, this does not affect the result of our
mathematical model. The model was derived, as shown subsequently, from
OTR based on unit gas volume in the dispersion.
Effect of Different PP Cycles on OTR
The OTRs at various PP cycles were determined in triplicate at 25°C
(see Table 2). Generally, if a longer PP cycle (lower frequency) was applied,
the measurement of sulfite oxidation gave a lower OTR. For example, the
OTR with a 15-s PP cycle was 43% larger than that with a 50-s PP cycle. With
a higher frequency of PP, there was more frequent, quick expansion of gas
bubbles during the depressurization phase, which would create more fresh
bubble surface area. According to Danckwerts's (13) surface renewal model,
a freshly formed surface allows much faster interfacial mass transfer rates
including OTR.

Applied Biochemistry and Biotechnology Vol. 105-10B, 2003


474 Huang et al.
Table 2
Effect of Different PP Cycles
on OTR Measured by Sulfite Oxidationa
Cycle, t2 b (s) Aeration rate (vvrn) SE
15 1.3 3.4E-3 1.9E-04
20 1.2 3.1E-3 2.0E-04
25 1.1 3.0E-3 2.8E-05
30 0.88 2.8E-3 5.4E-05
35 0.82 2.8E-3 1.5E-04
40 0.74 2.6E-3 1.1E-04
50 0.63 2.4E-3 2.3E-04
"Same peak pressure, Pm = 9 psig, was applied.
bWhere t 1/ t2 = 4/5.
cOIR (M/min) measured at 25°C with a working volume of 41.

Table 3
Effect of Different Constant Pressures
on OTR Measured by Sulfite Oxidation
P (psig) Aeration rate (vvrn) OTRa SE
4.5 0.93 2.7E-03 1.7E-04
9 1.1 3.1E-03 1.9E-04
12 1.1 3.3E-03 1.7E-04
15 1.0 3.3E-03 1.7E-04
"OIR (M/min) measured at 25°C with a working volume of 41.

Effect of Different Constant Pressures on OTR


The effect of different PPs was compared with OTR measured at con-
stant pressure that was selected to equal the time-integrated average pres-
sure during a PP cycle. The OTRs at different constant pressures were
measured in triplicate runs at 25°C. Table 3 shows that increasing gas
pressure increased the OTR owing to the higher solubility of oxygen in the
broth, according to Henry's law. The OTR with 9 psig of PP in Table 1 is
4% higher than that of the corresponding time-integrated average pres-
sure of 4.5 psig. However, the OTR at the 50-s PP cycle in Table 2 is 13%
lower than that at a constant pressure of 4.5 psig. The main reason should
be that the aeration rate, 0.63 vvm, under the 50-s PP cycle was below the
gas-loading point. This is similar to the situation when the OTRs with
3 psig of PP were smaller than that without PP.
Discussion
Mathematical Model of OTR (Effects of PP)
Our studies were conducted at different average gas flow rates. Chang-
ing the volumetric flow rate did not change the inlet pressure, but the

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Effects of PP on OTR by Sulfite Method 475
pressure dropped across the sparger nozzle. This also changed the size of
bubbles and the gas flow rate throughout the pulsation. However, assum-
ing no coalescence of bubbles, the number of bubbles was the same owing
to the proportionality between the size of the bubbles and the gas flow rate.
Thus, it is reasonable to take an average of the size of the bubbles and
assume that the number of bubbles was the same throughout the pulsation
for mathematical modeling.
During the pressurization phase, the bubbles had longer residence
time owing to the smaller radius of the bubbles; by contrast, the bubbles
flew upward rapidly during the depressurization phase. However, we kept
the ratio of the period from the pressurization to the depressurization phase
the same. In other words, the effect of bubble residence time should not be
different in our studies.
The total surface area and volume of bubbles inside the fermentor are
written respectively as
(1)

(2)
in which n is the total number of bubbles, and r is the assumed average
radius of bubbles.
Then by dividing Eq. 1 by Eq. 2, the gas-liquid interfacial area, a, is
defined as follows:
a =A/V = 3/T (3)
Assuming the gas in the dispersion obeys the idea gas law,
pV=K (4)
in which K is a constant.
From Eq. 2 and 4, assuming no coalescence of bubbles, so that n is a
constant, and taking the reciprocal of radius, then
1/T = Bp 1/3 (5)
in which B, a constant, is defined as [4mt j(3K)F/3.
Substituting Eq. 5 into Eq. 3, and taking the first-order time derivative
of interfacial area, a,
a = 3Bp1/3 (6)

da/dt = Bp-2/3 dp/dt (7)


Applying Henry's law (HC* =p) and taking the first-order time deriva-
tive of saturated DO concentration, C*,
(8)

(9)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


476 Huang et al.
Meanwhile, the pressure change inside the fermentor by PP can be
approximated as linear functions for each cycle of the pressurization phase
(interval t from zero to tl ) and the depressurization phase (interval t from
tl to t 2), respectively:
Pl = (pJtl)t + Pa (10)

P2 = (pJ[t l - t2])t + (t2/[t2 - tl])P m+ Pa (11)


in which Pl and P2 are the pressure inside the fermentor during the pressur-
ization and depressurization phase, respectively; Pm is the peak pressure of
PP; and Pa is the atmospheric pressure.
For the fast reaction, such as sulfite oxidation, CL =0, the OTR equation
is simplified and differentiated as
d(OTR)/dt =kL(da/dtC' + adC'/dt) (12)
Substituting Eqs. 6-9 into Eq. 12, then
d(OTR)/dt =kpl/3 da/dt (13)

OTR = 3f4kp 4/3 + 1 (14)


in which constant k is defined as 4kLBH-t, and I is an integrating constant.
From Eqs. 10,11, and 14, the OTR equation can be written as
OTR l = 3f4k([pJt l ]t + pl/3 + 1 (15)

OTR 2 =3/Jc([Pm/{t l - t2}]t + [t2/{t2 - tl}]Pm + pl/3 + 1 (16)


in which OTRl and OTR2 are the OTRs during the pressurization and
depressurization phase, respectively.
For each PP cycle, the time-integrated average OTR can be given from
the integration of Eqs. 15 and 16,

OTR avg = (f;l OTRldt + L:2 OTR2dt )/t2 (17)

_ 9k ([ Pm+Pa ]7/3 -Pa 7/3) +1


OTRavg-zs-- (18)
Pm
If Danckwerts's (13) surface renewal model is satisfied, Eq. 18 can be
written as

OTR avg = MTZO.5p~l ([Pm + pi/ 3- Pa7/3) + 1 (19)


in whichM =9D l°.5 B/7H; and Dl =t 2 kL2 /60,modified diffusion coefficient.

Model Fitting
The model of OTRs with different PPs, Pm = 3-15 psig (same cycle,
30 s), was fitted with nonlinear regression provided by Flying Raichu's

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Effects of PP on OTR by Sulfite Method 477
OTR = 9.5e-5 * ~-0.5 Pm-' [( Pm + p. )7f3 - Pa713 ] + 7.0e-4
Rsqr= 0.98
3.4e-3 , - - - - - - - - - - - , - - - - - - - - - - - - - - ,
• Experimental Data
- - Model Curve

3.2e-3

3.0e-3
,-..

1:E
~ 2.8e-3
'-'

2.6e-3

2.4e-3

2.2e-3 L -_ _--l.-_ _--L_ _ _..I.-_ _--L-_ _--'L-_-----'

o 3 6 9 12 15 18

Pm (psig)

Fig. 1. Model of OTR with different PPs and same cycle, t2 =30 s.

SigmaPlot 2000 ver 6.10 (SPSS Science). Assuming that Danckwerts's (13)
model is obeyed, the experimental data in Fig. 1 show good agreement
with the model curve from Eq. 19 (M = 9.5e-5, I = 7.0e-4, Rsqr = 0.98, in
which Rsqr = 1.0 if 100% matching).
In the case of different PP cycles, t2 = 15-50 s, fitting with the same
computer program and assumption, the result of fitting also gave good
agreement between experimental data and the model curve (M = 7.0e-5, I
= 1.3e-3, and Rsqr = 0.97; see Fig. 2). However the parameter M was 26%
smaller, and I was 46% larger than that estimated from different PPs. This
can be caused by deviation from the assumptions of ideal gas law, unifor-
mity of the bubbles' size, and/or Danckwerts's (13) model.
Danckwerts's (13) model was checked by applying the parameters M
and I computed from the model with different PPs, as shown in Fig. 3. The
fitted power of t2 is -0.50 and Rsqr is 0.85, which should be reasonable
enough. However, some assumptions may need to be modified to get a
better fit.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


478 Huang et al.
OTR = 7.0e-5 * tz~.5 Pm-I [( Pm + p. )713 - p.713 ] + 1.3e-3
Rsqr=O.97
3.00-3 , - - - ' - - - - - - - - - - - - , - - - - - - - - - - ,
• Experimental Data
- - Model Curve

3.4e-3

3.2e-3

-..
3.0e-3
~
E:
~
!l:: 2.8e-3
S
2.00-3

2.4e-3

2.2e-3 L -_ _ _.L..-_ _ _.L..-_ _ _-'---_ _ _-'---_ _- - 1

10 20 30 40 50 60

tz (sec)
Fig. 2. Model of OTR with different cycles and same PP, Pm = 9 psig.

The OTR above 9 psig of PP showed significantly more enhancement


than that above the corresponding fixed time-integrated average pressure
of 4.5 psig (see Fig. 4). It is obvious that the OTRs above 9 psig of constant
pressure showed a similar shape of saturation curve. By contrast, the
experimental data and model curve of different PPs has the trend to rise
linearly. Equation 7 gave the ratio of the bubble surface area change to
the hydrostatic pressure change. This ratio was smaller if we elevated the
pressure drop. Namely, the bubbles' area change was less important than
the mass transfer coefficient at higher pressures of PP. The surface renew-
able dominated in this case. This proves that PP not only affects the eleva-
tion of DO level by higher pressure, but also increases the interfacial area
and mass transfer coefficient by different PP cycles.
Conclusion
The OTR increased with peak pressure or frequency of PP. During
the depressurization phase, it created more fresh bubble surface area with

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Effects of PP on OTR by Sulfite Method 479
OTR =9.5e-5 * ~ .(l.SO Pm· l [( Pm + p. )713 _ p.713 ] + 7.0e-4
Rsqr = 0.85
3.8e-3
• Experimental Data
- - Model Curve
3.6e-3

3.4e-3

3.2e-3
"""'
~
E
3.0e-3
~
S 2.8e-3

2.6e-3

2.4e-3

2.2e-3
10 20 30 40 50 60

~ (sec)

Fig. 3. Model of OTR with different cycles and same PP, Pm = 9 psig, by applying
parameters M and I from Fig. 1.

higher peak pressure. From the surface renewal model, fresh interfacial
area had a much higher rate of oxygen transfer than the old bubble surface
area. In addition, the higher frequency of PP created a higher level of
turbulence next to the bubble surface, which enhanced the value of kL by
increasing the frequency of surface renewal. Our studies indicated that a
good agreement of measurement and theoretical model was achieved,
and that the OTR above 9 psig of PP showed significantly more enhance-
ment than that above the corresponding time-integrated average pres-
sure at 25°C. Futhermore, keeping the aeration rate above the loading
point can provide better improvement of OTR by PP.
Acknowledgments
We acknowledge the assistance provided by Tom Huang for the de-
sign of the fermentor. This research was supported in part by a grant from
the National Science Foundation.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
480 Huang et al.

3.4e-3
• Expt. Data (Const. p)
0 Expt. Data (Integ. Avg. Pm)
- - Model Curve
• •
3.2c-3

3.0e-3
........
1:
.~
2.8c-3
~
~
b
2.6e-3

2.4e-3

2.2e-3 L -_ _- - ' -_ _ _- ' - -_ _- - - '_ _ _- ' -_ _ _- ' - -_ _- - '

o 3 6 9 12 15 18

P (psig)

Fig. 4. Comparison of OTRs, different PPs, and constant pressures.

References
1. Finn, R. K. (1954), Bacteriol. Rev. 18,254-274.
2. Hixson, A W. and Gaden, E. L. Jr. (1950), Ind. Eng. Chem. 42, 1792-1801.
3. Kilburn, D. G. and Webb, F. C (1968), Biotechnol. Bioeng. 10,801-814.
4. Kataoka, H., Sato, 5., Mukataka, 5., et al. (1986), Biotechnol. Bioeng. 28, 663-667.
5. Trager, M., Qazi, G. N., Buse, R., and Onken, U. (1992), J. Ferment. Bioeng. 74,282-287.
6. Rhiel, M. and Murhammer, D. W. (1995), Biotechnol. Bioeng. 47, 640-650.
7. Huang, W.-C, Gong, C 5., and Tsao, G. T. (2002), Appl. Biochem. Biotechnol. 98-100,
909-920.
8. Trujillo-Roldan, M. A, Pena, C, Ramirez, O. T., and Galindo, E. (20m), Biotechnol.
Prog. 17, 1042-1048.
9. Yoshida, F., Ikeda, A, Imakawa, 5., and Miura, Y. (1960), Ind. Eng. Chem. 52,435-438.
10. Shuler, M. L. and Kargi, F. (1992), Bioprocess Engineering, Prentice Hall, Englewood
Cliffs, NJ, p. 165.
11. Schultz, J. S. and Gaden, E. L. Jr. (1956), Ind. Eng. Chem. 48,2209-2212.
12. Cooper, C M., Fernstrom, G. A, and Miller, S. A (1944), Ind. Eng. Chem. 36,504-509.
13. Danckwerts, P. V. (1951), Ind. Eng. Chem. 43, 1460-1467.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0481/$20.00

Permeation Associated
with Three-Phase-Partitioning Method
on Release of Green Fluorescent Protein

THEREZA CHRISTINA VESSONI PENNA,*


EB CHIARINI, AND ADALBERTO PESSOA JUNIOR
Department of Biochemical and Pharmaceutical Technology,
School of Pharmaceutical Science, University of Sao Paulo,
Rua Antonio de Macedo Soares, 452, 04607-000, Sao Paulo/SP, Brazil,
E-mail: tcvpenna@usp.br

Abstract
Transformed cells of Escherichia coli expressing recombinant green fluo-
rescent protein (GFPuv) were subjected to two methods of extraction:
(1) freezing/ thawing/ sonication (FTS) cycles prior to the three-phase par-
titioning (TPP) method, or (2) directly to TPP extraction. The amount of
GFPuv released by the FTS plus TPP method varied: 374 Ilg/ mL (first cycle),
93-442 Ilg/mL (second cycle), 32-359 Ilg/mL (third cycle), 18-115 Ilg/mL
(fourth cycle). The GFPuv yields by the second method (TPP only) were,
23-54 Ilg/ mL for the first extract and 33-91 Ilg/ mL for the second. The FTS
plus TPP method released similar amounts of GFPuv to that extracted by
TPP; and provided a better mixture elution through the hydrophobic inter-
action column: 13-63 Ilg/ mL for FTS plus TPP methods, and 2.5-13 Ilg/ mL
for TPP. The results showed that although selective permeation is a more
laborious methodology, it was more efficient for obtaining of GFPuv in
relation to the direct extraction of the cells for TPP.
Index Entries: Recombinant green fluorescent protein; Escherichia coli;
protein purification; three-phase-partitioning method.

Introduction
The native form of the green fluorescent protein (GFP), extracted from
jellyfish Aequorea victoria, emits a brilliant green fluorescent light when
exposed to ultraviolet (uv) light (1..-360-400 nm) (1). The recombinant form
of the green fluorescent protein, GFPuv, can be successfully expressed in
prokaryotic and eukaryotic cells (2).

"Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 481 Vol. 105-108,2003
482 Penna et al.
The expressed GFPuv in the cytoplasm of Escherichia coli strains (DH5-
a, JMI09, and TB1) can be liberated from the cells by enzymatic digestion
(lysozyme) of the cellular wall, by chemical lysis (t-butanol, ethanol), or by
selective physical permeation through the freezing, thawing, and sonica-
tion methods (1,3-8).
Direct extraction from E. coli by the three-phase partitioning (TPP)
method has been used to extract, concentrate, and purify proteins and
enzymes (5,9-12). The TPP method combines several extraction advan-
tages: salting-out, isoeletric precipitation (13), solvent precipitation (14),
and osmolytic (15) and kosmotropic precipitation of proteins (16).
t-Butanol, which is miscible in water, when added to the concentrated
aqueous solution in ammonium sulfate (about 50%) partitions the mixture
in two phases: t-butanol upper layer and aqueous lower layer. t-Butanol
associates with the proteins present in the aqueous phase, precipitating
them in the interface between the aqueous lower layer and t-butanol upper
layer (9). The relationship between t-butanol and ammonium sulfate con-
centration should be appropriate in order to allow synergism between the
phases during the precipitation of proteins (11). Yakhnin et al. (5) observed
that an aqueous solution of 2 M ammonium sulfate (approx 50% of satura-
tion) provided a better recovery (90%) of extracted GFPuv in relation to the
crude extract, relative to recovery when extracted with ethanol (58%). On
the other hand, Sharma and Gupta (12) recovered 85% of an amylase inhibi-
tor using 30% saturated ammonium sulfate plus t-butanol, representing a
purification of 25-fold. When ammonium sulfate is removed, proteins dis-
solve in the aqueous phase again, maintaining their physical and biologic
properties (15-17).
The pH of a solvent determines the charge state of a globular protein.
According to Rothstein (17), the charge state of a protein can be manipu-
lated experimentally by varying the solvent properties, especially the pH.
Proteins with zero net charge (isoeletric point [pI]) show a greater tendency
to precipitate, since minimum solubility usually occurs at or near the pI.
Proteins are soluble in the conditions they evolved in, for E. coli cytoplasm,
low ionic forces, between 0.1 and 2.0 M, and neutral pH value.
The purposes of this work were to compare physical and chemical
methods in the isolation, extraction, and purification of GFPuv expressed
in the cytoplasm of E. coli DH5-a cells, and to determine the stability of
GFPuv at different pHs.

Materials and Methods


Transformation
E. coli DH5-a (3,6) was transformed with pGFPuv (Clontech), which
is a high copy number plasmid (2), by the standard calcium chloride method
(3). The transformed cells were stored (at -75°C) into Luria-Bertani (LB)
broth (USB) supplemented with 100 ~g/mL of ampicilin (Boehringer
Mannheim) and 50% glycerol.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Three-Phase-Partitioning Method 483

Expression
An overnight culture of E. coli in LB broth/ ampicillin was transferred
to the surface of LB / ampicillin/ isopropyl-~-D-thiogalactopyranoside
(IPTG) USB) agar and incubated (37°C for 24 h at 100 rpm). Using a hand
held uv lamp (395 nm, model UVL-4; UVP), four brilliant fluorescent colo-
nies were selected and transferred, each one to a tube containing LB / ampi-
cillin broth and incubated (37°C for 24 h). The 8 mL suspension of (from
four tubes) was mixed and transferred into 200 mL of LB / ampicillin broth.
The seeded broth was divided into 50-mL (cultures 1-4) lots and each trans-
ferred to a 250-mL Erlenmeyer flask. The flasks were incubated (100 rpm
at 37°C for 3 h) to an OD66onm of 0.8 (10 8 CFU /mL) when IPTG was added
(0.5 mM final concentration). After 21 h (100 rpm at 37°C), the GFPuv
expressed by induced cells was confirmed under UV light (395 nm). For all
cultures, the cellular concentration was obtained by the gravimetric method
of the biomass (mg/L) held on the surface of a 0.22-!!m membrane
(Millipore) and submitted to 105°C for about 24 h, attaining constant weight.
The biomass concentration related to isolated GFPuv was expressed by
specific productivity (!!g of gfpuvlmg of dry cell weight [DCW]).

Extraction of GFPuv by TPP Extraction Method (9)


The cultures were centrifuged (1000g for 30 min at 4°C), and the
pellets were observed under UV light (395 nm) and resuspended in cold
extraction buffer solution (XE: 25 mM Tris-HCl, pH 8.0, Trizma® Base
[Sigma, St. Louis, MO] 1.0 mM ~-mercaptoethanol W-ME] [Pharmacia
Biotech, Uppsala, Sweden]; 0.1 mM phenylmethylsulfonyl fluoride
[PMSF] [USB]. For culture 1, 10 mL of XE was cultured and then divided
into three aliquots of 450 !!L (lA, 1B, 1C), and the remaining aliquot (8.65
mL) was lyophilized and the pellet resuspended in 1.0 mL of XE (l-D).
Cultures 2-4 were resuspended in 1.0 mL of XE.
The samples of culture 1 were subjected to direct extraction by the
TPP method. For samples lA, 1C and 1D, to each 450 !!L of cell suspen-
sion, 300 !!L of 4 M (NH4)2S04 (1.6 M final concentration) and 750 !!L of
t-butanol (ratio 1:1) were added. For sample 1B, 450!!L of cell suspension
was mixed with 250 !!L of 4 M (NH4hS04 (1.4 M final concentration) and
700 !!L of t-butanol (ratio 1: 1). The mixtures were vortexed for 1.0 min and
centrifuged at 6000g for 10 min. The three phases formed were collected.
After the t-butanol upper layer and the white interfacial precipitate were
removed, another equal volume of t-butanol was mixed with the lower
aqueous layer and centrifuged. The upper layer of the system was dis-
carded. The interfacial green layer was collected and dissolved in 450 !!L
of XE buffer. While the lower layer was fluorescent, it was subjected to
repeated TPP. In every TPP extraction, the subsequent white interfaces
that showed fluorescence under UV light were dissolved in 450 !!L of XE,
and 250 !!L (sample 1B) or 300 !!L (samples lA, 1C, 1D) of (NH4)2S04 and
700 or 750 !!L of t-butanol were added, respectively.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


484 Penna et al.

Permeation (18)
Selective permeation of the pelleted cultures 2-4 was performed by
four cycles of slow freezing (-20°C; 0.S3°C / min), followed by thawing (20°C)
and sonication (or freezing/thawing/sonication [FTS]) (High Intensity
Ultrasonic Processor, Vibram cells, model 100; Sonic & Materials). The freez-
ing/thawing cycles were performed in a freeze-dryer (FTS System™,
Secfroid; Lyolab G) chamber (Dura StopTM MP). With a PT-100 thermo-
couple inserted into a pellet, the freezing/ thawing temperature was regis-
tered every minute using the software Lyphoware for Windows. Between
the freezing/thawing cycles, the microtube was kept immersed in an ice-
salt bath, and a 3.0-mm microtip ultrasonic processor was placed into the
sample, which was submitted to threefold pulse sonication at O°c. Each
pulse was at 25 vibration amplitude at alternating cycles of 6.0 s on and 1.0
s off. Thus, the permeate was subjected to TPP extraction.
Partial Purification of GFPuv with Hydrophobic Interaction Column
Extract (250 ilL) was mixed with 250 ilL of 4 M (NH4}2S04 and trans-
ferred to the top of a fast-flow, methyl support hydrophobic interaction
column (HIC). The column was previously equilibrated with 2 M
(NH4)zS04. After the column was loaded, GFPuv was retained near the
top of the column by affinity binding to the HIC resin. The loaded column
was washed first with 250 ilL of 1.3 M (NH4hS04 to elute proteins other
than GFPuv that bind with low affinity. GFPuv was eluted with 750 ilL of
buffer solution (lOmMTris-HCI,lOmMEDTA, pHS.O) and stored at 4°C.
The progress of the GFPuv through the column was observed with a UV
light, as well as confirmation of the eluted material.
Fluorescence Intensity
The fluorescence intensity of GFPuv was measured in the eluted
samples in the spectrofluorophotometer (RF-5301 PC; Shimadzu, Kyoto,
Japan), with an excitation filter of 394 nm and an emission filter of 509 nm.
Purified recombinant GFPuv purchased from Clontech ("standard
GFPuv") was used to generate a standard curve to relate protein concen-
tration to fluorescence intensity. A standard curve was prepared using
known amounts (1.52, 2.44, 3.90, 6.25, and 10.00 Ilg/mL) of standard
GFPuv diluted in buffer solution (10 mM Tris-HCI, pH S.O; 1.0 mM ~-ME;
0.1 mM PMSF). The fluorescence intensity of the samples was compared
to the standard curve from the standard GFPuv (GFPuv Ilg/mL = 0.0256
x [fluorescence intensity] + 0.S576; R2 = 0.99) and was expressed as
micrgrams of GFPuv per milliliter.
Total Protein Concentration Released
The total protein concentration released in the eluted samples was
compared to purified bovine serum albumin (BSA) in buffer solution (mol
wt of 66 kDa; Sigma) at A 280 in a spectrophotometer and expressed in mil-

Applied Biochemistry and Biotechnology Vol. 105-70B, 2003


Three-Phase-Partitioning Method 485
ligrams of BSA per milliliter. The total protein concentrations in the buffer
solution ranged from 100 to 1000 ~g/mL, the maximum A 280 being 0.615.
The relationship between total proteins and BSA was made through the
standard curve (total protein ~g/mL = 1727.2 X [OD280nm1 -26.863;R2= 0.99).
The specific GFPuv mass was expressed as micrograms of GFPuv BSA.
Electrophoresis
Samples of the extracted GFPuv were run on a 12% sodium dode-
cylsulfate polyacrylamide gel electrophoresis (SDS-PAGE). The first
30 min was performed at 50 V (70 mA/100 W). The voltage was increased
to 200 V for the last 40 min. The protein bands were visualized with
Coomassie Brilliant Blue gel staining.

Excitation and Emission Spectra


The spectral ranges of the excitation and emission for the extracted
and purified GFPuv were compared to standard GFPuv: 500-~L aliquots in
buffer (10 mM Tris-HCI, 10 mM EDTA, pH 8.0) containing (1) standard
GFPuv (2.52 ~g/mL), (2) extracted GFPuv (8.68 ~g/mL), and (3) extracted/
purified GFPuv (8.84 ~g/mL) each transferred to a quartz cuvet (3 x 10 mm
light path length x 45 mm height). Using a spectrofiuorometer, samples
were first excited from 300-450 nm, and emission intensity was recorded at
509 nm at every 0.2-nm excitation wavelength and then excited at 394 nm
and the emission scanned from 450 to 550 nm.
Stability of GFPuv at Different pH Values
Samples of the standard GFPuv and extracted/purified GFPuv
(approx 5.0 ~g of gfpuJmL) were exposed (24 h at 4°C) to different pH
values (4.0 ± 0.2 to 8.0 ± 0.2) in buffer solutions: (1) 10 mM acetate (pH 4.0
and 5.0), (2) 10 mM phosphate (pH 5.5, 6.0, and 7.0), and (3) 10 mM Tris
(pH 8.0). To each 1175 ~L of buffer solution, a 25 ~L sample was mixed.
The mixture (1200 ~L) was stored at4°C overnight, and centrifuged (6000g
for 30 min). Precipitation was not observed. The pH values were mea-
sured and correlated with the corresponding GFPuv concentration by the
standard calibration curve (Fig. 3). The pHmeter AR-20 (Fisher) was pre-
viously adjusted with standard buffers (Synth) of known pH values: 4.0,
7.0, and 9.0.
Expression of the Results
The GFPuv contents were expressed as micrograms of GFPuv per
milliliter of sample. The amount of total proteins was expressed as milli-
grams of BSA per milliliter of sample. The specific mass of GFPuv as related
to the total protein concentration (BSA) was expressed as milligrams of
GFPuv per milligram of BSA. The specific productivity was obtained from
the relationship between the GFPuv concentration and dried cell weight
(micrograms of GFPuv per milligram of DCW).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
486 Penna et al.
30~-----------------------------------,

25

O+-----~------~------~--~~------;
300 350 400 450 500 550
~~I------------------------------------~I
Wavelength (nm)

Fig. 1. The spectrum of excitation (maximum peak: 394 nm) for (a) standard GFPuv,
(b) extracted GFPuv, and (c) extracted/purified GFPuv is shown. Also shown is the
spectrum of emission (maximum peak; 509 nm) of (d) standard GFPuv, (e) extracted
GFPuv, and (f) partially purified.

,----,--,-l)J
6,00 -r-----,------,------,----,------,-----y------,

~ A. ..... ~
"b~~§tif:~~~r«..,!~,.--rc:::;;;~-c.r1
i
S,OO
1fI?~
4,00 +----+-----t-J"R!i!!!'-+----II----t---+----t
~ 3,00 +---+---A--I~'1IJ---+----+----i----+------l

2,00 i----+-~rF7.r'9------+----+---__II-----+_---I

[~
--
1,00 hi~g~pIL-r--_j---1--__t--__t--__1

3,0 4,0 s,o 6,0 7,0 8,0 9,0 10,0


pH

Fig. 2. GFPuv extracted from E. coli at different pH values: (-/1-) extracted GFPuv;
(-0-) extracted/purified with HIe; (-D-)standard GFPuv.

Results and Discussion


The characteristic spectra of excitation and emission for the standard
GFPuv, and extracted and purified (withHIC), are shown in Fig. 1. Optimal
excitation was at 394 nm and emission at 508.8 nm, for the samples tested,
similar to standard GFPuv.
The effect of pH on the samples (standard and extracted/purified
GFPuv) is shown in Fig. 2. A quite good stability of the molecule was
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Three-Phase-Partitioning Method 487
observed at pH of about 5.0-5.2, the inflection region of the curves, related
to GFPuv concentration vs pH. According to Rothstein (17), the solubility
of a protein in aqueous solution is minimum at the pH at which it is
isoeletric, the pI of the protein. For this range of pH value, where pI is
located, the protein has a zero net charge (1,2). This is observed by an
abrupt decrease in the fluorescence intensity and solubility. Ward (1) veri-
fied that the pI for the native GFP was between pH 4.7 and 5.1, correspond-
ing with our finding on GFPuv extraction efficiency. According to Scopes
(16), the pI for most plants and bacteria proteins is characterized between
pH 4.5 and 5.0, while the pI of most animal proteins is between 5.5 and 6.0.
The maximum and stable fluorescence intensity of GFPuv was shown
in the pH range of 6.0-9.8, declining abruptly between 5.5 and 4.5. For pH
values <4.5, the fluorescence intensity was minimum (Fig. 2). Bokman and
Ward (19) verified that native GFPuv maintains stable fluorescence in the
pH range of 5.5-12.0; however, fluorescence intensity decreases between
pH 5.5 and 4.4, and drops sharply above 12.0. The red-shifted green fluo-
rescent variants (EGFP and GFP-S65T) exhibit a narrow range of pH stabil-
ity between 7.0 and 11.5, drops sharply above 11.5, and decreases between
7.0 and 4.5, thusretainingabout50%offluorescenceatpH6.0 (W. W. Ward,
personal communication, 2000).
The GFPuv yields obtained by TPP extraction at different concentra-
tions of (NH4)2S04' are shown in Table 1. Samples lA, 1e, and 1D were
extracted with 1.6 M (NH4)zS04' Sample 1B was extracted with 1.4 M
(NH4)zS04'
The pool concentration of 98 f-lg of GFPuv / mL in sample 1B was about
1.2, 4.0, and 2.0 times greater, respectively, than obtained for samples 1A
(85 f-lg of GFPuv /mL), 1e (25 f-lg of GFPuv /mL), and 1D (47 f-lg of GFPuv /
mL). However, the specific mass for 1B (7.4 f-lg of GFPuv /mg) was, respec-
tively, 2.0, 5.0, and 2.0 times lower than attained for 1A (15 f-lg of GFPuv /
mg), 1e (40 Ilg of GFPuv /mg), and 1D (13 f-lg of GFPuv /mg), because the
concentrations of impurities extracted with GFPuv were 3-20 times larger.
Therefore, the final concentration of 1.6 M (NH4)2S04 in the aqueous phase
favored the GFPuv extraction yield, in accordance with Dennison and
Lovrient's (9) observations.
Even though sample 1D was 20 times more concentrated (lyophilized
and rehydrated) than lA, 1B, and 1e, the thickest white interface (from the
first TPP step) improved the specific mass in the pool (13 f-lg of GFPuv / mg
of BSA), which was close to that of sample 1A (15 f-lg of GFPuv /mg) and
three times lower than that of sample 1e (40 f-lgofGFPuv /mg). This yielded
the lowest total protein concentration (0.63 mg of BSA/mL).
The GFPuv contents obtained from FTS permeation cycles applied to
the pelleted culture followed by TPP extraction are given in Table 2. For
every permeation and extraction step applied to the sample 2, GFPuv con-
centrations were larger than those obtained for samples 3 and 4, although
the GFPuv contents in the pool of FTS cycles were close for all samples.
However, the specific masses obtained from sample 3 by TPP extractions
Applied Biochemistry and Biotechnology Vol. 105-108,2003
~ Table 1
:g GFPuv Contents Obtained from Direct Application of TPP Extractions to the Pelleted Culture 1-
~
Q..
0;)
0'
(NH4)2S04 GFPuv Extraction Total protein Specific mass Specific productivity
I')
:::- Sample Extractione (M) (llg/mL) (%) BSA (mg/mL) (Ilg GFPuv /mg BSA) (Ilg GFPuv / mg DCW)
(b
3
tn' 1A First 1.6 33 15 8.3 4.0 23
~ Second 1.6 91 42 7.1 13 62
""Q..:;, Third 1.6 92 42 5.2 18 63
0;) Pool 85 5.5 15 58
0'
;;; HIC 2.5 1.5 1.6 1.7
I')
:::-
:;, First 1.4 44 16
0
1B
Second 1.4 83 30 10 8.3 57
~
'<: 1.4 54 20 16
Third 3.4 37
Fourth 1.4 93 34 5.1 18 63
Pool 98 13 7.4 67
.j::..
co HIC 3.7 1.2 3.1 2.5
co
1C First 1.6 23 36 0.65 35 15
Second 1.6 33 53 0.85 39 23
Third 1.6 7.2 11 0.25 28 4.9
Pool 25 0.63 40 17
HIC 5.2 0.33 16 3.6
1D Firstb 1.6 54 36 6.0 9.1 37
Secondb 1.6 77 52 2.6 30 53
Thirdb 1.6 17 12 2.0 8.8 12
Pool 47 3.6 13 32
~
:-
HIC 13 2.7 4.8 8.8
a
- "Samples lA, Ie, and lD, with 1.6 M (NH4hS04 in aqueous phase, sample lB with 1.4 M (NH4)2S04 in aqueous phase and sample, 1D
'i" lyophilized and rehyrated.
a
,0;)
- bYield (%) recuperation of GFPuv in crude extract (pool) after eluted through HIe column = (HIe/pool) x 100.
N epool = mixture of equal volumes of permeates or extracts; HIe = hydrophobic interaction column (!!g GFPuv /mL of eluted material).
a
8 dYield = HIe/pool
Three-Phase-Partitioning Method 489
Table 2
GFPuv Contents Obtained from Association of Physical Permeation
by FTS Cycles Applied to Pelleted Cultures Followed by TPP Extractions
Specific
Total protein Specific mass productivity
GFPuv Yield BSA (Ilg GFPuv/ (Ilg GFPuv/
Treatment" (llg/mL) (%) (mg/mL) mgBSA) mgDCW)
Sample 2 (500 ilL of 10 mM Tris-HCl pH 8.0)-dry mass of 1.40 mg/mL
Permeation
First FTS 833 42 95 8.8 595
Second FTS 504 25 77 6.6 360
Third FTS 351 18 40 8.7 251
Fourth FTS 175 8.8 25 7.0 125
Pool 198
Extraction
First FTS 374 29 35 11 267
Second FTS 442 34 24 18 316
Third FTS 359 28 8.4 43 257
FourthFTS 115 8.9 3.6 32 82
Pool 218 23 9.4 156
HIC 63 3.5 18 45
Sample 3 (1000 ilL of 10 mM Tris-HCl pH 8.0)-dry mass of 0.92 mg/mL
Permeation
First FTS 315 33 87 3.6 342
Second FTS 411 43 136 3.0 446
Third FTS 137 14 31 4.3 149
FourthFTS 82 8.7 12 6.5 89
Pool 946 94 10 1028
Extraction
First FTS 255 41 1.7 150 277
Second FTS 257 41 5.6 45 279
Third FTS 96 15 4.6 21 104
Fourth FTS 18 2.8 1.0 17 19
Pool 196 2.4 83 213
HIC 52 32 1.6 56
Sample 4 (1000 ilL of 10 mM Tris-HCl pH 8.0)-dry mass of 1.44 mg/mL
Permeation
First FTS 429 66 62 6.9 294
Second FTS 174 27 28 6.1 119
Third FTS 39 6.0 8.6 4.5 27
Fourth FTS 7.8 1.2 4.6 1.7 5.3
Pool 233 24 9.6 160
Permeated mixture
First extraction 20 14 2.2 9.2 14
Second extraction 93 64 1.0 90 64
Third extraction 32 22 1.6 21 22
Pool 43 2.4 18 29
HIC 13 3.4 3.7 8.7
"Pool = mixture of equal volumes of permeates or extracts; HIe =hydrophobic interac-
tion column (Ilg of GFPuv /mL of eluted material).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


490 Penna et al.

200kDa

97,4 kDa

29kDa
27 kDa (GFPuv)

2 3 4 5 6 7 8

Fig. 3. SDS-PAGE of sample 3. Lane 1, molecular mass weighed 18-200 kDa (Gibco);
lane 2, standard GFPuv (20 !lg/mL); lane 3, permeate of First-cycle FIS (32 !lg/mL);
lane 4, permeate of second-cycle FIS (60 !lg/mL); lane 5, permeate of third-cycle FIS
(39 !lg/mL); lane 6, permeate of fourth-cycle FIS (46 !lg/mL); lane 7, mixture of per-
meate after IPP extraction (13 !lg/mL); and lane 8, mixture of eluted extracts from
column HIe (60 !lg/mL).

increased up to 42% (150 Ilg of GFPuv fmg for first FTSfTPP) resulting in
the largest sample concentration in the pool (83 Ilg of GFPuv fmg).
The combination of first and second FTS cycles followed by TPP
extractions resulted in yields>SO% GFPuv, respectively: 67% for sample 2,
77% for sample 3, and 93% for sample 4, indicating that the third and fourth
FTS plus TPP steps can be eliminated.
In sample 4, the pool obtained from the first through fourth FTS
cycles (equal volumes) and subjected to the first through third TPP extrac-
tions showed 10 times less total proteins (2.4 mg of BSAf mL) and specific
mass two times larger (18 Ilg of GFPuv fmg) than for sample 2 (9.4 Ilg of
GFPuv fmg), owing to the effectiveness of TPP extractions.
The samples run on gel electrophoresis showed a progressive
increase in impurity from the first up to the fourth FTS permeation (Fig. 3,
lanes 4-6), owing to a simultaneous increase in the intracellular release of
molecules other than GFPuv by successive FTS cycles (Fig. 1). A significant
removal of molecules other than GFPuv, by TPP extraction of permeated
sample 3, was confirmed in lanes 7 (before HIC) and 8 (after HIC) (see Fig. 3).
A single band between 27 (standard GFPuv) and 29 kDa (standard mol wt)
can be visualized. Therefore, the HIe procedure did not improve TPP effec-
tiveness on the GFPuv purification, confirming Sharma and Gupta's (11,12)
and Yakhnin's (5) observations.

Conclusion
The recombinant GFPuv, which was expressed by transformed cells of
E. coli DHS-u, showed similarities in fluorescence, stability, and solubility

Applied Biochemistry and'Biotechnology Vol. 105-108,2003


Three-Phase-Partitioning Method 491

to standard GFPuv. The slow FTS permeation followed by TPP extraction


provided better GFPuv yields compared with those obtained by directly
applied TPP extraction.

Acknowledgments
We thank biologist Irene A. Machoshvili for providing technical
assistance, and Olivia Cholewa for technical revision of the manuscript.
This study was made possible by financial support provided by Brazilian
Committees for Scientific Technology Research (Conselho Nacional de
Desenvolvimento Cientifico e Tecnol6gico and Funda<;ao de Amparo a
Pesquisa do Estado de Sao Paulo).

References
1. Ward, W. W. (1998), in Green Fluorescent Protein: Properties, Applications and Protocols,
Chalfie, M. and Kain, S., eds., Wiley-Liss, New York, NY, pp. 45-75.
2. Chalfie, M., Tu, Y., Euskirchen G., Ward, W. W., and Prasher, D.C. (1994), Science 263,
802-805.
3. Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989), Molecular Cloning: a Laboratory
Manual, 2nd ed, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.
4. Johnson, B.H. and Hecht, M.H. (1994), Bio!Technology 12, 1357-1360.
5. Yakhnin, A. V., Vinokurov, L.M., Surin, AK., and Alalhov, Y.B. (1998), Protein Expres-
sion Purif. 14,382-386.
6. Naglak, T.J., Hettwer, D.J., and Wang, H.Y. (1990), in Separation Process in Biotechnol-
ogy, Asenjo, J. A, ed., Marcel Dekker, New York, pp. 143-175.
7. Wheelwright, S.M. (1994), Protein Purification: Design and Scale Up of Downstream
Processing, Wiley, New York.
8. Endow, S.A and Piston, D. W. (1998), in Green Fluorescent Protein: Properties, Appli-
cations and Protocols, Chalfie, M. and Kain, S., eds., Wiley-Liss, New York, NY,
pp.271-294.
9. Dennison, C. and Lovrient, R. (1997), Protein Expression Purif 11, 149-161.
10. Dennison, c., Moolman, L., Pillary, C. S., and Meinesz, R. E. S. (2000), Afr. J. Sci. 96,
159-160.
11. Sharma, A and Gupta, M. N. (2001), Process Biochem. 37, 193-196.
12. Sharma, S. and Gupta, M. N. (2001), Protein Expression Purif. 21,310-316.
13. Bell, D. J., Hoare, M., and Dunnill, P. (1983), Adv. Biochem. Eng. 26, 1-72.
14. Conn, E. E. and Stumpf, P. K. (1980), Introdufiio it Bioquimica, Edgard Blucher, Sao
Paulo, Brazil.
15. Lehninger, A L. (1990), Principios de Bioquimica, Sarvier, Sao Paulo, Brazil.
16. Scopes, R.K. (1994), Protein Purification: Principles and Practices, 3rd ed., Springer-
Verlag, New York, NY.
U. Rothstein, F. (1994), in Protein Purification Process Engineering, vo1.18, Harrison, R. G.,
ed., Marcell Dekker, New York, NY, pp. 115-208.
18. Vessoni Penna, T. c., Chiarini, E., Machoshvili, 1. A, Ishii, M., and Pessoa, A. Jr.
(2002), Applied Biochem. Biotechnol. 98-100,791-802.
19. Bokman, S. H. and Ward, W. W. (1981), Biochem. Biophys. Res. Commun.101, 1372-1380.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0493/$20.00

Comparison of Growth Characteristics


of Panax ginseng Hairy Roots
in Various Bioreactors

GWI-TAEK JEONG,l DON-HEE PARK,*,1,2


BAlK HWANG/ AND JE-CHANG WOo4
1Faculty of Chemical Engineering,

2Biotechnology Research Institute,


3Department of Biological Sciences,
Chonnam National University, Kwangju 500-757, Korea,
E-mail: dhpark@chonnam.ac.kr; and
4Department of Biology, Mokpo National University,
Chonnam 534-729, Korea

Abstract
This study investigated the effects of flask-to-liquid volume ratio on
the growth of Panax ginseng hairy root, transformed by Agrobacterium
rhizogenes, in flask cultures and compared the characteristics of various
bioreactors for scale-up. The flask-to-liquid volume ratio was optimum at
1.5 mL of air / mL of medium in flask cultures, and hairy root growth was
not affected above the optimum ratio. In 500-mL flask culture, hairy root
showed two growth phases. After the first exponential growth, specific
growth rate decreased. The growth characteristics of P. ginseng hairy root
in various bioreactors were investigated. Hairy root growth was about 55-
fold ofinoculum after 39 d in a 5-L bioreactor and about 38-fold of inoculum
after 40 d in a 19-L bioreactor. Carbon yield was higher in a 19-L bioreactor
than in others, but it did not show any linear relationship to the growth rate
of hairy roots in bioreactors.
Index Entries: Panax ginseng; transformed hairy root; ginseng polysaccha-
ride; bubble bioreactor; flask-to-liquid volume ratio.

Introduction
Plants are the potential source of a large number of important bio-
chemical constituents such as pharmaceuticals, food additives, pigments,
flavors, fragrances, and fine chemicals. Large-scale plant cell and tissue
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


493
494 Jeong et al.
cultures for producing useful products have been considered as an attrac-
tive alternative to whole-plant extraction for obtaining valuable chemi-
cals (1-3). Large-scale culture of plant cell and tissue has been hampered
by a low growth rate, sensitivity to shear stress, surface adhesion, cell
aggregation, and contamination (4).
Transformed root cultures provide a promising alternative for bio-
technological exploitation and a method for the constant and standard-
ized production of useful metabolites of plant cells. Hairy roots are
induced by the genetic transformation of plant cells by Agrobacterium
rhizogenes and are characterized by a high growth rate; high secondary
metabolite productivity; and inherent genetic stability, reflected in stable
growth and reproduction (5). Several bioreactor designs for large-scale
production of hairy roots have been demonstrated (6,7). Bioreactors used
for hairy root culture are more complex owing to continuous growth of
hairy root. They must have a unique configuration to compensate for the
heterogeneous, cohesive, structured, and entangled nature of fibrous
roots (4). In flask and bioreactor cultures, oxygen transfer to hairy roots
submerged in medium has three external mass transfer resistances: the
gas-liquid, liquid-liquid, and liquid-solid resistances. In shake-flask cul-
tures, the important parameters affecting flask properties are flask size,
shaking speed, closure type, and medium volume (8).
Panax ginseng c.A. Meyer, which belongs to the Araliaceae family, is
one of the most famous oriental medicinal plants and is mainly distrib-
uted in Korean peninsula and China. Field cultivation of ginseng plant is
a time-consuming and labor-intensive process. It takes 4-6 yr from seed-
ing to final harvesting and requires much care because the growth is
subjected to several conditions such as soil properties, climate, patho-
gens, pets, and inplant diseases (6).
Ginseng plants have many beneficial bioactive effects on human
health, such as their hemostatic qualities and abilities to promote blood
circulation, relieve pain, cure bleeding wounds and trauma, relieve stress,
and improve immune functions (6,9,10). Many chemical, biochemical, and
pharmacologic studies of ginseng plant have been conducted. The major
compounds of pharmaceutical interaction in ginseng have been isolated
and identified to be saponin (ginsenosides), polysaccharides, antioxidants,
peptides,fattyacids, alcohols, vitamins, and phenolic compounds (6,10). In
recent years, ginseng polysaccharides have been regarded as useful com-
pounds with important pharmacologic effects such as immune stimula-
tion, as well as antitumor, antihepatitis, mitogenic, and hypoglycemic
activities (11). As cell-wall components (primary metabolite), the total cellu-
lar content of ginseng is fairly stable and a high productivity is more easily
obtained compared with other components.
In the present study, we investigated the effect of liquid-to-flask vol-
ume ratio on the growth of P. ginseng hairy roots in flask cultures and
compared culture characteristics and metabolite production in shake-flask
and various bioreactors.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Growth of Panax ginseng Hairy Roots 495

Materials and Methods


Plant Materials and Maintenance
Hairy roots of P. ginseng c.A. Meyer, induced and established by the
root-disk method (12), were used. They were maintained on hormone-free
1/2 MS (l/2-macro MS) (13) liquid and solid medium containing 30 giL of
sucrose at 23°C under dark conditions, and subcultured once every 3 wk.

Medium and Culture of Hairy Roots


In all experiments, hairy roots were cultivated in liquid hormone-free
112 MS medium containing 30 giL of sucrose. The pH of the medium was
adjusted to 5.8 with 2 N NaOH, and the medium was autoclaved at 121°C
for 15 min prior to use. Cultivations were carried out at 23°C in dark or light
conditions, using the bioreactors described next.
Shake-Flask Cultures
For shake-flask cultures, a 500-mL Erlenmeyer flask with 200 mL of
MS medium was shaken at 70 rpm in dark conditions on a rotary shaking
incubator (Vision Scientific). About 1 g fresh wt of hairy roots was inocu-
lated into the flask. All data were obtained from three samples.
Solid Culture of Hairy Roots
Solid culture experiments were performed in 250-mL flasks contain-
ing 50 mL of 1/2 MS solid medium containing 30 giL of sucrose and 0-
8 giL of agar incubated in dark, static conditions at 23°C.
Stirred Bioreactor
A 1-L bioreactor (800-mL working volume) was used for agitated
cultivation. The agitator was a magnetic bar (5 mm id, 25 mm length).
About 1 g hairy roots was inoculated in the bioreactor, and filtered air was
supplied at a rate of 0.1 vvm at the bottom.
Bubble Bioreactors
A 3-L column (2.5 L [w Iv]), 5-L bioreactor (4 L [w Iv]), and 19-L
bioreactor (17 L [w Iv]) were used as bioreactors. These bioreactors have
height I diameter aspect ratios of 7.14, 1.41, and 1.48, respectively. The
bubble bioreactors are free of internal mechanical agitation parts. Each
bioreactor was inoculated with 4, 4, and 36 g fresh wt of hairy roots, respec-
tively. The supplied aeration rate was 0.1 vvm.

Experiment Using Different Liquid-to-Flask Volume Ratios


Erlenmeyer flasks (250 or 500 mL) contained 50,100,150, and 200 mL
of liquid medium were covered with aluminum foil or a cotton stopper.
About 1 g fresh wt of hairy roots was inoculated into every flask and cul-
tured in a rotary shaking incubator (70 rpm) at 23°C under dark conditions
for 26 d.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


496 Jeong et al.
Analytical Methods
To determine cell mass, hairy roots were harvested, rinsed with dis-
tilled water, and the extra water was eliminated. Treated hairy roots were
measured as fresh wt and dry wt. The dry wt was measured gravimetically
after drying the roots at 60°C for 24 h. In the medium, reducing sugar was
measured colorimetric ally by the dinitrosalicylic acid method (14) using a
spectrophotometer (OR/4800; HACH) and glucose was used as the stan-
dard. Total sugar was measured by the phenol-sulfuric acid method (15),
and standard curves were made by sucrose. The conductivity of culture
medium was carried out at 20°C using a Model CM-20E conductivity meter,
(cell constant k = 1.013; TOA Electronics, Japan).

Extraction and Analysis of Intracellular Polysaccharide


To determine intracellular polysaccharide, 100 mg of powdered dry
hairy roots was suspended in 10 mL of distilled water, sonicated for 10 min,
and centrifuged twice at S030g for 10 min. The collected supernatant was
used to determine intracellular polysaccharide by the phenol-sulfuric acid
method (12).

Calculation of Growth Rate and Metabolite Productivity Yields


Cell growth was evaluated in terms of average cell growth rate (GR).
Average GR was defined as follows:
GR = (final cell mass - initial cell mass)/initial cell mass/time
The yield coefficients (carbon) of cell in the hairy root cultures were
calculated as follows:
Y x / s = mass of cell increased/substrate consumed

Results and Discussion


Effect of Liquid-to-Flask Volume Ratio
In flask cultures,liquid-to-flask volume ratio had a significant effect
on the biomass growth, requiring nutrients and oxygen. Figure 1 shows
the results of hairy root growth according to liquid-to-flask volume ratio
in 250- and SOO-mL flask after 26 d. The 2S0-mL flask containing 100 mL
of medium showed higher growth than the others. The final cell mass in
flasks containing 50 and 150 mL of medium was lower than that in flasks
containing 100 mL of medium. In the 2S0-mL flasks containing 150 mL of
medium, the flasks covered with a cotton stopper showed about l.4-fold
higher growth than those with aluminum foil. In the case of SOO-mL flask
cultures, flasks containing 100 and 200 mL of medium were comparable
with 2S0-mL flasks containing 100 mL medium. These results indicated
that initial liquid volume ratio was optimum at 1.5 of mL air/mL of
medium in the flask culture of hairy roots. With Atropa belladonna hairy

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Growth of Pan ax ginseng Hairy Roots 497

+
14

12 r+ r+
10

r+
~
g
0
.~
8
+
oS
~
6 r%-
...
0
0
4

0
50/250 lOon5O 15On5O 150/250+ 100/500 200/500

Liquid-lo-na k volume ratio (Medium volume/Aa k volume)

Fig. 1. Effect of medium volume size on growth of hairy roots cultured for 26 d in
250 and 500-mL flasks. All flasks were covered with aluminum foil except for 100/250+
(covered with cotton stopper).

root cultures, initial specific growth rates were higher at a SO-mL medium
volume than at 100 mL (16). In addition, Auro et al. (8) and van Suijdam
et al.(17) reported that reducing the liquid volume in flask culture allows
an increase in oxygen transfer rate. These reports suggested that air /
liquid volume ratio is a limiting factor affecting the plant cell growth, and
this parameter influences the effectiveness at mass transfer of gas-liquid
and liquid-solid in flask cultures.
Characteristics of Hairy Roots in Flask Cultures
Figure 2 shows the growth characteristics, such as biomass growth on
semilogarithmic coordinates, sugar consumption, changes in pH, and con-
ductivity of the medium, of P. ginseng hairy roots in SOO-mL Erlenmeyer
flask culture. Two phases of lag and exponential growth were observed
from a semilogarithmic representation of the data. Diauxic growth in plant
cell and tissue cultures may occur when the cell is offered a catabolizable
energy source in the presence of a more readily catabolizable energy source
(18). After a short lag period of about 4 d, the first exponential growth
phase occurred between d 4 and 12 of the culture periods, and hairy roots
had a specific growth rate ("") of 0.057 d-1• The second growth phase was
between d 12 and 42 of the culture periods ("" = 0.019 d-1). Generally, expo-
nential growth has been reported to occur only during the first few d of
hairy root culture periods by several researchers and for a range of plant
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
498 Jeong et al.

6.0
s.o i
30.0 4.0 6

20.0 30

...... S
.....
..J
2S::J'
.....
S tOI
c
0
:g
:8
7.0
s
c i
~
c
6.0
20j 41
.....,
CII
u
c
4.0
IS
CII
!:! .g.~
8
0
u 3.0 0 3.
III
:a 2.0
...
U

8
E 10 ;
0 -0- Biomass (/)
iii -+- Reducing sugar 2
- 0 - Talal Sugar S
~1
0.7
--.op- pH
- - ConducIivity
0.6 0
0 10 20 30 40 SO

Time (Day)

Fig. 2. Growth properties of P. ginseng hairy roots in SOO-mL flask culture.

species (16). After 42 d, the cultures entered stationary phase by substrate


depletion and insufficiency of culture space. Finally, at 48 4, hairy roots
grew about 16.5-fold of inoculum. The pH of the medium dropped from
5.3 to 4.5 after 12 d, but it gradually increased to 5.6 towards the end of the
growth. Total sugar and medium conductivity decreased inversely to the
increase in biomass. This decrease in medium conductivity appeared to
reflect the amount of nutrients and electrolytes consumed by the cell (16).
Reducing sugar increased during the first 20 d of the culture and decreased
afterward. The average specific growth rate of hairy roots was about 0.34
g dry wt/(g inoculum· d) for total culture periods.
Figure 3 shows the growth of hairy roots on 1/2 MS solid medium
containing 0-8 giL of agar in a solid culture environment. Hairy roots grew
about 5.8-fold on medium containing 8 giL of agar after 28 d. Solid medium
conditions containing <3 giL of agar showed lower growth. These results
occurred by mass-transfer resistance, which was caused by hairy roots
immersed in liquid or semisolid medium in static culture conditions.
Comparison of Performance of Various Bioreactors
Stirred Bioreactor
In the stirred bioreactor, agitation is performed by magnetic bar and
the filtered air is introduced from the bottom of the bioreactor. During 27

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


Growth of Pan ax ginseng Hairy Roots 499

+
6

rr
2

rr r+ rr
o
o 2 4 6 8
Agar concentration (gIL)

Fig. 3. Growth of hairy roots cultured for 27 d in 1/2 MS solid medium containing
agar as solidifier.

d of cultivation, hairy roots grew about 24-fold of inoculum. The growth


rate of hairy roots, 0.85 g dry wt/(g inoculum·d), was about 2.65-fold
higher than in the 250-mL flask culture.
Bubble Bioreactor
In the bubble bioreactor, aeration and mixing are achieved by air
sparging. Like an airlift bioreactor, the bubbles create less shear stress in a
bubble column, so that the bubble reactor is useful for hairy root culture.
Bubble column bioreactors are structurally very simple and therefore
require only a low initial investment and maintenance capital, and have a
low possibility of contamination (4). In a 3-L bioreactor with a 7.14 height/
diameter aspect ratio, the growth rate of hairy roots, 0.44 g dry wt/
(g inoculum·d), showed low growth. This result explained that the culture
space was limited by the floating of hairy roots on the upper side of
the bioreactor by a high aspect ratio and air bubbles. Air bubbles were
entrapped in the hairy root mat and resulted in floating of the root mat and
inefficient contact with the liquid phase during the culture period.
Figure 4 shows the growth characteristics of P. ginseng hairy roots in
a 5-L bioreactor for 39 d. Hairy roots increased about 55-fold of inoculum
and the growth rate was 1.38 d-1• The intracellular polysaccharide content,
0.11 g/ g on a dry wt basis, was lower than in flasks. For extended culture
periods, hairy roots caused several problems in the bubble column reactor.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
500 Jeong et at.
3.5 35 6

3.0 30
5 pH

2.5 25 "......
'""""'
e .... ~ 4

l 2.0
··0
20 .9
'OJ
c::

.~ ~Gl
.~
t)
::s 1.5 15 g
0
"80 u
:a
u ~
1.0 10 ~
___ Conductivity
0.5 -+- Reducing sugar 5
···0·· Total sugar
~pH

0.0 0
0 5 10 15 20 25 30 35 40
Time (Day)

Fig. 4. Growth properties of hairy roots cultured for 39 d in 5-L bioreactor.

First, hairy roots grew on the upper partition of the bioreactor, causing a
more limited use of culture space than suspension cultures. Second, at
liquid culture, the meristem-dependent growth pattern of hairy roots con-
structed a root ball or mat, which consists of old tissue at the core, and grew
young lateral roots around it. This root mat inhibited the mass transfer of
nutrients and oxygen to the root mat core. Hairy roots in large-scale culture
constructed greater root mats than those in small-scale culture at latter
periods of cultivation.
Figure 5 shows the time course of the growth and nutrition consump-
tion of P. ginseng hairy roots in a 19-L bioreactor. After 40 d, hairy roots
increased about 38-fold of inoculum and showed a growth rate of 0.96 d-1•
This result was about 2.6 times higher than that observed in the 250-mL
flask. Kim (19) reported that hairy roots grew about 37 times (2.7-101 g dry
wt) in a 20-L airlift bioreactor using Phytolacca esculeuta hairy root culture.
These results showed the possibility that the mass cultivation of P. ginseng
hairy roots could be accomplished in a bioreactor. Sucrose was first hydro-
lyzed and continuously consumed afterward from 32.1 to 16.8 giL during
the culture periods. The pH of the medium was consistently maintained at
4.87 after the first 20 d and slowly increased to 5.2 afterward. Medium
conductivity decreased from 2.82 to 0.82 ms I em during the culture period.
The intracellular polysaccharide content was 0.13 gl g on a dry wt basis.
This value was lower than that of natural roots (12).
Table 1 summarizes the results of hairy root growth characteristics
and metabolite production in bioreactors. The growth rates obtained from

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Growth of Pan ax ginseng Hairy Roots 501
r-------------------------------------~~ 6
3.0
0··········· ··· ......... 0 ........ . pH
30
5
2.5
25 ~
'-'
c:
20·S!
.~
.~ 1.5
I
g 15 ~
8
§
U 1.0
10 ~
CIl

__ Conductivity
0.5 - . - Total sugar 5
···0·· Reducing sugar
~pH
0.0 ...I.--.----.....----.------.r-----.----.....----.-------,r----,--.L. 0
o 5 10 15 20 25 30 40
Time ( Day)

Fig. 5. Growth properties of hairy roots cultured for 40 d in 19-L bioreactor.

bioreactors were higher than from flask cultures. Intracellular polysaccha-


ride content of hairy roots on bioreactors was in a range of 0.11-0.13 g/ g on
a dry wt basis. These results were lower than that of natural P. ginseng roots,
0.45-0.79 g/ g on a dry wt basis. Natural P. ginseng root generally contained
a lot of ginseng starch at the main root part (12). Carbon yield was greatest
in the 19-L bioreactor, but it did not show any linear relationship to the
growth rate of hairy roots in bioreactors. The relationship between growth
of carbon yield and the ionic effect needs to be studied in more detail.

Conclusion
In flask and bioreactor cultures, oxygen transfer to hairy roots sub-
merged in medium have three external mass-transfer resistances. Thus, the
most important parameters affecting flask culture properties are flask size,
shaking speed, closure type, and medium volume. The liquid-to-flask vol-
ume ratio was optimum at 1.5 mL of air / mL of medium in flask cultures.
In 500-mL flask culture, hairy roots showed two growth phases. After the
first exponential growth phase, specific growth rate was decreased. Hairy
roots showed a growth rate of about 1.38 and 0.96 d-1 in 5 and 19-L
bioreactors, respectively. The yields of sugar and conductivity were higher
in the 19-L bioreactor than in the others, butthere was no linear relationship
to the growth rate of hairy roots in bioreactors. Intracellular polysaccharide
content of hairy roots on various bioreactors was in the range of 0.11-0.13
g/ g on a dry wt basis. It was lower than those of natural ginseng roots, 0.45-
0.79 g/ g on a dry wt basis.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
~
[
Il:J
c'
g.
~
iii'
q-
-.::
~
::J
0..
Il:J Table 1
c' Comparison of Growth Kinetics and Metabolite Production
~
::J of p, ginseng Hairy Roots in Bioreactors·
o
~
-.:: Culture Aspect Working Growth Growth Intracellular Carbon
Type of time ratio volume ratio rate polysaccharide yield
bioreactor (d) (height/ diameter) (L) (fold) (d-1) (g/g) (g dry wt/ g sugar)
V1
o Static culture 28 0.05 5.8 0.21 NT NT
N
Shake-flask culture 48 0.20 16.5 0.34 0.19 0.19
Stirred type 27 1.43 0.80 24.0 0.85 NT 0.19
Bubble column type 25 7.14 2.60 11.0 0.44 0.11 0.13
Bubble column type 39 1.41 4.00 55.0 1.38 0.11 0.22
Bubble column type 40 1.48 17.00 38.0 0.96 0.13 0.18
aNT, not tested.

~
:--

o
-~
~
N
§
Growth of Pan ax ginseng Hairy Roots 503
References
1. Stafford, A., Moris, P., and Fowler, M. W. (1986), Enzyme Microb. Technol. 8,578-587.
2. Dicosmo, F. and Misawa, M. (1995), Biotechnol. Adv. 13(3),425-453.
3. Chattopadhyay, 5., Farkya, 5., Srivastava, A. K., and Bisaria, V. S. (2002), Biotechnol.
Bioprocess Eng. 7, 138-149.
4. Giri, A. S. and Narasu, M. L. (2000), Biotechnol. Adv. 18, 1-22.
5. Hooykaas, P. J. J. and Schilperoort, R. A. (1992), Plant Mol. BioI. 19,15-38.
6. Nam, K. Y. (1996), Recent Korea Ginseng, Korea Ginseng & Tobacco Research Institute,
Taejon, Korea.
7. Nobuyuki, V. and Takesh, K. (1994), in Advances in Plant Biotechnology, Ryu, D. D. Y.
and Furusaki, 5., eds., Elsevier, Amsterdam, The Netherlands, pp. 307-338.
8. Auro, M. A., Hodge, H. M., and Roth, N. G. (1957), Ind. Eng. Chem. 49, 1237-1238.
9. Jung, N. P. and Jin, S. H. (1996), Korean J. Ginseng Sci. 20(4),431-471.
10. WU, J. and Zhong, J. J. (1999), ,. Biotechnol. 68,89-99.
11. Baek, N. I., Park, C H., and Park, C E. (2000), Biotechnol. Bioprocess Eng. 6(6),433-437.
12. Jeong, G.T., Park, D.H. and Hwang, B. (2002), Appl. Biochem. Biotechnol. 98-100,
1129-1139.
13. Murashige, T. and Skoog, F. (1962), Physiol. Plant. 15,473-497.
14. Miller, G. L. (1959), Anal. Chem. 31(3),426-428.
15. Chapline, M. F. and Kendy, J. (1986), Carbohydrate Analysis: A Practical Approach, IRL
Press, Oxford, UK.
16. Kanokwaree, K. and Doran, P. M. (1997), J. Ferment. Bioeng. 84(4), 378-381.
17. van Suijdam, J. C, Kossen,N. W. F., and Joha,A. C. (1978), Biotechnol. Bioeng. 20, 1695-
1709.
18. Peraza-Luna, F., Rodriguez-Hedida, M., Arios-Castro, C, Bessiere, J. M., and Calva-
calva, G. (2001), J. Agric. Food Chem. 49,6012-6019.
19. Kim, Y. H. (1994), PhD thesis, Chungbuk National University, Chungbuk, Korea.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0505/$20.00

Heterogeneous Aspects
of Acid Hydrolysis of a-Cellulose

QIAN XIANG, l Y. Y. LEE, *,1


PAR O. PETTERSSON,2 AND ROBERT W. TORGET3
1 Department of Chemical Engineering,
230 Ross Hall, Auburn University, Auburn, AL 36849,
E-mail: yylee@eng.auburn.edu;
2Mid Sweden University, 891 18 Ornskoldsvik, Sweden; and
3 National Renewable Energy Laboratory,
1617 Cole Blvd., Golden, CO 80401

Abstract
Hydrolysis of a-cellulose by H 2S04 is a heterogeneous reaction. As such
the reaction is influenced by physical factors. The hydrolysis reaction is there-
fore controlled not only by the reaction conditions (acid concentration and
temperature) but also by the physical state of the cellulose. As evidence of
this, the reaction rates measured at the high-temperature region (above
200°C) exhibited a sudden change in apparent activation energy at a certain
temperature, deviating from Arrhenius law. Furthermore, a-cellulose, once
it was dissolved into concentrated H 2S04 and reprecipitated, showed a reac-
tion rate two orders of magnitude higher than that of untreated cellulose,
about the same magnitude as cornstarch. The a-cellulose when treated with
a varying level of H 2S04 underwent an abrupt change in physical structure
(fibrous form to gelatinous form) at about 65% H 2S04, The sudden shift of
physical structure and reaction pattern in response to acid concentration and
temperature indicates that the main factor causing the change in cellulose
structure is disruption of hydrogen bonding . Finding effective means of
disrupting hydrogen bonding before or during the hydrolysis reaction may
lead to a novel biomass saccharification process.

Index Entries: Acid hydrolysis; cellulose; hydrogen bonding; kinetics;


crystallinity.

Introduction
Acid-catalyzed cellulose hydrolysis is a complex heterogeneous reac-
tion. It involves physical factors as well as the hydrolytic chemical reaction.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 505 Vol. 705-108,2003


506 Xiang et al.

~\~~~~
nr +H· m

- DomiaantPalhway

Fig. 1. Mechanism of acid catalyzed hydrolysis of JH -4 glucan.

The molecular mechanism of acid-catalyzed hydrolysis of cellulose (cleav-


age of f3-1-4-glycosidic bond) follows the pattern outlined in Fig. 1 (1). Acid
hydrolysis proceeds in three steps. The reaction starts with a proton from
acid interacting rapidly with the glycosidic oxygen linking two sugar units,
forming a conjugate acid. The cleavage of the C-O bond and breakdown of
the conjugate acid to the cyclic carbonium ion then takes place, which
adopts a half-chair conformation. After a rapid addition of water, free sugar
and a proton are liberated (2-5). The formation of the intermediate car-
bonium ion takes place more rapidly at the end than in the middle of the
polysaccharide chain. In accordance with this, the yield of monosaccha-
rides after partial hydrolysis is higher than that calculated on the basis of
a random bond cleavage (1).
The global kinetics of acid hydrolysis was first described by Saeman (6)
as two pseudo-homogeneous consecutive first-order reactions. Hydrolysis
of glycosidic bonds also follows a first-order reaction (7,8). The first-order
rate constants obey the Arrhenius equation with a modification to include
the acid dependence term
E;
- RT
k i = kio x Ami X e
where kio is the pre-exponential factor, A is the concentration of acid, mi is
an exponent indicating the acid effect, and Ei is the activation energy.
The first-order kinetic equation generally applies to reactions in
homogeneous phase. Therefore, first-order reaction is justifiable for
hydrolysis of oligosaccharides that are soluble in the hydrolyzing medium.
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Acid Hydrolysis of a-Cellulose 507
In actual hydrolysis with dilute acids, a heterogeneous reaction takes place,
yielding "hydrocellulose," a product with reduced degree of polymeriza-
tion (DP) but higher crystallinity (9). The rate of hydrolysis of cellulose in
crystalline form is one to two orders of magnitude lower than that of homo-
geneous hydrolysis of soluble model compounds. The hydrolysis of cellu-
lose is strongly influenced by the degree of crystallinity and the swelling
state of cellulose (5). The reactivity of cellulose is also affected by mechanical
disintegration and/ or decrystalli-zation procedures (5,10).
It is quite obvious that acid-catalyzed cellulose hydrolysis is a hetero-
geneous reaction in which the nonreaction factors (e.g., crystallinity, diffu-
sion barrier, physical conformation) represent a major part of the overall
resistance (11). The present investigation was undertaken to verify the na-
ture of the nonreaction resistances in cellulose hydrolysis and to provide
further understanding of the heterogeneous aspects of this reaction.

Materials and Methods


a-Cellulose
Chemicals were of analytical grade purchased from Sigma-Aldrich. a-
Cellulose (product no. C8002) was analyzed for sugars, moisture, and ash
content according to the National Renewable Energy Laboratories (NREL)
procedures (12). a-Cellulose contained 92.2% glucan, 3.4% xylan, and 3.2%
mannan on dry basis. Ash content was negligible.

Experimental Apparatus and Procedures


Hydrolysis of a-cellulose was performed in both batch and bed-
shrinking flow through reactors. For diluteacid-catalyzed hydrolysis at
high-temperature conditions, experiments were performed using sealed
tubular reactors. The reactors (19 .0-cm3 internal volume) were constructed
of Hastelloy C-276 tubing (0.5 in. id x 6 in. length) capped with Swagelok
end fittings. Approximately 0.6 g of a-cellulose and 12 g of 0.07% H 2S04
solution were loaded into a reactor. The solid/liquid ratio was main-
tained at 1/20 in all batch operations. Two separate fluidized sand baths
were used: one for preheating, the other for reaction temperature control.
The reactors were first submerged into the preheating sand bath set at
30°C above the desired reaction temperature for rapid preheating. After
1.5 min, the reactors were then quickly transferred into another sand bath
set at the precise desired reaction temperature. After the desired reaction
time, each reactor was taken out of the sand bath and quenched in a cold-
water bath. The time variation of the solid composition and the mono-
meric sugar concentrations in liquid were determined. Batch hydrolysis
with 4% H 2S04 at 120°C was conducted in an autoclave using plastic-
capped Pyrex glass bottles.
For treatment of a-cellulose with concentrated H 2S04, 3 g of cellulose
was added to 50-mL solutions with varying H 2S04 concentrations (50-72%).
The treatment was conducted at 25°C for 4 h. After treatment, distilled
Applied Biochemistry and Biotechnology Vol. 105-108,2003
508 Xiang et al.
water was added to dilute the acid and then the remaining or reprecipitated
cellulose solid was filtered and washed. The pretreated a-cellulose was
stored wet with 50-70 % moisture content for further experiments includ-
ing X-ray, scanning electron microscopy (SEM), and acid hydrolysis tests.
Analytical Methods
Solid samples were analyzed for glucan content according to the NREL
procedures (12). Oligomeric sugars in the hydrolysate liquor were con-
verted to monomers using 4 wt% H 2S04 hydrolysis at 120°C for 60 min.
Sugars and other compounds were determined by high-performance liq-
uid chromatography (HPLC) using a Bio-Rad Aminex HPX-87P column.
The X-ray diffraction test on original and pretreated a-cellulose was con-
ducted using a Rigaku X-ray D I Max-B Diffractometer. SEM was done using
a Zeiss DSM 940 scanning electron microscope.

Results and Discussion


To assess the heterogeneous nature of the hydrolysis reaction, experi-
ments were carried out beyond the range of reaction conditions normally
applied for hydrolysis. The intent here was to apply reaction conditions
severe enough to bring about changes of the physical structure of cellulose
while the reaction is taking place.
The first series of experiments on acid hydrolysis of a-cellulose was
conducted at temperatures above 200°C and extremely low acid conditions
(0.07%) using batch reactors. The results of these experiments are summa-
rized in a semilog plot for a-cellulose as shown in Fig. 2, the slope of the plot
representing the first-order rate constant. The first-order reaction was con-
firmed by the linearity of the plots for all temperatures.
If the hydrolytic reaction is the controlling resistance in the overall
process, the Arrhenius equation should apply over the entire temperature
range. However, the experimental data in Fig. 2 contradict this hypothesis,
showing a rather abrupt increase in reaction rate between 210 and 225°C.
The rate constants obtained in this experiment were also put into the
Arrhenius plot, In(k) vs 1 IT (Fig. 3). The sudden departure of the rate con-
stants from normal Arrhenius pattern is also seen in Fig. 3 with a breaking
point near 215°C. The actual measured rate constants at 215-245°C are
much higher than those predicted from the data taken over 185-205°C. For
example, the measured rate constant at 235°C is 3.6 times the value extrapo-
lated from the Arrhenius equation. We have also observed a similar behav-
ior from hydrolysis of pretreated yellow poplar.
This finding indicates that the kinetic behavior of cellulose is strongly
dependent on the physical state of the substrate. Within the low-tempera-
ture region (below 215°C), the a-cellulose is in crystalline structure; thus,
it retains the kinetic behavior unique to this structure. Above 215°C, the
data in Figs. 2 and 3 show a different kinetics with higher apparent activa-
tion energy (higher sensitivity to temperature). The chemical reactions are

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Acid Hydrolysis of a-Cellulose 509

205"C

215 D C

10+-----~------~----_,------~------r_----~------~----_,~

o 6 10 15 20 26 30 35 40
Time (min)

Fig. 2. Effect of temperature on hydrolysis rate (0.07 % H 2S04 , batch reactor) .

0.5.---------------------------------------------------------,

·05

·15 •
sc •
:"
-- ·25
.E
....... """"'" "'"''""'"'' "'" .. .. ".. ......
-35

23~ · C 21G OC 195 · C 185 · C


-55+---------r--------,--------~--------~--------_r------~
00019 000195 0002 000205 00021 000215 00022
1fT (11K)

Fig. 3. Arrhenius plot for first-order rate constants in hydrolysis of a-cellulose


(185-245°C, batch reactor, 0.07% H 2S04 ).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


510 Xiang et al.
generally more sensitive to temperature than the nonchemical reaction
factors (physical factors). At high temperature (215°C), the physical factors
seem to be eliminated, making the true chemical reaction the major influ-
ential factor in the overall process. The overall process therefore becomes
more temperature sensitive. We believe that the sudden change in the
reaction behavior (especially the activation energy) resulted from a tem-
perature-induced disruption of the physical structure of cellulose.
Sasaki et. al. (13,14) reported a similar observation for cellulose
hydrolysis in supercritical water. They found that cellulose is rapidly dis-
solved and depolymerized in supercritical water at 300-320°C with no
acid. The cellulose disappearance rate under these conditions is far above
the prediction by extrapolation of the Arrhenius equation from the data at
260-280°C. They reported a one-order-of-magnitude jump of hydrolysis
rate when the temperature was raised from a sub critical temperature to
near or above one critical temperature (approx 300°C).
To further prove the effect of physical state of a-cellulose on hydroly-
sis, we attempted to verify whether the kinetics is affected when the cel-
lulose is physically altered before the reaction. Various solvents can alter
the physical structure of cellulose. Some solvents including concentrated
H 2S04 can dissolve cellulose. In the subsequent experiments, the cellulose
substrate was pretreated with concentrated H 2S04 for 4 h at 25°C. Acid
concentrations of 50, 55, 60, 65, 70, and 72% were applied. The resulting
slurry / solution was then diluted with water to 4% acid. The substrates
thus prepared were then subjected to further hydrolysis at 120°C and 4%
acid concentration.
Figure 4 presents the hydrolysis profiles of cellulose pretreated with
concentrated H 2S04• When pretreatment was done with 60% or less H 2S04,
the hydrolysis of cellulose was extremely slow, basically at the same level
as untreated cellulose. However, when a-cellulose was pretreated with
65% H 2S04 or higher, most of the cellulose was dissolved. When it was
diluted with water, part of the dissolved cellulose was reprecipitated. The
hydrolysis rate of the reprecipitated cellulose was two orders of magni-
tude higher than that of untreated a-cellulose. Beyond 65% H 2S04, the
increase in hydrolysis rate was again gradual with respect to acid concen-
tration. The reprecipitated cellulose was hydrolyzed at about the same
rate as cornstarch under the identical hydrolysis condition. It is therefore
reaffirmed that the hydrolysis reaction is indeed strongly influenced by
the physical state of the cellulose. The acid hydrolysis as a rate process has
two different types of resistance: reaction and physical. If so, the physical
resistance is two orders of magnitude greater than the reaction resistance.
X-ray diffractograms were taken for the cellulose and reprecipitated
cellulose. As shown in Fig. 5, the highly crystalline structure of untreated
cellulose was totally disrupted and a completely different diffraction pat-
tern with near zero crystallinity appeared after dissolution into 65% H 2S04
and reprecipitation. SEM photographs were taken for untreated a-cellu-
lose and those treated with 55,60, and 65% H 2S04• Untreated a-cellulose

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Acid Hydrolysis of a-Cellulose 511

Treated with 55% HoSO.


50% H2 SO•

.............................................. ........ ·S·St,4· i-i;SO:······

1 .......................................................................... . ......... . .

0. 1 +-----~------~-----.------r-----_r------r------r------r---~~
10 10 30 40
Tune (min)
50 10 70 10 '0

Fig. 4. Hydrolysis profiles of a-cellulose pretreated with various concentrations of


H 2S04 , Pretreatment was conducted at 25°C for 4 h. Hydrolysis was carried out at
120°C and 4% H 2S04 ,

7000

6000 Alpha-Cellulose

5000

4000 Alpha Cellulose


~
III
c:
OD
Treated with 65';' H2SO.
~ 3000

2000

1000

0
10 12 14 16 18 20 22 24 26 28 30
Angle

Fig. 5. X-ray diffractograms of a-cellulose (original and treated with 65% H 2S04 ),

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


512 Xiang et al.

Fig. 6. SEM photographs of a-cellulose treated with different levels of concen-


trated H 2S04 •

and samples treated with 55 and 60% acid are seen to retain the original
fibrous structure although the fibers were broken into smaller fragments by
acid treatment (Fig. 6). However, the sample treated with 65% acid shows
a completely different picture. The original fibrous form of cellulose disap-
peared and changed into a gel-like substance. When the cellulose fibers are
dissolved into concentrated acid, the bundles of glucan chains are sepa-
rated into multiple single chains. As the acid is diluted, the dissolved glucan
chains reassociate. When this happens, the glucan chains do not go back to
the original orderly structured fibrillar form but form an irregular bundle.
We note that the change in cellulose structure owing to acid treatment is
gradual to a certain point (60% acid in this case) and then undergoes a
drastic change beyond that point. At higher temperature, it requires less
concentrated acid to undergo this drastic change. For example, at 70°C, it
requires only 50% H 2S04 • We mentioned earlier a similar behavior with
respect to temperature: a drastic increase in hydrolysis rate at a certain
temperature. We have confirmed that these sudden changes in kinetic
behavior and crystallinity are owing to structural changes in the cellulose.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Acid Hydrolysis of a-Cellulose 513
The cellulose structure is closely related with the hydrogen bonding exist-
ing inside the cellulose chains. The existence of hydrogen bonds in cellulose
molecule is well documented (1). The hydrogen bonding exists within a
single chain of glucan (intramolecular hydrogen bonding) and between
the adjacent glucan chains (intermolecular hydrogen bonding ). The inter-
molecular hydrogen bonds are believed to be the primary factor holding
the cellulose chains together forming the fibrous structure. The state of
hydrogen bonding in cellulose also determines other physical properties
of cellulose, such as the extent of crystallinity.
Among all possible nonreaction factors (e.g., physical conformation,
diffusion, crystallinity, chemical composition), the state of hydrogen bond-
ing stands out as the primary factor controlling the main resistance in acid
hydrolysis of cellulose. There is yet another support for this contention. Of
all the physical factors, only hydrogen bonding can undergo such an abrupt
change in reaction rate and structure in response to temperature and con-
centration of acid as seen in our experiments.
The state of hydrogen bonding is the primary factor determining the
molecular level structure of cellulose. Kinetics of acid hydrolysis of cellu-
lose is therefore strongly dependent on the state of hydrogen bonding. A
better understanding of hydrogen bonding as to how it relates to the
molecular structure of cellulose and finding an effective means to disrupt
the hydrogen bonding may prove to be a fruitful way to establish acid
hydrolysis as a viable biomass saccharification process.
Acknowledgments
We wish to acknowledge the help of Dr. Changshin Sunwoo from
Chonnam National University, Korea; Dr. Mike Miller from Auburn Uni-
versity for SEM photographs; and for David Wood and Wei Zhang assis-
tance in the reaction experiments. This research was sponsored by the
US Department of Energy under Cooperative Agreement DE-FC36-
01GOl1072, and by NREL under subcontract ACO-1-31003-01.
References
1. Fengel, D. and Wegener, G. (1984), Wood Chemistry, Ultrastructure, Reactions, Walter
de Gruyter, Berlin, Germany.
2. Shafizadeh, F. (1963), TAPPI J. 46, 381-383.
3. Timell, T. E. (1964), Can. J. Chem. 42, 1456-1472.
4. Harris, J. F. (1975), Appl. Polym. Symp. 28,131-144.
5. Philipp, B., Jacopian, V., Loth, F., Hirte, W., and Schulz, G. (1979), in Hydrolysis of
Cellulose: Mechanisms of Enzymatic and Acid Catalysis, Advances in Chemistry Series
No. 181, Brown, Jr. R. D. and Jurasek, L., eds., American Chemical Society, Washing-
ton, DC, pp. 127-143.
6. Saeman, J. F. (1945), Ind. Eng. Chem. 37,43-52.
7. Springer, E. L. (1966), Tappi 49,102-106.
8. Daruwalla, E. H. and Shet, R. T. (1962), Text. Res. J. 32,942-954
9. Nelson, M. L. (1960), J. Polym. Sci. 43, 351-37l.
10. Millett, M. A., Effland, M. J., and Caulfield, D. F. (1979), in Hydrolysis of Cellulose:
Mechanisms of Enzymatic and Acid Catalysis, Advances in Chemistry Series No. 181,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


514 Xiang et al.
Brown, Jr. R. D. and Jurasek, L., eds., American Chemical Society, Washington, DC,
pp.71-89.
11. Torget, R. W., Kim, J. 5., and Lee, Y. Y. (2000), Ind. Eng. Chern. Res. 39,2817-2825.
12. (1995) NREL Chemical Analysis and Testing Standard Procedure, No. 001-014,
National Renewable Energy Laboratories, Golden, CO.
13. Sasaki, M., Kabyemela, B., Adschiri, T, Malaluan, R., Hirose,S., Takeda, N., and Arai,
K. (1997), unpublished poster presentation in Fourth International Symposium on
Supercritical Fluids, Sendai, Japan.
14. Sasaki, M., Fang, Z., Fukushima, Y., Adschiri, T., and Arai, K. (2000), Ind. Eng. Chern.
Res. 39,2883-2890.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0515/$20.00

Characterization of Molecular Weight


Distribution of Oligomers from Autocatalyzed
Batch Hydrolysis of Xylan

XIA LI,· ALVIN O. CONVERSE, AND CHARLES E. WYMAN


Thayer School of Engineering, Dartmouth College,
8000 Cummings Hall, Hanover, NH 03755,
E-mail: xia.li@dartmouth.edu

Abstract
Oat spelt xylan was treated with water in a batch reactor at temperatures
of 180 and 200°C. Ion-moderated partition (IMP) chromatography was then
applied to separate oligomers in solution according to their molecular size.
Calibration of the IMP measurements based on peak height was found to
quantify dissolved monomer and oligomer yields welL Oligomer concentra-
tions in the liquid hydrolysate were also determined from the difference in
xylose monomer concentrations measured by high-performance liquid chro-
matography before and after posthydrolysis of dissolved xylooligosaccha-
rides to xylose. Delayed formation and then rapid disappearance of oligomers
from DP10 to DP2 was observed by IMP, and total oligomer yields measured
by IMP and posthydrolysis were very similar at these times. However, while
IMP detected virtually no oligomers initially, posthydrolysis measurements
gave significant amounts of soluble oligomers at these times, indicating that
oligomers with chain lengths> 10 were in solution but not detectable by the
IMP system used.
Index Entries: Autohydrolysis; hydrolysis; ion-moderated partition chro-
matography; oligomers; thermochemical; xylan.

Introduction
The fractionation of lignocellulosic materials for production of a va-
riety of marketable chemicals is an attractive approach for biomass utili-
zation, and the hemicellulose portion can be hydrolyzed into sugars for
fermentation to a variety of valuable products. Xylooligomers are impor-
tant intermediates in hemicellulose hydrolysis for pretreatment and sugar
recovery. They also have potential applications in many fields including
pharmaceuticals, feed formulations, agricultural applications, and func-
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 515 Vol. 105-108,2003


516 Li et al.

tional foods. However, their comparatively high production costs hinder


such uses. A better understanding of the role of xylooligomers in hemi-
cellulose hydrolysis may lower the cost of biomass pretreatment and
expand the use of xylooligomers.
Production of xylooligomers from lignocellulosic materials or xylan
can be carried out in a single step by reaction with steam or water, often
termed autohydrolysis, producing solubles consisting of primarily xylo-
oligomers, xylose, furfural, and other degradation products, and leaving
a solid residue made up of primarily cellulose and lignin(1-3). The pres-
ent study characterizes the molecular weight distribution of oligomers
released during the batch autohydrolysis of xylan at 180 and 200°C and
follows xylooligomer time profiles for xylan autohydrolysis at 200°C for
a 5% solids concentration.

Materials and Methods


Substrates
Oat spelt xylan was obtained from ICN (Aurora, OH). Its composition
was measured and found to contain about 82.7% xylan. Monomeric xylose
was obtained from Sigma (St. Louis, MO). Xylobiose, xylotriose, xylo-
tetraose, and xylopentaose were obtained from Magazyme International
Ireland (Bray, County Wicklow, Ireland).
Reaction System
Tubes made of made of 316 stainless steel with a O.5-in. OD,5-in.
length, and 0.035-in. wall thickness were obtained from Swagelok (Bangor,
ME) and used for the reaction system shown in Fig. 1. Xylan at a 5% solids
concentration was added to the tubes, which were then sealed with O.5-in.
diameter Swagelok caps. The tubes were dropped into boiling water for 1
min before being transferred to a 4kW sand bath (Cole-Parmer, Vernon
Hills, IL) controlled at a target temperature of 180 or 200°C. After reaction
for a specified time of between 3.5 and 30 min, the tubes were removed and
quenched in an ice bath. The liquid and solid contents were then trans-
ferred to microcentrifuge tubes and separated by centrifuging at 6022g for
10 min.
Measurement of Xylose Monomer and Oligomer Yields
To measure the yield of monomeric xylose, the liquid decanted from
the centrifuge tubes was passed through a 0.2-I-tm syringe filter before being
analyzed by a Waters (Milford, MA) high-performance liquid chromatog-
raphy (HPLC) separation module 2695, equipped with a Waters refractive
index (RI) detector model 2414 and a Bio-Rad (Hercules, CA) Aminex HPX-
87P column. The HPLC pump was set at a flow rate of 0.6 mLI min, and the
HPLC heater was operated at 85°C.
The yield of oligomers was determined by adding 72% H 2S04 to the
centrifuged liquid from autohydrolysis to achieve a final acid concentra-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Oligomer Molecular Weight Distribution 517

Fig. 1. Tubular reactor system used to hydrolyze xylan.

tion of 4%. These samples were then autoclaved for 1 hat 121°C in sealed
autoclave bottles. Next, the hydrolysate was neutralized with calcium
carbonate to a pH of 5.0-6.0 before being analyzed by HPLC. A set of
sugar standards of known concentration was taken through the same
analytical procedure in parallel to determine losses owing to the destruc-
tion of sugars and to provide a correction factor for adjusting the sugar
concentrations following posthydrolysis(4). The increase in monomer
concentration resulting from the posthydrolysis procedure provided a
measure of the fraction of the total dissolved xylose that is oligomeric(5,6).
The solid residues from the centrifuge tubes were vacuum dried and
subjected to quantitative saccharification to measure the remaining xylose
content (7).
Molecular Weight Determination for Oligomers
A Waters model 717 chromatography system, equipped with a Waters
RI detector 410 and a Bio-Rad Aminex HPX-42A ion-moderated partition
(IMP) column was connected to a Waters Fraction Collector II. Centrifuged
samples were filtered and analyzed by IMP at a flow rate of 0.2 mL/min and
a column temperature of 85°C. Concentrations were converted to yields on
a weight basis.

Results and Discussion


Table 1 and Fig. 2 show the sugar recovery for autohydrolysis of
xylan at 180 and 200°C for a 5% solids concentration. As expected, xylan
and xylooligomers were hydrolyzed faster at the higher temperature: the
maximum xylose monomer yield was about 11% at 200°C but <3% at
Applied Biochemistry and Biotechnology Vol. 105-108,2003
518 Li et a/.
Table 1
Sugar Recovery for Autohydrolysis of Xylan
at 180 and 200°C with 5% Solids Concentration
Time (min)
Yield (% of potential xylose) 0.00 5 10 20 30
Temperature = 180°C
Xylan in solid residue 100.00 43.96 36.96 26.47 16.53
Xylose in liquor 0.00 0.00 0.32 1.43 2.39
Xylooligomers in liquor 0.00 37.04 41.53 42.92 50.17
Total 100.00 81.00 78.81 70.82 69.09
Temperature = 200°C
Xylan in solid residue 100.00 35.08 12.55 7.67 11.12
Xylose in liquor 0.00 0.61 2.59 11.40 10.52
Xylooligomers in liquor 0.00 41.06 48.52 25.12 2.76
Total 100.00 76.75 63.66 44.19 24.40

100
-SOil d resl duel 18OC)
-o-SOII d resl due(200C)
~ 80 ••• A· •• Xylose I n t he II quor( 18OC)
.. ·6· .. XyI ose I n t he II quor (2OOC)
~
jij - -+- - Xyl 0011 goners I n t he II quor (18OC)
:c 60 - -0- - Xvi 0011 CIO/II!rs I n t he II auor (2000
c
i
,----0., __ ---
40
':---- .. :::::. . "'111:=------- . . . --------
'0
'#
-d' 20
Qi
>= .. .... .. -.... ....................................... .
0
0 5 10 15 20 25 30
Time, min

Fig. 2. Sugar recovery for uncatalyzed xylan hydrolysis at 180 and 200°C with 5%
solids concentration.

180°C over the 30-min run time. At 200°C, the yield of xylooligomers
reached a maximum of about 50% of potential xylose at 10 min and then
dropped. At 180°C, the xylooligomer yield reached about 50% of poten-
tial xylose at 30 min, the end of our run time, but appeared to be still
increasing. On the other hand, the amount of xylose in the solid residue
decreased with increasing reaction time, reaching a nearly constant value
after 10 min at 200°C but still continuing to drop after 30 min at 180°e.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Oligomer Molecular Weight Distribution 519

0.00 5.00 10.00 15.00 20.00 25.00

Fig. 3. IMP chromatogram of hydrolysate from uncatalyzed hydrolysis of oat xylan


30,00

-35.00 «too 40.00 50.00 G5.DD 80.00 85.00 70.0:

for a 5% solids concentration at 200°C for 10 min. I, DPl; 2, DP2; 3, DP3; 4, DP4.

''.
,<0.
,30.
'20.
110.

'00.

4
3
2

'O. ""j-,-.......,~,.......,,...,...-......,~-,-.-.-....,,.-.-.-..,r-.-.-..,...,-....,..,.-.-.,.....,,.......-.,....,.......,..,~,...........,,.....,.....,.,,..........,,.....,....,....,.....,,~,.....-.-.-I
0.00 5.00 10.(1) 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 MOO 80.00 85.00 70.00 75.00 10.00 85.00 .00 15.00 100.00

Fig. 4. IMP chromatogram of hydrolysate from uncatalyzed hydrolysis of oat xylan


for a 5% solids concentration at 200°C for 20 min. I, DPl; 2, DP2; 3, DP3; 4, DP4.

Figures 3 and 4 show IMP chromatograms of the liquid from


uncatalyzed hydrolysis of xylan with a 5% solids concentration run at
200°C for 10 and 20 min, respectively. As shown, IMP separated oligo-
mers up to DPIO, and less monomer and more long-chain oligomers were
observed at a 10-min than at 20 min. The average DP decreased as reaction
time increased.
Pure xylose, xylobiose, xylotriose, xylotetraose and xylopentaose
standards were used to calibrate the IMP system for species of DPI-DP5,
Applied Biochemistry and Biotechnology Vol. 105-108,2003
520 Li et al.
Table 2
Known Concentrations of Xylooligomers from DP1 to DP5
and Concentrations Calculated Based on Peak Height Relative
to Peak Height for Measured Concentration of Xylobiose
Actual Concentration Concentration based on peak
(mg/mL) height ratio (mg/mL)
Xylose 1.00 0.94
Xylobiose 1.02 1.02
Xylotriose 1.13 1.18
Xylotetraose 0.79 0.79
Xylopentaose 0.69 0.71

allowing us to calculate the concentration of DPI-DP5 in the liquors from


chromatograms at each sample time of the type shown in Figs. 3 and 4.
By testing a variety of samples of known composition, we found we could
obtain better accuracy if we based the determination of concentrations on
peak height rather than area. We also found that we could determine the
concentration of each of these species well by taking the ratio of each peak
height to the peak height for xylobiose and multiplying this ratio by the
measured concentration of the latter; this procedure proved more accu-
rate than using xylose monomer as a basis.
An example of this calculation follows. A known sample containing
1.02 mg/ mL of xylobiose gave a peak height signal of 33,252 m V, while a
known sample of 0.79 mg/ mL of xylotetraose gave a peak height signal of
25,894 m V. By a ratio of the peak heights, the xylotetraose concentration is
calculated to be (25,894/33,252) x 1.02 = 0.794 mg/mL, a value that is in
good agreement with the known amount in the standard.
Table 2 summarizes a test with known concentrations of DPI-DP5
and the calculated concentration based on the ratios to the height of the
xylobiose peak. This approach was extended to quantifying the concentra-
tions of DP6-DPIO based on chromatograms such as in Figs. 3 and 4 for each
sample time. However, we were unable to obtain standards to confirm the
concentrations of DP6-DPIO.
Based on the aforementioned approach, Fig. 5 shows the yield distribu-
tion of oligomer chain lengths from DPI-DPIO vs time for uncatalyzed xylan
hydrolysis at 200 0 e with a 5% solids concentration. No oligomers in the DPI-
DPIO range were observed until about 7 min into the run, and even then the
concentrations were very low. At the to-min mark, each of the concentrations
suddenly rose to about 4% of the total potential sugar, although the mono-
mer, DP2, DPto, and DP9 yields appeared to be somewhat lower than for the
other oligomers. The highest oligomer concentration appeared to be for DP5,
but the yields of a number of species were very similar.
A much greater separation in yields of the various chain lengths was
measured at the IS-min mark, as shown in Fig. 5. Furthermore, the yields
decreased in the exact order of the degree of polymerization with the

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Oligomer Molecular Weight Distribution 521
14 -1P1
12 ... c- .. 1P2
"iii
~ 10 --6--1P3
CIJ
8. ~
8
-"'-[p4
-"'-IPS
o~ 6 ....... 1P6
:>R
,,' 4
CI
--'>--[p7
Qj
2 -IPS
~
0 --0 ... + .. rpg

0 10 20 30 --O--1P10
Time, min

Fig. 5. Oligomer time profiles for uncatalyzed xylan hydrolysis at 200°C with a 5%
solids concentration.

- a Igone.s I n t he II quor. 2OOC( [1'2: !PIO)


···6-··algone.s In the Ilquor-2OOC(post hydrolysis)

....... - -6, •• ".

/'-

10
O--~-'c----'----'----'------'------""

o 5 10 15 20 25 30
Time, min

Fig. 6. Total oligomer yields determined by IMP and posthydrolysis.

highest yields being for monomers and the lowest for the DPI0 oligomers.
We also observed that those species from DP4-DPI0 dropped from the
peak seen at the previous time while xylose monomer yields continued to
increase until the 25-min time. DP2 and DP3 reached their maximum
levels at the 15-min sample time.
Because no significant yields of oligomers were observed via IMP
until about 10 min into the run, we wished to determine whether very little
solubilization occurred in these earlier times or whether the chains lengths
were too long for our IMP system to measure them. Accordingly, we mea-
sured overall oligomer yields at each time by the posthydrolysis procedure
and compared the results to the total oligomer yields measured by adding
the yields of DP2-DPlO, as shown in Fig. 6. Very good agreement was
found between the total oligomer yield measured by both approaches from
about 10 min on, suggesting that the IMP did a good job of capturing most
of the oligomers in solution at longer run times and that use of the peak
height ratios is reasonable. On the other hand, posthydrolysis revealed
levels of oligomers in solution atthe 3- and 7-min marks similar to at 10 min
while almost no oligomers were measured via IMP for these earlier times.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
522 Li et al.
These differences suggest that the chain length of the oligomers in solution
is too long for the IMP column used to measure.

Conclusion
The oligomer yield profiles measured by IMP follow the expectation
that longer chains depolymerize to form shorter chains that ultimately
result in release of soluble oligomers and monomer. However, it is some-
what surprising how rapidly oligomers and monomer from DPI0 to DPI
appeared and then mostly disappeared. Virtually no oligomers or mono-
mers were observed with IMP until several min into the reaction, and at
that time, the total yields measured by IMP and posthydrolysis of oligo-
mers were very similar, suggesting that the peak height method of deter-
mining oligomer concentrations is reasonably accurate. However, while
IMP provided no evidence of significant yields of oligomers at early times,
posthydrolysis indicated that significant amounts of soluble oligomers with
degrees of polymerization> 10 were in solution but not detectable by the
IMP system used. Accordingly, we plan to try other systems in an attempt
to find columns that can measure longer soluble oligomers.

Acknowledgment
We are very grateful to the National Science Foundation Division of
Bioengineering and Environmental System for supporting this research
through contract BES-9985351.

References
1. Garrote, G., Dominguez, H., and Parajo, J. C. (2001), Bioresour. Technol. 79, 155-164.
2. Garrrote, G., Dominguez, H., and Parajo, J. C. (1999), Holz Roh Werkst 57, 191-202.
3. Heitz, M., Capek-Menard, E., Koeberle, P. G., Gagne, J., and Chornet, E. (1991),
Bioresour. Technol. 35,23-32.
4. Ruiz, R. and Ehrman, T. (1996), NREL Chemical Analysis and Testing Standard
Procedure, No. 014, National Renewable Energy Laboratory, Golden, CO.
5. Garrrote, G., Dominguez, H., and Parajo, J. C. (2001), Process Biochem. 36,571-578.
6. Shevchenko, S. M., Chang, K., Robinson, J., and Saddler, J. N. (2000), Bioresour. Technol.
72(3),207-211.
7. Ruiz, R. and Ehrman, T. (1996), NREL Chemical Analysis and Testing Standard
Procedure, No. 002, National Renewable Energy Laboratory, Golden, CO.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0523/$20.00

Conversion of Sugarcane Bagasse


to Carboxylic Acids Using a Mixed Culture
of Mesophilic Microorganisms

PIYARAT THANAKOSES, NAGAT ABD ALLA MOSTAFA,


AND MARK T. HOLTZAPPLE*
Department of Chemical Engineering,
3122 TAMU, Texas A&M University,
College Station, TX 77843-3122,
E-mail: m-holtzapple@tamu.edu

Abstract
Using the MixAlco process, biomass can be converted into carboxylic
acids, which can be chemically converted into mixed alcohol fuels. This
study focused on the use of countercurrent fermentation to anaerobically
convert sugarcane bagasse and chicken manure to mixed carboxylic acids
using a mixed culture of mesophilic microorganisms from terrestrial and
marine sources. Bagasse was pretreated with lime to increase digestibility.
The continuum particle distribution model (CPDM) simulated continuous
fermentors based on data collected from batch experiments. This model
saves considerable time in determining optimum operating conditions. For
an 80% bagasse/20% chicken manure fermentation with terrestrial inocu-
lum at a volatile solids loading rate (VSLR) of 7.36 g/(L of liquid·d) and a
liquid residence time (LRT) of 8.88 d, total carboxylic acid productivity, total
acid selectivity, and yield were 2.49 g/(L of liquid·d), 0.581 g of total acid/
g of VS digested, and 0.338 g of total acid/ g of VS fed, respectively, at a
concentration of 18.7 g of total acid/L. At the same VSLR and LRT, fermen-
tation with marine inoculum gave higher total acid productivity, total acid
selectivity, and yield than fermentation with terrestrial inoculum. For an
80% bagasse/20% chicken manure fermentation with marine inoculum at a
VSLR of 3.83 g/(L of liquid·d) and an LRT of 12.1 d, total carboxylic acid
productivity, total acid selectivity, and yield were 1.38 g/(L of liquid-d),
0.667 g of total acid/ g of VS digested, and 0.359 g of total acid/ g of VS fed,
respectively, at a concentration of 16.2 g of total acid/L.
Index Entries: Bagasse; carboxylic acids; lime pretreatment; sugarcane;
fermentation.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 523 Vol. 105-108, 2003


524 Thanakoses et at.
Lime
pretreatment

Mixed a1cohols~-fC1

Fig. 1. MixAlco process.

Introduction
In 1995, more than 250,000,000 t of agricultural waste including
bagasse was generated in the United States (1). The conversion of waste
biomass into liquid fuels has attractive environmental benefits, such as
reducing the greenhouse effect and increasing resource availability. The
most common method to convert lignocellulosic biomass to useful prod-
ucts is simultaneous saccharification and fermentation, which enzymati-
cally hydrolyzes lignocellulose to sugars that are fermented to alcohol.
Currently, the cost of cellulase is too expensive. Moreover, this process
requires sterile operation. To solve these problems, Holtzapple et al. (2)
developed the MixAlco process (Fig. 1).
The MixAlco process converts any biodegradable material into mixed
alcohol fuels. The biomass is treated with lime to increase digestibility.
Then it is fermented with a mixed culture of acid-forming microorganisms
that degrade the major substrates (e.g., polysaccharides, proteins, and lip-
ids) to carboxylic acids under nonsterile, anaerobic conditions. To maintain
the pH, calcium carbonate is added, which reacts with the acid to form
carboxylate salts. The carboxylate salts can be converted to ketones, which
may be hydrogenated to alcohol fuels that are clean burning and do not add
net carbon dioxide to the environment (3,4).
Both high conversion and high product concentrations are possible by
using a countercurrent fermentation in which solid and liquid pass in op-
posite directions through a series of fermentors (Fig. 2). The countercurrent
flow arrangement is beneficial because the inhibitory effects of high prod-
uct concentration (in F1 in Fig. 2) is partially offset by the fresh, highly
reactive substrate, and the lower reaction rate of highly digested biomass
(in F4 in Fig. 2) is partially offset by a lower product concentration (3,4).
In mixed-culture acid fermentations, methane production can be
inhibited by methane analogs such as iodoform and bromoform (5-7) or the
coenzyme M analog, 2-bromoethanesulfonic acid (6,8,9). By inhibiting
methane, a potential hydrogen sink is eliminated and the reducing power
is used to produce higher carboxylic acids, such as butyric and caproic
acids (4-6).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Carboxylic Acids from Sugarcane Bagasse 525

Product Fresh
liquid liquid

Undigested
bioma s

Fig. 2. Four-stage countercurrent fermentation.

Biomass feed consists of volatile solids (VS) and ash. Fig. 3 shows that
the biomass VS are converted into gaseous and liquid products, with some
remaining solid residues. In the liquid products, VS consist of carboxylic
acids and other components, extracellular proteins, and energy storage
polysaccharides (4). The following definitions are employed throughout
this article:
. ld _ total acids produced
Yle - VS field (1)

. VS digested
converSIOn = VS fed (2)
total acids produced
total acid selectivity = ---~--­ (3)
VS digested
·d d .. _ total acids produced
tota1 aCl pro UCtiVlty - totall··d
lqUl vo1ume m . a11 fermentors . time
. (4)

VSLR = VS fed to the fermentor train


total liquid volume in all fermentors . time (5)

LRT = total liquid volume in all fermentors


flow rate of liquid out of fermentation train (6)

mass balance closure = mass out (7)


mass in + water of hydrolysis
To save time determining optimum operating conditions, Loescher
(10) developed the continuum particle distribution model (CPDM), which
can simulate continuous fermentor systems based on the data collected
from batch experiments.
Bagasse is low in nutrients, whereas animal manure contains larger
amounts of nutrients such as nitrogen, vitamins, and minerals (11). Conse-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
526 Thanakoses et al.
Ca.
CO2 Gas
Volatile Total Acids
solids
Dissolved Liquid
(VS)
digestion VS
Undigested
VS Undigested
solids
ash ash

Fig. 3. Digestion of biomass.

quently, our research combined two substrates: sugarcane bagasse (a car-


bohydrate-rich biomass) and chicken manure (a nutrient-rich biomass).
From previous studies, Playne (12) studied carboxylic acid production from
bagasse by nonsterile, mixed-culture fermentation in continuous fermen-
tation. Total acid productivity was 4.9 g/ (L of liquid·d) and yield was 0.25
g of total acid/ g of bagasse fed at a concentration of 15.5 g of total acid/L.
To improve total acid productivity and yield, we studied the use of
SO% bagasse/20% chicken manure countercurrent fermentation with ter-
restrial and marine microorganisms. The effect of volatile solids loading
rate (VSLR) and liquid residence time (LRT) on total acid productivity,
yield, conversion, and total acid selectivity were investigated. In addition,
CPDM was used to predict total acid concentration and conversion for
countercurrent concentration based on batch fermentation.

Materials and Methods


Substrates
Sugarcane bagasse (SO%) and chicken manure (20%) were used as
substrates. Bagasse was obtained from a sugar facility in Raceland, LA. It
was stored in a shed and allowed to dry. Using conditions recommended
by Chang et al. (13), the bagasse was pretreated with 0.1 g of Ca(OH)2/ g of
dry biomass at 65 C for 24 h and then hammer milled. The average mois-
D

ture content of treated bagasse was 0.070 g of water / g of wet bagasse, and
the average ash content was 0.140 g of ash/ g of dry bagasse. The treated
bagasse consisted of 0.S60 g of VS/ g of dry bagasse (0.651 g of carbohy-
drate/ g of dry bagasse, 0.029 g of crude protein/ g of dry bagasse, and 0.lS0
g of lignin/ g of dry bagasse).
Chicken manure was collected from the Poultry Science Center, Texas
A&M University, College Station, TX. The manure was air-dried and then
pretreated with 0.1 g of Ca(OH)z/ g of dry biomass at lOODC for 1 h. The
treated chicken manure consisted of 0.052 g of water / g of wet chicken
manure and 0.340 g of ash/ g of dry chicken manure. The average moisture

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Carboxylic Acids from Sugarcane Bagasse 527
content of treated chicken manure was 0.052 g of water I g of wet chicken
manure, and the average ash content was 0.340 g of ashl g of dry chicken
manure. The treated chicken manure consisted of 0.660 g of VS/g of dry
chicken manure (0.482 g of carbohydrate I g of dry chicken manure, 0.178 g
of crude protein I g of dry chicken manure, and no lignin). The chicken
manure was collected in three batches throughout the research; unfortu-
nately, the quality varied from batch to batch.

Medium
The liquid medium used in the bagasse I chicken manure system was
the modified Caldwell & Bryant (C&B) medium deoxygenated water,
which consists of a dry nutrient mixture in deoxygenated water at a concen-
tration of 1.4 g of dry nutrient mixturelL of deoxygenated water (11);
Deoxygenated water was prepared by boiling distilled water under a
nitrogen purge for 5 min. The medium was allowed to cool to room tem-
perature, and then 0.275 giL of cysteine hydrochloride and 0.275 giL of
sodium sulfide was added under a nitrogen purge. The medium solution
was stirred and poured into storage bottles with a nitrogen purge.
Dry nutrient mixture contained (g/100 g of mixture) K2HP04 (16.3),
KH2P04(16.3),(NH4)2S04(16.3),NaCl(32.6),MgS04·7H20(6.8),CaC12·2H20
(4.4), HEPES (0.86), hemin (0.71), nicotinamide (0.71), p-aminobenzoic acid
(0.71), Ca-pantothenate (0.71), folic acid (0.35), pyridoxal (0.35), riboflavin
(0.35), thiamine (0.34), cyanocobalamin (0.14), biotin (0.14), EDTA (0.35),
FeS04·7H20 (0.14), MnCl2(0.14), H 3B03 (0.021), CoCl2(0.014), ZnS04·7H20
(0.007), NaMo04 (0.0021), NiCl2 (0.0014), and CuCl2 (0.0007) (4).

Inoculum
Two primary sources of inoculum were used: terrestrial and marine.
The terrestrial source included the rumen fluid from a forage-fed fistula ted
steer located at the Meat Center, Department of Animal Science, Texas
A&M University, College Station, TX. In addition to rumen fluid, compost
from Dr. Mark Holtzapple's residence and swamp material from Wolf Pen
Creek, College Station, were used. They were collected in bottles filled with
deoxygenated medium, which consisted of distilled water, 0.275 giL of
sodium sulfide, and 0.275 giL of cysteine hydrochloride.
Another source was marine microorganisms, a salt-resistant inocu-
lum, from sediments at East Beach, Harborside Street, 51st Street, and
Sportsman Road on Galveston Island, TX. The sediment was collected from
O.5-m-deep holes and placed in bottles filled with deoxygenated medium,
which consisted of distilled water, 0.275 giL of sodium sulfide, and 0.275
giL of cysteine hydrochloride.
Inhibitors
The inhibitor in this experiment was iodoform (CHI3) prepared in a
solution containing 20 g of inhibitor IL of ethanol. The inhibitor was kept

Applied Biochemistry and Biotechnology Vol. 705-708,2003


528 Thanakoses et at.
Metal bars

Fig. 4. Assembled fermentor.

in a tinted bottle. The cap was replaced immediately after use owing to light
and air sensitivity.
pH Control
To control the pH between 6.0 and 6.5, excess CaC03 was added to the
fermentors. In addition, as a nitrogen supplement, urea was added if the
pH was <6.5.
Fermentor
Figure 4 shows the assembled fermentor used in this research. The
fermentor was a 1-L polypropylene copolymer centrifuge bottle (98 x 169
mm),Nalgene brand NNI 3120-1010, 7100g maximum force. The centrifuge
bottle was placed on a Wheaton Modular Cell Production Roller Apparatus
(Model III). The apparatus, which was located in an incubator, consists of
rollers that rotated the fermentors horizontally at 1 rpm. The bottles were
capped with a size 11 rubber stopper with a glass tube inserted through it.
The glass tube was capped with a rubber septum for gas sampling and
release. The septum was replaced when there was a visible hole resulting
from frequent gas venting. Two pieces of 0.25-in. stainless steel tubing,
with ends welded shut, were inserted into holes in the stopper. The tubes
formed a stir bar to assist in mixing the components inside the bottle; thus,
this fermentor could use a high solids concentration. The fermentor could
not stand a pressure >2 atm (15 psig); above this pressure, it would leak or
explode if the gases were not released.
Experimental Procedure
Continuous countercurrent fermentations were performed in anaero-
bic fermentors at 40°C. The transfer process occurred every 1 or 2 d,
depending on the system. To maintain anaerobic conditions, a nitrogen
purge was used during transferring, or whenever the fermentors were open.
Two grams of calcium carbonate were added every 1 or 2 d to each fermen-
tor to neutralize the carboxylic acids.
Countercurrent fermentations were conducted at varying LRT and
VSLR, as described in Tables 1 and 2. The operating parameters for each
Applied Biochemistry and Biotechnology Vol. 105-708,2003
).
:g
~
0..

r")
'"o·
::l-
rt>
3
<n' Table 1
'<:
II>
'" Operating Parameters for Bagasse/Chicken Manure Countercurrent Fermentation with Terrestrial Inoculum
::,
0..

o· Fermentation train A B C D E F G H I J K L M
'"
<ii
r")
::l- LRT (d) 11.7 11.4 13.5 9.70 10.8 13.1 8.87 8.88 12.1 20.5 19.0 20.0 19.7
::,
Cl
VSLR
~
'<: (g VS/[L of liquid in all fermentors·d]) 10.1 10.3 17.9 11.2 14.4 10.1 10.9 7.36 3.81 2.13 2.00 4.82 6.54
VS feed to F1 at each transfer (g VS) 9.3 7.3 19.6 12.2 15.7 7.3 11.7 7.40 3.90 3.90 3.90 12.4 16.5
VI Liquid feed to F4 at each transfer (L) 0.1 0.08 0.15 0.15 0.15 0.08 0.15 0.15 0.10 0.10 0.15 0.15 0.15
N
<.0 Frequency of transfer" lid lid lid lid l/d lid lid l/d lid E2d E2d E2d E2d
Liquid volume in all fermentors (L) 0.920 0.708 1.10 1.10 1.04 0.728 1.08 1.00 1.02 0.916 0.976 1.28 1.26
Iodoform addition rate
(mg iodoform added to each
fermentor/L of liquid fed to F4) 16.0 20.0 10.7 5.3 10.7 20.0 10.7 10.7 24.0 24.0 16.0 10.7 10.7
Urea addition rate
(g urea added to each fermentor/L
of liquid fed to F4) (if pH < 6.5) 1.00 1.25 0.67 0.67 0.67 1.25 0.67 0.67 1.0 1.0 0.67 0.67 0.67
a1 I d, once a day; E 2 d, every 2 days.
6=
,....
a
I
a
-'"
,Co
I\.)
a
a
w
)..
:g
[
OJ
0'
3-
ro
3
;:n'
~ Table 2
OJ
;:,
0... Operating Parameters for Bagasse/Chicken Manure Countercurrent Fermentation with Marine and Terrestrial Inoculum
OJ
0'
fii Fermentation train N P Q R I J
3-
;:,
o LRT (d) 12.1 12.1 20.5 20.5 12.1 20.5
~
'<:: VSLR (g VS/[L of liquid in all fermentors·dD 3.83 3.84 2.13 2.13 3.81 2.13
Marine inoculum (wt%) 100 40 100 40 a a
Ul Terrestrial inoculum (wt%) a 60 a 60 100 100
W
o VS feed to F1 at each transfer (g VS) 3.9 3.9 3.9 3.9 3.9 3.9
Liquid feed to F4 at each transfer (L) 0.1 0.1 0.1 0.1 0.1 0.1
Frequency of transfer Daily Daily Every 2 d Every 2 d Daily Every 2 d
Liquid volume in all fermentors (L) 1.02 1.02 0.916 0.916 1.02 0.916
Iodoform addition rate (mg iodoform added 24.0 24.0 24.0 24.0 24.0 24.0
to each fermentor/L of liquid fed to F4)
Urea addition rate (g urea added to each 1.0 1.0 1.0 1.0 1.0 1.0
fermentor /L of liquid fed to F4) (if pH < 6.5)
c;::
;--

~
Cl
,C>:>
~
Cl
8
Carboxylic Acids from Sugarcane Bagasse 531

+. . . . . . . . . .
liquid
~
..
liquid
..
.. _... ............ ... .
liquid
..........................
liquid
.......................
: :

Collection Fre h
bottle medium
F3

Fre h .............. . : ....................... L...................; ~ .................... J L............ ollection


bioma s solid olid olid olid bottle

Fig. 5. Single-centrifuge procedure.

fermentation train with 100% terrestrial inoculum are given in Table 1.


The operating parameters for each fermentation train with 40% and 100%
marine inoculum, compared with 100% terrestrial inoculum, are given in
Table 2. All fermentation trains consisted of four countercurrent stages
and the single-centrifuge procedure (Fig. 5).
After the system reached steady state, fermentation data were col-
lected for 40 d to determine carboxylic acid concentration, acid productiv-
ity, selectivity, yield, conversion, and CH4 productivity.
Analytical Methods
The total gas volume from each fermentor was measured using an
inverted, graduated glass cylinder apparatus (water displacement appara-
tus) filled with an aqueous solution of 30 wt% CaCl2 (14). The methane
content was determined using a gas chromatograph (Agilent Model 6890)
with a thermal conductivity detector and packed column (CarboxenlO04,
Supelcol-2390; J&W Scientific). Samples were taken directly from the fer-
mentors using a 5-mL syringe. To ensure calibration of the column and
instrument, a standard consisting of 10.06 mol% methane, 29.99 mol% car-
bon dioxide, and the balance nitrogen was periodically run through the
column before running the gas samples. Methane and carbon dioxide are
both products of microbial digestion. Knowledge of the amount of total gas
and methane allowed CO2 production to be calculated. CO2 production is
of two kinds: biotic and abiotic. The abiotic CO2 is produced by neutralizing
the carboxylic acids with calcium carbonate. It is assumed that 1 mol of
abiotic CO2 is produced for every 2 mol of acid produced. The biotic CO2
produced directly from the fermentation can be calculated from the total
CO2 subtracted from the abiotic CO2,
Liquid from the first reactor and solid from the last reactor were
collected during each transfer. Liquid samples were analyzed for car-
boxylic acids and VS. Solid samples were also collected to determine the
mass of undigested VS. Acid analysis was performed using an Agilent
6890 gas chromatograph with capillary column O&W Scientific, model
DB-FFAP). Liquid samples were combined with 1.162 giL of internal
standard (4-methyl-n-valeric acid) and acidified with 3 M phosphoric
acid before injection into the gas chromatograph. VS in the fed substrates

Applied Biochemistry and Biotechnology Vol. 105-108,2003


532 Thanakoses et al.
and from liquid and solid samples were determined by drying at 105°C
and then ashing at 550°C.
A complete mass balance for the fermentation systems was obtained
on the entire train over a steady-state period. The mass balance closure
determined the difference between the mass entering the system and the
mass exiting the system.
The mass balance closure was calculated as follows:
mass balance closure =

undigested VS + non-acid dissolved VS + acids out + biotic CO 2 + CH 4 (8)


VS in + water of hydrolysis
The system should theoretically have 100% closure; any discrepancies
in the closure result from errors in the measurements and transfer process.
Ross (4) suggested that the biomass could be represented as cellulose, which
has a monomer weight of 162 g/ moL When cellulose is hydrolyzed to sugars,
it gains a molecule of water (mol wt of 18 g/ mol) per monomer that accounts
for the increase in mass. An approximation for the water of hydrolysis is

water of hydrolysis = VS digested x 1~2g:/:~l (9)

Continuum Particle Distribution Modeling


Ross (4) describes a continuum particle as a collection of particles that
has a VS wt of 1 g upon entering a fermentor. To apply the CPDM method,
batch experiments were used at a variety of initial substrate concentrations.
The batch experiments consisted of five fermentors at varying initial
substrate concentrations (20, 40, 70, 100, and 100+ g of dry substrate/L of
liquid). The 100 and 100+ fermentors had the same initial substrate concen-
trations, but the 100+ fermentor contained a medium with a mixture of
carboxylate salts (70 wt% calcium acetate, 20 wt% calcium propionate, and
10 wt% calcium butyrate) at a concentration of 20 g of carboxylic acids/L
of liquid. The inoculum for the CPDM batch experiment was 20% of liquid
that was taken from countercurrent fermentation operating with the same
substrate. The medium for the CPDM batch experiments was modified
C&B medium, the same as the countercurrent fermentation. Initially, 2.0 g
of calcium carbonate and 0.1 g of urea were added to each fermentor. Daily
samples of the liquid were taken from each batch fermentor. From these
batch experiments, the reaction rate at varying acid concentrations and
biomass digestion were determined.
After analyzing the liquid, the carboxylic acid concentration was con-
verted to acetic acid equivalents (Ae) by the following equations:
a (mol/L) = acetic (mol/L) + 1.75 x propionic (mol/L) +
2.5 x butyrie (mol/L) + 3.25 x valerie (mol/L) + (10)
4.0 x caproic (mol/L + 4.75 x heptanoic (mol/L)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Carboxylic Acids from Sugarcane Bagasse 533
This can be expressed on a mass basis as follows:
Ae (giL) = 60.05 (g/mol) x a (mol/L) (11)
The acetic acid equivalent unit is based on the reducing power of the
acids produced in the fermentation. From the batch experiments, the data
were fit to an empirical equation by least-squares analysis. The concentra-
tion of acetic acid equivalents in each batch experiment was fit to Eq. 12:
A =a+-.l!L (12)
e 1 + ct
in which a, b, and c are constants fit by least-squares analysis. The equation
derived from the curve fit was then differentiated to give the rate of acid
production", by Eq. 13:
r = dAe = b
(13)
dt (1 + ct)2

This rate in Eq. 13 .was then converted into specific rate, (g of Ae r


produced/ g of VS·d), by dividing it by the initial substrate concentration,
So (g of VS/L), in the respective batch fermentation.
(14)

The specific rate equation in the mixed-acid fermentation is empirical.


Equation 15 can be determined by using least-squares analysis:

" e(1-x) f
r pred = h (15)
1 + g{<I>Ae)

in which x is the fraction conversion of volatile solids; e, f, g, and hare


empirical constants; and <I> is the ratio of total grams of actual acid to grams
ofAe •
Ross (4) included <I> to decrease the inhibitory effects of the higher acids
that would overestimate the specific rate. The (1- xy term in the numerator
is the conversion penalty function described by South and Lynd (15). The
conversion, x, is calculated using

(16)

in which a is the selectivity (g of Ae produced/ g of VS digested), and So is


the initial substrate concentration (g of VS/L).
In the CPDM model, the selectivity a was taken as a constant. The
selectivity a for CPDM was calculated from the selectivity s from the coun-
tercurrent experiment (g of total acid produced/ g ofVS digested) by divid-
ing by <I> (the ratio of total grams of actual acid to grams of Ae). The
relationship between s and a follows:
5=<1>0 (17)

Applied Biochemistry and Biotechnology Vol. 105-708,2003


534 Thanakoses et al.
Equation 15 was used in a Mathematica program (14) to predict acetic
acid equivalent concentration (Ae), which could be converted back to total
acid concentration, and conversion (x) for the countercurrent fermentations
at various LRT and VSLR. Other system-specific parameters needed for the
Mathematica program were the average selectivity (0), holdup (ratio of g of
liquid to g of VS wet solid), moisture (ratio of g of liquid to g of wet solid in
feed), solids concentration (g VS/L of liquid), and liquid volume (L).

Results and Discussion


Countercurrent Fermentations
Table 3 presents the results of the countercurrent fermentations with
terrestrial inoculum. The highest acid productivity, yield, and acid selectiv-
ity were 2.49 g/ (L of liquid ·d), 0.338 g of total acid/ g of VS fed, and 0.581 g
of total acid/ g of VS digested, respectively, at a concentration of 18.7 g of
total acid/L in fermentation train H (LRT =8.88 d and VSLR =7.36 gilL of
liquid·d]). Fermentation train J (LRT = 20.5 d and VSLR = 2.13 gilL of
liquid·d]) had the highest conversion (0.60 g of VS digested/ g of VS fed). Its
yield oW.338 g of total acid/ g ofVS fed (0.336 g of total acid/ g of dry bagasse
fed) at a concentration of 18.7 g of total acid/L was higher than the yield
from Playne (12) (0.25 g of total acid/ g of bagasse fed at a concentration of
15.5 g of total acid/L). The difference could result from the benefits of coun-
tercurrent fermentation and addition of chicken manure in this research.
The correlations between VSLR and acid productivity (p), selectivity
(s), yield (Y), and conversion (x) are shown in Figs. 6-9, respectively. Acid
productivity increased with increasing VSLR. Moreover, the selectivity,
yield, and conversion decreased with increasing VSLR. These correlations
can be described by the following equations:
p = 0.0497 VSLR + 1.06 (18)

s = -0.0077 VSLR + 0.483 (19)

y = -0.0088 VSLR + 0.253 (20)

x = -0.0142 VSLR + 0.578 (21)


At higher solid loading rates, the microorganisms have a higher acid
productivity; however, the conversion is lower. Moreover, the mixed cul-
tures of microorganisms digest primarily the easy" portions and produce
/I

non-acid soluble VS that presumably are energy-storage products (11).


Consequently, the ratio of digested to undigested VS (conversion) was low.
With the higher VS loadings, the selectivity (g acid/ g of VS digested) and
yield (g acid/g of VS fed) were lower. On the other hand, at smaller VS
loading rates, the acid productivity was lower. Furthermore, the microor-
ganisms must consume both the easy and difficult portions and produce
less of the soluble energy-storage products. With the lower VS loading, the

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


).
:g Table 3
Results for Bagasse/Chicken Manure Countercurrent Fermentation with Terrestrial Inoculum"
[
0;)
o· Fermentation train A B C D E F G
1"\
:J-
t1)
:3 Average pH in all fermentors 6.2 ± 0.41 6.4 ± 0.19 6.4 ± 0.31 6.1 ± 0.38 6.3 ± 0.25 6.4 ± 0.20 6.1 ± 0.21
iii·
~
Total acid productivity 1.22 1.41 1.55 1.56 1.84 1.51 2.12
til
:J (g/[L of liquid in all fermentors·dD
Q..
0;) Total acid concentration (giL) 14.2 ± 0.76 16.6 ± 1.06 21.0 ± 0.48 15.2± 0.73 20.1 ± 0.80 20.0 ± 0.36 19.0 ± 0.87

fb
1"\
Acetic acid (wt%) 35.7 ± 5.10 35.4 ± 3.21 39.6 ± 2.48 35.6± 4.09 39.3 ± 3.15 40.1 ± 1.68 40.2 ± 3.06
:J-
:J Propionic acid (wt%) 26.9 ± 3.25 20.2 ± 2.74 18.1 ± 1.82 26.5 ± 2.91 17.2 ± 3.44 18.3 ± 1.68 18.7 ± 1.71
0
~ Butyric acid (wt%) 17.0 ± 2.91 19.0 ± 2.96 19.3 ± 2.28 15.9 ± 2.16 20.1 ± 3.22 18.6 ± 1.34 17.5 ± 2.28
'<
Valeric acid (wt%) 13.6 ± 3.06 13.2 ± 2.31 10.2 ± 2.14 14.0 ± 1.68 11.7 ± 2.66 10.3 ± 1.05 11.6 ± 1.18
Caproic acid (wt%) 5.54 ± 1.84 9.61 ± 2.41 9.82 ± 1.82 6.76 ± 0.77 9.29 ± 3.10 9.86 ± 1.91 8.24 ± 1.06
V1
w Heptanoic acid (wt%) 1.26 ± 0.29 2.63 ± 0.99 2.87 ± 0.91 1.18 ± 0.36 2.36 ± 1.38 2.86 ± 0.74 3.76 ± 1.81
V1
VS digested (g VS I d) 4.00 3.30 8.10 4.70 4.30 3.20 4.50
Yield (g total acid I g VS fed) 0.118 0.137 0.087 0.139 0.127 0.151 0.197
Selectivity 0.275 0.303 0.210 0.362 0.465 0.344 0.511
(g total acid I g VS digested)
Non-acid dissolved VS 0.078 0.100 0.091 0.074 0.216 0.100 0.180
(g VS/g VS digested)
Conversion (g VS digested I g VS fed) 0.430 0.452 0.413 0.385 0.274 0.438 0.385
Biotic CO2 productivity 1.38 NIA 0.567 1.27 0.645 NIA 0.765
~
:- (g C02 /[L of liquid in all fermentors·dD
c
-1" CH4 productivity 0 NIA 0 0.0028 0 NIA 0
(g CH4 /[L of liquid in all fermentors·dD
c
-
,?l Mass balance closure (g VS outl g VS in) 0.819 NIA 0.707 0.860 0.929 NIA 0.913
N
C
c a
All errors are ± 1 SO. N I A, not available. (Continued)
'"
).
:g Table 3 (Continued)
~ Results for Bagasse/Chicken Manure Countercurrent Fermentation with Terrestrial Inoculuma
t:l..
txl

I')
Fermentation train H I J K L M
::J-
(1)
3
;;;. Average pH in all fermentors 6.3 ± 0.19 6.4 ± 0.29 6.3 ± 0.31 6.4 ± 0.21 6.2 ± 0.29 6.2 ± 0.28
q-
'<: Total acid productivity 2.49 0.791 0.547 0.615 0.623 0.634
II>
:;,
t:l.. (g/[L of liquid in all fermentors·d])
txl
0· Total acid concentration (g/L) 18.7 ± 0.49 9.6 ± 1.05 13.2 ± 0.64 11.1 ± 0.79 13.1 ± 0.88 15.2 ± 0.68
rb
I')
::J-
Acetic acid (wt%) 40.4 ± 2.63 35.4 ± 4.10 35.1 ± 4.95 44.5 ± 3.99 38.5 ± 3.88 35.8 ± 3.66
:;,
0 Propionic acid (wt%) 19.9 ± 1.91 20.2 ± 2.50 25.0 ± 3.31 24.3 ± 5.08 22.5 ± 3.14 21.8 ± 2.41
~
'<: Butyric acid (wt%) 17.1 ± 2.02 10.7 ± 2.06 16.5 ± 3.01 9.6 ± 3.43 17.8 ± 3.58 18.8 ± 2.15
Valeric acid (wt%) 11.5 ± 1.52 14.1 ± 2.51 13.1 ± 2.10 12.2 ± 2.94 11.9 ± 2.68 12.7 ± 2.27
{JJ Caproic acid (wt%) 8.27 ± 1.09 8.38 ± 1.35 5.60 ± 0.64 5.10 ± 1.35 6.83 ± 1.71 8.17 ± 1.77
w Heptanoic acid (wt%) 2.84 ± 0.65 11.2 ± 2.66 4.66 ± 1.16 4.34 ± 1.16 2.45 ± 0.92 2.72 ± 1.05
(j'I

VS digested (g VS/d) 4.30 2.10 1.20 1.06 3.10 3.50


Yield (g total acid/ g VS fed) 0.338 0.208 0.250 0.300 0.125 0.094
Selectivity (g total acid/ g VS digested) 0.581 0.386 0.417 0.566 0.258 0.229
Non-acid dissolved VS 0.133 0.157 0.225 0.226 0.181 0.160
(g VS/g VS digested)
Conversion (g VS digested/ g VS fed) 0.581 0.538 0.600 0.530 0.484 0.412
Biotic CO2 productivity 0.809 0.628 0.601 0.435 0.466 0.515
(g CO2 / [L of liquid in all fermentors·d])
a:
:-
CH4 productivity 0 0.0001 0.0001 0.0001 0.0004 0.0008
-~ (g CH4 /[L of liquid in all fermentors·d])
c
- Mass balance closure (g VS out/ g VS in) 0.886 0.863 1.01 1.02 0.786 0.791
~
N aAll errors are ± 1 SD. N/ A, not available.
c
8
Carboxylic Acids from Sugarcane Bagasse 537
• Experimental data
3.0
- Best line fit

2.5 •
0
'>·E~ 2.0 •
="9
'8~
'"Ag= 1.5
'"C)
'~d
S~ 1.0 •
~ •
0.5

0.0
0 5 10 15 20
VSLR (g VS/(L liquid' d»

Fig. 6. Correlation of total acid productivity with VSLR (R2= 0.271).

0.7 • Experimental data


""'
'"C)
2:l 0.6 • - Best line fit
'" •

a)
0.5
...
OJ)

.ecn;a •
-
0.4
.~:>
UOJ)
• • •
..!:l:O
a) .....
0.3
• • ••

(/J~
0.2
S
B 0.1
~
0
0 5 10 15 20
VSLR (g VS/(L liquid' d»

Fig. 7. Correlation of selectivity with VSLR (R2 = 0.0677).

0.4 • Experimental data


"0'
<2:l • - Best line fit
(/J
:>
0.3 •
:a
'0
0.2

S
'"
B
OJ)
'-' 0.1
:s
~
0
0 5 10 15 20
VSLR (g VS/(L liquid·d»

Fig. 8. Correlation of yield with VSLR (RZ = 0.275).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


538 Thanakoses et al.
0.7
• Experimental data
0.6 • • - Best line fit
0.5
§
'1i.! 0.4
~
i= 0.3
8 •
0.2

0.1

0
0 5 ill ~ W
VSLR (g VS/(L liquid'd))

Fig. 9. Correlation of conversion with VSLR (R2 = 0.540).

30
.-..
~
'-'
25
§
'l:) 20
!;l

~
u
15

:9 ill 100% terrestrial


~
'3 5
~
0
20 30 40 50 60 70 80 90
Time (days)

Fig. 10. Total acid concentration for baggasse/ chicken manure fermentation with
different inoculum sources at LRT =12.1 d and VSLR =3.8 g/(L·d).

selectivity (g of acid/ g of VS digested) and yield (g of acid/ g of VS fed)


were higher. Moreover, the ratio of digested to undigested VS (conver-
sion) was higher.
Figures 10 and 11 show the steady-state total acid concentration for
the fermentations with 100% marine, 40% marine + 60% terrestrial, and
100% terrestrial inocula. The total acid concentration for the fermentation
with 100% marine inoculum was higher than the fermentations with 40%
marine + 60% terrestrial and 100% terrestrial inocula at VSLR = 2.13 and
3.8 g/ (L·d). Other results are given in Table 4. The acid productivity,
selectivity, and yield in the fermentations with marine inoculum (trains
N, P, Q, and R) were higher than fermentation trains I and J with solely
terrestrial inoculum. Fermentation trains Nand Q with 100% marine
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Carboxylic Acids from Sugarcane Bagasse 539
30

~ 25
= ~ 100 % marine
.g 20 l------------.~O<=..
~ .....~:-:::*:,.,..-~~~Ik:-'~'>E!II;l~1r
~§ 15 •
-V~!!inH
Mg ........... 11'""'-... A
......, .......... --
-....r,.- - 100% terrestrial 60% terrestrial

~ 10
'5 5
F:
O+---.---.---.---,---.---.--~
20 30 40 50 60 70 80 90

Time (days)

Fig. 11. Total acid concentration for baggasse/chicken manure fermentation with
different inoculum sources at LRT = 20.5 d and VSLR =2.13 g/(L·d).

inoculum had slightly higher acid productivity, selectivity, and yield than
fermentation trains P and R with a 40% marine + 60% terrestrial inoculum.
Conversion in fermentation trains I, N, and P were similar at LRT =12.1 d
and VSLR = 3.8 g/(L·d). However, conversion increased with increasing
percentage of marine inoculum at LRT = 20.5 d and VSLR = 2.13 gl (L·d).
The carboxylic acid produced in the fermentation is combined with cal-
cium carbonate and transformed to carboxylate salt. At high salt con-
centrations, the terrestrial microorganisms could not grow well compared
to the marine inoculum. Therefore, the fermentation with 100% marine
inoculum resulted in the highest acid productivity, yield, and acid selec-
tivity in this experiment.
Continuum Particle Distribution Modeling
The data from the batch fermentation at varying initial substrate con-
centrations can be found in Thanakoses (14). The values of e, f, g, and h for
Eq. 15 and selectivity used in the CPDM Mathematica program are given
in Table 5. Other parameters used in the CPDM Mathematica program are
presented in Table 6. From the Mathematica program, the predicted total
acid concentration and conversion for the countercurrent fermentation at
various LRT and VSLR were obtained. Figure 12 shows the CPDM "maps"
for the 80% bagasse/20% chicken manure fermentation system (solids con-
centration = 127 g of VS I [L of liquid]) with different inoculum sources. The
results indicated that the fermentation with marine inoculum yielded
higher acid concentration and conversion than the fermentation with ter-
restrial inoculum at any VSLR and LRT. Furthermore, the CPDM "map" for
fermentation with marine inoculum showed that 90% conversion is pos-
sible and that the carboxylic acid concentrations could reach 23.4 giL at a
VSLR of 2 gl (L·d) and LRT of 20.5 d.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


~
~
~ Table 4
Q..
III
Results for Bagasse/Chicken Manure Countercurrent Fermentation with Marine and Terrestrial Inoculum

!"')
~ Fermentation train N P Q R I
<b J
3
~.
q- Average pH in all fermentors 6.3 ± 0.18 6.3 ± 0.16 6.3 ± 0.2 6.3 ± 0.2 6.4 ± 0.29 6.3 ± 0.31
'<:
III 1.38 1.18 0.928 0.841 0.791 0.547
:J Total acid productivity
Q..
III (g/[L liquid in all fermentors·d])
o· Total acid concentration (g/L) 16.2 ± 1.08 14.0 ± 1.15 19.7 ± 0.70 18.8 ± 0.73 9.6 ± 1.05 13.2 ± 0.64
Ib
!"')
~ 45.1 ± 2.12 44.9 ± 0.69 35.4 ± 4.10 35.1 ± 4.95
:J Acetic acid (wt%) 45.5 ± 3.60 44.9 ± 4.93
0
Propionic acid (wt%) 21.7 ± 4.37 22.2 ± 4.50 15.7 ± 2.12 22.2 ± 0.63 20.2 ± 2.50 25.0 ± 3.31
~ Butyric acid (wt%) 12.9 ± 2.68 12.9 ± 2.69 14.3 ± 1.32 12.9 ± 0.38 10.7 ± 2.06 16.5 ± 3.01
Valeric acid (wt%) 9.8 ± 1.48 9.5 ± 1.36 11.2 ± 0.87 9.5 ± 0.19 14.1 ± 2.51 13.1 ± 2.10
VI Caproic acid (wt%) 5.9 ± 2.01 6.6 ± 1.57 8.0 ± 0.87 6.6 ± 0.22 8.38 ± 1.35 5.60 ± 0.64
.j:::.
0 Heptanoic acid (wt%) 4.2 ± 1.59 3.9 ± 0.98 5.7 ± 2.70 3.9 ± 0.14 11.2 ± 2.66 4.66 ± 1.16
VS digested (g VS / d) 2.20 2.17 1.52 1.39 2.10 1.20
Yield (g total acid/ g VS fed) 0.359 0.308 0.425 0.385 0.208 0.250
Selectivity (g total acid/ g VS digested) 0.667 0.553 0.559 0.554 0.386 0.417
Non-acid dissolved VS (g VS/g VS digested) 0.286 0.304 0.336 0.281 0.157 0.225
Conversion (g VS digested/ g VS fed) 0.538 0.556 0.760 0.695 0.538 0.600
Biotic CO2 productivity 0.389 0.576 0.448 0.437 0.628 0.601
(g CO2 /[L of liquid in all fermentors·d])
&
:-
CH4 productivity 0 0.0002 0.0002 0.0011 0.0001 0.0001
(g CH4 /[L of liquid in all fermentors·d])
Mass balance closure (g VS out/ g VS in) 1.02 1.01 1.03 1.03 0.863 1.01
-~
c
-
~ "All errors are:!: 1 SD.
to..>
c
C
....,
).
:g
[
tlJ

~
~
in"
q-
'<:
III
::.
Cl..
tlJ

[
::.
o
& Table 5
'<:
The Values of e, f, g, h and Selectivity Used in CPDM Mathematica Program for Bagasse/Chicken Manure Fermentation
Inoculum Selectivity e f g h
V1
~ sources (g AJ g VS digested) (g AJ[g VS·d]) (dimensionless) (L/ g total acid)l/h (dimensionless)
.....
100% Terrestrial 0.512 0.069 2.60 0.003 2.57
40% Marine + 60% Terrestrial 0.675 0.156 2.80 0.015 1.99
100% Marine 0.752 0.153 2.55 0.047 1.38

~
~
-..
C
,QQ

§'"
542 Thanakoses et al.
Table 6
Average Parameter Constant Values for Bagasse/Chicken
Manure Fermentation with Terrestrial and Marine Inocula
Used in CPDM Mathematica Program
Parameter constant Value
Holdup (g liquid/g VS wet cake) 4.56 ± 0.828
Moisture (g liquid/ g wet solid) 0.066 ± 0.008
FI-F4 solids concentration (g VS/L) 127 ± 17.4
FI-F4liquid volume (L) 0.250 ± 0.062
<I> (g total acid/ g Ae) 0.815 ± 0.033

"All errors are ± 1 SO.

- - 100% marine
60
....... 40% marine + 60% terrestrial
,....,
t::! 50 - - - - 100% terrestrial
~
=
0
.;j 40
§
=
<1l
30
=
t)

0
t)

~ 20
<il
~ 10

0
0 0.2 0.4 0.6 0.8
Conversion

Fig. 12. CPOM maps for bagasse/chicken manure countercurrent fermentation


(127 g of VS/L).

Verification of CPOM Method


Using Eq. 15 with the parameter constants for each case in the CPDM
Mathematica program, the acid concentration and conversion were pre-
dicted at varying VSLR and LRT from fermentation trains A to R. Tables 7
and 8 compare the predicted and actual acid concentration and conversion.
The average error between the experimental and predicted acid concentra-
tion in the fermentations with 100% terrestrial, 40% marine + 60% terres-
trial, and 100% marine inocula was 20.1, 25.2, and 33.1 %, respectively. The
average error between the experimental and predicted conversion in the
fermentations with 100% terrestrial, 40% marine + 60% terrestrial, and 100%
marine inocula was 24.1, 33.2, and 35.5%, respectively. The model variation
could have resulted from variations in chicken manure quality because
other studies show much less variation (14).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


).
:g
~
t:l..
Q:)
o· Table 7
1"\ Comparison of Experimental and Predicted Carboxylic Acid Concentration and Conversion
~
<!>
3 for Bagasse/Chicken Manure Fermentation with 100% Terrestrial Inoculum
iii·
~ Fermentation train Average
til
::.
t:l..
Q:) A B C 0 E F G H I K L M (%)"
o· J
fb
1"\ Experimental carboxylic 14.2 16.6 21.0 15.2 20.1 20.0 19.0 18.7 9.6 13.1 11.1 13.0 15.1
~
::.
0 acid concentration (g/L)
~
'< Predicted carboxylic 17.2 17.0 17.2 16.1 16.6 17.9 15.5 14.9 14.7 16.0 14.5 13.7 20.4
acid concentration (g/L)
(from CPOM)
U1
~ Error (%)b 20.8 2.5 -18.3 5.8 -17.4 -10.6 -18.4 -20.2 53.7 21.9 30.7 5.4 35.2 20.1
w
Experiment conversion 0.430 0.452 0.413 0.385 0.274 0.438 0.385 0.581 0.538 0.600 0.530 0.484 0.412
Predicted conversion 0.393 0.392 0.238 0.395 0.309 0.373 0.420 0.566 0.770 0.877 0.902 0.693 0.431
(from CPOM)
Error (%)b -8.6 -13.3 -42.4 2.6 12.8 -14.8 9.1 -2.6 43.1 46.2 70.2 43.2 4.6 24.1

Expected conversion 0.343 0.342 0.220 0.347 0.278 0.325 0.368 0.484 0.641 0.722 0.741 0.581 0.366
(from Eq. 23)
Error (%)< -20.3 -24.3 -46.6 -9.8 1.6 -25.7 -4.3 -16.8 19.2 20.3 39.9 20.1 -11.2 20.0
~
:-
.... "Average = average absolute error.
bError = (predicted - experimental) x 100/experimental.
....c~ cError = (expected - experimental) x 100/experimental.
~

""....,cc
~
[
g.\Xl
::r-
3<;:;.
~
:;,
""Cl.. Table 8
\Xl Comparison of Experimental and Predicted Carboxylic Acid Concentration and Conversion
o· for Bagasse/Chicken Manure Fermentation with 40 and 100% Marine Inoculum
~
::r-
:;,
o 100% Marine inoculum 40% Marine + 60% terrestrial inoculum
~ Train N Train Q Average (%)" Train P Train R Average (%)"
Experimental carboxylic acid concentration (g/L) 16.2 19.7 14.0 18.8
U1
~ Predicted carboxylic acid concentration (g/L) 22.9 24.6 19.4 21.0
~
Error (%)b 41.3 24.9 33.1 38.6 11.7 25.2
Experimental conversion 0.538 0.760 0.556 0.695
Predicted conversion 0.809 0.917 0.769 0.876
Error (%)b 50.4 20.6 35.5 38.3 21.0 32.2
aAverage = average absolute error.
bError = (predicted - experimental) x 100/ experimental.

-~
,~
N
Cl
:::3
Carboxylic Acids from Sugarcane Bagasse 545
• Result data
- Best line fit


0.5
• ••
O+---r--~·~-F~L-.---.-on,--,
0.1 0.2 •0!3· 0.4 0.5 0.6 0.7
-0.5 •
Selectivity (g total acid/g VS digested)

Fig. 13. Correlation of error with selectivity for bagasse/ chicken manure counter-
current fermentation:

Figure 13 shows the error between the predicted conversion, x pred' and
the experimental conversion, x exp' as a function of selectivity. Linear regres-
sion of the data gives the correlation in Eq. 22.
X pred - xexp
error = x = 1.12s - 0.307 (22)
exp

Substituting Eq. 19 into Eq. 22 gives the correlation among VSLR, x pred'
and xexp:

(23)
Xexp = _0.0086 VSLR + 1.23
In Eq. 23, xexp may be interpreted as the "expected conversion." The
expected conversions for fermentation trains A-M are presented in Table 7.
The results indicate that the average expected conversion error from Eq. 23
(20.0%) was less than the average predicted conversion error from the CPDM
map (24.1%). Applying these corrections reduces the error in conversion
that was directly obtained from the CPDM model, in which the selectivity
was kept constant.

Conclusion
The 80% bagasse/20% chicken manure countercurrent fermentation
with terrestrial inoculum operated at LRT = 8.88 d and VSLR = 7.36 g/ (L of
liquid·d) had the highest acid productivity (2.5 g/[L·d]) with the highest
acid selectivity (0.581 g of acid/ g of VS digested) and highest yield (0.338
g of acid/ g of VS fed). The fermentation operated at a LRT = 20.5 d and a
VSLR = 2.13 g/(L of liquid·d) had the highest conversion (0.60 g of VS
digested/ g ofVS fed). Using marine inoculum, instead of terrestrial inocu-
lum, increased the total acid productivity, yield, and acid selectivity. CPDM
predicted the total acid concentration and conversion within 20.1 and 24.1 %,
respectively, of experimental results. By applying a correction for varying
selectivity, the experimental conversion could be predicted within 20%.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


546 Thanakoses et al.
In summary, 90% conversion of biomass to carboxylic acid at a concen-
tration of 23.4 gIL is possible by using marine inoculum (solids concentra-
tion of127 gl [L of liquid]). Furthermore, applying CPDM could predict the
optimum condition for countercurrent fermentation that should be veri-
fied at the pilot plant and industrial scales.

References
1. Klass, D. L. (1998), Biomass for Renewable Energy, Fuels, and Chemicals, Academic Press,
New York, NY, p. 150.
2. Holtzapple, M. T., Ross, M. K., Chang, N. S., Chang, V. S., Andelson, S. K., and Brazel,
C. (1997), in Fuels and Chemicals from Biomass, Saha, B. C. and Woodward, J., eds.,
American Chemical Society, Washington, DC, pp. 130-142.
3. Holtzapple, M. T., Davison, R R, Ross, M. K., Aldrett-Lee, S., Nagwani, M., Lee, C.
M., et al.. (1999), Appl. Biochem. Biotechnol. 77-79,609-631.
4. Ross, M. K. (1998), PhD thesis, Texas A&M University, College Station, TX.
5. Russell J. B. and Martin S. A. (1985), J. Anim. Sci. 59, 1329-1338.
6. Bauchop, T. (1967), J. Bacteriol. 94, 171-175.
7. Chapula, W. (1980), in Digestive Physiology and Metabolism in Ruminants, Ruckebusch,
Y. and Thivend, P., eds., MTP, London, UK, pp. 325-347.
8. Martin, S. A. and Macy J. M. (1985), J. Anim. Sci. 60,544-550.
9. Sauer, F. D. and Teather, R M. (1987), J. Dairy Sci. 70, 1835-1840.
10. Loescher, M. E. (1996), PhD thesis, Texas A&M University, College Station, TX.
11. Domke, S. B. (1999), PhD thesis, Texas A&M University, College Station, TX.
12. Playne, M. J. (1981), Adv. Biotechnol. 2, 85-90.
13. Chang, V.S.,Nagwani,M.,andHoltzapple,M. T. (1998), Appl. Biochem.Biotechnol. 74,
135-159.
14. Thanakoses, P. (2002), PhD thesis, Texas A&M University, College Station, TX.
15. South, C. Rand Lynd, L. R(1994), Appl. Biochem. Biotechnol. 45/46,467-481.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 / 0547 /$20.00

Alternative Approach for Utilization


of Pentose Stream from Sugarcane Bagasse
by an Induced Flocculent Pichia stipitis
HEIZIR F. DE CASTRO, *,1 SAMUEL C. OLiVEIRA, 2
AND SANDRA A. FURLAN 3
Department of Chemical Engineering, 2Department of Biotechnology,
1

Faculty of Chemical Engineering of Lorena, PO Box 116, 12606-970,


Lorena, SP, Brazil, E-mail: heizir@dequUaenquil.br; and
3University of join ville, PO Box 246, 89201-972, joinville, SC, Brazil

Abstract
A new approach for the utilization of hemicellulosic hydrolysate from
sugarcane bagasse is described. This approach consists of using the hydroly-
sate to dilute the conventional feedstock (sugarcane juice) to the usual sugar
concentration (150 giL) employed for the industrial production of ethanol.
The resulting sugar mixture was used as the substrate to evaluate the perfor-
mance of a continuous reactor incorporating a cell recycle module, operated
at several dilution rates. An induced flocculent pentose-fermenting yeast
strain was used for this bioconversion. Under the conditions used, the reactor
performance was satisfactory at substrate feed rates of 30 g/(L·h) or less,
corresponding to an ethanol productivity of about 11.0 gl (L· h) and an overall
sugar conversion >95%. These results show real advantages over the existing
alternatives for a better exploitation of surplus bagasse to increase industrial
alcohol production.
Index Entries: Sugarcane bagasse; sugar mixture; Pichia stipitis; floccula-
tion; ethanol.

Introduction
The main objective of the studies related to pentose fermentation is to
establish favorable process conditions, which can also be applied to hemi-
cellulosic hydrolysate obtained from agricultural residues. The prepara-
tion of this kind of hydrolysate often generates inhibitory components
(acetic acid, hydroxymethylfurfural), which have negative effects on the
fermentation process (1,2). Because the formation of these components
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 547 Vol. 105-108,2003


548 de Castro et at.
cannot be avoided, several methodologies have been employed to mini-
mize their undesirable effects. These include, detoxification strategies for
removing these components (3-4), improvement of the medium composi-
tion by adding special nutrients (1) or microorganism acclimatization to the
toxic inhibitors (1,5).
Under batch cultivation, these methodologies have proven to be effi-
cient using several types of microorganisms, although they increase the
cost of the process. In the evolution of the studies carried out in our labo-
ratory, these strategies showed limited efficiency to achieve satisfactory
reactor performance under continuous operation when the most promis-
ing pentose-fermenting yeast (Pichia stipitis) was used. This was related not
only to the chemical environmental composition available to the biocata-
lyst but also to the poor ability of the selected yeast strain to maintain high
cell concentrations inside the reactor. A different behavior, however, was
observed when the pentose stream was combined in different proportions
with sugarcane juice and used as feed medium in the fermentation system
(6). Under these conditions, the selected yeast strain was shown to have
very strong flocculating properties, which allowed maintenance of high
cell concentrations inside the reactor. This suggested that sugarcane juice
could be used as an inducer of flocculation in P. stipitis.
Taking advantage of this property, a new approach for the utilization
of the pentose component obtained from acid hydrolysis of sugarcane ba-
gasse is proposed (Fig. 1)'. By diluting the sugarcane juice with sufficient
hemicellulose hydrolysate to attain the usual level of sugar concentration
of ethanol plants (150 giL), a substrate containing a sugar mixture (sucrose,
glucose, and xylose) was used to evaluate the performance of a continuous
reactor incorporating a cell recycle module, operated at several dilution
rates. Since high cell densities were achieved, high conversion of the mixed
substrate was possible. In this article, we give the flocculating properties of
P. stipitis growing under continuous operation. We also present the steady-
state concentrations of relevant process variables.
Materials and Methods
Preparation of Microorganism and Inoculum
P. stipitis strain CBS 5773 was employed. Stock culture was maintained
at 4°C on malt agar slants. Yeast was grown semiaerobically at 30°C for 24 h
in Erlenmeyer flasks containing 20 mL of presterilized 10% sucrose broth
(sugarcane juice) at pH 5.5. Amounts of additional nutrients/L of media
prepared were 5.0 g of (NH4 h 504 ,1.0 g of KH2P04 , 1.5 g of MgS04·7H20,
0.1 g of CaCl2 , and 0.5 g of yeast extract.
Preparation of Sugarcane Bagasse Hydrolysate
and Fermentation Medium
Acid hydrolysis of sugarcane bagasse was carried out at 190°C for 15
min in a 25-L working volume steel reactor (AISI 316), with agitation. The

Applied Biochemistry and Biotechnology Vol. 105-708,2003


Utilization of Pentose Steam by P. stipitis 549

Sugarcane

1
--
Sugar cane juice Medium Preparation
Sugar extraction
Total sugars> 20%
• (Sucros9+glucos9+xylose)

1
(sucrose, hexose)

Sugar cene bagasse Total sugars


2% (pentose, hexose)

Acid hydrolysis Pentose stream


~l Hemicellulose
Hydrolyzate J
H2S04
190°C 115 min
Ethanol

1
Lignin

Fig. 1. Flow chart for ethanol process by P. stipitis based on utilization of sugarcane
juice diluted with pentose stream from sugarcane bagasse.

Table 1
Substrate Characteristics"
Substrate Reducing Total Furfural Acetic acid
type sugars (giL) sugars(g/L) (giL) (giL)
Sugarcane juice 15.00 237.00 0.49
Bagasse hydrolysate 23.00 23.00 1.10 3.37
Sugarcane juice diluted 19.80 149.90 0.66 0.92
with bagasse hydrolysate
"Average values are given.

reactant mixture consisted of 70 mg of concentrated (98% [w Iv]) H 2S04 /g


(dry wt) of sugarcane bagasse and sufficient water to provide a liquid-
to-solid ratio of 10:1. After hydrolysis, the liquid fraction was used to dilute
fresh crushed sugarcane juice to a desirable sugar concentration (150 giL
of total reducing sugars). Then, the medium was supplemented with nutri-
ents at the same concentrations used for the inoculum preparation. The
chemical composition of the feedstock sources and the resulting mixed sub-
strate is shown in Table 1. Under the conditions used, a sugar mixture con-
taining 20% monosaccharide and 80% disaccharide was obtained. The
concentrations of acetic acid and furfural originally present in the bagasse
hydrolysate were significantly reduced to levels <1 giL.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


550 de Castro et al.

F+FR
VF
X
P Vs
TS

F F
TSo
Fig. 2. Schematic diagram of experimental system used in continuous fermentation
runs (VF = 0.221, Vs = 0.025 L; FR = recycle rate).

Fermentation Conditions
Fermentation runs were carried out in a simple column reactor fitted
with a separate settling device and working volume of 0.245 L. The fer-
mentation system utilized appropriate sensors to control the temperature
and pH, as shown schematically in Fig. 2. Continuous fermentation was
started by loading the reactor with substrate medium at a low sugar con-
centration and adjusting the control parameters (pH 5.0, temperature of
30°C, and airflow rate of 0.1 vvm). Then the reactor was inoculated with
cell suspension using five precultured conical flasks. A start-up period of
4-6 d was required to accumulate a working cell density (X> 80 giL, dry
wt), and for this purpose, the fermentation system was continuously fed
with substrate medium at increased sugar levels ranging from 50 to 150
giL. Several dilution rates (D = 0.1 to 0.4 h-1) were imposed in order to
determine the limit performance of the reactor for this specific substrate.
During the continuous runs, periodic samples were taken from the reac-
tor vessel for analysis of the relevant variables, such as sugar output,
biomass density, and product concentration. For each tested dilution rate,
the fermentation parameters were determined when the concentrations
within the reactor remained relatively constant over a period correspond-
ing to at least three residence times. For comparison, the reactor was also
run with pure sugarcane medium containing 150 giL of total sugars.
Analytical Procedures
The cell concentrations in the overflow were determined from a cali-
bration curve relating the optical density (600 nm) to cellular biomass. Cell
density in the reactor was directly determined by centrifuging 10 mL of
culture, resuspending cells in distilled water, recentrifuging, and then dry-
ing at 105°C to constant weight. The degree of flocculation, expressed as

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Utilization of Pentose Steam by P. stipitis 551
sedimentation rate, was determined by measuring the height of sediment
cells in a millimetric ally graduated cylinder as it settled at 1 min intervals
(7). Total reducing sugars were assayed following the method described by
Nelson (8). Xylose and acetic acid concentrations were analyzed by high-
performance liquid chromatography (HP 1082B) using a refractive index
detector and a Bio-Rad HPX 87H (300 x 7.8 mm) column, and employing the
following conditions: H 2S04 (5·10-3 M) eluent, 0.6 mL/min flux, 45°C col-
umn temperature, 16X detector attenuation, 20-mL sample volume. Etha-
nol concentration was measured by gas chromatography (GC model 35)
equipped with an FFAP 20% Chromosorb W column and flame ionization
detector with an injection temperature of 110°C (9).

Results and Discussion


Flocculation and Cell Retention Within Reactor
Flocculation is a widespread phenomenon in the microbial world, that
is frequently used as an alternative way to obtain a high-ceIl-density fer-
mentation system (10,11). The most common application of flocculating
microorganisms can be found in the production of ethanol for beverage and
nonbeverage uses, employing several reactor configurations (12,13). Con-
trary to hexose-fermenting yeasts (i.e., typically Saccharomyces cerevisiae),
which are reasonably well-understood systems (10,11), flocculation in pen-
tose-fermenting yeasts is a very recent method (14-17). In these yeasts,
flocculation has been observed in cells of Pachysolen tannophilus (15) and
Pichia stipitis (16,17). However, the feasibility of using such yeasts in con-
tinuous fermentations is still limited by their poor stability and weak floc-
culation tendencies (17,18). If floc retention inside the reactor could be
increased, then the phenomenon would have a great technological impor-
tance for ethanol production from hemicellulosic wastes. This has been
achieved by using the system described here. High cell concentrations
within the reactor were attained in a period no longer than 4 d. The ratio of
the cell concentration inside the reactor (X) to that in the overflow (Xe)
defined by Denverell and Clark (18) was used to measure the flocculation
intensity of P. stipitis cells during the start-up period. As shown in Table 2,
there was a noticeable increase in the yeast settling accompanied by a
buildup of dense particle flocs. Therefore, a gradual increase in the floccu-
lation intensity (XI Xe) was achieved, resulting in a cell concentration ratio
of about 36.5 after 48 h. This allowed the reactor to run with virtually total
biomass retention and clear effluent. A sample of the flocculation culture,
taken from the reactor, was used to conduct a set of standard flocculation
tests (7), as shown in Table 3. No attempt was made to explain the mecha-
nism by which cell aggregation occurred; however, the characteristics
of this cellular aggregate are similar to those attained by the flocculent yeast
strain S. cerevisiae (19,20). Despite the characteristics of the substrate
used, no negative effects on yeast flocculation properties or activity were
observed during the continuous runs.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


552 de Castro et al.
~--------------------~25

60

~ 45
L::
o
~
~ 3
L::
8
5
TOlalre~

O+--r~--~~-r~--~~~O
0.0 0.1 0.2 0.3 0.4
Dilution rale' 1)(h)

Fig. 3. Steady-state variables of continuous ethanol production by P. stipitis using


sugar mixture substrate.

Continuous Runs
The performance of the fermentation system was evaluated by plot-
ting ethanol productivity (Qp), residual sugar concentration (5), and prod-
uctconcentration (P) as a function of the dilution rate (D) (Fig. 3). An average
ethanol concentration of 56 giL was achieved at dilution rates ranging
from 0.12 to 0.4 h-1, which corresponded to ethanol productivities from 6.6
to 22.6 gl (L·h).
Table 4 shows that sugars metabolized as hexoses (sucrose and glu-
cose) were almost fully converted, while pentose (xylose) conversion
markedly decreased with the increase in dilution rate. Although both
sugars were consumed simultaneously, the hexose consumption rate by
P. stipitis was much higher than the pentose one. As a consequence, a
progressive loss of sugars in the output stream (as pentose) was observed
when the dilution rate increased. This is in agreement with findings that
xylose metabolism in P. stipitis is inhibited but not repressed by hexoses.
Thus, in our case, the rate of xylose fermentation was found to be the
limiting step of the process, and maximum dilution rate for effective con-
version was 0.20 h-1 • Increasing dilution rate above this level resulted in
a progressive loss of sugars, namely, pentose in the output stream.
No significant differences were observed in the apparent product
yield (Yp/s) throughout the continuous runs (Table 5), with an average
value of 0.39 gl g. Since most of the ethanol was produced from hexoses,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Utilization of Pentose Steam by P. stipitis 553
Table 2
Retention and Accumulation of Pichia stipitis Cells
Within Reactor Using Substrate Based
on Sugarcane Juice Diluted with Pentose Stream
Duration Cell concentration inside Output cell concentration XIXe
(h) reactor, X (giL) Xe (giL) ratio
7 5.2 NO 5.2
24 6.9 1.1 6.3
48 14.6 0.4 36.5
72 37.7 NO 37.7
96 55.2 2.5 22.1
120 76.6 1.7 45.0
144 97.0 2.8 34.0
148 107.4 4.1 26.1
aND, not detected.

Table 3
Flocculent Properties of P. stipitis Using Sugarcane Juice
Diluted with Hemicellulose Hydrolysate
Characteristics Values
Sedimented cells in 5 min (%) 75
Flocculation velocity (cml s) 0.14
Specific rate of sedimentation (mLI g) 27.5
Washout resistant D = 0.4 1h-1 , X = 93.0 giL

Table 4
Influence of Substrate Feed Rate on Individual and Overall Sugar Conversion
Substrate Substrate Substrate
input (giL) output (giL) conversion (%)
o (h-1) Xyl Suc+Glu IS Xyl Suc+Glu IS Xyl Suc+ Glu Overall
0.12 18.5 129.2 147.7 4.9 0.3 5.2 73.5 99.7 96.5
0.14 13.8 139.1 152.9 5.9 0.6 6.5 57.2 96.6 95.7
0.20 11.4 141.9 152.4 4.9 0.5 5.4 57.0 99.6 96.4
0.37 20.2 131.0 151.2 7.4 0.5 7.9 62.0 99.7 94.7
0.41 15.1 141.3 156.4 10.1 0.7 10.8 33.1 99.5 93.1
aXyl, xylose; Sue, sucrose; Glu, glucose; IS, total sugar.

the average (Yp / s) value was unexpectedly lower than the one observed for
hexose-fermenting yeasts, such as S. cerevisiae (Yp/s = 0.45 g/ g) (20). This
suggests that there are distinct physiologic differences between pentose
and hexose yeasts.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
554 de Castro et al.
Table 5
Performance of Bioreactor Characteristics
on Different Substrates Using Pichia stipitis
Substrate type
Reactor Sugarcane Sugarcane juice diluted
parameters juicea with pentose stream
Dilution rate (h-1) 0.36 0.37
Substrate input (giL) 147.9 151.2
Residual sugars (giL) 3.3 7.9
Product (giL) 59.9 55.9
Biomass (giL) 90.0 94.7
Sugar conversion rate (%) 97.8 94.7
Y p/ s (gig) 0.41 0.39
Qp (g/L·h) 21.6 20.7
f.tp(g/g·h) 0.24 0.22
DSo (g/[L·hD 53.2 55.9
aData obtained using the same reactor system running on sugarcane medium.

Conclusions
Research and development studies of alcohol fermentation technol-
ogy have been conducted over the last two decades to make alcohol pro-
duction more efficient. In this context, there are clear advantages in using
surplus bagasse from ethanol plants as a raw material for the same end
product. Alcohol fermentation of pentose has become an attractive research
topic and publications are numerous. In principle, biochemical and micro-
biologic problems, such as finding strains and establishing the somewhat
unusual (e.g., semiaerobic) optimal conditions, have been solved at a labo-
ratory-research leveL However, so far, a suitable process technology has
not been established.
Given our background in developing very effective ethanol technol-
ogy based on disaccharide feedstock (19,20), it would be logical to con-
tinue this research program using pentose feedstocks. The present study
investigated the feasibility of using hemicellulosic hydrolysate to dilute
the conventional raw material (sugarcane juice) to a standard sugar con-
centration employed for the industrial production of ethanoL Under the
conditions used here, the reactor performance was satisfactory at sub-
strate feed rates of 30 g/ (L·h) or less, corresponding to an ethanol produc-
tivity of about 11.0 g/ (L·h) and an overall sugar conversion >95%. On a
laboratory scale, this process has shown to be an attractive alternative for
utilization of the bagasse generated by the ethanol plants. Moreover, this
process has the potential to attain high ethanol levels; avoid treatment for
removing inhibitory compounds normally found in the hemicellulosic
acid hydrolysate; decrease the addition of microbial nutrients; and require

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Utilization of Pentose Steam by P. stipitis 555
minimal changes on ethanol plants for its implementation, i.e., substitu-
tion of the yeast genus S. cerevisiae by P. stipitis. However, a successful
application of this novel approach is limited by the implementation of an
adequate dilution rate on the reactor in order to achieve an efficient pen-
tose conversion. Optimization of the dilution rate is currently the main
potential of further developments of this technology.

Acknowledgments
We thank Maria Eunice M. Coelho for revising the manuscript. We
also gratefully acknowledge Conselho Nacional de Desenvolvimento
Cientffico e Tecnol6gico (CNPq) for financial support.

References
1. Olsson,1. and Hahn-Hagerdal, B. (1996), Enzyme Microb. Technol. 18, 312-33l.
2. Clark, T. and Mackie, K 1. (1984), J. Chem. Biotechnol. 34B, 101-110.
3. Palmqvist, E., Hahn-Hagerdal, B., Szengyel, Z., Zacchi, G., and Rczey, K (1997),
Enzyme Microb. Technol. 20,286-293.
4. Silva, S. 5., Ribeiro, J. D., Felipe, M. G., and Vitolo, M. (1997), Appl. Biochem. Biotechnol.
63/65,557-564.
5. Parekh,S., Parekh, R, and Wayman, M. (1987), Process Biochem. 22, 86--9l.
6. de Castro, H. F., Castro, 1. A B., and Nunes, A 1. 1. (1989), in Biomass for Energy and
Industry, vol. 2, Grassi, G., Gosse, G., and dos Santos, G., eds., Elsevier Applied
Science, London, UK, pp. 2318-2322.
7. de Castro, H. F. and Oliveira, P. C. (1998), Revista Microbiologia 29, 27-30.
8. Nelson, N. A (1944), J. Bioi. Chem. 153,357-380.
9. Roberto, I. c., Mancilha, I. M., Felipe, M. G. A, Silva, S. 5., and Sato, S. (1994), Arq. Bioi.
Tecnol. 37,55-63.
10. Calleja, G. B. (1987), in The Yeasts, vol 2, Rose, A H. and Harrison, J. S. , eds.,
Academic Press, New York, NY, pp.165-20l.
11. Stratford, M. (1996), Cerevisiae 1, 38-45.
12. Grenshields, R N. and Smith, E. 1. (1974), Process Biochem. 9,11-28.
13. Jones, S. T., Korus, R. A, Admassu, W., and Heimsch, R C. (1984) Biotechnol. Bioeng.
26,742-747.
14. Weeb, S. R. and Lee, H. (1990), Biotechnol. Adv. 8,685-697.
15. Chung, I. S. and Lee, Y.Y. (1986), Biotechnol. Bioeng. Symp. 17,391-400.
16. Pereira Jr., N. and Bu'Lock, J. D. (1993), Revista Microbiologia 24, 132-139.
17. Groo~en, D. R J., Vleesenbeek, R, Windmeijer, M. G. A, van der Lans, R G. J. M.,
and Luyben, KC.A M. (1991), Enzyme Microb. Technol. 13,734-739.
18. Denverell, K F. and Clark, T. A (1985), Biotechnol. Bioeng. 27, 1608-161l.
19. Paiva, T. C. B., 5ato, 5., Visconti, A E. 5., and Castro, 1. A B. (1996), Appl. Biochem.
Biotechnol. 57/58, 535-54l.
20. Oliveira,S. c., Paiva, T. C. B., Visconti, A E. 5., and Giudici, R. (1999), Appl. Biochem.
Biotechnol. 74(3),161-172.

Applied Biochemistry and Biotechnology Vol. 105-708,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 / 0557 /$20.00

Hydrogen Production
from Paper Sludge Hydrolysate

ZSOFIA KADAR,1,2 TRUUS DE VRIJE,1 MIRIAM A. W. BUDDE,1


ZSOLT SZENGYEL, 2 KATI RECZEY, 2 AND PIETERNEL A. M. CLAASSEN*,1
lAgrotechnological Research Institute (A TO B. V.),
PO Box 17, 6700 AA Wageningen, The Netherlands,
E-mail: p.a.m.c/aassen@ato.wag-ur.nl; and
2Budapest University of Technology and Economics,
Department of Agricultural Chemical Technology,
Szent Gellert ter 4., H-1521 Budapest, Hungary

Abstract
The main objective of this study was to develop a system for the produc-
tion of "renewable" hydrogen. Paper sludge is a solid industrial waste
yielding mainly cellulose, which can be used, after hydrolysis, as a feed-
stock in anaerobic fermentation by (hyper)thermophilic organisms, such as
Thermotoga elfii and Caldicellulosiruptor saccharolyticus. Tests on different
medium compositions showed that both bacteria were able to produce
hydrogen from paper sludge hydrolysate, but the amount of produced
hydrogen and the requirement for other components differed. Hydrogen
production by T. elfii strongly depended on the presence of yeast extract
and salts. By contrast, C. saccharolyticus was less dependent on medium
components but seemed to be inhibited by a component present in the
sludge hydrolysate. Utilization of xylose was preferred over glucose by
C. saccharolyticus.

Index Entries: Hydrogen production; paper sludge; Thermotoga elfii;


Caldicellulosiruptor saccharolyticus.

Introduction
The Earth's climate has changed because of human activities. Since
the beginning of the industrial revolution, atmospheric concentrations of
the greenhouse gases-carbon dioxide, ~ethane, and nitrous oxide-
have increased continuously. Transportation specifically contributes to
global warming through the burning of gasoline and diesel produced
from fossil feedstock. The use of noncarbonaceous fuels, produced from
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 557 Vol. 105-108,2003


558 Kadar et al.
renewable feedstock, reduces greenhouse gas emission, during fuel pro-
duction and fuel combustion.
The idea of using hydrogen as a fuel is not a novel one, and interest in
hydrogen production has grown in recent years. The major advantage of
energy from hydrogen is the lack of polluting emissions since the utiliza-
tion of hydrogen for energy production, either via combustion or via fuel
cells, results in pure water. Renewable hydrogen can be produced from
biomass by either gasification or fermentation (1).
At present, one of the greatest drawbacks to production of renewable
hydrogen is cost. The use of cheap, renewable raw materials, such as agri-
cultural or industrial wastes or byproducts, could reduce the cost.
In Hungary, 50,000 t of paper sludge is produced annually. It is a
solid industrial waste, arising from the pulping and paper-making indus-
tries, yielding mainly cellulose. Landfilling is the most preferred han-
dling option for paper sludge, but costs have risen dramatically over the
past decade and space is also limited (2). However, companies from vari-
ous industrial sectors have special responsibilities and tasks, according to
their fields, in protecting and developing the environment. This forces the
companies to find new solutions. Technologies for the production of
valuable products would be attractive. Paper sludge, after being hydro-
lyzed, could be used as a renewable feedstock for hydrogen production
during anaerobic fermentation by thermophilic microorganisms.
Several (hyper)thermophilic microorganisms that convert sugars to
hydrogen, carbon dioxide, and organic acids at the theoretical efficiency
of 4 mol of hydrogen/ mol of glucose consumed have been described (1).
Members of the order Thermotogales were first found in natural ecosys-
tems associated with active volcanism (3). Thermotoga elfii, which was
isolated from an African oil field, is a thermophilic, glucose-fermenting,
strictly anaerobic bacterium, and it is able to grow on different carbohy-
drates (4). Caldicellulosiruptor saccharolyticus, originating from New
Zealand thermal springs, is an obligate anaerobic, extremely thermo-
philic, cellulolytic bacterium (5). The present study addresses the require-
ments of T. elfii and C. saccharolyticus for growth and hydrogen production
from paper sludge hydrolysate.

Materials and Methods


Microorganisms and Culture Media
Thermotoga elfii DSM 9442 and Caldicellulosiruptor saccharolyticus DSM
8903 were obtained from the Deutsche Sammlung von Mikroorganismen
und Zellkulturen.
The medium for growth of T. elfii contained per liter of demineralized
water: 1.0 gofNH4Cl, 0.3 gofK2HP04, 0.3 gofKH 2P04, 0.2 gofMgCl2 ·6H20,
0.1 g of CaCI2 ·2H20, 0.1 g of KCl, 10 g of NaCl, 0.5 g of Na-acetate, 0.5 g of
cysteine-HCI,4 g of yeast extract, 0.5 mg of resazurin,20 mL of 10% Na 2C03,
and 10 mL of trace element solution.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
H Production from Paper Sludge Hydrolysate 559
The trace element solution contained per liter: 1.5 g of nitrilotriacetic
acid, 3 g of MgS04·7H20, 0.5 g of MnS0 4·2H20, 1 g of NaCl, 0.1 g of
FeS04·7H20,O.18gofCoS04·7H20,O.lgofCaCl2·2H20,O.18gofZnSO4·7H20 ,
10 mg of CuS04·5H20, 20 mg of KAI(S04)2·12H20, 10 mg of H 3B03, 10 mg of
Na2Mo04·2H20, 25 mg of NiCl2·6H20, and 0.3 mg of Na2Se03·5H20.
The medium used for cultivation of C. saccharolyticus contained per
liter-of demineralized water: 0.9 g of NH4Cl, 1.5 g of K2HP04, 0.75 g of
KH2P04, 0.4 g of MgCl2·6H20, 0.9 g of NaCl, 2.5 mg of FeCl3·6H20, 19 of
yeast extract, 0.75 g of cysteine-HCl, 0.5 mg of resazurin, and 1 mL of trace
elements solution (SL-10).
The trace element solution, SL-10, contained per liter: 1.5 g of
FeCl2·4H20, 70 mg of ZnCl2, 100 mg of MnCl2·4H20, 6 mg of H 3B03, 190 mg
of CoCI2·6H20, 2 mg of CuCI2·2H20, 24 mg of NiCI2·6H20, 36 mg of
Na2Mo04·2H20, 15 mg of Na2W04, 15 mg of Na2Se03·5H20, and 10 mL of
25%HCL
The medium was flushed with N2 for 15 min and sterilized by auto-
claving at 120°C for 20 min at 1.2 bar. Sterile solutions of salts, phosphate,
yeast extract and trace elements were added separately after sterilization.
Prior to inoculation, the pH was adjusted to 7.2 with 1 M KOH or 1 M HCL

Preparation of Hydrolysate
As carbon and energy source, glucose or paper sludge hydrolysate
was used. The paper sludge (generally 45% carbohydrate content based
on dry matter; dry matter content is 60%) originated from Dunapack Pulp
and Paper Mill, Dunapack Paper and Packagings, Hungary. Large-scale
hydrolysis was performed at a substrate concentration of 4% (w Iv) in a
pH- and temperature-controlled 31-L Braun fermentor. The paper sludge
was suspended in water and the pH was set to 4.8 by the addition of
concentrated H 2S04 , The slurry was sterilized at 121°C for 20 min. After
cooling the mixture to 50°C, 15 filter paper units (FPU) I g of dry matter
(DM) of Celluclast 1.5L and 15 IU I g of DM of Novozym 188 were added
to hydrolyze the cellulose. The final paper sludge hydrolysate contained
12.8 giL of glucose and 2.4 giL of xylose after 48 h of hydrolysis. The
hydrolysate was pretreated by mixing with 3-morpholinopropane sul-
fonic acid (MOPS) (10 mM) or phosphate buffer (the concentrations of
phosphates were the same as used in the culture media) to a final pH of
7.2, and cleared by centrifugation (10 min, 13,200g) prior to the addition
of other components.

Fermentation Assays
Cultures were grown in duplicate in anaerobic 100-mL serum bottles
with 30-mL volumes. Active cultures for inoculation were prepared by
growing T. elfii and C. saccharolyticus for 72 h at 65°C and 24 h at 70°C,
respectively.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
560 Kadar et at.
Paper sludge hydrolysate was diluted to initial glucose and xylose
concentrations of 9.2 and 1.9 giL, respectively. The flasks were incubated
under different temperatures: 65°C for T. elfii and 70°C for C. saccharolyticus.
Flasks were periodically sampled during fermentation for determina-
tion of hydrogen production, soluble sugar and organic acid contents,
optical density, and pH.
Analytical Procedures
The optical densities of the cultures were determined at 5BO nm on an
Ultrospec 2000 Spectrophotometer.
Glucose, xylose, and organic acids were analyzed by high-perfor-
mance liquid chromatography with differential refractometry detection
and a Shodex ionpack KCBll column at BO°C. H 2S04 (3 mM ) was used as
eluent at a flow rate of 1 mL/min. The samples were 1:1 diluted in 1 M
H 2S04 with 250 mM propionic acid, as the internal standard, and filtered
through a 0.45-!!m filter prior to injection.
Hydrogen was determined by gas chromatography with thermal con-
ductivity detection and an RVS MolSieve SA, 60lBO mesh, 3 m x 0.125-in.
column at 50°C. The temperature of the detector and the injector were tOO
and BO°C, respectively. N2 was used as carrier gas.
Statistical analyses were done using MINITAB software.

Results
Effect of Phosphate and MOPS Buffer
In the paper sludge, the main mineral components are typically kaolin
and CaC03 (used for coating and/ or as filler) and Ti02(used for whitening)
(6). When paper sludge hydrolysate was mixed with the culture medium
used for growth of the microorganisms, probably precipitation of Ca-phos-
phate occurred. To avoid this precipitation, phosphate buffer was added
first and the pH was adjusted to 7.2. After centrifugation, the other medium
components were added to the clarified supernatant and no further pre-
cipitation occurred. In a control experiment, MOPS was used to replace the
phosphate buffer.
As can be seen from Table 1, hydrogen was produced from paper
sludge hydrolysate by T. elfii and C. saccharolyticus. Replacement of phos-
phate by MOPS buffer had no effect on glucose consumption or on hydro-
gen and acetate production. During fermentation, the pH decreased
continuously to 5.2. Although both T. elfii and C. saccharolyticus seemed to
grow well without supplementation with phosphate, in further experi-
ments the hydrolysate was always pretreated with phosphate buffer and
the precipitate was removed prior to fermentation.
Effect of Medium Composition
The medium components required for optimal hydrogen production
by T. elfii and C. saccharolyticus on paper sludge hydrolysate were studied.

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


H Production from Paper Sludge Hydrolysate 561
Table 1
Effect of Phosphate and MOPS Buffer on Glucose Consumption,
Hydrogen, and Acetate Production by T. elfii and C. saccharolyticus
After Growth on Paper Sludge Hydrolysate"

T. elfii C. saccharolyticus
Glucose H2 Acetate Glucose H2 Acetate
Medium (mM) (mM) (mM) (mM) (mM) (mM)

Phosphate 23.9 27.2 20.9 5.4 21.6 12.7


MOPS 20.8 27.1 19.8 3.1 20.5 11.5
•Samples were taken after 92 and 88 h, respectively. During fermentation, only glucose,
hydrogen, and acetate were measured.

Table 2
Scheme for Addition of Salts, Yeast Extract, and Trace Elements
to Media Used for Growth of T. elfii and C. saccharolyticus'

Yeast extract Saltsb Trace elements

1 glc pos cont + + +


2 psh pos cont + + +
3 - t.e. + +
4 - salts + +
5 - salts, t.e. +
6 -y.e. + +
7 - y.e., t.e. +
8 - y.e., salts +
9 - y.e., salts, t.e.
• gIc pos cont, positive control on glucose; psh pos cont, positive con-
trol on paper sludge hydrolysate; t.e., trace elements; y.e., yeast extract;
2-9, paper sludge hydrolysate as carbon and energy source; +, present;
-, absent.
b see Materials and Methods (cysteine-HCl, resazurin, and Na2C03

were always present).

The composition of the media is schematically presented in Table 2. To


assess the effect of supplementation on fermentation by the microorgan-
isms employed, the concentrations of carbohydrates, hydrogen, acetate,
and lactate were measured. Fermentations on paper sludge hydrolysate
with initial concentrations of 8.9 giL of glucose and 1.8 giL of xylose were
compared with a fermentation on glucose medium (9.3 giL of glucose).
Figure 1 shows the results of fermentations with T. elfii at 65°C.
Hydrogen production by T. elfii on paper sludge hydrolysate supplemented
with all media components was similar to hydrogen production by the
positive control. No significant difference in glucose consumption and
acetate production was observed. However, some lactate was produced
during fermentation of paper sludge hydrolysate. Omission of trace ele-
ments from the paper sludge hydrolysate medium had no significant effect.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
562 Kadar et al.

35

o
25

20
::s:
E 15

.-
- -
10
...
I- I- i--

. • • 1l-
5 - - - fo-

n
I--

o
Sic pos psh pos - Le. - ailS - all . le. - ye. - y.c.. I.e - y.e.. - y.c..
conI conI sailS salIs. le.

• 112 produced 0 glu acelale produced • lac\alc produced

Fig. 1. Effect of different medium components on glucose consumption and hydro-


gen, acetate, and lactate production during fermentation of paper sludge hydrolysate
by T. elfii after 72 h at 65°C; see Table 2 for abbreviations.

In the absence of all salts (except cysteine-HCl and resazurin) and plus or
minus trace elements, hydrogen production was lower. This is not surpris-
ing since T. elfii is known as a halophilic bacterium. However, it is surpris-
ing that hydrogen production decreased while glucose consumption and
acetate production remained the same. Other experiments showed that
when either NaCl or the other salts were added to this medium, hydrogen
production was partly restored but still lower compared to the complete
paper sludge hydrolysate medium (results not shown). The omission of
yeast extract had the greatest effect. Hydrogen production in cultures with-
out yeast extract was much lower or absent when salts were also omitted.
In the latter cultures, glucose concentration decreased even though no
products were detected. At present, it is unclear whether this is owing to
Maillard reactions or other nonspecific reactions. Maillard reactions have
been observed before but are hardly reproducible and difficult to quantify.
Lactate was only produced in paper sludge hydrolysate based medium
enriched with yeast extract.
The results were supported by statistical calculations (Fig. 2). Supple-
mentation of the medium with either yeast extract or salts had the greatest
effect, and trace elements were practically without any influence. The
effects of these variables on hydrogen production were independent.
A similar experiment was performed with C. saccharolyticus. The
results are shown in Fig. 3. Hydrogen production in the control culture on
glucose was comparable with hydrogen production on glucose by T. elfii.
However, the production of hydrogen by C. saccharolyticus on paper sludge
hydrolysate supplemented with all medium components was much lower
than the positive control on glucose. In contrast to T. elfti, both glucose and

Applied Biochemistry and Biotechnology Vol. 105-108,2003


H Production from Paper Sludge Hydrolysate 563

19

16

------------ ------------ ------------ ------------- ---.-==~---"""'""' ..

10

salts yeast extract trace elements

Fig. 2. Statistical analysis of effect of medium components on hydrogen production


from paper sludge hydrolysate by T. elfii.

35.0

30.0

25.0

20.0
:E
E
15.0 10.-

10.0 - f- - f- - f-

5.0 - - - - - - - -
f ~ tf
f-- f-

0.0 I ~I ~I ~I ~
'.
:I l:r
glc pos xyl po psb pos - c. - salts - salts, - y.. - y.c ., - y.c .. - y . ..
coni conI COOl l.c . I.c. . Its saJts.
I.c .

• H2 produced I!ll glucose consumed ill xylose consumed


• acetate produced • lactate produced

Fig. 3. Effect of different medium components on glucose and xylose consumption


and hydrogen, acetate, and lactate production during fermentation of paper sludge
hydrolysate by C. saccharolyticus after 42 h at 70°C. xyl pos cont; positive control on
xylose; see Table 2 for abbreviations.

xylose, which are present in small amounts in paper sludge hydrolysate,


were consumed by C. saccharolyticus. The control culture on xylose pro-
duced less hydrogen than the control culture on glucose. This could be
owing to a lower consumption of xylose in this experiment and to the
hydrogen yield, which is most probably lower per mol of consumed Cs
sugar (xylose) than per mol of consumed C6 sugar (glucose). Interestingly,
the omission of medium components did not further decrease hydrogen
production on paper sludge hydrolysate. Thus C. saccharolyticus, growing
on paper sludge hydrolysate, was able to produce a significant amount of
hydrogen in the absence of yeast extract. During fermentation, in the pres-
ence of yeast extract, lactate production was observed (Fig. 3).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
564 Kadar et al.
16
14 •
12
10
::E 8
Ei
6
4
2
0
----~$~----&

• •
0 20 40 60 80 100
time (h)

Fig. 4. Carbohydrate consumption (O, glucose, 0, xylose) during fermentation of


paper sludge hydrolysate by T. elfii (black symbols) and C. saccharolyticus (open
symbols).

In contrast to T. elfii cultures on paper sludge hydrolysate, glucose


consumption by C. saccharolyticus was much lower (Fig. 4). Furthermore, in
C. saccharolyticus cultures the utilization of xylose was preferred over glu-
cose. T. elfii only consumed xylose at the end of the exponential phase and
then only to a very small extent. However, especially for T. elfii, interpre-
tation of the results was complicated because in the absence of hydrogen
production, probably also in the absence of growth, xylose concentrations
were increasing. It is possible that because of increased activity of xylanase
or {b }-xylosidase present in the Celluclast preparation, hydrolysis of hemi-
cellulose and derived oligomers continued, owing to the pH and high tem-
perature during the fermentation.

Discussion
There is an abundance (50,000 t annually in Hungary) of paper sludge
(industrial byproduct) with high carbohydrate content, which, after enzy-
matic hydrolysis, could be a substrate for hydrogen fermentation. T. elfii
and C. saccharolyticus have been identified as extremely thermophilic micro-
organisms able to convert sugars to hydrogen, carbon dioxide, and organic
acids. Therefore, the growth of these two bacteria on paper sludge hydroly-
sate was studied.
Both T. elfii and C. saccharolyticus could grow and produce hydrogen
on paper sludge hydrolysate, but the levels of hydrogen production and
the medium requirements for optimal hydrogen production differed.
T. elfii produced a high amount of hydrogen, but needed yeast extract to do
so. Moreover, because it is a halophilic bacterium, 1% NaCl was required.
C. saccharolyticus seemed to be less dependent on additional medium com-
ponents, since the hydrogen production was not stimulated by the addition
of yeast extract, salts, or trace elements to paper sludge hydrolysate. In
contrast to T. elfii, hydrogen production by C. saccharolyticus seemed to be

Applied Biochemistry and Biotechnology Vol. 705-708,2003


H Production from Paper Sludge Hydrolysate 565
inhibited by a component present in the paper sludge hydrolysate. Both,
organisms produced not only acetate but also lactate, but only in a nitro-
gen-rich medium containing yeast extract.
The theoretical yield of hydrogen from glucose was 4 mol of hydro-
gen/mol of sugar. Approximately 46 and 48% of the maximum hydrogen
yield was obtained with T. elfii growing on glucose and on paper sludge
hydrolysate (based on glucose consumption), respectively. Since the theo-
retical hydrogen yield on xylose, a Cs sugar, has not been determined yet,
it is hard to calculate the hydrogen yields for C. saccharolyticus, which uti-
lizes both glucose and xylose. The hydrogen yield found with T. elfii on
glucose was significantly lower than previously reported (7). This could be
owing to the accumulated hydrogen in the closed bottles, which inhibits
further production. In future studies, fermentations on a larger scale under
controlled conditions and with N 2 sparging will allow more accurate deter-
minations of hydrogen yields.
We previously observed that T. elfii is able to utilize amino acids for
hydrogen production (results not shown). This phenomenon was also
checked with C. saccharolyticus, but here no hydrogen production from
yeast extract was observed. The utilization of yeast extract for hydrogen
production also hampered the determination of an accurate mass balance.
Besides this effect on obscuring the metabolic route, the dependence on
yeast extract is not in favor of commercial exploitation of T. elfii.
According to the results we have shown, it can be concluded that both
microorganisms are able to produce hydrogen from paper sludge hydroly-
sate. The dependence on additional nutrients was higher for T. elfii than for
C. saccharolyticus. C. saccharolyticus seemed to be inhibited by paper sludge
hydrolysate. It was clearly demonstrated that utilization of xylose by
C. saccharolyticus is preferred over glucose. T. elfii only consumed xylose at
the end of the exponential phase and to a very small extent. Optimization
of hydrogen production by extreme thermophiles from paper sludge will
be further investigated in future studies.

Acknowledgments
This study was financially supported by the Commission of the Euro-
pean Communities, Quality of Life and Management of Living Resources
(project no. QLK5-1999-01267), the Netherlands Organisation for Interna-
tional Cooperation in Higher Education (Huygens Programme), and the
Dutch EET Programme.

References
1. Claassen, P. A. M., van Lier, J. B., Lopez Contreras, A. M., van Niel, E. W. J., Sijtsma,
L., Starns, A. J. M., de Vries, S. 5., and Weusthuis, R. A. (1999), Appl. Microbiol.
Biotechnol. 52,741-755.
2. Dunapack Paper and Packagings Ltd. (2000), Environmental Report 2000, Mikl6s
Galli, Dunapack Ltd., Hungary.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


566 Kadar et al.
3. Ravot, G., Ollivier, B., Fardeau, M. 1., Patel, B. K. c., Andrews K. T., Magot, M., and
Garcia J. 1. (1996), Appl. Environ. Microbial. 62,2657-2659.
4. Ravot, G., Magot, M., Fardeau, M. 1., Prensier, G., Egan, A., Garcia, J. 1., and Ollivier,
B. (1995), Int. J. Syst. Bacterial. 45, 308-314.
5. Rainey, F. A., Donnison, A. M., Janssen, P. H., Saul, D., Rodrigo, A., Bergquist, P. 1.,
et al. (1994), FEMS Microbial. Lett. 120,263-266.
6. Lynd,L. R., Lyford, K., South, C. R., and Levenson, K. (2001), TAPPI J. 84,50.
7. van Niel, E. W. J., Budde, M.A.W., de Haas, G. G., van der Wal, F. J., Claassen, P. A. M.,
and Starns, A. J. M. (2002), Int. J. Hydrogen Energy 27,1391-1398

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0567/$20.00

Single-Stage Anaerobic Codigestion


for Mixture Wastes of Simulated Korean
Food Waste and Waste Activated Sludge

NAM Hyo HEO, l SOON CHUL PARK/,l JIN SUK LEE, l


Ho KANG/ AND DON HEE PARK3
1Biomass
Research Team, Korea Institute of Energy Research,
71-2 Jangdong, Yusongku, Daejon 305-343, Korea,
E-mail: bmscpark@kier.re.kr;
2Department of Environmental Engineering, Chungnam National University,
220 Gung-dong, Yuseong-gu, Daejeon 305-764, Korea; and
3Department of Chemical Engineering, Chonnam National University,
330 Yongbong-dong, Puk-gu, Gwangju 500-757, Korea

Abstract
Korean food waste was treated with a single-stage anaerobic codigester
(SSAD) using waste activated sludge (WAS) generated from a municipal
wastewater treatment plant. The stability and performance of the system
was analyzed. The C/N ratio was improved with increasing food waste
fraction of feed mixture. The pH, alkalinity, and free ammonia nitrogen
concentration were the parameters used to evaluate the digester's stability.
The experimentally determined values of the parameters indicated that there
were no methane inhibitions in the digester. Digester performance was
determined by measuring the total chemical oxygen demand TCOD), volate
solids (VS) removal, methane content in biogas, methane production rate
(MPR), and specific methane productivity. Methane content in biogas and
MPR were significantly dependent on hydraulic retention time (HRT) and
ratio of food waste to WAS. The methane content in biogas decreased at
shorter HRT or higher organic loading rate (OLR) with increased food waste
fraction. Concerning the performance of the codigester, the optimum oper-
ating condition of the SSAD was found to be at an HRT of 10 d with a feed
mixture ratio of 50% food waste and 50% WAS. A TCOD removal efficiency
of 53.6% and a VS removal efficiency of 53.7% were obtained at an OLR of
5.96kgofTCOD/(m3·d)and3.14kgofVS/(m3·d),respectively.Amaximum
MPR of 1.15 m3 CH4/(m3·d) and an SMP of 0.37 m 3 CH4/kg of VSfeed were
obtained at an HRT of 10 d with a methane content of 63%.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 567 Vol. lOS-lOB, 2003


568 Heo et al.

Index Entries: Alkalinity; ammonia nitrogen; anaerobic codigestion; food


waste; methane; volatile fatty acids; waste activated sludge.

Introduction
In 2000, Korea generated about 46,438 tid of municipal solid waste
(MSW). Korean food waste (KFW) constitutes approx 25% of MSW (1). Most
of the food waste is disposed by landfill (45.4%) and incineration (9.5%), and
the rest is recycled as feedstuff, composting, or anaerobic digestion (1). Food
waste is difficult to treat or recycle because it contains high levels of sodium
salt and moisture and is mixed with other collected wastes (2,3). In 2005,
landfilling of food waste, which causes various problems such as a foul odor
and ground and surface contamination by leachate, will be prohibited.
Therefore, for KFW, anaerobic digestion has been considered as a fea-
sible alternative to reduce waste volume and recover renewable energy as
methane. Presorted KFW with 15-30% of total solid (TS) has a high volatile
solid (VS) content (88-92% of TS), and the methane potential of KFW is
estimated to be about 0.44-0.47 m 3 of CH4/kg ofVSfeed (4,5). A characteristic
of food waste that contains highly soluble organics is that it is converted
rapidly to volatile fatty acids (VFA) in the early stage of digestion. A drastic
drop in pH corresponding to VFA accumulation may lead to irreversible
acidification of the digester, owing to poor buffering capacity (4-6). There-
fore, anaerobic digestion is often a two-phase system in order to separate
acid and methane-forming phases (4,5,7).
One of the most interesting options for improving biogas yield and
reducing the volume of organic solid wastes is anaerobic codigestion, i.e.,
codigestion of the organic fraction of municipal solid wastes (OFMSW)
together with sewage sludge (8-12). Cecchi et a1. (13) and Mata-Alvarez et a1.
(14) described the codigestion of OFMSW with sewage sludge in existing
digesters and the advantages of codigestion. The advantages include dilu-
tion of potential toxic compounds coming from wastewater or cosubstrate,
improved balance of nutrients, synergistic effects of microorganisms,
increased organic loading of biodegradable matter, and better methane gas
yields per unit of digester volume. In addition, using existing anaerobic
digesters to treat OFMSW together with sewage sludge in municipal waste-
water treatment plants could reduce capital and operating costs (15-19).
In Korean municipal wastewater treatment plants, the existing anaero-
bic digesters conventionally are two-stage anaerobic digesters operated at
mesophilic temperature (33-37°C); they do not operate properly because
the actual loading rate is much less than the design loading rate. Therefore,
using the extra capacity by adding OFMSW to existing municipal waste-
water treatment plants anaerobic digesters may be a good alternative,
especially for treating KFW and it may be possible to enhance the operating
efficiency of anaerobic digesters. The present study was performed in a
single-stage anaerobic codigester (SSAD) at 35°C, with feed mixture ratios
of simulated KFW (SKFW):WAS of 10:90, 30:70, and 50:50. The objectives

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Use of SSAD for Mixture Wastes of SKFWaste and WAS 569
of this study were to evaluate the feasibility of anaerobic codigestion of
KFW and waste activated sledge (WAS) and to investigate the stability and
performance of the codigester.

Materials and Methods


Experimental Apparatus
This study used an SSAD-a semicontinuously fed and mixed reac-
tor-made of Plexiglas. The effective volume of the digester was 3.5 Land
four SSADs were arranged in a water bath. Each digester was mechani-
cally stirred at 80 rpm with stainless steel paddles on a central shaft by an
electric motor with a speed controller. The digestion system was operated
with a fill-and-draw method. The water bath was maintained at 35 D C by
circulating water through a water jacket by a temperature-controlled cir-
culator (Haake, Karlsruhe, Germany). The biogas produced was collected
in Tedlar bags, and the volume was measured three times per week by a
wet gas meter (Sinagawa, Tokyo, Japan).
Feedstock and Feed Mixtures
To simulate KFW, a traditional Korean food called Bibimbab-SKFW,
which has a similar composition to the food waste, was used. It had a TS of
15-20% and a moisture content of 80-85%. Bibimbab consists of boiled rice
(10-15%), vegetables (65-70%), and meat and eggs (15-20%). The SKFW
was broken down to 2-4 mm using a cook mixer, and the final concentra-
tion of TS was controlled to 8 ± 0.5% by adding tap water. A considerable
fraction of the soluble organics of SKFW was solubilized reducing the par-
ticle size. WAS was collected from the secondary clarifier in municipal
wastewater treatment plants located in Daejon, Korea. The WAS sample
was thickened using a centrifuge and adjusted to TS of 3%, which is typical
of thickened sludge (TS of 2 to 3%) from gravity thickener in municipal
wastewater treatment plants. The feed mixture ratios of SKFW:WAS were
10:90,30:70, and 50:50 on the basis of VS contents of two solid wastes,
respectively. Each feed mixture was made once per week and stored in a
refrigerator at 4 DC; however, partial hydrolysis and acidogenesis of the
feed mixture was observed. Table 1 gaves the chemical characteristics and
elemental analysis of the inoculum used as the initial seed, SKFW, WAS,
and three feed mixtures.
Operating Conditions
The inoculum was used for a quick start-up without preacclimation to
feed mixtures, in which the microorgnisms were well adapted to the mix-
ture waste of KFW with thickened sludge (WAS plus primary sludge) was
obtained from the mesophilic (35 DC) SSAD at the Korea Institute of Energy
Research. All experiments were conducted with three feed mixtures as a
function of hydraulic retention time (HRT). The start-up periods of each
feed mixture were 50 d, and the HRT of four digesters was maintained from
10 d to 13, 16, and 20 d.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


~
:g
[
0;)

~ Table 1
3 Chemical Characteristics and Elemental Analysis
ii;' of Inoculum, SKFW, WAS, and Three Feed Mixtures·
q-
'<:
~
::J Feed Mixtures (g SKFW : g WAS)
Q..
0;)
0' Inoculum SKFW WAS 10:90 30:70 50:50
~
::J TCOD (giL) 23.8 100 (10) 36 (2.0) 44.3 (1.0) 45.4 (1.39) 58.5 (2.17)
o
~ SCOD (giL) 0.31 43.0 (3.8) 0.3 (0.2) 2.77 (0.51) 4.01 (0.93) 10.8 (0.77)
'<:
TS (giL) 21.7 80.0 (5.0) 30 (1.0) 31.6 (0.40) 33.1 (0.70) 37.6 (1.10)
VS (giL) 14.2 75.0 (5.0) 18.5 (0.5) 23.7 (0.60) 26.0 (0.60) 31.4 (0.90)
U1 pH 7.4 4.50 (0.5) 6.8 (2.0) 6.08 (0.15) 5.26 (0.24) 4.09 (0.26)
o Alkalinity as CaCO (giL) 3.65 0.15 (0.1) 0.8 (0.2) 0.95 (0.13) 0.51 (0.13)
" NH4+ (giL) 0.81 0.29 (0.1) 0.03 (0.01) 0.16 (0.04) 0.12 (0.04) 0.16 (0.02)
VFA (g/L)b 0 0.75 (0.15) 0 0.37 (0.13) 0.53 (0.32) 0.68 (0.16)
Elemental analysis (% TS)
C 45.76 30.05 30.92 34.02 36.43
H 6.68 5.10 3.4 4.22 5.21
0 38.80 20.94 24.55 25.55 29.09
N 2.84 5.03 4.44 4.34 4.13
S 0.24 0.97 0.94 0.89 0.78
C/N ratio 16.11 5.97 6.97 7.84 8.82
~
:-
• Numbers in parenthesis are the SD about the mean.
b VFAs: C 2-C 6 •

-~
~
to..>
§
Use of SSAO for Mixture Wastes of SKFWaste and WAS 571

Analytical Methods
The pH and temperature were monitored. Chemical oxygen demand
(COD), TS, VS, alkalinity, and NH/ of the samples were determined
according to standard methods (APHA, AWW A & WEF, 1998) and were
analyzed three times per wk during the start-up. In particular, the sample
for the analysis of soluble chemical oxygen demand (SCOD), NH4 +, and
VFA was prepared by filtration using a 0.45-f.-tm cellulose acetate mem-
brane after centrifugating at 15,000 rpm for 15 min. Elemental composi-
tion of the sample was analyzed with an elemental analyzer (CHN-1000;
Leco) and a sulfur analyzer (SC-432DR; Leco).
The percentage of methane and carbon dioxide was analyzed using a
gas chromatograph (HP-5890A GC) eqUipped with a thermal conductivity
detector and a 6-ft stainless column packed with Hayesep Q (80/100 mesh).
The injection and detector temperatures were 120 and 150°C, respectively,
and the column oven operated isothermally at 60°C. Helium was used as
the carrier gas at a flow rate of 30 mL/min. The concentration of VFA was
determined using the same gas chromatograph equipped with a flame
ionization detector and a capillary column (25 m x 0.2 mm x 0.33-f.-tm;
Hewlett Packard-FFAP), with helium as the carrier gas (flow rate of 0.8
mL/min). The injection and detector temperatures were 200 and 220°C,
respectively. The initial temperature of the column oven was 80°C and
increased gradually by 13°C/min, reaching a final temperature of 180°C.

Results and Discussion


Typical WAS has the following composition: 33% carbohydrates, 28%
protein, and 28% lipid (20) and is characterized by a low C/N ratio of 5.9
(21). On the other hand, food waste is reported to have the following com-
position: 78% carbohydrates, 6.5% protein, and 0.6% lipid (20). The anaero-
bic codigestion of OFMSW with sewage sludge result in a better balance of
nutrients and higher C/N ratio in organic wastes, which are the prerequi-
sites for a stable process performance. In other words, the high concentra-
tion of macro- and micronutrients in the sewage sludge could compensate
the lack of nutrients in OFMSW, and the low C/N ratio of sewage sludge
could be increased by the addition of OFMSW. Therefore, the addition of
cosubstrate may improve the process performance and the methane yield
owing to the better nutrient balance and C/N ratio.
As shown in Table 1, the C/N ratio of WAS used was similar to the
typical value of 5.9, and the C/N ratio of SKFW was nearly three times
higher than that of WAS. The C/N ratio of the three feed mixtures used as
the substrate was improved with increasing SKFW fraction. During the
start-up of each feed mixture, 15-20 d were required to reach the first steady-
state condition. Indicators of digester stability, such as pH, alkalinity, NH4+
and VFA, were monitored. The COD and VS reduction, methane content of
the biogas, MPR, and SMP were investigated for the operating performance
oftheSSAD.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
572 Heo et al.
8 s
7
III 10'90 5050 A III 10:90 • 30:70 • 50 50 B
>. 4
~ 6 >.
01

~
'0

1:
..... 3
0 en
>
I-
Jf2
ci
....l
0

0
10 13 16 20 10 13 16 20
HRT, day. HRT, days

Fig. 1. OLRs of SSAD at different HRTs: (A) TeOD basis; (B) VS basis.

40 20
III 10' 90 • 30:70 .. 50:50 A III 10'90 • 30'70 • 50"50 B
30 IS
~
ci ~
§ 20 ell
> 10
.,
C
...:::>
C :::>
E
E 10 ~ 5
~

0 0
10 13 16 20 10 13 16 20
HRT, days HRT, clay

Fig. 2. Effluent TeOD and VS concentrations from SSAD at different HRTs: (A)
TeOD; (B) VS.

Figure 1 presents the organic loading rate (OLR) expressed as total COD
(TCOD) and VS on three feed mixtures as the function of HRT. The OLR was
gradually increased as the HRT became shorter or the SKFW fraction of the
feed mixture increased. Figure 2 shows the TCOD and VS concentration of
the effluent; and the levels were nearly similar without the effect of HRT.
However, the effluent VS concentration in the mixture ratio of 50:50 feed was
the lowest compared with the other feed mixture, in spite of the higher OLR
and shorter HRT. This was owing to the high biodegradability of SKFW.

Stability of SSAD System


In the anaerobic digestion process, the parameters that indicate sta-
bility are pH, alkalinity, NH4+, and VFA concentration in the digester. The
pH of the methane bioreactor is usually controlled to a set point within the
range of 6.5-7.5 (22). During anaerobic treatment of organic solid wastes,
a major pH drop in the digester is owing to the accumulation of VFA
produced from the hydrolysis/ acidogenesis of organic wastes, especially

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Use of SSAD for Mixture Wastes of SKFWaste and WAS 573
76 . -__________________ ~
5.----__________________

13 10.90 11 30:70 • 5050 A


7.5
~ 46

74 0-
~ 42
::r: 7.3
U
Cl. ~
.~ 3.8
72
"
~
-'" 3.4
7. 1
~

7 3 .......................
10 13 16 20 10 13 16 20
HRT, doy. HRT. dsJ

Fig. 3. pHs and alkalinities of SSAD at different HRTs: (A) pH; (B) alkalinity.

15 50
I!l 10:90 . 30.70 . 50:50 A Ell 10:90 . 30.70 . 5050 B
1.2 40
..J
~ Ob
E 30
C 0.9
.2 .;
E 'c0

'"" ..E
0.6 E 20
0
E
E ~
-< 0.3 "- to

0 0
10 13 16 20 10 13 16 20
IIRT, doy. IIRT, days

Fig. 4. Ammonium ion and FAN concentrations of SSAD at different HRTs: (A)
ammonium ion; (B) FAN.

during an overloaded condition. Figure 3 shows the digester pH and


alkalinity as a function of HRT with three feed mixtures. It is clear that pH
and alkalinity of the SSAD proportionally increase as the HRT becomes
longer or the OLR at an identical HRT increases. In particular, the alkalin-
ity was compared among three feed mixtures; the maximum alkalinity
was 4.91 giL as CaC03 for a feed mixture of 50:50. Even at a short HRT of
10 d with an OLR of 5.96 kg of TCOD I (m3·d), an alkalinity of 4.37 giL as
CaC03 was maintained. Lay et a1. (23) reported that the methane produc-
tion rate in a high-solid sludge digestion was good at a pH range of 6.6-
7.8, and the digester may fail if the pH was <6.1 or >8.3. During all the
experiments, the digester pH was maintained within the stable range of
7.2-7.5 with sufficient alkalinity ranging from 3.1 to 4.91 giL as CaC03 •
In addition, the accumulation of VF A was not observed in all experiments
after reaching the first steady state.
Figure 4 shows the ammonium ion (NH4+) and free ammonia nitrogen
([FNA], NH3) concentration produced from the digester as a function of
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
574 Heo et al.

7 0 r - - -- - - - - - - - - - - -__~ 70 r-----------------~
60 EI 1090 D 30.70 • 50,50 A 60 III 10:90 • 30:70 • 50:50 B
50
!jo!
e; 40
~~ 30
C
8I- 20

10

o
10 13 16 20 10 13 16 20
HRT, cia" II RT, dB,..

Fig. 5. TCOD and VS removals of SSAD at different HRTs: (A) TCOD; (B) VS.

HRT on three feed mixtures. The ammonia nitrogen, which is formed by the
anaerobic biodegradation of organic nitrogenous compounds such as pro-
teins or amino acids, significantly affects methanogenic activity. The inor-
ganic nitrogen produced, the ammonium ion, and FAN exist in two forms
in the liquid phase. In particular, because the FAN concentration depends
on pH, it is important to control the pH of an operating digester. Braun
et al. (24) and Bhattacharya and Parkin (25) suggested that the maximum
tolerable FAN concentration is 55-80 mg/L for a stable digestion. Lay et al.
(23) investigated the influence of ammonia nitrogen on methanogenic
activity in the wide pH range of 6.5-8.5. The methanogenic activity
decreased with increasing NH/ concentration and dropped 10% at a con-
centration of 1670-3720 mg/L, 50% at 4090-5550 mg/L, and 0 at 5880-6600
mg/L. In the present study, at an HRT of 20 d with a 50:50 feed mixture, the
ammonium ion concentration was the highest at 1160 mg/L, and the FAN
concentration was 37 mg/L as a function of pH 7.5 at 35°C. Therefore, the
FAN concentration was estimated to be below the concentration that inhib-
its the methanogenic activity mentioned in the literature. The FAN concen-
tration was calculated using Eq.1, presented by Kayhanian (26) as a function
of pH at 35°C. Judging from the stability parameters of an SSAD, there was
no inhibition of methane production owing to pH drop, insufficient alka-
linity, or inhibition by FAN for all the experiments.
TAN x (Ka + lHJ)
NH3(mg/L) = (Ka + [H]) + 1 (1)
in which NH3 is the free ammonia nitrogen concentration (mg/L), TAN is
the total ammonia concentration in solution including ammonium and free
ammonia (mg/L), [H] is the hydrogen ion concentration (lO-pH), Ka is the
temperature-dependent dissociation constant (1.097 x 10-9 at 35°C).
Performance of SSAD System
Figure 5 presents the TCOD and VS removal efficiencies of SSAD as
a function of HRT on three feed mixtures. For the TCOD and VS removal
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Use of SSAD for Mixture Wastes of SKFWaste and WAS 575

0.9
~

...8x 0.8
....'"c:
~
c: 0.7
8
GI
c:
(\I
oSGI 0.6
~

0.5
0 30 60 90 120 150
Operating time, days

-+- 20 day ____ 16 day ......- 13 day _____ 10 day

Fig. 6. Methane contents in biogas of SSAD at different HRTs.

efficiencies, the trend was similar and slightly increased as the HRT
became longer for the same feed mixture. However, the magnitude of the
TCOD and VS removal efficiencies was remarkably different among
the three feed mixtures at the same HRT. Therefore, it became clear that
the TCOD and VS removal efficiencies were more affected by the SKFW
fraction of the feed mixture than the influence of the HRT; that is, their
removal efficiencies depended on the biodegradability of the feed mix-
ture. Among three feed mixtures, the TCOD and VS removal efficiencies
were the highest at a feed mixture of 50:50. The maximum TCOD removal
of 57.9% and VS removal of 56.3% were achieved when the digester was
operated at an HRT of 20 d with an OLR of 2.98 kg of TCOD / (m3·d) and
1.61 kg ofVS/(m3 ·d), respectively. Even at a shorter HRT of 10 d with an
OLR of 5.96 kg of TCOD/(m3·d) and 3.14 kg of VS/(m3 ·d), the TCOD
removal was 53.6% and VS removal was 53.7%. These results are compa-
rable to previous research for the anaerobic codigestion of OFMSW and
sewage sludge. Mata-Alvarez et al. (14) investigated the mesophilic (35°C)
anaerobic codigestion of the mixture of 50% OFMSW and 50% sewage
sludge. They reported that VS removal was 57% at an HRT of 14.5 d with
an OLR of 2.8 kg of VS/(m3·d). Del Borghi et al. (11) investigated the
thermophilic (55°C) anaerobic codigestion of a mixture of 50% OFMSW
and 50% sewage sludge. The VS removal was 64% at an HRT of 12 d with
an OLR of 4.0 kg of VS/(m3 ·d).
Figure 6 shows the methane percentage in the biogas. The methane
content of biogas produced from all digesters was higher than the usual
values of 60% and had significant differences among three feed mixtures.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
576 Heo et al.
1.5 ,.-----_ _ _ _ _ _ _ _----, 05 ,.-----_ _ _ _ _ _ _ _----,

a 10'90 III 30:70 [] 50:50 A .. 10:90 • 30 '70 II 50:50 B


12 04
J
'"
>
Cl 03
:)
U
E 0.6 E 02
...
:i c:
::;
::;: 0.3 '" 0 1

o o
10 13 16 20

IIRT. days HRT, days

Fig. 7. MPRs and SMPs of SSAD at different HRTs: (A) MPR; (B) SMP.

As shown in Fig. 6, it became clear that the methane content of biogas


grad ually decreased as the HRT became shorter or the OLRs increased. The
methane content of the digester operated at different HRTs with a mixture
ratio of 50:50 was in the range of 63-70%, higher than those of 60 and 50%
obtained by Mata-Alvarez et aL (14) and Del Borghi et aL (11).
Figure 7 presents the MPR and SMP at different HRTs with three feed
mixtures. Two parameters increased with increasing SKFW fraction of
the feed mixture owing to the abundant carbohydrate of food waste. MPR
was expressed as the methane volume of biogas produced per volume of
digester per d. As shown in Fig. 7A, the MPR proportionally increased as
the HRT became shorter or the OLR increased; therefore, the MPR was
directly related to the OLR applied to the reactor and thus reflects the
COD and VS removal efficiency. The maximum MPRs of the three feed
mixtures were 0.48 of 10:90, 0.6 of 30:70 and 1.15 m 3 ofCH4/(m3 ·d) of 50:50
ratio at an HRT of 10 d with an OLR ranging from 2.37 to 3.14 kg of VS/
(m3 ·d). The SMP is a good parameter to estimate the biodegradability of
a substrate added to the digester. The SMP (m3 of CH4/kg of VS feed ) was
expressed as the methane volume ofbiogas produced per unit mass ofVS
added to the digester (Fig. 7B). HRT had little effect on SMP of each feed
mixture and decreased above an HRT of 13 d. We have already estimated
the SMP of three feed mixtures by the batch biochemical methane poten-
tial test (27). The SMPs were 0.20, 0.23 and 0.29 m 3 of CH4/kg of VS feed
at the mixture ratios of 10:90, 30:70, and 50:50, respectively. The SMPs
obtained in the present study on feed mixtures of 10:90 (0.19-0.22 m 3 of
CH4/kg ofVS feed ) and 30:70 (0.23-0.24 m 3 of CH4/kg ofVS feed ) were similar
to the SMPs of the biochemical methane potential test. On the other hand, .
the SMP (0.34-0.37 m 3 of CH4/kg ofVSfeed ) of the feed mixture of 50:50 was
larger than the SMP obtained in the biochemical methane potential test.
However, the maximum SMP of 0.37 m 3 ofCH4/kg of VS feed is nearly iden-
tical to those obtained by Mata-Alvarez et al. (14). Table 2 summarizes the
operating conditions, reactor characteristics and digester performances
Applied Biochemistry and Biotechnology Vol. 105-708,2003
~
:g
[
0:> Table 2

r"\ Results of SSAD Operated at Different HRTs with Three Feed Mixtures
::r
~
:3v;.
q-
Mixture ratios of SKFW to WAS
'<:
OJ
::J Parameter 10:90 (step 1) 30:70 (step 2) 50:50 (step 3)
0..
0:>
o· Operating conditions
fi)
r"\ HRT (d) 20 16 13 10 20 16 13 10 20 16 13 10
::r
::J 3.77 4.89 2.93 3.75 4.6 5.96
0 OLR (kg TCOD/[m 3 ·dD 2.27 2.79 3.42 4.43 2.51 3.07
~ OLR (kg TVS/[m3 ·dD 1.19 1.49 1.83 2.37 1.34 1.63 2.00 2.60 1.61 1.97 2.42 3.14
'<:

Reactor characteristics
TCOD (giL) 26.4 27.2 27.6 28.7 25.1 26.5 27.4 28.0 25.1 25.8 27.0 27.7
U1
'.J SCOD (giL) 0.89 0.85 0.85 0.85 0.72 0.75 0.79 0.79 0.74 0.81 0.76 0.76
'.J 22.7 22.8 23.5 21.9 22.9 23.0 23.4 20.6 20.8 20.7 20.8
TS (giL) 22.1
VS (giL) 14.6 15.3 15.6 16.1 14.6 15.1 15.2 15.7 13.7 14.0 14.2 14.5
pH 7.45 7.37 7.29 7.19 7.51 7.48 7.41 7.38 7.48 7.42 7.40 7.37
Alkalinity as CaC03 (giL) 3.54 3.4 3.28 3.11 4.04 3.95 3.91 3.74 4.91 4.75 4.64 4.37
NH4-N (giL) 0.76 0.74 0.72 0.68 0.93 0.91 0.91 0.89 1.16 1.14 1.09 1.01
VFAs (giL) 0 0 0 0 0 0 0 0 0 0 0 0
Digester performance
TCOD removal (%) 39.1 37.2 36.4 33.9 49.5 45.8 44 42.8 57.9 56.7 54.7 53.6
TVS removal (%) 38.6 36.5 36.1 33.9 43.9 41.1 40.0 38.0 56.3 55.3 54.8 53.7
~
:- Methane content (%) 85.7 82.6 78.5 76.5 80.2 77.0 72.5 70.4 69.4 66.8 64.5 63.3
....
<::> SMP, m 3 CH4/kg TVS feed 0.194 0.192 0.218 0.202 0.23 0.228 0.235 0.232 0.339 0.359 0.375 0.366
1"
.... MPR, m 3 CH4 1(m3 ·d) 0.230 0.286 0.4 0.48 0.298 0.356 0.471 0.601 0.532 0.708 0.907 1.150
<::>
,0:>
N
<::>
8
578 Heo et al.

of SSAD to treat the mixture wastes of SKFW and WAS. The data pre-
sented are the mean values at the steady-state condition.

Conclusion
The effects of HRT and mixture ratio of SKFW:WAS on the stability
and performance of the SSAD were investigated. The SSAD was quite
efficient for treating KFW.
Maintaining the proper pH in the digester is important to keep the
FAN concentration, a strong inhibitor, low. The digester pH was main-
tained within the stable range of 7.2-7.5 with sufficient alkalinity ranging
from 3.1 to 4.91 giL as CaC03• In addition, the accumulation of VFA was
not observed in all experiments after reaching the first steady state. Under
the investigated operating conditions, there were no inhibitions of the
methanogenic activity in the digester.
The optimum operating conditions of the SSAD, concerning the MPR
and SMP, was found to be at an HRT of 10 d with a feed mixture ratio of
50% WAS and 50% SKFW. The TCOD removal efficiency of 53.6% and VS
removal efficiency of 53.7% were obtained at the highest OLR of 5.96 kg of
TCOD/(m 3·d) and 3.14 kg ofVS/(m3·d),respectively. ThemaximumMPR
of 1.15 m 3 of CH 41(m3·d) and SMP of 0.37 m 3 of CH4/kg of VSfeed with the
methane content of 63% were obtained at an HRT of 10 d. Therefore, the
optimum operating conditions in the SSAD, and the corresponding per-
formance of the digester indicate that the anaerobic codigestion of KFW
with sewage sludge is a good option for food waste and WAS treatment.

Acknowledgment
This work was supported by a grant from the Ministry of Commerce,
Industry and Energy of Korea.

References
1. KMOE (2002), Korea Ministry of Environment (Website: http://www.me.go.kr).
2. Kim, P. J., Chang, K. W., and Min, K. H. (1995),J. Korea Organic WasteRecyclingCounc.
3,35-42.
3. Yun, Y. 5., Park, J.I., Suh, M.S., and Park, J. M. (2000), Bioresour. Technol. 73,21-27.
4. Cho, J. K., Park, S. c., and Chang, H. N. (1995), Bioresour. Technol. 52,245-253.
5. Lee, J. P., Lee, J. 5., and Park, S. C. (1999), Appl. Biochern. Biotech. 77-79,585-593.
6. Kang, H. and Jewell, W. J. (1990), /. Korean Soc. Environ. Eng. 12(1),55-64.
7. Cho,J. K., Lee,J. P., Lee,J. 5., Park, S. C.,and Chang, H. N. (1996),/. Korean Solid Waste
Eng. Soc. 13(5), 616-624.
8. Rintala, J. A and Jarvinen, K. T. (1996), Waste Manage. Res. 14, 160-170.
9. Converti, A, Drago, F., Ghiazza, G., and Del Borghi, M. (1997), J. Chern. Tech.
Biotechnol. 69,231-239.
10. Demirekler, E. and Anderson, G. K. (1998), Environ. Technol., 19, 837-843.
11. Del Borghi, A, Converti, A, Plazzi, E., and Del Borghi, M. (1999), Bioprocess Eng. 20,
553-560.
12. Dinsdale, R. M., Premeier, G. c., Hawkes, F. R., and Hawkes, D. L. (2000), Bioresour.
Technol. 72,159-168.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Use of SSAD for Mixture Wastes of SKFWaste and WAS 579
13. Cecchi, F., Traverso, P. G., Perin, G., and Vallini, G. (1988), Environ. Technol. Lett. 9,
391-400.
14. Mata-Alvarez, J. and Cecchi, F., Pavan, P., and Llabres, P. (1990), BioI. Wastes 33,
181-199.
15. Poggi-Varaldo, H. M. and Oleszkiewicz, J. A. (1992), Environ. Technol. 13,409-421.
16. Kayhanian, M. and Tchobanoglous, G. (1993), Water Sci. Technol. 27, 133-143.
17. Kayhanian, M. and Rich, D. (1996), Water Environ. Res. 68,240-252.
18. Hamzawi, N., Kennedy, K. J., and McLean, D. D. (1998), Water Sci. Technol. 38(2),
127-132.
19. Hamzawi, N., Kennedy, K. J., and McLean, D. D. (1998), Environ. Technol. 19,993-1003.
20. Mata-Alvarez, J. (2003), Biomethanization of the Organic Fraction of Municipal Solid
Wastes, IWA, London, UK.
21. Task Force on Thermal Destruction (1992), Sludge Incineration, Manual of Practice
Fd-19, Water Environment Federation, Alexandria, VA.
22. Malina, J. F. and Pohland, F. G. (1992), Design of Anaerobic Processes for the Treatment
of Industrial and Municipal Wastes, vol. 7, Technomic, Lancaster, PA.
23. Lay, J. J., Li, Y. Y., Noike, T., Endo, J., and Ishimoto, S. (1997), Water Sci. Technol. 36(6/7),
493-500.
24. Braun, R., Huber, P., and Meyrath, J. (1981), Biotechnol. Lett. 3, 159-164.
25. Bhattacharya, S. K. and Parkin, G. G. (1989), ,. Water Pollut. Control Fed. 61,55-59.
26. Kayhanian, M. (1999), Environ. Technol. 20,355-365.
27. Heo, N. H, Park, S. c., Lee,J. 5., and Kang, H., (2002), in Proceedings ofthe SecondIWA
Specialised Conference on Environmental Biotechnology, vol.2, International Water
Association, Massey University, Palmerston North, New Zealand, pp. 461-469.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0581/$20.00

Biosorption and Desorption


of Copper (II) Ions by Bacillus sp.

WAIHUNG LO,*,l LAU MEl NG,l HONG CHUA?


PETER H. F. YU, l SHIRLEY N. SIN, 2 AND PO-KEUNG WONG 3
Department of Applied Biology and Chemical Technology
1

and the Open Laboratory of Chiral Technology, and


2Department of Civil and Structural Engineering,
The Hong Kong Polytechnic University,
Hung Hom, Hong Kong SAR, E-mail: bctlo@polyu.edu.hk; and
3Department of Biology, The Chinese University of Hong Kong,
Shatin, Hong Kong SAR

Abstract
Batch biosorption experiments were conducted to investigate the removal
of Cu2+ ions from aqueous solutions by a series of bacterial strains isolated
from a local activated sludge process. The characteristics of 12 isolates were
identified and examined for their ability to bind Cu2+ ions from aqueous
solution. Among the isolates, two species exhibited biosorption capacity
>40 mg of Cui g of dry cell. Isotherms for the biosorption of copper on
bacterial cells were developed and compared, and the equilibrium data
fitted well to the Langmuir and Freundlich isotherm models. The bio-
sorption of copper increased significantly with increasing pH from 2.0 to 6.0
regardless of the species. More than 90% of copper sorbed on the cells of
Bacillus sp. could be recovered by washing with 0.1 M HN03 for 5 min. The
performance of two different desorption processes was also tested and com-
pared. The results show that five biosorption and desorption cycles are a
better operation process than five successive biosorptions followed by one
desorption to remove and recover copper from aqueous solution. The
biosorbent could be used for at least five biosorptions and desorption cycles
without loss of copper removal capacity. It can be concluded that the acti-
vated sludge or sludge-isolated bacteria could be a potential biosorbent for
copper removal.
Index Entries: Activated sludge; bacteria; bioremediation; copper; des-
orption; heavy metal; metal removal; wastewater treatment process.

"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 581 Vol. 105-108,2003


582 Lo et al.

Introduction
Heavy metals, namely copper, zinc, nickel, chromium, cobalt, and sil-
ver, are commonly found in the effluents from electroplating and metal-
processing industries. Heavy metals are also the major waste constituents
from the manufacturing of paints, plastics, batteries, alloys, and scientific
instruments. It is well documented that are at high concentrations, heavy
metals toxic to living organisms, particularly to those in aquatic environ-
ments. In the past few decades, metal-laden effluent discharges into munici-
pal sewers without treatment have been strictly prohibited. In recent years,
more stringent industrial effluent discharge standards have been promul-
gated. In Hong Kong, effluents from industrial sources are required to be
pretreated to substantially reduce the heavy-metal contents to an acceptable
level before discharging into municipal sewers, especially for Cu (1).
Conventional methods employed for the removal of Cu ions from
industrial effluents include chemical precipitation, filtration, electro-
chemical treatment, and ion exchange. Most of these methods are expen-
sive and incapable of removing trace levels of Cu ions. The chemical
precipitation method produces a large amount of sludge for disposal. In
addition to physical and chemical methods, the potential of biosorption
has been demonstrated (2,3).
Biomass of bacteria has been known to readily adsorb or accumulate
metal ions. Bacteria's ability to uptake metals has attracted much atten-
tion because of its potential use as an effective and economic means for the
remediation of wastewater polluted by heavy metals (4). Microbial cells
can be supplied as waste in industrial fermentation process and biologic
wastewater treatment processes (5-7). Hence, biosorption may provide
an economical and effective alternative to the conventional techniques for
Cu removal.
It is important to identify more microbial strains that could uptake
metals with high efficiency and specificity as well as to design better
bioprocesses that effectively remove or recover heavy metals from aquatic
systems. To optimize design and operation of a biosorbent system for Cu
removal, a thorough understanding of biosorption behavior and desorp-
tion kinetic characteristics of microbial cells is needed. This motivated us
to evaluate the feasibility and ability of the activated sludge or sludge-
isolated bacteria to remove Cu in wastewater. The Cu biosorption charac-
teristics of one of the bacterial isolates, Micrococcus sp., have been reported
previously (8).
In the present study, the biosorption characteristics of other bacterial
strains isolated (Bacillus sp. and Pseudomonas sp.) from activated sludge were
examined and compared with Micrococcus sp. The effects of Cu concentra-
tion and pH on biosorption were systematically examined. The desorption
kinetics, efficiency of Cu removal and recovery by repeated biosorption and
desorption operations, efficiency of the desorption of metals from metal-
loaded biosorbents, and reusability of Bacillus sp. were studied.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Biosorption of Cu Ions by Bacillus sp. 583

Materials and Methods


Isolation Procedures and Identification
Fresh activated sludge was collected from the return sludge channel
at a local sewage treatment plant. The sludge was serially diluted in dis-
tilled and deionized water. Aliquots (0.1 mL) were spread on nutrient agar
(Difco) and cultivated in an incubator at 30°C for 3 d. Twelve different
colonies were picked up and maintained on the same medium for the fol-
lowing metal biosorption test. The colonial characteristics of these isolates
were studied. The isolates were identified using the MIDI Sherlock Micro-
bial Identification System and API 20 NE as well as 20 E systems.
Preparation of Biosorbent
The bacterial cells of each isolate were grown in 1-L conical flasks
containing 100 mL of nutrient broth at 30°C with shaking at 200 rpm. The
72-h cultivated cells were harvested by centrifuging (Beckman J21-21
Model) at 10,OOOg for 30 min. After rinsing twice with distilled and deion-
ized water, the cells were then suspended in a designated volume of dis-
tilled and deionized water for preparing the biomass stock solution. The
concentration of biomass stock solution was determined by oven drying at
105°C for 24 h after filtration before and after biosorption experiments.
Biosorption
The biosorbent at a final concentration of 1 to 2 g of cell/L was sus-
pended in a 100-mL solution containing 100 mg of CulL in a polypropylene
bottle, which was gently agitated (250 rpm) at 25°C. The pH of the metal
solution was adjusted to 5.5 by 0.1 M NaOH and 0.1 M HN03 just before
experimentation and at 9 h during the experiment. Samples of 5 mL were
taken from the solution at 3 and 12 h and subsequently centrifuged at
lO,OOOg for 10 min. The concentration of each heavy metal in the superna-
tants was determined using a Perkin-Elmer atomic absorption spectropho-
tometer model 100. To determine the effect of pH on Cu biosorption, the
metal solutions and the bacterial suspensions were adjusted separately to
the desirable pH (2.0-6.0) by the addition of 0.1 M NaOH and 0.1 M HN03
and mixed. The mixture was then incubated at an initial Cu concentration
of 100 mg/L at 25°C on an orbital shaker for 12 h.
Desorption Kinetics
After the biosorption experiment on one selected isolate, Bacillus sp.,
the metal-loaded bacterial cells were harvested from the cell and metal
suspensions initially containing 0, 2, and 50 mg/L of Cu, respectively. The
bacterial cells were then rinsed with distilled and deionized water and
resuspended in 0.1 M HN03 solution. After gentle shaking, samples were
taken from the suspensions at designated time intervals. The samples were
centrifuged immediately, and the metal concentration in the supernatant
was determined.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


584 Lo et al.
Biosorption and Desorption Cycles
Bacterial cells with a final concentration of 1 to 2 g of cell/L were
suspended in metal solutions containing 0, 2, and 50 mg/L of Cu in cen-
trifuge tubes at pH 5.0. The tubes were shaken at 25°C for 12 h. After 12
h, the metal-sorbed cells were centrifuged, rinsed with distilled and deion-
ized water, and resuspended in 35 mL of 0.1 M HN03 for 3 h in order to
recover the metal ions from the cells. The regenerated biosorbents were
again suspended in metal solutions for the next biosorption run. The
biosorption and desorption steps were repeated five times. The metal
concentrations in the supernatants were determined after biosorption
and desorption.
Successive Biosorption and Desorption
Bacterial cells at a final concentration of 1 to 2 g of cell/L were sus-
pended in solutions containing 0, 2, and 50 mg/L in centrifuge tubes. The
pH of the cells and metal solutions was adjusted to 5.0 by adding 0.1 M
HN03 and NaOH. The tubes were shaken at 25°C for 12 h. The tubes were
then centrifuged, and the supernatant was decanted and analyzed for
unsorbed Cu. The centrifuged cells were resuspended in 35 mL of fresh Cu
solution and Cu sorption took place for a further 12 h. This procedure was
repeated five times. After the fifth sorption cycle, Cu from the cells was
desorbed by extraction with 0.1 M HN03 • The amount of sorbed or des-
orbed Cu was calculated by measuring the decrease or increase in Cu in the
contacting solution.

Results and Discussion


Isolation and Identification
The isolation procedure yielded about 12 strains. All the heterotrophic
aerobic isolates from the activated sludge showed a wide variety of species
including Gram-negative and -positive bacteria. Most of them were rods
and a few bacteria were filamentous and coccus. These bacterial isolates
were identified; their identities are listed in Table 1.
Biosorption
Table 1 shows that the Cu removal data of bacterial isolates are char-
acterized by large variations in biosorption capacity. The amount of
biosorption ranged between 2.9 and 42 mg/ g. Metal binding on the cell
surface is possible through phosphoryl, carboxyl, sulfydryl, and hydroxyl
groups of cell envelope (2). These different sorption capacities would be a
consequence of the structure and chemistry in the bacterial envelope. Three
isolates sorbed Cu more than 20 mg/ g: Bacillus sp. (21 mg/ g), Pseudomonas
sp. (37 mg/ g), and Micrococcus sp. (42 mg/ g). These three isolates were
chosen for biosorption isotherm study on the basis of their higher efficiency
in removing Cu from solution than the other isolates.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Biosorption of Cu Ions by Bacillus sp. 585
Table 1
Identification and Cu2+ Removal Capacity of Isolates
Cu removal capacity
Isolates (mg Cui g dry cell)
Bacillus sp. 21.0
Pseudomonas sp. 37.0
Micrococcus sp. 42.0
Neisseria sicca 5.6
Aeromonas hydrophila 2.9
Pseudomonas sp. 3.8
Xanthomonas maltophilia 3.1
Bacillus lentimorbus 2.9
Pseudomonas sp. 3.9
Bacillus subtilis 8.4
Gordona bronchialis 7.5
Kocuria varians 5.5

Effects of pH on Biosorption
eu uptake was negligible at pH 2.0 and then increased rapidly as the
pH was increased from 3.0 to 5.0 (Fig. 1). The pH of the solution obviously
affected eu removal by the three selected isolates. As suggested in previ-
ous studies (6,9), the biosorption process of metal is analogous to the ion-
exchange process. Therefore, metal cations and protons compete for
binding sites on the cell walls as pH decreases, lowering uptake of metal.
Additional support for this assumption was the finding that eu could
easily be desorbed from the cells by lowering the pH in the following
desorption study. This finding also indicated that the eu removed was
mainly bound to cell walls and external surfaces of the biomass.
Biosorption Isotherms
Biosorption isotherms for eu are presented in Fig. 2. The estimated
model parameter values and the correlation coefficients for the Langmuir
and the Freundlich isotherms are given in Table 2. The goodness of fit was
satisfactory over the range of the experiments. Values of Langmuir maxi-
mum biosorption capacity, qmax, for the different materials ranged from 21
to 47 mg/ g. The Freundlich coefficient k ranged from 4.9 to 15 mL/ g.
These values varied among the tested biosorbents, indicating large differ-
ence in their eu sorption behavior.
Based on the Langmuir biosorption maximum, qmax' the species of
Micrococcus and Pseudomonas showed the highest biosorption capacity.
However, the corresponding Freundlich coefficient, k, was similar among
the three species. The parameters k and qmax reflect different characteristics.
The Freundlich k represents the amount of eu sorbed when the solution

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


586 Lo et al.
120

100

i 80
t'
-r
"CI

60
t.l
III
.!.
c:r 40

20

0
0 2 3 4 5 6 7
pH

Fig. 1. Effects of pH of solution on Cu removal capacity of three isolates at initial


concentration of 100 mg of CulL.

concentration in the equilibrium is unity. On the other hand, qmax represents


the saturation level of sorbed eu at high solution concentrations. Similarly,
the parameters n and b are also not directly comparable. The parameter n
measures the extent of impact on biosorption of a change in residual solu-
tion concentration from unity (a high value for n implies a relatively large
change in sorbed eu when the residual eu concentration deviates from
unity, either above or below it). By contrast, the parameter b measures the
affinity of the biosorbent for the solute (a high value of b means a high
sorption level at low solution concentration). Thus, the Freundlich param-
eter k and the Langmuir parameter b both measure, in a sense, the effective-
ness of eu biosorption at low eu concentration in solution.
Micrococcus sp. showed the highest values for band qmax' indicating a
large capacity for eu biosorption at all solution concentrationS. The qmax
value for Pseudomonas sp. was slightly higher, indicating a high saturation
level at high eu concentration. The relatively low value of k indicates that
Pseudomonas sp. sorbed less eu at a low solution concentration. Thus, the
biosorption capacity is affected by solution eu concentration. Despite hav-
ing a lower qmax than the other two bacteria, Bacillus sp. sorbed more eu at
a low solution concentration. This indicated that Bacillus sp. is potentially
useful for removing eu from solutions at low concentration. Activated
sludge showed moderate biosorption capacity at low or high concentra-
tions in comparison with the other bacteria tested. The observed eu
biosorption capacities of Pseudomonas sp. and Micrococcus sp. were higher
than those reported (23 mg/ g dry cell) for isolated Pseudomonas aeruginosa
PU 21 from hospital sewage (10) and other microorganisms such as yeast
and fungal biomass (11,12).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Biosorption of Cu Ions by Bacillus sp. 587

A~------------------------~
o M"u:rococcus sp.
!:J. Btu:lIbtssp.
50 o helllilnMnllSJIIIt/ds
\7 Aetlvated sludge

o ro ~ ~ ~ 50 A ~ " H ~
Ce (mg CulL)

Fig. 2. Biosorption isotherms of three isolates and activated sludge at pH 5.0.

Table 2
Parameters of the Langmuir and Freundlich Isotherms
for Activated Sludge and Bacteria
Langmuir equation Freundlich equation

qmax b r2 k n r2
Activated sludge 31 0.20 0.99 4.9 0.46 0.99
Micrococcus sp. 43 2.06 0.99 14.0 0.37 0.96
Pseudomonas sp. 47 0.06 0.99 14.0 0.37 0.96
Bacillus sp. 21 1.0 0.99 15.0 0.12 0.94

Kinetics of Desorption
The kinetics of the desorption of Cu from the Cu-Ioaded cells of Bacil-
lus sp. is demonstrated in Fig. 3. It can clearly be seen that Cu desorbed very
rapidly, and the desorption reached equilibrium within 15 min regardless
of nearly or partially Cu-saturated biomass. Chang et al. (10) also found
that equilibrium was shortly reached after 5 min contact in the case of P.
aeruginosa PU 21 for Pb, Cu, and Cd desorption. The desorption efficiency
was about 95%. Diluted HN03 (0.1 M) is efficient for the recovery of Cu
from Bacillus sp.
Biosorption and Desorption Cycles
Figures 4 and 5 illustrate the metal removal and recovery efficiencies
of Bacillus sp. during five regeneration cycles. There was no significant
difference in the Cu biosorption capacity of the biomass from cycle 1 to 5.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


588 Lo et al.
24

o l-wL
.1III-cfL
18
,... ...
i=-
to
'1:1

r
...
~

!.
12

0::1'

I'>"~ "
0
0 500 1000 1500
TlDle(mJa)

Fig. 3. Desorption kinetics on metal-loaded Bacillus sp. With 2 and 50 mg of CulL


for 24 h followed by washing with 0.1 M HN03•

3,-------------------------------,
t?Z22 Sorptioa
c:J Desorptioa

1 1 3 4 5
Cycle

Fig. 4. Sorptionl desorption cycles for Cu at pH 5.0 and 2 ppm.

For all cycles, about 90% and 95% of sorbed eu could be recovered by 0.1 M
HN03-induced desorption at low and high eu-Ioaded biomass, respec-
tively. eu recovery from the biomass was very effective with 0.1 M HN03•
In addition, the biosorption capacity of the biomass in subsequent cycles
was not reduced by 0.1 M HN03•

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Biosorption of Cu Ions by Bacillus sp. 589
15~-------------------------------,
I2Zl Sorption
c::J Desorption

3 4 5

Cycle

Fig. 5. Sorption/ desorption cycles for Cu at pH 5.0 and 50 ppm.

50

=-...
;:;
0
e
2mgIL

..
50mgIL
>. 40
"1:1
1)1)
.....
U
=
1)1) 30
6
'-'
=
a..
0

20
0
'"
"1:1
~
01
:; 10
6
......=
-<
0
0 2 3 4 5 6
Cycle

Fig. 6. Successive sorption cycles for Cu with Bacillus sp. At pH 5.0.

Successive Biosorption and Desorption


Bacillus sp. were further evaluated by repeatedly contacting the mate-
rials with eu concentrations in order to gain a better insight into the long-
term eu sorption behavior. The accumulated sorption curves for Bacillus
sp. during repeated contacting with eu solutions over five cycles are shown
in Fig. 6. Repeatedly contacting Bacillus sp. with a high eu concentration
solution resulted in faster attainment of steady saturation level within three
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
590 Lo et at.

Table 3
Performance of Two Different Biosorption and Desorption Processes
Five successive biosorption Five biosorption
and one desorption cycles and desorption cycles
Total sorption Total desorption Total sorption Total desorption
(mg/ g) (mg/ g) (mg/ g) (mg/ g)
2mg/L 8.8 5.9 7.9 7.0
50mg/L 42.0 37.0 81.0 77.0

cycles and much higher sorption than indicated by the Langmuir sorption
maximum. At low Cu concentration contact, Cu uptake continued at a
steady rate in small increments even after five cycles since the amount of Cu
sorbed during repeated contacting was less than the Langmuir sorption
maximum. Therefore, the biomass could be used repeatedly without des-
orption to remove Cu from a low Cu solution.
The total amount of Cu sorbed onto the Bacillus sp. in the two different
biosorption and desorption processes is compared in Table 3. At high Cu
concentration, the alternative biosorption and desorption process could
remove more copper. More frequent regeneration of Cu sorption sites was
needed to maintain the removal efficiency at high Cu concentration (higher
than the Langmuir sorption maximum, qrnax). Unlike the high concentra-
tion, the amount of Cu sorbed for reused cells and regenerated cells was
quite similar at low concentration. Five successive biosorptions with one
desorption process is less expensive and more environmentally friendly in
the treatment of low effluent Cu. Less desorption solution will be used to
regenerate the sorption sites. However, the total amount of Cu desorbed
from successively reused cells without desorption was lower than that
from cells regenerated by desorption during each cycle. The unrecoverable
Cu in reused cells after five successive sorptions might be owing to the Cu
diffusion into the interior of the cell walL

Conclusion
Among the 12 isolates, Micrococcus sp., Pseudomonas sp., and Bacillus
sp. exhibited Cu biosorption capacity >20 mg/ g of dry cell. Copper
biosorption by these bacterial strains was strongly affected by Cu solution
concentration. Based on the Freundlich parameter k, Bacillus sp. was found
to have a biosorption extent comparable with that of Micrococcus sp. at
low Cu concentration. The biosorption of Cu increased significantly with
increasing pH from 2.0 to 6.0 regardless of the species, thereby suggesting
ion exchange as one of the dominant biosorption mechanisms. More than
90% of eu sorbed on the cells of Bacillus sp. could be recovered by washing
with 0.1 M HN03 for 5 min. Alternative biosorption and desorption cycles
was a better operation process than successive biosorption followed by

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Biosorption of Cu Ions by Bacillus sp. 591
one desorption to remove and recover Cu from aqueous solution that had
a total amount of Cu higher than the Langmuir sorption maximum. Bacil-
lus sp. was used for at least five sorption and desorption cycles without
loss of Cu removal capacity. In conclusion, activated sludge or sludge-
isolated bacteria could be a potential biosorbent for Cu removal.

Acknowledgments
We thank Dr. K. C. Cheung and Louisa Fok for isolating the micro-
bial strains. We also wish to express our sincere gratitude to the Hong
Kong Research Grants Council (grant no. PolyU 5001/00E) and the Hong
Kong Polytechnic University Area of Excellence Scheme for supporting
this research.

References
1. Environmental Protection Department (1991), Technical memorandum-Standards
for effluents discharged into drainage and sewerage systems, inland and coastal
waters, Hong Kong Government, Hong Kong SAR.
2. Volesky, B. and Holan, Z. R. (1995), Biotechnol. Prog. 11,235-250.
3. Butter, J., Evison, L. M., and Hamcock, I. C. (1998), Water Res. 32(2),400-406.
4. Lo, W., Chua, H., Wong, M. F., and Yu, P. F. (2003), Water Sci. Technol. 47,251-256.
5. Kasan, H. C. (1993), Crit. Rev. Environ. Sci. Technol. 23(1), 79-117.
6. Lo, W., Chua, H., Lam, K. H., and Bi, S. P. (1999), Chemosphere 39(15), 2723-2736.
7. Leung, W. c., Wong, M. F., Chua, H., Lo, W., Yu, P. H. F., and Leung, C. K. (2000),
Water Sci. Technol. 41(12), 233-240.
8. Wong, M. F., Chua, H., Lo, W., Leung, C. K., and Yu, P. H. F. (2001), Appl. Biochem.
Biotechnol. 91-93,447-457.
9. Zhang, L., Zhao, L., Yu, Y., and Chen, C. (1998), Water Res. 32(5), 1437-1444.
10. Chang, J. 5., Law, R., and Chang, C. C. (1997), Water Res. 31(7), 1654-1658.
11. Volesky, B. and May-Phillips, H. A. (1995), Appl. Microbiol. Biotechnol. 42, 797-806.
12. Niu, H., Xu, X. 5., Wang, J. H., and Volesky, B. (1993), Biotechnol. Bioeng. 42,785-787.

Applied Biochemistry and Biotechnology Vol. 105-1OB, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03 /l05-108 /0593 /$20.00

Flow Field in a Shrinking-Bed Reactor


for Pretreatment of Cellulosic Biomass

YINKUN WAN AND THOMAS R. HANLEY*


Department of Chemical Engineering,
University of Louisville, Louisville, KY 40292,
E-mail: tom.hanley@louisville.edu

Abstract
A shrinking-bed reactor was designed by the National Renewable Energy
Laboratory to maintain a constant bulk packing density of cellulosic biom-
ass. The high solid-to-liquid ratio in the pretreatment process allows a high
sugar yield and avoids the need to flush large volumes of solution through
the reactor. The shrinking-bed reactor is a promising pretreatment reactor
with the potential for scale-up for commercial applications. To scale up the
shrinking-bed reactor, it is necessary to understand the flow pattern in the
reactor. In this study, flow field is simulated with computational fluid
dynamics using a porous medium model. Different discrete" snapshots" and
multiple steady states are utilized. The bulk flow pattern, velocity distribu-
tion, and pressure drop are determined from the simulation and can be used
to guide reactor design and scale-up.

Index Entries: Shrinking-bed reactor; computational fluid dynamics; flow


field; cellulosic biomass; hydrolysis.

Introduction
Decades ago, the National Renewable Energy Laboratory (NREL)
investigated the possibility of producing high yields of glucose from cellu-
lose by thermochemical methods. Complications arose owing to the use of
strong acid and high temperatures. Strong acids require significant alkali
neutralization. High temperature requires a short residence time to pre-
vent glucose degradation into toxic chemicals. High acid solution flow
rates produce large volumes of acid solution that require treatment and
heating. Recently, NREL investigated a two-stage, two-temperature
dilute-acid pretreatment process (1-3). In the first stage, the temperature is
relatively low and much of the hemicellulose hydrolyzes. In the second
stage, the temperature rises, hydrolyzing more of the hemicellulose. The

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 593 Vol. 105-108, 2003


594 Wan and Hanley
Acid Solution Inlet

! Spring

Moveable end

Biomass

Fixed end

Acid Solution Outlet

Fig. 1. Structure of shrinking-bed reactor.

countercurrent process shortens the residence time, and two-temperature


processing avoids exposure to high temperatures, reducing the degrada-
tion and reopening of thermochemical hydrolysis of cellulose. To reduce
the acid solution volume and keep high solid-to-liquid ratio, NREL
designed a shrinking-bed reactor (4).
The shrinking-bed reactor was designed to keep a constant bulk pack-
ing density of solid biomass in the reactor, resulting in a high sugar concen-
tration owing to the high solid-to-liquid ratio, as well as a reduced liquid
volume. Figure 1 shows the structure of a laboratory-scale shrinking-bed
reactor. This reactor has one fixed end and one movable end. The movable
end is supported by a compressed spring. When the reaction processes, the
easily processed hemicellulose is hydrolyzed and removed from the packed
bed, decreasing the bulk packing-bed density. When the spring-loaded
movable end compresses the remaining biomass particles, the bulk pack-
ing density remains constant.
Chen et al. (5) developed a kinetic model to describe the shrinking-bed
reactor and found that shrinking-bed operation increases the sugar yield
by about 5% compared with non-shrinking-bed operation. Lee et al. (6) also
developed a mathematical model to simulate a countercurrent shrinking-
bed reactor and show that bed shrinking positively affects both hemicellu-
lose and cellulose hydrolysis. Converse (7) presents a model to simulate a
cross-flow shrinking-bed reactor for the hydrolysis of lignocellulosics and
finds a trade-off between cross-flow and plug-flow shrinking-bed reactors.
Cross flow allows higher yields than plug flow but generally with a slight
reduction in product concentration.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Flow Field in a Shrinking Bed Factor 595
Previous studies have shown the great potential of the shrinking-bed
reactor for commercial application. To scale up the shrinking-bed reactor,
the first step is to understand the flow field in a laboratory-scale shrinking-
bed reactor. In the present work, computational fluid dynamics (CFD) is
used to study flow behavior in a laboratory-scale shrinking-bed reactor.

Model and Method


Porous Media Model
The porous media model (8) is used to simulate flows through shrink-
ing-bed operation, (Le., packed-bed) in which biomass is taken as porous
media.
In the porous media model, a momentum source term Si is added to
the standard fluid flow equation.
3 3 1
$.=
1
l: D"IlV,+
j=l IJ
l: C .. -pIV·IV.
J j=l IJ 2 J J
(1)

The first part of Eq. 1 is a viscous loss term and the second is an inertial
loss term. For simple homogeneous porous media, Si can be written as
Il 1
$.1 = -a V·1 + C2 -2 P IV·I1 v.
1
(2)

in which a is the permeability and C2 is the inertial resistance factor.


In laminar flows through porous media, the pressure is proportional
to velocity and C2 can be taken as zero. Ignoring convective acceleration
and diffusion, the porous media model can be changed into Darcy's Law:

VP = 1: V (3)
a
in which VP is the pressure drop.
For turbulent flows and for cases in which the permeability term can
be eliminated, the porous media model can be rewritten as follows:

a-=
iJP
~3 C
Xi J-l
2ij
(1-2 pvjlvil) (4)

For packed-bed and turbulent flows, Darcy's law can be rewritten as


Ergun equation (9):

VP = 150ll (1- E)2 V + 1.75p (1- E) VV


D2 E3 D £3 (5)
P P
For laminar flow, the second term of the Ergun equation may be
dropped, resulting in the Blake-Kozeny equation:

150ll (1_E)2
VP=-2- 3 V (6)
DP £

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Table 1
Different Simulated Stages
Packed- Packed- Inlet Outlet
bed bed port port Flow
height diameter diameter diameter rate Grid
Status (m) (m) (m) (m) (m3 /s) (Xx Y)

Initial (100%) 0.160 0.040 0.008 0.008 1.256 X 10-7 20 x 53


80% of initial height 0.128 0.040 0.008 0.008 1.256 x 10-7 20 x43
60% of initial height 0.096 0.040 0.008 0.008 1.256 x 10-7 20 x 32
40% of initial height 0.064 0.040 0.008 0.008 1.256 x 10-7 20 x 32
20% of initial height 0.032 0.040 0.008 0.008 1.256 x 10-7 20 x 32

Comparing Eqs. 3 and 4 with Eq. 5,


D2
P E
(l=- - - -
(7)
ISO (1- E)2

3.5 (I-E)
C2 =v P
--2-
E
(8)

in which Dp is the mean particle diameter, and 10 is the void volume fraction
of the packed bed.
Experimental Method
The commercial CFD package FLUENT 5.5 was used to simulate a
laboratory-scale shrinking-bed reactor. Since the shrinking process cannot
be simulated inFluent 5.5, we took several discrete "snapshots" to simulate
the shrinking process assuming the biomass particle diameter and void
fraction remain constant.
The shrinking-bed reactor's operation conditions were as follows:
Flowing media: water (liquid); temperature: 298K; heattransfer: none;
flow regime: laminar; inlet velocity: 2.5 x 10-3 m/ s; outlet pressure: 1 atm;
particle diameter (Dp): 1 mm; packed-bed void fraction (E): 0.2; (l/a) =1.72
x 1011; and C2 = 3.5 X 105 •
Initial and subsequent snapshot states are given in Table 1.

Results and Discussion


Flow Pattern by Velocity Vector
The velocity vectors in 100, 80, 60, 40, and 20% initial bed height are
shown in Fig. 2. The flow in the shrinking-bed reactor is mainly plug flow
except for some radial flow in the bed comers. As the bed shrinks, the
percentage of plug flow volume in the whole bed decreases. Since plug flow
is a positive factor in the mass transfer process, the mass transfer efficiency
will decrease as the bed shrinks.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Flow Field in a Shrinking Bed Factor 597

Fig. 2. Flow patterns in the shrinking-bed reactor.

To keep plug flow in the entire shrinking bed, one option is to create
even fluid distribution in the inlet and outlet. Figure 3 shows the flow pat-
tern of the shrinking-bed reactor with the inlet acid solution evenly distrib-
uted over the bed diameter. Apparently, this change reduces radial flow in
the top comer and improves the mass transfer efficiency. If the outlet flow
could also be evenly distributed, the radial flow in the bottom comer would
disappear, enhancing the shrinking-bed reactor's performance.
Axial and Radial Velocity Distributions
In this simulation, the X direction is radial, with positive X velocity
toward the reactor wall; Y is for axial direction, with positive Y velocity
indicating upward flow.
X velocity distributions for 80% initial bed height are shown in Fig. 4.
Clearly, radial flow exists in the comers of the shrinking bed. The radial
flow can increase the residence time and has different impacts on produc-
tion of sugars. If the hydrolysis process is controlled by chemical kinetics
and has not reached equilibrium, then an increase in residence time can
enhance production of sugars; if the hydrolysis has reached equilibrium,
the increase in residence time will result in sugar degradation.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
598 Wan and Hanley

4.95e-04 IlillllllllIlIlillilllllJilIUlUlIUI
1lUlUIIIIIIIIUlUIIIIIIIIUlIIIIUI
IIUJlIIllIlIIlllIUllllJUlUllIIllI1
1IUllUIIIIIIIUlUIIIIJUlIIIIIIIUI
1IIIIlIIIIIJillUIUlIIIJIIIUIIIIIUI
4.476-04 U11IUIUIIIIIUIUIIIIIUIUIIIIIUI
IIUIUIIIIIIIIUIUIIIIIUIUIIIIIUI
IJUIUlUlllUUIUlllIlUlUlllilUl
IIUIUIIIIIIIIUIUIIlIIUIUIIlIIUI
IJUlUlllIlIlIUlllllillUllllllllUl
3.996-04 1J1liUllUUllillUIIlIIUIIlIIlIIlII
IIIIIUIUlJlIJUIUIIIIIUIUIIIlIUI
III11UIJIIUIIUIUIJIIJIIIUIlIlIUI
IIUJUIIUlIllUIUlJlIIUlUllUWI
1I11l1l1l1l11liUlUIIIIIUIUIUIIUI
3.51&-04 IIUJUIIUJlllUIUlIlIIUlIIlIlllUI
IIUJlIlIUllIllIIUlllllUllIlJlIlUJ
IUIUIIIIJIIIUUJIIIIIUIIJIIIIIUI
IIUlIIlIliJIllIlIUllIlJllIUlUliU
IIUIIIIIIIIIIIIIIUIIIIJltIUlIllIUI
3.02e-04 IIUIIIIIUJJlIIIIUIIIIJJJlUIIIIIUJ
1II111111111111111UlIIIJJIlUIIIIIUJ
IIUIIIIUIIJIIIUUIUIIIIIUIUlIJlJ
UUlllllllJlUlllilIUlJlUUlJlII1II
UlIIlJlIllllllUllIlIIlllIIlIlIUl1II
2.54&-04 1JIIIIIIIIIIJlIJlIIIIlIlIlllUIJJlIUJ
WllIllIIIllllIIlUlIlIllIIJlllJIUIJ
IJIIlIUJIlIIIJUIUlDlIIIIIIIIJIIIII
1IJJ11l1J111J1J11IUlIIlIIIIIIIIIIIUI
IUlllillIllIlIJlIJllUlIlIlJlIlIlUlI
2.06&-04 1IIIIIlIlHIIIJIIIUIJJlllllUIIIIIUI
IlIIlUlIIIJJllIIlUIlJUllilHllIJIII
IlJUIIIIUIIIIIIIIIIIJIIUIIIIIIIIlIJ
IIUJIlIIIllIIIUIUIIlIlUIIUIIIDtI
1IU1I1I1I11IIIIl1UIIlIIIIUlUUIlU
1.58&-04 1IIIIIIIIIIUIIUlUIIIIJllllUIIIIJJI
IIUlllIUUllilUlllIlIJUllUlIllI1I
IIJUllIllIlIlIUlllIllIlIlIUUlllJlI
llUlJJlUllIlllllUUlIllIlUUl1ll11
UUUUUlIllIUIUlIIlIllUllJllI1I
1.096-04 UIIIIlIUIIIIIIlIJlIlIlII1UIIIIIIUI
IJIUIlIIlIIIIUlIUIlIlIIUUIIIlUU
1JIlllIJJllUIIIIIIJIIIllllIIJIIIJIUI
1I1l1111UIIUIlIIUIJIIIIIUlUUIIII

Lx
1IIIIIIIIIIUIWIlliUlIJUlllIIIUlI
6.109-05 lUllIlIUlllIUIIWllUw/JIlIUli1
III1UIIIIIIIUJ IUIU/WII
IIII11UIII 11/11111111
um'm"~ /IIIIIIIIU
mun: 1111
1.28&-05

Fig. 3. Flow pattern and fluid velocities in meters per second in the shrinking-bed
reactor with optimized inlet.

• default-mtenor

1.40e+02

1.20e+02
... . ... .
1.00e+02

8.ooe+01

Position
(mm) 6.00e+01

4.00e+01

2.00e+01

0.00e+00 -i't~.'--'.y-.-'r-.'"**""'-.-''--'-*,-r'-'-T-'~--,-.--.---,-.--.-.-.--,-­
-1 - 0.8 .(l.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

X Velocity (mm/s)

Fig. 4. Radial velocity distribution at 80% bed height.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Flow Field in a Shrinking Bed Factor 599
I • dataylt-interior

Results from the Yvelocity distributions along the axis in 100, 80, 60,
40, and 20% initial bed height show a uniform Y velocity distribution, or
plug flow, in the middle of the bed. The uniform Yvelocity area decreases
rapidly from about 60 to 0% in the whole bed volume when the bed height
shrinks from 100 to 20%. Moreover, it is observed that with no positive Y
velocity, no backflow occurs in the shrinking-bed reactor. Thus, the shrink-
ing-bed reactor is ideal for mass transfer and, as a result, can stimulate the
hydrolysis reaction. In this aspect, the shrinking-bed reactor is suitable for
bioethanol production. Typical Y velocity distribution along the axis is
shown in Fig. 5.
Pressure Drop
Typical pressure distribution contours in the shrinking-bed reactor
are shown in Fig. 6, where it is indicated that the pressure distribution is not
even. The highest static pressure occurs in the fluid inlet while the lowest
pressure appears in the fluid outlet. The higher pressure in the upper bed
level may result in higher hydrolysis reaction rates in this area.
In Fig. 7, pressure distributions for different bed heights are plotted
in the central axis (symmetry line). As the bed shrinks, the pressure drop
decreases. As the bed height decreases from 100 to 20% of initial height,
the pressure drop decreases from 1.7 x 104 to 5.8 X 103 Pa. The decreased
pressure drop has a positive effect on biomass hydrolysis. Initially, a large
pressure drop produces greater fluid flow and counters the effect of larger
Applied Biochemistry and Biotechnology Vol. 105-108,2003
600 Wan and Hanley

8.5ge+06

7.73e+06

6.87e+06

6.02e+06

5.16e+06

· 4.30e+06

3.44e+06

2.58e+06

1.72e+06

Lx
8.5ge+05

O.OOe+OO

Fig. 6. Static pressure contour (Pa) at 40% bed height.

180
........ 20%
-0-40%
160 +-----------------------------~_6_W% ~----------------~~~
-_80%
-lIE-1000k
140

120

E 100
.s.c
.
0
E
0 80
CI.

60 -"

40

20

0
0 2 3 456 7 8 9 10 11 12 13 14 15 16 16.6
Sialic Pressure (KPa)

Fig. 7. Pressure distribution along the central axis.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Flow Field in a Shrinking Bed Factor 601

~ ~~

~\ tiL
40"10
,
[I] 20%

Fig. 8. Path lines in shrinking-bed reactor at five bed heights.

bed height. Increased residence time of fluid in the bed means higher risk
of decomposition of sugars into toxic chemicals. In the latter stages, the
smaller pressure drop increases residence time and allows more lignocel-
lulose to be hydrolyzed.
Flow Path
Pathlines are used to visualize the flow of massless particles in the
problem domain. The particles are released from one or more defined sur-
faces. In the present work, 10 massless particles were released from the
fluid inlet. Figure 8 shows the path lines of the shrinking-bed reactor at
different heights and confirms the flow pattern described in Fig. 2.

Conclusions
The CFD simulation for a laboratory-scale shrinking-bed reactor in-
dicated mostly plug flow with some radial flow in the bed corners. No
backflow was found. As the bed shrank, the pressure drop of the bed
decreased. One suggestion to optimize the flow in this reactor is to adjust
the fluid inlet and outlet to produce more even flow distribution. Com-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
602 Wan and Hanley

pared with the non-shrinking-bed reactor, simulation results of the shrink-


ing-bed reactor show that the shrinking process can reduce the residence
time of the fluid in the bed as well as the risk of decomposition of sugars
into harmful chemicals. Moreover, the bed shrinkage reduces the fluid
flow and enhances sugar concentration in the hydrolysate to improve
process economics. Therefore, the shrinking-bed reactor is an ideal reactor
for bioethanol production.

Nomenclature
C, C2, D = matrix
Dp = particle diameter (m)
P = pressure (Pa)
Sj = source term for ith momentum equation
Vj = velocity of i direction
a = permeability
E = void fraction
!..t =molecular viscosity (kg/m·s)
p = density (kg/m3)

Acknowledgments
This research was supported by NREL (contract no. XCO-1-31016-01).

References
1. Tucker,M. P.,Farmer,J. D., Keller,F. A,Schell, D.J., and Nguyen, Q. A (1998),Appl.
Biochem. Biotechnol. 70-72, 25-35.
2. Nguyen, Q. A, Tucker, M. P., Keller, F. A, Beaty, D. A, Connors, K. M., and Eddy,
F. P. (1999), Appl. Biochem. Biotechnol. 77-79, 133-142.
3. Nguyen, Q. A, Tucker, M. P., Keller, F. A, and Eddy, F. P. (2000), Appl. Biochem.
Biotechnol. 84-86, 561-576.
4. Torget, R W., Hayward, T. K., and Elander, R (1997), presented at the 19th Sympo-
sium on Biotechnology for Fuel and Chemicals, Colorado Springs, CO.
5. Chen, R, Wu, Z., and Lee, Y. Y. (1998), Appl. Biochern. Biotechnol. 70-72,37-49.
6. Lee, Y. Y., Wu, Z., and Torget, R. W. (2000), Bioresour. Technol. 71,29-39.
7. Converse, A O. (2002), Bioresour. Technol. 81, 109-116.
8. Fluent, Inc. (1998) FLUENT 5 User's Guide, vol. 1., Fluent, Inc., Lebanon, NH.
9. Ergun, S. (1952), Chern. Eng. Prog. 48(2),89-94.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0603/$20.00

Lactic Acid Production Through


Cell-Recycle Repeated-Batch Bioreactor

HUROK OH, YOUNG-JUNG WEE,


JONG-SUN YUN, AND HWA-WON Rvu*
Faculty of Applied Chemical Engineering,
Institute of Bioindustrial Technology, Chonnam National University,
Gwangju 500-757, Korea, E-mail: hwryu@chonnam.ac.kr

Abstract
The effect of various nitrogen sources on cell growth and lactic acid pro-
duction was investigated. The most effective nitrogen source was yeast
extract; more yeast extract gave higher cell growth and lactic acid productiv-
ity. Yeast extract dosage and cell growth were proportional up to a yeast
extract concentration of 30 giL, and lactic acid productivity was linearly cor-
related up to a yeast extract dosage of 25 giL. However, increasing the yeast
extract content raises the total production cost of lactic acid. Therefore, we
attempted to find the optimum yeast extract dosage for a repeated-batch
operation with cell recycling. The results show that when using Enterococcus
faecalis RKYI only 26% of the yeast extract dosage for a conventional batch
fermentation was sufficientto produce the same amount oflactic acid, whereas
the lactic acid concentration in the product stream (92-94 giL) and lactic acid
productivity (6.03-6.20 g/[L·hD were similar to those of a batch operation.
Furthermore, long-term stability was established.
Index Entry: Lactic acid; repeated-batch; cell-recycle; hollow-fiber mod-
ule; nitrogen source; yeast extract.

Introduction
Lactic acid (CH3CHOHCOOH) is an organic hydroxy acid with wide-
spread industrial applications. Recently, much attention has been paid to
lactic acid as a monomer for biodegradable plastic. Between the two meth-
ods for producing lactic acid, fermentation is reported to be more favorable
than the chemical synthetic method, because of its environmental friendli-
ness and selectable production of a single enantiomer (1-4).
However, there are two major problems that make the biological
production of lactic acid economically unfavorable. One is the fastidious
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 603 Vol. 105-108, 2003


604 Oh et al.
nutrient requirements of lactic acid bacteria, because of their inability to
biosynthesize some essential amino acids and vitamins. The other is end-
product inhibition by the accumulated lactic acid (5-8).
Most lactic acid bacteria require complex nutrient sources, such as
yeast extract and peptone, which are usually expensive. According to
Tejayadi and Cheryan (9), the nitrogen source costs about 38% of the total
production cost.
To solve these problems some investigators studied inexpensive
nutrient sources, such as barley, wheat, cell autolysate, and soybeans (10-
12). Others tried to develop a more efficient process, mainly by improving
the lactic acid productivity with high-ceIl-density cultures using cell recy-
cling with or without an in situ separation apparatus (13,14).
In the present study, we characterized a cell-recycle repeated-batch
operation using a hollow-fiber filtration module to simultaneously reuse
the produced biomass and reduce the yeast extract dosage. The study
focused on the effects of various nitrogen sources, followed by character-
izing the effects of the yeast extract dosage during batch operation. The
minimum yeast extract dosage during a repeated-batch operation with
cell recycling giving high cell growth and lactic acid productivity was
also determined.

Materials and Methods


Strains and Culture Media
The homofermentative L (+ )-lactic acid producer, Enterococcus faecalis
RKYI (15-18) was used. Stock cultures were stored in the culture medium
with 50% glycerol at -20°C. The composition of the growth medium or
inoculum was 30 giL of glucose, 10 giL of yeast extract, and 10 giL of
K2HP04 • The fermentation medium was composed of 30-150 giL of glu-
cose, 0-40 giL of yeast extract, and no additional nutrients.
Batch Fermentation
In preparation for experiments, 1 mL of a stock culture was inocu-
lated into a 20-mL vial containing 15 mL of growth medium, and culti-
vated at 38°C and 200 rpm for 10 h. After three consecutive propagation
steps, 2 mL of the culture broth was transferred to a 50-mL vial containing
40 mL of sterilized growth medium for vial cultivation or inoculum prepa-
ration for 2.5-L jar fermentor experiments. In the vial experiments, the pH
was adjusted by manually adding 2 M Na 2C03 • Temperature and agita-
tion speed were 38°C and 200 rpm, respectively.
Fermentations were performed using a 2.5-L jar fermentor (KF-2.5L;
Korea Fermentor, Inchon, Korea) with automatic temperature, pH, and
agitation speed controls. During the batch fermentations, 4% (v Iv) inocu-
lum was transferred to a 2.5-L jar fermentor containing 1 L of sterilized
medium. The temperature was maintained at 38°C, with a constant agi-
tation speed of 200 rpm, and pH of 7.0, controlled by 10 M NaOH.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Lactic Acid Production by E. faecalis RKYI 605

~
® <D

.... ®
--
--
®
<D
®
Fig. 1. The schematic diagram of cell-recycle membrane bioreactor. I, Bioreactor;
2, hollow-fiber membrane filtration module; 3, product storage tank; 4, fresh medium
storage tank; 5, neutralizer reservoir; 6, peristaltic pump; and 7, three-way valve.

Cell-Recycle Repeated-Batch Operation


Repeated-batch operation with cell recycling was performed in the
same fermentor as the batch operation, but with a hollow-fiber module
(SKUF-103-0830; SK Chemical, Suwon, Korea). Figure 1 shows a sche-
matic diagram of the membrane cell-recycle bioreactor. The filtration
module contains about 100 fibers, with a 25-mm id and 300-mm length.
The nominal molecular weight cutoff of the membrane was 30 kDa and
the total surface area was 0.06 m 2•
Nutrient-rich medium (containing 15 gil of yeast extract) was used
during the first batch operation. After depletion of glucose of the first batch
run, 90% (v Iv) of the culture broth was withdrawn through the hollow-
fiber module and the same volume of fresh medium was fed into the reac-
tor. Subsequent batch cultures were performed in the same manner.
Analytical Methods
To determine cell growth the optical density was measured using a
spectrophotometer (UV-160A; Shimadzu, Tokyo, Japan) and converted to
a dry cell weight with a precalculated standard curve.
Glucose concentration was enzymatically measured by the glucose
oxidase-peroxidase method using a glucose E-kit (Young-Dong Pharma-
ceuticals, Seoul, Korea).Lactic acid concentration was quantified using a
high-performance liquid chromatograph (Waters486; Millipore, Bedford,
MA) with an Aminex HPX-87H ion-exclusion column (Bio-Rad, Hercules,
CA) and a UV detector set at 210 nm. The mobile phase was 5 mM H 2S04,
at a flow rate of 0.6 mL/min. The temperature of the column was main-
tained at 35°C.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
606 Oh et al.
Table 1
Effect of Various Nitrogen Sources on Cell Growth and Lactic Acid
Productivity During Batch Cultivation of E. faecalis RKYl"
Nitrogen sources Cell growth (giL) Lactic acid (giL)
None 0.20 3.42
Yeast extract 6.71 23.87
Bactopeptone 4.69 18.35
Polypeptone 4.62 16.89
Corn steep liquor 3.55 16.82
Corn steep solid 1.07 9.14
Beef extract 3.17 16.80
Pharmamedia 0.74 8.36
Soybean flour 0.68 7.22
Cottonseed flour 0.56 6.87
Malt extract 0.48 5.22
Sodium-L-glutamate 0.33 4.37
Ammonium sulfate 0.43 5.24
Ammonium chloride 0.35 5.23
Urea 0.30 4.18
aCulture conditions: 50-mL serum bottle containing 40 mL of medium (with
10 giL of each nitrogen source) at 38°C, 200 rpm, for 10 h.

Results
Effect of Nitrogen Sources
To investigate the effect of various nitrogen sources on lactic acid
fermentation, organic, inorganic, and several cereal-derived nitrogenous
materials were tested. Table 1 presents the results of 50-mL serum bottle
experiments containing 10 giL of each nitrogen source.
The highest dry cell weight (6.7 giL) and overall lactic acid produc-
tivity (2.4 g/[L·h]) were recorded when yeast extract was used as the
nitrogen source. Bactopeptone, polypeptone, corn steep liquor and beef
extract showed comparable results. Next to those nitrogen sources were
corn steep solid and some cereal-based materials, which showed some
possibilities as substitutive nitrogen sources. However, adding single
organic or inorganic substrates, such as sodium L-glutamate, ammonium
sulfate, ammonium chloride, and urea, promoted little in both the cell
growth and lactic acid production.
Effect of Yeast Extract Dosage on Batch Cultivation
Because yeast extract was proven to be the most effective nitrogen
source, we investigated the influence of yeast extract dosage during batch
cultivation with the 2.5-L fermentor experiments. Figure 2 presents cell
growth and lactic acid production in the medium with 100 giL of glucose
and 0-40 giL of yeast extract. As the yeast extract dosage was increased, the
dry cell weight and lactic acid production rate increased linearly up to 30
and 25 giL of yeast extract, respectively. For yeast extract dosage >30 giL,
however, both cell growth and lactic acid production rate did not increase.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Lactic Acid Production by E. faecalis RKYI 607

-0-- y=o
20 ____ Y=2
-0-- Y=5
---+--- Y = 10
--t:r- Y=15
15 --4- Y=20
-v- Y=25
----T- Y=30
-<>- Y=40

100

80

::r
--
~
"0
60

.~
CJ
.;:: 40
~
....1
20

25
Time [h]
Fig. 2. Effect of yeast extract dosage on cell growth and lactic acid production
during batch cultivation of E. faecalis RKYl. Medium composition: 100 giL of glu-
cose, 0-40 giL of yeast extract.

Cell Recycle Repeated Batch Operation


A study on repeated-batch cultivation with cell recycling was per-
formed to reduce the yeast extract requirement in the medium. Figure 3
shows the results for cell-recycle repeated-batch operation supplemented
with 100 giL of glucose and no yeast extract. Cell growth rate was lower
than cell lysis rate from the second batch run owing to limitations in nitro-
gen sources. Lactic acid productivities of each subsequent batch were 6.19,
3.42, and 1.98 g/(L·h), with a gradual decrease in cell activity.
Figure 4 presents the results for cell-recycle repeated-batch cultiva-
tion using fresh medium containing 100 giL of glucose and 3 giL of yeast
Applied Biochemistry and Biotechnology Vol. 105-70B, 2003
608 Oh et al.
14

12

10

~ 8
te
CO 6

=i3
u 4

0
120 -+- Lactic acid
-0- Glucose
100
~
.....
~ 80
0
g
Oh
't:l 60
!a
....'t:lg 40
(.)
'p
g 20
...:l

0
0 15 30 45 60 75
Time [h]

Fig. 3. Time course for cell-recycle repeated-batch operation with no yeast extract.
Culture conditions: 2.5-L jar fermentor containing 1 L of medium at 38°C, 200 rpm;
composition of fresh medium: 100 giL of glucose, 0 giL of yeast extract.

extract. After the fifth batch, the accumulated cell growth was about 17
gIL, but the lactic acid productivity and the substrate consumption rate
were lower than those of the first batch by 14 and 13%, respectively.
Figure 5 shows the fermentation profiles of repeated-batch opera-
tion with cell-recycle using fresh medium containing 100 gIL of glucose
and 4 gIL of yeast extract. A consistent cell growth was seen from the
second to tenth batches, with the maximum cell growth of 28.5 gIL.
Through the repeated-batch runs, the lactic acid productivities and the
substrate consumption rates were stable, in the range of 6.03-:-6.39 gl (L·h)
and 95-100%, respectively.
Figure 6 shows the results of the cell-recycle repeated-batch opera-
tion using fresh medium supplemented with 150 gIL of glucose and 7 gl
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Lactic Acid Production by E. faecalis RKYI 609
20.--------------------------------,

0
120 _ Lactic acid
-0- Glucose

~ 100
.......
Q)

'"0 80
.E
01)

] 60

.....
"0
~ 40
()
.J::
~ 20
~

Time [h]

Fig. 4. Time course for cell-recycle repeated-batch operation with 3 giL of yeast
extract. Culture conditions: 2.5-L jar fermentor containing 1 L of medium at 38°C,
200 rpm; composition of fresh medium: 100 giL of glucose, 3 giL of yeast extract.

L of yeast extract. The substrate consumption rates were 95-99% through-


out the experiment. In other words, the system was stably maintained
under these nutritional conditions, as with the operation with 100 giL
of glucose and 4 giL of yeast extract. The maximum cell growth was 28
giL, and the lactic acid productivities of each batch were between 5.75
and 5.90 gl (L·h).
Discussion
Both cell growth and lactic acid productivity were low when single
organic or inorganic nitrogen source was added to the medium. When
cereal-based nutrients were tested as nitrogen sources, cell growth was low,
but the specific productivity based on cell mass (1.06-1.12 g of lactic acidl
Applied Biochemistry and Biotechnology Vol. 105-108,2003
610 Oh et al.

25

20
~
~
ofi 15
~
I)Il
:=1 10
41
U

0
120 ___ Lactic acid
--0- Glucose
.....,
100
~(1)
rfJ
0 80
U
..=OIl
'e 60
§
'e
.~ 40
.;:l
~ 20
.....:I

15 30 45 60 75 90 105 120 135 150


Time [h]

Fig. 5. Time course for cell-recycle repeated-batch operation with 4 giL of yeast
extract. Culture conditions: 2.5-L jar fermentor containing 1 L of medium at 38°C,
200 rpm; composition of fresh medium: 100 giL of glucose, 4 giL of yeast extract.

[g of cell·h]) was higher than any other nutrient sources. The highest cell
growth and lactic acid productivity were found with undefined organic
substrates. Like most studies on lactic acid biosynthesis, the most effective
nitrogen source was yeast extract (1,2,7).
During batch operation, when the effects of yeast extract dosage on
lactic acid fermentation were tested, it was found that more added yeast
extract caused higher cell growth and lactic acid productivity. Figure 7 pre-
sents the relationship between the yeast extract dosage and maximum cell
growth, and the relationship between yeast extract dosage and overall pro-
ductivity. Figure 7 also shows that the yeast extract dosage and cell growth
were proportional up to a yeast extract dosage of 30 giL, and that lactic acid
productivity was linearly correlated up to a yeast extract dosage of 25 giL.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Lactic Acid Production by E. faecalis RKYI 611
30~~~~~~~~~~~~~~~~~~~

25

-
'U
U
10

___ Lactic acid


-0- Glucose
~150
~
~ 120
o
.2OIl
'0 90
§
'0
.~ 60
u
'):1
~ 30
.....:l

0
0 24 48 72 96 120 144 168
Time [h]

Fig. 6. Time course for cell-recycle repeated-batch operation with 150 giL of glucose
and 7 giL of yeast extract. Culture conditions: 2.5-L jar fermentor containing 1 L of
medium at 38°C, 200 rpm; composition of fresh medium: 150 giL of glucose, 7 giL of
yeast extract.
20r-~~~--------------------------'

__ Cell growthmax
-+- LA productivity
10 20 30 40
Yeast extract dosage [gIL]

Fig. 7. Effect of yeast extract dosage on cell growth and volumetric productivity of
lactic acid during batch cultivation of E. faecalis RKYl. LA, lactic acid.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


612 Oh et al.
These results are similar to those of Kulozik and Wilde (6). When Lactobacillus
helveticus was grown on 60 giL of lactose, the lactic acid production rate
increased linearly with increases in the yeast extract concentration up to 10
giL of yeast extract, whereas the cell growth was maximized at 15 giL of
yeast extract.
However, with the higher yeast extract concentration in the media, the
cost of raw material increases. Therefore, cell-recycle repeated-batch opera-
tion was characterized, because it was believed that it could reduce the yeast
extract requirement while sustaining the productivity of a batch operation.
If the medium contained 100 giL of glucose and 4 giL of yeast extract,
43 g of yeast extract was required to produce 1 kg of lactic acid. This is only
26% the yeast extract dosage of batch fermentation, whereas the lactic acid
concentration in the product stream (92-94 giL) and lactic acid productiv-
ity (6.03-6.20 g/[L·hD were similar to those of a batch operation. Further-
more, cell-recycle repeated-batch operation gave long-term stability in the
lactic acid production.
When the fresh medium containing 150 giL of glucose was used, 50 g
of yeast extract was required to produce 1 kg of lactic acid. The lactic acid
concentration in the product stream was 138 to 142 giL, and the lactic acid
productivitywas5.75-5.90g/(L·h). The initial concentration of lactic acid of
each repeated-batch operation was in the range of 8.6-14.5 giL owing to
retained medium, but inhibition by the initial lactic acid was not observed
during fermentation.

Conclusion
Among the various nitrogen sources tested, the most effective nitro-
gen source was yeast extract. As well, with greater addition of yeast
extract, higher cell growth and lactic acid productivity were achieved.
When the repeated-batch operation with cell-recycle was performed with
glucose and yeast extract concentration of 100 and 4 giL, respectively,
only 26% of the yeast extract dosage required for batch fermentation was
needed to produce 1 kg of lactic acid, whereas the lactic acid concentra-
tion in the product stream (92-94 giL) and the lactic acid productivity
(6.03-6.20 g/[L·hD were similar to those of batch operations. Further-
more, long-term stability was established.

Acknowledgments
This work was supported by the Ministry of Commerce, Industry and
Energy through the Korea Institute of Industrial Technology Evaluation
and Planning.

References
1. Hofvendahl, K. and Hahn-Hagerdahl, B. (2000), Enzyme Microb. Technol. 26,87-107.
2. VickRoy, T. B. (1985), in Comprehensive Biotechnology, vol. 3, Moo-Young, M., ed.
Pergamon Press, Tarrytown, NY, pp. 761-776.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Lactic Acid Production by E. faecalis RKYI 613
3. Amass, W., Amass, A, and Tighe, B. (1998), Polym. Int. 47,89-144.
4. Chung, E. c., c.,
Kim, S. SO, J. 5., and Yun, H. S. (2001), Biotechnol. Bioprocess Eng. 6,
128-132.
5. Olmos-Dichara, A, Ampe, A, Uribelarrea, J.-1., Pareilleux, A, and Goma, G. (1997),
Biotechnol. Lett. 19(8),709-714.
6. Kulozik, W. and Wilde, J. (1999), Enzyme Microb. Technol. 24,297-302.
7. Hujannen, M. and Linko, Y.-Y. (1996), Appl. Microbiol. Biotechnol. 45,307-313.
B. Amrane, A and Pringent, Y. (1997),]. Biotechnol. 55,1-8.
9. Tejayadi, s. and Cheryan, M. (1995), Appl. Microbiol. Biotechnol. 43, 242-248.
10. Javanainen, P. and Linko, Y.-Y. (1995), Biotechnol. Technique 9(8), 543-548.
11. Hsiesh, C. M., Yang, F.-C., and Iannotti, E. 1. (1999), Process Biochem. 34, 173-179.
12. Amrane, A (2000), World]. Microbiol. Biotechnol. 16,207-209.
13. Bibal, B., Vayssier, Y., Goma, G., and Pareilleux, A (1991), Biotechnol. Bioeng. 37,
746-754.
14. Ye, K., Jin, 5., and Shimizu, K. (1996),]. Chern. Technol. Biotechnol. 66,223-226.
15. Yun, J. S. and Ryu, H. W. (2001), Process Biochem. 37,235-240.
16. Kang, K. H. and Ryu, H. W. (1999),]. Microbiol. Biotechnol. 9(2), 191-195.
17. Kang, K. H., Yun, J. 5., and Ryu, H. W. (2000),]. Microbol. Biotechnol. 10(1),1-7.
lB. Ryu, H. W., Kang, K. H., Pan, J. G., and Chang, H. N. (2001), Biotechnol. Bioeng. 72(1),
119-124.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0615/$20.00

Limits for Alkaline Detoxification


of Dilute-Acid Lignocellulose Hydrolysates

NllS-OlOF NllVEBRANT, *,1 PER PERSSON/ ANDERS REIMANN, l


FllIPE DE SOUSA, l Lo GORTON, 2 AND LEIF J. JONSSON 3
ISTFI, Swedish Pulp and Paper Research Institute,
PO Box 5604, SE-114 86 Stockholm, Sweden,
E-mail: non@stfi.se;
2Department of Analytical Chemistry,
Lund University, PO Box 124, SE-221 00 Lund, Sweden; and
3Biochemistry, Division for Chemistry,
Karlstad University, SE-651 88 Karlstad, Sweden

Abstract
In addition to fermentable sugars, dilute-acid hydrolysates of lignocel-
lulose contain compounds that inhibit fermenting microorganisms, such
as Saccharomyces cerevisiae. Previous results show that phenolic compounds
and furan aldehydes, and to some extent aliphatic acids, act as inhibitors
during fermentation of dilute-acid hydrolysates of spruce. Treatment of
lignocellulose hydrolysates with alkali, usually in the form of overliming
to pH 10.0, has been frequently employed as a detoxification method to
improve fermentability. A spruce dilute-acid hydrolysate was treated
with NaOH in a factorial design experiment, in which the pH was varied
between 9.0 and 12.0, the temperature between 5 and 80°C, and the time
between 1 and 7 h. Already at pH 9.0, >25% of the glucose was lost when
the hydrolysate was treated at 80°C for 1 h. Among the monosaccharides,
xylose was degraded faster under alkaline conditions than the hexoses
(glucose, mannose, and galactose), which, in turn, were degraded faster
than arabinose. The results suggest that alkali treatment of hydrolysates
can be performed at temperatures below 30°C at any pH between 9.0 and
12.0 without problems with sugar degradation or formation of inhibiting
aliphatic acids. Treatment with Ca(OH)z instead of NaOH resulted in more
substantial degradation of sugars. Under the harsher conditions of the
factorial design experiment, the concentrations of furfural and 5-hydro-
xymethylfurfural decreased while the total phenolic content increased.
The latter phenomenon was tentatively attributed to fragmentation of
soluble aromatic oligomers in the hydrolysate. Separate phenolic com-
pounds were affected in different ways by the alkaline conditions with

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 615 Vol. 105-108,2003


616 Nilvebrant et al.
some compounds showing an increase in concentration while others
decreased. In conclusion, the conditions used for detoxification with alkali
should be carefully controlled to optimize the positive effects and mini-
mize the degradation of fermentable sugars.
Index Entries: Bioethanol; lignocellulose hydrolysates; fermentation inhibi-
tors; alkali detoxification; sugar degradation.

Introduction
Consumption of fossil fuels in the transportation sector generates
carbon dioxide, which may act as a greenhouse gas promoting global warm-
ing. Ethanol produced from renewable resources serves as an alternative to
fossil fuels and may not result in a net addition of carbon dioxide to the
atmosphere. Ethanol can also be used as an additive to gasoline instead of
methyl tert-butyl ether. Lignocellulose (e.g., forestry wastes), is an abun-
dant and inexpensive raw material considered for production ofbioethanol
(1). Cellulose and hemicellulose can be degraded to monosaccharides that
are subsequently converted to ethanol by microorganisms, such as baker's
yeast, Saccharomyces cerevisiae.
One problem associated with degradation of lignocellulose polysac-
charides to monosaccharides by dilute-acid hydrolysis is the simultaneous
formation of compounds that are toxic to the fermenting organism. These
inhibitory compounds include furan derivatives, phenolics, and aliphatic
acids (2-11). Since the inhibitors decrease the ethanol productivity and
consequently make the production of ethanol more expensive, it is desir-
able to decrease the concentration of the toxic compounds in the hydroly-
sates prior to fermentation. Various detoxification methods for removal of
inhibitors have previously been carefully investigated, such as alkaline
treatment at pH values of about 10, enzymatic treatment, and treatment
with ion-exchange resins (5,7-9). Overliming-i.e.,treatmentwithCa(OH)2
at pH values of about 10, and sometimes at elevated temperature-is a
widely used method for detoxification of lignocellulose hydrolysates
(5,7,8,10,12). However, the center of interest in these investigations has
been on fermentability as well as on concentration of inhibitors before and
after treatment, and little effort has been made to optimize the parameters
during alkali treatment. A possible drawback with alkali detoxification
that has so far not been carefully investigated is the concomitant decrease
in the concentration of fermentable sugars.
It is well known that sugars degrade in aqueous alkaline solutions (13-
19). The parameters that influence the different degradation pathways and
rearrangements of sugars is also known. However, most of the investiga-
tions (14-19) providing this evidence were performed on standard solution
mixtures, and therefore, the knowledge gained might not necessarily apply
for dilute-acid hydrolysates because of their complex nature. Furthermore,
alkali treatment ofhydrolysates has proven to be an efficient detoxification
method; thus, it would be desirable to perform an investigation to clarify

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Alkaline Detoxification of Hydrolysates 617

which factors are the most important during the alkali treatment. The main
goal would be to find experimental settings in which the detoxification
effect is at a maximum and in unison with the degradation of fermentable
sugars at a minimum.
In the present study, a factorial design experiment was used to eluci-
date the limits for alkaline detoxification of a dilute-acid hydrolysate of
spruce regarding the individual and synergistic effects of temperature,
concentration of hydroxyl ions (pH), and duration of treatment on the con-
centrations of inhibitors and fermentable sugars. The variables-tempera-
ture, pH, and time-were changed simultaneously. Normally, Ca(OHh is
used in alkali treatments of hydrolysates, so called overliming, but to avoid
solubility problems, NaOH was used to raise and maintain the pH, using
a pH-control unit. In addition, some experiments were performed with
Ca(OH)2 to examine whether sodium and calcium had different effects on
the hydrolysate composition at the applied conditions. The concentrations
of different monosaccharides were analyzed as well as changes in the con-
centration of furan aldehydes, aliphatic acids, total phenolics, and selected
separate phenolic compounds. Finally, the results were evaluated using
response surface modeling.

Materials and Methods


Chemicals
Methanol (p.a.), acetic and formic acid (p.a.), acetonitrile (high-perfor-
manceliquidchromatography[HPLC]quality),4-hydro-xybenzoicacid(for
synthesis), and vanillin (4-hydroxy-3-methoxy-benzaldehyde) (for synthe-
sis) were supplied by Merck (Darmstadt, Germany). Catechol, furfural, and
vanillic acid were from ICN Biomedicals (Aurora, OH). Arabinose (~99%),
galactose (~99%), glucose (~99%), mannose (~99%), xylose (~99%), and cin-
namic acid (~99%) were from Fluka (Buchs, Switzerland). Coniferyl alde-
hyde, S-hydroxymethylfurfural (HMF) (99%), and fucose (~99%) were
purchased from Sigma-Aldrich (Steinheim, Germany). All water used was
of Milli-Q quality, i.e., ion exchanged and filtered to yield a low conductiv-
ity, here 0.05 !lSI cm (Millipore, Bedford, MA).

Preparation of Hydrolysates
A two-step dilute-acid hydrolysate of Norway spruce (Picea abies) was
used. Chipped Norway spruce was impregnated with H 2S04 (0.5% [w Iv])
prior to loading in a 2S0-L batch reactor. Steam at a pressure of 12 bar
(190°C) was loaded and kept for 10 min. Subsequently, the liquid and solid
fractions were separated, after which the solid fraction was washed with
water, reimpregnated with H 2SO4 and loaded into the reactor again. Steam
at a pressure of 21 bar (215°C) was loaded and kept for 10 min. After filtra-
tion, the liquid fractions from steps 1 and 2 were pooled to form the final
hydrolysate. The pH of the final hydrolysate was 1.9.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
618 Nilvebrant et al.
Procedure
Optimal conditions for preventing decomposition of sugars during
alkali detoxification were investigated using a fractional factorial experi-
mental design with MODDE 4.0 software from Umetrics AB (Umea, Swe-
den). The experimental design was used to investigate the influence of pH,
temperature, and duration of treatment. To cover the entire area of interest
with as few experiments as possible, all variables were changed simulta-
neously and the influences of the factors on the responses were investi-
gated (n = 17 and degrees of freedom = 11). All variables were given
minimum and maximum values based on previous knowledge on the topic.
The acidity was varied between pH 9.0 and 12.0, temperature between 5
and 80°C, and time between 1 and 7 h, at the high (H) and low (L) levels with
an additional midpoint (M) at pH 10.5, 42.5°C, and 4 h, respectively. The
MODDE software (Umetrics AB) generated the plan for the factorial
experiment (17 experiments; LLL, HLL, LHL, HHL, LLH, HLH, LHH,
HHH, LMM, HMM, MLM, MHM, MML, MMH, and triplicate of MMM to
yield the reproducibility).
Portions of the hydrolysate (40 mL) were treated according to the plan.
A pH-stat, TIM 900 Titration Manager (Radiometer Analytical A/S,
Copenhagen, Denmark) was used in titration mode where the titrator kept
a constant pH by continuously adding a titrant. The titrant used was 5 N
NaOH. It was necessary to use a pH-stat since degradation of carbohy-
drates will occur with formation of various organic acids that could alter
the pH of the samples during the experiment. The experiments were per-
formed in the presence of air to resemble real-life conditions for detoxifica-
tion of hydrolysates. After treatment the solution volume was adjusted to
100 mL with Milli-Q water prior to analysis.
To compare the effect of using either NaOH or Ca(OHh for the alkali
treatments, an additional set of experiments was conducted at six selec-
tions of pH, temperature, and reaction time. These settings comprised treat-
ment at pH 10.0 and 25 or 60°C for 1 h, pH 9.0 and 80°C for 1 and 7 h, and
pH 10.5 and 42.5°C for 4 and 7 h.
Ultraviolet Absorbance
The absorbance of the hydrolysates (largest peak in the UV spectrum)
was measured at 282 nm using an HP 8453 UV-VIS spectrophotometer
(Agilent, Palo Alto, CA). The hydrolysates were diluted 1000 times with
Milli-Q water prior to determination of absorbance.
Measurement of Total Phenolics
The total amounts of phenolics in the hydrolysates were estimated
with a spectrophotometric method based on the Folin & Ciocalteu reagent
(Sigma, Steinheim, Germany). The samples were diluted 25 times with
Milli-Q water, and 1 mL of the diluted sample was transferred to a 50-mL
volumetric flask, to which 3 mL of the Folin & Ciocalteu reagent (2 mol/L)
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Alkaline Detoxification of Hydrolysates 619
and 30 mL of Milli-Q water were added followed by thorough mixing.
After 5-8 min, 10 mL of 20% sodium carbonate solution was added and the
volume was adjusted to 50 mL. The mixture was stirred for 2 h, after which
the absorbance was measured at 760 nm. The amounts of phenolics were
determined from an external calibration curve based on vanillin, since it
was the most abundant phenol in the hydrolysate.
HPLC-Analysis of Furans and Phenolics
Analysis of furfural, HMF, catechol, coniferyl aldehyde,
4-hydroxybenzoic acid, vanillic acid, vanillin, and cinnamic acid was per-
formed using an HP 1100 Series HPLC-system consisting of an automatic
degasser; a binary pump; an auto-sampler; a UV detector; and an HP 1050
Series diode-array detector, DAD (Hewlett-Packard, Palo Alto, CA). The
analytes were separated on an XTerra™ MS C18, 5 ~m, 2.1 x 150 mm analyti-
cal column withanXTerraMSC18 5 ~,2.1 x 10 mmguard column (Waters,
Milford, MA). A mobile phase gradient, consisting of Milli-Q water and
acetonitrile both containing 75 ~L/L of formic acid, started with 5% aceto-
nitrile for 5 min, after which the acetonitrile content increased linearly to
10% after 10 min; 30% after 20 min; 50% after 40 min; and, finally, 90%
acetonitrile after 50 min. The column was equilibrated with 5% acetonitrile
for 10 min before every injection. The injection volume was 5 ~L. All analy-
ses were performed as duplicates.
Five point calibration curves for each individual compound were used
for quantification. Furthermore, to quantify the furans, the samples had to
be diluted 75 times prior to injection. The UV detector was set at 254 nm and
the DAD was set at 210,254,280, and 330 nm. Wavelengths for quantifica-
tion were selected for each compound on the basis of absorption maximum
for the analyzed compound as well as absorption patterns for coeluting
compounds. The furans, furfural and HMF, as well as vanillic acid and
cinnamic acid were quantified at 254 nm, coniferyl aldehyde and catechol
at 280 nm, 4-hydroxybenzoic acid at 210 nm, and vanillin at 330 nm. More-
over, sum integrations of two retention time windows were performed at
280 nm to obtain rough estimations of the relative furan and phenolic con-
tent of the samples. The first retention time window ranged from 0 to 13.5
min and the second from 13.5 to 50 min. The former gave an approximate
value of the furan content and the latter an approx value of the phenolic
content of the samples.
The analyzed phenolics were chosen owing to their high abundance
in the hydrolysates. Together, they account for >90% of the phenolics
known to be present in this type of hydrolysate (12).
Analysis of Aliphatic Acids
Analysis of aliphatic acids was performed ona Beckman P / ACE MDQ
capillary electrophoresis instrument with a 60 cm x 50 ~m id fused silica
capillary. All samples were filtered through a 0.45-~m Whatman cellulose
acetate filter prior to hydrodynamic injection at 15 psi for 4 s. The voltage

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


620 Nilvebrant et al.
was set to 20 kV at reversed polarity. The electrolyte had a pH of 9.8 and
was composed of 5.0 mM trimellitic acid, 50 mM tris(hydroxymethyl)-
aminomethane, 1.0 mM tetradecyl trimethylammonium bromide, and
0.5 mM Ca2+. The electrolyte was filtrated through a 0.2 -11m cellulose nitrate
filter and degassed using helium before use. Detection was performed
50 cm from the injection site by indirect UV at 220 nm.
Analysis of Sugars
Arabinose, galactose, glucose, mannose, and xylose were determined
by high-performance anion-exchange chromatography with a Dionex DX
500 chromatography system coupled to pulsed amperometric detection
(Dionex ED 40) using a CarboPac PA-1 column (all from Dionex, Sunny-
vale, CA). Initially, the column was activated with a mixture of 200 mM
NaOH and 70 mM sodium acetate for 5 min. The eluent was then changed
to Milli-Q water (1 mLI min) and the sample was injected after 7 min. After
sample injection, an isocratic elution (1 mL/min) with pure Milli-Q water
and post column addition of 300 mM NaOH was applied. L-Fucose was
used as the internal standard.
Results
Hydrolysate
The absolute concentrations of the individual compounds in the
untreated hydrolysate were 21.9 giL glucose, 3.3 giL galactose, 16.4 giL
mannose, 1.7 giL arabinose, 8.5 giL xylose, 3.1 giL acetic acid, 0.9 giL
formic acid, 1.1 giL levulinic acid, 2.0 giL HMF, 0.5 giL furfural, 1.9
mglL catechol, 1.1 mglL cinnamic acid, 54.0 mglL coniferyl aldehyde,
39.2 mglL 4-hydroxybenzoic acid, 17.3 mglL vanillic acid, and 107.3
mglL of vanillin. The initial concentration of total phenolics measured by
the Folin & Ciocalteu method was 3.1 giL when vanillin was used as the
standard. All results presented in the next section were calculated by the
MODDE software, which generated models by a multivariate equation fit.
Values for the fit were given as the explained variation (R2) and the pre-
dicted variation (Q2). All changes in concentrations provided later in the
text are given in percent of the initial values. Furthermore, all reported
significance values in the text are based on a 95% confidence interval.
Factorial Design Using NaOH
The models had very good fits with R2 and Q2 values ranging from 0.8
to 0.95 for the different monosaccharides (arabinose, galactose, glucose,
mannose, and xylose). Furthermore, increased temperature and pH had
significant effects, temperature being more important, on the degradation
of all sugars according to the coefficient plots.
All sugars were degraded already after 1 h at the highest tempera-
ture and pH applied. However, more moderate conditions indicated that
the degradation of the individual sugars was somewhat more extensive
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Alkaline Detoxification of Hydrolysates 621

the longer time the treatment lasted (data not shown). Moreover, degra-
dation of all the individual monomeric sugars followed practically the
same pattern at the various temperatures and pH values (Fig. 1A). How-
ever, xylose had a tendency to degrade slightly more easily than the other
sugars at increased temperature and pH, while arabinose tended to be the
most resistant (Fig. 1B,C). Furthermore, the individual sugar concentra-
tions seemed to be independent of the pH at low temperatures (5°C), but
the pH dependency increased with increased temperature (Fig. 1A). That
is, higher pH resulted in more extensive degradation; all sugars totally
disappeared at 80°C and pH 12.0, whereas 40-50% disappeared after treat-
ment at 80°C and pH 9.0.
The results on aliphatic acids showed that increased temperature
and pH had significant effects, temperature being more important, on
the formation of formic acid, but only temperature had a significant effect
on the formation of acetic acid and neither temperature nor pH had sig-
nificant effects on the concentration of levulinic acid according to the
coefficient plots (not shown). The model had a very good fit for formic
acid, with an R2 value of 0.92 and a Q2 value of 0.80, an acceptable fit for
acetic acid (R2 = 0.68 and Q2 = 0.36), whereas there was a poor fit for
levulinic acid.
Formic acid and acetic acid showed an increase in concentration at
elevated temperature and pH according to the response surfaces (Fig. 2A,B).
However, there was an approx fourfold increase in formic acid and only a
twofold increase in acetic acid at the harshest conditions (80°C and pH
12.0). Furthermore, formic acid and acetic acid were not degraded to any
larger extent since the relative concentration never fell below 100% in the
response surface plots (Fig. 2A,B).
The concentration changes in levulinic acid were far more moderate
than those of acetic and formic acid, that is, the concentration changes were
no more than ±20% around the initial value for levulinic acid (Fig. 2C).
Both the temperature and pH were significant for the changes in con-
centrations of furfural and HMF according to the coefficient plots (not
shown). The R2 and Q2 values were 0.85 and 0.58 for furfural and 0.81 and
0.48 for HMF. Both furfural and HMF were more easily degraded, probably
to acetic and formic acid, when pH was increased from 9.0 to 12.0 and the
degradation increased with increasing temperature Fig. 3A,B). Furfural
and HMF were absent after 1 h at 80°C and pH 12.0.
Regarding separate phenolic compounds, determined with HPLC,
the temperature and pH were only significant (R2 =0.87 and Q2 =0.68) for
the concentration of vanillic acid. Roughly a threefold increase in the con-
centration of vanillic acid was observed at 80°C and pH 12.0 (Fig. 3C).
According to the UV measurements at 282 nm both the temperature
and pH (R2 =0.87 and Q2 =0.68) were significant for the concentration of the
compounds that absorbed light at this wavelength. At low temperatures, a
decrease in absorptivity from 100 to 70% was observed when the pH was
increased from 9.0 to 12.0 (Fig. 4A). When the temperature was increased
Applied Biochemistry and Biotechnology Vol. 105-108,2003
~ A B C
~
Q.. 100
t1:J IS tOO c: 100
'Q 0 IS
';:1 .:z
~ § 80 80
i fi
3iii' 60 60
~ 60
~ iii ~ _. - Alabinose
~
tu .:z iii 40 - • - Alabin_
::l ';:I 40 ......... Oalactose
] ......... Galactose
Q.. :§ :§
... :§ --01_ -01_
t1:J 0 12 ... ...
o· 0 0 20
~ 20 - - - Mannose - - - MIII\lIose
-;t. ~
~ o ....•.•.. Xylose o ....•.•.• Xylose
::l
o to 20 30 40 50 60 70 80 9 10 11 12
~ Temperature ("C) pH
'<:

Fig. 1. Response surface (4-h treatment) for glucose (A) as well as models for degradation of monosac-
0"1
charides as a function of increasing temperature at pH 10.0 for 1 h (B) and increasing pH at 60°C for 1 h (C).
N
N

A B C
400 CI
120
IS IS 0
.:z .:z '.c!

i .sc 110
I 300
~ ~
8 8 8 100
iii 200 iii 140 iii
'Q .:z '.c!
:!11 :§ :§
ci=
:- ~
....0
12
...
0
~ -;t. ~
-~ 2~
- 60 9
~ Temperature ("C) 80
§'" Fig. 2. Response surfaces (4-h treatment) for (A) formic, (B) acetic, and (C) levulinic acid.
Alkaline Detoxification of Hydrolysates 623
from SoC, there was an initial decrease in absorbance. By contrast, at tem-
peratures above 40°C the absorbance increased drastically at high pH.
The content of phenolic compounds could be estimated by integration
of the area in the HPLC chromatogram corresponding to compounds elut-
ing after 13.5 min. High temperature and pH gave an increase in total
phenolics (Fig. 4B).
The Folin & Ciocalteu measurements, used to estimate the total con-
centration of phenolics, largely showed an increase with increased tem-
perature and pH (Fig. 4C). The temperature and pH were significant (RZ =
0.94 and Q2 = 0.86) for the total phenolic content. A fivefold increase in
phenolic concentration was seen under the harshest conditions, 80°C and
pH 12.0. However, note that a small decrease could be found at 2SoC and
pH 10.0, i.e., under normal overliming conditions.
Comparison of NaOH and Ca{OH)2
Under mild conditions, no significant degradation of sugars could be
determined with either NaOH or Ca(OH)z. At 60°C and pH 10.0, approx
10% of the monosaccharides were lost and a small difference between
NaOH and Ca(OHh was observed (Table 1). When the conditions were
more severe, Ca(OH)z treatment clearly resulted in more extensive degra-
dation than treatment with NaOH (Table 1).

Discussion
It is well known that sugars can mutarotate under mild alkaline con-
ditions in aqueous solutions. Moreover, it is also known that harsher alka-
line conditions catalyze enolization reactions, provided the sugars have an
a-hydrogen atom, which, in tum, might lead to isomerization reactions
(13,20). Furthermore, ~-elimination reactions, leading to the formation of
dicarbonyl compounds, might occur simultaneously or subsequent to an
enolization reaction (13,20). In time, the ~-eliminations will lead to forma-
tion of saccharinic acids (13,20). In addition to saccharinic acids, compounds
with fewer carbon atoms than the monosaccharides, such as formic acid,
are formed. These reactions can occur through retroaldolization as well as
splitting or benzilic acid rearrangement of dicarbonyl compounds (13,14).
The degree of degradation or rearrangement of the sugars depends on
a variety of factors, such as concentration and type of sugar, temperature,
duration of treatment, as well as concentration and type of alkali (14,16,18-
20). Because of the high reactivity of ketoses, they appear to be more impor-
tant as intermediates in degradation of monosaccharides than aldoses.
Furthermore, enolization seems to be the rate-limiting step during isomer-
ization and degradation of monosaccharides in aqueous alkaline solutions
(15). Higher pH values and the presence of calcium ions increase the eno-
lization and retroaldolization rate (14,15).
The most critical area of question of this study was the possible degra-
dation of fermentable sugars during the treatments. When the temperature
Applied Biochemistry and Biotechnology Vol. 105-108,2003
).. 0'1
:g A B C N
~ ~
[ c 340
0
OJ '0
0' c
n ,8 80 ~
c
::,- '.g, 80 ...
~ E60 ~6O 8
<n' 8 g 240
~ "ii
8 40 '0
~4O .-J
;:,
"" Ol 20 ~ 20 :5
Q.. '0 ...
12 0
OJ :5 0 12 :9 0 140
0' ... .... -;;.
0 0
<ii
g. -;;. -;;.
;:,
o 64
Temperature ("C) Temperature ("C)
~
Fig. 3. Response surfaces (4-h treatment) for (A) furfural, (B) HMF, and (C) vanillic acid.
0'1
i',.)
~
A B C
160
...... 8 500 8 500
~ 400 ~ 400
120
j. ~ 300
.. ~ 300
.
«I
] '0 ~ 200
:s :9.... :5
....
...
0 0
100 12 0
?
"I- -;;. -;;.
~ ~
0-
~ 60 iil
Temperature ("C) :::J
......
~ !1l
N ......
Fig. 4. Response surfaces (4-h treatment) for (A) UV absorbance at 282 nm, (B) total phenolics determined by ~
§ HPLC (after 13.5-min elution), and (C) total phenolics determined by the Folin & Ciocalteu method. :-
Alkaline Detoxification of Hydrolysates 625
Table 1
Comparison of Relative Concentrations of Monosaccharides
(Percent Compared to Untreated Hydrolysate)
After Treatment with NaOH or Ca(OH)2 at Various Settings
Temperature Time
Alkali pH (0C) (h) Arabinose Galactose Glucose Mannose Xylose
NaOH 10.0 25 1 97 97 97 97 98
Ca(OHh 10.0 25 1 99 99 100 100 100
NaOH 10.0 60 1 92 92 89 87 86
Ca(OH)2 10.0 60 1 88 83 83 81 77
NaOH 9.0 80 1 85 78 76 71 67
Ca(OH)2 9.0 80 1 74 61 60 59 49
NaOH 9.0 80 7 59 42 45 38 20
Ca(OH)2 9.0 80 7 21 8 12 8 4
NaOH 10.5 42.5 4 89 84 80 79 73
Ca(OH)2 10.5 42.5 4 85 83 79 88 78
NaOH 10.5 42.5 7 81 79 75 74 69
Ca(OHh 10.5 42.5 7 65 46 65 63 42

exceeded 30°C, an extensive degradation of the different sugars started


and pH became an important factor in the degradation (Fig. 1). The higher
the pH, the more extensive was the degradation. Note that both hexoses
and pentoses were degraded similarly (Fig. 1). These results should be
compared with the temperature and pH values of previously reported
overliming experiments; pH 10.0 at room temperature (5), pH 9.4-10.7 at
25°C or pH 7.5-9.B at 60°C (7), and pH 10.0-10.5 at 50°C (10). The formation
of inhibiting compounds such as aliphatic acids, furan derivatives, and
possibly also phenolic compounds during the treatments was also impor-
tant. According to the response surfaces (Fig. 2), there was a clear increase
in acetic and formic acid concentration when the temperature and pH were
raised. Because the initial concentration of the sugars was much higher
than the concentration of acids, partial sugar degradation resulted in major
changes in the concentration of formic and acetic acid. In the case of levulinic
acid, the picture was somewhat more complicated. Here, it seems that the
rate of breakdown oflevulinic acid was equal to the rate of formation owing
to breakdown ofhexoses. A difference between levulinic acid, on one hand,
and formic and acetic acid, on the other, is that levulinic acid has a keto
group that could be attacked by hydroxyl ions at higher pH values.
Because the aliphatic acids might act as inhibitors (4), it would be
desirable to keep the concentration of these compounds low. The response
surfaces (Fig. 2) suggestthatthis could be accomplished at any pH between
9.0 and 12.0 as long as the temperature does not exceed 30°C. It was clear
that a general degradation of furaldehyde and HMF took place at high
temperature and pH (Fig. 3A,B).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


626 Nilvebrant et al.
Forsskcihl et al. (17) detected 11 different forms of phenolics after treat-
ment of separate solutions of o-glucose and o-xylose under nitrogen with
0.63MNaOH for4h at 96°C. This finding implies that phenolics potentially
could be formed in lignocellulose hydrolysates during alkali treatment as
a result of degradation and rearrangement of monosaccharides.
The effect of the various treatments on the concentrations of individual
phenolics was difficult to interpret since the response surfaces showed a
great discrepancy except for the concentration of vanillic acid (Fig. 3C).
Nevertheless, there was an obvious formation of vanillic acid at increased
temperatures and pH values. Vanillic acid could possibly be formed through
oxidation of vanillin, however, the response surface for vanillin did not show
any evidence for this. Furthermore, air was present during the treatments,
which could cause oxidation of the phenolics. The oxidations could in tum
result in further reactions and/or degradation of phenolics. In addition,
soluble lignin fragments were probably present in the hydrolysate. These
lignin oligomers could most likely be degraded at high pH and cause forma-
tion of a variety of phenolic compounds at the applied conditions. gel perme-
ationchromatography analyses have shown that sprucehydrolysates contain
soluble oligomeric compounds absorbing UV light (unpublished results).
Different aromatic compounds are known to have very different
inhibitory effects on S. cerevisiae (6). Therefore, interconversion between
specific phenolic compounds may be of major relevance regarding the
combined inhibitory effect of the total phenolic content. The data indi-
cated that treatment under alkaline conditions resulted in a decrease in
the concentration of some phenolic compounds, such as coniferyl alde-
hyde, and an increase in the concentration of others, such as vanillic acid.
Because the response surfaces of separate phenolic compounds gen-
erally were hard to evaluate, a sum integration was performed in a reten-
tion time window (>13.5 min) in the UV chromatogram (280 nm)
corresponding to the retention times where the monomeric phenolics
eluted, according to the HPLC analyses. The response surface that re-
sulted from the sum integration showed an increase in total phenolic
content with increasing temperature and pH (Fig. 4B).
The Folin & Ciocalteu method showed a very clear increase in phe-
nolic content at high temperature and pH (Fig. 4C). There was a fivefold
increase in phenolic concentration at the harshest conditions compared to
the initial concentration. However, in agreement with previous results
(5,12) the formation of phenolics did not start at temperatures below 30°C.
Consistent with the results of the Folin & Ciocalteu measurements, the
sum integrations at 280 nm showed a fivefold increase in phenolic concen-
tration. It is reasonable to state that the formation of phenolics became
essential when the temperature exceeded 30°C. The decrease in total phe-
nolic content observed under milder conditions, approx 25°C and pH 9.0-
10.0, is in agreement with previous results (5,12).
UV absorbance is known to be a good tool to obtain approximate
values for the furan content of acidic hydrolysates (21). The UV response
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Alkaline Detoxification of Hydrolysates 627
surface (Fig. 4A), showed an initial decrease in absorptivity when the tem-
perature was raised. However, at temperatures above 40°C the absorptiv-
ity increased. On the other hand, the concentrations of furfural and HMF
decreased at high temperature (Fig. 3A,B). Hence, the absorbance at 280 nm
cannot be solely attributed to furfural and HMF. Consequently, compounds
that could account for the increase in absorptivity must have been formed
during the treatments. These compounds probably encompassed phenolic
compounds with high molar absorptivity around 280 nm since the phenolic
retention time window in the HPLC separation (Fig. 4B) showed a clear
increase in phenolic concentration at the higher temperatures and pH val-
ues, as did the Folin & Ciocalteu measurements (Fig. 4C).
Treatment with Ca(OHh resulted in more substantial sugar degrada-
tion than N aOH (Table 1). This could probably be attributed to the catalytic
effect of calcium ions on the enolization reaction (14,15), which may influ-
ence both the kinetics of the following benzilic acid rearrangement and the
product pattern.
No analyses of inhibitors were performed after the Ca(OH)2 treatments.
However, previous results indicate that sodium and calcium hydroxide
have similar effects on the concentration of inhibitors (5).

Conclusion
The sugars in a dilute-acid hydrolysate of spruce treated with NaOH
showed no major changes in concentrations between pH 9.0 and 12.0 as
long as the temperature did not exceed 30°C. However, at 80°C and pH 12.0
all the sugars were completely degraded. Substituting NaOH for Ca(OH)2
led to more rapid sugar degradation. Along with the degradation of the
sugars, furfural and HMF were degraded whereas acetic and formic acid
were formed. However, the concentration of levulinic acid did not increase.
Separate phenolic compounds were affected differently by treatment with
alkali. Severe conditions resulted in an increase in total phenolics, possibly
as a combined result of sugar degradation and fragmentation of lignin
oligomers. The results suggest suitable limits for the conditions that should
be used for treatment with alkali and reveal the effect of the treatment on
specific sugars and inhibitory compounds.

Acknowledgments
This work was supported by grants from the Swedish National Energy
Administration.

References
1. Wheals, A. E., Basso, L. c., Alves, D. M. G., and Amorim, H. V. (1999), Trends
Biotechnol. 17,482-487.
2. Ando, 5., Arai, I., Kiyoto, K, and Hanai, S. (1986), J. Ferment. Technol. 64,567-570.
3. Clark, T. A. and Mackie, K L. (1984), J. Chern. Technol. Biotechnol. 34B, 101-110.
4. Larsson,S., Palmqvist, E., Hahn-Hagerdal, B., Tengborg, c., Stenberg, K, Zacchi, G.,
and Nilvebrant, N.-O. (1999), Enzyme Microb. Technol. 24, 151-159.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


628 Nilvebrant et al.
5. Larsson,S., Reimann, A, Nilvebrant, N.-O., and Jonsson, L. J. (1999), Appl. Biochem.
Biotechnol. 77-79,91-103.
6. Larsson,S., Quintana-Sainz, A, Reimann, A, Nilvebrant, N.-O., and Jonsson, L. J.
(2000), Appl. Biochem. Biotechnol. 84-86,617-632.
7. Martinez, A, Rodriguez, M. E., York, S. W., Preston, J. F., and Ingram, L. O. (2000),
Biotechnol. Bioeng. 69, 526-536.
B. Martinez, A, Rodriguez, M. E., Wells, M. L., York, s. W., Preston, J. F., and Ingram,
L. O. (2001), Biotechnol. Prog. 17,287-293.
9. Nilvebrant, N.-O., Reimann, A, Larsson,S., and Jonsson, L. J. (2001), Appl. Biochem.
Biotechnol. 91-93,35-49.
10. Ranatunga, T. D., Jervis, J., Helm, R. F., McMillan, J. D., and Wooley, R. J. (2000),
Enzyme Microb. Technol. 27,240-247.
11. Ranatunga, T. D., Jervis, J., Helm, R. F., McMillan, J. D., and Hatzis, C. (1997),
Biotechnol. Lett. 19,1125--1127.
12. Persson, P., Andersson, J., Gorton, L., Larsson,S., Nilvebrant, N.-O., and Jonsson, L.
J. (2002), ]. Agric. Food Chern. 50(19),5318-5325.
13. Pigman, W. and Anet, E. F. L. J. (1972), in The Carbohydrates, vol. lA, Pigman, W. and
Horton, D., eds., Academic Press, New York, NY, pp. 165--194.
14. De Bruijn,J.M.,Kieboom,A P.G.,and VanBekkum,H. (1986), Reci. Trav. Chim. Pays-
Bas 105, 176-183.
15. De Bruijn, J. M., Kieboom, A P. G., and Van Bekkum, H. (1987), Reci. Trav. Chim. Pays-
Bas 106, 35-43.
16. De Bruijn, J. M., Kieboom, A P. G., and Van Bekkum, H. (1987), StarchlStaerke 39,
23-28.
17. Forsskcihl, I., Popoff, T., and Theander, O. (1976), Carbohydr. Res. 48, 13-21.
lB. Yang, B. Y. and Montgomery, R. (1996), Carbohydr. Res. 280,27-45.
19. Yang, B. Y. and Montgomery, R. (1996), Carbohydr. Res. 280,47-57.
20. De Bruijn, J. M., Kieboom, A P. G., and Van Bekkum, H. (1986), Sugar Technol. Rev.
13,21-52.
21. Martinez, A, Rodriguez, M. E., York, S. W., Preston, J. F., and Ingram, L. O. (2000),
Biotechnol. Prog. 16, 637-641.

Applied Biochemistry and Biotechnology Vol. 105-708,2003


SESSION 4
Genetics and Genomics
in Bioenergy and Bioproducts
Session 4

Genetics and Genomics


in Bioenergy and Bioproducts

JAMES s. McLAREN 1 AND THOMAS W. JEFFRIES 2


llnverizon International Inc., Chesterfield, MO; and
2USOA Forest Products Laboratory, Madison, WI

There is now widespread acknowledgment that renewable bio-


resources, as a platform for raw inputs and as a source of powerful
biocatalysts, have considerable potential to support sustainable economic
growth, increase national energy security, strengthen rural communities,
and minimize anthropogenic effects on the environment. Clearly, the tran-
sition to renewable resources from fossil fuels must meet certain replace-
ment criteria in the economic arena. However, significant advances are
required in science and technology development to help meet such criteria
and to ensure sustainable enterprises.
As we enter the 21"1 century, biotechnology is one such advance that
is poised to become a major platform for driving significant progress over
the next 20-50 years. The knowledge and understanding of genome
sequences, gene function, gene expression, protein interactions, and meta-
bolic control mechanisms, will allow a sound scientific basis for a healthier,
more reliable food supply combined with much improved design of renew-
able resources. The panel participants in this special session provided a
state-of-the-art overview of how such advances are being utilized and ap-
plied in a range of renewable resource projects.
Tom Jeffries (USDA) discussed the genome-wide expression analysis
of Saccharomyces metabolically engineered for the fermentation of xylose. S.
cerevisiae transformed with the Pichia stipitis genes for xylose reductase
(XYLl), xylitol dehydrogenase (XYL2), and D-xylulokinase (XYL3) will
assimilate xylose under aerobic conditions and will produce limited
amounts of ethanol. However, they do not metabolize xylose anaerobically
and they tend to accumulate xylitol. Affymatrix genechip studies showed
that even though these yeasts have the enzymatic machinery necessary for
xylose assimilation, they do not possess the regulatory mechanisms to rec-
ognize xylose as a fermentable sugar. Metabolism on xylose is oxidative.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 631 Vol. 105-108,2003


632 Introduction to Section 4

The great strength of genome-wide expression analysis is being able to


recognize regulatory patterns.
Frank Larimer (ORNL) covered the role of annotation in microbial
genomics, highlighting the important effort required between having a
draft sequence and being able to make a function call for individual genes.
High-throughput draft sequence information costs were reported at $0.04/
base for a lOX coverage, while the cost rises to approx $0.1O/base for fin-
ished, annotated sequence. Adding important functional information in-
cludes assembling inputs from comparative genomics, proteomics, and
cross-checking via reconstruction of metabolic activity. The annotated
output has implications for designing oligonucleotide arrays, building
protein expression constructs, understanding functional clustering, and
uncovering metabolic networks.
Sharon Shoemaker (UC-Davis) discussed genomics and proteomics
as tools for biocatalysis, including the importance of interrelationships
between genomic and proteomic approaches. While the genomics revolu-
tion is well underway, the proteomics market is less well developed but is
expected to grow to $2.8 billion by 2005. Examples presented included the
use of yeast proteome microarrays as screening tools, and the application
of exploratory tools to improve the cellulose multienzyme complex for the
conversion of cellulose to cellobiose and glucose. A major conclusion was
that a combination of approaches may be required to decrease cellulose
cost to the economic threshold, including mining natural diversity, gene
shuffling, directed evolution, and site-directed mutagenesis.
Jerry Tuskan (ORNL) presented the status of the Populus genome
project, including plans to add additional shot-gun cloning and BAC-end
sequencing to provide up to 6X total coverage. The overall objective is to
improve poplar hybrids as renewable sources for bioenergy and bio-based
products. Approaches discussed included molecular beam mass spectro-
metric analysis of lignin, population genetics, microarrays, and compara-
tive genomics for gene identification. The multidisciplinary work integrates
the information on tree feedstock quality with map-based cloning in order
to improve germplasm stocks for future commercial use.
Jim McLaren (lnverizon) discussed the development of sorghum as a
renewable bioresource, through genomics-based tools to enhance starch
as a feedstock for a future glucose sugar platform. Initial steps include the
application of methyl-filtration to select the genespace for sequencing and
the integration of this information with quantitative trait loci, and expres-
sion microarrays. Starch-enhanced sorghum hybrids, with elite agronomic
characteristics, will be processed via selected biocatalytic technologies in
a modified milling system. The integration of these approaches is expected
to allow an improved biorefinery design, with an output portfolio that
includes biofuels and a selected range of economically acceptable bio-
based products.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


SESSION 5
Development of Biobased Products
Session 5

Development of Biobased Products

Bio-based product development is an old concept that is rapidly


expanding and entering new levels of commercialization. The current goal
is economic and renewable suites of products from new biorefineries,
which will include production of primary products and co-products
together. The primary bio-based products can include oils, commodity or
specialty chemicals, and materials. Combined with power and fuel etha-
nol, bio-based products are critical for enhanced value in the various mixed
product streams envisioned emerging from either current agricultural
products, agricultural and forest residues; or direct biomass crops. This
session, organized by Dr. David Glassner (Cargill Dow, Minnetonka, MN)
and Dr. Todd Werpy (Pacific Northwest National Laboratory, Richland,
WA) focused the oral session on examples of methods to capture this
potential mix of products.
Kyle Beery of Archer-Daniels Midland Company gave a progress
report on a hydrolysis sugar platform from corn fiber. Key issues were
hydrolysis methods, inhibitors, and the variability of corn fiber. David
Glassner of Cargill Dow presented a corporate vision of a sustainable
industry building on their lactide fermentation platform. Their 400 mil-
lion lb/year lactic acid plant is under construction with a planned start
date in early 2003. Kevin Gray of Diversa Corporation spoke of novel
genetic techniques including gene shuffling to harness the broad diver-
sity that exists in culture libraries. In particular, he spoke to issues of
improving functional hydrolytic enzymes (e.g., cellulases and xylanases)
for biomass conversion. Brian Carr of Athenix Corporation also spoke on
the power of genetic engineering, rational genomics and screening to
improve biomass conversion and enzymes.
Curtis Scharf of Terresolv Technologies focused on another bio-based
platform-vegetable oils. He reported on properties development and for-
mulations to allow expanded uses of biodegradable bio-based oils as lubri-
cants that can match or exceed petroleum oil properties. James Barber of
Metabolix then reported on polymer production (polyhydroxyalkonates)

Applied Biochemistry and Biotechnology 635 Vol. 105-108, 2003


636 Introduction to Section 5
by fermentation or, ultimately, directly in genetically engineered plants.
Glucose can also be thermochemically converted by hydrogenation into
sorbitol, followed by dehydrogenation into isosorbide, another renewable
monomer. Dennis Magyar described DuPont's process for this conversion,
emphasizing the importance of purification at each step. The process is
operating in France to make polyethylene isosorbide terephthlate where
the isosorbide ratio can allow varied properties.
The oral presentations were followed by 20 posters covering other
aspects of bioproducts and biorefineries. These included process integra-
tion, economics, life-cycle, and biorefinery modeling and assessment.
Bioproducts ranged from ethanol, glycerol, propandiol, bio-octane, lactic
acid, acetic acid and PHAs as well as amino acids and specialty materials.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0637/$20.00

Production of Lactic Acid from Food Wastes

KWANG It KIM,l Woo KVUNG KIM,l DEOK KI SEO,l


IN SANG YOO,l EUN KI KIM/ AND HVON HEE YOON*,l
Department of Chemical Engineering, Kyungwon University,
1

Sungnam, Korea 461-701, E-mail: hhyoon@kyungwon.ac.kr; and


2Department of Biological Engineering,
Inha University, Incheon, Korea 402-751

Abstract
Conversion of food wastes into lactic acid by simultaneous saccharifica-
tion and fermentation (SSF) was investigated. The process involves sacchari-
fication of the starch component in food wastes by a commercial amylolytic
enzyme preparation (a mixture of amyloglucosidase, a-amylase, and pro-
tease) and fermentation by Lactobacillus delbrueckii. The highest observed
overall yield of lactic acid in the SSF was 91% of theoretical. Lactic acid
concentration as high as 80 giL was attainable in 48 h of the SSF. The opti-
mum operating conditions for the maximum productivity were found to be
42°C and pH 6.0. Without supplementation of nitrogen-containing nutrients,
the lactic acid yield in the SSF decreased to 60%: 27 giL of lactic acid from
60 giL of food waste. The overall performance of the SSF, however, was not
significantly affected by the elimination of mineral supplements.
Index Entries: Food waste; lactic acid; simultaneous saccharificaiton and
fermentation; Lactobacillus delbrueckii; amyloglucosidase.

Introduction
Approximately 5,000,000 t of food wastes are generated annually in
South Korea. Most of the food wastes are landfilled or incinerated, and
groundwater contamination and emission of noxious gases and dioxins
have been frequently cited. Food waste management has therefore been an
important issue from the standpoint of environmental protection and con-
servation of resources.
Among the major components in food wastes are carbohydrates (starch
and cellulose) and proteins. The starch and cellulosic components of the
food waste can be hydrolyzed into monomeric sugars. The sugars from the
food wastes can be utilized as substrates for fermentative production of a
variety of chemicals including lactic acid. Lactic acid is widely used in the
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 637 Vol. 105-108,2003


638 Kim et al.

food and chemical industries. Current demand for lactic acid in production
of biodegradable polymers (polylactic acid) (1) is bringing its status from a
specialty chemical to a commodity chemical.
Traditionally, hydrolysates of corn or potato starch have been used
as the feedstock for the fermentative lactic acid production process (2).
Lignocellulosic biomass has been currently tested as an alternative feed-
stock for lactic acid production (3-5). Municipal solid wastes have also
been evaluated as feedstock forlactic acid (6,7). Recently, Leeetai. (8) and
Loh et al. (9) studied lactic acid production from food wastes and reported
a product concentration of 20-40 giL from direct fermentation of food
wastes lasting 4 to 5 d. Amylolytic lactobacillus strains and other lactic
acid-producing organisms, such as Rhizopus oryzae (10,11), can directly
metabolize starch to produce lactic acid. However, they do so with a very
low fermentation rate giving a relatively low product yield, and low prod-
uct concentration (12). Enzymatic hydrolysis of starch and fermentation
of glucose to lactic acid are well-established industrial processes. Sepa-
rate enzymatic and microbial processing of food wastes can improve lac-
tic acid production over that of direct microbial conversion. In the present
study, we were interested in applying the enzymatic hydrolysis and fer-
mentation for conversion of food wastes to lactic acid in order to achieve
high product yield and production rate. The hydrolysis and fermentation
can be carried out separately or simultaneously. Simultaneous sacchari-
fication and fermentation (SSF) has been extensively investigated in con-
nection with ethanol production from starch or cellulosic feedstock.
Recently, it has been evaluated as a means of producing lactic acid (13).
Our research was undertaken to assess the technical feasibility of an SSF
process based on starch-hydrolyzing enzymes and lactic acid bacteria for
production of lactic acid from food wastes.

Materials and Methods


Substrates
Food wastes were obtained from a university cafeteria (Kyungwon
University, Korea). They consisted mainly of cooked rice, flour products,
vegetables, fishes, and meat. They were ground and kept in a refrigerator
before. The water content of the food waste was about 80% (w Iw). The
wet food waste sample was used directly as a substrate. The food waste
was analyzed for its chemical composition; typical composition is given
in Table 1. Total carbon, nitrogen, hydrogen, and oxygen contents were
analyzed by an elemental analyzer (CHN-lOOO). Mineral elements such as
P, K, Ca, Mg, and Na were analyzed by ICP-AES (JY ULTRACE 138).
Enzymes
A commercial enzyme mixture, SAN Super 240L (Novo Nordisk), was
used. This enzyme preparation consists of mainly amyloglucosidases and
a balanced amount of a-amylases and proteases. The activity of SAN Super
Applied Biochemistry and Biotechnology Vol. 705-108,2003
Production of Lactic Acid from Food Wastes 639
Table 1
Chemical Composition of Food Wastes·
T-C T-N T-H T-O Ash P K Ca Mg Na
pH (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%)
4.8 41.9 2.14 6.40 42.4 1.06 0.23 0.58 0.37 0.07 2.3
"T-C, total carbon; T-N, total nitrogen; T-H, Total hydrogen; T-O, total oxygen.

240L was specified to be 240 AGU ImL by the supplier, in which 1 AGU is
defined as the amount of enzyme hydrolyzing 1 !lmol of maltoselmin at
25°C and pH 4.3. It also contains controlled amounts of fungal amylase and
protease activities.
Microorganisms and Inoculumn Preparation
Homofermentative Lactobacillus delbrueckii NRRL B-445 (KCCM 40069,
renamed as Lactobacillus rhamnosus) was obtained from Korean Culture Cen-
ter of Microorganisms (KCCM). The lyophilized culture was transferred to
plates of solid Elliker broth (Difco) medium. The plates were incubated at
37°C for 36 h. The grown colonies either were used to prepare inoculum or
were stored at 4°C for later use. The microorganism on agar slants was trans-
ferred to a liquid Elliker broth medium and cultivated at 37°C for 36 h to be
used as inoculum in lactic acid fermentation experiments.

Enzymatic Hydrolysis of Food Wastes


The ground food waste sample was diluted with water to a final
concentration between 65 and 150 giL (dry wt), which was the extent of
selected experimental conditions used. SAN Super 240L was added and
the mixture was placed in an incubator shaker maintained at a given
temperature. Samples were withdrawn periodically and boiled for 1 min
to arrest enzyme action. Reducing sugars and glucose were measured.
Experiments were carried out to determine the effect of enzyme dosage
on the glucose yield.
Glucose Fermentation
Glucose fermentation was performed to determine the lactic acid
productivity of the strain used. Fermentation medium contained 15 giL
of yeast extract, 0.5 giL of dipotassium hydrogen phosphate, 0.5 giL of
potassium dihydrogen phosphate, LOlL of sodium acetate trihydrate,
0.5 giL of magnesium sulfate heptahydrate, 0.05 gL of manganese sulfate
monohydrate, 1 giL of ammonium citrate, and 0.03 giL of iron sulfate
heptahydrate. Three weight percent of glucose was added. The fermen-
tation medium was sterilized at 121°C for 20 min. A 10% (v Iv) inoculum
was used in this fermentation. The fermentation was carried out at 4rC
and 150 rpm in a shake flask. An excess of CaC03 powder was used to
neutralize the lactic acid produced.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
640 Kim et al.

Simultaneous Saccharification and Fermentation


The food wastes prepared as described above were used as substrates
in SSF experiments. The same level of nutrient supplements as used in
glucose fermentation was used in this SSF experiment. The SSF solution
was autoc1aved at 121°C for 20 min. It was then inoculated and SSF was
carried out after adding SAN Super 240L enzyme (1.8 mL/100 g of dry
substrate). All experiments, with the exception of the pH optimization
studies, were carried out in 250-mL shake flasks with a 100-mL working
volume at a temperature of 42°C and an agitation rate of 150 rpm, unless
specified otherwise. The initial pH was adjusted to 6.0 and controlled by
the addition of CaC03 powder. An oxygen-free environment was main-
tained by initially sparging CO2 through a 0.3-Ilm filter. A New Brunswick
Bioflo model III fermentor was used for the pH optimization studies. It
was operated with a pH control and at a 1000 mL working volume. The pH
was controlled with 5 N NaOH. Samples were boiled for 1 min to eliminate
any bacterial activity and then stored at 4°C for further analysis.
Analytical Methods
Samples taken from the hydrolysis and fermentation experiments
were analyzed for lactic acid and glucose by a high-performance liquid
chromatography equipped with an refractive index detector. A Bio-Rad-
HPX-87H column was used at 65°C with a 0.005 M H 2S04 mobile phase at
a flow rate of 0.6 mL/min. The total carbohydrate content was deter-
mined by hydrolyzing the food waste to sugars followed by measuring
the concentration (i.e., reducing sugars) of the resulting sugars. Hydroly-
sis was performed by a primary 72% H 2S04 hydrolysis and a secondary
dilute-acid hydrolysis, as described in ref. 14. Reducing sugars were de-
termined according to Nelson (15) using glucose as a standard. The yield
was expressed as a percentage based on initial available sugars.

Results and Discussion


Enzymatic Hydrolysis of Food Wastes
Food wastes were hydrolyzed enzymatically and yields of glucose
were examined under various hydrolysis conditions. From 100 gil of food
waste (dry basis), 75 gil ofreducing sugars was produced by acid hydroly-
sis. This reducing sugar concentration was used as the initial available
sugar concentration.
Figure 1 shows the time course of the saccharification of 65 gil of food
waste at 55°C and pH 6.0 with varying enzyme (SAN Super 240L) concen-
trations. The amyloglucosidase activity of this enzyme was 240 AGU / mL,
as mentioned earlier. The highest glucose yield, about 67%, was obtained
in 24 h with at least 1.8 mL of the enzyme mixture/IOO g of food waste. The
overall optimal working conditions for SAN Super 240L are a temperature
of 40-55°C and a pH of 4.5-6.0, as determined by the supplier. The hydroly-
sis and the SSF experiments were therefore performed at these conditions.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Production of Lactic Acid from Food Wastes 641

Enzyme loading --.- 3.6 mL


____ 1.8 mL
100
(per 100 9 food waste)
~0.5mL
0.2 mL
----T-
--.- 0.1 mL
~BO --.- w/o enzyme
!...
'0
'i BO
':;'
I 40
8:::I
a 20

o 6 12 18 24

Time (h)
Fig. 1. Effect of enzyme loading on the enzymatic hydrolysis of food waste (65 giL)
at 55°C and pH 5.5.

40

-
-
~30

~\;:\
,
C
0
;:
......as
c
20

\
\
II)
u - lactic acid
c --- glucose
o 10
0

0
0 12 24 36 48 60 72

Time (h)
Fig. 2. Time course of lactic acid fermentation with pure glucose (30 giL) by L. del-
brueckii at 42°C and pH 6.0 using two different nitrogen supplements: (e) 15 giL of
yeast extract; (A) 15 giL of peptone.

SSF of Food Wastes


Glucose fermentations were performed to determine the lactic acid
production capability of the strain L. delbrueckii. As shown in Fig. 2,25.5 giL
of lactic acid was produced from 30 giL of glucose in 24 h in the fermen-
tation carried out at 42°C and pH 6.0. When peptone, instead of yeast
extract, was used, the lactic acid production rate decreased significantly,
but almost the same level of the final lactic acid concentration was obtained.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
642 Kim et al.
100

---
0
..J
C)

C
80

60

-
:;
I!
C 40
CD
()
C __ glucose
0 --- lactic acid
0 20

0
0 12 24 36 48

Time (h)
Fig. 3. Time course of batch SSF run with food waste (120 giL) at 42°C.

The typical time course of lactic acid production from food waste by
SSF with SAN SUPER 240L enzyme and L. debrueckii is illustrated in Fig. 3.
During the SSF process, the food waste was saccharified to glucose, and the
glucose was then metabolized by the microorganism and converted into
lactic acid. Accumulation of glucose was seen in the initial phase of the SSF.
The fermentation then proceeded at a rate comparable with the fermenta-
tion rate of lactic acid by the same organisms under similar conditions
using glucose (Fig. 2). At the given SSF conditions, 70.1 giL of lactic acid
was produced from 120 giL of food waste in about 48 h, giving a yield of
78%. From this preliminary experiment, it was clear that food wastes can be
used as a resource for lactic acid production and that SSF could be an
effective bioprocess for producing lactic acid from the food wastes. To
further develop an SSF process for lactic acid production from food waste,
the effect of various operating parameters on performance was examined.
The parameters studied were pH, temperature, substrate concentration,
and nitrogen and mineral supplements.
Effect of pH and Temperature
In the SSF process, both bioreactions of carbohydrate hydrolysis by
enzymes and glucose fermentation by microorganisms occur simulta-
neously. Thus, optimum conditions of both bioreactions should be coinci-
dent for effective SSF operation. Generally, the SSF process requires
operating conditions that represent a compromise between the optimum
conditions of the enzyme and the microorganisms.
The optimum temperature of the L. delbrueckii was reported to be 37°C
by the supplier (KCCM). On the other hand, the optimum temperature for
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Production of Lactic Acid from Food Wastes 643
100 T'"""-------------------,

---
-lactic acid
---glucose
80
...I
C)

C 60
0
!...
A-~

(~~----
C
CD
U
40
I \
~
""-----
,i
c
-- ---- -- ----"1
.:a-_----...
0
0 20
----.............................
----
0
0 12 24 36 48

Time (h)
Fig. 4. Effect of temperature on batch SSF of food waste (145 giL): (e) 35°C;
(_) 45° C; (A) 55°C.

the SAN Super 240L enzyme was specified to be 40-55°C, as stated earlier.
We therefore selected 35, 45, and 55°C for the optimization study. As shown
in Fig. 4, at 35 and 45°C, 79.7 (yield 73%) and 77.2 giL (yield 71 %) of lactic
acid, respectively, were produced from 145 giL of food waste in 48 h. At
55°C, however, the fermentation rate was significantly reduced. The final
lactic acid concentration was reduced to 50.7 giL (yield 47%), leaving a
higher residual glucose concentration within the 48-h experiment. From
these studies, the optimum temperature would appear to be between 35
and 45°C. An operating temperature of 42°C was chosen for subsequent
SSF experiments. Note that this optimum temperature might vary with
different enzyme loadings and thus needs further study.
The optimum pHs for both the enzyme and the microorganisms were
similar, at about 6.0. The optimum pH of the SSF process was therefore
expected to be about 6.0. Figure 5 depicts the effect of pH on the lactic acid
production rate and yield. Lactic acid concentrations produced from 120
giL of the food waste in 48 h were 11.5, 54.2, 63.4, and 57.6 giL at the
operating pHs of4.0, 5.0, 6.0, and 6.5,respectively. The corresponding lactic
acid yields were 13,60, 71, and 64%. Maximum lactic acid production rate
was indeed observed at pH 6.0. In the SSF at pH 4.0, glucose was accumu-
lated throughout the reaction time, and lactic acid was formed very slowly.
This indicates that the SSF process was limited by the fermentation, prob-
ably owing to suppressed cell growth at this pH. In this set of experiments,
the pH was continuously controlled with 5 N NaOH in a 1.0-L (working
volume) fermentor. Interestingly, a slightly higher lactic acid yield (78%)
was produced in the shake-flask operation in which pH was controlled by
Applied Biochemistry and Biotechnology Vol. 105-108,2003
644 Kim et at.

~~-----------------------------------,
___ pH4
__ pH5
_ 60
~pH6

-
.J __ pH 6.5
0,

o 12 24 36 48

Time (h)
Fig. 5. Effect of pH on batch SSF of food waste (120 giL) at 42°C.

CaCOJl as shown in Fig. 3, than in the bioreactor controlled at similar pH.


This result could not be explained clearly and needs to be investigated
further. However, since there was no significant difference in lactic acid
production and because of its ease of operation, the shake-flask system was
used throughout the optimization study.

Effect of Nitrogen and Mineral Supplements


One of the major costs in lactic acid fermentation is the consumption
of large quantities of expensive nitrogen sources, such as yeast extract.
They are required as the growth nutrients and can affect the lactic acid
productivity. Thus, the effect of nitrogen supplements on lactic acid pro-
duction from food wastes was studied. A series of SSF experiments was
conducted using yeast extract or peptone as the nitrogen source. An SSF
run without nitrogen supplement was also performed. As shown in Fig. 6,
33.5 (yield 75%) and 30.3 giL (yield 67%) of lactic acid were produced after
2 d of SSF with 60 giL of food waste when 15 giL of yeast extract and
peptone, respectively, were used as the nitrogen supplement. Without
supplementation of nitrogen sources, 27.2 giL (yield 60%) of lactic acid was
produced from 60 giL of food waste. Since the food waste used contained
nitrogen compounds, as indicated in Table I, it was thought that those
nitrogen compounds in the food waste were used as the growth nutrients'
sources. Proteins contained in food wastes can be hydrolyzed to nutrients
for growth and lactic acid production of L. delbrueckii by the enzyme that
contains a controlled amount of protease activity. This result indicates a
minimal requirement of nitrogen supplements in lactic acid production
from food wastes using SSF.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Production of Lactic Acid from Food Wastes 645

~~----------------------------------~

--
. .J
c;,
-a
30

'uas 20

:nas
U

..J 10 _____ yeast extract


_____ peptone
--A- wlo nitrogen supplement

o+-------~------~------~------~--~
o 12 24 36 48

Time (h)
Fig. 6. Effect of nitrogen sources on batch SSF of food waste (60 gil) at 42°C.

100

--
. .J
c;,
80

60
-a
'uas
u
:= 40
u
as
..J
_____ with mineral supplement
20
--- wlo minerai supplement

0
0 12 24 36 48

Time (h)
Fig. 7. Effect of mineral supplements on batch SSF of food waste (130 gil) at 42°C.

To determine the effect of mineral supplements on the lactic acid pro-


duction, a set of SSF runs was carried out with and without mineral supple-
ments. The composition of mineral supplements was indicated in the
fermentation medium as described earlier. As shown in Fig. 7, in the early
stages of SSF up to 24 h, 72.4 giL of lactic acid was produced in the SSF
conducted with mineral supplements. A lower lactic acid concentration of
57.2 giL was produced in the SSF conducted without mineral supplements.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
646 Kim et al.
100 - , - - - - - - - - - - - - - - - - - - - - , - 1 0 0

-a,
-
.J

"C
80

60
BO

60
--
ffl.
"C
'ii
':;'
'0 'C
co '0
,~ CO

...
40 40
ti U
j U
20 20 CO
__ lactic acid concentration .J
- - lactic acid yield

60 70 80 90 100 110 120 130 140 150

Food wastes concentration (gil)

Fig. 8. Effect of food waste concentration on lactic acid yield in batch SSF at 42°C.

However, after 36 h of SSF operation, an almost identical level of lactic


acid concentration (72.5 giL, yield 74%) was produced from 130 giL of
food waste in both SSF runs. Since the food waste contained some miner-
als and phosphates, as shown in Table 1, which might be adequate for
bacterial growth and lactic acid production, additional supplements may
not be required. This experimental result along with the aforementioned
results from the nitrogen supplements study demonstrates a high eco-
nomic feasibility of lactic acid production from food waste.
Effect of Substrate Concentration
The effect of initial food waste concentration on the final concentration
of lactic acid produced was examined. The initial food waste concentra-
tions were varied from 65 to 145 giL in the SSF runs. As shown in Fig. 8,
the final lactic acid levels increased with each increase in food waste level,
but, the conversional yield decreased; during the 2-d SSF, the lactic acid
yields decreased from 91 to 73%. The highest yield of 91 % was obtained
from 65 giL of food waste with a final lactic acid concentration of 44.3
giL. On the other hand, the highest lactic acid concentration of 79.7 giL
was obtained from 145 giL of food waste with the least yield of 73%. This
result may be owing to incomplete hydrolysis of food waste, glucose and
lactic acid inhibition on the microorganisms' activity for a higher level of
food waste, and I or formation of other byproducts. Further work is needed
to clarify this observation.

Conclusions
We have demonstrated that lactic acid production from food wastes
by way of SSF is technically feasible. The highest yield of lactic acid on the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Production of Lactic Acid from Food Wastes 647
basis of available carbohydrates in the food waste was 91 % of theoretical:
44.3 giL of lactic acid from 65 giL of food waste. A total lactic acid concen-
tration of 80 giL was attainable from 145 giL of food waste within 48 h of
the SSF, which is remarkably higher than the lactic acid concentration of
20-40 giL obtained in 4-5 d in a direct fermentation of food wastes (8,9).
Within the scope of our study, the optimum operating conditions for the
production of lactic acid were 42°C and pH 6.0. Without supplementation
of nitrogen sources, the yield of lactic acid in the SSF decreased to 60%:
27 giL of lactic acid from 60 giL of food waste. Elimination of all the
noncarbon nutrients other than yeast extract from the SSF medium did not
adversely affect lactic acid production. Yeast extract therefore appears to be
the only additional nutrient required to maintain optimum performance of
the SSF. This clearly indicates that the food waste used in this investigation
contains a sufficient amount of mineral sources needed for bacterial growth
and lactic acid production, which would be one of the economic benefits
associated with food waste. Although further, detailed economic analysis
is needed, the preliminary results of our study demonstrate that lactic acid
production using food wastes can be considered a promising way of food
waste management.

Acknowledgment
This research was supported financially by Kyonggido through
Kyonggi Regional Research Center at Kyungwon University, Korea.

References
1. Lipinsky, E. S. and Sinc1are, R. G. (1986), Chern. Eng. Prog. 82,26-32.
2. Vickroy, T. B. (1985), in Comprehensive Biotechnology, vol. 3, Blanch, H. W., Drew, S.,
and Wang, D.1. c., eds., Pergamon, New York, NY, pp. 761-776.
3. Schmidt, S. and Padukone, N. (1997), J. Ind. Microb. Biotechnol. 18,10-14.
4. Parajo, J. c., Alonso, J. L., and Santos, V. (1996), Process Biochem. 3,271-280.
5. Chen, R. and Lee, Y. Y. (1997), Appl. Biochem. Biotechnol. 63-65,435-447.
6. McCaskey, T. A., Zhou, S. D., Britt, S. N., and Strickland, R. (1994), Appl. Biochem.
Biotechnol. 45/46, 555-568.
7. Zhou, S. D. and McCaskey, T. A. (1996), Appl. Biochem. Biotechnol. 57/58,517-524.
8. Lee, B. S., Yoon, H. H., and Kim, E. K. (2001), Korean J. Biotechnol. Bioeng.16,207-211.
9. Loh, C. W., Fakhru'l-Razi, A., Hassan, M. A., and Karim, M. I. A. (1999), Artif. Cells
Blood Substit. Immobil. Biotechnol. 27,455-459.
10. Hang, Y. D. (1989), Biotechnol. Lett. 11,299-300.
11. Hamamci, H. and Ryu, D. D. Y. (1994), Appl. Biochem. Biotechnol. 44, 125-133.
12. Bigelis, R. and Tsai, S. P. (1995), in Food Biotechnology-Microorganisms, Hui, Y. H. and
Khachatourians, G. G., eds., VCH, New York, NY, pp. 239-280. .
13. Linko, Y. Y. and Javanainen, P. (1996), Enzyme Microb. Technol. 19, 118-123.
14. Ehrman, T. (1992), NREL Chemical Analysis and Testing Standard Procedure, No. 002,
National Renewable Energy Laboratory, Golden, CO.
15. Nelson, N. (1944), J. BioI. Chern. 153,375-380.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Burnana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0649/$20.00

Microbiologic Oxidation
of Isosafrole into Piperonal

ALBERDAN SILVA SANTOS,1,2 NEI PEREIRA JR., 3


IRACEMA I. DA SILVA, 3 MARIA INES SARQUIS,4
AND OCTAVIO A. C. ANTUNES;,1
llnstituto de Quimiea, Un ivers ida de do Brasil, CT Bloeo A Lab 641,
Rio de janeiro, Rj 21945-970, Brazil,
E-mail: oetavio@iq.ufrj.br;
2Departamento de Engenharia Quimiea, Universidade Federal do Para,
Belem, Para, 66075-900, Brazil;
3Eseola de Quimiea, Universidade do Brasil, CT Bloeo E,
Rio de janeiro, Rj 21945-970, Brazil; and
4Laborat6rio de Colec;ao de Cultura de Fungos,
Departamento de Mieologia, Instituto Oswaldo Cruz,
Rio de janeiro, Rj 21945-970, Brazil

Abstract
The biotransformation ofisosafrole by Cladosporium sphaerospermum yielded
piperonal, which is a compound of great commercial importance in the flavor
and fragrance industries. The experiments were performed in SOO-mL conical
flasks containing 100 mL of Czapek-modified medium in an orbital shaker
with controlled agitation and temperature. Spores of C. sphaerospermum were
used as inocula, and after 96 h of incubation the substrate was added to the
culture. Samples of 2 mL were withdrawn at 24-h intervals and analyzed by
gas chromatography, (GC) and/or GC/MS spectroscopy.

Index Entries: Biotransformation; Cladosporium sphaerospermum; phenyl-


propanoids; isosafrole; piperonal.

Introduction
Oxidation is one of the most studied processes in the biologic transfor-
mation of organic compounds besides reduction (1). The simplest and most
suitable microbiologic process is the direct incorporation of oxygen to an
exogenous (or xenobiotic) substrate from molecular oxygen, because
dioxygen is the cheapest oxidant in the chemical industry (2). Therefore,
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 649 Vol. 105-108,2003


650 Santos et al.
alcohols can be produced from hydrocarbons, phenol derivatives from ben-
zene, and allylic alcohols and/ or epoxides from alkenes. Of course, thermo-
dynamically, these reactions are quite different owing to the different redox
potentials involved. Kinetically, considering the same feedstock (e.g.,
limonene) different catalytic pathways might take place. Therefore, multiple
product formation is possible. Although many examples of these biologic
oxidations have been described so far (3), not any enzymes responsible for
these transformations have been identified, and few have been fully charac-
terized. However, specific investigations have indicated that at least three
types of enzymes can be involved in microbiologic oxidation: heme-depen-
dent monooxygenases, W-hydroxylases, and methane monooxygenases
(4). Biologic incorporation of one oxygen atom is catalyzed by a group of
enzymes-cytochrome P450-dependent monooxygenases. This group of
enzymes drives the reactions under mild conditions, and it is one of the most
important classes of enzymes for the biotransformation (via oxidation) of
xenobiotics. These enzymes, cytochrome P450-dependentmonooxygenases,
are easily detected in all types of living cells. Many studies have been carried
out in mammalian enzymes owing to their involvement in the metabolism
of pharmaceuticals, mainly in the liver cells. The greatest feature of these
enzymes is the ability to oxygenate lipophilic xenobiotics leading to water-
soluble metabolites to facilitate their excretions. Based on this phenomenon,
many groups have been working on the oxidation-based biotransformation
of lipophilic compounds (xenobiotics) by different microorganisms, living
cells, or crude enzymatic preparations.
In addition, it has been vastly demonstrated that this enzymatic or
biomimetic system is able to conduct this oxidation enantioselectively (5-7).
As extensively emphasized in the literature (8), the screening of microorgan-
isms is a very important tool to find the best microorganism capable of effect-
ing a desired chemical transformation. Furthermore, the best reaction
conditions as well as the optimum process operation must be investigated if
commercial interest is to be pursued. The empirical basis of such a piece of
work is based on the fact that most, if not all, microorganisms have the nec-
essary machinery to effect a given chemical transformation. Of course, when-
ever working in such a way the substrate is commonly xenobiotic to the
microorganism and, consequently, potentially toxic to it. Therefore, very
careful choice of the reaction's conditions is needed.
Although oxidations have been known for more than a century, the first
systematically studied reaction effected by microorganisms was the
(enantioselective) reduction of activated ketones, which has been extensively
studied (9,10). When bearing the potential difficulties in mind, the screening
of microbiologic transformations of chemical compounds (xenobiotic sub-
strates) certainly seems to be the method of choice to assay a desired bio-
chemical transformation of an organic compound (4). Whole or fractions of
microorganisms, animal and plant cells, or crude enzymatic preparations
have been used (and extensively tested) as selective (or specific) biocatalysts.
The methodology is relatively rapid; is versatile; and, in principle, can be

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Oxidation of Isosafrole into Piperonal 651
used by any trained chemist (respecting the biosecurity-associated proto-
cols). Several assays of microbiologic transformations have been described in
the literature. The different classes of xenobiotic compounds and the differ-
ent microorganisms are incredibly great and growing rapidly for biotrans-
formation purposes (1,11-15). The main appeal of using microrganisms
instead of the traditional chemical transformations is associated with various
factors; for example, the solvent is usually water (not very cheap but always
friendly), and reaction conditions are generally mild and can be engineered
to recycle the biotransformation agent (e.g., immobilized). Consequently,
the overall process is normally an energy-saving one, compatible with the
new (and justified) claims of society. Based on these facts, the oxidation of
isosafrole (l,2-methylenedioxy-4-[1 '-propenyl]-benzene) to the correspond-
ing aldehyde, piperonal (4-carboxyaldehyde-1,2-methylenedioxybenzene)
was envisaged, although careful perusal of the literature indicated that com-
pounds bearing the methylenedio-xybenzene would behave as cytochrome
P450 inhibitors (16,17). On the other hand, the process itself is very attractive
from an industrial point of view because it provides an alternative to the
commonly used highly energy demanding process-ozonolysis. Addition-
ally, once the product is mainly used both as a fixateur in the fragrance
industry and as a flavor additive, good manufacturing practices and pro-
cesses will be highly desirable. The use of a clean, mild, and energy-saving
process to obtain a product of considerable economic importance was there-
fore the objective of the present work.
Isosafrole belongs to an important class of natural products, the
arylpropenoids. However, reports concerning the microbiologic oxidation
of these compounds are not common (4,18). For instance, microbiologic
oxidation of safrole (l,2-methylenedioxy-4-[2'-propenyl]-benzene) and the
corresponding isomer, isosafrole (1,2-methylenedioxy-4-[1'-propenyl]-
benzene), has not been the subject of many biotransformation studies. Evi-
dence pointing to the high toxicity of these types of compounds toward
microorganisms and their known properties of acting as cytochrome P450-
dependent monooxygenase inhibitors (19) are probably the main reasons
for the scarce interest in the microbiologic oxidation of arylpropenoids
with the methylenedioxyl group. However, metalloporphyrin biomimetic
oxidation of arylpropenoids has been reported (17).
The present article reports a biotransformation process of a natural
product compound originated from a plant source, isosafrole, as the
xenobiotic substrate to be oxidized by Cladosporium sphaerospermum. To our
knowledge, ours is the first study in which a compound belonging to the
arylpropenoid class containing a methylenedioxy group (Fig. 1) has been
successfully submitted to a microbiologic oxidation process.
Materials and Methods
Chemicals
The isomeric mixture of Isosafrole (E- and Z-, 85:15 by high resolution
gas chromatography) was kindly supplied from Geroma do Brasil (Parana,
Applied Biochemistry and Biotechnology Vol. 105-108,2003
652 Santos et al.

oyyCHO
~-V
E- and Z-Isosafrole Safrole Piperonal

Fig. 1. Arylpropenoids containing methylenedioxyl group.

Table 1
Compositions of Different Media for First Adaptation
of Lyophilized Microorganisms
Luria-Bertani Czapek-Dox Sabouraud-dextrose
Compound broth (giL) agar (giL) agar(g/L)
NaN03 3.00
K2HP04 1.00
MgS04·7H20 0.50
KCl 0.50
FeS04·7H20 0.30 0.01
Sucrose 30.00
Glucose 40.00
Agar 30.00 30.00
Tryptone 10.00
NaCl 10.00
Yeast extract 5.00
Neopeptone 10.00

Brazil). The solvents used were of analytical grade and were supplied from
VETEC (Rio de Janeiro, Brazil). All other chemicals and those used in the
broth medium were of analytical grade and purchased from Sigma/ Aldrich
(St. Louis, MO).
Microorganisms
The strains of lyophilized microorganisms belonging to different gen-
era were obtained from the different institutions: Aspergillus flavus, C.
sphaerospermum, Aspergillus niger, Pseudomonas putida, and Pseudomonas
aeruginosa were generously supplied by the Funda<;ao Instituto Oswaldo
Cruz (Rio de Janeiro, Brazil). Bacillus subtilis and Saccharomyces cerevisiae
were obtained from the Instituto de Microbiologia of the Universidade do
Brasil (Rio de Janeiro, Brazil).
Bacteria were cultivated in Luria-Bertani broth (20), and fungi and
yeast inCzapeck-Dox agar andSabouraud-dextrose agar (21),respe ctively.
Bacteria and yeast were cultivated at 26°C for 48 h and fungi at 26°C for 120
h. We used this procedure in the first inoculum condition (Table 1).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Oxidation of Isosafrole into Piperonal 653
Screening of Biotransformation Assay
Glass tubes (1 x 20 cm) were filled with 5 mL of the different broth
media. Each strain was inoculated in an individual tube, by a platinum
loop, containing its respective culture medium. Only spores were trans-
ferred from fungi. Bacteria and yeast were grown normally. The tubes were
properly closed with a wad of hydrophilic cotton and maintained in an
orbital shaker (150 rpm, 26°C). Under these conditions bacteria and yeast
were grown for 12 h and filamentous fungi for 96 h. After these periods, 0.2
mL of substrate emulsified with a 1% tween-80 suspension was added to
each respective tube. These cell cultures were incubated in the same initial
cultivation conditions. After 24 h, two tubes of each strain were removed
from the culture and submitted to the extraction procedure. All experi-
ments were carried out in parallel with controls, in the same conditions,
without the presence of microorganism.
Fungal Growth and Culture Medium
The strain of lyophilized spores of C. sphaerospermum 2950 was sus-
pended in 1 mL of 1% NaCl solution. This suspension was used to inoculate
Czapeck-modified agar plates. The growth and sporulation of mycelia was
maintained at 26°C for 5 d. Cultivation conditions were accomplished in
Czapeck-modified broth medium, containing 20 giL of sucrose, 2.0 giL of
NaN03, 1.0 giL ofKH2P04,1.0 giL of MgS04·7H20, 0.010 giL of (NH4)S04
and 0.50 giL of FeS04·7H20. The salts and carbon source were dissolved in
distilled water, made up to 100 mL, and transferred to 250 mL Erlenmeyer
flasks. The flasks containing the medium were sterilized at 121°C (1 atm)
for 15 min.
Several strains of different filamentous fungi microorganisms were
inoculated in a slant test tube containing 20 mL of Czapeck-modified agar
(30 giL) by a platinum loop. Only spores were transferred from the agar
plate. They were cultivated at 26°C for 5 d. After this period, the spore
suspensions were prepared with 10 mL of 1% NaCl solution. One milliliter
of the microbial suspension was used to inoculate 100 mL of sterilized
medium in a baffled 250-mL Erlenmeyer flask and maintained in a shaker
at 150 rpm and 26°C.
Addition of Substrate
After 4 d, 2 mL of substrate emulsified in a 1% Tween-80 and 1 mL
of 30% (v Iv) H 20 2 were added to the Erlenmeyer flask, and the pH was
adjusted to 3.0 with glacial acetic acid. The flask was returned to the initial
incubation conditions. Samples of 2 mL were withdrawn at 24-h intervals
from the medium and submitted to an empty test tube for extraction of the
product, as described next.
Extraction Procedure
The addition of 2 mL of ethyl acetate to the cell cultivation allowed
extraction. In each experiment, the mixture was submitted to agitation in
Applied Biochemistry and Biotechnology Vol. 105-108,2003
654 Santos et al.
a vortex shaker (without glass beads) for 20 s. Then it was left to decant for
3 min, and after organic phase separation, a lo5-mL aliquot was withdrawn,
filtered (through a Millipore 0.45-~m membrane), and transferred to a
4-mL glass septum capped viaL The samples were then submitted to gas
chromatography (GC) and/or GC/mass spectroscopy (MS) analyses.

Methods of Product Identification


Gas Chromatography
An HP5890 II gas chromatograph equipped with an HP5 WCOT
capillary column (25 m x 0.32 mm ID), with a film thickness of 0.52 ~m,
was used. H2 was used as carrier gas (1 mL/min, 32 cm/s). The oven
programming temperature started from 100 to 250°C at 3°C/min. The
injector was held at 150°C and detector at 240°C. Injection was operated
in the splitless mode (l-~L injection). In addition to mass spectra, reten-
tion times (and literature retention indexes) of authentic standards were
used as identification parameters (22).
Gas Chromatography/Mass Spectroscopy
A 5973HP gas chromatograph-mass spectrometer equipped with an
HP5 capillary column (as just described) was used. H2 was used as carrier
gas a flow rate of at 1 mL/min (32 cm/s) under the chromatographic con-
ditions just described. Mass spectra were obtained under electron impact
at 70 eV (scan mode: 1 scan/min; acquisition m/z: 40-400 units). Chromato-
graphic analysis revealed the presence of a peak at 8.85 min, evidencing the
presence of 4-carboxaldehyde-l,2-methylenedioxybenzene through com-
parison with an authentic sample, confirmed by comparison of the respec-
tive retention index and mass spectrum.

Results and Discussion


Several strains of microorganisms including filamentous fungi, yeast,
and bacteria were submitted to assay for the biotransformation of the
isosafrole into piperonal in the presence and absence of H 20 2. A. niger (laC
4222), A. fLavus (laC 3974), C. sphaerospermum (laC 2950), P. aeruginosa
(INCQS 00311), or P. putida (INCQS 00113) were able to oxidize the substrate,
whereas S. cerevisiae and B. subtilis were not. Among these microorganisms,
fungi presented the better capacity to oxidize isosafrole. Therefore, one strain
of each specie of fungi was selected for further study, and C. sphaerospermum
showed the highest selectivity and conversion yield. The biotransformation
of isosafrole was made possible through the action of C. sphaerospermum,
which had the ability to oxidize this substrate into piperonal with 100%
selectivity and 9.8% conversion, after 7 d of incubation. Figure 2 represents
the schematic reaction. Chromatographic analysis revealed the presence of
a peak at 8.85 min (Fig. 3), evidencing the presence of 4-carboxaldehyde-l,2-
methylenedioxybenzene through comparison with an authentic pattern, in
addition to comparison with the result of the analysis and sample control
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Oxidation of Isosafrole into Piperonal 655

<.
o~ I C. sphaerospermum..
o ~
4-( 1- propenila- E)-I,2- methylenedioxybenzene 4-carboxaldehyde-l,2methylenedioxybenzene

Fig. 2. C. sphaerospermum as oxidizer catalyst agent in biotranformation process.

2.4e4 (I)

2.2e4
c!O'

2.0e4 (j

1.8e4 ii
1.6e4
'"~
(j

1.4e4 ~
1.2e4
51>-l
l.0e4
§
~
.s>.
8000 b
6000
4 6 8 10 12 14 16
Time (min.)

Fig. 3. GC analysis showing the biotransformation product peak at 8.84 min.

(I)
c!O.
7.0e4
5'
6.0e4 (j

ii
S.0e4
'"
(j

~
4.0e4
~
3.0e4 51>-l
2.0e4
§
t:l
>-'

1.0e4 b
4 6 8 10 12 14 16
Time (min.)

Fig. 4. GC analysis of control sample: 85.34% of 4-{1'-propenyl-E)-1,2-methylene-


dioxybenzene at 10.26 min; 13.32% of 4-(1'-propenyl-Z)-1,2-methylenedioxybenzene
at 9.30 min; 1.34% of 4-(2'-propenyl)-1,2-methylenedioxybenzene at 8.52 min.

(Fig. 4). Results of the GC/MS analysis confirmed the presence of the prod-
uct, which was characterized through the mass spectrum, compared with
standards in the Wiley /NBS database libraries.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
656 Santos et al.
Table 2
Results from Screening of Biotransformation
Microorganisma Strain Product yield (%)
P. aeruginosa INCQS 00311 3.01
P. putida INCQS 00113 1.38
Control C1 0.00
A. niger IOC 4222 3.71
A.flavus IOC 3974 3.53
C. sphaerospermum IOC 2950 4.74
Control C2 0.65
aCl was the control for bacteria and C2 was the control for fungi.

However, C. sphaerospermum was not able to oxidize isosafrole with-


out the addition of H 20 2• Therefore, the supply of dioxygen by atmospheric
air was not enough to conduct the desired biotransformation, which points
to the action of a peroxidase instead of a cytochrome P450-dependent
monooxygenase. On the other hand, the formation of artifacts was pre-
vented since experiments in the absence of the microorganism did not show
any chemical transformation of the substrate. The other microorganisms
displayed the potential to oxidize the substrate (Table 2).
The literature provides strong evidence that substances containing the
methylenedioxyl group act as cytochrome P450 inhibitors (19,23). Accord-
ingly, the biomimetic oxidation of compounds containing the methylene-
dioxyl group suggests that these kinds of compounds are potential substrates
for lignin peroxidase in appropriate conditions (15). Recent publications sum-
marize the important advances achieved in the last few years on peroxidase-
catalyzed transformations with major emphasis on preparative applications.
In one of these works, a chloroperoxidase-catalyzed epoxidation of a synthetic
methylenedioxybenzene compound is described (15,24). It is worth pointing
out that the addition of H 20 2to the reaction medium to promote the desired
oxidation with C. sphaerospermum has not yet been described in the literature.

Conclusion
Our results show that C. sphaerospermum, in the presence of H 20 2,
effected the formation of 4-carboxaldehyde-1,2-methylenedioxybenzene
with 100% selectivity. This microorganism is a promising biotransforma-
tion agent capable of reaching high conversion yields of isosafrole in the
desired product. The need of for H 20 2 indicates the action of a peroxidase
instead of a monooxygenase, wich should be specifically involved since it
is normally expressed in conditions of stress.

Acknowledgments
We acknowledge financial support from CAPES, PROCAD, CNPq,
FAPERJ, and PRONEX.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Oxidation of Isosafrole into Piperonal 657

References
1. Roberts, S. M. (2000), J. Chem. Soc. (Perkin Trans. 1) 611-633.
2. Moro-oka, Y, and Akita, M. (1998), Catalysis Today 41, 327-338.
3. DeMontellano, P. R. O. (1995), Cytochrome P-450 Structure, Mechanism and Biochemis-
try, 2nd ed., Plenum, New York.
4. Archelas, A. and Furstoss, R. (1997), Annu. Rev. Microbiol. 51,491-525.
5. Groves, J. T. and Neumann, R. (1989) J. Am. Chem. Soc. 111,2900-2909.
6. Mansuy, D. (1994), Pure Appl. Chem. 66(4), 737-744.
7. Narimatsu, S; Kobayashi, N.; Masubuchi, Y., Horie, T., Kakegawa, T., Kobayashi, H.,
et al. (2000), Chem. Bioi. Int. 127, 73-90.
8. Faber, K (1995), Biotransformations in Organic Chemistry, 2nd ed., Springer-Verlag,
Berlin, Germany.
9. Prelog, V. (1964), Pure Appl. Chem. 9,119-130.
10. Seebach, D., Sutter, M. A, Weber, R. H., and Zuger, M. F. (1984), Org. Synth. 63,1-10.
11. Welsh, F. W., Murray, W. D., and Williams, R E. (1989), Crit. Rev. Biotech. 9(2),105--169.
12. Costas, M., Chem, K, and Que, L. Jr. (2000), Coord. Chem. Rev. 200-202,517-544.
13. Gonsalves, A M. d' A R, and Pereira, M. M. (1996), J. Mol. Catal. A: Chem. 113,209-211.
14. Kato, Y. and Asono, Y. (2001), J. Mol. Catal. B: Enzymat. 13,27-36.
15. Conesa, A, Punt, P. J., and van den Hondel, C. A M. J. J. (2002), J. Biotechnol. 93,
143-158.
16. Delaforge, M., Jaouen, M., and Bouille, G. (1999), Envir. Toxicol. Pharm. 7, 153-158.
17. Keseru,G.M., Balogh,G. T.,Arvai,G.,and Bert6k, B. (1999), Tetrahedron 55,4457-4466.
18. Shimoni, E.; Ravid, V., and Shoham, Y. (2000), J. Biotechnol. 78, 1-9.
19. Kakko, I., Toimela, T., and Tahti, H. (2000), Chemosphere 40, 301-305.
20. Cappuccino,J. G. and Scherman, N. (1992), Microbiology: A Laboratory Manual, 3rd ed.,
Beijamings Curmmings, New York, NY.
21. Demain, A L. and Solomon, N.A (1986), in Manual of Industrial Microbiology and
Biotechnology, John Wiley & Sons, New York, NY, 97-121.
22. Adams, R P.(1995), Identification of Essential Oil Components by Gas Chromatography/
Mass Spectroscopy, Allured Publishing, Carol Stream, IL.
23. Barreiro, E. J. and Fraga, C. A M. (2001), in Quimica Medica: As Bases Moleculares da
Apio dos Ftirmacos, Artmed Editora, Porto Alegre, Brazil, pp. 182-185.
24. Adam, W., Lazarus, M., Saha-M6ller, C. R, Weichold, 0., Hoch, V., Haring, D., and
Schreier, P. (1999), Adv. Biochem. Eng. Biotechnol. 63,73-108.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0659/$20.00

Breathing Air from Protein Foam

DOUGLAS M. ACKERMANN JR., DAVID N. JEWELL,


MATTHEW L. STEDMAN, VORAKAN BURAPATANA,
PRABHANI V. ATUKORALE, MICHELLE L. PINSON,
ALISON E. WARDLE, WENYAN ZHU, AND ROBERT D. TANNER*

Chemical Engineering Department, Vanderbilt University,


Nashville, TN 37235,
E-mail: rtanner@vuse.vanderbilt.edu

Abstract
Protein foams can be used to extinguish fires. If foams are to be used to
extinguish fires where people are present, such as in high-rise buildings or
ships, then a method for allowing people to breathe in a foam-filled envi-
ronment is needed. It is proposed that the air, used to create the foam be
used for breathing. A canister that will break incoming air-filled foam has
been designed for attachment to a standard gas mask, in order to provide
breathable air to a trapped person. Preliminary results for the modified
mask indicate feasibility of breathing air from air-filled protein foam.
Index Entries: Protein foams; egg albumin; canister; polypropylene sheet;
airflow; breakthrough.

Introduction
Protein foams, such as egg albumin foam, are comprised ofhydropho-
bic proteins that are efficient and inexpensive products for extinguishing
smoldering or other difficult-to-put-out fires. These fires can even be those
fed by large amounts of fuel such as in oil, coal, and chemical fires. These
types of fires also have the ability to give off considerable heat, which
creates resistance toward conventional fire suppression with water. Unlike
water, protein foam has the ability to extinguish a fire through its ability to
separate the oxygen supply from the fire. To exclude air from a fire, the
foam must completely fill the fire zone and the fire zone must remain com-
pletely filled to allow cooling to proceed.
When people are present in fires such as those in warehouses, air-
planes, ships, or even space capsules, an additional problem occurs: people

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 659 Vol. 105-108,2003


660 Ackermann et al.
cannot breathe freely because the protein foam used to extinguish the fire
has excluded the air in the area of the fire. We propose here to begin to solve
this problem by making that air in the foam available to the trapped persons
by using a device to collapse the foam and simultaneously release the air for
breathing. To implement the proposed solution, a breathing apparatus was
constructed for the purpose of allowing a person to survive in the enclosed
fire region until either help arrives or it is possible to escape.
The first breathing device tested was a surgical mask (typically found
in drug stores) manufactured by 3-M. The mask was constructed ofpolypro-
pylene and cotton fibers. It proved to be ineffective as a foam-breaking
device, because the cotton fibers absorbed water from the albumin solu-
tion, impeding airflow through the mask.
The proposed alternative is a foam-breaking canister for use with a
military type C-3 Chem/Bio Gas Mask (1). When equipped with its stan-
dard filtering canister, this mask effectively protects the wearer's respira-
tory tract, face, and eyes against certain chemical and biologic agents such
as HCN, S02' mustard gas, CN, GB, CS, CR, radioactive dust, and smoke
from fires (1). Here the mask was fitted with a 15-cm-high, 7-cm-diameter
airtight steel can in place of its standard filter canister. The standard gel-
activated carbon filter canister is not desirable because the material in the
filter is not water resistant and makes breathing difficult once it begins to
absorb the collapsed foam liquid. The airtight can was tested with a variety
of combinations of water-resistant filters used to break the foam and simul-
taneously release the air for breathing and reject the water from the mask.
Performance failure of the canister filter pack was defined as when either
the entering protein foam or released water broke through the filter and
reached the mask's breathing space. This mask was tested using foam of
various bubble sizes, made from various concentrations of egg albumin.

Materials and Methods


Equipment
Egg albumin (lot no. CB 951) was purchased from Matheson Coleman
and Bell. Fiberglass screen of 1 x 1 mm grid size was purchased from West
End Hardware (Nashville, TN). Shop-VacFoamSleeve (905-85),D.C. May's
Painter's Polypropylene Coveralls, and Phifer Super Solar Screening were
purchased from Lowe's (Nashville, TN). TRONY SC1025ID5*1603 photo-
cell was used in conjunction with an OPCOM laser pointer to determine the
bubble collapse time. A 5-cm-diameter polyvinyl chloride (PVC) elbow
was purchased at Lowe's.
Mask Design
A Chinese green tea steel cylinder 15 cm long and 7 cm in diameter
was used as the framework for the foam-breaking canister. Foam-break-
ing screening devices were placed alone or in series inside the canister.
The canister was attached (with an airtight seal made of duct tape) to the

Applied Biochemistry and Biotechnology Vol. 705-708,2003


Breathing Air from Protein Foam 661

Model No. 1
I
Screen Cone
12
t
I

Model No. 2

I ~ ~ ~ I
'\ t t
Screen Cone

Model No. 3
I [> I
t
Shop-Vac Foam Sleeve Cone

Model No. 4

I ~ ~ I
t '\
Shop-Vac Foam Polypropylene\
Sleeve Cone Screen Cone

Model No. 5 PVC


Elbow

I ~
J
~ I
'\pOlypropylene\
Screen Cone
Shop-Vac Foam
Sleeve Cone

Fig. 1. Diagrams of each model as described in Materials and Methods. Each filter-
ing device is labeled and indicated with an arrow. Diagrams are not drawn to scale.

air intake port of the gas mask. The other end was open and submerged
in foam.
The foam-breaking devices included a 1 x 1 mm grid size fiberglass
screen rolled into a cone, a piece of the foam sleeve rolled into a cone, and
a polypropylene sheet, or a combination of the three. The foam sleeve is a
porous material approx 1 cm thick with openings about 0.5-1 mm in diam-
eter. The five canister models tested with the corresponding foam-breaking
methods are illustrated in Fig. 1.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
662 Ackermann et al.
Table 1
Tested Canister Model Types and Corresponding Foam-Breaking Methods·
Model Foam-breaking Perceived resistance
no. method used (Borg CRI0 scale) (mean :t: SD)

1 One 15-cm fiberglass screen cone 2 :t: 0 (with and without foam)
2 Three 8-cm fiberglass screen cones 2 :t: 0 (with and without foam)
loaded in series
3 One 15-cm Shop-Vac foam sleeve cone 2.5 :t: 0 (with and without foam)
4 One to-cm Shop-Vac foam sleeve 3.25 :t: 0.35 (with and without
cone, one 10-cm polypropylene/ foam)
screen (skeleton) cone loaded in series
5 One 10-cm Shop-Vac foam sleeve cone, 3.75 :t: 0.35 (no foam)
one to-cm polypropylene/ screen
(skeleton) cone, polypropylene sheet
loaded in series, 5-cm-diameter
PVC elbow
6 One to-cm foam sleeve cone, one to-cm 8.5 :t: 0.71 (foam)
(model 5 polypropylene/screen (skeleton) cone,
with foam) polypropylene sheet loaded in series,
5-cm-diameter PVC elbow
•All perceived resistance values are based on evaluation by two of the authors (DA, DJ).
There was no noticeable difference between the perceived resistance for both dry and wet
foam-filled models for models 1-4. The perceived resistance with and without foam was
noticeable for model 5 and is indicated above.

The model types for the foam-breaking canisters are described in


Table 1 in the following format: the filtering devices are listed in the order
that they are loaded, starting at the open end of the canister and moving
in the direction of the mask. The foam-breaking method used for model I
consists of a IS-cm fiberglass screen cone made flush with the canister wall
with duct tape. The foam-breaking configuration used for model 2 com-
prises three 8-cm fiberglass screen cones loaded in series all made flush
with the canister walls with duct tape. The foam-breaking configuration
used for model 3 consists of one IS-cm-Iong Shop-Vac Foam Sleeve made
flush with the canister walls with duct tape. The foam-breaking configu-
ration for model 4 consists of an 8-cm-Iong Shop-Vac Foam Sleeve cone
and an 8-cm-Iong fiberglass screen cone lined with a polypropylene sheet
loaded in series. The two cones were made flush with the canister walls
with ducttape. The foam-breaking method for modelS consists of a 10-cm-
long Shop-Vac Foam Sleeve cone, and a 10-cm-Iong fiberglass screen cone
lined with a polypropylene sheet and a polypropylene sheet stretched
across the canister loaded in series. The two cones were made flush with
the canister walls with duct tape. The polypropylene sheet was stretched
across the canister at the junction between the steel cylinder and a S-cm-
diameter 90° PVC elbow. The 90° PVC elbow was added in modelS to
increase drainage of solution resulting from broken foam.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Breathing Air from Protein Foam 663
Air Outlet
_c r -
Laboratory Air Canister
Supply (I:nIet)
/
I Rotameter
----.
-
I I
Cap Cap

Fig. 2. Diagram of configuration used to measure airflow through freestanding


canister.

Rate of Airflow Through Canister


Caps were placed on the ends of a 15- to 17-cm-diameter steel Chinese
tea can in order to test the airflow rate through the canister separate from
the gas mask. The canister was sealed to be airtight, as shown in Fig. 2. A
laboratory air tube was attached and sealed to each cap. One air hose was
attached to the laboratory air supply, and the other to an inline rotameter
to measure the airflow rate. The outlet port of the rotameter was allowed
to vent to the laboratory. A baseline measurement of airflow through the
empty canister (control experiment) was run first. Foam-breaking devices
were added to the canister as described for models 1-5. The corresponding
airflow rates through the canister were then measured.

Perceived Resistance To Breathing For Model Types


It is important to be able to breathe comfortably through the mask, the
filtering media and foam under a variety of conditions. The perceived resis-
tance to breathing through the gas mask was tested with the Borg CRI0
breathing scale (2) for perceived exertion. Using the mask equipped with the
canister, the user took several breaths and reported a perceived exertion
level on the Borg CRIO scale. It was established in preliminary experiments
that the perceived exertion was about the same for the mask with or without
the foam. Models 1-4 were therefore each tested without the introduction of
foam for simplification. ModelS proved to be an exception for testing. It was
the only model that seemed to limit the number of breaths that could be
taken when subjected to foam. After approx 70 breaths with the canister
submerged in foam, the perceived exertion became too difficult for the sub-
ject to continue. Perceived exertion data were therefore considered with
modelS saturated with foam. The method used to determine this breathing
ability is based on the self-assigned Borg CRI0 exertion scale, which was
originally conceived as a way to measure the respiration of individuals with
impaired lungs.

Foam Col/apse Time


The time required for natural foam collapse in a glass tube was deter-
mined to establish how fastthe foam collapses and how wetthe foam is. The
Applied Biochemistry and Biotechnology Vol. 705-108,2003
664 Ackermann et al.

speed of bubble collapse is important to know since much breakage is


needed in the filter-aided mask to collapse the bubbles fast enough so that
a person could survive on the air from the broken foam. It is important to
consider foam wetness in choosing the appropriate filtering media. A 1.0
giL stock solution of egg albumin protein was used in the experiments on
foam collapse time. Collapse time was measured using laser light transmit-
tance through the foam in a glass column in a dark room.
The laser beam was directed through the foam fractionation column
and onto a TRONY photocell (housed in a Project Enclosure, purchased
from Radio Shack, to shield from ambient light) connected in parallel to a
voltmeter (Fig. 3). The voltage across a photocell is directly proportional to
the intensity of the light striking the photocell (3). The relative transmit-
tance of the laser light through the column was thus determined by the
voltage, as measured by the voltmeter, across the photocell as the foam
collapsed. Scattered light from the laser and intruding outside light is as-
sumed, which is why a maximum voltage (2.35 ± 0.13) was determined by
shining the laser through the column without any foam. This maximum
voltage was used as the control. The voltage across the photocell increased
as the foam collapsed and more light reached the photocell. Once the solu-
tion was foamed, the voltage change across the photocell was measured
every minute for 15 min. Percentage of transmittance was inferred by per-
centage of the maximum voltage at each minute. Percentage of collapse was
then inferred from percentage of transmittance.
Foamate Liquid Recovery
Foam stability is related to water content in the foam. A strong or
stable wet foam is best for extinguishing fires (zone 2 in Fig. 4), and a weak
or unstable dry foam (zone 1 in Fig. 4) is best for breathing through the
foam-breaking breathing mask. To measure the stability or strength of the
foam, the volume of liquid in the foamate was measured as a function of
time. Foamate liquid recovery was determined by foaming a 1 giL egg
albumin solution in a foam fractionation column to a foam height of 71 cm.
This height was measured by reading off of a ruler placed on the foam
fractionation column (Fig. 3). The height of solution in the base of the col-
umn was measured every 30 s up to a total run time of 15 min. The original
solution height before foaming was used as the maximum liquid height
value. Once the liquid was foamed, the height of the foam before it began
to drain was used as the zero height value. Volume of foamate liquid recov-
ery was determined by [the change in height (cm)] x 3t x [radius of the
column (in cm»)2 in mL of solution. Percentage of foamate solution recovery
was determined by V( t) I V max' in which V( t) is the volume of foamate liquid
recovered at any given time, t.
Testing of Mask using Foam
Three liters of a 1.0 giL egg albumin solution was prepared and placed
in a 5-gal open-top Nalgene container. Air entered via a sparger, which was
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Breathing Air from Protein Foam 665

Fig. 3. Laser foam fractionation column setup. (Bottom right) Laser beam shining
through foam in dark.

Zone 3
Best operating zone
for both extingui hing
fire and for breathing
through a rna k

Foam
Bubble
Diameter

Albumin
oncentration

Fig. 4. Conceptual diagram of optimal zones both breathing through a foam-


breaking mask (zone 1) and for extinguishing fires (zone 3). The overlapping region
(zone 3) represents the best operating zone for extinguishing fires and for breathing
through a mask.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


666 Ackermann et al.
made of approx 100 glass beads (about 4 mm in diameter) in a tightly
wrapped nylon mesh at the bottom of the container. The sparger filled the
S-gal container with foam of bubble diameter of about 5-10 mm. The can-
ister of models 1,2,3,4, or 5 was attached to the mask. The canister, attached
to the mask, was then vertically submerged in the foam in order to allow
excess water to be drained. The user began breathing normally, pulling the
foam into the canister. All model types broke the foam, providing breath-
able air to the user. The user continued to breath until "breakthrough" was
achieved. Breakthrough is defined as the breath at which the foam pen-
etrates the gas mask via the intake valve. The number of breaths required
to reach breakthrough was then recorded. ModelS was the exception. After
approx 70 normal breaths, the perceived exertion became too great for
normal breathing. The test was then stopped even though breakthrough
was not achieved.

Results and Discussion


As shown in Table 1, the perceived resistance, evaluated by using the
Borg CR10 scale (2), was found to increase with model number with the
exception of models 1 and 2 (whose resistances were found to be equal). The
Borg CR10 value is the self-assigned value on a scale of 1-10, with 1 being
very easy and 10 being very hard. In general, there is an increase in foam-
breaking material loaded into the canister in each consecutive model type,
and this increase leads to greater breathing resistance.
The evolution of foam-breaking models is described as follows:
Model 2 was created in response to the achievement of breakthrough of
model 1 after only one breath. The number of breaths until breakthrough
increased to three (three times that of model 1) by placing three series
screen cones of grid size 1 x 1 mm. Model 3 was then tried because models
1 and 2 were ineffective in limiting foam breakthrough of bubbles of small
diameter (less than about 5 mm). The small cell size and thick nature of the
Shop-Vac foam sleeve (the foam was required to travel through more
than one cell) improved the effectiveness of the canister. It was observed
that as foam was broken by the Shop-Vac foam sleeve, albumin solution
began to accumulate in the foam sleeve cells. As the user inhaled, foam of
bubble diameter smaller than the original foam (about 0.5-1.5 mm) was
created behind the Shop-Vac foam sleeve cone owing to the coalescence
of the solution that had accumulated there (Fig. 5). Model 3 achieved
breakthrough at 39 ± 1 breaths. To combat this small-diameter foam, a
second cone was installed to create model 4. This second cone consisted
of a polypropylene-lined screen cone. It was observed that the second
cone broke about 0.5- to l.S-mm-diameter foam created from the Shop-
Vac foam sleeve. In model 4, the second cone caused coalescence of the
accumulated solution in a fashion similar to that that occurred in the
Shop-Vac foam sleeve cones in models 3 and 4. The resulting foam diam-
eter did not differ noticeably from about 0.5- to 1.S-mm-diameter foam

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Breathing Air from Protein Foam 667

Colle

Fig. 5. Evolution of foam through each model as described in Results and Discus-
sion. Foam flows from left to right. Denser bubbles indicate small-diameter foam. Each
filtering device is labeled and indicated with an arrow.

created by the Shop-Vac foam sleeve cones. Model 4 achieved break-


through after 127.5 ± 4 breaths. ModelS was designed to provide addi-
tional breakage of about 0.5- to 1.S-mm-diameter foam resulting from the
second cone in model 4. To help break this small bubble foam, a polypro-
pylene sheet was stretched across the canister (Fig. 1). The 90° PVC elbow
(Fig. 1) was added to exploit gravity for drainage. However, the PVC
elbow was not enough to keep the polypropylene sheet from being satu-
rated and in effect further restricting breathing. When about 0.5- to 1.5-
mm-diameter foam resulting from the second cone in model 4 reached the
polypropylene sheet, breathing resistance became too great for normal
breathing after 77.5 ± 8 breaths. No foam breakthrough was reached using
modelS.
As shown in Fig. 6, the perceived resistance to breathing (by a subject
using the gas mask with the canister) is lessened as the airflow rate through
the canister is increased. The flow rate through the canister, as determined
Applied Biochemistry and Biotechnology Vol. 105-70B, 2003
668 Ackermann et al.
4.5

!c 4
!
:i'rJ
II: 3.5

.=0
OI(1)
.c..- 3
15
me'
'tJ 0 2.5
GIlD
> .....
•~ 2
~
1.5 +----.-------r-------r-----,
7.8 8.3 8.8 9.3 9.8
Flow Rate (Llmln)

Fig. 6. Correlation of airflow through canister to perceived breathing resistance as


determined using Borg 10 scale. The trend shows a decrease in flow rate with an
increase in perceived breathing resistance.

by the method described in Materials and Methods, decreased as a function


of perceived resistance as evaluated using the referenced Borg CRlO scale.
Mask models 1 and 2 are exceptions to this trend. It is believed that this is
owing to experimental error and is allowed within 1 SD of each mean
airflow rate (Fig. 6). This confirms the typical resistance for each model type
as approximated by the Borg CRlO scale.
The typical foam collapse time data, quantified in Fig. 7, describe how
fast the foam can collapse. Figure 8 shows a typical drainage experiment.
The photocell could be used to determine that approx 9% of V max is achieved
in 15 min. This implies that approx 9% of the laser light was transmitted to
the photocell, confirming what is seen in a visual assessment (the foam does
not look to be significantly collapsed). The data in Figs. 7 and 8 can be cross
plotted to give Fig. 9. Figure 9, in turn, can be used to define whether there
is wet foam or dry foam. The wet foams have little of their initial water
drained, whereas the dry foams have much of their water drained. Foam
wetness determines how well a person can breathe air from it and how well
it can extinguish a fire (Fig. 4).
Mask model types 1 and 2 provide little perceived resistance but offer
little foam-breaking ability and are therefore impractical since foam enters
the mask almost from the first breath. Although mask modelS offers a high
level of bubble-breaking ability, the perceived resistance to breathing is
beyond a feasible level (when filled with foam). This leaves mask model 4,
which allows for many breaths (about 120) until foam breakthrough, com-
bined with the characteristic of a perceived resistance that allows for com-
fortable breathing, as shown in Fig. 10. Model 4 is, therefore, the best overall
canister of models 1-5.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Breathing Air from Protein Foam 669

12

10

>
~ 6
'0
'#.
4

0+---------.--------.,--------.---------,
o 5 10 15 20
Tim. (min)

Fig. 7. Percentage of maximum voltage from photocell reading transmitted laser light
with respect to time. The voltage across the photocell climbs to approx 9% of V max (mean
Vmax ± SO = 2350 ± 127.92 m V) after 15 min. This is a measure of the resistance to the
penetration of laser light through the foam in the column at approx 40 cm above the
initial bulk liquid/ foam interface. The initial solution contains an albumin protein con-
centration of 1 giL. The solution is allowed to foam to a height of 71 cm above the initial
bulk liquid/foam interface.

100
90
80

l§! 70

0 60
i
a: 50
c
0
:0:;
:::I 40
;g 30
~
20

10
0
0 5 10 15 20
Tim. (min)

Fig. 8. Percentage of foamate liquid recovered (100 x Volume of albumin solution


recovered/volume of albumin solution initially in foam). The percentage of foamate
liquid recovered rises with respect to time and planes to approx 86% recovery. The
foam was under the conditions given in Fig. 7.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


670 Ackermann et al.
12

10

8
>C
ca
E
> 6
'0
'#.
4

0
0 20 40 60 80 100
% Foamate Liquid Recovered

Fig. 9. Cross plot of percentage of foamate liquid recovered and percentage of maxi-
mum voltage from photocell reading transmitted laser light. This figure can be used to
determine the relative foamate liquid content as a function of laser light transmittance.

140
.c

....
at
::I 120
0
.s:

•..
0lil:
It
100

II
E
80

"
~ 60
'j!
:::)
40
..•
.c
..•"
II
20

0
0 1 2 3 4 5 6 7 8 9 10
Perceived Breathing Resistance (Borg 10 Scale)

Fig. 10. Correlation of breaths until foam breakthrough into mask to perceived
breathing resistance as determined using the Borg 10 scale. The trend shows an
increase in breaths until breakthrough with an increase in perceived breathing resis-
tance. The datum point to the right represents the type 5 model when saturated by
foam. Model no. 5 did not reach breakthrough but exceeded a feasible breathing
resistance on the Borg CR10 scale.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Breathing Air from Protein Foam 671

Conclusion
Further tests need to be run to design a canister model that allows for
the breakage of smaller-diameter and wetter foams. We have shown that it
is feasible to breathe air from an air-generated foam reaching 120 breaths
in succession. Further work is needed to extend the working time and
breakage ability of the canister-modified gas mask breathing device.

References
1. Snyder, D. (2001), C-3 Chem/bio Gas Mask. Gas Mask USA. (Website: www.gasmask
usa.com)
2. Borg, G. (1998), Borg's Perceived Exertion and Pain Scales, Human Kinetics, Champaign,
IL.
3. Daley, R. (2002), Exp 137: Electromagnetic Spectrum. Lab on Legs: Energy and Change
March 28,2002. CSIRO South Australia Education Programs. Available at Website:
www.csiro.au/adelcsirosec/Pdf/Exp137.PDF. Accessed April 27, 2002.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


SESSION 6
Enzymatic Production and Conversions
Session 6

Enzymatic Production and Conversions

JOEL R. CHERRyl AND ROBERT DICOSIM02

Novozymes Biotech, Inc., Davis, CA; and


1

2E. I. DuPont de Nemours, Wilmington, DE

This session consisted of 7 oral and 35 poster presentations focusing


on various aspects of microbial enzyme production and performance for
biomass conversions. Representing research conducted in 15 countries,
the research touched on the use of enzymes for all areas of biomass conver-
sion, including pretreatment, hydrolysis, and conversion to chemical feed-
stocks. Applications of enzymes for the production of oleo chemicals and
pharmaceuticals were also described. To be commercially viable, any
enzymatic process must utilize an efficient mix of process-compatible
enzyme activities; the enzymes must be produced at high level from the
host organism, recovered with minimal cost, and result in a minimum of
unwanted side products. The hydrolysis of different biomass materials
will require different enzyme mixes, and the future challenge will be to
meet this requirement in a cost-effective manner. Improved methods were
presented to analyze complex enzyme mixtures secreted by efficient cel-
lulolytic organisms that will likely prove critical to meeting this require-
ment. While the characterization of enzyme mixtures secreted by naturally
occurring organisms presented in this session will likely greatly improve
our understanding of how naturally occurring organisms degrade various
biomass substrates, it is generally recognized that pretreated biomass may
require different, synthetic mixes of enzymes from diverse organisms.
Making such mixtures depends on building a toolbox of candidate
enzyme activities that can work under similar process conditions, and the
identification and cloning of novel enzyme activities remains a primary
focus of research groups worldwide. Other work exploring the immobili-
zation of enzymes, the use of alternate hosts such as plants, and changes
in biomass structural characteristics during enzymatic hydrolysis were
also presented and add to our knowledge base for future explorations of
the use of enzymes in converting renewable resources to bulk and fine
chemicals, agrochemicals, and pharmaceuticals.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 675 Vol. 705-708,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0677/$20.00

Partial Purification and Kinetic


Characterization of Acid Phosphatase
from Garlic Seedling

BEGUM YENIGUN· AND YUKSEL GUVENILIR

Istanbul Technical University, Department of Chemical Engineering,


80626 Maslak, Istanbul, Turkey,
E-mail: yenigun@itu.edu.troravcibasi@itu.edu.tr

Abstract
The objective of this study was to obtain purer acid phosphatases than
produced by prior art by operating under conditions that improve the final
product. The study features are the use of a mild nonionic detergent, 40-80%
saturation with (NH4)2S041 maintained at low temperature to remove impu-
rity, and the use of chromatografic columns to concentrate the acid phos-
phatase and remove non-acid phosphatase proteins with lower or higher
molecular weights. Acid phosphatase was isolated and purified from garlic
seedlings by a streamline method without the use of proteolytic and lipolytic
enzymes, butanol, or other organic solvents. Grown garlic seedlings of 10-
15 cm height were homogenized with 0.1 M acetate buffer containing 0.1 M
NaCI and 0.1% Triton X-100. After homogenization, the supernatant was
filtered with paper filters. Filtrated supernatant was cooled to 4°C, followed
by a threestep fractionation of the proteins with ammonium sulfate. The
crude enzyme was isolated as a green precipitate that was dissolved in a
small amount of 0.1 M acetate buffer containing 0.1 M NaCI and 0.1 % Triton
X-100. Garlic seedling acid phosphatase was purified with ion-exchange
chromatography (DEAE cellulose). The column was equilibrated with 0.1 M
acetate buffer. Acid phosphatase was purified 40-fold from the starting
material. The specific activity of the pure enzyme was 168 U / mg. A variety
of stability and activity profiles were determined for the purified garlic seed-
ling acid phosphatase: optimum pH, optimum temperature, pH stability,
temperature stability, thermal inactivation, substrate specificity, effect of
enzyme concentration, effect of substrate concentration, activation energy,
and effect of inhibitor and activator. The molecular mass of acid phosphatase
was estimated to be 58 kDa by sodium dodecyl sulfate polyacrylamide gel
electrophoresis. The optimum pH was 5.7 and the optimum temperature was
50°C. The enzyme was stable at pH 4.0-10.0 and 40-60°C. Activation energy
was between 10 and 20 kcal, and as Michaelis Menten coefficients, Vm values

"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 677 Vol. 105-108, 2003


678 YenigOn and GOvenilir
were 100 and 20 mM/ sand Km values were 21.27 and 8.33 mM for paranitro-
phenylphosphate and paranitrophenyl, respectively. Studies of the effect of
metal ions on enzyme activity showed both an activating and a deactivating
effect. While Cu, Mo, and Mn showed strong inhibitory effects, Na, Ca, and
K were the significant activators of acid phosphatase.
Index Entries: Enzyme; acid phosphatase; enzyme purification; garlic
seedlings.

Introduction
Phosphate esters are widely distributed in any organism. While many
metabolic intermediates are activated through the transfer of phosphate
groups, it is equally important that phosphate esters can also be rapidly
broken down. The hydrolytic removal of phosphate groups from phos-
phoesters is catalyzed by phosphatases. Depending on the pH at which
such phosphatases have optimal activity, one can distinguish between
acidic phosphatases (also called acid phosphatases) and alkaline phos-
phatases. The latter enzymes require divalent metal ions as co factors and
are common in animal tissues and bacteria (1).
Acid phosphatases are enzymes that hydrolyze the terminal phosphate
of phosphomonoesters, thus releasing inorganic phosphate. The optimum
pH for hydrolysis is in the range of 4.0-6.0. These enzymes are ubiquitous in
bacteria, fungi, animals, and plants (2). Studies of acid phosphatases from
various plant sources suggested their roles during the solubilization of mac-
romolecular organic phosphates in soils and the mobilization of phospho-
rous reserves during germination (3). Most studies of acid phosphatase,
particularly those from plant sources, are largely of a descriptive nature.
With a few exceptions (notably human prostatic acid phosphatase), the acid
phosphatases occur in very small quantities, are unstable in dilute solution,
and are subject to surface denaturation when purified. These factors,
together with a tendency to occur in multiple forms or as isoenzymes, make
the isolation of highly purified acid phosphatase difficult (4).
Materials and Methods
Garlic seedlings were grown from Kastamonu Seedlings, Turkey.
Chemicals (paranitrophenyl (p-NP), paranitrophenylphosphate (p-NPP),
4-nitrophenylphosphate) were obtained from Sigma, England. All other
chemicals (ammonium sulfate, NaOH, ammonium, sodium acetate, acetic
acid, citric acid, NaCl, ammonium hydrogen phosphate, sodium hydrogen
phosphate, potassium dihydrogen phosphate, sodium pyrophosphate
decahydrate, sodium pyrophosphate, D-glucose-6-phosphate, pyridoxal-
5-phosphate) were obtained from Merck, Germany. DEAE-cellulose anion
exchanger (Sigma) was used for column chromatography.
Specific Activity Assay
For substrate p-NPP, hydrolysis was determined by measuring inor-
ganic phosphate released by the acid phosphatase reaction. Phosphatase
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Purification of Acid Phosphatase from Garlic 679
activity was assayed in 0.1 M of sodium acetate buffer (pH 5.5) containing
0.1 M NaCl and 0.1% Triton X-100, and 50 mM substrate at pH 5.5.
Enzyme 0.25 mL was added to 2.25 mL of substrate solution. Buffer solution
(0.25 mL) instead of enzyme was used for blank sample. The reaction mix-
tures were incubated at 37°C for 30 min. Reactions were quenched by
sequential addition of 2.5 mL of 20 mMNaOH solution. The absorbance was
read at 405 nm. A standard curve of p-NPP was constructed. One unit of
phosphatase activity was defined as the amount of enzyme required to
produce 1 (mol of free Pi/min from 5 mM of p-NPP at pH 5.5 and 37°C.
Protein concentration was determined with an ultraviolet spectrophoto-
meter at 280 nm.
Purification of Enzyme
Garlic seedlings were grown in 3-,6-, and 9-wk periods. Crude extract
was obtained and ammonium sulfate saturation was applied for each
period. It was observed that 3- and 6-wk-old seedlings were quite unstable
and lost their activities in a short time. On the other hand, 9-wk-old seed-
lings did not show activity loss during almost 10-12 wk. As a result, the
purification procedure was carried out with 9-wk-old seedlings. The steps
for purification were homogenization, centrifugation, saturation, and ion-
exchange chromatography.
Garlic seedlings (9 wk old) were first washed free from soil and
weighed. They were homogenized in 0.1 L of 0.1 M sodium acetate buffer,
pH 5.5 (O.lMNaCl,O.l % TritonX-100), with a Waring blender atlow speed
for 20 s and then high speed for 30 s. The crude extract was filtered through
black filter paper and white filter paper, respectively. The crude extract after
filtration was centrifuged for 5 min at 20,000 rpm to get a clear extract. The
supernatant and precipitate were obtained. The precipitate was then resus-
pended in a minimum amount of buffer solution. Solid ammonium sulfate
was added to the supernatant to change the solubility of proteins and to
precipitate the proteins. First 40% was brought to saturation with ammo-
nium sulfate (25 g/100 mL). Stirring at about 4°C was continued for 2 h, and
the mixture was then centrifuged for 30 min at 20,000 rpm. The supernatant
and precipitate were separated. The supernatant was retained and ammo-
nium sulfate concentration was increased first to 60% (39 g/ 100 mL) then to
80% (52.3 g/100 mL). After stirring and centrifuging as just described, the
second supernatant and precipitate were obtained, which were then resus-
pended in a minimum amount of buffer.
A 20-mL sample of supernatant obtained after ammonium sulfate satu-
ration and centrifugation was loaded onto a DEAE cellulose ion-exchange
chromatography column (2 x 60 cm) that had been preequilibrated with
0.1 M sodium acetate buffer, pH 5.5. The protein was eluted with the same
buffer using a fraction size of 5 mL at the rate of 1/7 mL/min. Fractions
containing the majority of the acid phosphatase activity were pooled for
activity assay. The activity of acid phosphatase at the end of each of the
preparative steps was measured by spectrophotometric method.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
680 YenigOn and GOvenilir

Stability of Acid Phosphatase and Activity Profiles


Optimum pH, optimum temperature, pH stability and temperature
stability, thermal inactivation, substrate specificity, effect of enzyme con-
centration, effect of substrate concentration, effect of cations, kinetic con-
stants, and activation energy were determined for the partially purified
acid phosphatase of tomato seedlings. p-NPP was used as substrate.
Optimum pH
Acid phosphatase activity was tested in a pH range of 3.0-10.0 by
changing the ionic strength of the substrate. Ammonium and citric acid
were used to regulate pH. Sample without enzyme but buffer was used as
blank solution. Activity was determined at 37°C for 30 min.
Optimum Temperature
Optimum temperature was determined by measuring the activity in
a temperature range (0, 20, 30, 37, 50, 60, and 70°C) using the activity assay
procedure at related temperature.
pH Stabi Iity
Buffer solutions at different ionic strengths in the range of 2.5-10.0
were incubated at 30°C for 24 h after enzyme solution was added. Activity
was determined at 37°C with activity assay procedure. Distilled water with-
out enzyme was used as blank sample.
Temperature Stability
Diluted enzyme fractions of 15 mL were incubated for 15 min at dif-
ferent temperatures (0,30,40,50,60,70, and BO°C). After cooling for 5 min,
their activities were determined at 37°C according to activity assay.
Thermal Inactivation
Thermal inactivation studies were carried out at 65, 67, and 69°C.
Samples of 2 mL of diluted enzyme solution were incubated 0, 10,20,40,
and 60 min at each temperature. Activity assay at 37°C was applied imme-
diately after incubations.
Substrate Specificity
Acid phosphatase was assayed for activity toward 5 mM diammonium
phosphate, disodium phosphate, potassium phosphate, sodium pyrophos-
phate, sodiumpyrophosphate decahydrate, o-glucose-6-phosphate, pyri-
doxal-5-phosphate, and o-phosphorylethanolamine. Each substrate had its
own blank sample consisting of buffer solution instead of enzyme. The
same activity procedure was applied.
Effect of Enzyme Concentration
Enzyme was first brought to a 17 V/mg concentration and then 8.5,
4.25, and 2.125 V/mg concentrations by diluting with buffer solution.
Activities of samples were determined at 37°C.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Purification of Acid Phosphatase from Garlic 681

Effect of Cations on Enzyme Activity


Studies of the effect of metal ions on enzyme activity showed both an
activating and a deactivating effect. While Cu, Mo, and Mn showed strong
inhibitory effects, Na, Ca, and K were the significant activators of acid
phosphatase. Experiments were carried out with different amounts of
inhibitors over which 2.25 mL of p-NPP substrate was added. After 5 min
at 37°C, 0.25 mL of enzyme was added to the 30-min reaction.
Kinetic Constants
Substrate concentration is an important factor in determining the
rate of enzyme reaction. When an initial velocity is plotted against sub-
strate concentration while enzyme concentration is kept fixed, a satura-
tion curve (Michaelis-Menten curve) can be obtained. The specificity of
garlic seedling acid phosphatase was established by determining the
Michaelis constants for two different substrates. p-NPP and p-NP were
preferred substrates. Substrate solutions were prepared by diluting of
concentrated solutions with the acetate buffer. Apparent Km and V max
values were estimated using substrate concentrations ranging from 2.5 to
20 mM. Km values were calculated using a Lineweaver-Burk plot.
Activation Energy
Activation energy, A, of acid phosphatase is calculated by Arrhenius
equation
k = A. e-E,){T

by substituting the initial rates, V, instead of k. Two different sub-


strates (p-NPP and p-NP) with different concentrations were used.

Results and Discussion


This will be the first report on purification and characterization of acid
phosphatase, which was isolated from garlic seedlings, as shown in Table l.
Acid phosphatase of garlic seedlings was purified from crude homogenate
by the following steps: centrifugation, treatment with ammonium sulfate,
and DEAE cellulose ion-exchange chromatography.
The specific activity of the crude homogenate was found to be 4.33
U / mg of protein, which is a quite good result when compared with previous
results from different plant sources in the literature (2,4-7). The specific
activities at each of the isolation and purification steps are as follows:
74 U / mg of protein after saturation to 40-80% with ammonium sulfate, and
168 U /mg of protein after DEAE Sepharose column. After a 40-fold purifi-
cation, the enzyme was obtained in an 18% yield when assayed with p-NPP.
After DEAE cellulose column, activity was quite high although the
protein was low. This resulted in the enzyme being almost brought to
homogeneity. However, most of the acid phosphatase literature is con-
cerned with the seedling and leaf source of the enzyme. Purifications of
100-fold are typical, and although some 700- to 1000-fold purifications
Applied Biochemistry and Biotechnology Vol. 105-108,2003
).
:g
[
OJ
0'
C"l
::J-
<b
3
i;;'
~
'":::.0..
OJ
0'
~
::J-
:::.
o
Table 1
~ Isolation and Purificaiton Steps of Garlic Seedling Acid Phosphatase
Total protein Specific activity Total activity Yield Purification
0"1
co Step (mg) (U fmg protein) (U) (%) (fold)
I'-.)

Crude extract 1392 4.33 6029 100 1


Saturation to 40-80% with (NH4)zS04 DEAE
cellulose ion-exchange column 6.5 168 1090 18 39

6-
;--

-~
,~
I\.)

§
Purification of Acid Phosphatase from Garlic 683

100

80

ti 60
<r;
~ 40
Is'<
20

0
0 20 40 60 80 100
TEMPERATURE, C

Fig. 1. Optimum temperature.

120

O+---~~~~----r---~----~--~----~---'
3 4 5 6 7 8 9 10 11

pH

Fig. 2. Optimum pH.

have been described, few present any evidence for the catalytic purity of
the preparations (8).
To determine the effect of some parameters on partially purified acid
phosphatase enzyme and its stability, pH and temperature assays were
done. Activity of garlic acid phosphatase toward p-NPP was investigated
since some enzymatic activity toward this substrate is present in many
plant-sourced acid phosphatases. p-NPP could be obtained in an optically
active form, with the chirality at phosphorus.
Enzyme activity was stable between pH 5.5 and 7.5 and 40-60°C.
Maximum activity was obtained at pH 5.7, and garlic seedling acid phos-
phatase presented high activity at 50°C, which is higher than described for
some other plant phosphatases, such as barley roots (30-35°C) and cotton
seed (37°C), when p-NPP was used as substrate (9) as shown in Figs. 1-4.
The activity of acid phosphatase toward various substrates is shown
in Table 2. We used some inorganic and organic phosphates that were not
commonly used in the literature, to examine whether the isolated enzyme
could be used in remediation as phosphate decomposer. Garlic acid phos-
phatase did not show activity with most of these substrates, a few of
which were utilized successfully with other acid phosphatases reported
in the literature (7).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
684 Yenigiin and Giivenilir
120
><
~ 100
!-I 80
~
~ 60

~40
~ 20
~ 0 +-------r------.------~------,_----_.
o 20 40 60 80 100
TEMPERATURE. C

Fig. 3. Thermal stability.

120
>-
t: 100
>
~ 80
~ 60
>
~ 40
5lc:: 20
~
0+----..----.----.-----.----.----.----,
4 5 6 7 8 9 10 11
pH

Fig. 4. PH stability.

Substrate concentration is an important factor in determining the


degree of enzyme reaction. The results of our study are given in Table 3.
Generally, 5 mM p-NPP is used as substrate for acid phosphatase stud-
ies reported in the literature. In our study, 2.5 and 5 mM p-NPP showed the
same characteristics in the first 10 min, but, on the other hand, 1.25 mM
p-NPP showed quite higher activity. Activities came closer to each other
after 30 min, and after 50 min all substrate concentrations showed the same
characteristics. This led us to believe that a 1 mM p-NPP concentration
could be used in further studies.
Garlic seedling acid phosphatase lost its activity after 60°C. It can be
seen in Fig. 5 how fast this inactivation was, especially for 65°C. Inactiva-
tion occurred nonlinearly and fast at 62, 65, and 67°C. It can be concluded
that temperature control could be effectively possible up to 65°C. Proteins
in enzyme structure will be denatured after 65°C.
The results of changes in enzyme concentration are shown in Fig. 6.
Enzyme concentrations of 17, 8.5, and 4.25 U / mg gave linear results but a
concentration of 2.125 U /mg showed diversions from linearity. It was
thought that there was a relationship between free water molecules coming
from dilution and active sites. The same situation had been reported for
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of Acid Phosphatase from Garlic 685
0,4
~ 0,3
~ 0, 2 .,;;;;:~---.:....... -+-65 C
a ~~~~----~ -67C
.3 0,1 ---.-69 C
O+---~--~----~==~~
o 20 40 60 80
Time (min.)

Fig. 5. Thermal inactivation.

2,5

E 2
E --17 U/ml
II')
1,5 ___ 8.5 U/ml
~
CIl --4.25 U/ml
a:I
« ___ 2. 125 U/n
0,5

0
0 10 20 30 40 50 60 70
TIME. min

Fig. 6. Enzyme concentration.

Table 2
Substrate Specificity
Specific activity
Substrate name (U Img)
Di.-ammonium phosphate 0
Di.-sodium phosphate 0
Potassiium phosphate 0
Sodium pyrophyosphate 0
Tetrasodium pyrophosphate decahydrate 0
D-Glucose-6-phosphate 0
Pyridoxal-5-phosphate 1.5
0-Phosphorylethanolamine 0

Table 3
Substrate Concentration
Time (min) 1.25 mM p-NPP 2.5mMp-NPP SmMp-NPP
10 0.625 0.517 0.577
30 1.388 1.464 1.528
50 1.579 1.552 1.661

Applied Biochemistry and Biotechnology Vol. 105-108,2003


686 Yenigiin and Giivenilir

-1,5 -1 -0,5 0 0,5


1/[S]

Fig. 7. Lineweaver-Burk plot of acid phosphatase activity with p-NPP and p-NP as
substrate at pH 5.5.

Table 4
Effect of Metal Ions on Enzyme Activites

Metal ions (1 mM) Relative activity (%)a

None 100
Copper 66
Molybdate 40
Manganese 66
Dodium 163
Calcium 165
Potassium 215
"Relative activity is expressed as percentage of the
activity of no addition.

lipases. For further information about this situation, amino acid composi-
tion of this enzyme should be investigated.
Only two substrates, p-NPP and p-NP, were used for Michaelis kinetic
constants. Figure 7 shows that V mfor p-NPP was 100 mM/ s and Km was
21.27 mM, while V mfor p-NP was 20 mM/ sand K". is 8.33 mM. It is obvious
that the reaction was faster with p-NPP. Nonlinear plots may be owing to
diesterase contaminants. p-NPP could be a substrate for some possible
diesterase contaminants since it has been proposed as a specific substrate
for certain phosphodiesterases (5).
The effects of several substances on enzyme activity with p-NPP as
substrate were measured (Table 4). Inhibition of acid phosphatases by cop-
per, manganese, and molybdate has been observed in sweet potato, cul-
tured tobacco cells, rice bran, gladiolus bulbs (7), Japanese radish (7),lentil
seeds (10) and wheat germ (3). On the other hand, the activators sodium,
calcium, and potassium, are also listed Table 4. Calcium has been observed
in wheat germ (3), sweet Spanish (2), and rice bran (11).
Applied Biochemistry and Biotechnology Vol. 705-108,2003
Purification of Acid Phosphatase from Garlic 687
. The activation energies were between 10 and 20 kcal. Substrate con-
centrations ranged between 1.25 and 20 mM for p-NPP and p-NP, a result
quite parallel with findings in the literature. Activation energy was slightly
higher with p-NP.
Initial velocity rates were determined with different substrate con-
centrations of 5,10,15, and 20 mM. Activities after 5 min were found to
be 2.8 x 10-3, 4 X 10-3, 6.4 X 10-3, and 6.9 x 10-3, respectively.
Considering the specific activities despite the little weight of the
seedlings and small volume of the crude extract, it can be concluded that
the seedlings consist of a significant amount of acid phosphatase enzyme.
Further studies using a more appropriate purification procedure may
lead to the use of garlic seedlings as an economic source of acid phos-
phatase enzyme.

References
1. Price,N. C. (1982), Fundamentals ojEnzymology,Oxford University Press, Oxford, UK.
2. Guo, J., Pesacreta, T. C. (1997), Plant Physiol., 151, 520-527.
3. Kawarasaki, Y., Hiedo, N., and Yamane, T. (1996), Plant Sci. 119,67-77.
4. Van Etten, R. 1. and Waymack, P. P. (1991), Arch. Biochem. Biophys. 288,621--623.
5. Hye-Shin, C. P. and Van Etten, R. 1. (1986), Phytochemistry 25(2), 351-357.
6. Deveci, N. and Guvenilir, Y. (1995), Appl. Biochem. Biotechnol. 53.
7. Yoshimoto, M., Kimura, T., Miyamoto, T., Sakamoto, J., and Hatano, S. (1992), Biosci.
Biotech. Biochem. 56(1),147-148.
8. Van Etten, R. 1. and Waymack, P. P. (1991), Arch. Biochem. Biophysics 288(2), 634--645.
9. Verissima Ferreira, c., Granjeiro, J., Taga, E., and Aoyama, H. (1986), Biochem. Biophys.
Res. Commun. 242, 282-286.
10. Bose K. S. and Taneja V. (1998), Biochem. Biophys. Res. Commun. 25, 629--634.
11. Hayakawa, T., Toma, Y., and Igaue, I. (1989), Agric. Bioi. Chem. 53(6), 1475-1483.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0689/$20.00

Automated Filter Paper Assay


for Determination of Cellulase Activity

STEPHEN R. DECKER,* WILLIAM S. ADNEY, EDWARD JENNINGS,


TODD B. VINZANT, AND MICHAEL E. HIMMEL

National Bioenergy Center, National Renewable Energy Laboratory,


1617 Cole Boulevard, Golden, CO 8040 I,
E-mail: steve_decker@nrel.gov

Abstract
Recent developments in molecular breeding and directed evolution have
promised great developments in industrial enzymes as demonstrated by
exponential improvements in ~-lactamase and green fluorescent protein
(GFP). Detection of and screening for improved enzymes are relatively
easy if the target enzyme is expressible in a suitable high-throughput
screening host and a clearly defined and usable screen or selection is avail-
able, as with GFP and ~-lactamase. Fungal cellulases, however, are difficult
to measure and have limited expressibility in heterologous hosts. Further-
more, traditional cellulase assays are tedious and time-consuming. Mul-
tiple enzyme components, an insoluble substrate, and generally slow
reaction rates have plagued cellulase researchers interested in creating
cellulase mixtures with increased activities and/ or enhanced biochemical
properties. Although the International Union of Pure and Applied Chem-
ists standard measure of cellulase activity, the filter paper assay (FPA), can
be reproduced in most laboratories with some effort, this method has long
been recognized for its complexity and susceptibility to operator error. Our
current automated FP A method is based on a Cyberlabs C400 robotics deck
equipped with customized incubation, reagent storage, and plate-reading
capabilities that allow rapid evaluation of cellulases acting on cellulose and
has a maximum throughput of 84 enzyme samples per day when perform-
ing the automated FPA.
Index Entries: Filter paper assay; cellulase; cellulose; Trichoderma reesei;
filter paper unit.

Introduction

Cellulose, an unbranched ~-l,4-linked homopolymer of glucose, is the


most abundant renewable fuel resource on Earth, accounting for about half
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 689 Vol. 105-108, 2003


690 Decker et al.
of the organic material in the biosphere, and it is the major polysaccharide
found in plant biomass. The hydrolysis of cellulose, aided by endocellulase,
exocellulase, and ~-D-glucosidase catalysis, produces glucose, an easily
fermentable monosaccharide. Intense research is currently aimed at the
conversion of cellulose to sugars and ethanol via fermentation because this
process has great economic potential and is environmentally friendly (1-3).
Unfortunately, the main impediment for ethanol production via enzymatic
saccharification of cellulose is the low activity of native cellulases (4,5).
Cellulose is insoluble and crystalline; hence, it is largely resistant to
enzymatic hydrolysis. In many biomass utilization schemes, the raw mate-
rial is first treated with dilute acid at moderate temperatures (6) to remove
hemicellulose and to speed up subsequent cellulose hydrolysis by enzymes.
The pretreated biomass can then be subjected to carefully chosen mixtures
of endo- and exoglucanases for maximum cost-effectiveness (7,8).
The advent of in vitro molecular evolution techniques has opened up
a new area for enzyme improvement. Although these techniques have dem-
onstrated improvements in several enzymes, the axiom "You get what you
screen for" remains the dominant truth in the high-throughput screening
(HTS) arena (9,lD). This certainty, however, is conditional. You must have
both an HTS-compatible expression host and a suitable screening method in
order to accomplish anything. Typically, the host is Escherichia coli, and the
screening involves either improvement-dependent growth or colorimetric
discrimination between improved and unimproved enzyme. For cellulase
enzymes, HTS-compatible host expression has been limited to either bacte-
rial enzymes or fungal endocellulases. Currently, fungal exocellulases have
not been demonstrated to be capable of native-like expression in an accept-
able HTS host. Additionally, coexpression of an intact whole cellulase sys-
tem has not been demonstrated outside of native systems. Therefore,
detection of improvements to enzyme activity is limited to single enzymes
expressed in non-HTS hosts.
Worldwide, screening of whole cellulase preparations is done pre-
dominantly using the International Union of pure and Applied Chemists
(IUPAC) standard filter paper assay (FPA). This assay, developed by
Ghose in 1984 (11), uses the dinitrosalicylic acid (DNS) method to deter-
mine reducing sugars released from 50 mg of Whatman #1 filter paper by
a complex cellulase mixture. The method requires simple reagents and
equipment, but it is plagued by long assay times, exacting dilutions, and
many manual manipulations. It is valid only at low levels of hydrolysis
and generally requires several iterations to pin down a valid activity mea-
surement. Although Whatman #1 filter paper is fairly ubiquitous and
readily available in most laboratories, it requires careful manual manipu-
lation in order to partition it evenly into 1 x 6 cm strips (-50 mg). Incon-
sistencies in size, shape, and folding methods can all lead to errors in
activity determination. In addition, because of the filter paper's physical
properties, it is very difficult to distribute it evenly through automated
methods, especially if the scale is reduced to microtiter plate levels. We
Applied Biochemistry and Biotechnology Vol. 105-108,2003
FPA to Determine Cellulase Activity 691

examined several cellulose compounds for their hydrolytic similarities to


Whatman #1 filter paper, including Solka-Floc®, SigmaCell-20, Avicel,
and cotton linters. Although these substrates differ in their physical prop-
erties from Whatman #1 filter paper, they have the distinct advantage of
being capable of forming pipetable slurries in aqueous solution.
One of the most confounding problems is the determination of the
initial dilution needed to bracket 3.6% hydrolysis (The 4% usually stated
does not account for the water added during hydrolysis). Because cellu-
lase activity in whole broths is nonlinear in regarding enzyme concentra-
tion, this atypical assay dictates dilution of the cellulase preparation to a
point where 2.0 mg of reducing sugar equivalents is released in 1 h at 50°C
and pH 4.8. This amount of enzyme is defined as one filter paper unit
(FPU). Traditionally, a wide dilution range is used and then the assay is
repeated to fine-tune the level of confidence. For commercial broths, the
dilution required is usually considerable, since the starting titers are often
>50 FPU /mL. These high dilutions can result in significant error if not
done carefully.
Recent advances in HTS equipment have led us to automate this
traditionally manually intensive assay. Although the intent behind the
project was to streamline cellulase production determination from fungal
strains producing complete cellulase systems, it should be possible to
assess the effect of alterations in a single component by augmenting the
single enzyme with the other required components. The goals of the
present study were to reduce operator involvement and potential opera-
tor error in determining the activity of cellulase preparations, to reduce
the amount of reagent usage and associated disposal costs, and to allow
high throughput of samples through automation of the assay.

Materials and Methods


Reagents
The enzyme used in this study was a commercial broth obtained from
Genencor (Palo Alto, CA) formulated for commercial sale and storage.
Protein concentration was 108 mg/mL and the traditional FPU value was
determined to be 38.6 FPU / mL. The buffer used was 50 mM citrate, pH 4.8.
The reducing sugar reagent was the standard FPA DNS reagent containing
1416 mL of deionized water, 10.6 g of 3,5-dinitrosalicylic acid, 19.8 g of
NaOH, 306 g of Rochelle salts (Na-K tartrate), 7.6 mL of phenol, and 8.3 g
of sodium metabisulfite. Glucose (3.0 mg/mL) was used as the reducing
sugar stock standard and was diluted to produce a standard curve. All
chemicals were ACS reagent grade or better and were obtained from Sigma-
Aldrich (St. Louis, MO).
Substrates
The substrates assayed included Whatman #1 filter paper (1 x 6 cm
strips for traditional method, 1/4 in. diameter circles for microtiter plate
Applied Biochemistry and Biotechnology Vol. 105-108,2003
692 Decker et al.·

Fig. 1. Layout of Cyberlabs C400 robotics deck. 1, probe head; 2, ALPS plate sealer;
3, tip storage; 4, plate storage; 5, chilled reagent/plate rack; 6, plate reader; 7, incubators.

scale), Solka-Floc (Brown, Berlin, NH), SigmaCell-20 (Sigma-Aldrich),


Avicel PH101 (FMC, Philadelphia, PA), and cotton linters (Fluka/Sigma-
Aldrich, St. Louis, MO). The cellulose powders were suspended in distilled
H 20 at 2.65% (w Iv). Glycerol was added to 1.5% (v Iv) to prevent the dried
pellets from electrostatically jumping out of the wells.
Hardware
The assay was developed on a Cybedabs C400 XYZ chassis robotics
deck (Cyberlabs, Inc., Brookfield, CT) (Fig. 1). The deck consisted of a probe
head with single-, 8-, and 96-channel pipettor probes as well as a center-pull
gripper hand. The deck of the robot was laid out with racks for pipet tip,
assay plate, and dilution plate storage; a 96-channel tip installation station;
a 96-channel O-ring greasing station; reagent reservoirs; an ALPS-100 auto-
mated plate sealer (Advanced Biotechnologies, Surrey, UK); a barcode
reader; a waste chute; a chilled elevated rack for assay setup and enzyme
and reagent storage; three independently controlled and electronically
heated microtiter plate incubators (Fig. 2A), and a SpectraMax 190 microtiter
plate reader (Molecular Devices, Sunnyvale, CA). The system was controlled
through CYBOS operating software. The incubators were originally devel-
oped at Cyberlabs and modified at National Renewable Energy Laboratory
by our in-house machine shop. This modification included the addition of
machined aluminum form-fitting inserts designed to enhance heat transfer
to the bottom of the assay plates, brass insert guides to assist in placement
of the plates on the form-fitting inserts, and brass plate lids designed to
passively heat the plates from the top and weigh down the plates to ensure
good contact with the inserts and prevent plates from warping at high tem-
peratures (Fig. 2B). The incubators were controlled through independent

Applied Biochemistry and Biotechnology Vol. 105-108,2003


FPA to Determine Cellulase Activity 693

Fig. 2. Custom-modified microtiter plate incubators showing (A) brass insert guides,
lid, and aluminum form-fitting inserts and (B) closeup of brass plate lid and aluminum
insert. The gripper arm of the probe head moves the incubator lids and the individual
brass plate lids.

Digi-Sense temperature controllers (Cole-Parmer, Vernon Hills, IL) using


Type J thermocouples. Output data from the SpectraMax plate reader were
exported as ASCII text and analyzed in Excel. The chilled deck was cooled
by circulating a of coolant through the deck using a Julabo F240 recirculating
cooler Gulabo, Allentown, PA).
Calculations
The standard FPA uses 50 mg of filter paper (FP) in 1.5 mL of assay
solution. For the automated assay, the reaction was carried out in a 96-well,
polypropylene, round-bottomed microtiter plate with a capacity of 300
f..tL/well. The volume of reaction mix in each well of the automated assay
was set at 0.080 mL:
50 mg FP 0.080 mL
1.50 mL x well = 2.67 mg cellulose/well

Applied Biochemistry and Biotechnology Vol. 105-108,2003


694 Decker et al.
A standard 1/ 4-in. office paper punch yielded filter paper disks with
an average weight of 2.65 mg, so this amount was used for each substrate.
The target concentration for each well was 2.65 mg of cellulose/0.080 mL
or 33.1 mg/mL, which compares with 33.3 mg/mL for the standard FPA.
The standard FPA targets 2 mg of reducing sugar released from 50
mg of filter paper, equivalent to 3.6% hydrolysis. Note that this is not
given in units of concentration. In the standard FPA, the reaction volume
is 1.5 mL and a concentration of 1.333 mg/mL of glucose (glc) equivalents
comprises 2.0 mg in 1.5 mL. Because the standard assay looks for the
release of a predefined amount of reducing sugar (2 mg) and not the
concentration of product, a conversion is needed to adjust the standard
2.0 mg of glucose equivalents released to a concentration equivalent to
3.6% hydrolysis. Owing to the reduction in substrate amount and reaction
volume, the required release of sugar to give 3.6% hydrolysis needed to
be determined:
2 mg glc/mL 1.0 mg cellulose 0 •
50 11 I x 111 I x 100 =6.3 Yo converSIOn
mgce uose . mggc

2.65 mg cellulose 1.11 mg glc well -1325 I / L


0.036 x we11 x mg ce11u Iose x 0.080 m L - . mg g c m

Because the FPA is based on total product present at a specific time in


a nonlinear reaction, a conversion factor is used to approximate a specific
activity-type unit. The standard FPA bases this conversion on 0.5 mL of
enzyme, even though the total reaction mix is 1.5 mL. In the automated assay,
the enzyme and buffer are premixed and added together, resulting in a dif-
ferent volume factor. The two conversion factors are calculated as follows:
Standard:
2.0 mg glc 1 !-tmol glc 1 .
0.5 mL x 0.18016 mg glc x 60 min = 0.37 !-tmol glc/(mm . mL)

Automated:
1.325 mg glc 1 !J.ffiol glc 1 .
mL x 0.18016 mg glc x 60 min = 0.123 !J.ffiol glc/(mm . mL)

Evaporative Losses
Methods were examined to determine the extent of water loss from
the plates under various incubation conditions, including uncovering the
plates; humidifying the incubator with water-saturated filter pads; and
using plastic lids, mineral oil overlays, or preheated brass lids. Each
method was tested by filling the plate with 80 or 300 ilL of water, weigh-
ing, and heating it at 50°C for 1 h. After incubation, the plates were
weighed and average water loss per well was determined.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
FPA to Determine Cellulase Activity 695

20 •
0
NREL grown - T. reesei Strain L27
Roal Oy Commercial Preps

V
Genencor Spezyme CP
logen 100L

15
•• CPN
Rhodia Penicillium funicu/osum cellulase
0

,2 = 0.9625
.e
...J
0
:::I
a.. 10
u..

O~~~~-r~~~~~~-r~,-~~~~~~~~

o 10 20 30 40 50
BeA Protein (mg/mL)

Fig. 3. Commercial and in-house-produced T. reesei cellulase preparations assayed


for protein content and FPU activity plotted to determine correlation between the two
factors. FPUs were determined using the standard FPA, and protein concentration
was determined using the Pierce BCA method following desalting chromatography
into 20 mM acetate, 100 mM NaCl, pH 5.0 buffer to remove interfering compounds.

Protein Determination
Cellulase preparations were assayed using the traditional FPA and the
FPU value was plotted against total protein as determined using the Pierce
micro-bicinchoninic acid (BCA) method (Pierce Endogen, Rockford, IL)
standardized using bovine serum albumin (Fig. 3). The protein concentra-
tion of the cellulase preparations was measured following desalting of the
enzyme sample into 50 mM acetate, pH 5.0 buffer using a HiPrep 26/10
desalting column prepacked with Sephadex G-25 with a nominal exclusion
limit of 5 x 103 as specified by the manufacturer (Amersham Pharmacia
Biotech, Uppsala, Sweden). The samples were desalted to eliminate interfer-
ing compounds in the preparations, and to standardize the sample buffers.
Only the macromolecular fractions were collected and assayed according to
the manufacturer's instructions.

Enzyme Assays
The traditional FPA was carried out according to Ghose (12). The paper
was carefully rolled and inserted into the assay tubes in order to minimize
irregular and broken fibers that can affect the assay. For the scaled-down
automated version, the assays were carried out in 96-well polypropylene
microtiter plates containing a 1/4-in.-diameter filter paper disk in each well.
The disks were cut with a standard office paper punch and loaded into each

Applied Biochemistry and Biotechnology Vol. 105-108,2003


696 Decker et al.
well by hand using forceps. Since this became very tedious very quickly,
alternative powdered cellulose substrates were explored. These substrates
were also loaded at2.65 mglwell in a 96-wellpolypropylene microtiterplateo
A 2.65% (w Iv) slurry containing 1.5% glycerol was stirred overnight at 4°C
and dispensed by hand from a constantly mixing reservoir using a 12-chan-
nel pipettor equipped with wide-bore pipet tips. One hundred microliters
was dispensed into each well of the assay plate, and the plates were dried for
2 d at room temperature. Initially, the plates were made without glycerol, but
after drying, many of the cellulose pellets spontaneously jumped out of the
wells, either while in storage on the bench or while being moved around the
deck. Glycerol was added in an attempt to alleviate this problem. A concen-
tration of 1.5% was determined to be the minimum needed to prevent the
substrate from popping out of the wells on drying. Regardless of the sub-
strate used, each assay was run with duplicate substrate plates and a
nonsubstrate-containing control plate. Each assay required two assay plates
to be made, and the validation consisted of four assays per run.
The automated assay was carried out as follows. Two.substrate plates
and a blank plate were set on the deck. The enzyme preparation to be tested
was diluted 1:100 or 1:200 and placed on the deck in the first seven rows of
each column of the enzyme master plate. A glucose standard of 3.0 mg/mL
was placed in row eight of each column. Using an eight-channel pipettor, the
robot proceeded to make further dilutions of the enzyme and glucose stan-
dardina 96-well polypropylene master plate. Column 1 contained undiluted
enzyme, columns 2-10 contained dilutions of 1:1.11,1:1.25,1:1.43,1:1.67,1:2,
1:2.5, 1:3.33, 1:5, and 1:10, respectively. Columns 11 and 12 contained only
buffer. Dilutions were made with 50 mM citrate buffer, pH 4.8. After making
the dilutions, the 96-channel pipettor was used to mix the diluted samples by
titration and dispense 80 f.tL to each well of each assay plate: two with sub-
strate and one control plate without substrate. After dispensing, the plates
were moved to the 5°C incubator, covered with preheated brass lids, and
incubated for 60 min. The plates were then transferred back to the deck and
150 f.tL of DNS reagent was added to each plate using the 96-channel pipettor.
After mixing by titration with the 96-channel pipettor, the plates were incu-
bated at 98°C for 10 min, again with preheated brass lids. Plates were then
relocated back to the deck, and 96-well polystyrene dilution plates were set
up containing 200 ilL of deionized water. A 10-f.tL aliquot from each assay
plate (with and without substrate) was aspirated using the 96-channel
pipettor, dispensed into a dilution plate, and mixed by titration (repeated
aspiration and dispensing). A Spectramax 190 microtiter plate reader read
the dilution plates at a wavelength of 540 nm. The data collected was sent to
an ASCII file and imported into Excel. The assay plates, dilution plates, and
master plate were all sent to the waste stacks, and the process was started
over with the next set of enzymes. This process was repeated for up to six
replicates before the robot ran out of tips and plates. Forty-two samples per
run (six replicates x seven samples per replicate) were assayed, with two runs
being carried out per d.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


FPA to Determine Cellulase Activity 697
1.BOO
1.600
1.400
1.200
u
CI 1.000
...J
O.BOO
....ECI
0 .600
E
0.400
0 .200
0 .000
-0 .200
dilution

Fig. 4. Tests of evaporation control carried out in the microtiter plate incubators.
Ninety-six-well polypropylene microtiter plates were filled with the indicated amount
of distilled water, weighed, and incubated at 50°C under the various conditions for 1 h,
and weighed again. Average water loss per well was calculated as (initial wt - final wt) /
96 wells and converted to a percentage based on initial water volume per well. Each bar
represents data from a single 96-well plate (three plates per condition).

Data Analysis
The data exported to Excel contained absorbance data for duplicate
substrate assay plates and a no-substrate control plate, each with its own
internal glucose standard and blanks. To correct for any potential uneven
heating in the three plates, each plate was calculated based on its own inter-
nal glucose standard. Background color development was accounted for by
subtracting the average of the last two columns (no enzyme) from each col-
umnfor each plate. The no-substrate control plate was then subtracted from
each substrate assay plate. The resulting reducing sugar concentrations were
plotted against enzyme dilution, and the dilution yielding 1.325 mg/mL
glucose (3.6% hydrolysis) was determined by interpolation of the closest two
points. This dilution was then divided by the conversion factor to attain the
FPUnumber.
Results
Evaporation Studies
Tests of evaporation control carried out in the microtiter plate incuba-
tors indicated that no lids gave a base loss of approx 45% from 80 f..tL at 50°C
for 1 h (Fig. 4). To minimize this loss, several additional methods were
evaluated. Layering mineral oil over the sample was attempted but showed
a decrease to only a 25% loss. Placing standard polystyrene microtiter plate
lids on the plates or humidifying the chamber by placing water-saturated
filter paper pads along the sides reduced the loss to approx 13%. The result
of using heated brass lids was a water loss of only 7% during the test con-
ducted at 50°C for 60 min.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


698 Decker et al.
Table 1
Summary of FPU Determination
by Automated Assay on Different Cellulosic Substrates
Automated Measurement
Substrate No. of replicates Average FPU /mL SO
Whatman #1 FP 28 62.6 17.2
Whatman #1 FP 28 58.2 6.8
SigmaCell-20 28 33.4 3.9
SigmaCell-20 21 37.1 3.2
Avicel PH101 28 27.8 3.4
Solka-Floc 28 34.7 6.2
Cotton linters 21 14.1 1.2
Whatman #1 FP traditional method 38.6

Automated FPA
Determination of the degree of hydrolysis of Whatman #1 filter
paper by a known cellulase was carried out in both normal scale and
miniature scale. The standard method, carried out by hand, resulted in an
FPU activity of 38.6 FPU/mL in the commercial preparation. Assaying
the same cellulase mixture on Whatman #1 filter paper and other cellu-
lose substrates in microtiter plates resulted in a varying range of activi-
ties. In the microtiter plate-based assays carried out on the C-400, the
commercial cellulase had apparent FPA activities of 60.4 FPU/mL on
filter paper, 35.2 FPU /mL on SigmaCell-20, 34.7 FPU /mL on Solka-Floc,
27.8 FPU / mL on A vicel PH101, and 14.1 FPU / mL on cotton linters. These
results are summarized in Table 1. Representative curves for each sub-
strate are shown in Fig. 5.
Discussion
Cellulase Assays
Current literature describing the assay of total cellulase activity (or of
individual component enzymes) has broadened considerably since the
first reports by Mandels et al. (13) that reducing sugar release and sub-
strate weight loss could serve as suitable cellulase assay methods. To some
extent, and for appropriate substrates, these methods are still considered
generally adequate and the basis for numerous product surveys (14).
However, considering the focus in biomass biotechnology on cellulase
improvement, and the desire to compare cellulase preparations rapidly
using smaller qualities of sample, application of laboratory automation to
cellulase performance measurement is important. To put the automated
assays in proper context, we review next the current state of the art for
cellulase assays.
Applied Biochemistry and Biotechnology Vol. 105-708,2003
FPA to Determine Cellulase Activity 699

Autom~ FPA on WhMman 11 Filter PIII* Autom.~ FPA on Solka Floc


2...,. ,.....----__ --~__,
1.800 .. '
,.~ ...L
1.800
lAOO "". ;:i ..·~. ~~.: .. ,: T.:;r~

, 1.500
-¥.
1.200
:~
~.
~"Y
'-'"
....
.::-~ T
. "'.
-.'
'~' .

iE .
1.000
~ l 'OOO~~~~~~~~~~~ 0 .800
:~ !- "-"-2.
~; ~.

t o!OO~~~----~~~~~~ OJIDC
0._ ~

0.200
0 .000
~~ ~~--------~ ~.200

d llullon

Autom~ FPA on Avlcel

1800 ......- - - - - - - - - - - - ,
11100 1---------:-.,.-----:-""--1
I AOO t----------:-~-t-__;

=.
1.200 1-'----..,.-'---:-r~--'----1
-¥. l'OOO~~~~~----~~__;

I t=~~~::::::~=::==:=:::::--I
0_
0.200 ....... - - , - -........- - - - - -
0.000 t--....-~-~-~--+
~.200 ~-M~~--~.-~~MM~~
dllUlion

Autom_ FPA on Slgmacell.20


2000 ,....._ _ _• _ _ _ _ _ _ _. ,

1.800
1800
I-
ii. ,.200
i 1.000
t 0.Il00 I--.--.,----~-
0.Il00
0._
Il.2OO 1--_ _ ._____. __. ____._ _ _ ----1
0000 l - -_ _ _ _ _ _ _ _ _ _-'

0.00 1 0.002 0003 0.0D4 0.0D6 0.0D6


dUlion

Fig. 5. Microtiter plate-based assays carried out on modified Cyberlab C400 system.
Reducing sugar equivalents concentrations (mg/mL of glucose) were determined
using the DNS method and plotted against various enzyme dilutions used to generate
the reducing sugar equivalents.

IUPAC Methods
As a result of significant effort by an international committee of cel-
lulase researchers and the IUPAC, a procedure was published in 1987
describing the use of filter paper and measurement of reducing sugar by
the DNS method of Miller (15) in the context of a highly specific assay
protocol (12). In fact, the text of this protocol must be followed carefully
to achieve comparable results. The rationale developed in this IUP AC
method is that to be maximally useful, all assays for cellulase activity
must be applied to an identical cellulosic substrate-Whatman #1 filter
paper-and that exposure of enzyme preparation to substrate must be
Applied Biochemistry and Biotechnology Vol. 105-108,2003
700 Decker et al.
permitted to proceed until 3.6% (w /w) of the cellulose in a SO-mg test
coupon; that is, 2 mg, is converted to glucose after a 60-min incubation at
50°C. The concentration (or actually dilution) of enzyme preparation
required to effect this is converted, through a somewhat indirect proce-
d ure, to the cellulase activity in filter paper units per milliliter. For example,
an undiluted cellulase preparation that yields exactly 2 mg of glucose dur-
ing the IUPAC assay has 0.37 FPU /mL. This fractional unit is the lowest
cellulase activity measurable with the IUPAC assay. Note that because the
IUP AC FPU assay is nonlinear owing to hydrolysis of an insoluble sub-
strate of variable structural composition, the use of traditional interna-
tional units of enzyme activities based on initial velocities is invalid. Here,
a single incubation time and temperature are used for all samples.
The IUPAC cellulase assay has many significant limitations; it merely
serves as the best existing method. The IUP AC commission warns, e.g., that
extrapolation of required glucose release from highly dilute or concen-
trated solutions of enzyme is not permitted. Indeed, the assays used to
confirm the release of 2 mg of glucose must be conducted with enzyme
dilutions that closely bracket the actual value. The implication is that cellu-
lase solutions too dilute to release 2 mg of glucose must be either concen-
trated to an appropriate level or pronounced unassayable by the IUP AC
method. This latter issue is important for consideration of the assay minia-
turization dictated by automation in micro titer plates.
Non-IUPAC Methods
Many cellulase enzyme preparations are simply not concentrated
enough to cause the required release of 2 mg of glucose from the SO-mg
filter paper sample in 60 min. If these samples cannot be concentrated
accurately (which is often the case) traditional FPU cannot be measured. In
such cases, however, the IUPAC committee recommends that the reducing
sugar release per unit time be accepted as a "provisional" measure of
enzyme activity. This is similar to the pseudo-initial rate approach often
used in the decade previous to the IUPAC report to measure cellulase
activity from a wide variety of substrates. These substrates may include
filter paper (16), Avicel (17), dewaxed cotton (18), or phosphoric acid-swol-
len cellulose (19). Methods based on the use of antibiotic disks (20) and
turbidity development (21) also predated the IUP AC study. More recently,
Johnson et al. (22) have developed methods to use cellulose solvents to
solubilize cellulose treated with purified enzymes and characterize these
products using size-exclusion chromatography.
Automated Cellulase Assays
To automate cellulase assays, several requirements are necessary,
including the following:
1. Creation of substrate plates.
2. Correct dilution of enzyme stock.
3. Scale-down of assay volume/ substrate.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
FPA to Determine Cellulase Activity 701

4. Consistent and rapid heating during hydrolysis and color


development.
5. Prevention of evaporative losses at low-volume scale.
6. Simultaneous reagent addition/ sampling in the assay plates.
7. Automated absorbance reading of samples.
8. Data analysis.
Of these, precise distribution of substrate and prevention of evapora-
tive losses at low volumes are the most important. The target reaction vol-
ume of 80 ,"",L required evaporative losses to be minimized and substrate
concentration be tightly controlled. Ensuring that these requirements are
met in an automated FPA that can assay up to 84 enzyme preparations per
day cuts reagent use and disposal20-fold and requires only -10% as much
researcher time as the traditional method.
Evaporation Studies
Initially, we expected to be able to seal the plates with a polypropy-
lene / foil laminate heat seal using the ALPS sealer integrated into the deck.
This plate sealing method worked well, but because ofthe frequentpipeting
to and from the sealed plates, gaining access to the well contents with a
pipet tip proved problematic. Even after solving this problem by integrat-
ing a custom seal-piercing tool onto the deck, the tips from the 96-channel
pipettor would get stuck in the holes during pipeting and the plate would
ride up with the pipettor when it retracted. Hold-down pins were then used
to help retain the plates on the deck after pipeting, but this made placement
of the plates difficult, because the plates tended to flex. Increasing the hand
grip enough to hold down the plates resulted in increased difficulty in
inserting the plates onto the deck as well. Additionally, itwas observed that
water tended to condense on the undersurface of the seal during incubation
steps. Sometimes these droplets would return to the well (sometimes not),
thus increasing the scatter in the sample analysis.
As the solutions to these problems began compounding themselves,
it was determined that a simpler method needed to be developed. Mineral
oil was overlaid on the surface of the assay mix with the intention of mini-
mizing evaporation. The oil had an apparent affinity for polypropylene
and tended to creep up the sides of the plate wells, leaving the middle of
the aqueous surface exposed to the air in the chamber. Additional oil was
added, but eventually the volume of oil became prohibitive by diminish-
ing the available assay mixture volume. Polystyrene and polypropylene
lids were used to cover the plates, and similar condensation problems
were noted. In addition, the lids were flimsy and difficult for the C400
gripper arm to handle without distortion. Humidification of the incuba-
tors was attempted by inserting thick sections of blotting paper inside the
chambers and saturating with water. The results showed an improve-
ment, but the blotting paper tended to fall apart and get in the way during
plate insertion and retrieval. There was also a concern that high humidity
in the incubators, which are heated with electric heat tape, would cause
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
702 Decker et al.
problems with the system's electronics. Water also tended to condense on
the incubator lids and drip off as the gripper arm moved around the lids.
Heated lids, similar to techniques developed for polymerase chain reac-
tion machines, were considered in order to minimize evaporation and
condensation. Initial thoughts regarding electrically heated lids with wires
attached were quickly discarded in favor of simply maintaining the lids in
the hot incubator and allowing their retained heat to control condensation.
The prototypes were constructed of lead and had numerous problems;
low heat capacity, easy deformation by the gripper, and their weight
stressed the gripping system. The final design was based on brass lids,
which solved all of the problems associated with lead and allowed us to
minimize water loss from the cellulase assay plates, both at 50°C for 60
min, and at 98°C for 15 min.
FPU-to-Protein Correlation
We have determined that at least for fungal cellulase mixtures, FPU
activity is closely correlated to total protein content, allowing us to approxi-
mate FPU activity quickly and give a good starting point for determining
a hard FPU value. Analysis of commercially available cellulase prepara-
tions and in-house-produced Trichoderma reeseibroths demonstrated a wide
variance in both protein content and FPU activity. A direct comparison of
these two properties yielded an interesting correlation: the number of FPU
was directly related to the protein content (Fig. I). This holds for both
T. reesei-derived cellulases and at least one non-T. reesei commercial prepa-
ration. The samples were desalted prior to determination of protein by BCA
and FPU activity. Previous work in this area had shown a tendency for
stored commercial cellulases to contain a high amount of partial peptides
and low molecular weight stabilizing compounds. Several cellulase prepa-
rations were examined for filter paper activity before and after desalting
and no difference was detected (data not shown). Although the data have
some scatter in them, it is likely that such a plot can be used to give a first
approximation of the filter paper activity of a cellulase preparation based
on the protein content. This will allow fewer and narrower dilutions in
determining the FPU value of cellulase samples.

Acknowledgments
This work was funded in part by the Ethanol from Biomass Program
of the Biofuels System Division of the US Department of Energy.

References
1. Wyman, C. E., Bain, R. L., Hinman, N. ,D. and Stevens, D. J. (1993), in Renewable
Energy: Sources for Fuels and Electricity, Johansson, T. B., Kelly, H., Reddy, A. K. N.,
and Williams, R.,H., eds., Island Press, Washington, DC, pp. 865-924.
2. Sheehan, J. J. (1994), in Enzymatic Conversion of Biomass for Fuels Production, vol. 566,
Himmel, M. E., Baker, J. O. and Overend, R. P., eds., American Chemical Society,
Washington, DC, pp. 1-52.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


FPA to Determine Cellulase Activity 703
3. Bergeron, P. (1996), in Handbook on Bioethanol., Wyman, C. E., ed., Taylor & Francis,
Washington, DC, pp. 89-103.
4. Himmel, M. E., Ruth, M. F., and Wyman, C. E. (1999), Curro Opin. Biotechnol. 10,358-
364.
5. Sheehan, J.S. and Himmel, M.E. (2001), AGRO-food-INDUSTRY HI-TECH 12(5),
TeknoScienze, Milano, Italy, 54-57.
6. Hsu, T.-A. (1996), in Handbook on Bioethanol., Wyman, C. E., ed., Taylor & Francis,
Washington, DC, pp 179-212.
7. Himmel, M. E., Adney, W. S., Baker, J. 0., Elander, R, McMillan, J. D., Nieves, R A.,
Sheehan, J. J., Thomas, S. R, Vinzant, T. B., and Zhang, M. (1997), in Fuels and Chemi-
cals From Biomass, vol. 666, Saha, B., Woodward, J., eds., American Chemical Society,
Washington, DC, pp. 2-45.
B. Himmel, M. E., Adney, W. S., Baker, J. 0., Nieves, R A. and Thomas, S. R (1996),
in Handbook on Bioethanol., Wyman, C. E., ed., Taylor & Francis, Washington, DC,
pp.143-161.
9. Crameri, A., Whitehorn, E. A., Tate, E., and Stemmer, W. P. (1996), Nat. Biotechnol.14,
315-319.
10. Zaccolo, M. and Gherardi, E. (1999), J. Mol. Bioi. 285, 775-783.
11. Ghose, T. (1984), Measurement of Cellulase Activity, Commission on Biotechnology,
International Union of Pure and Applied Chemistry, New Delhi, India.
12. Ghose, T. K. (1987), Pure Appl. Chem. 59,257-268.
13. Mandels, M., Miller, G. L., and Slater, R.W. (1961), Arch. Biochem. Biophys. 93, 115-121.
14. Nieves, R A., Ehrman, C.L, Adney, W. S., Thomas, S. R, and Himmel, M. E. (1998),
World J. Microbiol. Biotechnol. 14,301-304.
15. Miller, G. L. (1959), Anal. Chem. 31,426-428.
16. Mandels, M., Andreotti, R, and Roche, C. (1976), Biotech. Bioeng. Symp. 6,21-33.
17. Wood, T. M. and McCrae, S. I. (1977), Carbohydr. Res. 57,117-133.
lB. Ghose, T. K., Pathak, A. N., and Bisaria, V. S. (1975), in Proceedings of the Symposium
on Enzymatic Hydrolysis of Cellulose, Bailey, M., Enari, T.-M., and Linko, M., eds., The
Finnish National Fund for Research and Development (SITRA), Helsinki, Finland,
pp.111-136.
19. Shoemaker, S. P. and Brown, R D., Jr. (1978), Biochim. Biophys. Acta 523, 147-161.
20. Sheir-Neiss, G. and Montenecourt, B. S. (1984), Appl. Microbiol. Biotechnol. 20,46-53.
21. Johnson, E. A., Sakajoh, M., Halliwell, G., Madia, A., and Demain, A. L. (1982), Appl.
Environ. Microbiol. 43, 1125-1132.
22. Johnson, D. B., Shoemaker, S. P., Smith, G. M., and Whitaker, J. R (1998), J. Food
Biotechnol. 22,301-319.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0705/$20.00

Adsorption of Components
of Enzymatic Synthesis of Ampicillin
on Different Hydrophobic Resins

MARCELO F. VIEIRA, l MARLEI BARBOZA, 2


AND RAQUEL DE LIMA C. GIORDANO*,l

1Departamento de Engenharia Qufmica,

Universidade Federal de Sao Carlos, Via Washington Luiz, Km 235,


CEP 13565-905, Sao Carlos, SP, Brazil, E-mail: raquel@deq.ufscar.br; and
2Departamento de Engenharia Qufmica, UNAERP,
Universidade de Ribeirao Pre to, Avenida Costabile Romano, 2201,
Cx. Postal 98, CEP 14096-380, Ribeirao Preto, SP, Brazil

Abstract
This work compared the performance of three hydrophobic resins for
the adsorption of ampicillin (AMP), D-phenylglycine (PC), D-phenylglycine
methyl ester (PCME), and 6- aminopenicillanic acid (6-AP A). The influence
of pH on adsorption efficiencies was assessed in the range of 4.5-8.5, at 4
and 25°C. The values at 4°C were slightly higher than those at 25°C. The
adsorption efficiency of AMP and 6-AP A decreased at higher pHs, for the
three resins. An opposite behavior was found for PGME, and the pH did
not affect PG adsorption efficiency. Isotherm models were fitted to experi-
mental equilibrium data and the best models were discriminated.

Index Entries: Ampicillin; purification; adsorption; ~-lactamics.

Introduction
Ampicillin (AMP) is one of the most widely used ~-lactam antibiotics,
with an annual production of 5.6 t (1). The enzymatic production of semi-
synthetic ~-lactam antibiotics has acquired a great relevance in order to
avoid the drawbacks of the conventional chemical processes, such as the
high toxicity of some reagents or the requirement of a high number of
synthetic steps (2).
The economic viability of a biochemical process depends not only on
the innovations achieved in reaction steps, but also on the innovations and
optimization of downstream processes (3).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 705 Vol. 105-108, 2003


706 Vieira et at.
Adsorption on polymer resins is a current method used for the
removal of chemicals and pharmaceuticals from dilute liquid mixtures
(4). Important advantages of these resins when used for such applications
are their easy regeneration and high selectivity. Several investigators con-
sidered the use of these resins to recover the antibiotic cephalosporin C
(5,6). Grzegorczyk and Carta(7) have reported equilibrium data for the
adsorption of penicillin-G by a number of porous polymeric adsorbents.
Chaubal et al., (8) studied the equilibrium for adsorption of penicillin V,
tetracycline, and cephalosporin C onto neutral polymeric adsorbents.
In the present study, the performance of the resins Amberlite®
XAD-4, Amberlite® XAD-7, and Duolite® XAD-761 for the adsorption of
AMP, D-phenylglycine (PG), D-phenylglycine methyl ester (PGME), and
6-aminopenicillanic acid (6-APA) was assessed. Temperature and pH
effects were also investigated. Adsorption equilibrium data of the four
components, for the three resins, were obtained and the respective iso-
therms discriminated.

Materials and Methods


Resins
Amberlites XAD-4 (polystyrene-divinylbenzene), XAD-7 (aliphatic
ester), and Duolite XAD-761 (phenol-formaldehyde) were obtained from
Rohm and Haas. AMP, PG, PGME, and 6-APA were obtained from Sigma
(St. Louis, MO).
Preparation of Resin
The adsorbents were pretreated with methanol to remove any trace of
ultraviolet-absorbing material. The beads were then washed with distilled
water and dried at 30°C for 24 h.
Influence of Temperature and pH on Adsorption Efficiency
The effect of temperature and pH was evaluated in batch experi-
ments. Preweighed amounts of hydrated adsorbent (0.25g dry basis, W)
were placed in test tubes containing 5 mL (V) of a solution with a known
initial solute concentration (Co). The tubes were then placed in a thermo-
static bath at the selected temperature and slowly stirred for a minimum
of 2 h at 25°C and 24 h at 4°C. From preliminary experiments, these time
spans were found to be sufficient to reach equilibrium. The compound
equilibrium concentration (C*) in the solution was determined by UV
spectroscopy at 240 nm. Experimental equilibrium adsorption data were
obtained by the same method. These data were correlated to linear,
Freundlich, and Langmuir models. The amount of compounds adsorbed
on the resin (q*) was estimated by mass balance using Eq. 1:

(1)

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Adsorption of AMP, PG, PGME, and 6-APA on Resin 707
The adsorption efficiency percent, %Ae, was calculated as follows:

(2)

Determination of Dissociation Constants


The dissociation constants of PG, AMP, PGME and 6-APA were deter-
mined at room temperature using a Titrine pHstat Metrohm. The compounds
(5-10 mM) were dissolved in water and titrated with 0.1 N NaOH or 0.1 N
HCl. All pKa and pKb values were obtained from the average of three experi-
ments. Isoelectric points (pI) were calculated as: pI = (pKa + pKb)/2.

Results and Discussion


Effect of Temperature
The results presented in Table 1 show that the adsorption efficiency
of the compounds on the three resins slightly increased as temperature
decreased. In most cases, adsorption efficiencies values obtained at 4°C
were about 10% higher than those at 25°C, for all resins. However, this
difference is too low to support a choice of a lower temperature for indus-
trial operation, in view of the higher energy costs to work at 4°C. The resin
XAD-4 performed best.
Effect of pH
As can be seen in Figs. 1-3, the pH has influenced in different ways the
adsorption equilibrium (at 25°C) of the compounds for the hydrophobic
resins that were tested. All pK values mentioned hereon were measured
according to the methodology previously described.
AMP presents values for pKa and pKb of 2.66 and 7.24, respectively. The
increment in pH above the isoelectric point (pI) of AMP (4.95) produces an
increase in the net negative charge of the compound. In these conditions, the
electrostatic interactions between AMP molecules and the aqueous mobile
phase become higher than the hydrophobic interactions with the resins,
reducing the adsorption capacity of the stationary phase. Futhermore,
the increase in the net negative charge of AMP molecules will increase
the repulsion among adsorbed molecules, and that effect may reduce the
adsorption efficiency of the resins.
PGME does not present an amino acid behavior, since it possesses
only the amine group (pKb = 6.89). Hence, an increment in pH leads to an
increase in the concentration of the ester neutral form; that is, more PGME
molecules become neutral when the pH increases to about 6.89, which may
cause a reduction of electrostatic interactions between PGME molecules
and the mobile phase, improving the adsorption capacity of the resins. For
the same reason, the repulsion among its molecules adsorbed on the resins
may be reduced, increasing the adsorption efficiency of PGME molecules.
This may explain the opposite behavior found for PGME and AMP.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
708 Vieira et al.
Table 1
Effect of Temperature on Adsorption Efficiency (%Ae)
on Three Resins at pH 6.5 and 25°C

XAD-4 XAD-7 XAD-761

25°C 4°C 25°C 4°C 25°C 4°C

PG 15 19 7 9 15 18
PGME 63 70 54 56 51 64
6-APA 19 20 10 10 17 25
AMP 76 79 32 48 49 56

75 -..r--APA
..-. ~PG
~ ---A--- PG ME
£)'60 --o--AMPI
c:
Q)
·u
~ 45
c:
.Q
a.
030
I/)

~
15

O+-~~~~--~~~~~~~-r--~
4,0 4,5 ,0 5,5 6;0 6;5 ,5 8;0
pH

Fig. 1. Influence of pH on the adsorption efficiency (%Ae) of compounds on XAD-


761 resin. T = 25°C.

The adsorption efficiency of 6-APA presented the same behavior as


AMP. 6-APA presents values for pKa and pKb of 2.47 and 4.93, respec-
tively. In consequence, it has a low pI (3.7), so in the pH range used herein
(4.5-8.5) 6-APA molecules always present a net negative charge. In this
way, the increment of pH produces an increase in this net negative charge,
causing the electrostatic interactions between 6-APA and the mobile
phase molecules to be higher than the hydrophobic interactions with the
resins, thus reducing the adsorption capacity of the stationary phase.
In the case of PG, the adsorption efficiency values showed a 15% boost
with increasing pH. PG has a pI of 5.49, higher than AMP, and its amine pKb
group is 9.02 and its pKa is 1.96. Thus, for the pH range studied, the concen-
tration of PG molecules with charged amine groups (and therefore posi-
tively charged) diminished very slowly over the entire pH range.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Adsorption of AMP, PC, PCME, and 6-APA on Resin 709

()' 45 ---fJ--N'A
c ----*"- PG
Q)
'13 ----A-- PGME
:E -u-AMPI
~ 30
o
...
~
o
'"
~ 15
9 c

5 6 7 8 9
pH
Fig. 2. Influence of pH on the adsorption efficiency (%Ae) of compounds on XAD-
7 resin. T = 25°C

i:>'
c
Q)
'13
:E ______"- fJPA
Q)
c ----*"- PG
o ----A-- PG ME
~
o
-n--AMPI
'"
~
:-=::: 0-
N ~
-K
0

4 5 6 7 8 9
pH

Fig. 3. Influence of pH on the adsorption efficiency (%Ae) of compounds on XAD-


4 resin. T = 25°C.

Adsorption Isotherms
Batch experiments were carried out in a stirred tank in order to study
the equilibrium of the adsorption of AMP, PGME, 6-APA, and PG on
the hydrophobic resins XAD-4, XAD-7, and XAD-761. All experiments
were carried out at 25°C and pH 6.5. The experimental data were fitted to
the linear, Freundlich, and Langmuir models (Eqs. 3-5). The Lenberg-
Marquardt algorithm for nonlinear least squares fitting was employed
Applied Biochemistry and Biotechnology Vol. 105-108,2003
)..
:g
~
c..
OJ
0'
9-
3v;'
~
::J
""c.. Table 2
OJ Best Adsorption Isotherm Models Fitted for Adsorption of AMP, 6-APA, PGME, and PG
0' on Resins XAD-761, XAD-7, and XAD-4, with Respective Parameter Values of Models
fii
9-
::J
o XAD-761 XAD-7 XAD-4
~ Compound Isotherm Parameter Isotherm Parameter Isotherm Parameter
AMP Freundlich KF = 21.0 Linear KH = 11.8 Langmuir qm = 376.2
.....I (r 2 = 0.9984) n = 0.66 (r 2 = 0.9991) (r2 = 0.9988) KL = 5.45
'"
o
6-APA Langmuir qm = 190.0 Linear KH = 2.4 Linear KH = 5.2
(r 2 = 0.9986) KL = 42.1 (r 2 = 0.9980) (r2 = 0.9992)
PGME Langmuir qm = 590.2 Freundlich KF = 15.5 Langmuir qm = 1184
(r 2 = 0.9992) KL = 21.4 (r 2 = 0.9992) n = 0.81 (r 2 = 0.9998) KL = 39.4
PG Linear KH = 4.4 Linear KH = 2.5 Linear K H =4.0
(r2 = 0.9976) (r2 = 0.9986) (r 2 = 0.9995)

~
:-
-~
Cl
,00
N
§
Adsorption of AMP, PC, PCME, and 6-APA on Resin 711

2~r-----------------------------~
220
./
• XAD-761
o XAD-7
... XAD-4
,-",.--__ or,)

/JY
",;4/

-- ~.-----
._---"

/~
f _x----<
I /-',.

I ~;/;;-/
6 0i..' -9'-Y"
40
20~
O~~-r~~~~r-~~~'-~~~~
o 2 4 6 8 10 12 14
C*(mg/mL)

Fig. 4. Adsorption isotherms of AMP on the three resins: (0, ., _) experimental


data; (---) fitting with models.

(using the software Microcal Origin 6.0). An example of the fitting of the
three tested isotherm models is shown in Fig. 4, for AMP. Tables 2 con-
tains the parameters of the best model fitted for each compound, on the
three resins. To discriminate which isotherm was the "best model," two
criteria were used: the minimum sum of the squares of the residues, and
a nonbiased distribution of these residues:
*
qm C
q* = (3)
KL +C *

q* =KF Con (4)

q* =KHC* (5)
in which KL (mg/ mL) and qm (mg/ g) are constants of the Langmuir equa-
tion, KF (mL/mg) and n are constants of the Freundlich equation, and
KH (mL/mg) is the constant of the linear equation.
In Fig. 4 we provide a comparison of the adsorption isotherms of
AMP for the three adsorbents. As can be seen, the adsorption capacity of
XAD-4 was much greater than for XAD-7 and XAD-761. For example,
when the equilibrium concentration in the solution was 6 mM, the
amount of AMP adsorbed was approx 190 mg/ g for XAD-4, being only
65 mg/ g for both XAD-7 and XAD-761. Table 2 presents the parameters
corresponding to the best model fitted for all the tested compounds, on
each resin.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
712 Vieira et al.
It can be observed from the results in Table 2 that for 6-APA, XAD-4
resin reached higher adsorption capacity values thanXAD-7 and XAD-761.
Using the fitted models, it is possible to calculate that at a concentration of
15 mM, the amounts of 6-APA adsorbed would be approx 75, 45, and 30
mg/ g, for XAD-4, XAD-7, and XAD-761 resins, respectively.
The results showed in Table 2 also demonstrate that for PGME,
XAD-4 resin presented the best performance in the adsorption of the
compound. For the sake of comparison, the calculated amount of PGME
adsorbed on the resin, in equilibrium with 15 mM PGME in liquid phase,
would be approx 320, 230, and 140 mg/g for XAD-4, XAD-761, and
XAD-7, respectively. The results for PG, given in Table 2 indicate similar
adsorption capacities for the three tested resins. The calculated amount
of PG adsorbed on XAD-4 and XAD-761 was about 11 mg/g, while on
XAD-7 it was 7 mg/ g (all values were taken at an equilibrium concentration
in the solution of 3 mM PG).
Because PG presents low solubility, the range of concentrations used
to determine its equilibrium isotherms was very low. This may explain
why linear isotherms were obtained for equilibrium adsorption data of PG,
for all resins.
L Comparison of the adsorption isotherms of 6-APA, PG, PGME, and

AMP for the different resins shows that the highest adsorption capacities
were attained for XAD-4, for all compounds. Higher selectivity was
achieved using XAD-4 resin, too: AMP /PGME =2.2 (at pH 4.5), AMP /PG
=7.0 (at pH 8.5), and AMP / 6-APA ( 4.5 (pH between 7.5 and 8.5).
Conclusions
The effects of temperature and pH on the adsorption efficiency of
6-APA, PGME, PG, and AMP on polymeric hydrophobic resins were stud-
ied. An improvement in the adsorption efficiency of about 10% was obtained
at 4°C, compared to the results obtained at 25°C, for all compounds on the
three resins tested. This enhancement of adsorption capacity, however, was
not enough to overcome the higher energy costs that the industrial operation
at low temperature would impose.
Adsorption was affected by the solution pH, as expected, reflecting
the interactions among polymer resin, adsorbates, and mobile phase. PGME
adsorption increased at high pH values, whereas AMP, and 6-APA pre-
sented an opposite behavior, for all resins used. No significant effect of pH
on the adsorption of PG was observed.
The equilibrium isotherms of 6-APA, PG, PGME, and AMP were
determined in a batch stirred tank. A linear isotherm described satisfac-
torily PG experimental equilibrium data for all resins. The same model
described well 6-APA and AMP adsorption on XAD-7 and 6-APA on
XAD-4. Other equilibrium data were satisfactorily represented by nonlin-
ear models (Freundlich and Langmuir).
The batch tests allowed screening of adsorbents on the basis of
adsorption capacity. The results presented demonstrate that the hydro-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Adsorption of AMP, PC, PCME, and 6-APA on Resin 713

phobic resins can be used in the downstream process of the enzymatic


synthesis of ampicillin.

Acknowledgments
We acknowledge the financial support of the Brazilian research-fund-
ing agency, CNPq, and the research program PADCT-CNPq.

References
1. Ospina,S., Barzana, E., Ramirez, 0.,T., and Lopez-Munguia, A. (1996), Enzyme Microb.
Technol. 19,462-469.
2. Hermindez-Justiz, 0., Terreni, M., Pagani, G., Garcia, J. L., Guisan, J. M., Femandez-
Lafuente, R. (1999), Enzyme Microb. Technol. 25,336-343.
3. Wheelwright, S. M. (1987), Bio/Techn%gy 5(8), 789.
4. Doulia, D., Rigas, F., and Gimouhopoulos, C. (2001), J. Chern. Technol. Biotechnol. 76,
83-89.
5. Kirkby, N. F., Slater, N. K. H., Weisengerger, K. H.; Addo-Yobo, F. and Doulia, D.
(1986), Chern. Eng. Sci. 41(8),2005-2016.
6. Casillas, J. L., Martinez, M., Addo-Yobo, F., AracH, J. (1993), Chern. Eng. J. Biochem.
Eng. 52(3), B71-B75.
7. Grzegorczyk S. and Carta, G. (1996), Chem. Eng. Sci. 51,819-826.
8. Chaubal, M. V., Payne, F. G., Reynolds, C. H., and Albright, R. L. (1995), Biotech.
Bioeng. 47, 215-226.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0715/$20.00

Xylanase Production
by Trichoderma reesei Rut C-30
on Rice Straw

ALEJANDRO COLlNA, l BETZABE SULBARAN-DE-FERRER, l


CATERYNA AIELLO, 2 AND ALEXIS FERRER*,l

Laboratorio de Alimentos, Departamento de Quimica,


1
Facultad de Ciencias, E-mail: aferrerl@cantv.net; and
2Facultad de Ingenieria, Universidad del Zulia, Av. Universidad,
Grano de Oro. M6dulo 2, Maracaibo, Venezuela

Abstract
Xylanase production of Trichoderma reesei Rut C-30 was examined at dif-
ferent initial pH values (4.8,5.9, and 7.0) on rice straw in shake flasks, and in
a fermentor, for the best pH condition. Enzyme performance was tested on
ammonia-treated dwarf elephant grass. The maximum xylanase activities,
92 and 122 IU / mL, were obtained atpH 4.8 in the shake flasks and fermentor,
respectively, in which good growth of the fungus was observed during the
first 24 h and consumption of proteins dissolved from the rice straw caused
the pH to rise later to values between 6.4 and 6.7 (optimal for xylanase pro-
duction). The xylanases from T. reesei were as effective as Multifect XL,
a commercial enzyme preparation, in hydrolyzing ammonia-treated
elephant grass.
Index Entries: Xylanase; Trichoderma reesei; rice straw.

Introduction
Potential uses of lignocellulosics are increasing. The development
of green technologies to use them usually includes several steps such as
enzyme production, substrate treatment, enzymatic hydrolysis of the sub-
strate, fermentation of hydrolysis products, and recovery of fermentation
products (1). Availability of cellulases and xylanases plays a very impor-
tant role in achieving this goaL
The capability of Trichoderma reesei strains to produce active and rela-
tively stable cellulases and xylanases is well known (2,3), but it is still under
active research because many factors affecting production and enzyme activ-
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 715 Vol. 105-108,2003


716 Colina et al.
ity are not completely understood (4,5). It has been reported that xylanase
production by T. reesei Rut C-30 is favored at pH values near neutrality (6,7),
whereas cellulase production has a broaderrange (5,6). However, the perfor-
mance and characteristics of the system are greatly influenced by the selec-
tion of the substrate. In the present work, rice straw was chosen for enzyme
production because it is the third most abundant residue in Venezuela (8).
The objectives of this work were to evaluate the effect of initial pH on xylanase
production by T. reesei Rut C-30 on rice straw as well as its hydrolytic perfor-
mance on ammonia-treated dwarf elephant grass, in comparison with the
performance of an available commercial xylanase.
The results of this research will be used by the Technological Park of
the University of Zulia (Venezuela) in the development of an ammonia
treatment technology.

Materials and Methods


Fermentation Substrate
Rice straw from medium grain grown in Zulia State (Venezuela) was
used as the substrate in all of the fermentations. A mixture of 72.4% amor-
phous cellulose (Solka Floc 40 FCC; International Fiber Corp., North
Tonawanda, NY) and 27.6% oat spelt xylan (Sigma, St. Louis, MO), which
has amounts equivalent to those of the cellulose and hemicellulose present
in rice straw, was used as a control substrate.
Microorganism and Culture Conditions
T. reesei Rut C-30 ATCC 56765 was used. Dilution flasks containing
potato dextrose agar (1 % dextrose) were used for maintenance and sporu-
lation (5-7 d). Spore suspensions were prepared by adding 10-20 mL of
Mandels and Weber's (9) mineral medium to the flasks containing spores
and stored at 4 ± 1°C until use.
Fermentations
Shake-flask experiments were carried out in 250-mL Erlenmeyer flasks
with 1% (w Iv) rice straw in 60 mL of Mandels and Weber's (9) mineral
medium at three initial pH values: 4.8, 5.9, and 7.0 adjusted with 0.1 N HCl
and 0.1 N NaOH. The flasks were kept at 29°C for 120 h at 200 rpm (7) in an
Innova 4300 orbital air incubator (New Brunswick Scientific, Edison, NJ).
The mineral medium and the solids concentration for the control experi-
ment were the same as for the rice straw fermentations; initial pH was
adjusted to 4.8. Spore inoculum was adjusted to 1 x 106 spores/ g of sub-
strate. All fermentations were carried out in duplicate.
A fermentor experiment was set at the best initial pH condition
obtained from the shake-flask experiments. Duplicate fermentations were
carried out batchwise in a 4-L Bioflo III model (New Brunswick) with 4 L
of mineral solution (9) at 29°C for 120 h at 150 rpm and aeration of 0.5 vvm.

Applied Biochemistry and Biotechnology Vol. lOS-lOB, 2003


Xylanase Production by T. reesei on Rice Straw 717
Dissolved oxygen was controlled at 20% saturation. Antifoaming A agent
(Sigma) was used.
Shake-flask sampling was carried out by taking the content of two
random flasks every 24 h. For the fermentor study, a 120 to 130-mL aliquot
was taken every 24 h. Samples were centrifuged at 12,OOOg for 30 min at 4°C.
Enzymatic activity, soluble protein, and pH were determined in the super-
natant, and Kjeldahl nitrogen was determined in the solids, which was
expressed as milligrams of crude protein/ milliliter of culture broth.

Chemical and Enzymatic Analyses


Cellulase and cellobiase activities were determined according to
Ghose (10). Xylanase activity was assayed according to Bailey et al. (11) but
using birchwood xylan (Sigma) instead of birch glucuroxylan. Lowry's
method was used to determine soluble protein (12). Kjeldahl nitrogen was
determined and used, previous nitrogen material balances, as an estimate
of microbial growth at some fermentation periods. The fungus has about
37% crude protein (13). Fiber content of rice straw and dwarf elephant grass
was determined by standard fractionation fiber analysis (14).

Treatment of Hydrolysis Substrate


Untreated and ammonia-treated dwarf elephant grass according to
Ferrer et al. (15)-air-dried, hammermilled to 1 mm, and stored at 4 ± 1°C
for 2 yr-was used to test the efficiency on enzymatic hydrolysis of the
xylanases produced by T. reesei Rut C-30.

Enzymatic Hydrolysis
Enzymatic hydrolysis was carried out at a solids loading of 5% (w / v) in
250-mL Erlenmeyer flasks containing 50 mL of 0.05 M citrate buffer (pH 4.8)
placed at 50°C at 100 rpm for 24 h in an incubator shaker (Innova 4300; New
Brunswick Scientific). Sodium azide was added for preservation (0.15%).
Samples (10 mL) were taken at 0, 6, 12, and 24 h and subjected to reducing
sugar analysis with the dinitrosalicylic acid method (16). Hydrolysis was
performed with the xylanases produced by T. reesei Rut C-30 on rice straw
and with xylanases produced by Trichoderma longibrachiatum (Multifect XL;
Genencor, Helsinki, Finland) at 1 IU / g of dry substrate loading. Both prepa-
rations were supplemented with 1 IU / g of dry substrate of cellulase
(Spezyme CP; Genencor, Rochester, NY) and 5.56 cellobiase units/ g dry
substrate of cellobiase (Novozym 188, Novo Nordisk, Franklinton, NC).

Results and Discussion


The pH increased in the rice straw fermentations (Fig. 1) and particu-
larly for an initial pH of 4.8 during the first 24 h. The fungus grew heavily
during this period, as seen under the microscope, and the growth was
better at lower initial pH (4.8 and 5.9), as the crude protein content in the

Applied Biochemistry and Biotechnology Vol. 105-108,2003


718 Colina et al.
10

~ - - -
~

-•
6 • - : y

~'--X--------l!---------
pH


2

o I I I I

o 24 48 72 96 120

time (h)
-+- pH 4.8 -+-- pH 5.9 --.- pH 7.0 ......X-..... Control pH 4.8

Fig. 1. Variations in pH in shake-flask fermentations of T. reesei Rut C-30 on rice


straw at different initial pHs and on control substrate.

solid fraction indicates (Fig. 2). Although Fig. 2 shows similar slopes for
initial pHs of 4.8 and 5.9, that does not mean that fungus growth was the
same since rice straw proteins were being solubilized in the fermentation
with the initial pH of 4.8 as pH increased to 5.81 (24 h), which makes about
26% of the protein in the solids to be solubilized, meaning that less protein
remains in the solids. In other words, more fungus biomass must be in the
solids for the initial pH of 4.8 compared to fermentation with the initial pH
of 5.9, in which no more protein is solubilized since pH did not increase
much (6.02) by 24 h. The growth of the fungus for the initial pH of 4.8
was as dense as for the control fermentation, in which the pH variation was
about 0.88 pH units (from 4.8 to 3.92), indicating high metabolic activity
as well, and twice as dense as the fermentation with the initial pH of 5.9.
Growth for an initial pH of 7.0 was poor. Crude protein of rice straw was
much higher at pH 4.8 (0.6 mg/mL) than at pH 5.9 and 7.0. This is owing
to protein solubilization from the straw at pH near neutrality, as has been
reported for grasses; the higher the pH, the greater the protein solubiliza-
tion (17). Since crude protein content of rice straw was 6.75% and crude
protein in the solids of a 120-mL sample at the initial pH of 4.8 was 0.6
mg/mL, only 74% of the initial protein was present in the solids; in other
words, 26% of the crude protein was solubilized. On the other hand, initial
crude protein in solids for the pH 5.9 and 7.0 fermentations was 0.38
mg/mL, which represents a 50% protein solubilization.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylanase Production byT. reesei on Rice Straw 719
0.80

0.70

0.60

c 0.50
G_
o-
.. E
Q. ...... 0.40
~
..
-g .....
CD
0.30
0

0.20 ........
x· .................
·X
0.10 .........

...... .......... .. ....... x//


~ '
0.00

o 24 48 72 96 120

time (h)
-+- pH 4.8 ---.- pH 5.9 ---.- pH 7.0···X·· Control pH 4.8

Fig. 2. Crude protein in solids from shake-flask fermentations of T. reesei Rut C-30
on rice straw at different initial pHs and on control substrate.

Figure 3, which shows the true protein (rice straw protein and/or
extracellular enzymes) released to the medium, confirms that soluble pro-
tein was about 26% (estimated by Lowry's method) atthe beginning ofthat
fermentation. Soluble protein decreased between 0 and 24 h in both the 4.8
and 5.9 initial pH rice straw fermentations, indicating protein consump-
tion. However, for pH 4.8 the decrease was about 25% (from 0.16 to 0.12
mg/mL), and for pH 5.9 was just 7.1 % (from 0.28 to 0.26 mg/mL), repre-
senting 100% more consumption of protein (0.04 vs 0.02 mg/mL). It is
known that the pH value in T. reesei fermentations decreases when the
fungus consumes cellulose and does not have either peptone or protein
(18). As pH decreased for the control substrate (no protein), one can infer
that solubilization of proteins from rice straw (6.75% protein) and subse-
quent consumption by the fungus were responsible for increasing the pH,
likely caused by protonation of amino groups released to the medium.
Therefore, since protein consumption is responsible for the increase in pH
in the broth (Fig. 1), this explains the greater metabolic activity at the initial
pH of 4.8, and thus greater fungus growth. Moreover, the difference
observed in protein consumption could be even higher since protein is
being solubilized at a greater extent in fermentation with the initial pH of
4.8, hence increasing the soluble protein in the broth. Later, protein in the
liquid phase started to increase, mainly owing to enzyme production. After
24 h, pH increased to 5.81, and more rice straw protein was released. This
Applied Biochemistry and Biotechnology Vol. 105-108,2003
720 Colina et al.
1.00

0.80

-
E
CI
0.60
.§.
.....
c
'Gi
0
c..
0.40

GI
:a:::J
"0 0.20
cn

o
o 24 48 72 96 120

time(h)

--+- pH 4.8 ___ pH 5.9 - . - pH 7.0 -X, Control pH 4.8

Fig. 3. Soluble pratein in shake-flask fermentations of T. reesei Rut C-30 on rice straw
at different initial pHs and on contral substrate.

was greater than the protein provided by fungus growth, and, as a conse-
quence, crude protein decreased.
The highest fungus growth was achieved in the control with a 0.3
mg/mL crude protein content in the solids by 72 h, which represents true
growth since there was no protein in the control substrate.
The results presented in Fig. 4 indicate that rice straw (initial pH of 4.8
and 5.9) is better as a substrate for xylanase production than commercial
pure cellulose and hemicellulose (control substrate) at the experimental
conditions used. Certainly, substrate selection plays a major role in enzyme
production. The results in Fig. 4 suggest that the pH developed in the rice
straw fermentations was more suitable for xylanase production since it
reached values between 6.0 and 7.0, and it has been reported that optimal
pH values for xylanase production are between 6.0 and 6.5 (7). On the other
hand, the low xylanase production on the control substrate can be explained
by pH va lues far from the optimal at the first stage of the fermentation.
The production of xylanases was higher at lower initial pH, with values
of 92, 46, and 8 IV /mL for initial pH values of 4.8,5.9, and 7.0, respectively
(Fig. 4). Since the pH values reached in the fermentations with initial pH
values of 4.8 and 5.9 were similar by 24 h, the difference in xylanase produc-
tion between both fermentations may be assigned to the greater fungus
growth in the fermentation starting at pH 4.8 (accelerated pH rise in the first
24 h, intense metabolic activity) since this is considered the optimal pH for
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylanase Production byT. reesei on Rice Straw 721
100
90
80

--
C" 70
E
:2 60
>-
:t:
> 50
U
• 40
•c•
II
30

~ 20
10
0
o 24 48 72 96 120
time (h)

---+-- pH 4.8 ___ pH 5.9 ~ pH 7.0 _...)E-.... Control pH 4.8

Fig. 4. Xylanase production in shake-flask fermentations of T. reesei Rut C-30 on rice


straw at different initial pHs and on control substrate.

growth. On the other hand, xylanase production at the initial pH of 7.0 was
very low, likely owing to poor growth during the fermentation. By 96 h,
xylanase production decreased and later increased again. This behavior is
associated with oxygen deprivation in the shake flasks.
Xylanase activity was higher than that reported by Dekker (19) for
T. reesei QM 9414 cultured on sugarcane bagasse (6.4 IV / mL), by Bailey and
Poutanen (20) for Aspergillus oryzae (90 IV / mL) and A. niger (49 IV / mL) on
wheat bran, and by Gomes et al. (21) for Trichoderma viride on rice straw
(72.4 IV/mL) and newspaper (92.4 VI/mL), although lower than that
reported by Gomes et al. (21) for T. viride (190 ill /mL) on sulfited pulp and
by Bailey et al. (7) for T. reesei Rut C-30 on wheat bran (138 ill/mL).
Cellulase activities measured at 120 h showed a different trend com-
pared to xylanases.
Cellulase activity decreased as the initial pH decreased, giving values
of 0.51 ill / mL, 0.29 ill / mL, undetected, and 0.62 ill / mL, for initial pHs of
4.8, 5.7, 7.0, and the control, respectively. When T. reesei Rut C-30 was cul-
tured on rice straw and on the control substrate, production of cellulase was
lower than that found in other studies with the same microorganism (22,23).
Since the cellulase activities of the rice straw fermentation with initial of pH
4.8 and the control fermentation (the same initial pH) were similar, it appears
that cellulase activity was not affected by the pH developed during the fer-
mentation, contrary to what was found in xylanase production.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
722 Colina et al.
7 1.0
- -
14
6 12
-
5 --
E
......
Cl
0.8
-
-
><
'<
- 10 iii
:::I
II
1/1
4 S-c 0.6 II

.--
-8 II
n
cr.
..
'u
J:
Q. <
3 0 6
-
-
Q. 0.4 '<
cu 2
2 :a:::I 4 3
.::

1
"0
en 0.2
- 2
-
0 0.0
- 0 24 48 72 96 120
0

time (h)

_____ Xylanase activity -+- Soluble protein --*- pH

Fig. 5. Xylanase production, pH variation, and soluble protein in BiofIo fermenta-


tion of T. reesei Rut C-30 on rice straw. Initial pH = 4.8.

When the best rice straw fermentation obtained in shake flaks (initial
pH 4.8) was run in a fermentor, xylanase production was greater and more
consistent (123 IV / mL) than in shake flaks, although it was similar by 72 h
(Fig. 5). Better culture conditions in the fermentor could explain this behav-
ior, mainly owing to better agitation and an adequate air supply. As a
result, the decrease in xylanase production found in shake-flask fermenta-
tion by 96 h was not observed in the fermentor. The trends in pH and
soluble protein were similar to corresponding values in the initial pH 4.8
shake-flask fermentation.
Xylanases produced in the fermentor were used to hydrolyze
untreated and ammonia-treated dwarf elephant grass (Table 1) and com-
pared with the performance of a commercial xylanase. As expected, the
treated substrate had greater sugar yields (279-301 mg/ g of dry matter
[DM]) than the untreated one (89 to 90 mg/ g of DM), confirming the effi-
cacy of the ammonia treatment in increasing the susceptibility of ligno-
cellulosics to enzymatic hydrolysis. Susceptible fiber in the untreated
material was almost completely hydrolyzed by 6 h of hydrolysis, whereas
a 24-h period was not sufficient to hydrolyze the available fiber in the
treated material. More important, there were no significant differences
(p < 0.05) in sugar conversion between commercial enzymes and enzymes
produced by T. reesei Rut C-30 on rice straw. A sugar yield of 56% of
theoretical is considered high for a 24-h hydrolysis period and a very low
enzyme loading of 1 IV / g of DM for xylanases and cellulases.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Xylanase Production by T. reesei on Rice Straw 723

Table 1
Reducing Sugar Yield (mg/ g DM) from Enzymatic Hydrolysis
of Untreated and Ammonia-Treated Dwarf Elephant Grass with Xylanases
Produced by T. reesei Rut C-30 on Rice Straw and Multifect XL"
Enzymes Hydrolysis time (h~
and substrates 0 6 12 24
T. reesei
Treated 83 ± 3.83 aA 210 ± 7.95 bA 244 ± 3.32 cA 301 ± 4.07 dA
Untreated 52 ± 6.75 aB 78 ±3.19 bB 94 ± 4.02 bB 89 ± 6.66 bB
Multifect XL
Treated 86 ± 1.34 aA 199 ± 9.91 bA 231 ± 16.6 cA 279 ± 14.75 dA
Untreated 53 ± 10.48 aB 78 ± 3.7 aB 92 ± 4.57 bB 90 ± 7.24 bB
"Results in rows with different capital letters are significantly different (p < 0.05). Results
in columns with different lower-case letters are significantly different (p < 0.05).

Conclusions
The differences observed in the fermentations are basically owing to
the nature of the substrates (lignocellulosic and commercial ones) and to
the effect of the initial and the developed pH on fungus growth and enzyme
production. Rice straw is an excellent substrate for the production of
xylanases.AninitialpHof4.8isappropriatefortheproductionofxylanases.
Xylanases produced by T. reesei Rut C-30 are effective for the hydrolysis of
ammonia-treated dwarf elephant grass.

Acknowledgments
We gratefully acknowledge financial support from the Technological
Park of the University of Zulia (Maracaibo, Venezuela), Fonacit (Caracas,
Venezuela), and Fundacite-Zulia (Maracaibo, Venezuela).

References
1. Kuhad, R and Singh, A. (1993), Crit. Rev. Biotechnol. 13, 151-172.
2. Ryu, D. and Mandels, M. (1980), Enzyme Microb. Technol. 2,91-102.
3. Wong, K. and Saddler, J. (1992), Crit. Rev. Biotechnol. 12,413-435.
4. Gibbs, P., Serviour, R, and Schmid, F. (2000), Crit. Rev. Biotechnol. 20, 17-48.
5. Domingues, F., Quiroz, J., Cabral, J., and Fonseca, 1. (2000), Enzyme. Microb. Technol.
26, 394-401.
6. Royer, J. and Nakas, J. (1989), Enzyme Microb. Technol. 11,405-410.
7. Bailey, M., Buchert, J., and Viikari, 1. (1993), Appl. Microbiol. Biotechnol. 40,224-229.
8. Ministerio de Producci6n y Comercio. (2001), Estadisticas, Caracas, Venezuela.
9. Mandels, M. and Weber, J., (1969), Adv. Chern. Ser. 95, 391-414.
10. Ghosh, V. (1987), Pure Appl. Chern. 59,257-268.
11. Bailey, M., Biely, Peters, and Poutanen, K. (1992), J. Biotechnol. 23,257-270.
12. Lowry, 0., Rosebrough, N., Farr, A., and Randall, R (1965), Anal. Chern. 16, 190-210.
13. Szakacs, G. and Tengerdy, R (1997), World J. Microbiol. Biotechnol. 13,487-490.

Applied Biochemistry and Biotechnology Vol. 705-708,2003


724 Colina et al.
14. Goering, H. and Van Soest, P. (1970), Agricultural Handbook, vol. 379, ARS-USDA,
Washington, DC.
15. Ferrer, A, Byers, F., Sulbaran-de-Ferrer, B., Dale, B., and Aiello, C. (2000), Appl.
Biochem. Biotechnol. 84-86, 163-179.
16. Miller, G. (1959), Anal. Chern. 31,426-428.
17. Bracho, R., Colina, A., Sulbaran de Ferrer B., Ferrer, A, Parra, P., Peters, J., and
Rumbos, C. (2001), in Memorias del V Congreso Venezolano de Quimica, Sociedad
Venezulan de Qufmica, Maracaibo, Venezuela, pp. 637-640.
18. Aiello, c., Ferrer, A, and Ledesma, A (1996), Bioresourc. Technol. 57, 13-18.
19. Dekker, R. (1983), Biotechnol. Bioeng. 25. 1127-1146.
20. Bailey, M. and Poutanen, K. (1989), Appl. Microbiol. Biotechnol. 30,5-10.
21. Gomes, I., Gomes,J., Steiner, W., and Esterbauer,H. (1992),Appl.M icrobiol.Biotechnol.
36,701-707.
22. Hayward, T., Hamilton, J., Templeton, D., Jennings, E., Ruth, M., Tholudur, A,
McMillan,J., Tucker, M., and Mohagheghi,A. (1999), Appl. Biochern.Biotechnol.84-86,
293-309.
23. Shin, c., Lee, J. P., Lee, J. S., and Park, S. (2000), Appl. Biochem. Biotechnol. 84-86,
237-245.

Applied Biochemistry and-Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0725/$20.00

A Different Method of Measuring


and Detecting Mono-
and Dioxygenase Activities
Key Enzymes in Hydrocarbon Biodegradation

ROBERTO ZAZUETA-SANDOVAL, * VANESA ZAZUETA NOVOA,


HORTENCIA SILVA JIMENEZ, AND ROBERTO CABRERA ORTIZ
Instituto de Investigaci6n en Biologfa Experimental,
Facultad de Qufmica, Universidad de Guanajuato, Noria Alta sin,
Apartado Postal 187, Guanajuato, Gto. 36000, Mexico,
E-mail: zazueta@quijote.ugto.mx.

Abstract
A spectrophotometric method of measuring oxygenase activity in cell
extracts or in zymograms was developed. It is an easy and cheap method
that allows spectrophotometric measurement of activity by a colored reac-
tion and reveals activity bands in a polyacrylamide gel electrophoresis
(PAGE) gel as brown bands. To prove its usefulness, we report on a study
with the oxygenase present in strain YR-l, isolated from petroleum-con-
taminated soils, that uses hydrocarbons as its sole carbon source. Soluble
oxygenase activity was detected (under our conditions of cellular homog-
enization) in the mycelium of a filamentous fungus strain named YR-l.
Oxygenase activity from aerobically grown mycelium was detected in
growth medium containing the hydrocarbons decane or hexadecane; the
enzyme activity exhibited similar optimum pH for the hydroxylation of
different aliphatic or aromatic substrates (decane, hexadecane, benzene,
and naphthalene) to the corresponding alcohols. Zymogram analysis con-
ducted with partially purified fractions from cell extracts from the aerobic
mycelium of the YR-l strain indicated the existence of only one oxygenase
enzyme. Partially purified samples of enzyme, analyzed by sodium dodecyl
sulfate PAGE, indicated the presence of one major protein band with a mol
wt of 56 kDa that can be a constituent of the native enzyme. In samples of
the enzyme, the 56-kDa protein gave a positive reaction in immuno-
detection experiments with antibodies directed against oxygenase from

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 725 Vol. 105-108,2003


726 Zazueta-Sandoval et al.
soybean. The partially purified enzyme oxidized different substrates,
although higher activity was displayed with benzene. Km values obtained
for benzene and decane indicated a higher affinity for the latter.
Index Entries: Oxygenases; filamentous fungi; hydrocarbon biodegrada-
tion; petroleum contamination.

Introduction
Hydrocarbons represent an enormous energy resource and are mainly
exploited as fossil fuels. As well as being the predominant energy source in
most countries, hydrocarbons are an important feedstock for the chemical
industry. Their potential impact on biotechnology is enormous, but as yet
their exploitation has been limited. The chemical nature of these compounds
is very diverse, varying from simple saturated aliphatic alkanes to complex
polycyclic aromatic compounds. Such a range of carbon substrates can sup-
port the growth of many microorganisms using diverse and often not well-
understood metabolic pathways. The potential for innovative industrial
application is therefore high. The absence of a basic understanding of the
biochemical nature of hydrocarbon metabolism has slowed down and in
some cases halted many research projects. In addition, a variety of problems
associated with the development of high-productivity fermentation strate-
gies for insoluble high-energy status substrates necessitate an approach
somewhat different from sugar-based fermentation technology (1).
Several possible biochemical pathways are involved in hydrocarbon
biodegradation. The first step in hydrocarbon biodegradation is hydroxy-
lation, which is catalyzed by the cytochrome P-450 protein complex (2).
Two mechanisms have been proposed to explain the incorporation of
molecular oxygen into the hydroxylated product: one atom in the case of
monooxygenases (3), and two atoms in the bacterial dioxygenase case (4).
Cytochrome P-450 is capable of using a wide range of xenobiotic com-
pounds as substrates and, by means of many types of chemical transforma-
tions, leads to the production of alcohols found in microorganisms as well
as in plants and animals (5).
In the present article, we describe an easy and useful spectrophotom-
eter method to measure the oxygenase activity in cell-free extracts and a
variation of the same method for the detection of this activity in zymo-
grams. We also present some biophysical properties of the oxygenase activ-
ity present in cell-free extracts of strain YR-l filamentous fungi isolated
from petroleum-contaminated soils.

Materials and Methods


Reagents and Chemicals
Molecular weights standards, phenylmethylsulfonyl fluoride (PMSF),
and yeast AO were purchased from Sigma (St. Louis, MO). The alcohol
substrates were fromJ.T. Baker (Phillipsburg, NJ) All other reagents were
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Measuring and Detecting Mono- and Dioxyenase 727
of the highest purity commercially available. Protein was measured by the
method of Lowry (11), with bovine serum albumin used as the standard.

Organisms and Culture Conditions


The isolation of filamentous fungi able to grow on hydrocarbons was
performed using as source petroleum contaminated soil samples collected
from the Salamanca refinery (Guanajuato, Mexico). The primary and sec-
ondary selections were achieved using minimal media supplemented
with 1% methanol or 1% hexadecane, respectively. The isolates were
named YR, and the particular strain used was named YR-1; in all cases,
colonial and microscopic morphologies were established as criteria for
the assessment of the isolated strains as filamentous fungi. Yeast-pep-
tone-glucose (YPG) complete media (6), and salt minimal medium supple-
mented with 0.1% peptone (named sMMP), containing the specified
amounts of glucose or hydrocarbons as carbon sources, were used to
cultivate the fungus. The strain was maintained in agar slant tubes and
spores were obtained after grown in YPG medium as described. Liquid
cultures (600 mL) were propagated in 2-L Erlenmeyer flasks inoculated
with spores at a final cell density of S x 105 /mL and incubated in a water
bath shaker at 28°C for different periods of time. To obtain aerobic myce-
lia, spores were inoculated in either YPG or sMMP media supplemented
with glucose (0.1 %), decane (1.0%), or hexadecane (1.0%) and the cultures
were incubated aerobically (6).

Preparation of Cell-Free Extracts


Mycelium cells were processed and broken as described by Torres-
Guzman et al. (7), with some modifications. Briefly, mycelia cells were
washed and suspended in buffer TP8.5 (20 mM Tris-HCI [pH 8.5] contain-
ing 1 mM PMSF). A volume of about 20 mL of cells was mixed with an equal
volume of glass beads (0.45-0.50 mm in diameter) and disrupted in a Braun
Model MSK cell homogenizer (Braun, Melsungen, Germany) for four 30-s
periods under a stream of CO2• The homogenate was centrifuged at 4360g
for 10 min to remove cell walls and unbroken cells. The cell wall-free su-
pernatant (crude extract) was centrifuged at 164,SOOg for 45 min; the result-
ing pellet, a mixed membrane fraction, was discarded and the 164,SOOg
supernatant (cytosolic fraction) was saved for enzymatic determinations.

Enzyme Assays
To evaluate the enzyme activity in cell extracts, it was necessary to
implement a spectrophotometric method. Then, some variations to the
lipoxygenase method were made to detect oxygenase in zymograms after
a PAGE run (8). Briefly, the basic principle was based on the property of
oxygenases to use molecular oxygen to oxidize the hydrocarbon-substrate
molecules by means of an electron transport system (2,4). Similarly, the
enzyme is capable of oxidizing o-dianisidine, producing a brown solution.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


728 Zazueta-Sandoval et al.
Enzyme activity was calculated by interpolating readings at 460 run in a
peroxidase activity calibration curve made with different concentrations of
hydrogen peroxide as substrate of the enzyme in the presence of
o-dianisidine (9). In the case of zymograms, activity bands were detected by
submerging the gel in the reaction mixture (o-dianisidine added to the
substrate and buffer) and incubating the gel at room temperature with
gentle agitation until the activity bands appeared.
The enzyme assays were performed at 25°C in reaction mixtures of 1.0
mL total volume containing 490 J.tL of 0.05M of potassium phosphate buffer
(pH 7.5),400 J.tL of reactant A (20 mg of o-dianisidine dissolved in 15 mL of
absolute ethanol and 0.025 M Tris-HCI buffer, pH 4.5, to 100 mL), 10 J.tL of
the appropriate substrate (aliphatic or aromatic hydrocarbons), and 100 J.tL
of cell-free extract (100-200 J.tg of protein). The reaction was started by
adding substrate, and color development was determined by measuring
the absorbance at 460 run in a Beckman DU-650 spectrophotometer.
In experiments in which the pH of the reaction was varied, phosphate
(50 mM) and Tris-HCI (50 mM) buffers were employed. One unit of enzyme
activity was defined as the amount of enzyme that leads to the production
of 1 J.tmol 02·min at 25°C. Oxygenase specific activity was expressed as
units per milligram of protein. Detection of oxygenase activity in zymo-
grams was performed by nondenaturing polyacrylamide gel electrophore-
sis (PAGE), following a variation of a spectrophotometric method. Briefly,
after nondenaturing 8% (w Iv) PAGE, the gel was submerged in the follow-
ing solution: 10 mL of o-dianisidine reactive (reactive A); 9.0 mL of 0.05 M
Tris-HCI, pH 8.5; and 1.0 mL of substrate (aliphatic or aromatic hydrocar-
bons). After incubating at 25°C for 60-120 min with gentle shaking, oxyge-
nase electromorphs were observed as brown bands.
Substrate Specificity
To test oxygenase specificity, enzyme activity was assayed in the
presence of one of the following substrates, each at a final concentration
of 100 mM:
Aliphatic: octane, decane, hexadecane, and tetracosane;
Aromatic: benzene, naphthalene, phenanthrene, anthracene, pyrene,
toluene, and chlorobenzene.

Partial Enzyme Purification


The 164,500g supernatant was used as the starting material for the
purification steps, which were performed at 4°C. The supernatant was
loaded onto a DEAE-Prep-Biogel (Bio-Rad) low-pressure chromatogra-
phy column (2.5 4.5 cm) previously equilibrated with 0.5 M Tris-HCI (pH
8.5) buffer. Elution was started with Tris-HCI buffer, followed by elution
with 0.1 M NaCI and subsequently 0.5 M NaCl (flow rate of 0.6 mL/min)
in the same buffer. Fractions (3 mL) were collected, and those containing
enzyme activity were pooled, concentrated by vacuum centrifugation
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Measuring and Detecting Mono- and Dioxyenase 729
(Savant system), dialyzed against 20 rnM Tris-HCl (pH 8.5), and stored at
-70°C. The concentrated fractions were applied to a preparative 6% (w Iv)
polyacrylamide gel and electrophoresed under nondenaturing conditions;
oxygenase activity was assayed, the active band was electro eluted from
the gel, and the eluted protein was used in spectrophotometric assays of
oxygenase activity.
Electrophoresis
Sodium dodecyl sulfate (SDS) PAGE analysis of samples was carried
out in slab gels using 10% (w Iv) polyacrylamide with the buffer system of
Laemmli (10). Standard proteins of 14.4-106.0 kDa were used as markers;
after electrophoresis, proteins were visualized in the gels using a Sigma
silver-staining kit.
Immunoblotting and Immunodetection
After SDS-PAGE, proteins were transferred to a nitrocellulose mem-
brane in a Mighty small Transphor unit (Hoefer TE22; Pharmacia Biotech,
San Francisco, CA). Detection was made with polyclonal antibodies raised
against commercial soybean lip oxygenase and revealed with a second
antibody coupled to peroxidase using 3,3'-diaminobenzidine (Sigma).

Results
Oxygenase Activity in Cell Extracts from Mycelial Cells
Oxygenase activity with hexadecane as substrate was analyzed in a
164,500g supernatant of aerobically grown mycelium of strain YR-1
obtained in sMMP containing 1.0% hexadecane as carbon source. The pres-
ence of oxygenase activity was clearly detected in the cytosolic fraction
(164,500g supernatant) of these-mentioned cells. The appearance of oxy-
genase activity as a function of incubation time in growth medium with
decane was estimated. Enzyme production reached its maximum after
22 h and then declined (Fig. 1); this decrease coincided with the onset of the
stationary phase of growth.
Oxygenase activity with hexadecane as substrate was only detected
when the fungus was grown in minimal media containing decane or
hexadecane as carbon sources. This suggests that the fungus contains an
oxygenase that recognizes hexadecane as substrate and is induced in the
presence of the hydrocarbons decane or hexadecane (not shown).
Oxygenase activity from aerobically grown mycelial cells was mea-
sured over a range of pH using hexadecane as a substrate. Fig. 2 shows that
the optimum pH for the oxidation of hexadecane to the respective alcohol
was approx 8.5 in Tris-HCI buffer. No effect of incubation temperature was
observed at 28 or 37°C (not shown).
To test the possibility of the presence of more than one oxygenase
activity in cell extracts of the strain YR-1, the zymogram for oxygenase
activity was obtained using different concentrations of the 164,500g super-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
730 Zazueta-Sandoval et al.
160 140

..'>
41
>-
:;:;
140
120
120
100
u 100 CD

..
as 80 c:::J
Q)
80
(I)
as 60 'ii
cQ) 60 e
CD
40 40 a..
~
0 20 20
0 0
0 10 20 30 40
Timeh

Fig. 1. Time course of oxygenase activity with respect to incubation time. Enzyme
activity was determined in the 164,500g supernatant from aerobically grown mycelial
cells in sMMP medium supplemented with decane. (A) Protein; (_) oxygenase activ-
ity. *Oxygenase activity expressed as 460 nm/min absorbance.

1.6~-----------------------------------------.

1.4
41
~ 1.2
.s;
t3as
.g 0.8
'0
CD
Q. 0.6
en
~ 0.4

0.2

O+------r-----r-----,------r-----,------r----~
6.5 7 7.5 8 8.5 9 9.5 10
pH
Fig. 2. Effect of pH on oxygenase activity. The enzyme activity was determined in
the 164,500g supernatant from aerobically grown mycelial cells in sMMP supple-
mented with decane as the sole carbon source. The reaction mixture contained 25 mM
potassium phosphate between pH 6.0 and 10.0, Tris-HCl between pH 6.0 and 10.0,
(0.5 M), 0.025 M decane 3,3'-diaminobenzidine in 0.1 M HCl and 100 Ilg of protein of
the 164,500g supernatant. (_) Tris-HCl buffer; (A) potassium phosphate buffer. *Oxy-
genase specific activity expressed as units per milligram of protein.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Measuring and Detecting Mono- and Dioxyenase 731

1 2 3 4 5

Fig. 3. Zymogram of oxygenase activity revealed with hexadecane as substrate.


After native PAGE, in 6% acrylamide, activity bands were revealed using the proposal
method. Lane I, 100 !1g of protein; lane 2, 200 !1g of protein; lane 3, 300 !1g of protein;
lane 4,300 !1g of 5-min boiling protein; lane 5, 300 !1g of protein extract from YR-1 strain
grown in sMMP-glucose medium. Arrow shows the band of oxygenase activity.

natant from mycelial cells grown in hexadecane as sole carbon source.


Figure 3 shows the oxygenase zymogram using hexadecane as the enzyme
substrate; as can be observed, under these conditions there was a dose
response in function of the increase in the amount of enzyme in the assay,
and only one major banq of oxygenase activity was detected in all cases.
The main activity band appeared only when the microorganism was grown
in the absence of glucose and in the presence of decane or hexadecane. In
addition to the major oxygenase activity band, a minor activity band can
be observed with hexadecane (Fig. 3, lanes 1-3). To determine whether the
enzyme was a monooxygenase, a dioxygenase, or a dioxygenase with the
activity of a monooxygenase as well (12), a PAGE was run, and half of it
was revealed with benzene and the other half with hexadecane. Figure 4A
shows only a single dioxygenase activity band (benzene), and Fig. 4B
shows mono oxygenase activity (hexadecane), both cases with a similar
relative electrophoretic mobility, indicating that the enzyme is a dioxy-
genase with monooxygenase capacity.
Partial Purification of Oxygenase
Oxygenase from mycelium grown aerobically for 24 h was partially
purified from the 164,SOOg supernatant by a combination of DEAE-Prep
Biogel column and preparative native PAGE of the pool of samples with
the highest activity level, followed by electroelution of the activity band.
After DEAE-Prep Gel chromatography, the samples with highest oxyge-
nase activity (fractions 22-26) were pooled and passed through a Superose
column in a high-performance liquid chromatography system. Figure S
shows that the bound proteins in the column were differentially eluted by
means of the change in NaCl concentration when a discontinuous NaCl
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
732 Zazueta-Sandoval et al.

A 1 2 B 3 4

Fig. 4. Zymogram of oxygenase activity revealed with different substrates. Oxyge-


nase activity was developed with (A) benzene or (8) hexadecane as substrates. Crude
extracts (164,500g supernatant) of YR-1 strain grown in sMMP-hexadecane were used
as sample. Lanes 1-4 were loaded with 300 !1g of each protein. Arrows shows the bands
of oxygenase activity.

5 0.6

4.5
0.5
4

3.5
ac: 3
0.4 8
=
0 0
IX)
N
2.5 0.3 ~
Q 2 Q
0 0.20
1.5

0.1
0.5

0 0
15 20 25 30 35 40 45 50 55 60 65 70 75

fraction number
Fig. 5. Oxygenase activity elution profile after ion-exchange chromatography. The
164,500g supernatant of mycelium cells grown aerobically in sMMP medium supple-
mented with hexadecane as the sole carbon source was applied onto a DEAE-PREP Gel
column. Elution was done with a discontinuous NaCI gradient (0-0.5 M). Oxygenase
activity is expressed as optical density (OD) at 460 nm (.). Protein content of YR-l is
expressed as 00 at 280 nm (e).

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Measuring and Detecting Mono- and Dioxyenase 733

2 3
A

Oxygenll5C activity

B
I 2 c
kDa

Oxygenase activity _I ~
66 _
4S -
-
29 -

Fig. 6. Heterologous immunodetection of oxygenase subunits from partially puri-


fied fractions of oxygenase. The oxygenase activity band obtained after PAGE was
electroeluted, submitted to SDS-PAGE, and the protein pattern was revealed with
silver stain (not shown) or blotted to a nitrocellulose membrane and immunodetected
using a heterologous antibody directed against lipoxygenase from soybean. (A) lanes
1-3 represent oxygenase activity revealed with hexadecane as substrate; (B) lane 1
represents oxygenase activity after PAGE, and lane 2 represents immunodetection
of native oxygenase; (C) immunodetection of putative oxygenase subunit after SDS-
PAGE. Arrows show oxygenase activity bands ([A] lanes 1 to 3; [B] lane 1); immuno-
detection of native form of oxygenase ([B]2); immunodetection of putative oxygenase
subunits under denaturing conditions (C).

gradient was used. This procedure allowed the removal of most of the
oxygenase activity from the column. As can be seen, there are two peaks
of oxygenase activity; one (fractions 22-26) showed highest activity when
benzene was used as the substrate. The second peak (fractions 64-71)
showed minor activity. Additional steps in the purification protocol
included the concentration of active fractions from the column, electro-
phoresis under nondenaturing conditions in a preparative PAGE gel, and
then electro elution of the enzymatic activity band. In later experiments,
only fractions 22-26 were used, because the other "peak" could be an
artifact owing the fact that its localization is in almost the final volume of
the column.
To gain information regarding the subunit composition of the native
enzyme, after PAGE a zymogram was performed and the activity band
(Fig. 6A) was cut and electroeluted from a preparative gel and submitted
Applied Biochemistry and Biotechnology Vol. 105-108,2003
734 Zazueta-Sandoval et al.
to SDS-PAGE. As shown in Fig. 6C, partially purified samples of oxygenase
were submitted to immunodetection experiments with polyclonal anti-
bodies against soybean lip oxygenase. The results obtained indicated that
two bands of 56 and 76 kDa were immunodetected, suggesting that these
protein bands could be related to lipoxygenase enzyme from other organ-
isms and may have some conserved epitopes. Additionally, the antibody
was capable of recognizing only one protein band in native conditions that
corresponded to the activity band in the zymogram (Fig. 6B).

General Properties:
Substrate Specificity and Preliminary Kinetic Parameters
The partially purified oxygenase was examined regarding substrate
specificity using different aliphatic and aromatic substrates in the enzy-
matic assay; oxygenase activity with different substrates was expressed as
specific activity (Table 1). The enzyme oxidized aliphatic and cyclic sub-
strates, although much higher activity was observed with cyclic substrates.
The complexity of the substrate molecule is important because when the
number of rings increased the activity was reduced. The same phenom-
enon was observed with the aliphatic hydrocarbons; that is, the activity
was lower with the increase in the number of carbon atoms in the molecule.
Km and V max values were determined for benzene and decane (Table 2); the
~ values for benzene (2.79 ~ and decane (0.56 JAM) suggest that the
higher affinity of the enzyme is for aliphatic substrates.

Discussion
We consider that the method implemented in our laboratory to detect
activity in zymograms and measure the oxygenase activity by spectropho-
tometric assay is easy and rapid. Furthermore when compared with the
electrode measurements of consumed oxygen in this oxidation reaction, it
is very inexpensive. This methodology allows us to make precise oxyge-
nase measurements and detect with high qualityactivity in zymograms.
Oxygenase activity in aerobically grown mycelium cells of strain
YR-1 obtained under different nutritional conditions was mainly found
in the 164,500g supernatant. This observation indicated that this fungus
oxygenase activity is cytosolic when using the ballistic homogenization
of cells method (drastic rupture). However, it is important to use a gentle
method of cellular homogenization, such as protoplast formation by lytic
enzymes and osmotic shock homogenization, to be sure of the intracellu-
lar localization of this enzyme. The enzyme was detected only when the
culture media contained a hydrocarbon as the sole carbon source, indicat-
ing a possible induction mechanism. However, the absence of activity
when the microorganism was grown in glucose as the sole carbon source
could indicate a possible regulatory effect of glucose.
The principal difference between the oxygenase from strain YR-1 and
other oxygenases in different organisms is the capacity of the former to use
Applied Biochemistry and Biotechnology Vol. 1O!!-108, 2003
Measuring and Detecting Mono- and Dioxyenase 735
Table 1
Oxygenase Activity from Strain YR-1
Using Different Substrates·

Oxygenase specific activity


Substrate (U / mg protein)

Aromatic
Benzene 18.58
Naphthalene 5.17
Phenanthrene 4.63
Anthracene 4.54
Pyrene 1.054
Toluene 14.35
Chlorobenzene 2.94
Aliphatic
Octane 12.09
Decane6.72
Hexadecane 3.59
Tetracosane 0.106
'Partially purified oxygenase activity by ion-exchange
DEAE-PREP chromatography was assayed as described in
Materials and Methods, except that the substrates listed here
were used in place of decane or hexadecane. All substrates were
of 60 mM final concentration. Activities are expressed as activ-
ity units. The given values are the mean of two independent
experiments with triplicate determinations in each instance.

Table 2
Kinetic Constants of Oxygenase from Strain YR-1·

Substrate Vmax (U /mg protein)


Benzene 2.79 260
Decane 0.56 352
"Oxygenase activity from peak 1 (fractions 23-26) from the
ion-exchange chromatography purification step was assayed
with benzene and hexadecane at pH 8.5, as described in Materials
and Methods. The given values are the mean of two independent
experiments.

complex cyclic hydrocarbons as well as different aliphatic hydrocarbons as


substrates. Kinetic parameters of oxygenase enzyme of strain YR-1 indi-
cated differences regarding the affinity for substrates, mainly the use of
decane over benzene (Table 2); however, maximum activity was obtained
with benzene as the substrate (Table 1). This could be explained in terms of
the Kcat (catalytic constant) of the enzymatic system. The enzyme has a low
1<". value for decane over benzene but could perhaps have a higher Kcat value
Applied Biochemistry and Biotechnology Vol. 705-708,2003
736 Zazueta-Sandoval et al.
for benzene over decane. Nevertheless, it is necessary to make a more
detailed determination of these constants, using samples with high purity
level. The enzyme seems to have two subunits in its native conformation,
made of two proteins of 76 and 56 kDa (Fig. SA); the 56-kDa protein band
was recognized by a heterologous antibody against lipoxygenase from soy-
bean. These results suggested that the oxygenase present in the YR-l strain
is a dioxygenase with the capacity to use both cyclic and aliphatic substrates,
i.e., a dioxygenase with both mono- and dioxygenase activities. In addition,
our study will be of importance in establishing the role of these enzymes in
hydrocarbon metabolism by the YR-l strain.

References
1. Alper, J. (1993), Biotechnology 11, 973-975.
2. Kellner, D. G., Maves, S. A, and Sligar, S. P. (1997), Curro Opin. Biotechnol. 8,274-278.
3. Lindley, N. D. (1992), in Handbook of Applied Biotechnology. Fungal Biotechnology, vol. 4,
Philip, K, Richard, A, Elander, P., and Mukeri, KG., eds., Marcel Dekker, Inc., New
York, NY, pp. 905-929.
4. Gibson, D. T. and Parales, R. (2000), Curro Opin. Biotechnol. 11,236-243.
5. Rogers, M. R. and Kaplan, A M. (1982), Dev. Ind. Microbiol. 23, 147-165.
6. Bartnicki-Garcia, S. and Nickerson, W. J. (1962), J. Bacteriol. 84,841-858.
7. Torres-Guzman, J. c., Arreola-Garcia, G. A, Zazueta-Sandoval, R., Carrillo-Rayas,
T., Martinez-Cadena, G., and Gutierrez-Corona, F. (1994), Curro Genet. 26, 166-171.
8. Funk, M. 0., Whitney, M. A, Hausknecht, E. c., and O'Brien, E. M. (1985), Anal.
Biochem. 146, 246-251.
9. Janssen, F. W., Kerwin, 1. M., and Ruelius, H. W. (1975), in Methods in Enzymology,
vol. 16, Kolowick, S. P. and Kaplan, N. 0., eds., Academic, New York, NY, pp. 364-369.
10. Laemmli, U. K (1971), Nature 227, 680-685.
11. Lowry, 0. H., Rosebrough,N. J., Farr, A 1.,andRandall,R.J. (1951),J. BioI. Chem.193,
265-275.
12. Wende, P., Berhardt, F. H., and Pfleger, K (1989), Eur. J. Biochem. 181, 189-197.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0737/$20.00

Xylanase Production by Penicillium canescens


10-10c in Solid-State Fermentation

YASSER BAKRI, PHILIPPE JACQUES, * AND PHILIPPE THONART


Centre Wallon de Biologie Industrielle,
Faculte Universitaire des Sciences Agronomiques,
Passage des Oeportes, 2, 5030 Gembloux, Belgium,
E-mail: Philippe.jacques@iaal.eudil.fr

Abstract
Filamentous fungi have been widely used to produce hydrolytic enzymes
for industrial applications, including xylanases, whose levels in fungi are
generally much higher than those in yeast and bacteria. We evaluated the
influence of carbon sources, nitrogen sources, and moisture content on
xylanase production by Penicillium canescens to-lOc in solid-state fermenta-
tion. Among agricultural wastes tested (wheat bran, untreated wheat straw,
treated wheat straw, beet pulp, and soja meal), untreated wheat straw gave
the highest production of xylanase. Optimal initial moisture content for
xylanase production was 83%. The addition of 0.4 g of xylan or easily metabo-
lizable sugar, such as glucose and xylose, at a concentration of 2 % to wheat
straw enhanced xylanase production. In solid-state fermentation, even at
high concentrations of glucose or xylose (10%), catabolic repression was
minimized compared to the effect observed in liquid culture. Yeast extract
was the best nitrogen source among the nitrogen sources investigated: pep-
tone, ammonium nitrate, sodium nitrate, ammonium chloride, and ammo-
nium sulfate. A combination of yeast extract and peptone as nitrogen sources
led to the best xylanase production.
Index Entries: Penicillium canescens; xylanase; solid-state fermentation.

Introduction
Xylans are one of the major components in wood and plant materials.
Xylan is a collective name of polysaccharides in which in most cases
~-l,4-linked o-xylopyranose residues are the main constituents. Depend-
ing on their origin, xylans may also contain variable amounts of ara-
binosyl- and 4-0-methylglucuronic acid residues and acetyl groups (1).

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 737 Vol. 105-108, 2003


738 Bakri et at.
The most important enzymes in the xylan degradation are the endo-I,4-~­
xylanases (EC3.2.1.8), which hydrolyze ~-I,4-glycosidic linkages between
xylopyranose units. The xylanases are of considerable interest as catalysts
in various biotechnological applications, such as in bleaching of kraft pulps
(2,3),inbaking(4),andanimalfeed(5).Avarietyofmicroorganisms,includ-
ing bacteria, yeast, and filamentous fungi, have been reported to produce
xylanolytic enzymes (6-8), with the latter being extensively studied (9-12).
The use of purified xylan as a substrate for bioconversion into
xylanases increases the cost of enzyme production. Consequently, for com-
mercial applications, there have been attempts to develop a bioprocess to
produce xylanase in high quantities from simple and inexpensive sub-
strates. Abundantly available lignocellulosic residues are an obvious choice
as substrates (10,11,13,14).
Among processes used for enzyme production, solid-state fermenta-
tion is an attractive one (15). Solid-state fermentation is characterized by
the complete or almost complete absence of free liquid. Water, which is
essential for microbial activities, is present in an absorbed or complexed
form with the solid matrix of the substrate (16,17).
The attraction of solid-state fermentation comes from its simplicity
and its closeness to the natural environment for many microorganisms.
This method has low capital costs for equipment and operating, high volu-
metric productivity, lower space requirements, and easier downstream
~cessing compared with that of submerged fermentation (15,18,19). In
addition, the reactor equipment is often simple and requires less process-
ing energy than for the corresponding liquid submerged fermentation. The
low moisture required to obtain maximum yields of product with fungi
excludes in most instances any problem of bacterial contamination (17,18).
In some instances, such as the delignification of agricultural residues for
increasing their digestibility, solid-state fermentation even offers the chance
for a direct applicability of the fermented product (15).
Therefore, the purpose of the present study was to investigate the
ability of a Penicillium canescens strain to produce xylanase enzyme in solid-
state fermentation using easily available different agricultural wastes such
as wheat straw, wheat bran, pulp beet, and soja meal.
Materials and Methods
Strain
P. canescens 1O-10c was supplied by G. I. Kvesidatse, Institute of Plant
Biochemistry, Academy of Sciences, Tbilisi, Georgia.
Fermentation Conditions
For solid-state fermentation, the medium was placed in a 250-mL
Erlenmeyer flask containing 5 g of substrate and nutrients (based on 100 mL
of liquid medium) plus distilled water to adjust the moisture content. The
fermentation medium consisted of: 10 giL of Na2HP04·2H20, 0.5 giL of
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Xylanase Production by P. canescens 739

KCl, and 0.15 gil of MgS04·7H20, plus nitrogen source. Initial experi-
ments were performed using 5 g of different agricultural residues (as car-
bon sources) and a combination of 0.75 g of yeast extract and 0.2 g of
ammonium sulfate (as nitrogen sources). The effect of the nitrogen sources
was tested using yeast extract, peptone, ammonium nitrate, sodium ni-
trate, ammonium chloride, and ammonium sulfate. The concentration of
nitrogen sources used was 52.5 mg of N / 5 g of wheat straw. In other experi-
ments, yeast extract combined with one of the previous nitrogen sources
was used at a concentration of 105 mgofN/5 g of wheat straw. The pH was
adjusted to 6.5 before sterilization. The flasks were sterilized at 120 0 e for
20 min, then inoculated with a 2-mL spore suspension (105-106 spores / mL).
For submerged cultures, the same medium as just described was
employed using 7.5 gil of yeast extract and 2 gil of ammonium sulfate,
as nitrogen sources. The pH was adjusted to 6.5 before sterilization. Erlen-
meyer flasks (250 mL) containing 50 mL of medium were inoculated with
a 1-mL spore suspension (105-106 spores/mL) and incubated at 30 0 e in a
rotatory shaker (120 rpm). At periodic intervals, the filtrate obtained was
assayed for enzyme activity.
Enzyme Extraction
The solid-state culture in each flask was picked up at different time
intervals. Water containing 0.1 % Tween-80 was added to make the volume
in flask equivalent to 100 mL. The flask's contents were stirred for 1.5 hours
on a magnetic stirrer at laboratory temperature and filtrated. The superna-
tant filtrates were used as the enzyme source.
Enzyme Assays
Xylanase activity was measured according to Bailey et al. (20) using
1% birchwood xylan as substrate; reducing sugars were assayed by a
dinitrosalicylic acid method with xylose as the standard (21). One unit of
enzyme activity is defined as the amount of sugar (in micromoles) pro-
duced per minute of reaction and per milliliter of enzyme solution, in the
assay conditions.
Results and Discussion
Effect of Different Lignocellulosic Materials on Xylanase Production
In the first experiments, we examined the effect of different lignocel-
lulosic materials-wheat bran, beet pulp, soja meal, and wheat straw
(treated with NaOH and untreated)-as carbon sources on xylanase pro-
duction. We observed that maximum enzyme activity was obtained by
using untreated wheat straw (Table 1). Table 1 shows that the time of
cultivation needed to obtain maximum enzyme (7448 U / g) production on
this substrate was 12 d. This substrate was used for further studies. Wheat
bran, beet pulp, and soja meal culture filtrates exhibited moderate activi-
ties. It has been reported that the ratio of cellulose to xylan of the growth
Applied Biochemistry and Biotechnology Vol. 105-108,2003
740 Bakri et a/.
Table 1
Production of Xylanase by P. canescens
in Solid Cultures on Various Lignocelluloses
Substrate (5 g) Day Xylanase (U / g)
Untreated wheat straw 4 2702
6 3024
8 3598
10 5628
12 7448
14 6398
Treated wheat straw 4 1134
6 1792
8 2282
10 2954
12 3332
14 3192
Wheat bran 4 1470
6 1582
8 1414
10 896
12 574
14 294
Pulp beet 4 714
6 938
8 1764
10 1792
12 1512
14 1148
Soja meal 4 1400
6 1932
8 1832
10 1610
12 1372
14 1274

substrate is important for the production of xylanase (22). Carbohydrates


present in untreated wheat straw were by far more effective for xylanase
production. Other studies (23,24) have also identified the same substrate
as being ideally suited for xylanase production in Aspergillus terreus and
Thermoascus aurantiacus cultures. Treatment of wheat straw with alkali
caused a 56% reduction in xylanase activity. Removal of the major part of
the hemicelluloses may be the reason for this performance. This alkali
treatment is also responsible for structural deformity in wheat straw (13).
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Xylanase Production by P. canescens 741

9000

Carbon source to minerai solution ratio

Fig. 1. Effect of carbon source-to-mineral solution ratio on xylanase production by


P. canescens in solid-state fermentation.

Using untreated wheat straw, higher crystallinity of the substrate or


reduced accessibility of hemicellulose apparently increased the induc-
tion of xylanase formation. By contrast, pretreatment, which made the
substrate more easily accessible and more rapidly depleted, caused a
drastic reduction in enzyme production (25,26).
Effect of Moisture Level
To achieve the best conditions to use in solid-state fermentation, the
effects of substrate moisture content on xylanase production were exam-
ined. Several investigators (27-29) have reported that moisture content is
a controlling factor for enzyme formation under solid-state fermentation.
This effect was therefore studied by changing the wheat straw to mineral
solution ratios. Five different ratios were tested. It was taken into consid-
eration that the concentration of soluble medium ingredients was not
changed. Figure 1 shows that the ratios 1:5 and 1:6 (initial moisture level
above 80%) yielded the highest xylanase production. Maximal activity
was attained in the medium with 83% initial moisture content (wheat
straw:nutrient solution ratio of 1:5 [v Iv]). This could be attributed
to faster growth of the organism at higher moisture content and the subse-
quent early initiation of the enzyme production (24). Many investigators
have reported a similar effect of moisture content on xylanase production
(19,30,31). Low moisture content is known to decrease the metabolic
and enzymatic activity, probably owing to reduced solubility of nutrients
from the solid substrate, low swelling, and higher water tension (24,28).
Therefore, the importance of moisture level in solid-state fermentation
media and its influence on microbial growth and product biosynthesis
Applied Biochemistry and Biotechnology Vol. 105-108,2003
742 Bakri et al.

10000

o 0.2 0.4 0.6 0.8


Xylan addltlon(g) to wheat straw

Fig. 2. Effect of addition of xylan to wheat straw on xylanase production by


P. canescens in solid-state fermentation.

may be attributed to the impact of moisture on the physical properties of


the solid substrate. A level of moisture higher than the optimum causes a
decrease in porosity, an alteration in substrate particle structure, a gummy
texture, a lower oxygen transfer, and an enhancement of the formation of
aerial mycelia (28).

Effect of Addition of Xylan


The effect of the addition of xylan on xylanase production was inves-
tigated. Our results show that the addition of xylan to the wheat straw
medium increased xylanase production (Fig. 2). The highest effect was
observed with the addition of 0.4 g of xylan. The addition of small amounts
of a purified xylan to complex lignocellulosic substrate has been found to
be advantageous and increased considerably the production of xylanase
(19,25), since xylan and its derivatives play an important role as inducer of
xylanase formation (10).

Effect of Addition of Glucose or Xylose


The expression of xylanases in fungi is subject to regulation by cata-
bolic repression. In submerged fermentation, the accumulation of reduc-
ing sugars has been reported to have a negative effect on the production
of xylanase (32,33). In our study, in submerged culture, we tested the
effect of the addition of 1% of glucose to 1% birchwood xylan on xylanase
production. The results shown in Fig. 3 indicate the inducible nature of
xylanase production by P. canescens. Whereas a very low xylanase activity
(0.34 IV/mL) was detected in the medium containing only glucose after
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Xylanase Production by P. canescens 743
25

-
......
E 20 .Xylan
Xylan+Glucose
2- 15 • Glucose
<II
C/I
(II
c(II
10

n ~
>. 5
><
0 ~
0 48 72 96 120
Time (h)

Fig. 3. Induction of xylanase activity on different carbon sources by P.canescens in


shake-flask culture (120 rpm) at 30°C.

10000

i
l:'!:
8000
-(/I

~-
-III
UQI 6000
1II..c:
QI~
• Glucose
(/1_
~o 4000 • xylose
11101
>''''
><8- 2000
2-
0
o 2 4 6 8 10
Sugar concentration (Ok)

Fig. 4. Effect of glucose and xylose concentrations added to wheat straw on xylanase
production by P. canescens in solid-state fermentation.

120 h of incubation, in medium containing xylan enzyme activity reached


a level of 24.17 IU / mL. The addition of glucose to the medium containing
xylan produced catabolic repression of xylanase production. We also
tested the effect of increasing concentration of glucose and xylose on
xylanase production by P. canescens in solid-state fermentation. Basal
medium was enriched with different glucose and xylose concentrations.
The results shown in Fig. 4 demonstrate that xylanase production increased
in the presence of 2% glucose and xylose concentrations. Increases in the
glucose or xylose concentrations from 4 to 10% caused a reduction in the
production of xylanase. However, no complete catabolic repression was
observed even at 10% glucose. Previously, P. canescens had been found to
grow well and produce high levels of xylanase in a submerged fermentation
system using lignocellulosic residues as substrate. The presence of glucose
caused severe catabolic repression and decreased xylanase production (32).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


744 Bakri et al.
The ability of solid-state fermentation to minimize catabolic repres-
sion has been reported. Souza et al. (34) demonstrated that the production
of xylanase by Aspergillus tamarii in solid-state fermentation using wheat
bran increased with glucose concentration from 0.1 to 1.0%. In addition,
the enrichment of the solid-state fermentation with different easily
metabolizable sugars such as xylose, fructose, galactose, maltose, and su-
crose at 1% (w /w) minimally affected xylanase production. Archana et al.
(28), compared xylanase production by Bacillus licheniformis A99 in sub-
merged and solid fermentation. They demonstrated that solid-state fer-
mentation served to minimize the repression caused by glucose in contrast
to that in submerged fermentation. Glucose, which completely repressed
xylanase synthesis in submerged cultivation, was tolerated up to 6% in
solid-state fermentation without causing any lowering in enzyme titer
while 8 and 10% glucose levels caused minimal repression. The addition
of glucose increased pectin esterase and polygalacturonase production by
Aspergillus niger in the solid-state system, but in submerged fermentation
the production was markedly inhibited as reported by Maldonado and
Strasser de Saad (35).

Effect of Different Nitrogen Sources


We tested the effect of nitrogen sources on xylanase formation by
P. canescens in solid-state fermentation. Various inorganic and organic
nitrogen compounds were used with a fixed concentration of nitrogen at
52.5 mg ofN / 5 g of wheat straw. The results shown in Table 2 demonstrate
that of the six organic and inorganic nitrogen sources used, yeast extract
and peptone were the best for achieving optimal xylanase production by
P. canescens, 5404 and 3878 U / g, respectively. Our results are in agreement
with those reported in the literature in which it was found that fungi-
produced highest xylanase activities were obtained with organic nitrogen
sources (36,37). Yeast extract has an important role in enzyme synthesis,
probably because this complex nitrogen source contains important ele-
ments that are necessary for the metabolism of fungus (14). Table 3 indi-
cates that using a combination of yeast extract with one of the other
nitrogen sources enhanced xylanase production. The combination of yeast
extract with peptone gave the best result (8932 U / g).

Conclusion
One of the factors determining the large-scale use of xylanases will
certainly be the cost of xylan-degrading enzyme preparation. The cost of
the carbon source, as well as the additional medium components, plays a
major role in the economics of xylanase production. The use of the purified
inducing substrate xylan for large production scale is far too expensive.
Alternatively, less costly carbon sources can be lignocellulosic biomass,
which is and will continue to be available in large quantities, as well as
residual products from industrial processing of lignocellulose. The results
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Xylanase Production by P. canescens 745

Table 2
Effect of Nitrogen Sources on Xylanase Production
Nitrogen Quantity Xylanase
source (g/5 g wheat straw) (U/g)
Yeast extract 0.5 5404
Peptone 0.39 3878
(NH4)2S04 0.24 1820
NaN0 3 0.32 2226
NH4N03 0.15 2072
NH4Cl 0.2 1848

Table 3
Effect of Yeast Extract Combined with Nitrogen
Sources on Xylanase Production
Nitrogen source Xylanase
(105 mgl5 g wheat straw) (U/g)
Yeast extract 4088
Yeast extract + peptone 8932
Yeast extract + (NH4hS04 6720
Yeast extract + NaN0 3 6006
Yeast extract + NH 4N03 5880
Yeast extract + NH4Cl 6916

reported herein indicate that P. canescens can be cultivated under solid-


state fermentation for the production of xylanase using, as carbon source,
available agricultural wastes. Untreated wheat straw gave the maximum
xylanase activity after 12 d of culture. Catabolic repression of simple sugar
such as glucose and xylose was minimized in solid-state fermentation with
wheat straw as substrate. Additionally, the results demonstrate the impor-
tance of initial moisture above 80% for xylanase production by P. canescens
in solid-state fermentation. Among nitrogen sources tested, yeast extract
supplemented with peptone gave the maximum xylanase production.
Although quantitative comparison of xylanase activities reported in
literature is not always possible because no standard enzyme substrate
has been adopted yet, the yields of xylanase productivity from P. canescens
observed in this work were approx 1.5-fold higher than optimum
productivities reported in the literature for some microorganism grown
in SSF (Table 4).
In conclusion, the fungus P. canescens, grown in solid-state fermen-
tation in a simple medium consisting of agricultural byproducts and a
low-cost mineral source, proved to be a promising microorganism for
xylanase production.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
)..
:g Table 4
if
Q..
Optimum Xylanase Production in Solid-State Fermentation by Filamentous Fungi
IJ:l

(")
Initial moisture Cultivation Activity Productivity
:;)-
<b Organism Substrate content (%) conditions (IV/g)a (IV/[Lh])b Reference
3<;;.
~ Allescheria terrestris Beet pulp 80 45°C,3 d 28.3 78.6 38
!l.>
:J Aspergillus ochraceus Wheat bran 60 28°C, 14 d 724 862 25
Q..
IJ:l + xylan
o· Wheat straw 60 28°C, 14 d 488 581
<b
(")
:;)-
:J Aspergillus ochraceus NG-13 Wheat straw 75 28°C, 14 d 2120 1580 39
0
a% Wheat bran 75 28°C, 14 d 2240 1670
'<:
Aspergillus niger 3T5B8 Wheat bran 60 32°C, 3 d 100.65 559 40
+ cellobiose
'..J Aspergillus awamori Sugarcane bagasse 92 30°C, 60 h 2500 3333 14
.j::..
0'>
Chaetomium cellulolytium ATCC 32319 Wheat straw 50 37°C, 10 d 580 1210 26
Corn stover 50 37°C, 10 d 350 729
Humicola langinosa Beet pulp 80 45°C,3 d 72.6 202 38
Penicillium capsulatum Beet pulp + bran 67 30°C, 9 d 279.9 428 41
Penicillium canescens Wheat straw 83 30°C, 12 d 9632 5551 This work
+ xylan
Sporotrichum thermophile Beet pulp 80 45°C, 3 d 27.8 77.2 38
Thermoascus aurantiacus Beet pulp 80 45°C,3 d 96.7 269 38
z;:
:- Thermophilic Wheat straw 67 45°C, 4 d 756 2620 13
0
- Melanocarpus albomyces IIS-68 Holocellulose 67 45°C,4 d 1084 3760
<.n
I
from rice straw
0
-
,Co
N aMaximum activity.
0
0 bVolumetric productivity; calculation is based on the initial moisture content.
w
Xylanase Production by P. canescens 747

Acknowledgments
The authors thank the Atomic Energy Commission Syria (AECS) for
their financial support.

References
1. Poutanen, K., Ratto, M., Puls, J., and Viikari, L. (1987), J. Biotechnol. 6,49-60.
2. Buchert, J., Tenkanen, M., Kantelinen, A., and Viikari, L. (1994), Bioresour. Technol. 50,
65-72.
3. Wong, K. K. L., James, c. S., and Campion, S. H. (2000), J. Pulp Paper Sci. 26,377-383.
4. Si, J. Q. (1995), Patent Cooperation Treaty (PCT), International patent application no.
WO 95/23515 AI.
5. Medel, P., Baucells, F., Gracia, M.I., BIas, c., and Mateos, G. G. (2002), Anim. Feed Sci.
Technol. 95, 113-122.
6. Gilbert, H. J. and Hazlewood, G. P. (1993), J. Gen. Microbiol. 139, 187-194.
7. Gokhale, D. V., Patil, S. G., and Bastawde, K. B. (1998), Bioresour. Technol. 63,
187-191.
8. Coughlan, M. P. and Hazlewood, G. P. (1993), Biotechnol. Appl. Biochem.17, 259-289.
9. Abdel-Sater, M. A. and El-Said, A. H. M. (2001), Intern. Biodeterior. Biodegrad. 47, 15-21.
10. Haltrich, D., Nidetzky, B., Kulbe, K. D., Steiner, W., and ZupanCic, S. (1996), Bioresour.
Technol. 58,137-161.
11. Sachslehner, A., Haltrich, D., Nidetzky, B., and Kulbe, K. D. (1997), Appl. Biochem.
Biotechnol. 63-65, 189-201.
12. Kvesitadze, E., Adeishvili, E., Gomarteli, M., Kvachadze, L., and Kvesitadze, G.
(1999), Intern. Biodeterior. Biodegrad. 43, 189-196.
13. Jain, A. (1995), Process Biochem. 30, 705-709.
14. Lemos,J. L. S., Fontes, M. C. A., and Pereira, N. J. (2001), Appl. Biochem. Biotechnol. 91-
93,681-689.
15. Pandey, A., Soccol, C. R., Nigam, P., and Soccol, V. T. (2000), Bioresour. Technol. 74,
69-80.
16. Cannel, E. and Moo-Young, M. (1980), Process Biochem. 15,2-7.
17. Durand, A. (1998), Biofutur 181, 41-43.
18. Weiland, P. (1986), in Treatment of Lignocellulosics with White Rot Fungi, Zadrazil, F.
and Reiniger, P., eds., Elsevier Applied Science, London, England, UK, pp. 64-76.
19. Ferreira, G., Boer, C. G., and Peralta, R. M. (1999), FEMS Microbiol. Lett. 173,335-339.
20. Bailey, M. J., Biely, P., and Poutanen, K. (1992), J. Biotechnol. 23,257-270.
21. Miller, G. L. (1959), Anal. Chem. 31,426-428.
22. Haltrich, D. and Steiner, W. (1994), Enzyme Microb. Technol. 16,229-235.
23. Ghanem, N. B., Yusef, H. H., and Mahrouse, H. K. (2000), Bioresour. Technol. 73,113-121.
24. Kalogeris, E., Christakopoulos, P., Kekos, D., and Macris, B. J. (1998), J. Biotechnol. 60,
155-163.
25. Biswas, S. R., Mishra, A. K., and Nanda, G. (1988), Biotechnol. Bioeng. 31, 613-616.
26. Dubeau, H., Chahal, D. 5., and Ishaque, M. (1986), Biotechnol. Lett. 8,445-448.
27. Xia, L. and Cen, P. (1999), Process Biochem. 34,909-912.
28. Archana, A. and Satyanarayana, T. (1997), Enzyme Microb. Technol. 21,12-17.
29. Gawande, P. V. and Kamat, M. Y. (1999), J. Appl. Microbiol. 87,511-519.
30. Abdulah, A. L., Tengerdy, R. P., and Murphy, V. G. (1985), Biotechnol. Bioeng. 27,20-27.
31. Dechamps,F.,Giuliano,c., Asther,M.,Huet,M.C.,andRoussos, S.( 1985),Biotechnol.
Bioeng. 27, 1385-1388.
32. Gaspar, A., Cosson, T., Roques, c., and Thonart, P. (1997), Appl. Biochem. Biotechnol.
67,45-58.
33. Kadowaki, M. K., Souza, C. G. M., Simao, R. c., and Peralta, R. M. (1997), Appl.
Biochem. Biotechnol. 66,97-106.
34. Souza, D. F., Souza, C. G. M., and Peralta, R. M. (2001), Process Biochem. 36,835-838.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


748 Bakri et al.
35. Maldonado, M. C. and Strasser de Saad, A. M. (1998), J. Ind. Microbiol. Biotechnol. 20,
34-38.
36. Bansod, F. M., Dutta-Choudhary, M., Srivasan, M. c., and Rele, M. V. (1993),
Biotechnol. Lett. 15, 965-970.
37. Steiner, W., Lafferty, R. M., Gomes, I., and Esterbauer, H. (1987), Biotechnol. Bioeng.
30,169-178.
38. Grajek, W. (1987), Biotechnol. Lett. 9,353-356.
39. Biswas, S. R., Jana, S. c., Mishra, A. K., and Nada, G. (1990), Biotechnol. Bioeng. 35,
244-251.
40. Couri, 5., Terzi, S.c., Pinto, G. 5., Freitas, S. P., and Costa, A. C. A. (2000), Process
Biochem. 36,255-261.
41. Considine,P. J., Buckley, R. J., Griffin, T. 0., Tuohy,M. G.,andCou ghlan,M. P. (1989),
Biotechnol. Techniques 3, 85-90.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0749/$20.00

Protease Production by Streptomyces sp.


Isolated from Brazilian Cerrado Soil
Optimization of Culture Medium
Employing Statistical Experimental Design

LUCIANA A. I. DE AZEREDO, 1 LEDA R. CASTILHO/


SELMA G. F. LEITE/ ROSALIE R. R. COELHO, 1
AND DENISE M. G. FREIRE*,4

71nstitute of Microbiology Prof. Paulo de G6es;


2Chemical Engineering Program, COPPE; 3School of Chemistry; and
40Bq/lnstituto de Qufmica, Universidade Federal do Rio de janeiro,
Predio do Centro de Tecnologia, Bloco A Lab. 549-2,
IIha do Fundao, 21949-900, Rio de janeiro/Rj, Brasil,
E-mail: freire@iq.ufrj.br

Abstract
Streptomyces are important microorganisms because of their capacity to
produce numerous bioactive molecules. In the present work protease pro-
duction, by Streptomyces sp. 594 isolated from a Brazilian Cerrado soil, was
maximized by optimizing a low-cost culture medium composition (casitone
and sugarcane molasses) using statistical experimental design. The final
protease activity (56 U /mL) was 2.8-fold and 58-fold higher than that
obtained in the beginning of this study, and in a previous work, using an
actinomycete selection medium, respectively. Protease production, not
growth associated, appeared to be modulated by an inducer system,
whereby the C/N ratio seemed to playa significant role.

Index Entries: Streptomyces; sugarcane molasses; protease production;


experimental design.

Introduction
The streptomycetes, abundantly found in soils, are a well-known bac-
terial group that are especially important because of their capacity to pro-
duce antibiotics and several classes of enzymes and enzyme inhibitors (1).
Brazilian soils under cerrado vegetation are very peculiar tropical soils that
are not sufficiently explored and probably rich in new actinomycete species
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 749 Vol. 105-108, 2003


750 Oe Azeredo et al.

(2). These soils may constitute an excellent source for the discovery of new
enzymes, including proteases, which are of considerable commercial value
for use in food, pharmaceutical, detergent, and tanning industries (3). Sug-
arcane molasses has not yet been reported in the literature as an adequate
substrate for protease production, but it is a very abundant byproduct of
the sugar industry in Brazil, and therefore is an attractive substrate.
The aim of the present work was to maximize the production of pro-
teases by an actinomycete, Streptomyces sp. strain 594, isolated from a Bra-
zilian cerrado soil. The optimization of casitone and sugarcane molasses
concentrations in the culture medium was performed using a central com-
posite experimental design. The kinetics of microbial growth and protease
production were also investigated.

Materials and Methods


Actinomycete Strain
Streptomyces sp. 594 was isolated from a red-yellow latosol (oxisol)
under cerrado vegetation cover, located on the Central Plateau, Brasilia,
Federal District, Brazil. The screening and identification of this proteolytic
strain are described elsewhere (4).
Preparation and Maintenance of Inoculum
Agar-grown mycelia were inoculated in a 1-L Erlenmeyer flask con-
taining 300 mL of malt extract-yeast extract (MY) medium, pH 7.0. Incuba-
tion was carried out under agitation, using a stirring bar, for 72 h at 28°C,
to obtain fragmented biomass. For quantification, several decimal dilu-
tions of the fragmented biomass suspension were prepared and inocula
spreaded over MY agar medium. After 10 d of incubation at 28°C, colonies
were quantified as the number of colony forming units (CFU) of Streptomy-
ces sp. per milliliter.
Experimental Design and Optimization of Culture Medium
The medium's components, sugarcane molasses (1.01 % Nand 29.6%
C) and casitone (11.3% Nand 45.4% C), were selected previously, using a
factorial experimental design (data not shown). In the present work, their
concentrations were optimized by employing a central composite experi-
mental design (5), with two factors and triplicate of the central point. The
concentrations (% [w Iv]) of molasses and casitone were normalized and
testedat-1,-0.71,O, +0.71, and +1 levels, as shown in Table 1. The software
Essential Regression and Experimental Design (6) was employed to gen-
erate the experimental design and to analyze the experimental results.
Experiments were carried out by seeding standard cell suspensions (105
CFU/mL) in the different culture media and incubating for 96 h at 30°C
under agitation (200 rpm). Enzyme production and overall yield coeffi-
cient (YP/X) (units of protease produced per milligram of cell dry weight)
were evaluated. An empirical model describing the proteolytic activity as
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Protease Production by Streptomyces sp. 751

Table 1
Optimization of Sugarcane Molasses and Casitone Concentration
(Xl and X2, respectively) for Protease Production by Streptomyces sp. 594a
Xl Normalized X2 Normalized Protease activity
Run (%w/v) Xl (% W/v) X2 (U/mL)
1 0.30 0 0.30 0 15.3
2 0.30 0 0.50 +1 12.2
3 0.10 -1 0.30 0 5.51
4 0.44 +0.71 0.44 +0.71 18.2
5 0.16 -0.71 0.44 +0.71 6.53
6 0.30 0 0.30 0 15.6
7 0.16 -0.71 0.16 -0.71 6.89
8 0.30 0 0.10 -1 6.27
9 0.44 +0.71 0.16 -0.71 14.5
10 0.5 +1 0.30 0 21.1
11 0.30 0 0.30 0 16.4
aA central composite design was used, and results are from fermentation experiments
carried out at 30°C for 96 h.

a function of molasses and casitone concentrations was obtained through


statistical analysis of the experimental results.

Analytical Methods
Determination of Biomass
Culture samples were withdrawn daily. Samples were filtered using
0.45-~m membranes, and microbial growth was measured by dry weight
determination. Biomass concentration was expressed as milligrams of dry
cells per milliliter of medium.
Reducing Sugar and Total Nitrogen
Reducing sugar and total nitrogen data were determined from fil-
tered samples according to the methods of Somogyi (7) and Bremner (8),
respectively.
Proteolytic Activity
Extracellular protease activity in filtered samples was determined by
azocasein (2%) assay in 50 mM phosphate buffer, pH 7.0 at 60°C (9). One
unit of proteolytic activity was defined as the amount of protease required
to produce an absorbance increase of 0.01 at 440 nm.

Results and Discussion


Experimental Design and Optimization of Culture Medium
As a consequence of an increase in molasses arid casitone concentra-
tion from 0.1 to 0.5% (w Iv), protease production by Streptomyces sp. 594
varied from 5.51 to 21.1 U ImL (Table 1). Through statistical analysis of the
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
752 Oe Azeredo et al.
results, an empirical quadratic model was obtained, as a function of molas-
ses (M) and casitone (C) normalized concentrations (Eq. 1). To guarantee
that only statistically significant terms were included in the correlation,
only those parameters with a critical significance (p value) < 0.05 were
considered. The regression coefficient of the polynomial model (R) was
0.972 and R2adjusted was 0.920, indicating a satisfactory model adequacy:
Protease (U / mL) = 14.67 + 7.305 x M + 2.073 x C - 5.716 x C 2 (1)
From Eq. 1, it can be seen that in the concentration range covered by
this correlation (0.1-0.5% [w Iv]), molasses concentration has a linearposi-
tive effect on protease activity, while the casitone effect is a more complex,
quadratic one. Through differentiation of the empirical model with respect
to casitone concentration, and by calculating the casitone value that would
equate this first derivative to zero, it was possible to calculate the optimum
normalized casitone concentration as 0.181, corresponding to a real casitone
concentration of 0.336% (w Iv). This behavior can be confirmed in Fig. 1,
where the response surface plot corresponding to Eq. 1 is shown. Therefore,
further experiments were carried out using a casitone level of 0.3% (w Iv).
Casitone concentration commonly cited in current literature for the
production of different enzymes by distinct microorganisms are between 0.2
and 0.5% (10). However, the present results show that casitone concentra-
tions >0.3% may exert a negative effect on protease production by Streptomy-
ces sp., probably owing to a feedback inhibition of enzyme synthesis (11).

Effect of Higher Sugarcane Molasses Concentrations


on Protease Production and Cell Growth
Since the previous experiments showed that molasses had a linear
positive effect on protease production in the whole range of 0.1 to 0.5% (w / v),
a univaried experiment was performed to investigate the effect of higher
molasses concentrations on protease production and cell growth. Casitone
was used at its optimized level (0.3% [w Iv]), and sugarcane molasses con-
centration was varied from 0.5 to 3.0% (w Iv). Figure 2 compares the biom-
ass concentration and protease activity obtained after 96 h of cultivation.
There is a clear positive effect of the increase in molasses concentration on
cell growth, indicating that the media tested previously were carbon lim-
ited. In response to the increase in molasses availability from 0.5 to 3.0%
(media C/N ratio from 7.3 to 15.95), biomass varied steadily from 1.01 to
2.32 mg/mL (Fig. 2A). The overall positive effect of molasses concentra-
tion on cell growth was not equally observed on protease production,
since a maximum protease activity (31 U ImL) was obtained at 1.0%
molasses, decreasing afterward (Fig. 2B). In conclusion, a threshold
molasses availability, related to the medium C/N ratio, apparently trig-
gers a metabolic shift in Streptomyces sp. 594 toward cell growth, to the
detriment of enzyme production. Similar results were reported for differ-
ent enzymes (e.g., lipases) produced by other microorganisms (12).

Applied Biochemistry and Biotechnology Vol. 105-708,2003


Protease Production by Streptomyces sp. 753

. 20-25 3
• 15-20 .€
2-
a 10-15 .~
.?:
a 5-10 '0
III
D 0·5 CI)
III 1,0
III

e
CI)

0...

Normalized casitone
concentration (.)

Fig. 1. Theoretical response surface plot obtained from central composite design for
protease production by Streptomyces sp. 594 as a function of normalized sugarcane
molasses and casitone levels. Experiments were carried out at 30°C for 96 h.

3
c/N - 15.95

1:
C>
'CD :J' 2
~E
Nl>
'OE
=~
CI)
u
0
0 .5 1.0 2 .0 3 .0
Molasses concentration (%. wlv)

40
3 B
~ 30
.~
>
""III0 20
CI)
CIl
III
CI)
10
(5
It 0
0 .5 1.0 2 .0 3 .0
Molasses concentration (%, wlv)

Fig. 2. Effect of higher sugarcane molasses concentrations on (A) cell dry weight and
(B) protease production by Streptomyces sp. 594 using 0.3% (w Iv) casitone in shake-
flask fermentations (170 rpm) carried out for 96 h at 30(C and pH 7.0.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


754 De Azeredo et al.
10 100

o+--__.,;;;...=~-____r--__.___-_+O
o 24 48 72 96 120
Fermentation time (h)

Fig. 3. Typical time course for cell growth, protease production, pH, and sugar
content in Streptomyces sp. 594 shake-flask fermentations (200 rpm, 30°C) using the
optimized medium (0.3% casitone and 1% molasses, pH 7.0) and an inoculum of 105
CFU/mL.

Fermentation Kinetics
The fermentation time course for protease production by Streptomy-
ces sp. 594 at 200 rpm is shown in Fig. 3. The maximum proteolytic activity
(56 U ImL) was observed after 96 h of cultivation, and its production was
shown to be not growth associated, occurring at the middle of the station-
ary phase (Fig. 3). Bascaran et a1. (13) and Porto et a1. (14) observed that
maximum protease production by S. clavuligerus occurred at the end of
exponential and beginning of stationary growth phase. Different results
were obtained by Kim and Lee (15), who observed protease production by
a strain of S. exfoliates in batch submerged fermentations as being associ-
ated with mycelium growth.
According to Fig. 3, sugar and nitrogen contents in the medium
decreased slowly but were not completely consumed along all growth
phases. Similar results for protease production by S. clavuligerus were
observed after nutritional shiftdown, indicating that the beginning of
protease production decreased with nutrient availability (13). By con-
trast, Aretz et a1. (16) and Brabban and Edwards (17), working with some
Streptomyces species, observed that sugars were rapidly consumed from
lag to exponential phase, and then fell to undetectable levels. The mecha-
nism by which control of protease production is achieved in many
prokaryotes systems is not known yet (16).
Employing the optimized medium (0.3% [w Iv] casitone and 1.0%
[w Iv] molasses), the maximum protease activity obtained (56 U ImL; Yp/x
=40 U Img) was 2.8-fold higher, when compared to previous experiments
(data not shown) employing molasses and casitone at a 0.5% (w Iv) con-
centration. When compared to protease activities obtained in a previous
work, using starch casein salt medium (18), a common medium for actino-
mycetes, a 58-fold increase in protease activity was obtained. Lower levels

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Protease Production by Streptomyces sp. 755
of protease were produced by S. viridosporus (0.38 U /mL) (19) when corn
oil was used in shake-flask fermentations.
In conclusion, the present results indicate that sugarcane molasses
and casitone are good carbon and nitrogen sources for protease production
by Streptomyces sp. 594. The kinetics of fermentation showed that protease
synthesis is not associated with biomass growth. Optimization of the con-
centration of medium components, using response surface experimental
design and statistical analysis, enabled a significant increase in proteolytic
activity. The good results obtained confirm this strain to be promising for
protease production using a low-cost carbon source.
Acknowledgments
This work was financially supported by Conselho Nacional de
Desenvolvimento CientHico e Tecnol6gico, Funda~ao de Amparo a
Pesquisa do Rio de Janeiro, Funda~ao Universitaria Jose Bonifacio, and
Programa Nacional de Excelencia.
References
1. Gilbert, M., Morosoli, R., Shareck, F., and Kluepfel, D. (1995), Crit. Rev. Biotechnol. 15,
13-39.
2. 5emedo, 1. T. A. S., Linhares, A. A., Gomes, R. c., Manfio, G. P., Alviano, C. S.,
Linhares, 1. F., and Coelho, R. R. R. (20ot), Microbiol. Res. 155,291-299.
3. Peczynska-Czoch, N. and Mordarski, M. (1988), inActinomycete Enzymes, Goodfellow,
M., Williams, S. T., and Mordarski, M., eds., Academic, New York, NY, pp. 219-283.
4. De Azeredo, 1. A. I., Leite, S. G. F., Freire, D. M. G., Benchetrit, 1. c., and Coelho, R.
R. R. (20ot),]. Microbiol. Methods 45, 207-212.
5. Neto, B. B., Scarminio, I. S., and Bruns, R. E. (1995), in Planejamento e Otimizariio de
Experimentos, Neto, B. B., Scarminio, I. S., and Bruns, R. E., eds., Editora da Unicamp,
Sao Paulo, Brazil, pp. 61-97.
6. Steppan, D. D., Werner, J., and Yeater, R. P. (1998), Essential Regression and Experimen-
tal Design (book and software downloadable, see Website: www.geocities.com/
SiliconValley /network/1032).
7. Somogyi, M. (1952), J. BioI. Chern. 195, 19-23.
8. Bremner, J. M. (1965), in Methods of Soil Analysis, Black, C. A., Evans, D. D., White, J.
1., Ensminger, 1. E., and Clark, F. E., eds., American Society of Agronomy, Madison,
WI, pp. 1149-1178.
9. Sarath, G., Motte, R., and Wagner, F. W. (1989), in Proteolytic Enzymes, Beynon, R. J.
and Bond, J. S., eds., IRL, Oxford, England, UK, pp. 22-25.
10. Lee, D., Lee, E. 1., and Lee, K. M. (2000), J. Microbiol. Biotechnol. 10,208-214.
11. Malathi, S. and Chakraborty, R. (1991), Appl. Environ. Microbiol. 57,712-716.
12. Baillargeon, M. W., Bistline, R. Jr., and Sonnet, P. E. (1989), Appl. Microbiol. Biotechnol.
30,92-96.
13. Bascaran, V., Hardisson, c., and Brana, A.F. (1990), Appl. Microbiol. Biotechnol. 34,
208-213.
14. Porto, A. 1. F., Campos-Takaki, G.M.,and Filho,J. 1. (1996), Appl. Biochem. Biotechnol.
60, 115-122.
15. Kim, I. S. and Lee, K. J. (1996), Microbiology 142, 1797-1806.
16. Aretz, W., Koller, K. P. and Riess, G. (1989), FEMS Microbiol. Lett. 65,31-36.
17. Brabban, A. D. and Edwards, C. (1996), J. Appl. Bacteriol. 80,651--658.
18. Mackay, S. J. (1977), Appl. Environ. Microbiol. 33,227-230.
19. Macedo, J. M. B., Gottschalk, 1. M., and Bon, E. P. (1999), Appl. Biochem. Biotechnol.
77-79,735-744.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0757/$20.00

Synthesis of Monocaprin Catalyzed by Lipase


MARCOA. M. DA SILVA,l VilORIA C. MEDEIROS,2
MARTA A. P. LANGONE/ AND DENISE M. G. FREIRE*,l
lDBq/lnstituto de Qufmiea, Universidade Federal do Rio de janeiro,
Predio do Centro de Teenologia, Bloeo A, Lab. 549-2,
IIha do Fundao, 21949-900, Rio de janeiro/Rj, Brasil,
E-mail: freire@iq.ufrj.br; and
21nstituto de Qufmiea, Universidade do Estado do Rio de janeiro,
Rio de janeiro, Brasil

Abstract
The production of monoglyceride emulsifiers commonly employed in the
food, cosmetic, and pharmaceutical industries can be catalyzed by lipases,
biocatalysts that are becoming increasingly attractive in the enzyme market.
The aim of this study was to produce monocaprin utilizing a commercial
immobilized lipase (Lipozyme 1M 20) through the direct esterification of
capric acid and glyceroL Experiments were performed for 6 h in an open
reactor and the products were analyzed by gas chromatography. The param-
eters investigated were the amount of enzyme, temperature, and molar ratio
between the reagents (capric acid/ glycerol). The experimental runs followed
an experimental design generated using Statistica® software. The results
showed that all the parameters were significant and that monocaprin pro-
duction was enhanced at the lower ranges of the tested variables. The best
conditions established were 55°C, 3% (w /w) enzyme concentration, and
molar ratio of 1. The final product, obtained after 6 h of reaction, was 61.3%
monocaprin, 19.9% dicaprin, and 18.8% capric acid. This composition satis-
fies the directives of the World Health Organization food emulsifiers, which
requires that these mixtures have at least 70% mono- plus diglyceride, and
a minimum of 30% monoacylglyceroL
Index Entries: Lipase; monocaprin; capric acid; glycerol; monoglycerides;
molar ratio; n-hexane; t-butanoL

Introduction
The lipases (Ee 3.1.1.3) constitute a group of enzymes, that is becom-
ing increasingly attractive for the oil industry. Stability and especially
selectivity are the properties that are responsible for their success in
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 757 Vol. 105-108,2003


758 da Silva et al.
biotransformations (1,2). These properties have been used in modification
of natural fats and oils, or for the production of monoglycerides.
Monoglycerides are nonionic emulsifiers that are widely used in the
food, cosmetic, and pharmaceutical industries. These substances are con-
sidered Generally Recognized as Safe products by the US Food and Drug
Administration. This fact has enhanced the development of new synthetic
methods and increased the use of monoglycerides in the food and pharma-
ceutical industries (3).
Mono- and diglycerides are consumed at an annual level of 85,000,000
kg in the United States, corresponding roughly to 70% of the total emulsi-
fiers used in the food industry (4). Recently, it has been proposed that some
fatty acids and mono glycerides, such as monocaprin, may also be used as
intravaginal microbiocides for protection against sexually transmitted dis-
eases (5,6) and as conservative agents in the dairy industries (7,8).
Presently, industrial-scale production of monoglycerides is carried out
by continuous chemical glycerolysis of fats and oils, employing inorganic
catalysts (e.g., Ca[OH]2) at high temperatures (220-250°C). This process is
characterized by a high energetic consumption and also may cause changes
in the sensorial properties of the products, such as color and taste (9).
The enzymatic synthesis of monoglycerides using lipase, besides the
advantages already cited, is carried out at lower temperatures, resulting in
higher-quality products, lower energy consumption, and effluents that are
less aggressive to the environment (10).
Different methods for the enzymatic synthesis of monoglycerides have
been investigated, such as selective hydrolysis using 1,3-regiospecific lipases,
esterification of fatty esters with glycerol, and glycerolysis of fats or oils (11).
The objective of the present work was to study the synthesis of monocaprin
by direct esterification of glycerol with capric acid in the presence of a com-
mercial immobilized lipase with and without the use of solvent.

Materials and Methods


Materials
Lipozyme IM-20 (Mucor miehei , lipase immobilized on a weak anion-
exchange resin) was kindly supplied by Novo Nordisk A/S (Bagsvaerd,
Denmark). Capric acid and gas chromatography (GC) standards (mono-,
di-, and tricaprin) were purchased from Sigma. (St. Louis, MO). Analytical-
grade glycerol, n-hexane, t-butanol, ethyl acetate, ethanol, acetone, lauric
acid (99.9%), and molecular sieves (with an effective pore diameter of 3 A)
were purchased from Merck (Darmstadt, Germany).
Measurement of Lipase Activity
The esterification activity of Lipozyme IM-20 was measured according
to the method described by Langone and Sant'Anna (12), which determines
the consumption rate of fatty acid at 60°C in a reaction system containing
glycerol, lauric acid, and a given amount of the commercial enzyme prepa-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Synthesis of Monocaprin Catalyzed by Lipase 759
Table 1
Variables and Levels of Variation
Level
Variable -1 o +1
T(°C) 40 55 70
R 0.5 1.0 1.5
E (% [w/wD 1 5 9

ration. One international unit of esterification activity is the quantity of


enzyme that consumes 1 JA.mol of lauric acid/min under the reaction condi-
tions. The enzyme used has an esterification activity of 20 IV/g.

Nonpolar Phase Analysis


Fatty acid and mono-, di-, and tricaprin were analyzed by capillary GC
according to the method described by Langone and Sant'Anna (12). All
concentrations were calculated as molar fractions from the peak area using
calibration curves.

Esterification Experiments
All experiments were performed in duplicate in a 20-mL open-batch
reactor with constant stirring and temperature control. The reaction system
contained a mixture of capric acid and glycerol and the biocatalyst
Lipozyme IM-20. In some experiments, a 1:1 mixture of n-hexane:t-butanol
was added to the reaction medium at different volumes (2.0, 4.0, and 6.0
mL), corresponding to 24,38, and 48% of the reaction volume, respectively.
The reaction's progress was followed by withdrawing 20-JA.L aliquots at
various time intervals and analyzing them by Ge, as previously described.
Experimental Design
The effects of different variables on a process can be determined using
experimental design methodology, which employs a reduced but meaning-
ful number of experiments (13). A 23 factorial design with central point was
used to analyze the effects of temperature (n, capric acid/glycerol molar
ratio (R), and enzyme concentration (E) on the selectivity of monocaprin. The
studied values for each variable are given in Table 1. The variables and levels
were chosen as reported in the literature and by preliminary studies (14).
Molecular Sieves
Several tests under the addition of 3-A molecular sieves (0.07 g/mL)
were performed in order to evaluate the effect of water removal on the reac-
tion medium. The'reactions were performed at 55°C with a stoichiometric
molar ratio of the reagents and enzyme concentrations of 3 and 5% (w /w).

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


760 da Silva et at.
~ .--------------------------------,
60

~ 40

20

Fig. 1. Effect of variables (T, R, E) on molar fraction and on selectivity in monocaprin


after 6 h of reaction.

Glycerol Addition
To test the influence of the reagent molar ratio, glycerol was added to
the reaction medium in a fed-batch mode. For the reactions performed with
capric acid/glycerol ratios of 0.5 and 1.0, 50% of the necessary glycerol
mass was added atthe beginning of the reaction (t =0 min), 25% after 30 min
of reaction, and the remaining 25% after 1 h of reaction.
Results and Discussion
Experimental Design
The results of the enzymatic synthesis of monocaprin were expressed
as a molar fraction of the components in the nonpolar phase (capric acid,
monocaprin, dicaprin, and tricaprin). The selectivity parameter, chosen to
define the best reaction conditions, was defined as the ratio of monocaprin
concentration to the sum of the concentration of the reaction products
(mono-, di-, and tricaprin). The results for the reactions are presented in
Fig. 1. As can be seen, the best results in terms of molar fraction (61 %) and
selectivity (70%) of monocaprin were obtained at 55°C, with a reagent molar
ratio of 1 and an enzyme concentration of 5% (w /w).
After 4 h of reaction, equilibrium was attained, so the influence of
variables on monocaprin selectivity was analyzed at this reaction time. The
results were analyzed using Statistica® for Windows, release 5.5, produced
by Statsoft. The main effects on selectivity in monocaprin and interactions
between factors were determined. Table 2 shows the regression coefficients,
standard errors, and effects. Most effects were negative, which agrees with
the decrease in monocaprin selectivity observed when the higher level (+ 1)
of each variable (T, R, E) was used (Fig. 1).
To evaluate whether the effects of the variables and their interactions
were statistically significant, a student's t-testwas employed. In Fig. 2, these

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Synthesis of Monocaprin Catalyzed by Lipase 761

Table 2
Statistical Parameters for Design 23
Experimental with Central Point
Factor Coefficient SE Effect
Average 65.495 1.066 65.495
(1) T -4.319 1.192 -8.638
(2) R -6.894 1.192 -13.787
(3) E -3.669 1.192 -7.338
TxR -0.369 1.192 -0.738
TxE -0.581 1.192 -1.162
RxE -2.744 1.192 -5.488
TxRxE -4.519 1.192 -9.038

p=,05

(2)RATIO

'·2"3

(I)TEMP

(3)ENZYME

2by3

I~

I~

·1 0 2 3

Elfed Estimate (Absolute Value)

Fig. 2. Pareto's graphic showing different effects of variables and their interactions
on monocaprin selectivity after 4 h of reaction.

evaluations are illustrated using a Pareto chart. The vertical line indicates
the magnitude of the minimum statistically significant effect for a confi-
dence level of 95%. Values shown in the horizontal columns are student's
t-test values for each effect. It can be observed that the molar ratio of the
reagents is the most significant variable influencing monocaprin selectivity.
Temperature, enzyme concentration, the interaction between Rand E, and
the interaction between the three variables (T, R, E) are also statistically
important. The results showed that all the parameters tested were signifi-
cant and favored monocaprin production at lower levels. This was further
confirmed in the subsequent experiments with varying temperature, molar
ratio of reagents, and enzyme concentration.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
762 da Silva et al.
1,2 .,-_ _ _ _ _ _ _ _ _ _ _ _----::---. 70
60
50
.~ 0,8
u:: 4O;?
:i: 0,6 !?-
30 ~
C 0,4
20
:>
~2 10
o+---+---~--~--~--_+o
40 45 50 55 60
Temperature ~C)

Fig. 3. Effect of temperature on monocaprin molar fraction (%) and on initial veloc-
ity (% MF / min) of monocaprin's production after 6 h of reaction using E = 5% (w /w)
andR=l.

Effect of Temperature
Increased temperature depressed the synthesis of monocaprin, as can
be observed in the reactions performed for 40 and 70°C (Fig. 1) and in the
statistical analysis. The temperature increase enhanced the synthesis of
di- and tricaprin, decreasing monocaprin molar fraction and selectivity.
Therefore, the influence of this parameter was studied at a lower range (40-
60°C), using an enzyme concentration of 5% (w /w) and the stoichiometric
molar ratio of the reagents (Fig. 3).
According to Fig. 3, the best condition for synthesizing monocaprin
was obtained at 55°C (63.25%), decreased at higher temperatures. How-
ever, the initial speed of monocaprin formation (Vi) increased with elevated
reaction temperature in the whole range tested (until 60°C). Thus, in spite
of the higher initial rate of monocaprin formation at 60°C, the final molar
fraction of the monoglyceride was lower at this temperature. This was
probably caused by di- and tricaprin formation during the reaction.
The optimum temperature (55°C) was not consistent with the data of
Kwon et al. (15), who found that R. miehei lipase (Lipozyme IM-20) was not
as active over 40°C on the synthesis of medium-chain glycerides (tri-, di-,
and monocaprin) in isooctane. They reported that during the esterification
of fatty acid and glycerol, water is Ploduced from condensation of the two
groups; this might act as a bridge for the exchange of ionization groups.
Thus, the conformation of lipase may not be so rigid that the structure of
lipase can be destroyed at high temperature. Therefore, considering the
results obtained in our work, it can be concluded that the use of a solvent-
free system in an open-batch reactor at 55°C was effective at eliminating the
water produced in the reaction.
Effect of Enzyme Concentration
As observed in the Pareto chart (Fig. 2), enzyme concentration influ-
ences monocaprin synthesis. Therefore, several experiments with different
Applied Biochemistry and Biotechnology Vol. 705-708,2003
Synthesis of Monocaprin Catalyzed by Lipase 763
80~ ________________________________--,

60

~ 40

20

o
3 5 7 9
Enzyme concentration (% wlw)

Fig. 4. Effect of the concentration of Lipozyme-IM 20 on monocaprin selectivity and


molar fraction at 55°C with R = 1 after 6 h of reaction.

concentrations were proposed, keeping the stoichiometric molar ratio of


the reagents and the temperature of 55°C.
The values obtained for the monocaprin molar fraction were similar
for the different enzyme concentrations tested (Fig. 4) after 6 h of reaction.
The best results were obtained with Lipozyme-1M at 3 and 5% (w / w); thus,
an increased enzyme concentration did not significantly increase the
monocaprin molar fraction. By contrast, a decreased monocaprin selectiv-
ity occurred, probably owing to increased synthesis of di- and tricaprin.
Additional advantages favoring lower enzyme concentrations include re-
duced operating costs and the fact that when 3% (w /w) Lipozyme-IM was
used, no tricaprin was produced.
Effect of Molar Ratio of Reagents
The experiments that used the stoichiometric molar ratio of reagents
resulted in higher molar fractions of monocaprin. The increase in capric
acid concentration (R = 1.5) enhanced the synthesis of di- and tricaprin.
Using an excess of glycerol corresponding to the capric acid/ glycerol molar
ratio of 0.5 did not result in a significant increase in monocaprin selectivity
and molar fraction (Fig. 1). The addition of glycerol in the fed-batch mode
was done considering that glycerol is a polar substance and could, in a high
concentration, inactivate the enzyme (14). As shown in Fig. 5, neither an
increase in the capric acid/ glycerol molar ratio nor addition of glycerol in
the fed-batch mode favored monocaprin synthesis in the reactions per-
formed at 55°C with Lipozyme at 3 and 5% (w /w). Thus, the best molar
ratio of the reagents for the synthesis of monocaprin is the stoichiometric
molar ratio, with all glycerol added at the beginning of the reaction.
Inhibition by the alcohol substrate was also reported by Wong et a1. (16).
They investigated the effect of molar ratio of substrates in the reaction mix-
ture on medium-chain glyceride synthesis from capric acid and glycerol by
lipase from C. rugosa. Excess glycerol in the reaction mixture seemed to in-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
764 da Silva et al.

80 ~-----------------------------------,
60

~40

20

Fig. 5. Effect of addition of glycerol on synthesis of monocaprin after 6 h of reaction


at 55°C. B, batch mode; FB, fed-batch mode.

80

60
t
.,u
c:
0
40 _ 3"10
~
-0-- 3%(with molecular sieves)
co
'0 20
::E - . - 5"10
-t:r-- 5%(with molecular sieves)

3 4 5 6 7
Time (h)

Fig. 6. Effect of addition of molecular sieve on molar fraction in monocaprin for


synthesis performed at 55°C using R = 1 and Lipozyme IM-20 concentrations of 3 and
5% (w/w).

hibit the activity of the enzyme. During this experiment, high volumes of
glycerol caused high solubility of enzyme in glycerol and thus decreased the
lipase concentration at the interface, probably decreasing the reaction rate.
Effect of Addition of Molecular Sieves
Molecular sieves have been used to lower the water activity in the reac-
tion medium, thus shifting the equilibrium toward the synthesis of esters
(17). Thus, molecular sieves were used in our work to evaluate the efficiency
of water removal from the reaction medium in comparison with the free
evaporation method. The results showed that monocaprin production was
not significantly affected by adding molecular sieves to the reaction medium.
Figure 6 presents the reaction progress curves for the two conditions (free
evaporation and addition of molecular sieves). The use of molecular sieves
for water removal, although effective, did not offer a significant advantage
over the simple evaporation technique (open batch stirred reactor at 55°C).

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Synthesis of Monocaprin Catalyzed by Lipase 765

OOr---------------------------------,
60

~ 40

20

o
Rl . E3% Rl . E5% Rl. E3%(s) R1. E5%(s)

Fig. 7. Effect of adding solvent on monocaprin selectivity and molar fraction at 55°C
with R = 1 at Lipozyme IM-20 concentrations of 3 and 5% (w /w) after 6 h. (s) = utili-
zation of 2.0 mL of the mixture of 1:1 n-hexane/t-butanol.

Giacometti et a1. ·(17) investigated the esterification of glycerol with


oleic acid catalyzed by lipozyme in a solvent system (n-hexane) in the pres-
ence of a molecular sieve (3000 mg of the 5-A molecular sieve) at 37°C.
Conversions of oleic acid increased more in the solvent system with the
molecular sieve than in the solvent system without the molecular sieve.
Again, it can be concluded that in the system used in our work (with-
out solvent, open reactor, at 55°C), water was completely evaporated.
However, in a solvent system, at lower temperatures, it is important to use
a desiccant.
Effect of Solvent Addition
Several investigators (18-20) have reported using organic solvents in
enzymatic synthesis of glycerides. Thus, comparison between optimized
reaction conditions with and without solvent was also evaluated. The
nature of the solvent influences the activity and stability of the enzyme to
a large extent. The polarity of the solvent plays a key role. Log P has often
been used to characterize solvents. Log P is defined as the logarithm of the
partition coefficient of a substrate in the standard l-octanol-water two-
phase system. Normally, solvents with high Log P values (Log P > 4)
(hydrophobic solvents) cause less inactivation of biocatalysts than more
hydrophilic solvents (19). The lipases have higher stability in a hydropho-
bic organic medium, such as hexane (Log P = 3.5). However, glycerol, the
reagent of monocaprin synthesis, is a very hydrophilic solvent and has
lower solubility in the nonpolar phase of this system. Thus, to obtain a
much more polar medium that permits a better glycerol solubilization, we
investigated adding t-butanol to a medium containing n-hexane.
To verify whether a mixture of 1: 1 n-hexane / t-butanol would increase
the selectivity in monocaprin, some experiments were performed using
increasing volumes of the solvent mixture. Figure 7 shows the reactions
with and without solvent, presenting final values of monocaprin selectivity
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
766 da Silva et al.
100
-+- capric aCI ___ monocapnn
80 ---t- dicaprin ___ tricaprin
C 60
c
0
:g
£ 40
~
"0
::2 20

0
0 2 3 4 5 6 7
Time (h)

Fig. 8. Time course for monocapin synthesis carries out at 55°C, with a capric
acid/ glycerol molar ratio of 1 and 3% (w /w) Lipozyme.

and molar fraction. The use of n-hexane/t-butanol (at a concentration of


24% [v / v] of the total reaction volume) as solvent mixture did not increase
the monocaprin selectivity. The best results, in terms of molar fraction
(61.3%) and selectivity (71.1%) in monocaprin were obtained with a 3%
(w /w) enzyme concentration without solvent. In addition, the use of
higher amounts of solvent (equal to 38 and 48% of the reaction volume) did
not result in any glyceride synthesis (results not shown). This effect can be
explained by the fact that the increase in the quantity of polar solvents, such
as t-butanol (Log P = 0.32), may deactivate lipase, because they strip off
essential water from the enzyme surface, leading to an insufficiently
hydrated enzyme molecule, which decreases enzyme activity (19). Arcos
and Otero (20) also found a less costly and more selective process in
the absence of solvents to produce mono- and dioleoylglycerols by direct
esterification of glycerol and oleic acid.
Time Course on Best Condition
Considering the results presented, the best conditions for synthesiz-
ing monocaprin were 55°C, an enzyme concentration of 3% (w /w), and
the use of the stoichiometric molar ratio of reagents. Figure 8 shows the
composition of the nonpolar phase for monocaprin synthesis under these
conditions.

Conclusion
The factorial design identified the most important parameters deter-
mining monocaprin selectivity under the tested conditions. The three
parameters studied (temperature, molar ratio of reagents, enzyme con-
centration) have statistical significance and the synthesis of monocaprin
was favored at lower values of these parameters. Using molecular sieves
for water removal, although effective, was not better than the simple
evaporation technique.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Synthesis of Monocaprin Catalyzed by Lipase 767
The results obtained in the present work indicate that monocaprin pro-
duction was feasible in a solvent-free medium. This reaction system avoids
the problems of separation, toxicity, and flammability of organic solvents,
lowering the cost of the final product and allowing product recovery without
further purification steps. The final composition obtained after 6 h of reaction
at 55°C with an enzyme concentration of 3% (w /w) and capric acid/ glycerol
molar ratio of 1 was 61.3% monocaprin, 19.9% dicaprin, and 18.8% capric
acid. This composition agrees with the directives of the World Health Orga-
nization (20) for use as food emulsifiers, which requires that these mixtures
have at least 70% mono- plus diglyceride, a minimum of 30% mono-
acylglycerol, and quantities of glycerol and triglycerides below 10%. The
development of monoglyceride production by lipases can be more competi-
tive than the chemical process, because less energy is consumed; simple
reaction media are used; and higher productivity, higher selectivity, and
higher-quality products are obtained in the bioprocess.

Acknowledgments
This project was financed in part by Funda~ao Carlos Chagas Filho de
Amparo aPesquisa do Estado do Rio de Janeiro (FAPERJ) (projects E-26/
170.731/2000 and E-26/150.356/2002), Funda~ao Universitaria Jose
Bonifacio (FUJB), and Coordena~ao de Aperfei~oamento de Pessoal de
Nivel Superior (CAPES).

References
1. Gao, X. G., Cao, S. G., and Zhang, K. C. (2000), Enzyme Microb. Technol. 27, 74-82.
2. Fevrier, P., Guegan, P., Yvergnaux, F., Callegari, J. P., Dufosse, L., and Binet, A.
(2001), J. Mol. Catal. B: Enzymatic 11, 445-453.
3. Otero, c., Arcos, J. A., Berrendero, M. A., and Torres, C. (2001), J. Mol. Catal. B:
Enzymatic 11, 883-892.
4. Ahmed, F. U. (1990), J. Am. Oil Chem. Soc. 67(1),8-15.
5. Kristmundsd6ttir, T., Arnad6ttir, S. G., Bergsson, G., and Thomar, H. (1999), J. Pharm.
Sci. 88(10), 1011-1015.
6. Bergsson,G., Steingrisson, 0., and Thomar, H. (1990), Antimicrob. Agric. Chem. 43(11),
2790-2792.
7. Isaacs, C. E., Litov, R. E., and Thomar, H. (1995), J. Nutr. Biochem. 6(7),362-366.
8. Petschow, B. W., Batema, R. P., Talbott, R. D., and Ford, L.L. (1998), J. Med.-Microbiol.
47(5), 383-389.
9. Berger, M. and Scheneider, M. P. (1992), J. Am. Oil Chem. Soc. 69,961-965.
10. B6rjesson, I. E. and HarrOd, M. (1999), J. Am. Oil Chem. Soc. 76(6),701-707.
11. Bornscheuer, U. T.(1995), Enzyme Microb. Technol. 17,578-586.
12. Langone, M. A. P. and Sant'anna, G. L. Jr. (1999), Appl. Biochem. Biotechnol. 77-79,
759-770.
13. Montgomery, D. C. (1997), Design and Analysis of Experiments, 4th ed., John Wiley &
Sons, New York, NY.
14. Langone, M. A. P., Abreu, M. E., Rezende, M. J. c., and Sant'Anna, G. L. Jr. (2002),
Appl. Biochem. Biotechnol. 98-100,987-996.
15. Kwon, D. Y., Song, H. N., and Yoon, S. H. (1996),J. Am. Oil Chem. Soc. 73(11), 1521-1525.
16. Wong, W. c., Basri, M., Razak, C. N. A., and Salleh, A. B. (2000), J. Am. Oil Chem. Soc.
77(1), 85-88.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


768 da Silva et al.
17. Giacometti, J., Giacometti, F., Cedomilla, M., and Racki, D. V. (2001), J. Mol. Catal. B:
Enzymatic 11, 921-928.
18. Bellot, J. c., Choisnard, L., Castillo, E., and Marty, A. (2000), Enzyme Microb. Technol.
28,362-369.
19. Koskinen, A. M. P. and Klibanov, A. M. (1996), Enzymatic Reactions in Organic Media,
1st ed., Blackie Academic & Professional, Bishopbriggs, Glasgow, UK.
20. Arcos, J. A. and Otero, C. (1996), J. Am. Oil Chem. Soc. 74(6),673-682.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0769/$20.00

Effect of Dose of Xylanase on Bleachability


of Sugarcane Bagasse Ethanol/Water Pulps

DENISE S. RUZENE AND ADILSON R. GON(:ALVES*

Departamento de Biotecnologia, FAENQUIL, CP 116,


CEP 12.600-970 Lorena, SP, Brazil,
E-mail: adilson@debiq.faenquil.br

Abstract
Pulps obtained from the ethanol/ water cooking of sugarcane bagasse were
bleached with the xylanase enzyme obtained from the fungus Thermomyces
lanuginosus IOC-4145 and with the commercial enzyme Cartazyme HS from
Sandoz. By changing the enzyme dose from 4.3 to 36 IU/g of pulp, kappa
number and viscosity were maintained when the xylanase from T.lanuginosus
was used. On the other hand, by using Cartazyme HS, kappa number de-
creased by 17%, reaching 35.5. This pulp was further extracted with NaOH
without a decrease in viscosity (10 cP), and pulp with a kappa number of 13
was obtained. X ylanases had no significant effect on the ethanol/ water pulps.
Index Entries: Organosolv pulping; sugarcane bagasse; ethanol/water
pulp; xylanase bleaching.

Introduction
Because of the increasing environmental need to eliminate the use of
chlorine in pulp-bleaching plants, the development of bleaching processes
that maintain the economic advantages of pulps has been intensively stud-
ied. An alternative method used in the bleaching process is the enzyme
treatment already studied for kraft pulps (1-3).
The use of enzymes (e.g., xylanases) for pulp treatment offers the ben-
efits, of environmental protection and improved pulp quality. Among the
various enzymes of interest to the paper industry, the hemicellulolytic
xylanases have been found to be commercially feasible for pulp bleaching (4).
Xylanase pretreatment facilitates chemical extraction of lignin from
pulp, reducing consumption of bleaching chemicals and the discharge of
toxic compounds into the environment (4). Xylanases catalyze the hydroly-
sis of xylose-xylose bonds within the xylan chain and solubilize only a
fraction of the total xylan present (5).
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 769 Vol. 105-108,2003


770 Ruzene and Gonc;alves
Studies using hemicellulases (xylanases in particular) have been car-
ried out for the prebleaching of nonwoody fibers, such as bamboo and
bagasse (4-8).
In the present work, ethanol/water pulps obtained from sugarcane
bagasse were bleached with the xylanase enzyme obtained from the fungus
Thermomyces lanuginosus IOC-4145, and the results were compared with the
commercial enzyme Cartazyme HS from Sandoz.

Materials and Methods


Pulping
Pulping of depithed sugarcane bagasse with a 1:1 (v Iv) ethanol:water
mixture was carried out in a closed and pressurized vessel at lS5°C for 3 h.
The product was filtered and the obtained pulp dried to determine yield.
Pulp samples were analyzed by kappa number and viscosity following
standard TAPPI methods (9,10).
Bleaching
Samples of bagasse ethanol/ water pulp (1.5 g) were suspended under
agitation in 50 mL of water (3% consistency) at 30°C for 10 min. Cartazyme
HS and xylanase enzyme obtained from T. lanuginosus IOC-4145 were
added at a dose from 4.3 to 90 lUI g of dry pulp. The samples were main-
tained in a shaker at 30°C for 4 h, followed by filtration and washing with
300 mL of distilled water at 30°C for removal of the enzyme. One set of
enzymatic bleached pulps was further submitted to alkaline extraction.
Samples obtained at different bleaching times (3 g of dry pulp) and original
pulp were extracted with 150 mL of 1 mol/L of NaOH at 65°C for 1 h under
magnetic stirring. After filtration, the pulps were washed with 150 mL of
1 mol/L of NaOH at 65°C and further with distilled water at 65°C until
reaching pH 6.0.
Dry pulps (3 g) were suspended in 150 mL of water (2% consistency)
and heated to 70°C. Sodium chlorite (3.9 mL of 40% aqueous solution) and
glacial acetic acid (0.6 mL) were added. The solution was further heated at
70°C for 5 min, and the bleached pulp obtained was exhaustively washed
with water. Pulps were oven-dried at 60°C for 15 min and analyzed with
respect to kappa number and viscosity by standard methods (9,10).
Hydrolysis of Pulp
One gram of dry pulp was treated with 10 mL of 72% H 2S04 with
stirring at 45°C for 7 min. The reaction was interrupted by adding 50 mL of
distilled water, and the mixture was then transferred to a 500-mL Erlen-
meyer flask, and the volume reached 275 mL. The flask was autoclaved for
30 min at 1.05 bar and 120°C for complete hydrolysis of oligomers (11). The
mixture was filtered and the volume of the hydrolysate was made up to 500
mL with distilled water. The sample (40 mL) of the hydrolysate was diluted
to 50 mL and the pH was adjusted to 2.0 with 2 mol/L NaOH. After filtration
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Bleachability of Ethanol/Water Pulps 771

in a Sep-Pak C18 cartridge to remove aromatic compounds, the hydrolysate


was analyzed in an Aminex HPX-87H column (300 x 7.8 mm) (Bio-Rad) at
45°C using a Shimadzu chromatograph and refraction-index detector. The
mobile phase was 0.005 mol/L H 2S04 at 0.6 mLI min. Sugar concentrations
reported as xylan and glucan were determined using calibration curves of
pure compounds.
Results and Discussion
Ethanol/ water pulps of sugarcane bagasse were bleached by xylanase
of two sources: T. lanuginosus and Cartazyme HS. Pulps were treated with
4.3-90 IV I g of xylanase at 30°C.
Values of the viscosity and kappa number of the ethanol/ water pulps
obtained in the xylanase treatment (X) and in the sequence xylanase-alka-
line extraction (XE) are given in Table 1. The kappa number of the ethanol/
water original pulp (42.8) was greater than that of Acetosolv pulps pre-
pared for a previous study (8) (kappa number of the original pulp = 28),
showing that the de lignification with ethanol/water was not as efficient.
The results of the viscosity and kappa number of the pulps bleached
with the enzyme were compared with the control and the bleached pulp
with sodium chlorite. It was found that the results of enzymatic bleached
pulps were poor, since there was no effect of xylanase treatment.
By using the endoxylanase from T. lanuginosus alone a slight decrease
in kappa number of 8% occurred (42.8-39.3). However, when the second
alkaline extraction step was added, the decrease in kappa number was 67%
(42.8-14.1). There was no increase in the viscosity with the enzymatic
bleaching followed by the alkaline extraction in comparison with the pro-
cedure with extraction alone. The cumulative effect of the alkaline extrac-
tion with the enzymatic treatment was not observed as it was found in the
bleaching of Acetosolv pulps (8).
The concentration of the endoxylanase obtained from T. lanuginosus
did not improve the efficiency of the treatment; however, for Cartazyme
HS the concentration of 36 U I g presented the best results in terms of kappa
number reduction (17%, 42.8-35.5). Viscosity ranged from 10 to 12 cP in all
experiments. Apparently, kappa number 13 was the limit reached in the
alkaline extraction or xylanase sequence of ethanol/water pulps. The re-
sults in Table 1 indicate that xylanase treatment had a small effect on
kappa number reduction and the xylanase-alkaline extraction sequence
had no effect when compared to alkaline extraction alone.
Yields for the pulps obtained after bleaching with the enzyme obtained
from T.lanuginosus IOC-4145 and with Cartazyme HS were about 88-92%
and for the bleaching with sodium hydroxide and sodium chlorite were
about 75%.
The content of xylan was preserved after the treatment with endo-
xylanase from T. lanuginosus. Only after alkaline extraction was a reduc-
tion in xylan detected, which means that xylanase acts over xylans (seerefs.
4 and 5), but the fragments formed did not seem to be easily released.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


).
:g
~
0..
OJ
0' Table 1
n
=>-
rt) Viscosity, kappa Number, Yield, and Chemical Composition of Different Treated Ethanol/Water Pulps'
3
ii;'
~ Endoxylanase from T. lanuginosus Cartazyme
I>l
:J
0.. Total Xylan/ Total Xylan/
OJ
0' Bleaching Viscositykappa Yield lignin Glucan Xylan glucan Acetyl Viscositykappa Yield lignin Glucan Xylan glucan Acetyl
til' sequence b (cP) no. (%) (%) (%) (%) ratio (%) (cP) no. (%) (%) (%) (%) ratio (%)
n
=>-
:J
0
C 10.8 5.6 74.6 5.9 73.8 11.6 0.15 0 10.8 5.6 74.6 5.9 73.8 11.6 0.15 0
~
'<: E 11.9 14.1 76.6 4.5 81.9 5.9 0.07 0 11.9 14.1 76.6 4.5 81.9 5.9 0.07 0
4.3 IU/g
X 10.1 39.3 90.0 9.7 72.0 10.8 0.15 2.6 9.9 36.4 88.4 7.0 49.1 7.4 0.15 0
XE 12.1 14.0 77.0 5.2 82.3 6.5 0.08 0 11.2 13.7 78.3 3.7 67.7 5.2 0.08 0
18 IU/g
X 10.2 39.3 92.0 9.8 71.7 10.7 0.15 2.6 10.1 38.5 91.6 8.8 55.2 8.1 0.15 0
XE 12.0 14.6 77.6 5.3 81.6 6.3 0.08 0 11.3 13.8 78.7 3.7 68.2 5.0 0.07 0
36 IU/g
X 10.1 39.6 90.0 11.3 66.9 10.7 0.15 4.2 9.8 35.5 87.6 11.5 68.7 10.5 0.15 5.9
XE 11.5 13.1 75.1 5.4 79.4 6.5 0.08 3.9 9.9 13.2 75.2 6.0 71.1 5.9 0.08 0
90 IU/g
X 10.1 39.8 87.7 11.3 69.4 10.7 0.15 3.9 10.5 38.1 90.0 10.9 68.8 10.5 0.15 4.7
XE 9.6 14.0 76.1 5.0 79.0 6.5 0.08 3.7 11.1 12.9 80.0 5.0 80.1 6.3 0.08 4.3
•Control: viscosity = 11.1 cP; kappa no. =36.9. Original pulp: viscosity =9.8 cP; kappa no. =42.8; 54.4% yield; 9.9% total lignin; 69.8% glucan; 10.7%
c xylan; xylan/glucan ratio = 0.15; 2.5% acetyl.
bC, chlorite bleaching; E, alkaline extraction; X, xylanase; XE, xylanase followed by alkaline extraction.
-'i"~
c
-
,CX>
N
C
C
W
B/eachability af Ethana//Water Pulps 773
Alkaline extraction makes solubilization of lignin and both xylans and
xylan fragments feasible. Application of Cartazyme HS at 4.3-18 IV / g
degraded 24-31 % of the xylan.
Little differences in the amount of xylans were detected when the
results of the original pulp were compared with those of the bleached
pulp with the xylanase obtained from T. lanuginosus. However, for
Cartazyme HS, higher xylanase preservation was obtained with a greater
concentration of the enzyme (36 and 90 V / g). Comparison of the results
of enzyme bleaching with those of chlorite bleaching revealed that xylans
were also preserved.
Comparison of the original and enzyme-treated pulps revealed that
the xylan/ glucan ratio was maintained. Xylan/ glucan ratio decreased only
after the treatment with enzyme followed by the alkaline extraction. The
decrease in the xylan/ glucan ratio is owing mainly to the alkaline extrac-
tion, and probably the main function of the enzyme is modification of
xylan structure.
The acetyl groups were preserved even after the use of endoxylanase.
After alkaline extraction, acetyl groups were totally removed. On the other
hand, Cartazyme HS had a greater bleaching action. High enzyme doses
seem to promote modification on xylan, maintaining acetyl groups that are
not removed after alkaline extraction. With sodium chlorite, the xylan/
glucan ratio was also preserved and the acetyl groups were removed.

Conclusion
The kappa number, viscosity, and hydrolysis values obtained were
close for the two enzymes. For the ethanol/ water pulps, a cumulative effect
of the alkaline extraction with the enzymatic treatment was not observed.
A kappa number of 13 was the limit reached with the ethanol/water pulp
after alkaline extraction with or without xylanase treatment. In future work
to elucidate the action of xylanase on ethanol! water pulps, the parameters
will be better evaluated at lower severity.

Acknowledgments
We acknowledge financial support from Funda~ao de Amparo a
Pesquisa do Estado de Sao Paulo and ConselhoNacional de Desenvolvimento
Cientifico e Tecnol6gico.

References
1. Valchev, I., Yotova, L., and Valcheva, E. (1998), Bioresour. Technol. 65, p. 57-60.
2. Farrel, R. L., Viikari, L., and Senior, D. (1996), in Pulp Bleaching: Principles and Practice,
Dence, C. W. and Reeve, D. W., eds., Tappi, Atlanta, GA, pp. 365-375.
3. Ni, Y. and Heiningen, A.R.P.V. (1998), TAPPI J. 81(4), 141-144.
4. Prasad, D. Y., Rajesh, K. 5., Praburaj, T., Mohan Rao, N. R., and Joyce, T. W. (1996),
Cellulose Chem. Technol. 30, 463-472.
5. Bajpai, P. and Bajpai, P. K. (1996), TAPPI J. 79(4),225-230.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


774 Ruzene and Gom;alves
6. Kulkarni, N. and Rao, M. (1996), J. Biotechnol. 51,167-173.
7. Shah, A. K., Sidid, S. S., Ahmad, A, and Rele, M. V. (1999), Bioresour. Technol. 68,
133-140.
8. Gon<;alves, A R. and Ruzene, D.S. (2001), Appl. Biochem. Biotechnol. 91-93,63-70.
9. TAPPI (1982), TAPPI Standard Methods, T 230 om-82, Technical Association of Pulp
and Paper Industry, TAPPI, Norcross, GA
10. TAPPI, (1985), TAPPI Standard Methods, T 236 cm-85, Technical Association of Pulp
and Paper Industry, TAPPI, Norcross, GA
11. Rocha, G. J. M., Silva, F. T., Curvelo, A A S., and Araujo, G. T. (1997), in Proceedings
of the 5th Brazilian Symposium on the Chemistry of Lignins and Other Wood Components,
vol. 6, Ramos, 1. P. and Mathias, A 1., eds., Curitiba, Brazil, pp. 3-8.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0775/$20.00

CelF of Orpinomyces PC-2 Has an Intron


and Encodes a Cellulase (CeIF)
Containing a Carbohydrate-Binding Module

HUIZHONG CHEN, l XIN-LIANG LI, *,2 DAVID L. BLUM, 3


EDUARDO A. XIMENES,4 AND LARS G. LJUNGDAHL 1
lCenter for Biological Resource Recovery,
Department of Biochemistry and Molecular Biology, Life Science Building,
The University of Georgia, Athens, GA 30602-7229;
2Fermentation Biotechnology Research Unit, NCAUR, USDN, ARS,
1815 N. University Street, Peoria, IL 61604, E-mail: lix@ncaur.usda.gov;
3Diversa Corporation, 4955 Directors Place,
San Diego, CA 92121-1609; and
4Laboratorio de Enzimlogia, Departamento de Biologia Celular,
Universidade de Brasilia, Asa Norte, Brasilia-OF, Brazil 70910-900

Abstract
A eDNA, designated celF, encoding a cellulase (CeIF) was isolated from the
anaerobic fungus Orpinomyces PC-2. The open reading frame contains regions
coding for a signal peptide, a carbohydrate-binding module (CBM), a linker,
and a catalytic domain. The catalytic domain was homologous to those of
CelA and CelC of the same fungus and to that of the Neocallimastix patriciarum
CELA, but CelF lacks a docking domain, characteristic for enzymes
of cellulosomes. It was also homologous to the cellobiohydrolase lIs
and endoglucanases of aerobic organisms. The gene has a Ill-bp intron,
located within the CBM-coding region. Some biochemical properties of the
purified recombinant enzyme are described.
Index Entries: Cellulase; anaerobic fungi; Orpinomyces; cellulose-binding
module; carbohydrate-binding module.

tNames are necessary to report factually on available data; however, the USDA neither
guarantees nor warrants the standard of the product, and the use of the name by USDA
implies no approval of the product to the exclusion of others that may also be suitable.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 775 Vol. 105-108,2003


776 Chen et al.

Introduction
Anaerobic fungi are part of the natural microflora of the alimentary tract
of many herbivorous animals. Since the discovery of the first anaerobic fun-
gus, Neocallimastix frontalis, by Orpin in 1975 from the rumen of sheep (1), at
least 17 different species of anaerobic fungi have been isolated from ruminant
and nonruminant herbivores. Anaerobic fungi produce highly active hydro-
lytic enzymes (2), such as endoglucanases, xylanases, lichenase, and esterases.
The fungi physically associate with the lignocellulosic tissue of plant frag-
ments. The hydrolytic enzymes exist as high molecular mass complexes
and as individual molecules (3,4). Several genes coding for the hydrolytic
enzymes have been cloned and sequenced from the monocentric fungi
N. patriciarum (5-8), N. frontalis (9), and Piromyces sp. (10,11) and from the
polycentric fungi Orpinomyces PC-2 (12-15) and Orpinomyces joyonii (16).
Analyses of the primary structures of endoglucanases, xylanases, and
lichenase from the anaerobic fungi have revealed that there are substantial
sequence similarities between them from the rumen anaerobic fungi and from
bacteria of the rumen, which suggests that these genes may have bacterial
origin (5,7,8,12-14). However, recently several family 6 cellulase genes have
been isolated from N. patriciarum (17) and Orpinomyces PC-2 (15), and their
amino acid sequences are homologous to enzymes of aerobic fungi. An eno-
lase gene from N. frontalis containing an intron (18) and a cyclophilin gene
from Orpinomyces PC-2 that is heavily interrupted by introns (3,19) have been
reported. So far no intronhas been found in any of the sequenced genes coding
for plant cell wall-degrading enzymes of the anaerobic fungi. Here we report
the cloning and sequencing of a cDNA (ceIF) encoding a family 6 cellulase
(CeIF) from Orpinomyces PC-2 and characterization of the cellulase obtained
by expression in Escherichia coli. Analysis of genomic DNA of the gene celF
revealed that there is an intron, indicating that celF has a fungal origin. This
appears as the first report for an intron-containing gene coding for an enzyme
from an anaerobic fungus involved in plant cell wall degradation.

Materials and Methods


Strains, Vectors, Construction, and Screening
of an Orpinomyces cDNA Library
Cultivation of the anaerobic fungus Orpinomyces PC-2, extraction and
purification of mRNA, cDNA library construction, screening for cellulase
clones, in vivo excision, and sequence determination and analysis have
been described previously (12,14). Cellulase-producing plaques were iden-
tified by incorporating 0.2% (w Iv) remazol brilliant blue-carboxymethyl-
cellulose (CMC) into top agar (15).
Enzyme Purification and N- Terminal Sequencing
A single colony of E. coli XL-l Blue harboring pCEL8 grown on Luria-
Bertani (LB)-ampicillin medium was inoculated into a flask containing

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Orpinomyces Celf 777
500 mL of LB-ampicillin (50 !-lg/mL) broth. The culture was shaken
(280 rpm) at 37°C and grown to an OD600 of approx 1.0. Isopropyl-1-thio-~­
D-galactopyranoside (1 mM) was added, and the culture was incubated for
another 4 h. Cells were harvested by centrifugation (5000g, 10 min), washed
with 50 mL of buffer containing 25 mM Tris-HCl (pH 7.0), and resuspended
in 30 mL of the same buffer. The cells were then disrupted by sonication, and
cell debris was removed by centrifugation. The recombinant enzyme was
purified using Q Sepharose, Mono S HR 10/10, and Superdex 75 10/30
columns on a fast performance liquid chromatography (FPLC) system.
Purified CelF was subjected to N-terminal amino acid sequencing using an
Applied Biosystems model 477A gas-phase sequencer equipped with an
automatic on-line phenylthiohydantoin analyzer.
Enzyme Characterization
All enzyme assays were carried out in duplicate in 50 mM sodium
citrate buffer at pH 6.0 and 40°C unless otherwise stated. Carboxy-
methylcellulase activity was assayed by mixing a O.2-mL aliquot of
appropriately diluted enzyme with 0.4 mL of buffer containing 1% (w Iv)
CMC (low viscosity; Sigma, St. Louis, MO). The reaction time was 30 min
and was terminated by the addition 1.2 mL of 31 mM dinitrosalicylic acid
reagent (20). Reducing sugars were measured by reading the absorbance
at 550 nm. Glucose was used as the standard. Hydrolysis of other polysac-
charides was tested similar to that for CMe. Activity measurements of
enzyme preparations toward p-nitrophenol (pNP)-linked substrates were
performed in 0.3 mL of buffer containing 2 mM substrates. Reactions
were terminated after 10 min by the addition of 0.8 mL of 1 M Na 2C03 •
The release of pNP was measured by reading the absorbance at 405 nm.
pNP was used as the standard. One unit of enzyme activity was defined
as the amount of enzyme required to release 1 !-lmol of glucose equiva-
lent/min. Protein was quantified by the Bradford (21) method using bo-
vine serum albumin as the standard.
The effect of pH on the activity and the pH optimum of the cellulase
were determined at 40°C using the following buffers: 0.1 M sodium acetate
(pH 4.2-5.4), 0.1 M sodium phosphate (pH 5.8-7.8), and 0.1 M Tris-HCl (pH
8.2). The effect of temperature on the cellulase was determined by assaying
the enzyme at temperatures from 20 to 65°e. Thermostability was measured
by incubating enzyme samples in 50 mM sodium citrate buffer, pH 6.0, for
up to 24 h at temperatures from 40 to 65°e. The enzyme solution was chilled
in an ice bath for 5 min and then assayed using the standard assay at 40°e.
In all these assays, CMC was used as the substrate. For determination of I\n
and V max values, enzyme assays were performed at pH 6.0 and 40°C with
CMC as the substrate at concentrations from 1.0 to 25 mg/ mL. I\n and V max
values were obtained from Lineweaver-Burk plots. Zymogram analysis was
done with lichenin (0.2% [w / v]) as the substrate according to Chen et al. (12).
Sugars released from cellooligosaccharides and polysaccharides were
analyzed with a Hewlett-Packard 1100 series high-performance liquid
Applied Biochemistry and Biotechnology Vol. 105-108,2003
778 Chen et al.
chromatograph equipped with an autoinjector and a 1047A refractive
index detector using a Bio-Rad Aminex HPX-42A carbohydrate column.
Water was used as the mobile phase at a flow rate of 0.6 mL/min, and the
column temperature was set at 80°e. Glucose, cellobiose, cellotriose,
cellotetraose, and cellopentaose were used as the standards.
For assays of cellulose-binding capability, enzyme samples of 50 I-tg
of protein were mixed with 30 mg of Avice I in a final volume of 1 mL of
buffer (50mM sodium citrate buffer, pH6.0) at4°e. After 1 h withcontinu-
ous shaking, the Avicel was removed by centrifugation (15,000g at 4°C).
Residual cellulase in the supernatant was assayed as already described.

Analysis of Genomic DNA


Oligonucleotides 5'ATGAAAATTTTACTTTTTGCCAG3' and 5'TTA-
GAATCCTGGTCTAGCATTTC3' corresponding to opposite strands of the
end regions of celF (GenBank accession no. U97154) were used as primers
with genomic DNA (12) and the cDNA libraries as templates for poly-
merase chain reactions (PCRs). Amplification was for 30 cycles with each
cycle consisting of 90 s of melting at 95°C, 60 s of annealing at 55°C, and 90
s of extension at 72°e. Amplified DNA was separated on an agarose gel,
and DNA bands were visualized by ethidium bromide staining. The PCR
products amplified from the genomic DNA were cloned into PCRlI vector
(Invitrogen) and sequenced as described previously (12,14).

Results and Discussion


Isolation and Sequencing ofOrpinomyces CelF
Screening of the cDNA library constructed in ",ZAPII yielded 19 cel-
lulase-producing plaques when 2 x 105 plaque-forming units were plated.
The positive plaques were further enriched and purified. PCR, restriction
enzyme digestion, and sequencing analyses revealed that 16 of these
plaques represented cDNAs of celA, celB, and celE, which were reported
previously (13-15). The other three plaques represented three different
new cellulase cDNAs (pCEL2, pCEL5, and pCEL8).
Sequencing of the inserts in the plasmids obtained from the plaques
by in vivo excision revealed that pCEL8 contained a 1520-bp cDNA (ceIF)
with a complete open reading frame (ORF) encoding a polypeptide (CeIF)
of 432 amino acids with a calculated mass of 46,736 Daltons. The detailed
sequence data are deposited to GenBank and can be accessed as U97154.
The putative translation start codon (ATG) for celF was assigned based on
the fact that there were stop codons in all three frames preceding the ORF;
there was no ATG codon upstream of the proposed ORF. After the ORF,
a 3' untranslated AT-rich end of 127 bp was observed, but no typical long
poly(A) stretch was found. The G+C content of the ORF of celFwas 36.4%
and that of the 5' and 3' noncoding regions was very low (11.8%). High
A+ T contents have also been found in other cDNAs from anaerobic fungi
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Orpinomyces Celf 779

,.
< •

I •
-
14 •

Fig. 1. 50S-PAGE and zymogram of the recombinant CeIF. (A) Coomassie brilliant
blue staining of505-PAG; (B) cellulase zymogram gel. Lane 5, protein molecular mass
standards; lane I, crude E. coli proteins (50 flg); lane 2, the purified CelF (2 flg).

(12-15). The codon usage for celF is similar to that of other Orpinomyces
PC-2 cellulase, xylanase, and lichenase genes.

Purification and N- Terminal Processing


of Recombinant Enzyme from E. coli
Several steps (see Materials and Methods) were required to purify the
recombinant enzyme to homogeneity. Starting from 650 mg of crude pro-
teins, we were able to obtain 2.5 mg of purified CelF with 22-fold purifica-
tion. Specific activity toward CMC of the purified enzyme was 120 U / mg.
Based on the specific activity of the purified cellulase, the recombinant CelF
protein constituted about 4.6% of the total E. coli proteins.
The first 10 N-terminal amino acid residues of the recombinant en-
zyme were AXGGAYAQXG. Except the two unidentified residues, the
sequence matched the region corresponding to amino acid residues 21-30
of the deduced protein sequence, suggesting that the enzyme is processed
in E. coli with a 20 amino acid signal sequence being removed to give a
mature active enzyme of 412 amino acid residues.

Enzyme Characterization
The purified recombinant CelF appeared as a single band on sodium
dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 1).
Zymogram analysis showed that the apparent molecular mass of CelF
produced in E. coli was approx 45.5 kDa (Fig. 1), which appears to be
consistent with the deduced molecular mass of the mature CelF lacking
the proposed signal peptide of 20 amino acid residues. No other activity
band with lower molecular mass was detected in the crude cell extracts,

Applied Biochemistry and Biotechnology Vol. 105-10B, 2003


780 Chen et al.
indicating that the Asn-rich linker region was relatively stable. This is
in contrast to several other anaerobic fungal hydrolytic enzymes contain-
ing the noncatalytic repeated peptide domains where the linker regions
between the catalytic domains and the noncatalytic repeated peptide
domains are susceptible to truncation (7,13,14).
The activity of the cellulase was determined from pH 4.2 to 8.6 using
CMC as the substrate. A typical pH profile was obtained with an opti-
mum between pH 5.8 and 6.2, which was similar to values of most other
hydrolytic enzymes from the anaerobic fungi (12,15) and was close to the
physiologic pH in the rumen. The effect of temperature on the cellulase
activity was determined in 50 mM sodium citrate at pH 6.0 from 20 to
65°C. Maximum activity was observed between 40 and 50°C, and the
activity decreased rapidly above 55°C. Thermal stability was investigated
by incubating the enzyme, up to 24 h, at different temperatures. Almost
no activity loss was observed between 40 and 45°C in the above men-
tioned buffer at pH 6.0 for 24 h, and 61 and 12.9% of the enzyme activity
were retained after 24 h of incubation at 50 and 55°C, respectively. Inac-
tivation occurred at 60°C with only 15.5% of the enzyme activity remain-
ing after 1 h of incubation. Km and V max values using CMC as the substrate
at 40°C and pH 6.0 obtained from Lineweaver-Burk plots were 4.37 mg/
mL and 173 U /mg of protein, respectively.
Activities of the purified CelF against various substrates are given in
Table 1. The enzyme rapidly hydrolyzed acid-swollen cellulose, CMC,
barley ~-glucan, and lichenin. The CelF had a low but detectable activity
toward microcrystalline cellulose (Avicel) with a specific activity of 0.039
U/mg of protein. No detectable hydrolysis was observed of pNP-~-D­
glucoside, oat spelt xylan, or pNP-~-D-xyloside.
Hydrolysis products formed during the action of the purified recom-
binant CelF on cellooligosaccharides, Avicel, acid-swollen cellulose, and
CMC were determined by hifh-performance liquid chromatography
(HPLC) (Table 2). Cellobiose was not hydrolyzed, whereas cellotriose was
slowly hydrolyzed to cellobiose and glucose. With cellotetraose as the sub-
strate, cellobiose was the sole product detected. Thus, the second gluco-
sidic linkage was mainly cleaved by the enzyme. Cellopentaose was largely
converted into cellobiose, cellotriose, and some glucose, indicating that
some cellotriose was further hydrolyzed to yield cellobiose and glucose.
The hydrolysis patterns of cellooligosaccharides by CelF are somewhat
similar to those by cellobiohydrolase II (CBHII) from Trichoderma reesei (22).
The end products of hydrolysis of Avicel, acid-swollen cellulose, and CMC
yielded mainly cellobiose, cellotriose, and minor amounts of glucose. It is
apparent that CelF had both endoglucanase and CBH activities, which is
similar to that for CelA and CelC from the same fungus (15).
Domain Structures of CelF and Its Homology with Other Cellulases
Analysis of the CelF sequence revealed that it contains a typical sig-
nal peptide sequence consisting of about 20 amino acid residues (23).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Orpinomyces Celf 781

Table 1
Substrate Specificity of Orpinomyces CelF Produced in E. eolia
Substrateb Specific activity (U /mg protein)
Avicel 0.039
Acid-swollen cellulose 83.1
CMC 120.0
Barley j3-glucan 1001.5
Lichenin 724.6
"Assays were performed at 40°C and pH 6.0 (50 mM sodium citrate)
for 10 min (PNP-~-glycosides), 15 min (oat spelt xylan, barley ~-glucan,
and lichenin), 30 min (CMC), and 4 h (Avicel), respectively.
"The activities with pNP-~-D-glucoside, pNP-~-D-cellobioside, oat spelt
xylan, or pNP-~-D-xyloside as substrate were < 1.0% of that against CMC.

Table 2
HPLC Analysis of Products of Cellooligosaccharide
and Polysaccharide Hydrolysis by Celpa
Products or residual substrates (Ilmol/mL)
Substrate G1 G2 G3 G4
Cellotriose (G3) 0.30 0.20 2.7
Cellotetraose (G4) 5.8
Cellopentaose 0.20 3.3 2.7
Avicel 0.20 0.11
Acid-swollen cellulose 0.037 0.72 0.34
CMC 0.017 0.31 0.12
"Reaction mixtures contained 0.25 VlmL of enzyme (CMC as sub-
strate) and 3 mM cellooligosaccharides, 0.7% (w Iv) CMC, or 0.5% (w Iv)
acid-swollen cellulose in 20 mM sodium citrate. Reactions were at pH 6.0
and 40°C for 4 h. In the case of Avice!, 1.0 U/mL of enzyme (CMC as
substrate) and 1% (w Iv) Avicel were combined and reactions wereatpH
6.0 and 40°C for 16 h.

Immediately after the signal sequence, amino acid residues 22-57 consti-
tute a typical fungal carbohydrate-binding module (CBM) (17,24). The
catalytic domain is located at the C-terminus (amino acid residues 106-
432). It is separated from the CBM by an extremely Asn-rich linker (amino
acid residues 67-105).According to classification of CBMs (24), the CBM
of CelF should be placed in family I, which is exclusive to fungal hydro-
lases. It has a high degree of similarity with the CBM of CBHII from
T. reesei (86% similarity and 53% identity). Six cysteine residues (nos. 22,
29,39,40,46, and 56) forming three disulfide bridges that stabilize the
polypeptide are conserved in CelF (25,26). Three highly conserved aro-
matic residues (Tyr26, Trp52, and Tyr53), which are conspicuous build-
ing blocks of the flat-face binding to the cellulose surface (23), are present
Applied Biochemistry and Biotechnology Vol. 105-108,2003
782 Chen et al.
in the CBM. Moreover, three invariant amino acids (Gln28, Asn50, and
Gln55) in suitable positions for hydrogen bonding with the cellulose sur-
face are also present in the CBM (23). The presence of a CBM in CelF is
consistent with the fact that about 80% of the cellulase activity of the
purified CelF adsorbed onto Avicel.
The deduced amino acid sequence of CelF of Orpinomyces PC-2 when
compared with protein sequences in the SWISS PROT and GP data banks
was homologous to several CBHIIs and endoglucanases belonging to fam-
ily 6 glycosyl hydrolases (27). The highest identity was with CELA of
N. patriciarum (17). The identity with it was 82.9% when the complete
sequences including signal peptide, CBM, linker, and catalytic domain were
compared. One deletion and/ or insertion between these two enzymes was
found in the linker region (after residue 86 in CeIF), whereas five amino
acids (residues 345-349) present at the carboxyl terminus of Orpinomyces
PC-2 CelF were not found in N. patriciarum CELA. The catalytic domain of
CelF of Orpinomyces PC -2 is homologous with the ca talytic domains of CelA
(75.2% similarity, 64.8% identity) and CelC (74.9% similarity, 63.6% iden-
tity) of the same organism (15). It also has substantial similarity with the
CBHIIs of T. reesei (52.6% similarity, 37.2% identity) (26) and Fusarium
oxysporum (54.7% similarity, 38.9% identity) (28), Cellulomonas fimi CENA
(49.2% similarity, 32.0% identity) (29), Streptomyces halstedii CELA1 (51.6%
similarity, 31.2% identity) (3D), and Thermomonospora fusca CELE2 (50.3%
similarity, 26.5% identity) (31).
The three-dimensional structures of the catalytic domains of T. reesei
CBHII (32) and T. fusca CELE2 have been determined (33). Mutagenesis
studies of T. reesei CBHII have shown that Asp245 is the likely proton
donor in the catalytic event and that the neighboring Asp199 is charged,
ensuring the protonation of Asp245. A function of the Tyr193 is to modu-
late the protonation states of the interacting carboxylates of Asp199 and
245 (34). These three amino acid residues are conserved in family 6 gly-
cosyl hydrolases and are present in the catalytic domain of CelF (Asp 181,
Asp223, and TyrI75), CelA, and CelC (15). Family 6 glycosyl hydrolases
contain both CBHIIs and endoglucanases. Endo- or exo-type activity
depends on whether the enzyme contains loops that form a tunnel active
site (32,34) or an open cleft (endoglucanase) (33). Compared with T. reesei
CBHII exoglucanase, there are deletions of two amino acids correspond-
ing to the carboxyl terminal loop (from Ser391 to Gly409) and four amino
acids corresponding to the second loop (from Pro178 to Ala191) for CelF
of Orpinomyces PC-2, which suggests that the loops might only partially
enclose the tunnel of the active site (15). This indicates that the loops of
CeIF, like those of CelA and CeIC, are distinct from those of CBHIIs and
endoglucanases in the same family, which may result in CelF having both
CBHII and endoglucanase activities (15).
Although the catalytic domain of CelF is very similar to those of CelA
and CelC, CelF has a CBM, whereas CelA and CelC contain a dockerin,
which is involved in neither catalysis nor cellulose binding (5,7,15). It has
Applied Biochemistry and Biotechnology Vol. 705-708, 2003
Orpinomyces Celf 783

I( 1 2 )

4.07
'.05 _.
2..04 -
1.U .

1.02 ..

0.51 -

B
GTATGGATATTTTATTTTTATAATTTAGGAAATAAATACTT
TTAAATAATTTAATTTAAGTATTGATTATTTTAAATTATTA
TATTACTTATAACAAATTAAGATATATAG

Fig. 2. Analysis of genomic sequence of celF. (A) Amplification of genomic DNA of


Orpinomyces PC-2 celF coding region by PCR. Reaction solutions (20 !lL) with genomic
DNA (lane 1) and cDNA (lane 2) as the templates and without template (lane 3) were
run on 1.5% agarose gel. DNA molecular standards were used in lane M. After elec-
trophoresis, DNA bands were visualized by ethidium bromide staining. (B) Nucle-
otide sequence of an intron found in Orpinomyces celF.

been suggested that dockerins function as docking domains in a fashion


similar to that of the dockerin domains of catalytic subunits of the
cellulosome of Clostridium thermocellum (4,10,35,36). The lack of a dockerin
in CelF suggests that enzyme is not a part of cellulase/hemicellulase com-
plexes found in Orpinomyces PC-2 (37) and other anaerobic fungi (38). The
presence of cellulases as both free and complex forms that are highly cata-
lytically similar may provide the fungus with an advantageous strategy for
the rapid decomposition of plant cell wall structures.
Coding Region ofCelF Contains an Intron
The ORF region of celF was amplified by PCR, and its size is larger
than that amplified from cDNA (Fig. 2A). Sequencing the DNA amplified
from the genomic DNA template revealed that it produced a 1410-bp
DNA fragment. Alignment of the sequences of genomic and cDNA of celF
revealed an intron of 111 bp (Fig. 2B) located in the CBM-coding region.
Splicing boundaries started with GT and ended with TA, which match the
general consensus sequences found for introns in filamentous fungi (39).
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
784 Chen et al.
In a similar experiment, the ORF regions of celA and celC genomic DNA
were amplified by PCR, and their sizes were the same as those amplified
from their cDNAs (data not shown), indicating that celA and celC are
devoid of introns. Other genes coding for cellulolytic enzymes of anaero-
bic fungi, which have been examined, are devoid of introns (8,12-14,16).
The fact that the ORF of celF contains an intron indicates that it has a
eukaryotic origin. Several other genes isolated from anaerobic fungi are
believed to have prokaryotic origins (3).
Using the software package PHYLIP (40), a phylogenetic tree based on
cellulase sequences was created from neighbor-joining bootstrap analysis.
It showed that CelF, CelA, CelC, and N. patriciarum CELA were clustered
together. It has been suggested that a common ancestral precursor of cel-
lulolytic aerobic fungi and rumen anaerobic fungi may have existed (17).
Our finding of an intron in celF provides solid evidence that the gene has
a eukaroytic origin. Thus, two types of cellulase genes have been discov-
ered in anaerobic fungi. One type includes those coding for endoglucanases
possibly transferred from rumen bacteria (7,13,14,16), and the other is those
of Orpinomyces CelA, CelC, CelF and Neocallimastix CELA, which appear to
have fungal origin (l15,171 this study). Moreover, current observationsseem
to support that celF CBM-coding region was replaced with those coding for
a dockerin-coding sequence during evolution, resulting in the formation of
celA and celC. Of course, the opposite could also be a possibility. In any
event, the high sequence similarity among celA, celC, celF, and Neocallimastix
celA strongly suggests the occurrence of gene transfer and duplication
among the fungi.

Acknowledgments
We thank Dr. Mike Cotta for critical comments and Marsha Ebener for
executive help. This work was supported by grants from the US Depart-
ment of Energy (DE-FG02-93ER20l2) and from the Consortium for Plant
Biotechnology Research, Georgia Research Alliance, and Aureozyme, Inc.

References
1. Orpin, G. C. (1975), J. Gen. Microbiol. 91,249-262.
2. Borneman, W. S., Akin, D. E., and Ljungdahl, L. G. (1989), Appl. Environ. Microbiol. 55,
1066-1073.
3. Ljungdahl, L. G., Li, X.-L., and Chen, H. (1998), in Thermophiles-The Keys to Molecular
Evolution and the Origin of Life, Wiegel, J. and Adams, M., eds., Taylor & Francis,
London, UK, pp. 187-197.
4. Steenbakkers, P. J. M., Li, X.-L., Ximenes, E. A., Arts, J. G., Chen, H., Ljungdahl, L. G.,
and Op Den Camp, H. J. M. (2001), J. Bacteriol. 183,5325-5333.
5. Black, G. W., Hazlewood, G. P., Xue, G.-P., Orpin, C. G., and Gilbert, H. J. (1994),
Biochem. J. 299, 381-387.
6. Dalrymple, B. P., Cybinski, H. D., Layton, I., McSweeney, C. S., Xue, G.-P., Swadling,
S. J., and Lowry, J. B. (1997), Microbiology 143, 2605-2614.
7. Gilbert, H. J., Hazlewood, G. P., Laurie, J. I., Orpin, C. G., and Xue, G.-P. (1992), Mol.
Microbiol. 6, 2065-2072.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Orpinomyces Celf 785

8. Zhou, L, Xue, G.-P., arpin, C G., Black, G. W., Gilbert, H. J., and Hazlewood, G. P.
(1994), Biochem. J. 297,359-364.
9. Durand, R., Rascle, C, and Fevre, M. (1996), Curro Genet. 30,531-540.
10. Fanutti, C, Ponyi, T., Black, G. W., Hazlewood, G. P., and Gilbert, H. J. (1995), J. Bioi.
Chem. 270,29,314-29,322.
11. Millward-Sadler, S. J., Hall, J., Black, G. W., Hazlewood, G. P., and Gilbert, H. J.
(1996), FEMS Microbiol. Lett. 141, 183-188.
12. Chen, H., Li, X.-L., and Ljungdahl, L. G. (1997), J. Bacteriol. 179,6028-6034.
13. Chen, H., Li, X.-L., Blum, D. L., and Ljungdahl, L. G. (1998), FEMS Microbiol. Lett. 159,
63-68.
14. Li, X.-L., Chen, H., and Ljungdahl, L. G. (1997), Appl. Environ. Microbiol. 63,628-635.
15. Li, x.-L., Chen, H., and Ljungdahl, L. G. (1997), Appl. Environ. Microbiol. 63,4721-4728.
16. Liu, J.-H., Selinger, L. B., Hu, Y.-J., Moloney, M. M., Cheng, K.-J., and Beauchemin,
K. A (1997), Can. J. Microbiol. 43,477-485.
17. Denman, S., Xue, G.-P., and Patel, B. (1996), Appl. Environ. Microbiol. 62,1889-1896.
18. Durand, R., Fischer, M., Rascle, C, and Fevre, M. (1995), Microbiology 141, 1301-1308.
19. Chen, H., Li, X.-L., and Ljungdahl, L. G. (1995),Proc. Natl. Acad. Sci. USA 92,2587-259l.
20. Miller, G. L. (1959), Anal. Chem. 31,426-428.
21. Bradford, M. M. (1976), Anal. Biochem. 72,248-254.
22. Harjunpaa, V., Teleman, A., Koivula, A, Ruohonen, L., Teeri, T. T., Teleman, a., and
Drakenberg, T. (1996), Eur. J. Biochem. 240, 584-59l.
23. Von Heijne, G. (1986), Nucleic Acids Res. 14,4683-4690.
24. Tomme, P., Warren, R. A. J., Miller, R. C Jr., Kilburn, D. J., and Gilkes, N. R. (1995),
in Enzymatic Degradation of Insoluble Carbohydrates, Saddler, J. N. and Penner, M. H.,
eds., ACS Symposium Series 618, American Chemical Society, Washington, DC
25. Hoffren, A-M., Teeri, T. T, and Teleman, O. (1995), Protein Eng. 8,443-450.
26. Teeri, T. T., Lehtovaara, P., Kauppinen, S., Salovuori, 1., and Knowles, J. (1987), Gene
51,43-52.
27. Henrissat, B. and Bairoch, A (1993), Biochem. J. 293,781-788.
28. Sheppard, P. a., Grant, F. J., Oort, P. J., Sprecher, C A., Foster, D. C, Hagen, F. S.,
Upshall, A, McKnight, G. L., and O'Hara, P. J. (1994), Gene 150, 163-167.
29. Wong, W. K., Gerhard, D. C, Guo, Z. M., Kilburn, D. G., Warren, R. A J., and Miller,
R. C Jr. (1986), Gene 44, 315-324.
30. Fernandez-Abalos, J. M., Sanchez, P., ColI, P. M., Villanueva, J. R., Perez, P., and
Santamaria, R. 1. (1992), J. Bacteriol. 174,6368-6376.
31. Lao, G., Ghangas, G. S., Jung, E. D., and Wilson, D. B. (1991), J. Bacteriol.173,3397-3407.
32. Rouvinen,I., Bergfors, T., Teeri, T. T.,Knowles,I. K. C,andJones, T. A. (1990), Science
249,380-386.
33. Spezio, M., Wilson, D. B., and Karplus, P. A (1993), Biochemistry 32, 9906-9916.
34. Koivula, A, Reinkainen, T., Rouhonen, L., Valkeajarvi, A, Claeyssens, M., Teleman,
a., Kleywegt, G. J., Szardenings, M., Rouvinen, J., Jones, T. A, and Teeri, T. T. (1996),
Protein Eng. 9, 691-699.
35. Choi, S.-K. and Ljungdahl, L. G. (1996), Biochemistry 35, 4906-4910.
36. Pages, S., Belaich, A., Belaich, J. P., Morag, E., Lamed, R., Shoham, Y., and Bayer, E.
A (1997), Proteins 29, 517-527.
37. Li, X.-L., Chen, H., He, H., Blum, D. L., and Ljungdahl, L. G. (1997), in Abstracts of the
97th General Meeting of the American Society for Microbiology, American Society for
Microbiology, Washington, DC, p. 424.
38. Dijkerman,R., Vervuren,M. B.F.,Op DenCamp,H.J.M.,and vander Drift,C (1996),
Appl. Environ. Microbiol. 62, 20-25.
39. Gurr, S. J., Unkles, S. E., and Kinghorn, J. R. (1987), in Gene Structure in Eukaryotic
Microbes, Kinghorn, J. R., ed., IRL, Oxford, England, UK, pp. 93-139.
40. Felsenstein, J. (1989), Cladistics 5, 164-166.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0787/$20.00

Partition Behavior and Partial Purification


of Hexokinase in Aqueous Two-Phase
Polyethylene Glycol/Citrate Systems

GEORGE G. G. OLIVEIRA, l DANIEL P. SILVA, l


INES CONCEI<:AO ROBERTO/ MICHELE VITOLO, l
AND ADALBERTO PESSOA JR. *,1

1Department of Biochemical and Pharmaceutical Technology,


Faculty of Pharmacy, University of Sao Paulo,
Av. Prof. Lineu Prestes, 580, B.16. 05508-900, Sao Paulo, SP, Brazil,
E-mail: pessoajr@usp.br; and 2DEBIQ-FAENQUIC
Rod. Itanjuba - Lorena, Km 74.5, 12,600-000Lorena, SP, Brazil

Abstract
This study dealt with the partition behavior and partial purification of
hexokinase (HK) from baker's yeast by liquid-liquid extraction using aque-
ous two-phase polyethylene glycol (PEG)/citrate systems. First, we investi-
gated the effect of agitation type (vortex and 8 rpm rotation) on the stability
of the system, and then the effects of sodium citrate concentration, PEG con-
centration, and molar mass of PEG on the partition coefficient of this enzyme
by using a 25 factorial experimental design. The results of this factorial experi-
ment showed the possibility of a partial purification of HK by using two
extraction steps, since the enzyme preferentially migrated to the top phase
and the total proteins (mainly contaminants) remained in the bottom phase.
The purification factor (PurTOP ) of the enzyme in the top phase was 1.87, and
the partition coefficient of the total proteins (K prot ) was 0.47.
Index Entries: Hexokinase; liquid-liquid extraction; aqueous two-phase
systems; Saccharomyces cerevisiae.

Introduction
Hexokinase (HK) (Ee 2.7.1.1) is the first enzyme of glycolysis to cata-
lyze the phosphorylation of glucose into glucose 6-phosphate (G6P). G6P,
in turn, is a key intermediate for several pathways such as g]uconeogen-
esis, shunt of pentoses, and glycogen metabolism. The same enzyme is
sensitive to the catabolic repression and plays an important role in the
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 787 Vol. 105-108,2003


788 Oliveira et al.
glucose uptake mechanism through the plasma membrane (1). Further-
more, HK is used in analytical methods to measure glucose, fructose,
manose, adenosine triphosphate (ATP), and creatine-kinase activity (2), in
the phosphorylation of pyranose and furanose analogs of glucose (3), and
in wine and fruit juice industries for the detection of illegal sugar addition
to the final products (4).
Although the HK enzyme can be found in almost every animal tissue
and in several microbial species, the use of Saccharomyces cerevisiae as an
enzymatic source and other products (5) makes sense in Brazil because the
large experience of the Brazilian chemical industry on handling this strain.
In addition, coupling the yeast processing with ethanol production should
probably have a positive effect on the profits of the distilleries. However,
production of HK can only be viable if large quantities of this enzyme are
available at competitive prices. The development of techniques for the sepa-
ration and purification of enzymes has been an indispensable prerequisite
for many of the advances made in the biotechnological industry (6,7).
According to Diamond and Hsu (8), 50-90% of the production costs
of biologic products are determined by the purification strategy. Purifi-
cation is troublesome because of the system complexity and the need to
retain biologic activity. Aqueous two-phase systems (ATPS) are liquid-
liquid extraction processes widely used for the extraction and purifica-
tion of many biologic products. Their technical simplicity and ease of
operation and scale-up make them very attractive for industrial applica-
tions (9, 10). Moreover, aqueous two-phase systems such as polyethylene
glycol (PEG) / citrate, PEG / phosphate and PEG / sulfate are adequate for
continuous large-scale purification of materials of biological origin and
allow the use of traditional liquid-liquid extraction equipment (11-13).
The systems are formed by the mixture of aqueous solutions of two
polymers or a polymer and an electrolyte solution. With the concentration
of each component of the system in one of the phases (top or bottom), there
is partition of biomolecules such as proteins, cells, cell particles or nucleic
acids. To achieve a significant extraction, the target product has to be pref-
erentially partitioned in favor of one phase, whereas the interfering sub-
stances (other biomolecules) should be partitioned to the other phase. It is
important to remember that higher water content in both phases avoids
protein denaturation. Differences in enzyme partitioning can be ascribed to
the interaction of the factors or mechanisms inherent in the system itself
(such as choice of system components, polymer molecular weight, concen-
tration of polymers and salts, ionic strength, and pH values) with those of
the target protein (such as hydrophobicity, charge, and molecular weight)
(14). Factors and mechanisms that cause the uneven distribution of proteins
between the two-phases are little understood, but empirical rules have been
developed. In practice, the technique requires further experimentation to
find an adequate system for each particular application, since many factors
influence the partition and purification of proteins. For this reason, the aim
of the present work was to verify the influence of some variables (PEG molar

Applied Biochemistry and Biotechnology Vol. 705-108,2003


Partition of Hexokinase by A TPS 789

mass, PEG concentration, and citrate concentration) on the partial purifica-


tion of HK of S. cerevisiae in aqueous PEG/ citrate systems.

Materials and Methods


Chemicals
Glucose-6-phosphate dehydrogenase (G6PDH), ATP, nicotinamide
adenine dinucleotide, nicotinamide adenine dinucleotide phosphate
(NADP) and G6P, utilized in the enzymatic analysis, were obtained from
Sigma (St. Louis, MO). PEG with a molar mass of 300,400, 1500, and 4000
g/mol was purchased from Labsynth (Sao Paulo, Brazil) and of 600 and
1000 g/mol from Merck (Darmstadt, FRG). Sodium citrate was purchased
from Grupo Qufmica Industrial (Sao Paulo, Brazil). All other chemicals
were of analytical grade.
Cell Homogenate
The solution containing disrupted cells (denominated cell homoge-
nate) was prepared through disruption of commercial baker's yeast, by
submission to a cell disruption in a mechanical grinder (bead mill) with
glass beads (diameter =0.5 mm). The cell suspension (wet cell cake and Tris-
HCl buffer) and glass beads were mixed in the volumetric proportion of
1: 1 (below 10°C). After disruption, cell debris and glass beads were removed
by filtration. Before each extraction in aqueous two-phrase system, the cell
homogenate, stored at -20°C, was thawed and centrifuged (2880g /10 min).
The supernatant was collected with HK and other proteins.
Binodal Curves and Equilibrium of PEG/Citrate System
The binodal curves were built by the titulation method, as described
by Albertsson (15), using stock solutions made of 50% (w / w) PEG and 30%
(w /w) citrate. Equilibrium studies of the PEG/citrate systems were con-
ducted for the extractions containing cell homogenate. Binodal curves are
present on the phase diagram and divide a region of component concentra-
tions that will form two immiscible aqueous phases (i.e., above the curve)
from those that will form one phase (Le., at and below the curve).
Homogenization of System
The extraction conditions were the same as those employed in the
study of the variation in pH, except only PEGs of 400,600, and 1000 g/mol
were used. The influence of agitation (vortex for 1 min, 8 rpm for 20 min or
vortex/rotation) was evaluated in the centrifugal-phase volumes.
Liquid-Liquid Extraction
Aqueous two-phase systems were prepared with different PEGs (300,
400,600, 1000, 1500, and 4000 g/mol) and sodium citrate (Na3C6H607·2H20).
In graduated centrifuge tubes (15 mL), 2.0 g of medium containing the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
790 Oliveira et al.
Table 1
Factors and Levels Used for Optimizing the Enzyme
Partition Coefficient by Aqueous Two-Phase System

Level of variables

Factor High (+) Central (0) Low (-)

MMPEG (gfmol) 1500 1000 400


PEG (% [wfw]) 24 22 20
Citrate (% [w fw]) 20 17.5 15

target enzyme (cell homogenate or pure enzyme; Sigma, St. Louis, MO) was
mixed with PEG (stock solution) and sodium citrate (solid). Deionized
water was then added to the mixture in order to adjust the final concentra-
tion desired for the components (system of 10 g). Next, the mixture was
agitated in vortex (1 min), rotated (8 rpm/20 min), and centrifuged (1500g /
10 min) to separate the phases from each other. Samples of the phases (top
and bottom) were removed and analyzed to verify the enzymatic activity,
total protein concentration, pH, and volume. During all the experiments,
the temperature was maintained at 25°C.
The factors and levels used for the HK extraction are provided in
Table 1 and were statistically analyzed by means of the program
5tatgraphics Plus 6.0 (Statsoft) according to the method of experimental
design proposed by Box et al. (16). After all the experiments, which con-
sisted of a 23 factorial experimental design with three repetitions in the
central point (totaling 11 extractions), the effect of molar mass of PEG
(MMPEG, g/ mol), PEG concentration (% [w /w]), and citrate concentration
(% [w / w]) were used to optimize the partition coefficient of the enzyme. For
each of the three factors, high (coded value: +1), central (coded value: 0), and
low (coded value: -1) set points were selected (Table 1).
Analytical Methods
HK activity was determined by spectrophotometric analysis (340 nm)
of reduced NADP+ at30°C, according to the method described by Bergmeyer
(2). One unit of HK was defined as the amount of enzyme that catalyzes the
reduction of 1 mmol ofNADP+ / min in the conditions of the experiment. The
concentration of total proteins was determined according to Bradford (17),
using patterns of bovine serum albumin (Sigma). The partition coefficient of
the enzyme (K) was calculated as the ratio between the HK enzymatic activi-
ties of the top and bottom phases (Eq. 1), and the partition coefficient of the
proteins (Kprot ) was calculated as the ratio between the total protein concen-
tration of the top and bottom phases (Eq. 2). Selectivity (5) was calculated as
the ratio between the partition coefficient of HK and the partition coefficient
of the total proteins (Eq. 3). The increase in purity in the top phase (PurTOP )
was obtained from the relationship between the specific activity of the
enzyme (U/mgprot) before extraction and after extraction (Eq. 4). The yield

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Partition of Hexokinase by ATPS 791
50~---------------------------------------------.

PEG 400

40

10

O+-------~------r_------r_----_,,_--~~======~
o 5 10 15 20 25 30
Citrate concentration (% w/w)

Fig. 1. Binodal curves obtained in PEG 400, 600,1000,1500, and 4000, and citrate (pH
8.5, 25°C) systems.

of extraction (%R) was obtained from the relationship between the total
proteins (mg) or total activity of the enzyme (U) in the phases obtained after
and before the extraction, using the volume of each phase:
HK TOP
K=-- (1)
HK BOT

ProtTop
K Prot --=---
- Prot (2)
BOT

s=~ (3)
K pmt

P ur
TOp = U TOp/mg Prot (4)
U initialmg Prot

Results and Discussion


Binodal Curves: PEG/Citrate System
The curves of the PEG/citrate system (Fig. 1) were obtained with the
intention of establishing extraction conditions (%PEG and %citrate). These
curves present a significant variation as a function of the molar mass of PEG
(MMPEG). It was observed that there was no formation of a two-phase sys-
tem with PEG 300. The binodal curves built with PEG 400, 600, 1000, 1500,
and 4000 g/mol were close to those presented by Vemau and Kula (18).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


792 Oliveira et al.
Table 2
Influence of Homogenization Type on Volume of Interface
(Occupied by Cellular Fragments) as Function of MMPEG

Type of homogenization

Vortex Rotation Vortex (1 min) +


1 min 8 rpm/20 min rotation (8 rpm/20 min)
MMPEG Volume of interface (mL)
400 1.1 0.3
600 1.7 1.0
1000 2.0 1.2
"Total volume of system was 8.5 mL.
bCitrate salt was not totally dissolved.

Studies of Equilibrium
Extractions in PEG / citrate systems with cell homogenate presented
cellular fragments occupying a large space between the top and bottom
phases of aqueous two-phase systems. This volume, according to
Genevieve et al. (19), can be called interface volume or be considered as
a third phase of the system, harming not only the transfer of proteins
between the phases (20), but also the total solubility of the citrate salt in
the system. For this reason, studies of equilibrium of the PEG/citrate
system were carried out in order to define extraction conditions precisely
with small or no interface volume. The results (Table 2) showed that the
association between vortex and rotation (vortex for 1 min + 8 rpm/20
min) makes it possible to decrease the interface volume as well as the total
dissolution of the citrate salt.
Experiments were conducted with another enzyme (G6PDH) to verify
the partition coefficient as a function of different times of rest of the PEG/
citrate system after its homogenization and centrifugation. The results
showed that only after 6 h of rest was the PEG / citrate system stable, which
is in agreement with the findings of Jain and Johri (21) in relation to the use
of different times of rest after the homogenization of the aqueous two-
phrase system.
Liquid-Liquid Extraction in PEG/Citrate System: HK
After obtaining the binodal curves and evaluating the equilibrium of
the PEG / citrate system, we conducted experiments to verify the effects of
citrate concentration (%citrate), PEG concentration (%PEG), and MMPEG
on the partition coefficient (K) of pure HK enzyme (Sigma). The statistical
analysis of the results employing t-tests and variance analysis showed
that MMPEG and %citrate were significant variables in the extraction
process, with 95% confidence. A negative effect of MMPEG and a positive
effect of %citrate on the response variable (K) were observed. This means

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Partition of Hexokinase by A TPS 793

1AOO
1~OO

1000
aOO
1- 600

Fig. 2. Response surface obtained in experimental design (2 3 factorial + central


point) with variables MMPEG and %citrate, and coefficient of partition (K) of HK as
response (22% [w/w] PEG, pH 8.5, and 2S°C).

Table 3
Partition Coefficient (K) of HK as Function
of MM PEG, %PEG, and %citrate, at pH 8.5 and 25°C

Experiment MMPEG %PEG %Citrate Kb

1 1500 24 20.0 1.46


2 400 24 20.0 1067.80
3 1500 20 20.0 0.36
4 400 20 20.0 1171.40
5 1500 24 15.0 0.03
6 400 24 15.0 NS
7 1500 20 15.0 0
8 400 20 15.0 NS
9 1000 22 17.5 1.87
10 1000 22 17.5 0.91
11 1000 22 17.5 1.78
a Agitation in vortex (1 min) Followed by Rotation (8 rpm/20 min) and
centrifugation (1500g /10 min) with time of rest of 3 h for PEGs 1000 and
1500, and 1 h for PEG 400.
bNS, no phase separation.

that the lowest MMPEG and the highest citrate concentration provided
the highest values of K, as observed in the response surface (Fig. 2). For
the purpose of statistical analysis, the values of the partition coefficients
obtained in experiments 6 and 8 (Table 3), which did not form two-phase
systems, were considered null (K = 0).

Applied Biochemistry and Biotechnology Vol. 105-708, 2003


794 Oliveira et al.
Table 4
Partition Coefficient (K)of HK as Function
of Growing Concentrations of citrate at pH 8.5 and 25°Ca
Experiment MMPEG %PEG %Citrate
1 400 22.0 15 NS
2 400 22.0 17.5 245
3 400 22.0 20 1000
Q Agitation in vortex (1 min) followed by rotation (8 rpm/20 min) and
centrifugation (1500g /10 min) with time of rest of 1.0 h after centrifugation.
b NS, no phase separation.

Owing to these results and to the absence of two-phase systems, when


MMPEG <400 g/mol was used, new experiments to evaluate the influence
of %citrate on the K value of HK were performed. The conditions were 22%
(w /w) PEG 400 and 15-20% (w /w) %citrate (Table 4). The highest K value
(1000) was attained when cell homogenate, 22% (w /w) PEG 400, and 20%
(w / w) citrate were employed. Experiments using higher values of %citrate
were not possible, because little or no enzyme was detected in the bottom
phase of the system.
Of all the variables tested, the low molar mass of PEG associated with
a high concentration of citrate salt mostly favored the transfer of the HK
enzyme to the top phase of the system. The high molar mass of this enzyme
indicates that the volume exclusion effect by the PEGs of higher molar mass
might have occurred. This means that in the top phase of the system there
was no space available for the enzyme when PEGs with a molar mass >400
glmol were used. Since it was not possible to obtain a two-phase system
with PEGs of lower molar mass, it was concluded that it would not be
possible to improve the partition coefficient as a function of the MMPEG.
However, in the case of the citrate, the transfer of the enzyme to the top
phase of the system can be attributed to the effect of "salting out," because
under high concentrations of this salt the enzyme is expelled from the
bottom phase. The combination of these two variables favored the extrac-
tion of the enzyme, since the enzyme excluded from the bottom phase was
transferred to the top phase.

Liquid-Liquid Extraction for ATPS PEG/Citrate


Cell Homogenate
The conditions that furnished the best results of pure enzyme extrac-
tion, 22% (w /w) PEG 400 and 20% (w Iw) citrate, were also used for extrac-
tions with HK of cell homogenate. However, the literature reports that
when cell homogenate is used, the substances that can interfere with the
characteristics of the system increase in number (22,23). This explains not
only the nonformation of the two-phase system during the preparation of
the systems with 22% (w /w) PEG 400, 20% (w /w) citrate, and cell homo-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Partition of Hexokinase by ATPS 795
25,-------------------------------------------,

'""'
~ 20
~
~
Q
0
15
'p
~
!:l
t::
CI) 10
()
t::
0
()

c:J 5
III
~

0
8 10 12 14 16 18 20
Citrate concentration (% w/w)

Fig. 3. Binodal curve built with PEG 1500 and citrate salt (pH 8.5 and 25°C). The
points represented by the triangles were used for the initial extractions of the system
PEG 1500/citrate in two stages.

genate, but also the formation of such a system when pure enzyme was
used (Table 4, experiment 3).
Experiments aiming the reduction of the system viscosity were per-
formed with a higher %citrate (significant variable on the value of K) and
a lower PEG concentration, until a close point to the binodal curve, because,
according to the results of the experimental design, this variable (%PEG)
does not interfere with the response of the system. These extraction condi-
tions and the systems in which there were separations of the phases pro-
vided a high value of enzymatic partition coefficient. However, the values
of protein partition coefficient were high, and the purification factor was
low. From this result it is possible to conclude that the system (12% [w / w]
PEG 400 and 25% [w /w] citrate, pH 8.5, and 25°C) prepared with cell
homogenate and following the homogenization procedures can be used to
prepurify the enzyme. However, to improve the purification factor of the
enzyme, new experiments were conducted. A second extraction stage was
performed and the top phase obtained in the first stage was used as the
initial medium and source of HK.
Second Stage of Extraction
The top phase obtained after the first extraction was added to the
second stage (PEG IS00/Citrate). This new stage of HK extraction was
performed in different concentrations of PEG and citrate, and these con-
centrations were defined as a function of the binodal (Fig. 3). The extrac-
tions were performed with the lowest concentrations of PEG and citrate,
but above the binodal curve. The results of the experiments are presented
in Table 5.
The results showed that to reach higher enzyme purity (PUYTOP ) and
partition coefficients (K), additional experiments should be performed.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


796 Oliveira et al.
Table 5
Extraction of HK in Aqueous Two-Phase System
Using Cell Homogenate in PEG 400/Citrate (Experiment 1)
and PEG 1500/Citrate (Experiments 2-9) at pH 8.5 and 25°C
%PEG %citrate Interface volume
Experiment (w/w) (w/w) K K prot PurTOP 5 (mL)
1 12 25 1000 54.15 1.25 0.00 0.0
2 17 15 1.01 1.07 1.57 0.94 0.3
3 10 20 0.66 1.80 0.83 0.37 0.5
4 10 15 0.13 0.34 0.39 0.38 0.4
5 17 20 19.47 7.53 1.32 2.58 1.5
6 13.5 17.5 0.51 0.47 1.87 1.09 0.4
7 13.5 20.0 1.76 2.12 1.28 0.83 1.0
8 13.5 22.5 11.07 2.59 0.82 4.28 1.5
9 13.5 25.0 79.07 16.38 1.02 4.83 1.7
a Agitation in vortex (1 min) followed by rotation (8 rpm/20 min) and centrifugation
(1500g/10 min).

These new extractions were performed with higher concentrations of cit-


rate (20,22.5, and 25 % [w fwD. However, the concentration of PEG should
be constant (13.5% [w /wD, because high concentrations of this polymer
can saturate the system, and low concentrations decrease the partition
coefficients and purification factors. The results (experiments 7-9) showed
that the augmentation of the citrate concentration (from 20 to 25%)
improved the values of K and PurTOp • On the other hand, the extraction
performed at 25% citrate provided a high interface volume and incomplete
dissolution of the salt.

Conclusion
The extraction of proteins by the aqueous PEG/ citrate system was
effective, with quite simple reagents and stages of process. Its use in the
partial purification of HK in two extraction steps is viable. In the first step,
using 12% (w /w) PEG 400 and 25% (w /w) citrate atpH 8.5 and 25°C, high
enzymatic recovery was obtained in the top phase of the system with high
values of partition coefficient. In the second step of extraction, performed
with 13.5% (w /w) PEG 1500 and 17.5% (w /w) citrate, a good purification
factor was obtained in the top phase (PurTOp = 1.87). This indicates a ten-
dency of the HK to stay in this phase in comparison with the undesirable
proteins (K prot = 0.47), which show a higher migration to the bottom phase
of the system. The exclusion effect, caused by the PEG with high molar
mass, did not allow the transference of the enzyme to the top phase. Fur-
thermore, the effect of salting out can also explain the transference of the
enzyme to the top phase of the system, because under high concentrations
of citrate the enzyme is expelled from the bottom phase. The results also

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Partition of Hexokinase by ATPS 797
showed that the response surface methodology is adequate to optimize the
enzyme partition coefficient by the aqueous two-phase system.

Acknowledgments
We thank Funda<;ao de Amparo a Pesquisa do Estado de Sao Paulo
(FAPESP) and Conselho Nacional de Desenvolvimento Cientifico e
Tecnol6gico (CNPq) for financial support, and Maria Eunice M. Coelho for
assistance in writing the manuscript.

References
1. Kovak, L.,Nelson, B. D., and Ernster, L. (1986), Biochem. Biophys. Res. Commum.134(1),
285-291.
2. Bergmeyer, H. (1984), Methods of Enzymatic Analysis, 3nd ed., Verlag Chimie,
Weinheim, Germany.
3. Chenault, H. K., Mandes, R F., and Hornberger, K. R (1997), J. Org. Chern. 62,331-336.
4. Whitaker, J. R (1991), in Food Enzymology, vol. 2, Fox, P. F., ed., Elsevier, New York,
NY.
5. Godfrey, T. and West, S. (1996) Industrial Enzymology, MacMillan, London,
England, UK.
6. Rodrigues, E. M. G., Pessoa-Jr., A., and Milagres, A. M. F. (1999), Appl. Biochem.
Biotechnol. 78, 779-788.
7. Rodrigues, E.M.G., Milagres, A.M.F., and Pessoa-Jr., A. (1999), Process Biochem. 34,
121-125.
8. Diamond, A. D. and Hsu, J. T. (1992), Adv. Biochem. Eng. Biotechnol. 47,89-135.
9. Costa, S. A., Pessoa Jr., A., and Roberto, 1. C. (1998), Appl. Biochem. Biotechnol. 70/72,
629-639.
10. Gaikar, G. V., Bodhankar, S. S., and Latha, V. J. (1996), J. Chern. Technol. Biotechnol. 67,
329-332.
11. Coimbra,J. S. R, Thommes,J.,Meirelles,A. J. A., and Kula,M. R (1995), Bioseparation
5,259-268.
12. Coimbra, J. S. R, Thommes, J., and Kula, M. R (1994), J. Chromatogr. 668,85-94.
13. Coimbra, J. S. R, Mojola, F., and Meirelles, A. J. A. (1998), J. Chern. Eng. Japan 31,
277-285.
14. Schmidt, A. S., Venton, A. M., and Asenjo, J. A. (1994), Enzyme Microb. Technol. 16,
131-142.
15. Albertsson, P. A. (1986), Partition of Cell Particles and Macromolecules, John Wiley,
New York, NY.
16. Box,G.P.,Hunter, W. G., and Hunter,J. S. (1978)Statisticsfor Experimenters,John Wiley,
New York, NY.
U. Bradford, M. A. (1976), Anal. Biochem. 72, 248-254.
18. Vernau, J. and Kula, M. R (1990), Biotechnol. Appl. Biochem. 12,397-404.
19. Genevieve,M.F., Walker,S. G., and Lyddiatt,A. (2000),J. Chromatogr. B743(1),409-419.
20. Cascone, 0., Andrews, B. A., and Asenjo, J. A. (1991), Enzyme Microb. Techno!. 13,
629-635.
21. Jain, A and Johri, B. N. (1999), Bioresour. Technol. 67,205-207.
22. Mom~, D. J., Mom~, D. M. (2000), J. Chromatography B 743,369-376.
23. Rito-Palomares, M. and Cueto, L. (2000), J. Chromatogr. B 743, 5-12.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0799/$20.00

Effect of Aeration on Lignin Peroxidase


Production by Streptomyces viridosporus T7A

LEDA M. F. GOTTSCHALK, l RONALDO NOBREGA, l


AND ELBA P. S. BON*,2
1Programa de Engenharia Qufmica, CapPE; and

2Departamento de Bioqufmica, IQ Universidade Federal do Rio de Janeiro,


RJ, Brazil, 21949-900, E-mail: elbaI996@iq.ufrj.br

Abstract
The effect of aeration on lignin peroxidase production by Streptomyces
viridosporus T7A was studied in a bench-scale bioreactor using a previously
optimized growth medium (0.65% yeast extract and 0.1 % corn oil, pH 7.0) at
37°C and natural pH. Airflow rates of 0.3,1.0, and 1.5 vvm and a fixed agita-
tion of 200 rpm were initially studied followed by 1.0 vvm and 200, 300, 400,
and 500 rpm. The use of 1.0 vvm and 400 rpm increased enzyme concentration
1.8-fold (100-180 U IL) and process productivity 4.8-fold (1.4-6.7 U I [L·h]) in
comparison with the use of 200 rpm and 0.3 vvm. The inexpensive corn oil,
used as carbon source, besides its antifoam properties, proved to be
nonrepressive for enzyme production.
Index Entries: Streptomyces viridosporus; lignin peroxidase production;
aeration; productivity; corn oil.

Introduction
Since World War II, human activity has introduced a great variety of
xenobiotic chemicals into the environment on a large scale. Every year some
1000 new chemicals are introduced on the market, many of them displaying
a rather poor biodegradability (1). Lignin peroxidase (LiP), which in nature
is responsible for lignin degradation in the presence of hydrogen peroxide
(2), also presents the ability to oxidize, besides lignin, a larger number of
aromatic substances, including highly polluting and recalcitrant com-
pounds, such as azo dye and pesticides (3,4). Although such properties
make LiP potentially useful in environmental pollution control (5), this
application in an industrial scale requires its production at low cost.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 799 Vol. 105-108,2003


800 Gottschalk et al.
Since the first characterization of liP by Ramachandra et al. (6), sev-
eral aspects of liP production by Streptomyces viridosporus T7A have been
studied. Yeast extract proved to be the best nitrogen source (6-8). Concern-
ing carbon sources, glucose; lactose; galactose; corn oil; and some sources
considered to be inducers, such as lignocellulose, cellulose, and xylans,
have been tested to optimize liP production. The use of glucose allowed
the highest liP activity in spite of showing a repressive effect during cell
growth (7-10). Corn oil, besides being a low-cost carbon source, favored
enzyme production in higher yields, improved enzyme stability during
fermentation, and prevented the formation of foam, a particularly inter-
esting feature considering large-scale fermentations (11). Calcium carbon-
ate increased the enzyme levels and the industrially relevant parameter
productivity (12,13). The positive effect of this salt might be related to
the enzyme molecule stabilization by the calcium ion and I or to a higher
enzyme release from the Streptomyces cells that displays a more filamen-
tous mycelium in comparison to the traditional pellet morphology (12).
The study of airflow rates is important in submerged bioreactors
because the microorganism grows completely immersed in the culture
medium without direct contact with the gas phase, which induces oxygen
mass transfer limitation both inside and outside the pellets (14). Thus, the
production of enzymes by aerobic filamentous microorganisms, such as
S. viridosporus, in submerged fermentations depends strongly on the
bioreactor airflow rates and agitation conditions, since oxygen, which has
a low solubility, is often a limiting nutrient (15). A better mix in the culture
medium, making oxygen and also nutrients more accessible for the cells,
would improve enzyme synthesis (16).
The present work studied the effect of airflow and agitation rates on
liP production by S. viridosporus in a bioreactor with a 3-L working volume.
MATERIALS AN D METHODS
Microorganism
S. viridosporus T7A (ATCC 39115) sporulating cultures were obtained
by growing the microorganism at 37°C for 6-8 d on an agar medium con-
taining malt extract (3 giL), yeast extract (3 giL), peptone (5 giL), glucose
(10 giL), and agar-agar (20 giL). Spores were suspended in a 20% (w Iv)
glycerol aqueous solution and maintained at -20°C (17).
Fermentations
liP was produced in batch fermentations using an optimized growth
medium (6.5 giL of yeast extract, 1 giL of corn oil,S giL of calcium
carbonate, 0.20 giL of MgS04·7H20; 0.20g/L of NaCI; 0.05 giL of CaCI2,
and a trace metal stock solution, all at pH 7.0 [18,191) ina bioreactormodel
Applikon, 3-L working volume, coupled to a biocontroller ADI 1030.
Inoculum was prepared in 500-mL shake flasks containing 100 mL of the
same medium that was inoculated with the spore suspension to a final

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Effect of Aeration on LiP Production 801
absorbance at 570 nm of 1.0. Flasks were incubated in a Shaker Tecnal BTC
9090 at 200 rpm and 37°C for 16 h. Three flasks (total volume of 300 mL)
were used to inoculate the bioreactor (10% [v/v]). Fermentations were
carried out at 37°C for approx 80 h. Airflow rates of 0.3,1.0 and 1.5 vvm
(1 L of air I[L of substrate·min]) and agitation of 200 rpm were initially
studied. Subsequently, an airflow rate of 1.0 vvm and agitation rates of
200,300,400, and 500 rpm were also evaluated. Samples, taken at differ-
ent time intervals, were centrifuged at 4000 rpm for 10 min. The cell cake
was used for cell growth evaluation, and the culture supernatant was
used for measurement of pH and LiP activity.
Cell Growth
Cell growth was estimated through the cell cake protein content
after pretreatment with 1 N NaOH (7) and expressed as grams of cell
protein per liter of culture. Protein concentration was estimated by the
modified Folin-phenol method and microassay procedure using a stan-
dard curve for bovine serum albumin (2-40 Ilg) (20).
LiP Activity
LiP was determined by the enzymatic oxidation of 2,4-dichlorophenol
(2,4-DCP) in the presence ofH20 2 and 4-aminoantipyrene (4-AAP) (21). The
1.0-mL reaction mixture contained 164 !J.M 4-AAP (Sigma, St. Louis, MO),
3 mM 2,4-DCP (Sigma), 4 mM hydrogen peroxide, and 200 ilL of the culture
supernatant in a 50 mM potassium phosphate solution (pH 7.0). The reac-
tion was initiated by the addition of hydrogen peroxide and the increase in
absorbance at 510 nm was monitored for 1 min at room temperature. One
unit of liP activity corresponded to the increase of one unit of absorbance
per minute under initial reaction rates. Extracellular LiP concentration was
expressed as units of enzyme per liter of culture. Normalized LiP activity
and productivity were expressed as units of enzyme per gram of cell protein
and units of enzyme per liter of culture per hour, respectively.
Results and Discussion
Effect of Airflow Rates
Figure 1 shows data for cell growth, as measured by protein content;
pH variation; and liP concentration, in fermentations carried out at 200
rpm and different airflow rates. Cell growth was not favored by the use
of 0.3 vvm; the peak biomass cell concentration of 0.229 giL was about
twofold lower in comparison to the findings observed for the use of
1.0 and 1.5 vvm: 0.404 and 0.489 giL, respectively (Fig. lA). Therefore,
this condition, which also presented the lower peak enzymes titers (100
U IL), resulted in oxygen andlor substrate transfer rate limitations.
Improved enzyme titers (peak values of 120 and 170 U IL) were
observed within 48 h of fermentation when airflow rate increased to 1.0 and
1.5 vvm, respectively (Fig. IB). Besides the positive response of cell growth
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
802 Gottschalk et al.

A OJ ~----------------------------T10

~~~~~~~~~~~~~~~:
0.6

~ 0.5
= 6
~...
...as"'-
0.4

0.3
5 :a
4
:§ 0.2 \a.-~~~::-ioOiiiJl 3
Ql
U 2
0.1
1
-=~~~~~~-r~~~~~~~0
20 40 60 80
Time (h)

B 180
160 -+--0.3
_ _ 1.0
140
~...
--1.5
120
,~ 100
;:
'<.I
as 80
~ 60
40
20
0
0 20 40 60 80
Time (h)

Fig. 1. Cellular protein (ptn), pH variation (A) and extracellular LiP activity (B)
profiles of S. viridosporus T7A grown in agitated submerged culture at 37°C, 200 rpm
using different airflow rates.

and enzyme production to the increase in aeration rate, maximum liP activ-
ity was also anticipated in 24 h in both cases. These results corroborate those
found in the literatpre; both lignin mineralization and LiP synthesis were
increased in cultures grown under high oxygen tension (14, 22).
The pH increased in all fermentations, reaching maximal values within
the range of 8.5-9.5 (Fig. lA). Enzyme titers decreased after 48 h of fermen-
tation with the use of 1.0 and 1.5 vvm, probably owing to pH inactivation
since previous work indicated pH 8.0 as a threshold value for enzyme
stability (8).

Effect of Agitation Rate


In filamentous microorganisms, such as S. viridosporus, pellet size and
morphology are greatly affected by the agitation or shear stress. It is also
reported that the average pellet size decreases strongly with the increase in
agitation rate (15).
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Effect of Aeration on LiP Production 803
Table 1
Comparison of Maximum liP Activity, Maximum Biomass Accumulation,
Maximum YLiP/)(I and Productivity in All Fermentation Conditions Tested"
Maximum Maximum
liP biomass Maximum
Fermentation activity accumulation YLiP/ X Productivity
conditions (U/L) (g cell protein/L) (U I g cell protein) (U/[L·h])
0.3 vvm, 200 rpm 100 0.229 518 1.4
1.0 vvm, 200 rpm 120 0.404 1600 2.5
1.0 vvm, 300 rpm 120 0.236 1071 3.3
1.0 VVffi, 400 rpm 180 0.162 3750 6.7
1.0 vvm, 500 rpm 140 0.129 2333 5.4
1.5 vvm, 200 rpm 170 0.489 1034 3.5
"Bold numbers indicate best parameter for each condition.

A previous work from our laboratory also showed that microorgan-


ism pellet size was affected by the presence of calcium carbonate in the
growth medium since the predominance of a more filamentous morphol-
ogy was observed (12). Because the reduction in pellet size might be rel-
evant for the mechanisms of mass transfer, facilitating nutrient uptake and
enzyme release into the culture broth, calcium carbonate was added to our
culture medium. Note also that calcium ions play important structural
roles in peroxidases of different organisms (23).
Although in the first set of experiments the airflow rate of 1.5 vvm
resulted in the highest liP production (170 U IL), the normalized produc-
tion was higher for the aeration of 1.0 vvm, the condition selected to study
the effect of agitation rate (Table 1).
Figure 2A shows that the increase in agitation rate from 200 to 500 rpm
using a fixed 1 vvm airflow rate resulted in a consistent decrease in cell
protein content that represented cell growth. This result was unexpected
because a better mix allowing oxygen and nutrients more accessible for
the cells should be beneficial for S. viridosporus growth. Therefore, it is pos-
sible that our mixing conditions resulted in high shearing effects that were
detrimental to mycelium integrity. Enzyme production, however, was
not likewise affected since the same peak concentration of 120 U IL was
observed for 200 and 300 rpm within 48 h of fermentation and a peak
enzyme concentration of 180 U IL was observed within 32 h when 400 rpm
was used, resulting in gain in productivity. Although liP also peaked within
the same time interval for 500 rpm, enzyme levels were lower (140 U/L).
The increase in agitation rate of 200-400 rpm, besides increasing liP produc-
tion, also anticipated maximum LiP activity in 16 h (Fig. 2B). Even though
we did not measure corn oil consumption during the fermentation, it is
conceivable that, since corn oil is an insoluble carbon source, better agitation
conditions favored this carbon substrate mass transfer (24-26). As previ-
ously observed for the airflow rate studies, pH increased in all fermenta-
tions, reaching maximal values of about 9.0 (Fig. 2A).
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
804 Gottschalk et al.
A 0.7 10
0.6 9
,-.. _ _ ptn200 8
~ 0.5 7
'-'
--ptn 400 ,
.~ 0.4 6
_plnSOO,
'0
IS.. 03
""\\I .
-+-pH 200 rp
_pH30Orpm
5
4
=
g..

'3 0.2 -pH40Orpm


3
=i 2
u
0.1
1
0 0
0 20 40 60 80
Time (h)

B 200
180
,-.. 160
g 140
'-' 120
f 100
'tCIS 80
~ 60
...:l
40
20
0
0 20 40 60 80
Time (h)

Fig. 2. Cellular protein (ptn), pH variation (A) and extracellular liP activity (B)
profiles of S. viridosporus T7A grown in agitated submerged culture at 37°C, 1.0 vvm
using different agitation rates.

Effect of Different Aeration/Agitation Rates


on Normalized LiP Production and Productivity
In submerged bioreactors, biomass grows with oxygen mass transfer
limitation, suggesting the necessity of designing a suitable bioreactor con-
figuration that provides good oxygen transfer in a low-shear environment
(14). The literature provides descriptions of different ways to increase oxy-
gen availability, such as immobilization of microorganisms in airlift
bioreactors, addition of a water-immiscible organic solvent in the medium,
and/or exposure of the culture medium to hyperbaric oxygen (27-29).
In the present work, different aeration and agitation rates in a sub-
merged bioreactor were evaluated in order to improve LiP production and
productivity. Figure 3 compares normalized LiP production (U / g of cell
protein). and Fig. 4 compares LiP productivity (U /[L·hD in all conditions
tested. The maximum value for normalized LiP activity (3750 U / g of cell
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Effect of Aeration on LiP Production 805

4000 TlC======~---------------'
• 200rpm/O.3vvm
.-. 3500 .200rpm/1.Ovvm
c:
.! 3000 .200rpm/1.5vvm
o
Q. 2500 .300rpm/l.Ovvm
. 400rpm/l.Ovvm
] 2000
2.
CI
1500
~ 1000
::J
> 500
0 +---,--
o 8 24 32 48 56 72 80
Time (h)

Fig. 3. Comparison of normalized LiP production of batch fermentations of S.


viridosporus T7A grown in agitated submerged culture at 37°C, using different airflow
and agitation rates.

7 ~====~--~----------------~
• 200rpm/O.3wm
6 .200rpm/l.0wm
~ 5 • 200rpm/l.Swm
::J .300rpm/1.Owm
;: 4 . 400rpm/1.Owm
:c
~ 3 • 500rpm/1.Owm
::s
'8.... 2
0..
1

o +-----,--
o 8 24 32 48 56 72 80
Time (h)

Fig. 4. Comparison of productivity of batch fermentations of S. viridosporus T7A grown


in agitated submerged culture at 37°C, using different airflow and agitation rates.

protein) was obtained for the aeration I agitation resulting from the use of
400 rpm and 1.0 vvm within 56 h of fermentation (Fig. 3). The maximum
value of liP productivity (6.7 U I [L·h]) was also obtained in the same con-
dition within 24 h of fermentation (Fig. 4).
Table 1 summarizes the maximum liP activity (U IL), maximum
biomass accumulation (gcell protein/L of culture), productivity (U I [L·h])
and maximum normalized liP production (U I g of cell protein) in the six
conditions studied. When the airflow rate increased from 0.3 vvm/200
rpm to 1.0 vvm/400 rpm, a 4.8-fold increase in productivity (from 1.4 to
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
806 Gottschalk et al.
6.7 u I [L·h]) and a 7.2-fold increase in normalized LiP activity (518-3750
U I g) were observed, suggesting an improvement in cell physiology
toward enzyme biosynthesis. Although an airflow rate of 1.5 vvm/200
rpm also favored LiP production (170 U IL), the normalized production
was lower (1034 U I g of cell protein). Thus, the higher cell growth, result-
ing from the better culture medium mix, allowing oxygen and nutrients
more accessible for the microorganism, had detrimental effects on
enzyme production, and by extension, on the normalized LiP production
data (16,29).
The increase in agitation rate from 400 to 500 rpm resulted in a severe
fall in normalized enzyme production (3750 to 2333 U I g of cell protein),
most probably owing to the negative effect of the shearing effects on the
mycelium's integrity. Productivity increased 2.7-fold when the agitation
rate increased from 200 to 400 rpm and decreased (6.7 to 5.4 U I [L·h]) when
it increased from 400 to 500 rpm, confirming that the best condition for LiP
production was 1.0 vvm and 400 rpm.

Conclusion
We studied in a bench-scale bioreactor the effect of aeration and agita-
tion on S. viridosporus LiP production using the nonrepressive, although
immiscible, com oil as carbon source and CaC03 to induce a more filamen-
tous cell morphology and, by extension, to improve nutrients and oxygen
mass transfer. Cell growth, enzyme production, and productivity responded
positively to the increase in aeration rate from 0.3 to 1.0 and 1.5 vvm. The
increase in agitation rate from 200 to 500 rpm using a fixed 1 vvm airflow rate
resulted in a consistent decrease in biomass accumulation; however,
enzyme production was positively affected. It was possible to increase en-
zyme levels 1.8-fold (100-180 U IL), productivity 4.8-fold (1.4-6.7 U I [L·h]),
and normalized enzyme production 7.2-fold (518-3750 U I g of cell protein)
using 1.0 vvm and 400 rpm in comparison with 0.3 vvm and 200 rpm. The
inexpensive com oil, used as carbon source, besides its antifoam properties
also proved to be nonrepressive for enzyme production.

Acknowledgments
This work was supported by the Brazilian Research Agency Conselho
Nacional de Desenvolvimento Cientffico e Tecnol6gico (CNPq) and Rio de
Janeiro Research Foundation (FAPERJ).

References
1. Miserez, K.,Philips, 5., and Verstraete, W. (1999), Water Sci. Technol.40(415), 137-144.
2. Odier, E. and Artaud, I. (1992), in Microbial Degradation of Natural Products,
Winkelmann, G., ed., VCH, Germany, pp. 161-191.
3. Pasti-Grigsby, M. B., Paszczynski,A., Goszczynski,S., Crawford, D. L., and Crawford,
R. L. (1992), Appl. Environ. Microbiol. 58, 3605-3613.
4. Goszczynski,S., Paszczynski, A., Pasti-Grigsby, M. B., Crawford, R. L., and Crawford,
D. L. (1994), J. Bacteriol. 176, 1339-1347.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Effect of Aeration on liP Production 807
5. Gilbert, M., Morosoli, R, Shareck, F., and Kluepfel, D. (1995), CRC Crit. Rev. Biotechnol.
15(1), 13-39.
6. Ramachandra, M., Crawford, D. 1., and Hertel, G. (1988), Appl. Environ. Microbiol. 54,
3057-3063.
7. Pasti, M. B., Pometto ill, A. 1., Nuti, M. P., and Crawford, D. 1. (1990), Appl. Environ.
Microbiol. 56, 2213-2218.
B. Lodha, S. J., Korus, R A., and Crawford, D. 1. (1991), Appl. Biochem. Biotechnol. 28/29,
411-420.
9. Zerbini, J. E., Oliveira, E. M. M., and Bon, E. P. S. (1999), Appl. Biochem. Biotechnol.
77-79,681-688.
10. Macedo, J. M. B., Gottschalk, 1. M. F., and Bon, E. P. S. (1999),Appl. Biochem. Biotechnol.
77-79,735-744.
11. Gottschalk, 1. M. F., Macedo, J. M. B., and Bon, E. P. S. (1999), Appl Biochem Biotechnol.
77-79,771-778.
12. Macedo, J. M. B., Gottschalk, 1. M. F., and Bon, E. P. S. (1999), Braz. J. Chem Eng. 16(2),
163-169.
13. Yee, D. c., Jahng, D., and Wood, T. K (1996), Biotechnol. Prog. 12,40-46.
14. Bosco, F., Ruggeri, B., and Sassi, G. (1996), J. Biotechnol. 52,21-29.
15. Jimenez-Tobon, G. A., Penninckx, M. J., and Lejeune, R (1997), Enzyme Microb.
Technol. 21,537-542.
16. Dominguez, A., Rivela, I., Couto, S. R, and Sanroman, A. (2001), Process Biochem. 37,
549-554.
17. Hopwood, D. A., Bibb, M. B., Chater, K F., Kieser, T., Bruton, C. J., Kieser, H. M.,
et al. (1985), in Genetic Manipulation of Strepto.myces: A Laboratory Manual, The John
Innes Foundation, Norwich, England, UK, pp. 3-5.
lB. McCarthy, A. J. and Broda, P. (1984), J. Gen. Microbiol. 130, 2905-2913.
19. Pometto ill, A. 1. and Crawford, D. 1. (1986), Appl. Environ. Microbiol. 51,171-179.
20. Peterson, G. 1. (1983), Anal. Biochem. 83, 346-356.
21. Pasti, M. B., Hagen, S. R, Korus, R A., and Crawford, D. 1. (1991), Appl. Microbiol.
Biotechnol. 34,661-667.
22. Faison, B. D. and Kirk, T. K (1985), Appl. Environ. Microbiol. 49,299-304.
23. Nie, G. and Aust, S. D. (1997), Arch. Biochem. Biophys. 337(2), 225-23l.
24. Prochazka, P., Nohynek, M., Vanek, Z., and Rokos, J. (1983), Folia. Microbiol. 28,
406-408.
25. Chen, H. C. and Whilde, F. (1991), Biotechnol. Bioeng. 37,591-595.
26. Choi, D. B., Tamura., S., Park, Y. S., Okabe, M., Seriu, Y., and Takeda, S. (1996), J.
Ferment. Bioeng. 82, 183-186.
27. Dominguez, A., Couto, S. R, and Sanroman, A. (2001), Biotechnol. Lett. 23,451-455.
2B. Ogino, H., Miyamoto, K, Yasuda, M., Ishimi, K, and Ishikawa, H. (1999), Biochem.
Eng. J. 4, 1-6.
29. Zacchi, L., Burla, G., Zuolong, D., and Harvey, P. J. (2000), J. Biotechnol. 78, 185-192.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0809/$20.00

Evaluation of Supports and Methods


for Immobilization of Enzyme
Cyclodextringlycosyltransferase

KELI A. SOBRAL, REGINA O. RODRIGUES,


ROGERIO D. OLIVEIRA, JOSE E. OLIVO,
FLAvlo F. DE MORAES, AND GISELLA M. ZANIN*

Chemical Engineering Department, State University of Maringa,


Av. Colombo 5790, CEP 87020-900, Maringa-PR, Brazil,
E-mail: gisellazanin@maringa.com.br

Abstract
An experimental design with factorial planning was used for the immobi-
lization of the enzyme cyclodextringlycosyltransferase (CGTase) from
Bacillus firmus (strain no. 37) to select the best combination of support, method
of immobilization, and conditions that gives primarily higher average
values for the specific immobilized enzyme activity, and secondarily, higher
average values for the percentage of protein fixation. The experimental
design factors were as follows: supports-controlled-pore silica, chitosan,
and alumina; immobilization methods-adsorption, and two covalent
bonding methods, either with y-aminopropyltriethoxysilane or hexa-
methylenediamine (HEMDA); conditions-7°C without agitation and 26°C
with stirring. The best combination of factors that lead to higher average
values of the response variables was obtained with immobilization of CGTase
in silica with HEMDA at 7°C. However, immobilization in chitosan at 7°C
gave the highest immobilized CGTase specific activity, 0.25 Ilmole of ~-CD /
(min·mg protein). Physical adsorption gave low specific enzyme activities,
and, in general, a high load of enzyme leads to lower specific enzyme activity.
Index Entries: Cyclodextringlycosyltransferase; immobilized enzyme;
controlled-pore silica; alumina; chitosan.

Introduction
Immobilized Enzymes
Enzymes are largely used as biocatalysts in the chemical, pharma-
ceutical,and food industries, and also as specific ligands in chemical and
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 809 Vol. 105-108,2003


810 Sabra I et al.
clinical analyses (1). Since the recovery and reuse of free enzyme in solu-
tion is limited by high processing cost, enzyme immobilization has been
pursued as an alternative technology for lower enzyme usage and higher
product throughput by weight of enzyme used. Although immobilized
enzymes can be easily reused many times in a batch reactor, their activity
might be limited by the immobilization technique and the presence of
mass transfer limitations (2). An optimum support must contain a large
surface area per unit volume (or mass), in which it should be able to
immobilize a large quantity of enzyme and offer little diffusional resis-
tance to either substrate or product, and it should be easily available,
nontoxic, and of low cost. The number of commercial applications of
immobilized enzymes has been growing steadily, and various enzymes
have already been immobilized in different polymers by several different
techniques, mainly in the form of granules or membranes (e.g., hollow
fibers), which are used for large-scale applications (3,4).
Immobilized Cyclodextringlycosyltransferase
Immobilization of the enzyme cyclodextringlycosyltransferase
(CGTase) has been pursued as a means of reducing the production cost of
cyclodextrins (CDs) from starch (5). CDs have many industrial applica-
tions, but many more would become commercially feasible if the price of
CDs could be lowered. Today's main CD applications are as complexants
for holding fragrances and flavors in the chemical, food, agricultural, and
pharmaceutical industries, and as emulsion stabilizers in food products,
jet printer inks, and so on. They have also specialized applications such as
catalysts, and in chromatographic separation of optical enantiomers (6,7).
There are not many publications describing immobilization of CGTase.
Tardioli et al. (5) have immobilized highly purified CGTase in controlled-
pore silica by covalent binding with y-aminopropyltriethoxysilane (CB-y-
APTS) and obtained a specific activity for the immobilized CGTase
(ICGTase) of 0.00185 !lmol of I3-CD I (min.mg of enzyme) at pH 8.0, 50°C,
28.96% protein fixation and 28.68% activity recovery, using maltodextrin
(5 giL) as substrate. Yang and Su (8) immobilized the CGTase from Bacillus
alkalophilic sp in chitosan by the method of covalent bonding with cross-
linking by glutaraldehyde and obtained 46% conversion of potato starch
(5%) to CDs at 60°C, pH 8.5. The main product was I3-CD (34%). In the
presence of ethanol, conversion increased to 58.3%. Abdel-Naby (9) immo-
bilized CGTase from Paenibacillus macerans NRRLB-3186 in aminated
polyvinylchloride (PVC) of three different hydrocarbon chain lengths, by
covalent binding with glutaraldehyde as a bifunctional reagent. The best
results were obtained with PVC of the largest chain length, in which 45.57%
of activity was retained, and best results for operational stability were found
at pH 6.0 and 75°C. In the reaction for CD production, ICGTase has shown
higher activation energy than the free enzyme (9).
Because the type of support and the method of immobilization used are
two important factors when enzymes are immobilized (10), we decided to
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Methods for Immobilization of CGTase 811

study various supports and enzyme immobilization methods for the CGTase
from Bacillus firmus (strain no. 37), seeking the most appropriate combination
of support, method of immobilization, and conditions that give primarily
higher average values for the specific immobilized enzyme activity, and,
secondarily, higher average values for the percentage of protein fixation.
Materials and Methods
Enzyme
CGTase from B. firm us (strain no. 37) was produced according to the
methodology of Matioli et a1. (11). The enzyme was cultivated at 37°C, pH
10.0, and 150 rpm, in 750 mL ofliqu,id medium containing: 1.0 (w I v) soluble
starch, 0.5 (w Iv) polypeptone, 0.5 (w Iv) yeast extract, 0.1 (w Iv) potassium
phosphate, 0.02 (w Iv) magnesium sulfate, and 1.0 (w Iv) sodium carbon-
ate. The cell-free culture medium had a specific enzyme activity of 5.1 !lmol
of ~-CD I (min·mg), determined as given in the section CGTase Activity.
Supports
Chitosan of pharmaceutical grade was obtained from Farmacon
(Maringa, PR, Brazil), with a particle diameter in the range of 0.212-0.425
mm. Controlled-pore silica (0.35-nm mean pore diameter) was supplied by
Sucrerie Vanciennes (Crepy in Vallois, France) with a mean diameter of
0.42 mm. Alumina (Aluminum oxide 90, acid activated) was supplied by
Merck (Darmstadt, Germany) as a fine powder.

Enzyme Immobilization: Cleaning and Hydration of Supports


Controlled-pore silica and alumina were cleaned and hydrated with
diluted nitric acid (1:20) applied at 10 mL of acid I g of support. The mix-
tures were agitated for 1 h at 82°C. Then the solids were washed with
distilled water in a Buchner funnel up to neutral pH in the filtrate and
vacuum dried for 15 min. The supports were later dried at 105°C for 15 h.
Chitosan was hydrated with successive washings with ethanol-acetone-
water (10:10:80) at 10 mL of solution I g of support, washed with distilled
water in a Buchner funnel up to neutral pH in the filtrate, and vacuum dried
for 15 min. Then, the hydrated chitosan was dried at 60°C for 24 h (12).
Immobilization by Adsorption
The cleaned and hydrated support was kept under vacuum for 15
min, and an enzymatic solution of CGTase (2 mg of proteinl g of support)
was added at 5 mLI g of support or, in the case of chitosan, at 10 mL of a
buffered solution containing 2 mM Tris-HCI buffer, pH 8.0, and 50 mM
CaCI2• The mixture of support and enzymatic solution were incubated for
24 h, at either 7 or 26°C (as was the case according to the experimental
design to be shown). The mixture was then separated in a Buchner funnel
and vacuum dried for 5 min. The filtrate was collected and a sample of it
was stocked at 4°C for later protein analysis. The solids of the immobi-
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
812 Sabra I et al.

lized enzyme were washed 10 times with distilled water and the filtrates
were collected for later protein analysis. After the last washing, the immo-
bilized enzyme was vacuum-dried for 20 min. A sample of the vacuum
dried solids was taken and dried 15 h at 105°C for determination of the
immobilized enzyme humidity. The value of protein concentration in all
filtrates was used for determination of the amount of protein that was not
fixed in the support, and from it, the percentage of protein fixation in the
immobilization procedure.
Immobilization by CB-y-APTS
Support Silanization
Silica and alumina supports were silanized with a 0.5% (v Iv) solution
of y-APTS, pH 3.0 to 4.0, added in the proportion of 3 mL of solution/ g of
support. The solids and solution were agitated for 5 min at room tempera-
ture and then kept at 75°C for 3 h. Later, the silanized support was vacuum
washed with distilled water to remove excess y-APTS and dried at 105°C
for 15 h. Chitosan, being an organic support, was not silanized.
Support Activation
Silanized supports were activated with a glutaraldehyde solution
(2.5% [v Iv] in 0.1 M sodium hydrogen phosphate buffer, pH 7.0; 3 mL of
solution/ g of solid). Support solids were kept under vacuum for 15 min,
and still under vacuum, the glutaraldehyde solution was added slowly
up to complete immersion of the solid. For chitosan, 15 mL of glutaralde-
hyde solution/ g of solid was used instead. The mixture of support and
activating solution was agitated at 26°C for 1 h. Then the solids were
washed with distilled water up to neutral pH to eliminate excess glutaral-
dehyde and vacuum dried for 20 min. CGTase was immobilized with the
same procedure as described in the section Immobilization by Adsorp-
tion. The humidity of the immobilized enzyme was determined by drying
a sample at 105°C for 15 h (5).
Immobilization by Covalent Bonding with Hexamethylenediamine
Approximately 2.5 g of cleaned and hydrated support was kept
under vacuum for 15 min and then a solution of Hexamethylenediamine
(HEMDA) (2.0% [w Iv]) was added under vacuum in the proportion of
12.8 mL of solution/ g of solid. The mixture was kept under agitation for
2 h at 40°C. Later, the liquid was decanted and an equal volume of glut-
araldehyde solution (3% [w Iv] in 0.1 M sodium hydrogen phosphate
buffer, pH 7.0) was added. The mixture was agitated for 15 min at 26°C.
Next, the suspension was transferred to a Buchner funnel and washed up
to neutral pH (10 successive washes of 12.8 mL of distilled water / g of
support), and the solids were vacuum dried for 15 min. A sample of the
immobilized enzyme was dried for 15 h at 105°C to determine humidity
(12). CGTase was then immobilized with the same procedure as described
in the section Immobilization by Adsorption.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Methods for Immobilization of CGTase 813

CGTase Activity
The activity of the free enzyme was determined at 50°C by assaying
the initial reaction rate of ~-CD production using the method of initial
velocities (13). The enzyme substrate was 10 gil of maltodextrin (Fluka,
Sigma-Aldrich, Buchs, Switzerland) in Tris-HCl buffer (pH 8.0), 0.01 M and
50 mM CaCI2• The substrate solution (1.5 mL) was warmed to reaction
temperature and mixed with an equal volume of enzyme solution. The
reaction was allowed to proceed for 30 min, and tubes were taken out of
the thermostatic bath each 5 min. The enzyme was inactivated by boiling
for 5 min, and the tubes were stocked at 4°C for later ~-CD assay (5).
The activity of the immobilized CGTase was also determined by
assaying the initial reaction rate of ~-CD production using the method of
initial velocities (13). However, in this case, a batch reactor fitted with a
stainless steel basket was used to hold the immobilized enzyme. The reac-
tion was carried out at 50°C with 50 mL of the same substrate solution just
described. After the substrate solution reached the reaction temperature,
a 1.5-mL sample was taken, this being the time zero point of the test, and
the basket containing about 1.5 g of immobilized enzyme particles was
introduced into the reactor. Samples of 1.5 mL were taken at 3-min inter-
vals up to 18 min. Then the basket was removed from the reactor and two
more samples were collected, at time 21 and 28 min, respectively. All
samples were treated as described for the free enzyme.

~-CD Assay
The concentration of ~-CD was determined by the dye extinction colo-
rimetric method with phenolphthalein, described by Vikmon (14), and
modified by Hamon and Moraes (15).

Protein Assay
The protein concentration of all enzyme solutions and collected fil-
trates was determined by the Bradford (16) method, usingCoomassie Blue
G-250, and bovine serum albumin as the standard protein.
Experimental Design
Evaluation of Supports and Immobilization Methods for CGTase
For evaluation of support/method influence on the immobilization
of CGTase, we chose a complete (32 x 21) factorial experimental design
with two qualitative factors (support and method) and one quantitative
factor (temperature). The factors and their levels were as follows: support
(factor Xl)-silica (level I), chitosan (level 0), and alumina (level -1);
methods (factor X2)-physical adsorption (level 1), CB-y-APTS (level 0),
and CB-HEMDA (level-I); and temperature (factor X3)-7°C without
agitation (level 1) and 26°C with stirring (level-I). The response variables
chosen were the percentage of protein fixation into the support used for
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
814 Sobra/ et at.
Table 1
Matrix for Full Factorial Design in Immobilization of Enzyme CGTase
Factor level code" Real factor level
Run Xl X2 X3 Support Method Temperature (DC)
1 1 1 1 Silica Physical Adsorption 26
2 1 1 -1 Silica Physical Adsorption 7
3 1 0 1 Silica CB-y-APTS 26
4 1 0 ~1 Silica CB-y-APTS 7
5 1 -1 1 Silica CB-HEMDA 26
6 1 -1 -1 Silica CB-HEMDA 7
7 0 1 1 Chitosan Physical Adsorption 26
8 0 1 -1 Chitosan Physical Adsorption 7
9 0 0 1 Chitosan CB-y-APTS 26
10 0 0 -1 Chitosan CB-y-APTS 7
11 0 -1 1 Chitosan CB-HEMDA 26
12 0 -1 -1 Chitosan CB-HEMDA 7
13 -1 1 1 Alumina Physical Adsorption 26
14 -1 1 -1 Alumina Physical Adsorption 7
15 -1 0 1 Alumina CB-y-APTS 26
16 -1 0 -1 Alumina CB-y-APTS 7
17 -1 -1 1 Alumina CB-HEMDA 26
18 -1 -1 -1 Alumina CB-HEMDA 7
'XI, support, X2, method, X3, temperature.

the immobilization of CGTase, and the specific activity of the immobi-


lized enzyme. Table 1 presents the experimental design in detail.
According to the experimental design, all 18 runs were made in tripli-
cate and carried out randomly. The mass ratio of enzyme to support at the
immobilization step was kept constant at 2.0 mg of protein/ g of support.

Results and Discussion


The influence of the factors-support, immobilization method, and
temperature-was evaluated based primarily on the response variable:
specific activity of immobilized enzyme.
ICGTase specific activity was calculated from the initial velocity data
for ~-CD production and the mass of protein fixed per gram of support. The
latter and the percentage of protein fixation were calculated from the dif-
ference between the amount of enzyme offered for immobilization and that
recovered in the washings of the ICGTase.
Table 2 presents the matrix of experimental results for the factorial
design runs and shows the results for ICGTase specific activity and per-
centage of protein fixation. The combination of immobilization method,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Methods for Immobilization of CGTase 815

Table 2
Matrix of Experimental Results from Experimental Design
Applied to Immobilization of Enzyme CGTase
Experimental conditions Results for ICGTase
Protein Specific
Temperature fixation activity
Run Support Method CC) (%) (U Img of protein)

1 Silica Physical Adsorption 26 40.13 0.15


2 Silica Physical Adsorption 7 55.37 0.14
3 Silica CB-y-APTS 26 81.67 0.19
4 Silica CB-y-APTS 7 57.77 0.16
5 Silica CB-HEMDA 26 84.13 0.15
6 Silica CB-HEMDA 7 67.60 0.20
7 Chitosan Physical Adsorption 26 39.37 0.10
8 Chitosan Physical Adsorption 7 40.27 0.15
9 Chitosan CB-y-APTS 26 44.67
10 Chitosan CB-y-APTS 7 30.67
11 Chitosan CB-HEMDA 26 98.37 0.10
12 Chitosan CB-HEMDA 7 87.57 0.25
13 Alumina Physical Adsorption 26 92.50 0.12
14 Alumina Physical Adsorption 7 77.33 0.18
15 Alumina CB-y-APTS 26 69.90 0.06
16 Alumina CB-y-APTS 7 68.23 0.11
17 Alumina CB-HEMDA 26 69.13 0.08
18 Alumina CB-HEMDA 7 66.13 0.10

support, and temperature that led to the single highest ICGTase specific
activity was the immobilization of CGTase in chitosan by covalent bonding
with HEMDA (CB-HEMDA) at 7°C. For this condition, the ICGTase spe-
cific activity was 0.25 U fmg of protein and the percentage of protein fixa-
tion was high, 87.57%.
However, since the goal of our study was to determine the level of
factors that gives primarily higher average values of the ICGTase specific
activity, we applied the analysis of variance (ANOVA) to the full model
using the software SAS for Windows® version 6.12. The results for the effect
of each factor and their interactions on the response variable ICGTase spe-
cific activity are presented in Table 3. Because the immobilization of CGTase
in chitosan by CB-y-APTS did not produce significant results, the set of data
resulting from this combination was removed from this analysis.
Allowing a level of statistical significance (p value) of 0.05, the results
of the ANOV A shown in Table 3 allow us to conclude that the significant
factors for attaining higher ICGTase specific activity are support (Xl) and
temperature (X3), which gave p values of 0.0057 and 0.0006, respectively.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


816 Sabra I et al.
Table 3
ANOVA Applied to Complete Model for ICGTase Specific Activity
Degrees Sum Mean
Factor of freedom of squares square Fvaluea p valueb
Xl 2 0.0238 0.0119 5.93 <0.0057
X2 2 0.0022 0.0011 0.56 0.5783
X3 1 0.0280 0.0280 13.95 <0.0006
X1·X2 1 0.0216 0.0216 10.75 <0.0022
X1·X3 1 0.0081 0.0081 4.03 0.0518
X2·X3 1 0.0022 0.0022 1.08 0.3044
X1·X2·X3 1 0.0074 0.0074 3.66 0.0634
Error 38 0.0763 0.0020
Total 47 0.1696
aF value = mean square of treatment/mean square of error.
b p value = level of statistical significance.

Table 4
Comparison of Difference Between Average Values of Response Variable
ICGTase Specific Activity for Factors Whose Levels are Statistically Significant
Difference between average values of ICGTase
Factors Levels specific activity (u/mg of protein)
Support 1 and-1 0.050
Temperature -1 and 1 0.048

According to this analysis, and within the range of the factorial design,
the choice of immobilization method is not significant to obtain high
ICGTase specific activity, but its interaction with the support is (Xl· X2;
P = 0.0022). The other interactions were not statistically significant.
The average values of the response variable (ICGTase specific activ-
ity) were analyzed next, for the levels of the factors that were of significance
(support and temperature), to determine whether there were significant
differences between response variable average values. Tukey test, with a
level of significance of 5%, was used and the results are shown in Table 4.
As Table 4 shows, the average values of ICGTase specific activity
differ with statistical significance when alumina (support, level 1) is
changed to silica (support, level-1), the difference being 0.050. In addi-
tion, when the temperature level 1 (26°C) is changed to temperature level
-1 (7°C), there was also a significant difference, 0.0483. Hence, the choice
of silica as support and the temperature of immobilization of 7°C, without
agitation, led to high average values of ICGTase specific activity. In rela-
tion to the immobilization method, there was not a significant difference
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Methods for Immobilization of CGTase 817

Table 5
Average Value of Response Variables Specific Activity
of ICGTase and Percentage of Protein Fixation for All Levels and Factors
of Experimental Design Applied to Immobilization of Enzyme CGTase
Average values for ICGTase Average values for
specific activity (U / mg of protein) protein fixation (%)
Level Support Method Temperature Support Method Temperature
1 0.164 0.147 0.12 64.44 57.49 68.87
0 0.152 0.128 56.82 58.82
-1 0.114 0.148 0.17 73.87 78.82 61.22

between the average values of ICGTase specific activity for the different
immobilization methods used.
To proceed in the selection of the most appropriate factors, the choice
of immobilization method was based on the second response variable: the
percentage of protein fixation. Table 5 shows the average values for the
response variables for each level of all factors. The immobilization method
that led to the highest average value for the percentage of protein fixation
(78.82%) was CB- HEMDA (immobilization method,level-1).
Therefore, the statistical analysis showed that the combination of fac-
tors including immobilization of CGTase in silica at 7°C by CB-HEMDA
results both in higher average values for the ICGTase specific activity, and
in higher average values for the percentage of protein fixation.
Further analysis of the data displayed in Table 1 shows that a too high
percentage of protein fixation is correlated with a lower ICGTase specific
activity. This is probably the result of steric hindrance of the active site of
ICGTase, because multiple enzyme layers may be formed, or excessive
crowding of the enzyme molecules into the support pores, making access
to the active sites difficult. In addition, the higher temperature (26°C) with
agitation that leads, in general, to higher percentage of protein fixation is
also associated with lower ICGTase specific activity. These two results
indicate the importance of extending this study to a range of lower enzyme
to support mass ratio, below 2.0 mg of protein/ g of support. A problem of
major concern, however, is the fact that for all factor combinations of this
study, the recovery of enzymatic activity for the ICGTase was lower than
5%, i.e., maximum ICGTase specific activity / free enzyme specific activity
x 100 < 5%. This may be the result of not using a sufficiently purified
enzyme, but it could also indicate that for greater activity recovery, a con-
trolled-pore structure with larger pores should be used. The latter hypoth-
esis is founded on the facts that CDs are large molecules, some linear
maltooligosaccharides synthesized by CGTase are of large chain size (17),
and the molecule of CGTase from B. firm us has a mol wt of 75 kDa (11).

Applied Biochemistry and Biotechnology Vol. 105-708,2003


818 Sobral et al.
Immobilization of CGTase by CB-y-APTS did not give good results,
and this combination should be removed from further studies. According
to Bon et al. (18), when silanization was applied to chitin, the reagent
y-APTS attacked chitin. The same might be happening to chitosan, because
chitin is the precursor of chitosan.

Conclusion
Statistical analysis of the experimental design data indicates that the
most important factors for guaranteeing, primarily, higher average val-
ues for the immobilized CGTase enzyme specific activities and, second-
arily, higher average values for the percentage of protein fixation is the
immobilization of CGTase in silica by the method of CB-HEMDA at 7°C,
without agitation. However, the highest ICGTase enzyme specific activ-
ity, 0.25 U / mg of protein, occurred for immobilization in chitosan by the
method of CB-HEMDA at 7°C, without agitation.
Further immobilization studies with CGTase should explore lower
enzyme-to-support mass ratios « 2 mg of enzyme/ g of support), supports
with larger controlled-pore sizes (> 0.35 nm), and new immobilization
methods seeking higher recovery of the free enzyme activity.

Acknowledgments
We are thankful for the financial support from Coordenacao de
Aperfei~oamento de Pessoal de Nivel Superior (CAPES), Conselho
Nacional de Desenvolvimento Cientifico e Tecnol6gico (CNPq), Funda~ao
Araucaria, and State University of Maringa.

References
1. Bulmus, v., Kesenci, K., and Piskin, E. (1998), Reactive Funct. Polym. 38, 1-9.
2. Wojcik, A, Lobarzewski, J., and Blaszczynska, T. (1990), J. Chem. Technol. Biotechnol.
90,287.
3. Hayashi, T., Hirayama, c., and Iwatsuki, M. (1992), J. Appl. Polym. Sci. 44, 143.
4. Hayashi, T. and Ikada, Y. (1994), J. Appl. Polym. Sci. 42,85.
5. Tardioli, P. W., Zanin, G. M., and Moraes, F. F. (2000), App!. Biochem. Biotechnol. 84-86,
1003-1019.
6. Bekers, 0., Uijtendaal, E. V., Beijnen, J. H., Bult, A., and Underberg, W. J. M. (1991),
Drug Dev. Ind. Pharm. 17, 1503-1549.
7. Korokolvas, A (1991), ENLACE Farmalab 2, 6-15.
8. Yang, C. P. and Su, C. S. (1989), J. Chem. Technol. Biotechnol. 46,283-294.
9. Abdel-Naby, M. A (1999), Process Biochem. 34,399-405.
10. Szejtli, J. (1988), in Cydodextrin Technology, Szejtli, J., ed., KIuwer, Dordrecht, The
Netherlands, pp. 79-185.
11. Matioli, G., Zanin G. M., Guimaraes, F., and Moraes, F. F. (1998), Appl. Biochem.
Biotechnol. 70-72,267-275.
12. Pereira, E. B. (1999), MS thesis, UEM, Maringa, PR, Brazil.
13. Dixon, M. and Webb, E. c., (1979), Enzymes, 3rd ed., Chap. 7, Longman Group, Lon-
don, UK, pp. 7-22.
14. Vikmon, M. (1982), in The First International Symposium on Cydodextrin, Szejtli, J., ed.,
D. Reidel Publishing Company, Dordrecht, The Netherlands, pp. 60-74.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Methods for Immobilization of CGTase 819
15. Hamon, V. and Moraes, F. F. (1990), Internal Report, Laboratorie de Technologie
Enzymatique, Universite de Technologie de Compiegne, Compiegne, France.
16. Bradford, M. (1976), Anal. Biochem. 72,248.
17. Matioli, G., Zanin G. M., Guimaraes, F., and Moraes, F. F. (2001), Appl. Biochem.
Biotechnol. 91-93,643-654.
18. Bon, E., Freire, D. G., Mendes, M. F., and Soares, V. F. (1984), Biotechnol. Bioeng. Symp.
14,447-455.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0821/$20.00

Active Nuclear Shuffling System


Using a Swollen Conidium
of Trichoderma reesei

HIDEO TOYAMA" MAKIKO YANO, AKANE GISUSHI,


T AKESHI HOIIA, AND NOBUO TOYAMA
Department of Food Science and Technology,
Faculty of Horticulture, Minamikyushu University,
Miyazaki 884-0003, Japan,
E-mail: gaf00771@nifty.com

Abstract
Cellulase hyperproducers of Trichoderma reesei can be constructed using
autopolyploidization and haploidization techniques. To increase the effi-
ciency of this method, the active nuclear shuffling system in a swollen
conidium was effective. A dried mature green conidium of a model strain,
T. reesei QM6a (lFO 31326), was swollen to make room for a larger autopoly-
ploid nucleus. After colchicine treatment, a larger autopolyploid nucleus
was produced in such a swollen conidium. Benomyl treatment of swollen
conidia generated multiple smaller nuclei from one larger autopolyploid
nucleus. Those smaller nuclei were transported through conidia to mycelia
after germination. This system could contribute to increasing the efficiency
of genetic shuffling.
Index Entries: Trichoderma reesei; cellulase; colchicine; cellulose; benomyl.

Introduction
Trichoderma is a well-known cellulolytic fungus (1). For the purpose of
breeding this fungus, chemical mutation and genetic engineering tech-
niques have been mainly applied (2,3). Therefore, we attempted to create
a new breeding technique using autopolyploid nucleus. Previously we
demonstrated that autopolyploidization and genetic recombination can
be carried out on this fungus using a mitotic arrester, colchicine, and a
haploidizing reagent, benomyl (4). Moreover, we reported the selection
system of cellulase hyperproducers using a double-layer selection medium
from conidia as genetic recombinants (5). In this article, we present our
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 821 Vol. 105-10B, 2003


822 Toyama et al.

attempt to increase the frequency of genetic recombination using swollen


conidia of this fungus.

Materials and Methods


T. reesei QM6a (IFO 31326) was used as a model strain. This fungus
was cultivated on potato dextrose agar (PDA) (BBL, Cockeysville, MD)
medium at 28°C and preserved at 4°C. For preparation of dried green
mature conidia, a mycelial mat (2 x 2 mm) was put on a PDA medium and
incubated for 2 wk at 28°C in order to generate a large amount of green
mature conidia on the colony. Those conidia were suspended in sterilized
water followed by filtration with a glass filter 3G-2 for the purpose of
removing mycelia. After filtration, those conidia were collected by cen-
trifugation (Sigma 2-15) at 5510g followed by drying in a desiccator. Those
treated conidia were used as dried mature green conidia.
At first, swollen conidia were prepared. As the medium for prepara-
tion of swollen conidia, Mandel's medium containing 1.0% (w Iv) glucose
(Wako, Osaka, Japan) and 0.5% (w Iv) peptone (Difco, Detroit, MI) (pH 6.0)
was used. The conidia of T. reesei QM6a are oval and mononucleate (6). After
Giemsa staining, one portion was stained in a conidium. The Giemsa-stained
portion was also stained with 4.6'-diamidino-2-phenylindole (DAPI) solu-
tion, such that the Giemsa-stained portion was regarded as a nucleus (7).
The inner volume of the original conidium was increased in order to pro-
duce a larger autopolyploid nucleus by swelling the conidium. Five loopfuls
of dried green mature conidia were added to 50 mL of the medium for
preparation of swollen conidia in a lOO-mL Erlenmeyer flask and incubated
for lO h at 28°C using a rotary shaker (Taitec R-30 mini) (160 rpm). After
incubation, swollen conidia were observed by microscope in the medium.
Those swollen conidia were collected by centrifuging at 55109 (Sigma 2-15;
Sigma, St. Louis, MO). Conidia were pretreated with 5 N HCl (Wako) for 40
min at 60°C. After washing with distilled water, nuclear staining was car-
ried out using Giemsa solution (Merck, Darmstadt, Germany) and DAPI
solution (Sigma) followed by photomicrography. Swollen conidia were
pretreated with 0.1 N HCl for 30 min at room temperature followed by
washing with distilled water and nuclear staining. When these swollen
conidia were stained with Giemsa solution, one small nucleus was observed.
The diameter of the nucleus was the same as that of the original conidium.
Next, colchicine treatment of the swollen conidia was carried out.
Swollen conidia were added to 20 mL of the medium in a 50-mL Erlenmeyer
flask and incubated at 28°C under stationary conditions. Mandel's medium
containing 1.0% (w Iv) glucose, 0.5% (w Iv) peptone, and 0.1 % (w Iv)colchi-
cine (Wako) (pH 6.0) was used. Swollen conidia were incubated in the
medium for 2 wk at 28°C under stationary conditions. After the colchicine
treatment, the swollen conidia were collected by centrifuging at 55109
(Sigma 2-15). When the colchicine-treated swollen conidia were stained
with Giemsa solution, one larger portion was stained in a swollen conidium.
Applied Biochemistry and Biotechnology Vol. 705-108, 2003
Active Nuclear Shuffling Using T. reesei 823

Table 1
Comparison of Number of White Colonies
No. of white colonies No. of green colonies
Experiment A-
I 1 904
2 0 885
3 1 923
4 3 954
Experiment Bb
1 7 948
2 11 908
3 9 865
4 15 981
aBenomyl treatment of treated swollen conidia in the solid medium.
bBenomyl treatment of treated swollen conidia in the liquid medium.

Since it was also stained with DAPI solution, this Giemsa-stained portion
was regarded as an autopolyploid nucleus.
The swollen conidia containing autopolyploid nuclei were spread on
the solid medium for benomyl treatment for 10 d at 28°C (experiment A).
PDA medium containing 0.1% (v Iv) Triton X-100 (polyoxyethylene-
octylphenylether) (Wako) and 0.6 Ilg/mL ofbenomyl(l-[butylcarbamoyl]-
2-benzimidazolecarbamate) (Sigma) was used.
Benomyl is a fungicide that deletes chromosomes from the polyploid
nucleus with or without chromosomal recombination (8). Therefore,
benomyl treatment was carried out in this experiment. Triton X-100 was
used for restraining mycelial elongation for ease of colony counting. After
incubation, the number of colonies that generated white conidia (white
colony) was counted. PDA medium containing 0.1 % Triton X-lOO (pH 6.0)
was used. The number of white colonies generated after incubation is given
in Table 1. It was suspected that the white colonies were produced through
genetic recombination. Thus, they were used for estimation of genetic shuf-
fling frequency.
Finally, benomyl treatment of swollen conidia in the liquid medium
was carried out (experiment B). Mandel's medium containing 1.0% (w Iv)
glucose,O.5% (w Iv) peptone, and 0.4 Ilg/mLofbenomyl (pH 6.0) was used.
Five loopfuls of swollen conidia containing autopolyploid nuclei were
added to 20 mL of the liquid medium in a 50-mL Erlenmeyer flask for
haploidization followed by incubation for 2 wk at 28°C under stationary
conditions. When the benomyl-treated swollen conidia were stained with
Giemsa solution, various types of smaller nuclei were observed in a swol-
len conidium. When such swollen conidia containing smaller nuclei were
incubated in Mandel's medium containing 1.0% glucose and 0.5% peptone
at 28°C, germination occurred and such smaller nuclei were transported
Applied Biochemistry and Biotechnology Vol. 105-108,2003
824 Toyama et al.

Table 2
Comparison of Frequency of White Colonies
White colonies/ Frequency of white
Total colonies colonies (%)
Experiment A 5/3671 0.14
Experiment B 42/3744 1.12

through mycelia. It was suspected that those smaller nuclei are transported
to conidia like the original nuclei.
The benomyl-treated swollen conidia were washed with sterilized
water and spread on the medium for counting the number of colonies and
incubated for 10 d at 28°C followed by the white colony count. As a result,
it appeared thatthe occurrence of white colonies was almost 10 times higher
than that in the solid medium containing benomyl (experiment A), as shown
in Tables 1 and 2.

Conclusion
From these results, we considered that a higher frequency of genetic
shuffling could be obtained by benomyl treatment in the liquid medium.
Consequently, such a technique was named "active nuclear shuffling,"
which we concluded can contribute to obtaining a higher frequency of
genetic recombination in T. reesei.

References
1. Reese, E. T. and Mandels, M. (1980), Biotechnol. Bioeng. 20(A2), 323-335.
2. Morikawa, Y., Kawamori, M., Shinsha, Y., Oda, F., Takasawa, S., and Ado, Y. (1985),
Agric.Biol. Chem. 49, 1869-187l.
3. Durand, H., Clanet, M., and Tiraby, G. (1985), Bioenergy 84, 246-253.
4. Toyama, H. and Toyama, N. (2000), Appl. Biochem. Biotechnol. 84-86,419-429.
5. Toyama, H., Yamagishi, N., and Toyama, N. (2002), Appl. Biochem. Biotechnol. 98/100,
257-263.
6. Rosen, D., Edelman, M., Galun, E., and Danon, D. (1974), J. Gen. Mirobiol. 83,31-49.
7. Yamada, M., Matsumoto, Y., Hamada, S., Fujita, S., and Yoshida, Y. (1986), Zbl. Bakt.
Hyg. 86, 6503-6507.
8. Bilinski, CA., Sills, A. M., and Stewart, G.G. (1984), Appl. Environ. Microbiol. 48,
813-817.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


SESSION 7
Downstream Processing of Bioproducts
Session 7

Downstream Processing of Bioproducts

MICHAEL COCKREM1 AND CESAR C. SANTANA 2


KiwiChem International, Madison, WI; and
1

2State University of Campinas, Sao Paulo, Brazil

The recovery and purification of the desired products from bioreactor


outflow is greatly influenced by the nature of the products themselves, the
size of the market, the need to achieve market and legally required specifi-
cation standards; and the market value of the product. As the scale increases,
economic methods of conducting product recovery and purification become
not only important but essential for competitive process integration. It is
therefore worthwhile to conceive a proper downstream scheme early in the
process development, which requires laboratory development and process
engineering efforts. Even small improvements can be very profitable in terms
of the overall process.
This session includes papers from presentations on such operations as
adsorption, membrane filtration and electrodeionization, which is gaining
importance in several new process developments such as enzyme recovery
and purification, elimination of suspended solids and colloidal materials
from biomass feed streams as well as salt removal in chemical, pharma-
ceutical, and biomass process industries. Other presentations focused on
thermochemical conversions of biological materials into higher-value com-
pounds. Esters and phase-change materials can be made from fats and oils.
Polyesters are a class of polymers that can be produced from biological
materials. Likewise lignin can be converted into fuel oxygenates. These pre-
sentations and those in the poster session illustrate that biotechnology for
fuels and chemicals is not a "stand alone" technology, but must be inte-
grated with separations and other conversion processes.

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 827 Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0829/$20.00

Phenylboronate-Chitosan Resins
for Adsorption of ~-Amylase
from Soybean Extracts

EDUARDO J. ARRUDA1 AND CESAR C. SANTANA*,2


1Pharmacy and Biochemistry, Dom Bosco Catholic University,

UCDB, CP. 1~O, CEP 79117-900, Campo Grande, MS, Brazil; and
2School of Chemical Engineering, DPB, UN/CAMP, CP. 6066,
CEP 13083-970, Campinas, SP, Brazil; E-mail: santana@feq.unicainp.br

Abstract
Isolation and purification of bioproducts from crude extracts can be
obtained by affinity methods based on reversible binding of a specific mol-
ecule to ligand immobilized in a porous matrix. In the present work,
nicrospheres based on chitosan matrix, which incorporated aminophenyl-
boronic acid as a derivative, were prepared and characterized, aimed at
developing a fJ-amylase adsorption process. Kinetic curves and adsorption
isotheriru of the crude extracts as well as the breakthrough curves for
a frontal chromatographic separation method of a commercial sample of
fJ-amylase from soybean are presented. These results were compared to simi-
lar data obtained with a comercial microspheres gel based-on agarose.
Index Entries: Purification; ~-amylase; soybean; phenylboronate; chitosan.

Introduction
Interest in bioproducts has been increasing as a result of biotechno-
logical development. Industrial demands for new applications, specific-
ity, and renewability of products have also increased dramatically. Because
of the expansion of downstream processing in biotechnology, the methods
aimed at concentration and purification of enzymes and biopolymers after
their production from a variety of sources require new advances (1-3).
~-Amylase (~-l,4-glucanmaltohydrolase, Ee 3.2.1.2) is an exoenzyrne that
removes units of nonreduced maltose terminals from polysaccharide
chains, producing ~-maltose and ~-limit dextrins. There is a considerable
industrial interest in this enzyme for the production of syrups rich in mal-

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 829 Vol. 105-108, 2003


830 Arruda and Santana
tose. Separation and purification of ~-amylase still require new methods
because of the expansion market.
The phenylboronate ligand, which is used in affinity separation of
some diol-containing analytes, presents the following main characteristics:
high connection capacity, good flexibility under conditions of adsorption
and elution (usually sorbitol), chemical and biologic stability, and lower
cost than that of naturallectins (4).
Chitosan, poly(o-glucosainine), is obtained by deacetylization of chitin,
poly(N-acetyl-D-glucosamme). Chitosan dissolves in acids of medium
strength and its molecule contains an anine group used extensively in chemi-
cal modifications (5) for several purposes. Compared with other immobili-
zation matrices, the chitosan matrix has wider pores and a larger surface
area, which allows new preparation conditions to be devised (6), and larger
amounts of enzyme to be immobiized under the same reaction conditions.
Compared to the other tested matrices, chitosan showed the best perfor-
mance, owing to lower resistance to diffusion and higher fiow rates (7).
Chitosan is similar to amylose in its structure, and it was chosen for its
potential in imparting specificity to the separation (8).
Batista-Viera et al. (9) described reversibility for adsorption and elu-
tion as well as covalent immobilization of soybean amylase on agarose-
phenylboronate in their discussion of the potential recovery possibility
for purification of ~-amylase from soybean extracts by affinity methods.
The present article describes our study of two ways to extract ~-amylase
from soybean to optimize the extraction mode. We have also synthetised
an affinity phenylboronate-chitosan (PBC) adsorbent and studied several
aspects of the interaction mechanism involved in the recognition of ~-amy­
lase on the PBC sorption material. Aiming at a possible scale-up of the
process, we studied the adsorption kinetics and isotherm and the break-
through curves in a frontal mode. The prepurification, adsorption, and
elution steps were dialyzed by quantification of the specific activity during
the various steps, aiming at the purification of soybean ~-amylase.

Materials and Methods


Chitosan-crab shells of a practical grade and sodium borohydride
were obtained from Sigma (St. Louis, MO), 3-aminophenylboronic acid
and dinitrosalicylic acid (DNS) from Aldrich (Milwaukee, WI), phenyl-
boronate-agarose (PBA-30) from Amicon (Denver, CO), and Coomassie
blue reagent from Bio-Rad (Richmond, CA). All other chemicals used were
of reagent grade.
Preparation of Chitosan Microspheies and Adsorbent
The required volumes (usually 250 mL) of an aqueous solution
containing chitosan-free base were prepared: Dissolving chitosan-free
base in acetic acid produced chitosan acetate salt. The spray method was
used with a standard 0.5 mm SSBR-JBC nozzle (Spraying Systems,

Applied Biochemistry and Biotechnology Vol. 105-108,2003


PBC Resins for Adsorption of~-Amylase 831

Campinas, Brazil). When the liquid was fed into the nozzle with a peristal-
tic pump, atomization occurred owing to the force of the compressed air,
separating the liquid into small droplets. The product was then collected.
Under standard conditions, the temperature, spray flow, and compressed
spray airflow (represented as the volume of air input per unit of time) were
set at 25°C, 6 mL/min, and 10 L/min, respectively (10). Noncrosslinked
chitosan atmospheres were prepared using a spray method by adding
2.5% (w /w) chitosan solution in 0.75 N acetic acid to 1 N NaOH and were
left to stand for 15 min. The micro spheres prepared in this way were highly
spherical and had a smooth but distorted surface morphology. Prepara-
tion conditions also had some influence on particle size. Porosity is a very
important property of chitosan matrices and can be changed using differ-
ent experimental conditions during preparation (stirring, ratio of chitosan
solution in AcOH/NaOH, temperature, stabilizer, and crosslinking).
Synthesis was carried out in three stages: preparation of the matrix,
activation, and derivatization (with reduction). Preparation of the matrix
was a process of obtaining porous particles from commercial chitosan.
Crosslinking and activation of the matrix involved reaction with a
dialdehyde agent. The derivatization was a reaction between aldehyde
groups in the matrix and amine groups in the ligand, 3-aninophenylboronic
acid, under mild conditions and the final reduction to eliminate unsatur-
ated sites. The resulting microparticulate gel (beaded noncrosslinked
chitosan) was conditioned overnight in 50 mM phosphate buffer at pH 8.0
and crosslinked and activated for 6 h at 4°C, using a ratio 0.45 mol/mol
aldehyde/ amino groups ratio. The activated matrix was washed with
phosphate buffer. Then a solution containing 16 Ilmol ofboronic acid/mL
was added, and the ligand was immobilized for 72 h at 4°C under stirring
(175 rpm). The immobilization density of the ligand in the chitosan matrix
was evaluated using ultraviolet (UV) absorption at 280 nm through the
determination of ligand concentration in solution before and after immo-
bilization (11). The PBC resin obtained was washed with distilled water
and reduced with sodium borohydride at 4°C.
Characterization of Adsorbent and Batch Experiments
Particle size distribution of the PBC adsorbent was obtained through
the light-scattering method using Malvern Mastersize equipment. The
total micropore and mesopore volumes of the dried and crosslinked
chitosan matrix were determined by nitrogen adsorption (Brunauer,
Emmet, and Teller [BET] method) and desorption methods. Batch adsorp-
tion experiments were carried out in Eppendorf tubes to characterize the
effects of pH and ionic strength on adsorption capacity.
Extraction of B-Amylase from Soybean
To verify the extraction efficiency and the content of ~-amylase in soy-
bean (PL-1/IAC, Brazil), extraction procedures developed by Smith et al. (12)
and Rai et al. (13,14) were used. Extraction was done with 1 g of defatted
Applied Biochemistry and Biotechnology Vol. 105-108,2003
832 Arruda and Santana
soybean flour, and diluted with buffer to obtain a protein solution used in
capillary electrophoresis.
Capillary Electrophoresis
The equipment used was Beckman P / ACE model 5010 in CGE mode
with an uncoated, fused silica capillary tube (47 cm x 75 J.tm). Samples were
dissolved in 120mMTris-HC1 buffer,pH6.6,containing 1% (w/w) sodium
dodecyl sulfate. Beckman provided the run gel buffer. Voltage was 14.1 kV
at 20°C. The extracts were lyophilized for further use.
Determination of Protein
The total protein content was determined by the Bradford (15) method,
using bovine serum albumin as a reference.
Determination of ~-Amylase Activity
~-amylase activity in the extracts was determined by the Bernfeld (16)
method, in 125 mM acetate buffer, pH 5.0. One unit of ~-amy1ase (activity
unit [AU]) was defined as the amount of enzyme that catalyzed formation
of 1 J.tIDol of maltose per minute using 1 mL of enzyme solution under given
conditions. In a typical batch determination, 400 J.tL of 2.5% (w /w) starch
solution was added to 400 J.tL of 125 mM acetate buffer, pH 5, with 200 J.tL
of the enzyme solution and the solution was immediately incubated at 60°C
for 30 min. After 4 mL of reagent DNS was added, the reagent was mixed
vigorously and boiled for 15 min in a bath, and the resulting sample was
collected in an ice bath. The UV absorbance was measured at 640 nm against
a blank with the same component mixture without the enzymatic solution.
The calibration curve of maltose (0.2-2 mg/mL) was prepared for
conversion of spectrophotometiic readings of the enzyme activity. The
specific activity was expressed as the activity of ~-amylase/mg of total
protein.
Kinetic Curves and Adsorption Isotherms
The experimental procedures for determination of both the kinetic
and isotherm curves were similar. Setups containing syringes and dispos-
able pistons with retention were used. Samples with the adsorbent and the
enzyme solution were placed in the syringes and conditioned in 25 mM
phosphate buffer, pH 6.8, containing o. 1 M NaCI (adsorption buffer) for 1
h with temperature controL In those experiments, I mL of the enzyme so-
lution mixed with the adsorbent and 15 mg of the lyophilized extract di-
luted in 50 mL of phosphate buffer were added and placed on a rotation
device (10 rpm). The revolution simulates an agitated tank and allows dis-
persion and homogenization of the suspension. Samples were withdrawn
at a few minutes' intervals for determination of the adsorption kinetics and
after 180 min for acquisition of the adsorption equilibrium data. The super-
natant was evaluated in tern of total protein and activity as the amount of
total protein or activity. For each concentration, the capacity was deter-
mined as the amount of total protein or activity (protein mg, or AU) con-
Applied Biochemistry and Biotechnology Vol. 105-108,2003
PBC Resins for Adsorption of~-Amylase 833
tained in 1 g of the adsorbent. The equilibrium data were fitted with the
Langmuir model (17).
Chromatographic Experiments
Isocratic Elution
The PBC and phenylboronate-agarose (Matrix gel PBA-30; Amicon)
adsorbents were conditioned in a column and tested. A Pharmacia HR 5/
5 column (0.5-cm diameter) was used and coupled to a Shimadzu high-
performance liquid chromatography system and a data acquisition system.
The tests were carried out with 0.85 g of each adsorbent (ie., 6 cm of the
adsorbent bed in the column) at a flow rate of 0.25 mL/min. A sample of a
prepurifled and dialyzed soybean ~-amylase was injected into the column
and preequilibrated with 25 mM phosphate buffer (pH 6.8). The same buffer
was used for isocratic elution with 0.1 M sorbitol and 0.1 M NaC1, or 0.5 M
NaCI for PBA-30 and PBC, respectively, at a flow rate of 0.25 mL/min. The
eluted phase was collected in 1-mL fractions and samples were analyzed
for protein concentration and activity.
Frontal Analysis
In the fixed-bed adsorption experiments, soybean extract was con-
tinuously fed into the column containing the solid phase, formed by beads
of the PBC adsorbent. The feed was maintained until the desired product
(~-amylase) appeared in the effluent at the feed concentration. The column
was packed with 0.5 g of the adsorbent beads and had a diameter of 0.5 cm,
a length of 2 cm, and a constant operational flow rate of 0.5 mL/min.

Results and Discussion


Physicochemical Characterization of Adsorbent
The particle size distribution of the prepared adsorbent was mea-
sured through a Malvern 3100 Light Scattering Instrument. The PBC
adsorbent contained particles in the range of 10-300 Ilm with a wide size
distribution profile and an average Sauter diameter of 98 Ilm. The particle
size distribution of the coninercial boronate-agarose adsorbent showed a
bimodal distribution with a predominance of particles in the range of 40-
200 Ilm, with an average Sauter diameter of 110 Ilm. The micropore and
mesopore size distribution characteristics of the various reticulated matri-
ces are presented in Figs. 1 and 2, for several molar ratios of glutaralde-
hyde (-CHO) / chitosan matrix (-NH2)' PBC adsorbent with a molar ratio of
0.45 was prepared. Cumulative pore volume increased with molar ratio,
and pore sizes between 50 and 100 nm were responsible for a great part of
the total pore volume.
During the drying of the swollen adsorbent containing water, the
pore structure collapses and data obtained do not describe porosity of the
original adsorbent. To study the contribution of the matrix-spacer arm and
ligand in the adsorption, the influences of pH and of ionic strength were
Applied Biochemistry and Biotechnology Vol. 105-108,2003
834 Arruda and Santana
~/M8IarR.-(..cHOI-NII.)

..
- 0 - RI 0.30
- 0 - R2 D,4S
1.0 - 6 - R3 Cl,!IS
-v- R4 1.45
'to -0-
a
~ 4~
-e- R6 S.so
~

I
- 0.5
~
j
0.0+--..--.......,...................---.-......-.,......,..........~....4--.-........
1 10 100
Pore diameter ( am)

Fig. 1. Cumulative pore volume of chitosan matrix crosslinked with glutaraldehyde.

2.5.,..----------------...,
~'MIW ..... (-CIIOI_MI,)

i
D··· R1 0,30
2.0 o· R2 CI,4'
6,.. R3 CI,95
9
'1
V R.4 I,~
... ~.. IS 2,IS ;\<>
1.5 •... R6 ,,,.

! _ I . .'. . IliIfcnIIIiII c.-


~~
~ 1.0

~~~o
e
£ 0.5

..
O.O~. . . . ._.~::::.____..........~~..___._.......I
1 10 100
Pore diameter <DID)

Fig. 2. Logarithmic differential pore volume of chitosan crosslinked with glutaraldehyde.

evaluated (Figs. 3 and 4), Figure 3 shows the effect of ionic strength on total
protein adsorption capacity and activity in terms of NaCI content at a fixed
pH 6.S. A similar behavior was observed for PBC, with a minimum capac-
ity at an NaCI concentration of about 0.5 M, in terms of both total protein
and activity. For PBA-30 there was a continuous increase in both capacities
with increasing ionic strength. Evaluation of the adsorbents in relation to
pH showed a typical behavior for the adsorption of ~-amylase at the iso-
electric point of the molecule and in its vicinity. As can be seen in Fig. 4, the
PBC adsorbent had an optimum pH value for adsorption for both total
protein and activity of about 5.5, which is equivalent to the isoelectric
point of soybean ~-amylase. For the PBA-30 adsorbent, minimum adsorp-
tion capacity of total protein occurred at a pH of 7.0, while a capacity in
terms of activity decreased continually with an increase in pH. The differ-

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


PBC Resins for Adsorption of ~-Amylase 835

1J
2S
. .....
-a-PDC
700 1 J
1 2.0 .. ·e· PBA 30 600 ..1

{..S soo '(


r 4009
i
t
1.0
~
300 i t
i.i i
u O.S PDC
-0- 200 u
··o··PBA30
O.O-t----r-....--r--r----r--...........---r--+lOO ~
~ 0.0 005 1.0 l.S 2.0 ~
IeDie ........ N.C1 (M)

Fig. 3. Effect of ionic strength on adsorption capacity of adsorbents (pH 6.8).

C'
-I•

:',
C'2.00
I ......... 600';a
\-Q\
c
j
-:. 1.7S

.J
iel. SO
.~~. 3OO~
p

i
r.~_
i l.2S
2001u
1 •

~ 1.00 • 0 -c-PBC 100 •

eo •• ·0 P8A30
0
eo
.;
i
ow 3 4 S 6 7 8 9 10 11 ~
~
pH

Fig. 4. Effect of pH on adsorption capacity of adsorbents (ionic strength of O. 1 M


NaCl).

ences in ionization of the adsorbents have an influence on the amount of


~-amylase adsorbed in the studied pH range.
Interestingly, because of so-called secondary interactions (ionic, charge
transfer, hydrophobic, and hydrogen bound), phenylboronate ligand can
be used for purification of nonglycosylated proteins, such as in the case of
soybean ~-amylase. Both adsorbents retained ~-amylase at low concentra-
tions of buffer and salt. This reflected a recognition mechanism based on the
formation of an ion pair between the ligand and ~-amylase (ionic interac-
tions). However, secondary interactions, such as hydrogen bridges, and
also hydrophobic interactions cannot be excluded. In fact, Figs. 3 and 4 show

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


836 Arruda and Santana
a maximum adsorption efficiency for PBC with no NaCl in the buffer; this
decreased to a minimum adsorption in the NaCl concentration range of
0.25-1 M, and again increased for an NaCl concentration higher than 1 M. By
contrast, for PBA-30 it was necessary to use at least 0.1 M NaCl to reach the
same yield. The hydrophobic contribution was more pronounced in PBA-
30, and an increasing amount was adsorbed as ionic strength increased. The
behavior of the total protein adsorbed is identical to the activity, which
shows that the adsorbed chemical species maintain a close relationship. A
similar behavior was reported (19) for the system immunoglobulin G on
agarose and histidine.
It is worth pointing out that nitrogen groups can be protonated,
depending on the pH in both cases reported (chitosanglutaraldehyde-
phenylboronate and agarose-carbodiimide-histidine), and, therefore, this is
a way to modify the intensity and the type of interaction of the matrix and
ligand (i.e., ion exchange).
Characterization of the Soybean Extract
Figure 5 displays an electrophoretogram for two soybean protein
extracts obtained with the two extraction methods described. The migra-
tion times of ~-amylase (58.8 kDa) and trypsin inhibitor (20.7 kDa)
molecules were 23.34 and 19.67 min, respectively. The composition of Ren
et al.'s (13) extracts was approx 77% trypsin inhibitor and 21.5% ~-amy­
lase. The extraction methods for the fractions of soybean ~-amylase,
described earlier, showed differences in selectivity. The method proposed
by Smith et al. (12) showed less selectivity, extracting the same number of
molecular species of low as well as of high molecular mass (3-300 kDa). In
addition to selectivity, the extraction method proposed by Ran et al. (13,14)
restricted the present molecular species to a quite narrow range of molecu-
lar masses (60 kDa).
Adsorption Kinetics and Isotherms
Kinetic curves were obtained for both adsorbents for total protein
and activity. Figure 6 shows typical results. A fast decrease in the protein
concentration and initial activity-70 and 50% for PBA-30 and 90 and 70%
for PBC, respectively-can be observed during the first 25 min. This dem-
onstrates a higher capacity and a faster kinetics of adsorption for PEC,
probably caused by a higher density of the ligand and a smaller resistance
to diffusion because of wider pores. Experimental data on equilibrium
adsorption in the batch procedure are shown in the isotherms depicted in
Fig. 7. The adsorption capacity of the adsorbents in terms of activity were
adjusted as a function of the enzyme activity by the model proposed by
Langmuir (17), described by Eq. 1 as

(1)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


PBe Resins for Adsorption of~-Amylase 837

0.025 !ley...... Il.:cncto


- BuffCTPO.4 10 111M pH 1,25 (Smilb't mctbod)
- EdIInoI..oIuIlIc . . - (ReD't mcIboc!)
0.020
~
....
-..
;::; 0.Q\5

~
:; 0.010
'f
:i
.D
....: 0.005

0.000 +-......~-r-.....-,,.....-.-.-r-"'-,-~....-......--.--t
o 5 10 15 20 25 30 35 40 45
MIgration time (minutes)

Fig. 5. Electrophoretogram for two soybean extracts obtained with extraction meth-
ods described.

6....-----------------------------~
A_rIMa..
5 -.-PBA30
-o-PBC

~---I':"I---,---
o~~~~~~~~~~
o 50 100 150 200 250 300 350 400
Migration time (minutes)

Fig. 6. Kinetic curves for PBA-30 and PBC adsorbents measured in batch experiments.

in which q' (mg/ g) is the total protein adsorbed per g of the adsorbent, c'
(mg/mL) is the concentration of the protein in the solution, qm (mg/ g) is
the maximum capacity of the adsorbent, and kd (mg/mL) is the dissocia-
tion constant. A fitting to experimental data was obtained by nonlinear
regression for the PBA-30 and PBC adsorbents for both the crude and
dialyzed extracts. Table 1 gives the parameters calculated for the iso-
therms obtained with the crude extract, and Table 2 displays the results
for the dialyzed soybean extract. The values of the parameters obtained
indicate that the adsorbents have similar adsorption capacities and PBC
has greater affinity characteristics than PBA-30.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


838 Arruda and Santana
~~---------------------------, "....,.
I~ _FBA30
_PBC

4
)'2000
e. lSOO
I
.. 1000
<
f~
O~~~~~--~-r--~-r--r--r~
o 100 200 300 400 SOO CiOO 700 100 900 1000
AcIivIly (au BIL-1)

Fig. 7. Activity isotherms at 25°C, pH 6.8, and ionic strength of O. 1 M NaCI carried
out in batch. Langmuir parameters for crude extracts for PBA-30 are: qm =1617 AU /
g, kd = 75.92 AU /mL; and for PBC are qm = 1626 AU / g, kd = 59.32 AU /mL. For dialyzed
extracts for PBA-30 they are: qm =2865 AU / g, kd =86.23 AU /mL; and for PBC are qm =
3082 AU/g, kd = 72.22 AU/mL.

Table 1
Langmuir Parameters of Adsorption Isotherms with Crude Extract
Total protein Total activity
qm kd qm kd
Adsorbent (mg/g) (mg/mL) R2 (AU/g) (AU/mL) R2
PBA-30 9.83 0.61 0.96 1.617 75.92 0.98
PBC 12.62 0.43 0.95 1.625 59.32 0.97

Table 2
Langmuir Parameters of Adsorption Isothern with Dialysed Extract
Total protein Total activity
qm kd qm kd
Adsorbent (mg/g) (mg/mL) R2 (AU/g) (AU/mL) R2
PBA-30 39.18 1.54 0.95 2.865 86.23 0.99
PBC 39.02 0.68 0.97 3.083 72.22 0.97

Breakthrough Curves
Figures 8 and 9 display the breakthrough curves for PBA-30 and PBC
adsorbents, determined in frontal chromatographic experiments. Curves
of total protein concentration and fl-amylase activities are shown for crude
and prepurified systems to demonstrate the effect of contaminants on com-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


PBC Resins for Adsorption of B-Amylase 839
2~~-----------------------------.
60
r
r
i200
r:i ~ "..1
e~
i
~I~ 40
i
/ #"J
II III
~

.jIljll
.e
a I d I
Pnteia 30
~
..
100 - without dialysis
£ ----0- with dialysis 20
~
i<
cJ Adivity
~ ~
/ -e- without dialysis 10
- 0 - with dialysis
0
I
0
0 10 20 30 40
Injected Volume (mL)

Fig. 8. Breakthrough curves for crude soybean extracts flowing through PBA30.
Column height is 2 cm, and column diameter is 0.5 cm. Mass of adsorbent is 0.5 g and
flow rate is 0.5 mL/min.

180 60
.r"'
~160 .-
so "I
.! 140
'I
II
1 120 40 i
! 100
..... i
•l
.! 80 30 ~
~

f
- Without dialyIia
60 20
~- Wdh dialysis
i 40
~ Actmty
_ Without dialyIia 10 <II!
20
- < l - With dialysis
0 0
0 5 10 20 2S 30 35
15 40 45
IDJeded Volume (mL)

Fig. 9. Breakthrough curves for PBC. Column height is 2 cm, and column diameter
is 0.5 cm. Mass of absorbent is 0.5 g and flow rate is 0.5 mL/min ..

petition at the adsorption sites. The behavior of the activity breakthrough


curves is typical of one-component ideal adsorption.
Enzyme Punfication Protocol
In a typical experiment with the aim of evaluating the quality of the
adsorbent (PBC) and comparing it with that of phenylboronate-agarose
(PBA-30), 200 ""L with 180 U of the enzyme and 0.93 mg of total protein was
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
840 Arruda and Santana
,----------------------------------r80
120
.r-
~ • Protei.
70.r-

n
S 100 -D-PBC
.s...u '..:I

. 60BS
II( PBA 30 ... lIE- ..
0
80 Actlylty
Do
. -.-PBC 50 =-
WI
...,
:l. ~
··.·PBA 30 5
.s.a 60 4O;:J

-
~
~
e 40 ~ 30.t>
~
~

...
ii
0 20
it! 20i
<
Eo< 10

o 0
o 5 10 15 20 25 30 35 40 45 50
Injeded Volume (mL)

Fig. 10. Comparison of elution curves of prepurified soybean extracts for PBA-30
and PBe. Isocratic elution was done with 0.1 M sorbitol and 0.1 M of 0.5 M NaCl,
respectively, and a flow rate of 0.25 mL/min.

injected into the columns filled with PBA-30 and PBC, respectively, both
the same adsorbent and bed height described in the procedures above. As
shown in Fig. 10, after elution, fractions of the eluted peak had a purifica-
tion factor at a specific activity of approx 5 for both PBA-30 and PBC.
For the pool of fractions eluted, the enrichment in specific activity was 4
(12.2%) and 4 (41.7%), respectively. The isocratic elution of the enzyme in
the column of PBA-30 showed a well-defined peak. The different behavior
of PBA-30 in the process of adsorption of the soybean proteins cannot be
explained only by the difference in density of the ligands of the adsorbents,
because the matrix and spacer are different. For the same matrix and spacer
the specificity of the adsorbent with pheny1boronate depends first on the
quality of the ligand, and second on ligand density (20).
To verify the contribution of each step in the whole purification pro-
cess, total protein content and activity were observed dialyzed extract all
the prepurification (concentration) and purification steps. Figure 11 dis-
plays a flow sheet of the complete purification scheme, including the
prepurification and adsorption steps. Table 3 shows the evolution of those
variables in the prepurification steps. The purification factor in these pre-
liminary schemes reached a value of 25. The results of the adsorption
experiments applied to the prepurification steps, presented in Table 4,
indicate a further 4-fold enrichment of the enzyme's specific activity, lead-
ing to a 97-fold enrichment in the overall enzyme purification process.

Conclusion
The affinity PEC adsorbent, which was synthesized with up to 100
flmol/ mL, was compared with the commercial phenylboronate-agarose
adsorbent (PBA-30), containing 51 flmol/mL with precipitated extract of
commercial soy and dialysate with ~-amylase. Both adsorbents showed
Applied Biochemistry and Biotechnology Vol. 105-108,2003
PBC Resins for Adsorption of ~-Amylase 841

Soybean grain I__ '--_


Defated whey
_ _---'
Buffer extract
(Smith et aI., 1962)

I Soybean p-amylase I Acetone-precipitate


t
Adsorption
Ethanol-soluble
proteins
PBA30 orPBC (Ren et aI., 1993 a,b)

t
Acetone-precipitate redissolved Diafiltration
in phosphate buffer (MW cut off 30 kDa)

Fig. 11. Flow sheet of prepurification and absorption steps.

Table 3
Prepurification Steps Starting from Deffated Soybean Flour
Total Specific
Volume Protein Activity activity activity Yield Purification
Fraction (mL) (mg/mL) (AU/mL) (AU) (AU/mg) (%) (fold)
Buffer extract 12.0 15.70 134.4 1,61 8.6 100 1
Ethanol-soluble 12.0 1.08 105.0 1,26 96.7 78.1 11
proteins
Diafiltration 8.6 0.86 139.0 1,19 161.6 74.1 19
(mol wt cutoff
of 30 kDa)
Acetone precipi- 1.0 4.67 1,00 1,00 215.2 62.3 25
tate redissolved a
a Acetone precipitate = prepurified soybean extract.

Table 4
Adsorption Steps Applied to Prepurified Extract
Total Specific
Volume Protein Activity activity activity Yield Purification
Fraction (mL) (mg/mL) (AU/mL) (AU) (AU/mg) (%) (fold)
Acetone 0.20 4.67 900 180 192 100 1
precipitation
PEA-30 1 0.029 22 22 759 12.2 3.9
PBC 1 0.100 75.1 75.1 752 41.7 3.9

viability for the enzyme purification procedures. The extraction of buff-


ered alcoholic solutions was shown to be advantageous for obtaining a
narrower range of molecular masses (60 kDa). The interaction was shown
to be complex and competitive as a result of formation of an ionic pair

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


842 Arruda and Santana
between ligand and matrix. Otherwise, interactions such as hydrophobic
and hydrogen bridges, should not be disregarded. The kinetic experiments
indicated the greater affinity of PBC for the extract and less restriction of
diffusion in PBC because of wider pores. The isotherms were fitted to the
Langmuir model over the whole range of concentrations tested and
showed high values for the dissociation constants, which suggests that
competition with another molecular species for the ligand in the adsorp-
tion of soybean B-amylase does exist. The breakthrough curves for total
protein suggest a competitive adsorption mechanism with a possible
multilayer modeling attack. The adsorption steps, performed in columns
and coupled to prepuriflcation steps, resulted in a total purification factor
of 97 for the specific activity of soybean B-amylase.

Acknowledgments
We gratefully acknowledge financial support from Coordenacao de
Aperfei<;oamento de Pessoal de Nivel Superior (CAPES)/Programa
Institucional de Desenvolvimento Cientifico e Tecnol6gico (PICDT), Brazil;
and Dom Bosco Catholic University (UCDB), Campo Grande, Brazil.

References
1. Paws, c., Fraguas, L. F., and Batista-Viera, F. (1994), Chromatographia 38,232-234.
2. Brena, B. M., Paws, c., Fraguas, L. F., and Batista-Viera, F. (1996), J. Chromatogr. B 684,
217-237.
3. Pazos, C. Fraguas, L. F., and Batista-Viera, F. (1998), Chromatographia 48, 209-214.
4. Bouriots, V., Galpin, 1. J., and Dean., P. D. G. (1981), J. Chromatogr. 210,267-278.
5. Shahidi, F., ArachchiJ. K V., and Jeon, Y. J. (1999), Trends Food Sci. Technol.10, 37-51.
6. Beppu, M, M, Arruda, E. J., and Santana, C. C. (1999), Polimeros: Ciencia e Tecnologia
9,163-169.
7. Kimura, T.,Yoshida, M, Oishi, K , Ogata M., and Nakakuki, T. (1989), Agric. BioI.
Chern. 53,1843-1848.
8. Miranda, E. A. and Bergiund, K A. (1990, Biotechnol. Prog. 6,214-219.
9. Batista-Viera, F., Barragan, F. B. B., and Busto, B. L. (1988), Biotechnol. Bioeng. 31,
711-713.
10. Davis, P. R S. S. and Illuni, L. (1999), Int. J. Pharm. 187,53-65.
11. Amoncita, R, Pneto, J. R., Kuelin, G. D., and Hageman, J. R (1980),J. Chromatogr. 189,
225-231.
12. Smith, A. K, Nash, A. M., Eldridge, A. c., and Wolf, W. J. (1962), Agric. Food Chern.
10(4), 302-304.
13. Ren, H., Madison, 1. T., and Thompson, J. F. (1993), Phytochemistry 33,535-539.
14. Ren, H., Madison, J. T., and Thompson, J. F. (1993), Phytochemistry 33, 541-545.
15. Bradford, M. (1976), Anal. Biochem. 72,248-254.
16. Bernfeld, P. (1955), in Methods in Enzymology, Colowick, S. P. and Kaplan, N. 0., eds.,
Academic Press, New York, NY, pp. 145-150.
17. Langmuir,1. (1916), J. Am. Chern. Soc. 30,2263-2295.
18. Brena, B. M, Batista-Viera, F., Ryden, L., and Porath, J. (1992),J. Chromatogr. 604,109-
115.
19. El-Kak, A, Manjmi, S., and Vijayalakshmi, M (1992), J. Chromatogr. 60,29-37.
20. Amicon (1981), in Operating Instructions, Amicon Corporation, Danvers, MA, pp. 1-46.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0843/$20.00

Ester Fuels and Chemicals from Biomass

EDWIN S. OLSON,* TED R. AULlCH,


RAMESH K. SHARMA, AND RONALD C. TIMPE
Energy & Environmental Research Center, University of North Dakota,
PO Box 9018, Grand Forks, NO 58202-9018, E-mail: eolson@undeerc.org

Abstract
Bench-scale research demonstrated that using an efficient esterification
step to integrate an ethanol with a carboxylic acid fermentation stream
offers potential for producing valuable ester feedstocks and fuels. Polar organic
acids from bacterial fermentations are difficult to extract and purify, but for-
mation of the ammonium salts and their conversion to esters facilitates the
purifications. An improved esterification procedure gave high yields of
esters, and this method will lower the cost of ester production. Fuel character-
istics have been determined for a number of ester-gasoline blends with prom-
ising results for lowering Reid vapor pressure and raising octane numbers.
Index Entries: Biorefinery; esters; esterification; alcohols; ammonium.

Introduction
New opportunities for enhancing the value of agricultural products
and improving their utilization require the development of innovative
processes that are highly efficient, economically competitive, and environ-
mentally acceptable. Potential applications in improved fiber and film tech-
nologies, safer biodegradable solvents, and less polluting oxygenate
additives for fuels are ripe for commercialization, except that processing
costs for the biobased esters that are the foundation of these applications
are currently high. The development of refineries to conduct this process-
ing of renewable resources will provide rural agricultural communities
with an opportunity to diversify their economies and shift the dependence
on petroleum-based chemical, polymer, and fuel markets.

Dual Fermentation Biorefinery


The dual fermentation biorefinery (DFB) concept is a unique oppor-
tunity to integrate two types of fermentation: the well-developed yeast

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 843 Vol. 105-108,2003


844 Olson et al.

Dual-Fermentation Biorefinery

Process

Condenser

ffRC fQ,Q,47COR

Fig. 1. Schematic representation of DFB.

fermentation of sugars to ethanol and the currently developing bacterial


fermentation of sugars to organic acids. The DFB process, as illustrated in
Fig. I, produces high-value organic esters from these streams. The key to
economic feasibility of the DFB is a high-yield esterification method that
facilitates the isolation and purification of the organic carboxylate stream.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Ester Fuels and Chemicals from Biomass 845

The fermentation of sugars to low molecular weight organic acids is


well documented. Economical methods for extraction and purification of
the acids, however, still present a challenge. Whereas ethanol is relatively
easy to concentrate and purify owing to its volatility, the recovery of
organic acids from fermentation is always problematic because of the
high water solubility and high polarity of the acids. Thus, it is difficult to
extract or distill some of these acids from the aqueous fermentation broth,
and the cost of extraction and purification is significant for acids and
derivatives prepared from them.
Current Methods for Ester Production
In the traditional process in which calcium is used for neutralization
during fermentation, the calcium salt of the organic acid product is acidi-
fied with H 2S04 , Gypsum is therefore obtained as a byproduct. The acid
form is extracted from the aqueous filtrate or concentrated by evaporation.
For example, lactic acid is concentrated by multieffect evaporation to give
the low-purity product. Heating the crude acid with ethanol and H 2S04
catalyst gives ethyl lactate. The yield is equilibrium controlled, and it is
difficult to remove the water and achieve higher than 60% yields. Improve-
ments in the yields of purified lactate esters were demonstrated at the US
Department of Agriculture (USDA) Eastern Regional Research Laboratory
(1). This method passed methanol vapor (92-100°C) through the crude 82%
lactic acid solution containing H 2S04 catalyst or used a packed-column
reactor tower with the methanol vapor. The condensate from the effluent
vapors consisted of the ester, water, and excess methanol, which was dis-
tilled to recover the ester (80-90% yields) or hydrolyzed to obtain pure
lactic acid (97% yield).
Cargill patents describe a concentration process for extracting lactic
acid from the fermentation broth using a nonpolar tertiary amine after
adjusting the pH with CO2 (2). The lactate is back-extracted from the amine
phase. Owing to the high solubility in water, partitioning the lactate back
and forth between phases requires very large amounts of solvent as well
as back-extractant solution. The lactic acid is converted to the cyclic dimer
(lactide), which is purified by vacuum distillation and subsequently poly-
merized in a melt process to polylactide, a biodegradable polyester resin
with excellent fiber-forming properties. Related methods using ion-
exchange columns for extraction have also been explored at the Energy
and Environmental Research Center (EERC) and many other laboratories.
Fouling of the resins is a major issue in these methods. Husson and King
(3) reported a successful method for recovering lactate from a resin with
tertiaryamines.
An electrodialysis process was used at Argonne National Laboratories
to concentrate lactic acid and regenerate base needed in the fermentation.
The lactic acid is esterified with an ethanol in a patented pervaporation
process that uses a membrane permeable to water and ammonia (4). The
water and ammonia are removed from the separator cell by using a sweep
Applied Biochemistry and Biotechnology Vol. 105-708,2003
846 Olson et al.
gas or vacuum and returned to the fermentation. The removal of water by
the selective membrane thus drives the equilibrium and allows high yields
of ester to form without extensive separation costs. The alcohol and product
ester are separated by distillation. This process should be efficient, but there
is concern that the membranes may foul with the large amount of impurities
in the crude fermentation products. These potential fouling impurities
include various biopolymers present, including reaction bypro ducts in the
acid-catalyzed heated solution, such as Maillard and Browning reaction
byproducts of the unfermented carbohydrates and proteins.
Direct Esterification in DFB Process
In addition to providing diversified products for food, chemical, poly-
mer, and fuel markets, the DFB concept is a unique opportunity for integrat-
ing recent and previous technical advances for processing carboxylic acids
and their ester products. Purification of the acid fermentation products is
needed and without esterification is very difficult. It is logical then to make
the alcohol needed for esterification simultaneously as well as to market the
purified ester products, rather than converting them back to acids.
A version of the DFB utilizing glucose hydrolysate is shown in Fig. l.
The glucose hydrolysate is divided into two streams: one for the conven-
tional ethanol process and one for a bacterial fermentation to carboxylate
salts. Ammonia is used for precise control of the pH, and thus the ammo-
nium salt is produced. Multieffect evaporators concentrate the fermenta-
tion product salt by removal of water.
Esterification effectively couples the carboxylate fermentation stream
(as a 70% aqueous solution of the ammonium carboxylate) with an ethanol
distillate stream to form the desired ester directly in one step. This integration
solves the carboxylic acid isolation problem without having to utilize the
low-efficiency amine extraction and back-extraction process as described in
the Cargill patents or other resin-based technologies that are subject to foul-
ing problems.
Direct esterification of the ammonium salts is the key to effective pro-
cessing of the organic acid stream. The problem is how to accomplish this
step in high yields with methanol or ethanol. In flask experiments,
Filachione and coworkers (5,6) demonstrated good yields in the esterifica-
tion of ammonium salts with higher alcohols. The reaction of ammonium
lactate with butyl alcohol gave 67% yields of ester, and higher alcohols gave
higher yields. By recycling, an 83% yield of butyl lactate was obtained, and
most of the ammonia was recovered. However, no esters were formed in
reactions on ammonium carboxylates with lower alcohols.
The EERC direct esterification process used a modification of the
published USDA method for acid substrates (1). In the EERC method,
esters are formed in the specialized tubular flow reactor from the 70%
ammonium salt concentrate and excess of alcohol and are removed in a
vapor phase without the need for a pervaporation membrane as described
in the Argonne patent (4) for preparing esters. The cost of membrane
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
Ester Fuels and Chemicals from Biomass 847
replacement owing to fouling by the polymeric species in the concentrate
is avoided. Initial bench-scale experiments gave ethyl lactate yields of
71 %, based on ammonium lactate feed. This is an order of magnitude
higher than those using the unmodified esterification tower.
In our initial studies, the entire overhead from the outlet condenser
was collected in cold traps, and excess ethanol used in the esterifications
was easily removed along with water and ammonia by distillation at atmo-
spheric pressure, leaving a relatively pure ester product. Further work is
needed to optimize this product separation and recover the ammonia for
recycle to the fermentation unit.
The DFB process allows tremendous flexibility in producing a variety
of products to optimize revenue through selective commodity production
to satisfy changing market demands. The EERC direct esterification pro-
cess is versatile and able to produce lactate esters for solvents and interme-
diates for high-value polylactide fibers or other esters for fuel additives,
solvents, and chemical intermediates. The fermentable component(s) could
be glucose derived from cornstarch as in most commercial ethanol plants
in the United States, from sucrose from cane or beet sugars, or from sugars
derived from other starches or lignocellulosic feedstocks. For example,
DFBs could be established through the modification of existing beet sugar
factories coupled with a new ethanol facility or the addition of a carboxylic
acid plant to an existing ethanol plant.
The esterification process can also utilize other alcohols, such as inex-
pensive methanol produced from syngas. In this case, the process would
not require ethanol fermentation, and the methyl ester would be produced
from the organic acid fermentation product. Another option is to produce
Guerbet alcohols (I-propanol, 2-methyl-I-propanol, 2-methyl-I-butanol)
by catalytic condensation of methanol and fermentation ethanol.
Utilization of Ester
Esters of the small organic acids resulting from the fermentation of
sugars have the potential for use in many applications. A relatively small
market is flavoring agents. For example, ethyl lactate is used as a flavoring
for cheese. In addition, the esters offer great potential as biodegradable
low-toxicity solvents, chemical intermediates for the preparation of poly-
mers and plasticizers, intermediates for the preparation of pure forms of
the corresponding acids, and fuel oxygenate additives. For the solvent and
fuel additive alternatives in particular, the production cost would have to
be substantially lower than the current values.
Ethyl lactate has been used for many years for solvent cleaning in the
electronic industry. Wider application as a substitute for toxic hydrocarbon
and halocarbon solvents in cleaning and degreasing; removal (stripping) of
coatings, and manufacture of solvent-based paints, adhesives, and printing
inks is being promoted. Ethyl or methyl lactate could be highly purified and
then used to prepare polylactide, a biodegradable polymer with excellent
fiber-forming properties. Plasticizers have been prepared by acylation of
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
848 Olson et al.
lactate esters. The selling cost would have to be reduced from about $2/1b
to about $1/1b for these applications to be commercialized.
Performance of Ester Gasoline Blends
Oxygenate additives are needed for the efficient combustion and low
emission properties of petroleum-based fuels. When added to gasoline,
oxygenates such as methyl tertiary-butyl ether (MTBE) and ethanol provide
octane and help reduce engine-out (prior to catalytic conversion) emissions
of hydrocarbons and carbon monoxide. However, neither MTBE nor etha-
nol represents an ideal gasoline additive. Because of its presence and per-
sistence as a contaminant in surface water and groundwater, MTBE has the
potential to negatively impact drinking water supplies and is considered a
human health hazard (7-9). Ethanol is not generally thought to represent
a human health hazard, but when added to gasoline at concentrations of 2-
15 vol%, it raises the Reid vapor pressure (Rvp) of the resulting blend by
about 1.0-1.51b / in. 2 over that of the base gasoline. Ignoring this increase in
Rvp, often referred to as the" ethanol bump," results in increased evapora-
tive emissions from ethanol-blended vs nonethanol-blended gasoline (10).
Compensating for the ethanol bump requires removal and/ or additional
refining of high-volatility butanes to generate a reduced-Rvp base gasoline
prior to ethanol addition, which adds to the consumer cost of gasoline.
Ethanol is also miscible with water and has a higher affinity for water than
gasoline, which can lead to ethanol separation from gasoline in the pres-
ence of water. Ethanol miscibility in water is the primary reason cited for
disallowing shipment of ethanol-blended gasoline in pipelines.
As MTBE is phased out, first in California and later throughout the
United States, the demand for clean octane will increase (11). Although
ethanol could eventually meet this demand, an ideal octane supply would
not require preblending volatility adjustment and would not preclude pipe-
line shipment of finished fueL An even more ideal octane supply would
effect an Rvp reduction on addition to a base gasoline, which would enable
utilization of fuels with higher butane levels.
In 1982, Beuther and Kobylinski (12) evaluated a few esters for their
potential as blending components of gasoline. They showed that both iso-
propyl acetate and an ethyl acetate / methyl acetate mixture provided good
blending octane numbers that were comparable to alcohols and ethers but
concluded that the cost of producing esters would keep them out of the
oxygenate market. More recent studies at the EERC showed that ethyl ac-
etate provided an increase in the (R + M)/2 value in a 10% blend with
gasoline, which was a little less than the increase observed for a 10% ethanol
blend for this gasoline. This was accompanied by a decrease in Rvp of 0.6
compared to the 1.0 increase for ethanoL
To evaluate the stability of the fuel-additive blend in the presence of
moisture, the extent of partitioning the additive from the fuel into a water
phase was determined by shaking 50 mL of water with an equal volume
of the fuel-additive blend. The volume increase for the aqueous phase was

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Ester Fuels and Chemicals from Biomass 849
Table 1
Fuel Characteristics of 10% Oxygenate-Gasoline Blends
Fuel Octane, Rvp Oxygenate partition
blend (R+M)/2 (lb/in.2) to water phase (%)
Base gasoline 86.9 13.2
10% Ethanol blend 89.4 14.2 100
10% Mixed alcohol blend 87.8 12.6 60
10% Mixed ester blend 88.3 12.4 0
10% Methyl propionate 88.7 12.4 0
10% Methyl butyrate 88.9 12.3 0
10% Ethyl acetate 20
10% Ethyl propionate 88.8 11.9 0
10% Ethyl butyrate 88.4 11.6 0
aNot tested with same base gasoline.

1.0 mL for the ethyl acetate blend, corresponding to 20% partitioning of


ethyl acetate into the water phase. By contrast, ethanol partitioned 100%
into the aqueous phase. Esters of longer chains should be expected to
partition much less than 20% because of their lower water solubility. This
would be a desirable fuel characteristic, since any partitioning may be
unacceptable for fuel pipeline transport.
With the prospect of achieving a lower production cost for esters in the
DFB process, a series of ester oxygenates was evaluated recently for use as
gasoline octane providers. These oxygenates included esters made from
fermentation-derived carboxylic acids (acetic, propionic, and butyric) in
combination with methanol and ethanol. In addition, the acetate ester was
prepared using a mixture of 3-, 4-, and 5-carbon alcohols derived from
methanol and ethanol via an alcohol condensation reaction using a novel
catalyst prepared at the EERC. The mixed ester was determined by gas
chromatography to be composed of methyl acetate (1.3%), ethyl acetate
(1.3%), n-propylacetate (11.5%), isobutyl acetate (81.7%), 2-methyl-1-butyl
acetate (1.2%), and other components (1.9%).
Each oxygenate was added to gasoline at a blend ratio of 10% and
assessed for viability as a gasoline additive on the basis of octane as (RON
+ MON)/2, Rvp, and resistance to water contact-induced phase separa-
tion. Gasoline-additive blends were analyzed for Rvp, research octane
number (RON), and motor octane number (MaN) according to American
Society for Testing and Materials Methods D5191, D2699, and D2700,
respectively. The results are given in Table 1. For comparison, Table 1 also
provides data for a 10% ethanol-blended gasoline, a blend of the alcohol
mixture that was used in preparing the mixed esters, and the gasoline
used as the base fuel for all the oxygenate blends.
All the esters exhibited good cold-weather blending characteristics.
The blends all appeared as a clear solution and exhibited no phase separa-
tion at 1°F after standing for 1 wk.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


850 Olson et al.
The additive partitioning test results listed in Table 1 indicate excel-
lent resistance to water-based phase separation. Although loss of ethyl
acetate was previously determined to be 20%, loss of esters to the water
phase was negligible for all of the ester blends with larger acyl components.
The extent of aqueous-phase partitioning decreased as expected with
increasing molecular size. The observed resistance to phase separation in
the presence of water means that pipeline transport of gasoline containing
these compounds may be possible.
The octane number test results (Table 1) indicated that the esters all
provided a significant increase in the mean octane rating, with methyl
butyrate achieving a 2.0 unit increase. The mixture of acetate esters gave
the smallest increase. The octane test results are consistent with results
reported in 1982 for several other esters (12). By comparison, the mixed
alcohol blend showed a smaller increase in octane numbers than that
observed for both the esters and ethanol blends.
The results listed in Table 1 show a beneficial depression in Rvp for the
ester blends of 0.8-1.6Ib/in. 2 • As expected, the reduction Rvp of the ester
blends varied inversely with molecular weight of the ester.
Opportunities for Ester Fuel Biorefineries
Bench-scale research demonstrated that the DFB process with an effi-
cient esterification step to integrate the fermentation streams offers poten-
tial for producing valuable ester feedstocks and fuels. The conversion of
cornstarch to sugars is currently the most economically viable scenario, and
the ethanol production operations are well defined; the economical produc-
tion of the organic acids and esters, the key technical focus of the current
research, still requires optimization and economic evaluation of the fermen-
tation, evaporation, esterification, and product purification unit operations.
The fact that the esters provide Rvp reduction along with an octane
boost could translate into an opportunity for oil refiners to reduce produc-
tion costs associated with stripping butanes from gasoline. Currently, high-
volatility butanes are removed from gasolines destined for sale in regions
with volatility restrictions. Because of the Rvp reduction that the esters
provide, the addition of ethyl propionate or ethyl butyrate to a gasoline
could offset all or a portion of the gasoline volatility contributed by butanes,
thereby reducing or eliminating the need for butane removal.
Although the Rvp of the inexpensive mixed alcohol blend was low,
conversion of this mixture to the ester resulted in a greater enhancement of
the octane numbers and provided a fuel with no loss to the aqueous phase.
Thus, it is clear that greater benefits in oxygenate blending are possible
with the esters, compared with higher alcohols.
This research as well as that reported earlier showed that esters are
excellent fuel additives, providing high-octane blends and low volatilities.
Emission characteristics are expected to be comparable to other oxygenate
additives. While production costs have heretofore prohibited their use in
fuels, DFB offers the opportunity to lower these costs.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Ester Fuels and Chemicals from Biomass 851

Acknowledgment
This work was supported by the Cooperative State Research, Educa-
tion, and Extension Service, USDA, under agreement no. 00-38819-8989.

References
1. Filachione, E. M. and Fisher, C. H. (1946), Ind. Eng. Chern. 38,228.
2. Baniel, A. M., Eyal, A. M., Mizrahi, J., Hazan, B., Fisher, R R, Kolstad, J. J., and
Stewart, B. F. (2000), US patent no. 6087532.
3. Husson, S. M. and King, C. J. (1998), Eng. Chern. Res. 37,2996-3005.
4. Datta, Rand Tsai, S.-P. (1998), US patent no. 5723639.
5. Filachione, E. M. and Costello, E. J. (1952), Ind. Eng. Chern. 44, 2189.
6. Filachione, E. M., Costello, E. J., and Fisher, C. H. (1951), J. Am. Chern. Soc. 73,5265.
7. Delzer, G. c., Zogoski, J. S., Lopes, T. J., and Bosshart, R 1. (1996), Water Resources
Investigations Report No. 96-4145, US Geological Survey, Rapid City, SD.
8. Squillace, P. J., Zogorski, J. S., Wilber, W. G., and Price, C. V. (1996), Environ. Sci.
Technol. 37,394-397.
9. Health Effects Institute (1996), Special report, Health Effects Institute Oxygenates
Evaluation Committee, Cambridge, MA.
10. Chang, T. (2001), Oil Gas J.
11. Meister, J. M.; Black, S. M., et al. (2000), Hydrocarbon Processing.
12. Beuther, H. and Kobylinski, T. P. (1982), in Proceedings of the Symposium on Chemistry
of Oxygenates in Fuels, American Chemical Society Kansas City Meeting.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0853/$20.00

Purification of Glucose-6-Phosphate
Dehydrogenase from Baker's Yeast
in Aqueous Two-Phase Systems
with Free Triazine Dyes as Affinity Ligands

YAN XU,l MICHELE VITOLO/


CRISTINA NORTHFLEET DE ALBUQUERQUE, 2
AND ADALBERTO PESSOA JR. *,2

1Department of Chemical Engineering, Hebei Institute of Technology,


Tangshan, Hebel, 063009, P.R. China; and
2Biochemical and Pharmaceutical Technology Department, FCF/USP,
PO Box 66083, CEP 05315-970, Sao Paulo-SP, Brazil,
E-mail: pessoajr@usp.br

Abstract
To improve the selectivity of glucose-6-phosphate dehydrogenase
(G6PDH) extraction by an aqueous two-phase system, a simple and inexpen-
sive affinity aqueous two-phase system using unbound reactive triazine dyes
as ligands was introduced. In a polyethylene glycol (PEG)/hydroxypropyl
starch (PES) system, the unbound free triazine dyes, Cibacron Blue F3GA
and Procion Red HE3B, partitioned unevenly in the top PEG-rich phase and
thus showed an affinity effect on G6PDH, but no influence on hexokinase.
The various parameters investigated were pH of the system, buffers, molecu-
lar weight of PEG, and ligand type and concentration. A two-step affinity
extraction process was established for the purification of G6PDH from
baker's yeast. The total yield of G6PDH was 66.9% and purification factor
was 2.35.
Index Entries: Aqueous two-phase systems; affinity partitioning; triazine
dyes; glucose-6-phosphate dehydrogenase; baker's yeast.

Introduction
Several procedures for the rapid purification of enzymes from yeast
and other microorganisms have been developed, mainly to avoid undesir-
able proteolytic degradation of the target enzymes in the course of their
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 853 Vol. 105-108, 2003


854 Xu et al.

isolation. In particular, the introduction of preparative affinity chromatog-


raphy (using natural ligands or artificial dye ligands) has led to significant
improvement in the efficiency and economy of time for these procedures
(1). However, when large-scale preparations are required, various micro-
biological and technological problems arise and often limit the use of
water-insoluble hydrophilic or hydrophobic resins (2). Aqueous two-phase
systems are finding increasing application in the recovery of proteins (3,4).
Moreover, the development of affinity partitioning for the purification of
proteins combined the advantages of biospecific adsorption of selective
compounds and the ability to work in a homogeneous phase (5,6).
Affinity partitioning as currently practiced requires the ligand to be
covalently attached to one of the phase-forming polymer components,
thereby causing the ligand/ polymer to partition predominantly into one of
the phases. Although triazine dyes covalently coupled to polyethylene
glycol (PEG) can be produced on a large scale, the process is complicated
and requires a chromatographic step and several organic solvent extrac-
tions (7). To simplify the affinity partitioning, Giuliano (8) used the free
dyes, uncoupled to the phase-forming polymers, as affinity ligands for the
partitioning of lysozyme in a polyvinylpyrrolidone / maltodextrin aqueous
two-phase system. In the PEG/phosphate system, it was reported that free
triazine dyes, partitioned predominantly into polymer phase, showed the
affinity effect on some dehydrogenases and kinases (9,10). Lin et aL (11,12)
also used free triazine dyes in affinity extraction of lactate dehydrogenase
in a PEG /hydroxypropyl starch (PES) aqueous two-phase system.
Glucose-6-phosphate dehydrogenase (G6PDH), the first enzyme in
the pentose phosphate pathway, is widely distributed in nature. It is a part
of the antioxidant enzymatic system that has an important role in tissue
protection against the destructive action of oxygen-free radicals (13). With
hexokinase, G6PDH presents great interest as analytical reagents for the
measurement of creatin-kinase activity, adenosine triphosphatase (ATP),
and hexose concentrations (14).
In this study, we investigated the partition of G6PDH with the free
Procion Red HE3B and Cibacron Blue F3GA, unbound to phase-forming
polymer, as affinity ligands in PEG/PES and PEG/phosphate aqueous
two-phase systems. The effects of various parameters, such as the pH of the
system, molecular weight of PEG, buffer solution, and concentration of
ligands, on partition coefficients of the enzymes were systematically stud-
ied. A two-step method for the purification of G6PDH from baker's yeast
was proposed for the first time.

Materials and Methods


Chemicals
PEG3000, PEG4000, and PEG6000 were purchased from LABSYNTH
Produtos para Laboratorios Uda. Hydroxypropyl starch (Reppal PESIOO)
was a kind gift of Professor J. A. Teixeira (Departmento de Engenharia
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of G6POH by Affinity Extraction 855
Biologica, Universidade do Minho, Portugal). Cibacron Blue F3GA and
Procion Red HE3B were purchased from Sigma. G6PDH, hexokinase,
~-nicotinamide adenine dinucleotide phosphate (NADP), ATP, D-glucose-
6-phosphate, and D-glucose were also from Sigma. All other reagents were
of analytical grade.
Commercial pressed baker's yeast (Saccharomyces cerevisiae) was used.
The cells were suspended in 50 mM Tris-HCl buffer (pH 7.5) containing
5 mMMgCl 2, 10 mM ~-mercaptoethanol, and 2 mM phenylmethylsulfonyl
fluoride.
Aqueous Two-Phase Systems
The systems were prepared from stock solutions: 50% (w /w) PEG,
30% (w /w) PES, 40% (w /w) ammonium sulfate, and 40% (w /w) phos-
phate (mixture of N aH 2PO4 and K2HPO4). The total weight was 3.0 g for the
polymer/polymer (PEG/PES) system, and 4.0 g for the polymer/salt sys-
tem. All concentrations were given in weight percentage except as indi-
cated otherwise.
Cell Disruption
A volume of 100 mL of glass beads (0.5 mm diameter) and 100 mL of
Tris-HCI cell suspension were introduced into the 250 mL chamber of a
vertical cell disrupter (15). Cell disintegration was carried out at 5°C under
agitation of 2300 rpm, and then the disrupted cell suspension was centri-
fuged at 18,000g for 30 min at 5°C. The supernatant was used in further
experiments.
Partitioning Experiments
The enzymes or the supernatant was added to the systems. After 15
min of inversion mixing to ensure partition equilibrium, phase separation
was accomplished by centrifuging at 2880g for 5 min, and samples were
withdrawn for analysis. The experiments were conducted at room tem-
perature for pure enzymes and at 4°C for disrupted cell supernatant.
Preparation of Dye Derivatives
According to Giuliano (8), the amino derivatives of chlorotriazine
dyes (Cibacron Blue F3GA and Procion Red HE3B) were prepared as
follows: Four grams of the dye was first dissolved in 50 mL of methanol
followed by the addition of 20 mL of concentrated ammonium hydroxide.
The solution was refluxed for 40 min. Solvents were removed via rotary
evaporation at reduced pressure (vacuum of 450 mmHg). The remaining
solid was washed with 300 mL of acetone and dried for use. Acetone was
recovered by rotary evaporation.
Analyses
G6PDH and hexokinase activity was measured spectrophotometri-
cally at 30°C by following the rate of NADP+ reduction at 340 nm in a
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
856 Xu et al.
coupled enzyme assay system (16). For both enzymes, one unit of enzyme
activity was defined as the amount of enzyme required to form 1.0 ~mol of
NADPH/min with the substrate in excess. The partition coefficient of the
enzymes (Ke) was defined as the ratio of the enzyme activities in the top and
bottom phases.
The concentration of protein was determined by Bradford's (17)
method using bovine serum albumin as standard. The concentration of the
dye and the derivatives was determined photometrically at 615 nm for
Cibacron Blue F3GA and its derivative, and 515 nm for Procion Red HE3B
and its derivative. The partition coefficient of the dyes (KL ) was defined as
the ratio of the concentrations in the top and bottom phases.
The binding of a ligand to an enzyme was evaluated by inhibition
constant, Ki, which is the dissociation constant of the enzyme-ligand com-
plex. This value was determined by Lineweaver-Burk plot. To calculate Ki,
it was assumed that the inhibition of the dyes to G6PDH and hexokinase is
competitive; thus, we obtained the linear relationship of 1/A - [1], in which
A is the activity of enzymes and [1] is the concentration of dyes. With the
intercept and slope of 1/A - [1], the Ki was calculated.
Enzymatic assay was carried out in the corresponding equilibrated
top and bottom phases.

Results and Discussions


Partitioning of Dyes in Aqueous Two-Phase System
Usually, the ligand was covalently coupled to the target phase poly-
mer in the affinity aqueous two-phase system, so as to confine the ligand in
one phase. To use uncoupled free dye molecules as ligand, their partition
behaviors in the system need to be estimated first.
Partitioning in PEG/Salt Systems
In PEG/salt aqueous two-phase systems, the dyes partitioned pre-
dominantly in the top PEG phase. The effect of pH on the partitioning of
Cibacron Blue F3GA and Procion Red HE3B in the PEG / phosphate system
is presented in Table 1. With the increase in pH, the partition coefficient of
both dyes increased. This may be caused by the repulsion of negatively
charged phosphate, because of more negative charges on the dye molecules
in the higher pH range. The effect of molecular weight of PEG on the par-
tition coefficient of Cibacron Blue F3GA and Procion Red HE3B in the PEG /
ammonium sulfate system is presented in Table 2. With the increase in
molecular weight of PEG, the partition coefficient of both dyes increased,
which is different from the behavior of biopolymers (3).
Partitioning in PEG/PES System
In the PEG /PES system, although the partition coefficients are smaller
than in the PEG/salt system, most of the dyes still partitioned in the top
PEG phase, i.e., the base for the affinity partitioning of the enzymes. For
affinity partitioning, the affinity ligands should have uneven distribution
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of G6POH by Affinity Extraction 857

Table 1
Effect of pH on Partitioning of Dyes
in PEG3000 (12.5% [w fwD/Phosphate (10% [w /wD System

pH Cibacron Blue F3GA Procion Red HE3B

3.45 7.3 6.3


5.97 319 310
7.48 707 >1000
9.70 >1000 >1000

Table 2
Effect of Molecular Weight of PEG
on Partitioning of Dyes in PEG3000 (12.5% [w /wD/
Ammonium Sulfate (10% [w/wD Systems

Molecular
wt (Daltons) Cibacron Blue F3GA Procion Red HE3B

3000 16.5 37.8


4000 27.6 133
6000 227 741

16

12

...J
~ 8

0
2 4 6 8
pH

Fig. 1. Effect of pH on partitioning of Cibacron Blue F3GA (1) and Procion Red HE3B
(2) in PEG3000 (8.3% [w /w])/PESlOO (15% [w /w]) system: phosphate buffer (2.67%
[w/w]).

in the aqueous two-phase system. If the affinity ligands partition evenly in


the two phase, no affinity effect will present.
Figure 1 shows the effect of pH on the partitioning of Cibacron Blue
F3GA and Procion Red HE3B in the PEG IPES system. With the increase in
pH, the partition coefficients of both dyes increased, which may also be
attributed to the more negative charges on dye molecules in higher pH
condition. Most ofthe buffer phosphate partitioned in the bottom PES phase.
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
858 Xu et al.
7~------------------~

~
:~
3 --+------~-
2
1
0+---~----~----~--_4

o 5 10 15 20
C (mM)

Fig. 2. Effect ofTris-HCl buffer on partitioning of blue dye in PEG6000 (8.3% [w /wD/
PESI00 (15% [w /wD system: pH 7.5.

Although the higher partition coefficient was needed for affinity par-
titioning, the pH should not exceed 7.5 because of the instability of the
enzymes. The influence of Tris-HCl buffer (pH 7.5) on the partitioning of
Cibacron Blue F3GA in the PEG/PES system was investigated (Fig. 2). At
the beginning, the increase in Tris-HCl concentration decreased the parti-
tion coefficient of the blue dye. A further increase had almost no effect on
the blue dye partitioning.

Effect of Molecular Weight of PEG on Partitioning of Enzymes


The effect of molecular weight of PEG on the partition coefficient of
G6PDH and hexokinase activity is shown in Fig. 3. An increase in molecular
weight of PEG brought about a reduction in partition coefficient of G6PDH
and hexokinase. This may be caused by the excluded volume effect of PEG
(3,9), and the different tie-line length of the systems.
Table 3 shows the effect of molecular weight of PEG on the partition
coefficient of G6PDH in the PEG/PESI00 system with and without blue
dye ligand. In the PEG6000 system, the partition coefficient was enhanced
from <1 without ligand to >1 with the blue dye ligand.
Effect of Buffer Solution on Partitioning of Enzyme
The effect of Tris-HCl concentration on the partitioning of G6PDH in
the PEG6000/PESI00 system with and without blue dye ligand was esti-
mated (Fig. 4). Generally, most negatively charged proteins partitioned to
the top PEG phase in the PEG/PES aqueous two-phase system. In the
presence of Tris-HCI buffer, the proteins would be transferred to the bot-
tom PES phase. With proper ligand, the G6PDH could be transferred to the
top phase again by the selective affinity effect, and the other proteins might
stay in the bottom phase. As can be seen, both partition coefficients, with
and without ligand, decreased with the increase in Tris-HCI concentra-
tion. However, at a concentration of about 9 mM, the G6PDH was trans-
ferred from the bottom phase (Ke < 1) to the top phase (Ke > 1) by the blue
dye ligand.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of G6POH by Affinity Extraction 859
2.5 ~------------------,

2.0

1.5

1.0

0.5

o +----.---.---r---,---~
2,000 3,000 4,000 5,000 6,000 7,000

Molecular weight (Oa)

Fig. 3. Effect of molecular weight of PEG on partition coefficient of enzymes in


PEG (8.3% [w /wD/PESI00 (15% [w /wD system: phosphate (2.67% [w fwD, pH 7.3.
HK, hexokinase.

,
6
5
4
~ 3
2
1
0
: :
0 5 10 15
C (mM)

Fig. 4. Effect of Tris-HCl buffer on the partition coefficient of G6PDH in PEG6000


(8.3% [w /wD/PESI00 (15% [w /wD system: pH 7.5, without (1) and with (2) 0.067%
blue dye.

Table 3
Effect of Molecular Weight of PEG on Partition Coefficient
of G6PDH in PEG/PESI00 Systema
Molecular Ke
wt (Daltons) No dye Cibacron Blue F3GA <pb

3000 4.46 4.96 1.11


6000 0.73 1.59 2.18
10,000 0.2 0.34 1.7
apEG (8.3% [w /wD/PESI00 (15.0% [w /wD system; 1.33% phosphate
(buffer), 0.033% blue dye, pH 7.3.
bRatio of partition coefficient in the presence to that in the absence
of dye.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


860 Xu et al.
2.0

1.6

1.2
Q)
:x::
0.8

0.4

0
0 0.02 0.04 0.06 0.08

C (%)

Fig. 5. Effect of dye concentration on the partitioning of G6PDH in PEG6000 (8.3%


[w fw])fPESlOO(15% [w fw]) system: 9mMTris-HCl, pH 7.5; (1) Cibacron BlueF3GA;
(2) Procion Red HE3B.

Effect of Ligand Type and Concentration


on the Partitioning of Enzyme
Figure 5 shows the effect of Procion Red HE3B and Cibacron Blue
F3GA concentration on the partitioning of G6PDH in the PEG6000/PESIOO
system. The red dye had a stronger affinity effect on the partitioning of
G6PDH than the blue one. Initially, when the ligand concentration was low,
increasing the dye concentration made more sites available for enzyme
binding and the partition coefficients increased rapidly. However, when
the dye concentration reached a certain quantity for a given enzyme con-
centration, a further increase in dye concentration had no effect on the
partition coefficients, which is similar to what has been observed when
dye-PEG (18) or dye-dextran (19) was used. A further increase in dye con-
centration resulted in a decrease in G6PDH partition coefficient and may be
induced by a static electric repulsive effect between the same negatively
charged dye ligands and the enzyme.
The triazine dyes may form covalent linkage with the proteins. To
eliminate the possibility of covalent modification of hexokinase or G6PDH,
the reactive chlorine atom of the dyes could be removed by hydrolysis
(20,21). According to Giuliano (8), Cibacron Blue F3GA and Procion Red
HE3B were converted to their amino derivatives to prevent the formation
of dye/protein covalent linkage. The affinity effect of triazine dye ligands
and their ammonia derivatives on the partitioning of G6PDH in the PEG
6000/PESIOO system is summarized in Table 4. The ligands could enhance
the partition coefficient of G6PDH, and the effect of red dye was stronger
than that of blue dye.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of G6PDH by Affinity Extraction 861
2.5

GBPOH
2.0

1.5

1.0

HK

0.5

0
0 0.05 0.10 0.15
C(%)

Fig. 6. Effect of Procion Red HE3B concentration on the extraction of G6PDH from
baker's yeast in PEG3000 (8.3% [w fwDfPESI00 (15% [w fwD system: 9 mM Tris-HCl,
pH 7.5. HK, hexokinase.

Table 4
Effect of Triazine-Reactive Dyes on Partitioning of G6PDH"

Dye ligand KL Ke <ph

No dye 0.54
Red 8.35 1.64 3.0
Red derivative 8.34 1.60 3.0
Blue 3.90 1.42 2.6
Blue derivative 3.93 1.44 2.7
'PEG6000 (8.3% [w /wD/PESI00 (15% [w /wD system; 9 mMTris-HCl,
pH 7.5, 0.033% dyes.
bRatio of partition coefficient in the presence of dye to that in the
absence of dye.

Purification of G6PDH from Baker's Yeast


Partitioning of Enzymes from Baker's Yeast
At 4°C, the PEG6000 system became too viscous to handle so the
PEG3000 system was chosen to investigate the partitioning behavior of the
enzymes in disrupted cell supernatant. Figure 6 indicates that the effect of
Procion Red HE3B concentration on the partitioning of G6PDH and hexoki-
nase in the PEG3000 jPESIOO system. The red dye ligand enhanced the par-
tition coefficient of G6PDH and showed no affinity effect on hexokinase.
The shape of the curve for G6PDH is similar to that of pure enzyme (Fig. 5),
but one point should be noted: the concentration of dye ligand is higher for
the affinity partitioning of G6PDH. In disrupted cell supernatant, there are
many other enzymes that may bind the red dye ligand, so higher ligand
concentration is needed for the affinity partitioning of G6PDH.
Applied Biochemistry and Biotechnology Vol. 705-708,2003
862 Xu et al.

Table 5
Effect of Triazine-Reactive Dyes on Partitioning of Enzymesa

Ke <ph

Dye ligand KL G6PDH Hexokinase G6PDH Hexokinase


No dye 0.017 0.68
Red 6.95 1.9 0.63 112 0.93
Red derivative 7.02 1.9 0.65 112 0.96
Blue 2.84 1.4 0.97 82 1.40
Blue derivative 2.85 1.5 1.00 88 1.50
apEG3000(8.3% [w fw])fPES100 (15% [w fwD system;9mMTris-HCl, pH 7.5,0.10% dyes.
bRatio of partition coefficient in the presence to that in the absence of dye.

The affinity effect of triazine dye ligands and their ammonia deriva-
tives on the partitioning of G6PDH and hexokinase in the PEG3000 /PESI00
system is summarized in Table 5. The ligands could enhance the partition
coefficient of G6PDH greatly. Without the dye ligand, the G6PDH was
concentrated in the bottom phase (Ke =0.017). With the red dye ligand, most
ofthe G6PDH partitioned into the top phase (Ke =1.9), but for pure G6PDH,
the partition coefficient changed from 0.54 to 1.64 by the red dye ligand
(Table 4).
To elucidate the mechanism (different effect of different dye ligands-
blue and red) of affinity aqueous two-phase partitioning of proteins
(G6PDH and hexokinase), Flanagan and Barondes (22) had proposed a
thermodynamic model (Eq. 1). This model describes the distribution equi-
librium of proteins in the affinity aqueous two-phase systems. It relates the
partition coefficient of proteins in the presence of affinity ligands with the
partition coefficient in the absence of affinity ligands, and partition coeffi-
cient of ligands and the dissociation constants of ligands with protein
molecules in the top and bottom phases. The equation was deduced from
Gibbs' free energy in equilibrium state:
K = Ko (KLKibl Kitt (1)
in which K and Ko are the partition coefficients of proteins in the presence
and absence of the ligand, respectively; KL is the partition coefficient of the
ligand polymer; Kit and Kib are the dissociation constants of the ligand with
protein molecule in the top and bottom phases, respectively; and a is the
number of ligand polymer molecules bound per protein molecule.
The binding strength of G6PDH to the dye ligands was similar (similar
K i) in the top and bottom phases (Table 6), but most of the dye partitioned
in the top PEG phase, so G6PDH could be brought from the bottom to the
top phase by the ligand. The KL of red dye was higher than that of blue dye,
and therefore the affinity effect was stronger. The binding strength of hex-
okinase to blue dye ligand was weaker than that of G6PDH (larger Ki value),
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Purification of G6POH by Affinity Extraction 863

Table 6
Inhibition Constants of G6PDH and Hexokinase
in Disrupted Cell Supernatant by Triazine Dyes"
K; (104 mM)
Cibacron Blue F3GA Procion Red HE3B
In presence In presence In presence In presence
Enzyme of top phase of bottom phase of top phase of bottom phase
G6PDH 5.22 4.56 3.99 1.41
Hexakinase 193 236 71.3 3.20
"In top or bottom phase ofPEG3000 (8.3% [w/w])/ PESlOO (15% [w/w]) system; 9mM
Iris-HCl, pH 7.5.

Table 7
Partitioning of G6PDH in PEG3000 (12.5% [w fwDf
Phosphate (10% [w fwD System"
Dye
Cibacron Blue F3GA 0.0043
Procion Red HE3B 0.0040
'pH 7.48; 0.05% dyes.

so the affinity effect was weakened. For red dye ligand, the binding strength
of hexokinase in the bottom phase was stronger than that in the top phase,
and no affinity effect was observed.
Two-Step Method for Purification of G6PDH from Baker's Yeast
Table 7 lists the partition coefficients of G6PDH in the PEG3000 / phos-
phate system in the presence of the dye ligands. Although the dyes were
concentrated in the PEG-rich top phase (Table 1), the enzyme stayed in the
salt-rich bottom phase. The dyes and the G6PDH could be separated in
the PEG/phosphate system, indicating that no covalent linkage formed in
the experimental condition.
To achieve purification of G6PDH from baker's yeast, a two-step pro-
cedure based on the aforementioned results was developed (Fig. 7). The
experiment was performed in a separatory funnel and the system was
enlarged to 15 g.
The disrupted cell supernatant was added into the PEG3000/PESlOO
system (first step). G6PDH was extracted into the top PEG phase by Procion
Red HE3B ligand. After phase separation, the top PEG phase was combined
with phosphate solution, forming a PEG/phosphate two-phase system
(second step). The enzyme was recovered in the phosphate-rich bottom
phase. After the "two-step" extraction, the total recovery of G6PDH was
66.9% with a purification factor of 2.35.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
864 Xu et al.
Cell disruption

B
Red dye supernatant Phosphate

~
PEG
~

PES Ph

System 1 System 2

Fig. 7. Diagram of two-step method for purification of G6PDH from baker's yeast:
System 1: PEG3000 (8.3% [w /wD/PESI00 (15% [w /wD, 9 mM Tris-Hel, pH 75; sys-
tem 2: PEG3000 (11% [w fwD/phosphate (18% [w /wD, pH 7.3.

Conclusions
The partitioning of G6PDH and hexokinase in PEG/PES and PEG/
phosphate aqueous two-phase systems was investigated with free triazine
dyes, Cibacron Blue F3GA and Procion Red HE3B, as their affinity ligands.
The dyes, not bound to phase-forming polymers, were concentrated
in top PEG phase in PEG/phosphate and PEG/ammonium sulfate sys-
tems. In the PEG/PES system, although the partition coefficients were
smaller than in the PEG / salt system, most of the dyes still partitioned in the
top PEG phase, which is the fundamental requirement for the affinity par-
titioning of the enzymes.
In the PEG/phosphate system, the dyes and enzyme were concen-
trated in the top PEG and bottom PES phase, respectively. In the PEG /PES
system, Procion Red HE3B changed the partition coefficient of pure
G6PDH from 0.54 to 1.64 with Tris-HCl buffer. For G6PDH in disrupted
cell supernatant, the partition coefficient increased from 0.017 to 1.9 with
the red ligand. The enzyme was transferred from the PES-rich bottom
phase to the PEG-rich top phase. To purify G6PDH from baker's yeast, a
two-step procedure was proposed. The total yield was 66.9% with a puri-
fication factor of 2.35. Thus, a simple, effective affinity partitioning method,
with the free dye as ligand for the purification of G6PDH from baker's
yeast was developed.

Acknowledgments
We thank Luis Fernando Peffi Ferreira and Luiz Carlos Martins das
Neves for their assistance in some of the experiments. This work was sup-
ported by Fundac;ao de Amparo a Pesquisa do Estado de Sao Paulo, Brazil.

References
1. Dean, P. D. G. and Watson, D. H. (1979), J. Chromatogr. 165,301-319.
2. Kopperschlager, G. and Johansson, G. (1982), Anal. Biochem. 124, 117-124.
3. Albertsson, P.-A. (1986), Partition of Cell Particles and Macromolecules, 3rd ed., John
Wiley, New York, NY.

Applied Biochemistry and Biotechnology Vol. 705-708, 2003


Purification of G6POH by Affinity Extraction 865
4. Walter, H. D., Brooks, E., and Fisher, D. (1985), Partitioning in Aqueous Two-Phase
Systems, Academic, New York, NY.
5. Shanbhag, V. P. and Johansson, G. (1974), Biochem. Biophys. Res. Commun. 61,
1141-1146.
6. Hubert, P., Dellacherie, E., Neel, J., and Baulieu, E. E. (1976), FEBS Lett. 65, 169-174.
7. Johansson, G. and Joelsson, M. (1985), Biotechnol. Bioeng. 27,621-625.
S. Giuliano, K. A. (1991), Anal. Biochem. 197,333-339.
9. Wang, W. H., Kuboi, R., and Komasawa, I. (1992), Chem. Eng. Sci. 47,113-121.
10. Bhide,A. A., Patel, R. M.,Joshi,J. B., andPangarkar, V. G. (1995), Separat. Sci. Techno!.
30, 2989-3000.
11. Lin, D. Q., Zhu, Z. Q., and Mei, L. H. (1996), Biotechnol. Techniques 10, 41-46.
12. Lin, D. Q., Zhu, Z. Q., and Mei, L. H. (1998), Separat. Sci. Technol. 33,1937-1953.
13. McElroy, M. c.,Postle, A. D., and Kelly, F. J. (1992), Biochim. Biophys. Acta 1117, 153-158.
14. Abrahao-Neto, J. P., Vitolo, M., and Infanti, M. (1996), Appl. Biochem. Biotechnol. 57/58,
407.
15. Ricci-Silva, M. E., Vitolo, M., and Abrahao-Neto, J. (2000), Process Biochem. 35,831-835.
16. Bergmeyer, H. (1974), Methods of Enzymatic Analysis, 2nd ed., Academic, New York,
NY.
17. Bradford, M. M. (1976), Anal. Biochem. 72,248-254.
18. Johansson, G. and Joelsson, M. (1986), Appl. Biochem. Biotechnol. 13,15.
19. Johansson, G. and Joelsson, M. (1987), J. Chromatogr. 393,195-208.
20. Moe, J. G. and Piskiewicz, D. (1979), Biochemistry 18, 2810-2814.
21. Farmer, E. E. and Easterby, J. S. (1982), Anal. Biochem. 123,373-377.
22. Flanagan, S. D. and Barondes, S. H. (1975), J. Bioi. Chem. 250, 1484-1489.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0867/$20.00

Continuous Clavulanic Acid


Adsorption Process

RENATA M. R. G. ALMEIDA,* MARLEI BARBOZA,


AND CARLOS O. HOKKA
Chemical Engineering Department, Federal University of Sao Carlos,
Via Washington Luiz, Km 235, 13565-905,
Sao Carlos, SP, Brazil, E-mail: prmrg@iris.ufscar.br

Abstract
Adsorption kinetics and equilibrium data of clavulanic acid, a ~-lactam
antibiotic, on ion-exchange resin Amberlite IRA 400 were utilized to carry
out the modeling and simulation of a continuous adsorption process. These
simulations allowed the estimation of yield, concentration, and purification
factors of the process utilizing the product final concentration. Experimental
runs of this process were carried out using the conditions pointed out by
simulation studies. Comparison of the experimental results and those calcu-
lated by the proposed model showed that the model could describe very well
the main features of the continuous process.
Index Entries: Clavulanic acid; continuous adsorption; purification process.

Introduction
The inhibitory activity of clavulanic acid (CA) against p-Iactamases,
enzymes that catalyze the p-Iactam ring hydrolysis reaction, have been
detected in cultures of the bacteria Streptomyces clavuligerus. The discovery
of this p-Iactam antibiotic was first reported in 1976 by Beecham (1). Nowa-
days the combination of CA with amoxicillin is the most successful example
of the use of a ~-lactam antibiotic sensitive to ~-lactamase together with an
inhibitor of these enzymes.
Primary extraction of CA from the clarified broth can be done either
by extraction from acidified broth into a water-immiscible organic solvent
or by adsorption in an anion-exchange resin followed by elution from the
resin with an aqueous salt solution (2). However, this antibiotic has no
strong hydrophobic groups and presents high degradation rates, which
leads to low recovery yields during the purification process. This antibiotic

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 867 Vol. 105-108, 2003
868 Almeida et al.
is highly unstable at a pH <5.0 and >7.0 and the degradation rate increases
remarkably with the increase temperature (3,4).
Continuous adsorption with adsorbent recycle appears to be an alter-
native process for purification of CA. The continuous process can reduce
the mass transfer resistances that have to be dealt with in fixed-bed adsorp-
tion. This process can minimize product degradation by reducing the con-
tact time required in batch operations with continuous withdrawal of the
antibiotic. In addition, since it takes place in well-mixed tanks, temperature
and pH gradients can be avoided, and, consequently, degradation is much
lower than in conventional columns. Futhermore, the continuous process
reduces the expenses with adsorbent because lesser amount of the resin is
needed (5).
Several investigators have studies similar continuous adsorption pro-
cesses dealing with separation of proteins, enzymes (5-7), and antibiotics
(8). Barboza et al. (8) proposed a mathematical model describing a continu-
ous adsorption process for purification of cephalosporin C on a hydropho-
bic resin.
In the present work, adsorption kinetics and equilibrium data of CA
on the ion-exchange resin Amberlite IRA 400 in the chloride cycle were
utilized to carry out the modeling and simulation of a continuous adsorp-
tion process. The yield, concentration, and purification factors of the pro-
cess were estimated utilizing the product final concentration obtained by
simulations. Experimental runs of the continuous adsorption process were
carried out using the simulation conditions, and the experimental results
were compared with those calculated by the proposed model.

Materials and Methods


Adsorbate
CA was obtained from fermentation broth produced by S. clavuligerus
ATCC 27064. Before the adsorption process, the fermentation broth was
pretreated as follows: The broth was centrifuged to obtain a clear solution
free of cells. Then the solution pH was brought down (3.5-4.0) to precipitate
the soluble solids and proteins. Finally, the solution containing the product
was centrifuged again and filtered through analytical paper filters to elimi-
nate suspended impurities.
Preparation of Adsorbent
The stationary phase was the anion-exchanger Amberlite IRA 400,
kindly supplied by Rohm and Haas. The resin was pretreated with 10%
(w Iv) NaCl and then washed several times with deionized water to elimi-
nate the excess ions. NaCI (2%) was used as the eluent throughout.
Apparatus and Process Start-up
The adsorption reactors were stirred mechanically and considered to
be perfectly mixed tank reactors. The process start-up was carried out by
switching the pumps on. The reactor feed started with CA solution in the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Continuous CA Adsorption Process 869
Feed Desorbing

~~l F! CEol F2
Frel qlCEI
kl k3
---+ Cn
---+
~ 2 FrC2 q2 CE2
Fl CEI Cr2 F2 CE2
Ct Cn C2 Cr2
Waste Product

Fig. 1. Schematic flow sheet of the continuous processes for CA purification.

first stage and 2% NaCl in the second. The temperature in the first reactor
was controlled at lOOC and in the second stage at 30°C. The outflowing
solutions of each reactor were independent during the runs. The concentra-
tions of CA with time were obtained by analyzing the samples in the exit
stream of each stage.
Analytical Methods
The clavulanic acid concentrations were determined by high-perfor-
mance liquid chromatography (HPLC), as described by Foulstone and
Reading (9) by reaction with imidazole solution, pH 6.8. The HPLC equip-
ment was operated at 28°C, with a flow rate of 2.5 mL/min. The mobile
phase was composed of a 0.1 M KH2P04 buffer solution containing 6% of
methanol and phosphoric acid, which brings the pH to 3.2. Contaminants
(CT) concentration were determined by spectrophotometer at 280 nm.
Mathematical Modelling
The basic principles of the theoretical model, previously described by
Rodrigues et a1. (7) for enzyme purification, were used. In the present work,
the model incorporates external liquid film mass transfer coefficient, pore
diffusion, and kinetic equations for adsorption and desorption in each stage.
It also accounts for the presence of inert contaminants in the CA feed stream.
These contaminants can be considered as usual fermentation broth compo-
nents such as sugars, pigments, amino acids, and proteins in solution.
The mathematical model of the process was obtained essentially by
accomplishing a mass balance in each reactor including the adsorption/
desorption equations. A schematic view of this process is illustrated in
Fig. 1. The process operates as follows: The sample is fed continuously to
the first reactor, the adsorbing stage, where it contacts the ionic adsorbent
beads and is adsorbed. The beads, with the adsorbed product, are then
pumped to the desorbing stage in the second reactor, where the addition
of the eluent (NaCl solution) causes the desorption of the antibiotic. The
spent beads are then recycled to the adsorption stage, while the antibiotic
is removed. Both reactors are well agitated.
Applied Biochemistry and Biotechnology Vol. 105-708, 2003
870 Almeida et al.
To reduce the number of variables, four global variables were previ-
ouslydefined, as shown by Eqs.l-4, for reactor residence time (8'1 and 8'2)
and solids residence time (88 1 and 882),
VI
8r 1 = - - - (1)
Fl + Fr Erl

(2)

(3)

(4)

The reactor residence time is the ratio between the liquid total volume
and the global liquid feeding rate. The solids residence time is the ratio of
the amount of resin in the reactor by the out flow rate of the resin from the
reactor. The values of the flow rates (Fl' F2I and Fr), as they are presented,
are calculated based on the established reactor residence times and solids
residence times, as described by Eqs. 1-4.
The differential equations representing the system behavior, in both
stages, are described next.

First Stage
CA BALANCE

iNTRAPARTICLE DIFFUSION

aCi!
- =D (a-2-Ci!+ - ac-i1 ) -
2 - ( ) aqil
p a2
r ar
E
pat eft
E l-E
p
-
at (5)
r

dqil
dt = kl eil (qml - qil) - k2 qil (6)

The initial conditions are


t = 0 ~ qil = Ci1 = 0 (7)
The boundary condition satisfying symmetry condition is
aC ir
r=O~-=O (8)
ar
The boundary condition at the surface of the resin is

(9)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Continuous CA Adsorption Process 871

LIQUID PHASE

SOLID PHASE

(11)

Here, iT 1 and Clare the mean value calculated by the weights of the Radau
quadrature (Eqs. 12 and 13) with contour in the surface of the solid or
Gaussian quadrature (Eq. 14) when some contour is not considered in the
surface (10).

(12)

_ N
~ C.Wo
C 1 = wo·C ]o( r--1) + j=l ] ]
(13)

N
q1=~qowo
j=l ] ]
(14)

in which N is the number of collocation points and Wj are the weights of the
Radau quadrature (Eqs. 12 and 13) or Gauss (Eq. 14).
C 1 can be easily calculated using the principle of the Eq. 13. In the
present study, iT 1 and iT 2 were obtained by Eq. 14.
ELUENT BALANCE

(15)

INERT CONTAMINANTS BALANCE

(16)

Second Stage
CA BALANCE

INTRAPARTICLE DIFFUSION

aC=i2 D
- (a-2-ci2+ -
2-ac-i2 ) - ( ) aqj2
E
p at ef2
E
p ar 2 r ar l-E
p
-
at (17)

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


872 Almeida et al.

(18)

The initial conditions are


t=0-Qi2=C i2 =0 (19)
The boundary condition satisfying the symmetry conditions is
ac-2
r=O--1 =0 (20)
or
The boundary condition at the surface is

(21)

LIQUID PHASE

SOLID PHASE

d7h 1 (7i -) k-
Tt= 8s 2 '11-Q2 - 3Q2 (23)

Here, q2 is calculated in the same manner as in the first stage.


ELUENT BALANCE

(24)

INERT CONTAMINANTS BALANCE

dC T2 CT2 (8s 2 - 8r 2 ) 1
dt = (8s 2 8r 2 ) + 8s 2 (Cn - CT2 ) (25)

The model is represented by a set of partial differential equations


(Eqs. 5 and 17), and they were reduced to a set of ordinary differential
equations by the method of orthogonal collocation. Two internal colloca-
tion points with two additional collocation points corresponding to the
two boundary conditions were used. The set of ordinary differential equa-
tions was solved by the Runge Kutta method of fourth order.
Equations 15, 16,24, and 25 were formulated by assuming that no
adsorption of eluent and contaminants occurs. The differential equations
can be solved in the steady state, and C1 and C2 can be evaluated for
defined operational conditions, which facilitates the process performance

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Continuous CA Adsorption Process 873

Table 1
Simulation Parameters
Co 8r I 8r 2 8s I 8s 2 Erl Er2 VI V2
Run (g/L) (min) (min) (min) (min) (-) (-) (mL) (mL)
1 0.100 71.3 44.1 108.0 154.8 0.86 0.60 130 130
2 0.100 71.3 44.1 108.0 107.2 0.86 0.60 130 90
3 0.100 81.0 30.6 101.1 100.7 0.86 0.60 130 90
4 0.100 81.0 30.6 101.1 65.2 0.86 0.60 200 90

Table 2
Operation Conditions
Run FI (mL/min) F2 (mL/min) Fr (mL/min)
1 0.62 2.10 1.39
2 0.62 1.20 1.39
3 0.32 2.04 1.49
4 0.49 1.56 2.30

analysis in terms of important parameters such as the performance para-


meters yield (Y), concentration factor (CF), and purification factor (PF),
defined by Eqs. 26, 27, and 28, respectively:
CF
Y=100~ (26)
COFI
C
CF=~ (27)
Co
C2 ·C TO
PF= (28)
Co· CT2
Y is the ratio between the mass of the desired product recovered and
the mass of the same product fed into the system. CF is the ratio between
product concentration in the prod uct stream and the product concentration
in the feed stream (Fig. 1). PF is the ratio between CF of the desired product
(C 2 /Co) and CF of the contaminants (CTI/C TO ).

Results and Discussion


The continuous adsorption process of CA was simulated to obtain
information about its dynamic behavior. The values of liquid phase ratio,
initial concentration of CA, reactor residence time, solids residence time,
and reactor volume used in the simulation runs are presented in Table 1.
These parameters were settled based on the flow rates and volumes of a
similar process described by Barboza et a1. (8).
The flow rates regarding each run are given in Table 2; they were
calculated utilizing Eqs. 1-4 and the values of Table 1.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


874 Almeida et al.
Table 3
Transport and Kinetic Parameters
ks kJ
(em/s) (L/[g·min])
Adsorption 9.0 X 10-5 5.30 X 10-2 1.70 0.13
Desorption 9.0 X 10-5 7.65 X 10-2 9.5 X 10-2

Table 4
Equilibrium Parameters
Parameter pH KD (giL)
Isotherm 10 6.2 1.14 X 10-2 7.90 X 10-2

Transport and kinetic parameters (Def' ks, kl1 k2, and k3) were deter-
mined by simulations of kinetic studies in batch runs for both adsorption
and desorption stages. These parameters are given in Table 3. Equilibrium
parameters of the Langmuir model are presented in Table 4. With the data
of Tables I, 3, and 4 the continuous adsorption process of CA on resin
Amberlite IRA 400 was simulated.
Figures 2A-D shows the CA and contaminants concentration profiles
in the exit stream from each stage for runs 1-4, respectively. CA was
adsorbed on the resin Amberlite IRA 400 in the first reactor and was des-
orbed by NaCl solution in the second one. Contaminants were present in
both stages and this fact may result in a low purification factor. Figure 2
shows that CF in all runs was very low because CA concentration at the
desorption stage was lower than at the adsorption stage. The process equi-
librium time was 5 h.
The performance parameters Y, PF, and CF were determined for all
runs and are presented in Table 5. This proposed process was able to fur-
nish a high yield but it was not able to give a concentrated product, as
shown by the low values of CF at all runs. It was observed that CA concen-
tration in the exit of the second stage was higher than contaminants concen-
tration, giving a PF >1. Recently, Mayer et a1. (11) reported values of Y of
about 60%, CF of about 2.0, and PF of approx 1.4 in a CA recovery process
in fixed-bed columns on the same adsorbent. Good CF values were found,
but PF values were in the same range and the Y values were quite low
compared with the results presented in Table 5.
Comparison of the yields of runs 4 (VI = 200 mL) and 3 (VI = 130 mL)
in Table 5 reveals that the highest yield was obtained utilizing a higher
volume at the first reactor. The same reactor residence time and solid resi-
dence time were used at both runs, but at run 4 the yield was 12% higher
than at run 3. For the second reactor, it was better to utilize a smaller vol-
ume, as shown by comparing runs 1 (V2 = 130 mL) and 2 (V2 = 90 mL).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Continuous CA Adsorption Process 875
0,8..-_ _ _ _ _ _ _ _ _ _---,
0 , 8 . . - - - - - - - - - :R,...un-:2,.-----.
0,7 A 0,7 B _CA (lsI stage)
___ CA (2nd stage)
0,6 0,6 -A-CT (lsI stage)
_ _ CT (2nd stage)
..a ........ ...a~d;:& i ..
0,5 "" ... "'" ~;::? Run 1 0,5
(5
0°,4 --CA(lslstage)
__ CA (2nd stage gO,4
........ CT (lsI stage)
_ _ CT (2nd stage) 0,3

0,2

0,1

2 2 4
Time (hours) Time (hours)

0,1I..r---------------. 0,8..-_ _ _ _ _ _ _ _ _ _---,

0,7 C A ~ ".".
0,7 0
"",,,,,, ...a~
0, 0,6
Run3
_CA (lsI stage) Run 4
0,5
_CA (2nd stage) _CA (lsI stage)
_ _ CT (lsI stage)
_ _ CT (2nd slage) 8°,4 -+-CA (2nd stage)
........ CT (lsI stage)
_ _ CT (2nd stage)
0, 0,3

0, 0,2

0,1

O,O-f---r--"""'T----,-.--r---r--I O,0f-----r---r--"""'T-""""'f--r--I
Time (hours) Time (hours)

Fig. 2, Simulations of CA adsorption continuous process: (A) run 1; (B) run 2; (C) run
3; (D) run 4.

Table 5
Simulated Results of Y, PF, and CF

Run Y (%) PF (-) CF (-)


1 41.1 0,95 0.12
2 48.0 1.23 0.25
3 79.0 1.29 0.12
4 90.0 1.57 0,28

The best results of Y, PF, and CF were obtained at run 4, but it is not
possible to say that run 4 had the optimum parameters because the process
used is multivariable. Optimum conditions can be obtained utilizing opti-
mization techniques included in the model. However, the optimization tech-
niques can lead to constrained optimum values of Y, CF, and PF, requiring
further pilot-plant and economical studies for a definite conclusion.
Experimental runs were carried out using the operation conditions
of the simulated runs 2 and 4 (Table 2) in order to validate the model. The
conditions of runs 2 and 4 are presented in Figs. 3 and 4, respectively.
Applied Biochemistry and Biotechnology Vol, 105-108,2003
876 Almeida et al.
Ao.s. B O•6
• CA (Exp.)

0.5. V a ill ~~ J:; -; .... ~ - .... --- 0.5 _CA(Sim.)


a CT(Exp.)
__ CT(Sim.)
0.4 ~ r ~ 0.4

....
• CA(EXP'l
a ," _CA(Sim.
Q 0.3. • I Q CT(Exp.)
() I __ CT(Sim.)
0.2 0.2 - - - , . . -II"" - ..

0.1 0.1

0.01-:to~~lr-~""!'2~-jr-~'-
4~-r~
Time (hours) Time (hours)

Fig. 3. Simulated and experimental results of CA adsorption continuous process of


run 2: (A) first reactor; (B) second reactor.

Ao.6
0.51. ~ q. ...JJ.r"'a r
. .. -a -
..
a- ; - 'D ....
B o.
• CA (Exp.)
_CA(Sim.)

0.4
V ; ~
~ .. • CA (Exp.)
Ig CT (Exp.)
__ CT(Sim.)
_CA(Sim.)
gO.3 J D CT (Exp.)
- - CT(Sim.)
a
~
..... - ... " , __
O.
J
- -
0.2 1 a lit
a a- -0"'11'.
0.1

O.O+-_"'Il_ _ ~_""~
_ _ _~_ _.,......I~ o. 1 2 4 5

Time (hours) Time (hou,",)


Fig. 4. Simulated and experimental results of CA adsorption continuous process of
run 4: (A) first reactor; (B) second reactor.

Figure 4 shows that the fits of CA and CT in both stages were good, but
Fig. 3 it shows that the model fitted CA data well in the adsorption reactor
but not in the desorption one. Since the volume of the first reactor in run
2 was lower than in run 4, it was expected that in run 2 a lower concentra-
tion than in run 4 in the desorption step would be achieved, around 0.25,
like simulation. However, the concentration was equal to run 4 and, con-
sequently, higher than obtained in the simulation. Further experiments,
in different conditions, will be necessary to elucidate this behavior. Fig-
ures 3 and 4 also shown that CA concentration in the second reactor was
lower than in the first, so, like the model predicted, a low CF was also
obtained in the experimental runs.
The parameters Y, PF, and CF were calculated for experimental runs
and are shown in Table 6, together with the parameters obtained by simu-
lation and the difference between the experimental and the simulated
results. The data in Table 6 show that the model predicted the performance
parameters with the differences between experimental and simulated data
lower than 17%. These differences can be related to some operational prob-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Continuous CA Adsorption Process 877

Table 6
Simulated and Experimental Results of Y, PF, and CF
and Difference Between Experimental and Simulated Results (e)
Y (%) PF (-) CF (-)
Run Exp. Sim. e (%) Exp. Sim. e (%) Exp. Sim. e (%)
2 58.10 48.00 17 1.48 1.23 17 0.30 0.25 17
4 96.90 90.00 7 1.69 1.57 7 0.31 0.28 9
"Exp., experimental; Sim., simulated.

lems in the experimental runs. For instance, a severe practical limitation of


this continuous process, as proposed here, is the recirculation of adsorbent
beads between the two reactors. This recirculation was sometimes manu-
ally operated because the beads frequently obstructed the pump. How-
ever, a steady flow could be attained most of the time.
The overall purpose of the model was to predict the general behavior
of the system and this purpose was attained. The model can be used to
determined the process parameters such as Y, PF, and CF obtained in
several different conditions and therefore can be used as a tool for opti-
mization of the multi variable process and process control and design. As
an example, the model can be used to optimize the process in order to
improve Y, CF, and PF, up to a level allowing economical operation of an
industrial plant.
From an industrial point of view, the continuous adsorption process
is feasible for use in combination with fermentation utilizing a filtration
system to remove cells and proteins in the fermentation broth. The size of
both fermentation and purification vessels should be established in such a
way that the fermentation step (batch) proceeds directly to a continuous
operation of the purification step.

Conclusion
Based on the CF, PF, and Y, the process performance was evaluated.
The experimental and simulated data of these parameters were compared,
and the differences between them were, in most cases,lower than 17%. This
fact shows the good agreement of the model with the experimental data:
A model of a continuous process for CA separation was described,
incorporating a kinetic model. This validated model, together with key
experimental data, can be a powerful tool to optimize and scale up this
purification process.

Nomenclature
Ck = CA concentration in the solution (giL)
Cik = CA concentration inside the particles (giL)
Applied Biochemistry and Biotechnology Vol. 105-108,2003
878 Almeida et al.
CSk = CA concentration at particle surface (g/L)
CEk = NaCl concentration (%)
CTk = contaminant concentration (g/L)
Defk = effective diffusivity of CA (cm2/ s)
e = difference between simulated and experimental data (%)
Fk = flow rate of feeding (mL/min)
Fr = flow rate of recycle (mL/min)
KD = Langmuir equilibrium constant (g/L)
KSk = mass transfer coefficient (cm/s)

kI = intrinsic kinetic constant for adsorption stage (g/[L'min])


k2 = intrinsic kinetic constant for adsorption stage (min-I)
k3 = intrinsic kinetic constant for desorption stage (min-I)
N = number of points of collocation (-)
qmk = kinetic equilibrium parameter (gCA/ gres)
qk = average amount of CA adsorbed in each stage (gCA/ gres)
qik = amount adsorbed in present particle along position r (gCA/ gres)
Vk = volume of present liquid phase (m/L)
Wj = weights of Radau quadrature (Eqs. 13 and 14) or Gauss (Eq. 12)
Erk = volume phase ratio, liquid to total phase volume (-)
Ep = porosity of resin (-)
8rk = reactor residence time (min)
8sk = solid residence time (min)

Subscripts
k = 0 = initial
k = 1 = first stage
k = 2 = second stage

Acknowledgment
We wish to acknowledge Funda<;ao de Amparo aPesquisa do Estado
de Sao Paulo (Process No. 99/07693-2, 99/03279-7, and 98/11596-0) for
financial support.

References
1. Brown, A. G., Butterworth, D., Cole, M., Hanscomb, G., Hood, J. D., Reading, c., and
Rolinson, G. N. (1976), J. Antibiot. 29,668-669.
2. Butterworth, D. (1984), in Biotechnology of Industrial Antibiotics, vol. 6, Vandamme, E.
J., ed., Marcel Dekker, New York, NY, pp. 225-235.
3. Haginaka, J., Nakagawa, T., and Uno, T. (1981), Chem. Pharm. Bull. 29,3334-3341.
4. Mayer, A. F., Hartmann, R., and Deckwer, W. D. (1997), Chem. Eng. Sci. 52,4561-4568.
5. Pungor, E., Afeyan, N. B., Gordon, N. F., and Cooney, C. L. (1987), Bio/Technology 5,
604-609.
6. Afeyan, N. B., Gordon, N. F., and Cooney, C. L. (1989), J. Chromatogr. 47, 1-19.
7. Rodrigues, M. I., Zaror, C. A., Maugeri, F., and Asenjo, J. A, (1992), Chem. Eng. Sci.
47(1),263-269.
8. Barboza,M., Hokka,C. 0., and Maugeri,F. (2002),Bioprocess Biosystem Eng. 25,193-203.
9. Foulstone, M. and Reading, C. (1982), Antimicrob. Agents Chemother. 22, 753-762.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Continuous CA Adsorption Process 879
10. Villadsen, J. and Michelsen, M. L. (1978), Solution of Differential Equation Models by
Polynomial Approximations, Prentice Hall, Englewood Cliffs, NJ.
11. Mayer, A. F., Anspach, F. B., and Deckwer, W. D. (1996), Bioseparation 6, 25-39.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0881/$20.00

Extraction of Nutraceuticals from Milk Thistle


I. Hot Water Extraction

JOSE F. ALVAREZ BARRETO,1 SUNNY N. WALLACE/


DANIELLE JULIE CARRIER, 2 AND EDGAR C. CLAUSEN*,1

1Department of Chemical Engineering, University of Arkansas,


3202 Bell Engineering Center, Fayetteville, AR 72701,
E-mail: ec/ause@engr.uark.edu; and
2Department of Biological and Agricultural Engineering,
University of Arkansas, 203 Engineering Hall, Fayetteville, AR 72701

Abstract
Milk thistle contains compounds that display hepatoxic protection prop-
erties. We examined the batch extraction of silymarin compounds from milk
thistle seed meal in 50,70,85, and 100°C water as a function of time. After
210 min of extraction at 100°C, the yield of taxifolin was 1.2 mg/ g of seed,
a 6.2-fold increase over the results obtained in a Soxhlet extraction with
ethanol on pretreated (defatted) seeds. Similarly, the yield of silychristin
was 5.0 mg/ g of seed, a 3.8-fold increase. The yields of silybinin A and
silybinin B were 1.8 and 3.3 mg/ g of seed, respectively, or roughly 30% of
the Soxhlet yield. The ratios of the extracted compounds, and particularly
the ratios at long extraction times, showed that the more polar compounds
(taxifolin and silychristin) were preferentially extracted at 85°C, while the
less polar silybinin was favored at lOO°e.
Index Entries: Milk thistle; extraction; water; silymarin; flavanolignans.

Introduction
Milk thistle (Silybum marianum) is an annual or a biennial plant native
to the Mediterranean and North Africa. It grows wild throughout Europe,
North Africa, the Americas, and Australia but can also be cultivated (1).
The plants can reach a height of 10 ft with dark and shiny leaves, and purple
to reddish flowers. Milk thistle has an indeterminate growth habitat, result-
ing in staggered flowering and maturity (2). The seeds of the plant contain
a group of flavanoid compounds commonly named silymarin (3).
The term silymarin usually encompasses the dihydroflavonol-
taxifolin-and the flavanolignans-silybinin, isosilybinin, silydianin, and
*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 881 Vol. 705-108, 2003


882 Barreto et al.

aI~O'~
o ~)6(Yt°""
~O
"'WO~,r6r~
O:~
,oK
(Ii
OCH,

seN SON SBN

~~i§X:
TXF
'"m'~
~
ISBN
.

Fig. 1. Structures of silychristin (SeN), silydianin (SDN), silybinin (SBN), taxifolin


(TXF), and isosilybinin (ISBN).

silychristin (Fig. 1). Some studies suggest that silybinin reduces the biliary
cholesterol concentration (4). It has also been demonstrated that silybinin is
useful in the intervention of hormone refractory human prostate cancer (5).
Furthermore, the combination of silybinin and silychristin has been found
helpful in decreasing the nephritic effects of chemical-induced injury (6).
The Deutsches Arzneibuch procedure for silymarin extraction is a
two-step process in which seeds are first defatted in a Soxhlet extraction
with petrol for 4 h, followed by a second Soxhlet extraction with methanol
for 5 h (7). Using this procedure, Benthin et a1. (8) reported silybinin yields
of 11 mg of silybinin/ g of seed. These investigators also extracted milk
thistle using pressurized liquid extraction techniques, in which 12 mg of
silybinin/ g of seed was obtained. In extracting OA-mm particle-size milk
thistle seed meal in a Soxhlet with petrol for 24 h, followed by an ethanol
Soxhlet for 4 h, Wallace et a1. (9) reported a silybinin yield of 16 mg/g of
seed meal. The differences in the results obtained by Wallace et a1. (9) and
Benthin et a1. (8) may not be significant, since the silybinin content of seed
batches varies significantly (2).
Wallace et a1. (9) reported the analysis of three off-the-shelf milk thistle
products, of which only two products contained silymarin compounds.
Inconsistency between herbal supplement label and product content is not
uncommon. For example, an analysis of ephedra products (10) showed a
broad range of ephedra alkaloid content, pointing most likely to manufactur-
ing problems. The lack of consistency among products can be owing in part
to the extraction step, where the desired molecules diffuse from the bulk herb
to a solvent phase, usually ethanol, methanol, acetone, hexane, or petroleum
ether. To increase the quality of products, the extraction step should be well
characterized, in terms of both rates and appropriate solvents.
The use of hot liquid water as an extraction solvent has recently caught
the attention of some researchers (11,12). Water is very useful in extracting
polar compounds and may also be useful in extracting polar compounds
from plant material without prior defatting. In increasing the water tem-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hot Water Extraction of Milk Thistle 883
perature up to its subcritical temperature, a decrease in the dielectric con-
stant is observed. For example, water at 250°C (523 K) displays a dielectric
constant of 27, which is the realm of that of methanol (33) and ethanol (24).
As a result, hot liquid (hot/liquid) water has solubility characteristics at
increased temperature that are similar to ethanol and methanol. The solu-
bilities of anthracene, pyrene, chrysene, perylene, and carbazole (13) and of
d-limonene, carvone, eugenol, 1,8-cineole, and nerol (14) were determined
in 289 and 498K (hot/liquid) water, where increases were observed with
temperature. Kubatovci et al. (12) showed that the extraction of peppermint
compounds using (hot/liquid) water at 175°C required 15 min, as com-
pared to 4 h with hydrodistillation. The use of (hot/liquid) water as an
extraction solvent shows promise as the search for milder and "greener"
solvents is intensified.
This article presents results from the extraction of silymarin com-
pounds from milk thistle seeds using (hot/liquid) water as the solvent, a
first step in process characterization. Silymarin compounds, ranging from
highly polar (taxifolin) to less polar (silybinin), were extracted in 50-100°C
water over a period of 17 h. A maximum temperature of 100°C was em-
ployed in order to maintain liquid water at atmospheric pressure. Increased
pressure experiments will be run in the future. In our study, we compared
the compounds extracted, as the temperature increased, and the dielectric
constant. We also examined the effect of water extraction temperature on
extraction yield.

Materials and Methods


Extraction Experiments
Milk thistle seeds were purchased from Frontier Herbs (Norway, IA)
and ground with a coffee grinder to an average particle diameter of 0.4 mm.
Extraction experiments were conducted at 50,70,85, and 100°C using 2 g
of seed (contained in a cheesecloth bag) in 200 mL of deionized water.
Leaching at 100°C was carried out in a 500-mL glass round-bottomed flask,
fitted with a condenser for total reflux. The flask was heated in an electric
mantel, and water was used to condense the vapor. The leaching experi-
ments at 50,70, and 85°C were carried out in 500-mL bottles in a shaker
water bath (Dubnoff Metabolic Shaking Incubator; Precision Scientific,
Winchester, VA) set at 80 strokes/min. Although the process conditions
were slightly different when operating at or below 100°C, the long diffu-
sion times observed in the experiments helped minimize the small differ-
ences in the systems.
Samples of extraction water, as a "tea," were taken in triplicate every
30 min, including time zero, using a l-mL pipet. Since the solids were held
in a permeable bag, they were not removed during sampling. Time zero was
arbitrarily set as the time when the water started boiling (100°C), or when
the temperature of the water in the bottles equilibrated with the set experi-
mental temperature (50, 70, 85°C). Aliquots were placed in preweighed test

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


,
884 Barreto et al.

,.il
i\
o
~ ~~ Il'~~~ ~~ .
0.00 ._. __ J._, ..-._.__.
i i i i
5.00
'
"--,H~~~;..~\.ll!l ~j,)jJ!- . J!-
j • I
10.00
• • iii
...

15.00
' i , ii'
20.00
, i • I i i ' -r--,--r--r-""T"'"" ,
25.00 30.00
i
35.00
' i •

Fig. 2. Typical chromatogram of milk thistle seed extract. Retention times of


taxifolin, silychristin, silydianin, silybinin A, and silybinin B were 10.059, 18.476,
21.264,24.313, and 25.330 min, respectively. Note that this particular seed lot con-
tained miniscule amounts of silydianin.

tubes and weighed to determine aliquot weight. Subsequently, the aliquots


were evaporated to dryness in a SpeedVac (Savant, Holbrook, NY). To the
dried sample, 1 mL of methanol was added to redissolve the residue. After
vortexing and centrifuging (lOg), the supematantwas filtered and analyzed,
as described next.
Chemical Analysis
The silymarin concentrations were determined by high-performance
liquid chromatography using a Waters system (Milford, MA) consisting of
an Alliance 2690 separations module and a 996 Photodiode Array, con-
trolled with Millennium32 chromatography software (9). Separation of the
silymarin compounds was obtained using a Symmerry® (Waters) C1S pre-
column placed in series with a Symmetry (Waters) C1S column (150 x 4.6 mm,
5 !lm), both at 40°C. A 10-!lL sample volume was injected. Solvent A was
20:80 methanol:water, while solvent B consisted of 80:20 methanol:water.
The gradient program was initiated with 85:15 solvent A:solvent B flowing
for 5 min, followed by a linear gradient of 45:55 solvent A:solvent B for 15
min. The proportions of 45:55 solvent A:solvent B were then held constant
for 20 min and brought back to 85:15 solvent A:solvent B over 10 min. The
flow rate was 0.75 !lL/min, and the silymarin compounds were monitored
at 290 nm. Peak identification was confirmed by mass spectrometry
(Pharmalytics, Saskatoon, Saskatchewan, Canada). Calibration curves were
prepared with silybinin from Sigma (St. Louis, MO), taxifolin from
Extrasynthese (Lyon, France), and silychristin and silydianin from PhytoLab
(Hamburg, Germany). No standard was available for isosilybinin, and thus
this compound was excluded from the analysis. The silybinin standard
obtained from Sigma contained two distinct peaks, which are further
referred to as silybinin A (first peak) and silybinin B (second peak). A sample
chromatogram from the extraction of milk thistle seeds is shown in Fig. 2.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hot Water Extraction of Milk Thistle 885

700C
2 -r---------, 1
5"-
• •
:co E

.. ;.1 1••
cao
• Satch 1 J:oo .Satch 1
.Satch2 S'a, .aatch2

o 100 200
ASatch3

300
S
0
E
0
0
••• 100 200 300
ASatch3

Tme,rrin Tme,nin

8SOC 100'C

.1 ·I.·
1
5_
A
5-

6
5 •
S4
:COB
go ....... 1 .Satch 1
i8 'a,~
:co
• Batch 1

:1-
S'a, .Satch2 .Batch2
5E ASatch3 E1 •• A Batch 3
0
0 01
0 100 200 300 0 100 200 300
Tme, nin Tme,rrin

Fig. 3. Silybinin B concentration as a function of extraction time at different tem-


peratures. Results show all the batches for all temperatures.

Results and Discussion


For all temperatures, three distinct experiments were conducted, of
which three samples were taken per time point (total of nine samples per
time point). Figure 3 demonstrates the reproducibility of the concentration-
time data at each temperature by showing the silybinin B concentration
in the extract water with time. The reproducibility of the data at each tem-
perature was generally quite good, with only a few "outliers" observed in
the data trend. In addition, the reproducibility of the data improved
with increasing temperature as the concentration of the extracted com-
pound increased.
Figure 4 shows typical results from the extraction of taxifolin,
silychristin, silybinin A, and silybinin B, presented as the yield of each
compound (mg/ g of seed) as a function of extraction time and temperature.
Each of the extracted compounds showed a consistent pattern of increasing
yield with temperature and time as equilibrium was approached. For each
of the silymarin compounds, extraction with 100°C water produced the
highest yield and concentration of compounds. After 210 min of extraction
at 100°C, the yield of taxifolin was 1.2 mg/g of seed, while the yields of
silychristin, silybinin A, and silybinin B were 5.0,1.8, and 3.3 mg/ g of seed,
respectively. Assuming that a 4 hr extraction of pretreated (defatted) seeds
with ethanol gives the highest yield of compounds, the percentage yield of
Applied Biochemistry and Biotechnology Vol. 105-108,2003
886 Barreto et at.

Taxlfolin Silychristin

• • • ••
..----------,r----i


1.5 6
• •••
••
.SO"C • WC
-a

•••
"i E4
=:
• .70"C
-;:2
.70"C


-;: 0.5 .. 85°C .. as"C
E .'00"C E .'00"C
100 200 300 300
TIme,mln TIme, min

SIIybIninA SlIyblnln B

••
••
"
• • •••
.wc

• ••
1:
•••
.70"C
J!I .. as"C
eo 1
0
• . . . . ill , • • .'00"C
o 100 200 0 100 200 300
TIme. min Time. min

Fig. 4. Silymarin as a function of time at different temperatures. Results are based


on the first experiment of each batch.

taxifolin after 210 min was 720%, while the percentage yields of silychristin,
silybinin A, and silybinin B were 480, 30, and 33%, respectively. Note that
the extraction yields of the polar compounds ta'Xifolin and silychristin were
significantly higher than the extraction yields in ethanol, indicating that
water is a better solvent for extracting polar compounds from milk thistle.
After 300 min of extraction, the yields of taxifolin, silychristin, silybinin A,
and silybinin B were 0.92 (550% of Soxhlet results), 4.7 (440%), 1.8 (30%),
and 3.4 (34%) mg/g of seed, respectively (data not shown). A slight
decrease in the yield of taxifolin was observed after 150 min, perhaps indi-
cating the onset of decomposition. Overall, water extraction at lOoDe
yielded about 65% of the amount of the total silymarins obtained in the two-
step Soxhlet extraction (with defatting) performed by Wallace et a1. (9).
The ratios of the concentrations of taxifolin, silychristin, and silybinin
A to the concentration of silybinin B at 85 and lOoDe as a function of time
are shown in Fig. 5. These temperatures were chosen because the silymarin
concentrations were not as large at temperatures below 85 C. As noted in D

Fig. 5A, at 85 e the ratio of taxifolin to silybinin B increased rapidly to 0.7


D

g/ g and then held constant at that level. At lOODC, the ratio reached a
maximum of 0.65 g/ g and then gradually fell with time to 0.35 g/ g. This
reduction in the ratio at lOODC shows that the taxifolin concentration
reached its maximum faster than silybinin B. A similar behavior for the
ratio of silychristin to silybinin B is noted in Fig. 5B. At 85 e, the ratio D

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Hot Water Extraction of Milk Thistle 887

A TAlSB B SiIyISB

0.8 2.5
2.0
0.6
m m 1.5 ......-esc
~ 0.4 ......-85C
.~ 1.0
_100C _100C
I- 0.2 ti) 0.5
0 0.0
0 100 200 300 0 100 200 300
Tme,nin Tme,nil

C SAlSB

rr
0.8

1 0.6
0.4
----~-- ......-85C
rn _100C
0.2
0.0
0 100 200 300
Tme,nin

Fig. 5. Compound ratio as function of time for 85 and 100°C experiments. (A) Taxifolin
to silybinin Bratio; (B) silychristin to silybinin B; ratio. (C) silybinin A to silybinin Bratio.

rapidly rose to just above 2.0 g/ g and then gradually increased before
leveling out at 2.2 g/ g. At 100°C, the ratio increased to a maximum of 2.0
g/ g and then gradually fell to 1.5 g/ g. Figure 5C shows that, excluding an
initial sharp increase, the ratio of silybinin A to silybinin B at 85°C was
constant at 0.65 g/ g. At 100°C, the ratio was constant at about 0.6 g/ g, again
excluding the initial period of sharp increase.
These ratios, and particularly the ratios at long extraction times, show
that the more polar compounds (taxifolin and silychristin) are preferen-
tially extracted at 85°C, while the less polar compounds (silybinin A and B)
are more easily extracted at 100°C (see also the data in Table 1). The data
reported by Wallace et al. (9) showed that the ratios of taxifolin to silybinin
B, silychristin to silybinin B, and silybinin A to silybinin B were 0.02, 0.1,
and 0.6, respectively. Thus, the ratios of extraction products using water at
100°C more closely resemble the Soxhlet extraction results than the water
extractions at temperatures below 85°C. More dramatic differences in polar
and nonpolar compound extraction with water are expected as the tem-
perature of liquid water is further increased, thereby lowering the dielec-
tric constant.
Although the yields of taxifolin, silychristin, silybinin A, and
silybinin B using water were less than what is reported in ethanol (9), this
technology shows promise because of the omission of the defatting step.
An oil-removal step was found necessary in the extraction procedures
proposed by Kahol et al. (15) and Benthin et al. (8). It is hoped that the
Applied Biochemistry and Biotechnology Vol. 105-108,2003
888 Barreto et al.
Table 1
Calculated Ratios of Compound/Silybinin B as Function of Temperature"
Water Dielectric Ratio of compound to SBb
temperature (0C) constant E Taxifolin/SB Silychristin/5B 5ilybinin A/5B
50 70 0.916 2.006 0.615
70 ·64 0.594 1.869 0.639
85 60 0.661 2.237 0.630
100 56 0.352 1.546 0.551
"These ratios were calculated at the last sampling point.
"SB, silybinin B.

work of Wallace et a1. (16) comparing the extraction of nondefatted and


defatted milk thistle seed meal using ethanol as the solvent will shed
more light on this subject matter.

Conclusions
Water is not only an interesting alternative solvent because of its low
operating and disposal costs, but is also effective in extracting the more
polar silymarin compounds from milk thistle seed. For each of the com-
pounds, extraction with 100°C water gave the highest yield and concen-
tration. After 210 min of extraction at 100°C, the yield of taxifolin was 1.2
mg/ g of seed, a 6.2-fold increase over Soxhlet extraction of pretreated
seeds with ethanol, while the yield of silychristin was 5.0 mg/ g of seed,
a 3.8-fold increase. The yields of silybinin A and silybinin B were 1.8 and
3.3 mg/ g of seed, respectively, or roughly 30% of the yield in the Soxhlet
extraction. The ratios of the extracted compounds, and particularly the
ratios at long extraction times, showed that the more polar compounds
(taxifolin and silychristin) were preferentially extracted at 85°C, while
the less polar compounds (silybinin A and B) were favored at 100°C.

References
1. Hamid,S., Sabir, A., Khan,S., and Aziz, P. (1983), Pakistan J. Sci. Ind. Res. 26,244-246.
2. Carrier, D. J., Crowe, T., Sokhansanj, 5., Katrusiak, A., Whahab, J., and Barl, B. (2003),
J. Herbs, Spices Med. Plants, 10(3), in press.
3. Tittle, G. and Wagner, H. (1978), J. Chromatogr. 153,227-232.
4. Duke, J. (1999), J. Med. Food 2, 73-76.
5. Zi, X. and Agarwal, R. (1999), Proc. Natl. Acad. Sci. USA 96, 7490-7495.
6. Sonnenbichler, J., Scalera, F., Sonnenbichler,l., and Weyhenmeyer, R. (1999), J. Pharm.
Exp. Ther. 290, 1375-1383.
7. DAB 9 (1986) Deutsches Arzneibuch, 9th ed., Deutscher Apotheker-Verlag,
Stuttgart.
8. Benthin, B., Danz, H., and Hamburger, M. (1999), J. Chromatogr. A 837, 211-219.
9. Wallace,S., Carrier, D. J., Beitle, B., Clausen, E., and Griffis, C. (2002), J. Nutraceut.
Funct. Med. Foods, submitted.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Hot Water Extraction of Milk Thistle 889
10. Gurley, B., Gardner, S., and Hubbard, M. (2000), Am. J. HeaIth-Syst. Pharmacy 57,
963-969.
11. Basile, A, Jimenez-Carmona, M., and Clifford, A (1998), J. Agric. Food Chem. 46,
5205-5209.
12. Kuh<itova, A, Miller, D., and Hawthorne, S. (2001), in 10th International Symposium
and Exhibit on Supercritical Fluid Chromatography, Extraction and Processing, Chester, T.
L., Taylor, L. T., and King, J. W., eds., Supercritical Conferences, Cincinnati, OH, p.
E-05.
13. Miller, D., Hawthorne, S., Gizir, A, and Clifford, A (1998), J. Chem. Eng. Data 43,
1043-1047.
14. Miller, D. and Hawthorne, S. (2000), J. Chem. Eng. Data 44, 315-318.
15. Kahol, A, Singh, K., Tandon, S., and Kumar, S. (2001), Indian patent no. 06309678.
16. Wallace, S., Carrier, D. J., and Clausen, E. C. (2003), Appl. Biochem. Biotechnol., in press.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0891/$20.00

Extraction of Nutraceuticals
from Milk Thistle
Part II. Extraction with Organic Solvents

SUNNY N. WALLACE, 1 DANIELLE JULIE CARRIER, 1


AND EDGAR C. CLAUSEN*,2

1Department
of Biological and Agricultural Engineering,
University of Arkansas, 203 Engineering Hall, Fayetteville, AR 72701; and
2Department of Chemical Engineering, University of Arkansas,
3202 Bell Engineering Center, Fayetteville, AR 72701,
E-mail: eclause@engr.uark.edu

Abstract
Seeds from milk thistle (Silybum marianum Gaert L.) contain flavanolignan
and dihydroflavanol compounds that have interesting and important thera-
peutic activities. The recovery of these silymarin compounds generally
involves a two-step defatting and extraction process using organic solvents.
This study examined the batch, single-stage extraction of whole and defatted
seeds using ethanol, methanol, acetonitrile, and acetone as the solvents. In
extracting defatted milk thistle seeds with organic solvents, extraction with
ethanol resulted in the highest silymarin yield, although some potential deg-
radation was observed. The maximum yields of taxifolin, silychristin,
silydianin, silybinin A, and silybinin B in ethanol were 0.6, 4.0, 0.4, 4.0, and
7.0 mg/ g of defatted seed, respectively. However, if silybinin A were the
diastereoisomer of choice, methanol would be the preferred extraction sol-
vent because it yielded the highest silybinin A to silybinin B ratio. Interest-
ingly, lipid removal is an important extraction step, because defatted material
yields twice the silymarin concentration.
Index Entries: Milk thistle; extraction; ethanol; methanol; silymarin;
acetone; acetonitrile.

Introduction
Seeds from milk thistle (Silybum marianum Gaert L.) contain flavano-
lignan and dihydroflavanol compounds that have interesting and impor-
tant therapeutic activities. Milk thistle flavanolignan and dihydroflavanol

*Author to whom all correspondence and reprint requests should be addressed.


Applied Biochemistry and Biotechnology 891 Vol. 105-108,2003
892 Wallace et al.

are also referred to as silymarin. The flavanolignan silybinin shows poten-


tial for reducing bilary cholesterol concentrations (1), and in the interven-
tion of hormone refractory human prostate cancer (2). In addition,
combinations of the flavanolignans silybinin and silychristin have been
shown to decrease the nephritic effects of chemical injury (3). Hence, there
is substantial evidence that indicates the therapeutic benefits of silymarin.
Interestingly, silymarin products are sold under different categories in
Europe than in North America. American milk thistle products are sold as
dietary supplements in a $45,000,000 industry (4), while the European
Union regulates them as therapeutic products (1).
With more North Americans using botanical preparations, there is
growing concern about their safety and efficacy. A quick analysis of
American off-the-shelf milk thistle products showed that the labels and
the actual content often do not match, to the point that one product was
actually devoid of flavanolignans (5). Similar testing of off-the-shelf
gingko (6) and ephedra (7) products also yielded inconsistencies between
product content and product label. Botanicals fall under the dietary
supplements umbrella, and, unfortunately, the corresponding products
vary much more than what is allowed by the pharmaceutical manufactur-
ers. However, as clinical evidence demonstrates the efficacy of specific
compounds, standardization, potency, safety, and dose response will
become important parameters, requiring extensive documentation.
The production of milk thistle products involves a two-step extrac-
tion process. Before being extracted for their flavanolignan content, milk
thistle seeds must first be defatted to remove the -25% lipid content (5,8).
However, the open scientific literature on milk thistle extraction is scarce.
The Deutsches Arzneibuch procedure for flavanolignan extraction stipu-
lates that the seeds must first be extracted in a petrol Soxhlet for 4 h,
followed by a methanol Soxhlet for 5 h (9). The work of Benthin et a1. (9)
reported on the adaptation of the Deutsches Arzneibuch procedure to
pressurized liquid extraction, using petrol and methanol as the extraction
solvents. A procedure in which the seeds were frozen, ground, and defat-
ted with hexane and then extracted with acetonitrile for maximal
flavanolignan recovery has also been reported (10). Thus, industrial
extraction procedures imply the use of organic solvents such as petrol,
methanol, and acetonitrile for flavanolignan extraction. Although spe-
cific solvents are proposed, information on liquid-to-solvent ratios,
extraction temperatures, and extraction rates is scarce. To increase the
quality of products and the efficiency of extraction, the extraction step
should be well characterized, in terms of both extraction conditions and
appropriate solvents.
The purpose of this study was to investigate flavanolignan extraction
(silymarin) from milk thistle seeds using ethanol, methanol, acetonitrile,
and acetone as the extraction solvents. Batch extraction data were obtained
at the normal boiling points of the solvents as a function of time. The extrac-
tion of whole and defatted seed was also compared.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Solvent Extraction of Milk Thistle 893

Materials and Methods


Extraction Experiments

Milk thistle seeds were purchased from Frontier Herbs (Norway, IA)
and ground with a coffee grinder to an average particle size of 0.4 mm.
Extraction experiments were conducted at the normal boiling point of
the respective solvent (methanol, ethanol, acetonitrile, or acetone). For all
experiments, 2 g of seed (contained in a cheesecloth bag) was extracted in
200 mL of the corresponding solvent. The boiling flask was heated in an
electric mantel, and water was used to condense the vapor.
Extraction samples were taken in triplicate every 30 min, including
time zero, using a 1 mL pipet. Time zero was set as the time when the
solvent started boiling. The aliquots were placed in preweighed test tubes
and weighed to determine aliquot weight. Subsequently, the aliquots
were evaporated to dryness under a stream of nitrogen. To the dried
sample, 1 mL of methanol was added, after which the solution was
vortexed and centrifuged (lOg). The supernatant was filtered and ana-
lyzed, as described next.

Chemical Analysis
Silymarin concentrations were determined by high-performance liq-
uid chromatography (HPLC) using a Waters system (Milford, MA) com-
posed of an Alliance 2690 separations module and a 996 Photo diode
Array, controlled with Millennium32 chromatography software. Silymarin
compounds were separated using a Symmetry® (Waters) C I8 precolumn
placed in series with a Symmetry (Waters) C I8 column (150 x 4.6 mm, 5
!-tm), both at 40°C. A 10-!-tL sample volume was injected. Solvent A was
20:80 methanol:water, while solvent B consisted of 80:20 methanol:water.
The gradient program was initiated with 85:15 solvent A:solvent B flow-
ing for 5 min, followed by a linear gradient of 45:55 solvent A:solvent B
for 15 min. The proportions of 45:55 solvent A:solvent B were then held
constant for 20 min and brought back to 85:15 solvent A:solvent B over 10
min. The flow rate was 0.75 !-tL/min, and the silymarin compounds were
monitored at 290 nm. Peak identification was confirmed by mass spec-
trometry (Pharmalytics, Saskatoon, Saskatchewan, Canada). Calibration
curves were prepared with silybinin from Sigma (St. Louis, MO), taxifolin
from Extrasynthese (Lyon, France), and silychristin and silydianin from
PhytoLab (Hamburg, Germany). No standard was available for iso-
silybinin, and thus this compound was excluded from the analysis. The
silybinin standard obtained from Sigma contained two distinct peaks,
which are further referred to as silybinin A (first peak) and silybinin B
(second peak). Sample chromatograms from the extraction of whole and
defatted milk thistle seeds are shown in Fig. 1, in which the HPLC proce-
dure was previously described (5).

Applied Biochemistry and Biotechnology Vol. 105-108,2003


894 Wallace et al.

5.00 10.00 40.00 45.00 !!II.OO

t:~~~~~~~~,.,.
......
'_,-,- ,,-,--.-.' ~,.,.-..--,--,r-r-I
5,00 10.00 15.00 20.00 25.00 30.00 315.00 40.00 045.00 !II.OO

Fig. 1. Typical chromatogram of milk thistle seed extract: (A) whole seeds; (B)
defatted seeds. Retention times of taxifolin, silychristin, silydianin, silybinin A, and
silybinin B were 8.7,15.4,17.6,22.9, and 23.8 min, respectively. Note thatthis particular
seed lot contained miniscule amounts of silydianin.

Results and Discussion


Extraction Reproducibility
Experiments were performed to illustrate the similarities and differ-
ences in extracting whole and defatted seeds with various organic solvents
at their normal boiling points. Figure 2 demonstrates the reproducibility of
the concentration-time data in each solvent by showing the silybinin B
yields in the extract solvent as a function of time. Three distinct experi-
ments were conducted with each solvent, and three samples were taken at
each time point for a total of nine samples per time point. The reproducibil-
ity of the data for each of the extracted compounds was generally quite
good, with only a few outliers observed in the data trend.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


).
:g Ethanol Methanol
A B
[
OJ 9.0
o· 3.5
g. " 8.0 • ••
! 7.0 3
3 "! •
(;;. " •
"i 6.0 • ~ 2.5
.t
~ ~
= 5.0 ii , .~ •
I\; -;
::J
co
, • •• 2 ".
Q.. ! 4.0 ,t ,. • • •I. Batch
I ~I
Batch 2
.Batch 2
I~BatCh ;1 1.5
OJ • "~ Batch 3
o· }3.0 Batch 3 E ,
<ii ~ 2.0 • •
• ". • •
g. "s:0;
O.S
::J
o >: l ·~r
0.0 0
~ 0 200 400 600 800 0 200 400 600 800
'. Tlme(min) Tlme(min)


1..0
V1 Aceton itrile 0 Acetone
C
4.5 3.0
" 4.0
2.5
! 3.5 "! .
"i 3.0 • 2.0
• 1\
• • • ~ • .- ,..
~co 2.5 • • • S.t :.. • • • :!co • •
1.S /' , :. -.
" 2.0 • • Batch 2
I~Batch ~I
Batch 3
'" •• "~ 1.0
•• • •
t1.5 • E
" 1.0
"-;>: 0.5

~ 0.5
~ 0.0 0.0
0 200 400 600 800 0 200 400 600 800
~ Tlme(min) n"", (min)
a
,0>

a
""' Fig. 2. Silybinin B yield (mg of compound/ g of defatted seed) as function of time in ethanol, methanol, acetone, and acetonitrile at their
3 normal boiling points. Results show all the batches for all of the solvents. (A) Ethanol; (B) methanol; (C) acetonitrile; (D) acetone.
896 Wallace et al.

A Taxlfolln

"t:I 0.7
do
do
<II 0.6 •
"t:I

E o.s
do

do
0.4 • •
. Ethanol

.-..". -.
"t:I
C)
• Methanol
Acetonitrile

.~a~ •• •-
~ 0.3
D\cetone
'; 0.2
E

"t:I 0.' t:t t:t
t:t t:t
a;
:;:: o.
0 200 400 600 800
TIme (min)

B Silychristin

"t:I
4.5

,,

do
do
<I>
4
"t:I
do

!
3.S
• •
a•
. Ethanol
do
"t:I
2.5 • Methanol
C)

Z
U
2 • Acetonitrile
cet one
!II 1.5
C)
E
"t:I
a; 0.5
:;::
200 400 600 800
limo ( mi n)

C SlIydlanln

"t:I 0.6
g:
••
<II
O.S
a1 a
~do 0.4 a a . Ethanol
"t:I t t:t
• - Methanol

• :.. m
~ 0.3 f\ Acetonitrile
z
5l a IPAcetone

• •• • •
0.2
C) t3
E 0.'
"t:I ~.
a;
>= o•
0 100 200 300 400 500 600 700
n me (min)

Fig. 3. Silymarin yields (mg of compound/ g of defatted seed) as function of time in


ethanol, methanol, acetone, and acetonitrile at their normal boiling points. (A) Taxifolin
(TAX); (B) silychristin (SeN); (C) silydianin (SON). Continued on next page.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Solvent Extraction of Milk Thistle 897

0 Silyb inin A

al 4.5
:
. 4.0 •

'tI
3.5
:§. 3.0 • . Ethanol
'tI

~
2.5 • • Methanol

-
2.0 Acetonitrile
cI:
Z t:¥.cetone
CD 1.S

-
(/)
CI 1.0
E
'tI 0.5
iii
0.0
>=
0 200 400 600 800
Time ( min)

E S lIyb lnl n B

'tI
$ 8.0
co
'tI
!!!
7.0

.
~ 6.0
S.O • • . Ethanol

'tI

~ 4.0 - Methanol
til Acetonitrile
z 3.0 cet one
til
(/)
CI 2.0
E 1.0
'tI
iii 0.0
>= 0 200 400 600 800
Time (min)

Fig. 3. Continued from previous page. (0) silyhinin A (SBNA); (E) silyhinin B (SBNB).

Extraction with Organic Solvents


Figure 3 shows typical results from the extraction of taxifolin,
silychristin, silydianin, silybinin A, and silybinin B, as the yield of each
compound as a function of time in each solvent. As expected, the extraction
of each of the silymarins showed a general increase in yield with time,
regardless of solvent, as equilibrium between the extracted compounds
and the solvent was approached. Some potential degradation of silymarins
was observed when extracting with ethanol (as noted by the decrease in
yield at long extraction times), a trend that was not observed when extract-
ing with the other solvents. In comparing the yields of the individual
silymarins, it is observed that ethanol was clearly the preferred solvent for
extraction, followed closely in order by methanol, acetonitrile, and acetone.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


):..
:g A Taxifolin I Silybinin B B Silychristin I Silybinin B
[
O:l 0.1 CD 1.4
0'
r') CD
:>- z ~ 1.2 ~ ••••• •
<1l In 0.08
VI ••••••• • Ethanol • Ethanol
:3 I • E
tn· g' 0.06 • • Methanol -• 0.81 •• • Methanol
~ •• ••••• • • z
:il 0.6
i ···,,_ .. 1\ 1\1\. •
, ..
'.' ~

0.04 • Acetonitrile Acetonitrile


CI
-~
X
'":oJQ..
CI • XX X X Acetone g 0.4 Acetone
O:l g 0.02 (v X X
0' 3 0.2
iii 3 1;
r') 1; 0 , a:
:>- a: 0
:oJ
o 0 200 400 600 800 0 200 400 600 800

~ Time (min) Time (min)


--

00
<..0 Silybinin A I Silybinin B
00 C Silydianin I Silybinin B 0
CD 0.12 CD 1.2
z z
(I
ID •• •
In 0.1 VI I
VI .-
• en 1 • Ethanol
• Ethanol 1.: 1 =...
~ 0.08 • • •,. <!< • E
p¥ 0.8 .i · • Methanol
,- • Methanol c
~.,
~
- 0.06
Acetonitrile
-
z
ID
0,6 ~.~N"'7"'i".H~""jj .... ,..
Acetonitrile
VI
~ 0.04 • en 0.4
, Acetone Acetone
a
E
0.02
..... ".. . g 0.2
3
ii 0
0
ia:
6- a:
:- 0 200 400 600 800 0 200 400 600 800

Time (min) Time (min)


~
,~
N Fig. 4. Compound ratio as function of time in ethanol, methanol, acetone, and acetonitrile at their normal boiling points. (A) Taxifolin
§ to silybinin B ratio; (B) silychristin to silybinin B ratio; (C) silydianin to silybinin B ratio; (D) silybinin A to silybinin B ratio.
Solvent Extraction of Milk Thistle 899
The maximum yield of taxifolin extracted with ethanol as the solvent was
0.6 mg/ g of defatted seed, while the maximum yields of taxifolin with
methanol, acetonitrile, and acetone were 0.3, 0.2, and 0.1 mg/ g of seed,
respectively. The maximum yields of silychristin were 4.0, 2.1, 1.6, and
2.5 mg/ g of defatted seed in ethanol, methanol, acetonitrile, and acetone,
respectively. In comparing these yields with the yields obtained in a 4-h
Soxhlet extraction with ethanol on defatted seeds (0.22 mg of taxifolin and
1.40 mg of silychristin/ g of defatted seed), it is seen that the yields of
taxifolin and silychristin actually exceeded the yields in the 4-h Soxhlet
extraction, which further illustrates the potential for silymarin degrada-
tion. The maximum yields of silydianin in the solvents were all about
0.4 mg/ g of defatted seed (23% of the Soxhlet results), while the maximum
yields of silybinin A in ethanol, methanol, acetonitrile, and acetone were
4.0 (49%), 1.9 (24%), 1.7 (21 %), and 2.0 (25%) mg/ g of defatted seed, respec-
tively. Finally, the maximum yields of silybinin B in ethanol, methanol,
acetonitrile, and acetone were 7.0 (52%),3.2 (24%),3.1 (23%), and 3.2 (24%)
mg/ g of defatted seed, respectively. The yields of silymarins extracted in
ethanol in the present work compare well with the results of pressurized
liquid extraction (9).
It is interesting to compare the ratios of the extracted compounds in
the various solvents. Figure 4 shows the ratios of taxifolin and of each of the
silymarins to the quantity of silybinin B, extracted from defatted seeds, as
a function of time for each of the solvents at their normal boiling points.
Excluding the first few minutes of extraction, the ratios of taxifolin to
silybinin B for acetone, acetonitrile, methanol, and ethanol remained rela-
tively constant at 0.02, 0.05, 0.06, and 0.08 mg/mg, respectively. The ratios
of silychristin to silybinin B remained constant at about 0.7 mg/mg for
ethanol, acetonitrile, and acetone and gradually increased from 0.8 mg / mg
and then held at 1.2 mg/ mg for methanol. Likewise, the ratios of silybinin
A to silybinin B held constant at 0.6 mg/mg for ethanol, acetonitrile, and
acetone, while increasing slowly from 0.6 to 1.0 mg/mg for methanol. The
ratio of silydianin to silybinin B varied from 0.02 to 0.1 mg/mg for the
solvents but generally held in the range of 0.08 mg/mg.
There are three possible rate-controlling steps in the extraction of com-
pounds from a solid matrix: (1) overcoming an energy barrier such as the
passage through a cell wall or desorption from a surface, (2) transport through
the matrix by diffusion through the bulk material or its pores, or (3) removal
by partitioning into the solvent. Basile et al. (11) showed that the controlling
mechanism for the extraction of monoterpenes, sesquiterpenes, waxes, and
lipids from rosemary using superheated water was the partitioning of the
compounds into the solvent. They also found a rough correlation between
the solubility of the compounds in the solvent and the extraction rates. Goto
et al. (12) noted that compounds that have similar structures are extracted
similarly. Each of the silymarin compounds in the present study are similar
in structure and in their solubilities in the extracting solvent. Thus, similar
extraction rates should be observed, which is indeed the case.
Applied Biochemistry and Biotechnology Vol. 105-108,2003
900 Wallace et al.
Table 1
Ratios of Taxifolin, Silychristin, Silydianin, and Silybinin A
to Silybinin B in Ethanol, Methanol, Acetonitrile and Acetonea
Taxifolinl Silychristinl Silydianinl Silybinin AI
Solvent silybinin B silybinin B silybinin B silybinin B
Ethanol (defatted seeds) 0.09 0.48 0.07 0.59
Ethanol, (whole seeds) 0.04 0.73 0.02 0.56
Methanol (defatted seeds) 0.06 1.20 0.08 1.00
Acetonitrile (defatted seeds) 0.05 0.06 0.08 0.60
Acetone (defatted seeds) 0.03 0.60 0.10 0.60
"Ratios were calculated at time 240 min. Ethanol extractions were conducted on defatted
and whole seeds.

A clearer view of the ratios of the extracted compounds can be seen in


Table 1, where the ratios are shown after an extraction time of 240 min, a
time prior to any compound decomposition. The ratios are nearly constant
except for two notable exceptions: methanol extraction shows significantly
higher silychristin/ silybinin Band silybinin A/ silybinin B ratios in com-
parison with the other solvents. Differences in the taxifolin/ silybinin B
ratios were also observed but appear to be more closely related to experi-
mental scatter owing to low extract concentrations. These observed differ-
ences in the ratios with methanol point toward methanol being a better
solvent if silychristin and/ or silymarin A are the desired compounds for
extraction. This point was apparently also observed by Benthin et al. (9),
who recommended methanol as the extraction solvent in pressurized liq-
uid extraction.
Extraction of Whole and Defatted Seeds
Recommended extraction procedures from the literature emphasize
the importance of removing the lipids from milk thistle seeds, prior to
silymarin extraction. Petrol (9) and hexane (10) have been suggested; how-
ever, no information has been reported on the advantages of a defatting
pretreatment step. Hence, extraction experiments were performed on
whole and defatted seeds using ethanol as the solvent at its normal boil-
ing point. Figure 1 shows liquid chromatograms for the whole and defat-
ted seeds, obtained under identical extraction conditions. As noted, the
chromatogram patterns are identical except for the larger areas of the
silymarin compounds obtained while extracting defatted seeds. This
result is better illustrated in Fig. 5, where the yields of silymarin com-
pounds (mg/ g of equivalent whole seed) are plotted as a function of time
for both whole and defatted seeds. In comparing the maximum yields of
the silymarins, defatted seeds yielded on average twice the quantity of
compounds obtained for whole seeds under identical extraction condi-
tions. The most striking difference was with silydianin (5X), and the least

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Solvent Extraction of Milk Thistle 901

A Taxlfolln

-
0.7
CI
0,6 •
~
0,5 •
..
I-~

Clal
EO 0 ,4
~
• • , . Defatted - Ethanol
!5al
;:::: 0.3 • Whole - Ethanol

• ••• •
.....
~~
;'C
-0
0.2
U
!5
()
0.1
••
,
• • •
0
0 200 400 600 800
TIme (mn)

B Silychrlatin

4.5


CI
4.0
z
() 3,5
• •

rn~

e>al
EGO
~ . 3.0
2.5
•• • , . Delatted - Ethanol ,

--;'C
§~
~.;
2.0
1.5

•••
•••• • • • Whole- Elhlnol

u
1.0
••
!5
()
0,5
0.0
0 200 400 600 800
TIme (min)

C SlIydlanln

-
0 ,6
e>
z
c
rn~
O.S
••
~
e>il
EGO . 0.4
• , . Defatted - Ethanol ,
!5al
-::
10 to
0 ,3
• • Whole- Ethanol
_~-GO 0 ,2
;'C
• • ••• • ••
• .. ,. •• • •
u 0.1
!5
()
0
0 200 400 600 800
TIme (min)

Fig. 5. Comparison between ethanol extraction of whole and defatted milk thistle
seeds. Silymarin yields are expressed as milligrams of compound/ gram of defatted
seed as a function of time. (A) taxifolin; (B) silychristin; (C) silydianin.
Continued on next page.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


902 Wallace et al.

D SlIyblnln A

4.5

CJ>
4 .0
<
Z 3.5


III "0
(/)a> 3.0
CIa>
Em 2.5 • 1~ Defaned - Eth.1 nol

..
-;;"5: 2_0
••
•••• •• ••
0:::: • Whole- Ethanol

.:•
;:.,
fa; 1.5
E"O 1.0
a>
<>
c
0
0.5 •
U 0_0
0 200 400 600 800
TIme (min)

E SlIyblnln B

-
8.0
CI
CD
7,0


Z
CD 6.0

(1)"0
CI: 5.0
Em
• I.Defa ned - Etha nol

.. 4.0
~"O

•• •
ca> • Whole - Ethanol
0::::
.,-a>
;: 3.0
• •• • • • •
••••• •
~
E"O 2.0
'c<>" 1.0 •
0
u 0.0 •
0 200 400 600 800
TIme (min)

Fig. 5. Continued from previous page. Comparison between ethanol extraction


of whole and defatted milk thistle seeds. Silymarin yields are expressed as milli-
grams of compound/ gram of defatted seed as a function of time. (D) silybinin A;
(E) silybinin B.

striking was with silychristin. The compound ratios (silymarin compound


to silybinin B) in Table 1 for whole and defatted seeds show that only
small differences were observed.
In extracting menthone, I-menthol, menthyl acetate, and other essen-
tial oil components from peppermint, Goto et aL (12) noted that the extrac-
tion rates were slowed because essential oil components must be desorbed
from lipid in which the components are absorbed. Fat removal apparently
helps to release the compounds from the plant matrix without preferen-
tially affecting the release of the individual compounds. It is possible that
pretreatment protocols for fat removal may affect product ratios.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Solvent Extraction of Milk Thistle 903

Conclusions
In extracting defatted milk thistle seeds with organic solvents, extrac-
tion with ethanol resulted in the highest silymarin yield, although some
potential degradation was observed. The maximum yields of taxifolin,
silychristin, silydianin, silybinin A, and silybinin B in ethanol were 0.6, 4.0,
004,4.0, and 7.0 mg/ g of defatted seed, respectively. However, if silybinin
A were the diastereoisomer of choice, methanol would be the preferred
extraction solvent because it yielded the highest silybinin A-t<rSilybinin B
ratio. Interestingly, lipid removal is an important extraction step, because
defatted material yields twice the silymarin concentration.
Future work in extracting silymarins from milk thistle with organic
solvents will focus on temperatures below the normal boiling point. Less
compound degradation should occur as the temperature is lowered. Multi-
stage extraction will also be considered in an effort to improve silymarin
yields.

References
1. Duke, J. (1999), J. Med. Food 2,73-76.
2. Zi, X. and Agarwal, R. (1999), Proc. NatI. Acad. Sci. USA 96, 7490-7495.
3. Sonnenbichler, J., Scalera, F., Sonnenbichler, I., and Weyhenmeyer, R. (1999),
J. Pharmacol. Exp. Ther. 290, 1375-1383.
4. "Top US Selling Herbs in 1999 and 2000," Nutrition Business Journal, San Diego, CA
5. Wallace, S., Carrier, D. J., Beitle, R., Clausen, E. and Griffis, C. (2002), Journal of
Nutraceutricals, Functional and Medical Foods, submitted.
6. Baker, A., Brymer, c., and Borrie, M. (2002), Geriatr. Today 5,13-16.
7. Gurley, B., Gardner, S., and Hubbard, M. (2000), Am. J. Health-Syst. Pharm. 57,
963-969.
8. Hamid, S., Sabir, A, Khan, S., and Aziz, P. (1983), Pakistan J. Sci. Ind. Res. 26,244-246.
9. Benthin, B., Danz, H., and Hamburger, M. (1999), J. Chromatogr. A 837, 211-219.
10. Kahol, A, Singh, K., Tandon, S., and Kumar, S. (2001), Indian patent no. 06309678.
11. Basile, A, Jimenez-Carmona, M. M., and Clifford, A A (1998), J. Agric. Food Chern. 46,
5205-5209.
12. Goto, M., Sato, M., and Hirose, T. (1993), J. Chern. Eng. Jpn 26(4),401-407.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0905/$20.00

Effect of Lidocaine on Ovalbumin


and Egg Albumin Foam Stability

VORAKAN BURAPATANA, ERNEST E. BUTLER,


GAURAV CHAUHAN, SEAN HARTIG, HELEN KINCAID,
TONG WANG, SHAYRIZAL SAMSUDIN, AND ROBERT D. TANNER*
Chemical Engineering Department,
Vanderbilt University, Nashville, TN 37235,
E-mail: rtanner@vuse.vanderbilt.edu

Abstract
Foam fractionation is a simple separation process that can remove and
concentrate hydrophobic molecules such as proteins, surfactants, and or-
ganic wastes from an aqueous solution. Bovine serum albumin and ovalbu-
min have been widely used as model proteins due to their strong foaming
potential and low price. Here, we study the effect of lidocaine on albumin
foam, since drugs like lidocaine are known to bind with albumin. We ob-
served that lidocaine not only enhances the amount of foam produced but
also the stability of that foam as well. The foam stability was evaluated as the
decay rate constant of the foam, determined from a change in height (or
volume) of the foam over a given time period.
Index Entries: Foam; egg albumin; ovalbumin; lidocaine; foam stability.

Introduction
Foam fractionation has been in use in some forms since the early 1960s.
Its practical application was for the removal of surfactants in waste treat-
ment plants. The underlying process is the removal of surfactant molecules.
Another possible application for foam fractionation is to produce a stable
foam for extinguishing fires while providing a medium for breathable
oxygen. Du et a1. (1) describe the process as follows:
Foam is a gas-liquid dispersion system in which liquid is considered the
continuous phase while gas bubbles are the noncontinuous phase. Foam
fractionation is a separation technique based on the surface activity of
solutes in a solution. Foam fractionation is carried out in a column that
consists of two parts separated by a distinct interface during foaming.
"Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 905 Vol. 105-108,2003


906 Burapatana et al.
The lower part is the bulk liquid solution while the upper part is the
foam phase.
There are two aspects that need further attention in order to develop
a successful foam fractionation. The first is foam drainability and the sec-
ond is foam stability. Drainability is defined as the ability of the bulk liquid
residue to flow away from foam formation. Surface tension plays a major
role in the drainage mechanism. Foam stability is defined as the resistance
to drainage, without bubble rupture. The foam must be stable enough to be
removed from the bulk solution.
Several factors influence foam stability, such as the initial solute con-
centration, pH, and ionic strength. Column conditions, such as gas velocity
and bulk liquid height, also play significant roles. Ovalbumin, the pre-
dominant protein in egg albumin to be foamed in the present study, con-
tains hydrophobic end groups and hydrophilic head groups, making it a
surfactant. Surfactants can interact with water in a variety of ways, each of
which disrupts or modifies the hydrogen-bonding network of water. When
a high concentration of ovalbumin, the surfactant, is placed in water, the
long, nonpolar hydrocarbon tails tend to aggregate because of favorable
intermolecular interactions of hydrophobic ends. The surfactant molecules
thereby organize themselves into three-dimensional spheres called micelles
that have a hydrocarbon core and sulfate groups around the outer surface.
Surfactants can also form other structures. Rather than form a sphere, some
surfactants can coat the surface of the water to form a layer one molecule
thick (a molecular monolayer) at low concentrations.
It is known that many drugs, such as lidocaine, can nonspecifically
complex with proteins such as those in egg albumin (2). In our previous
experiments, lidocaine was found to increase albumin foam volume. It was
hypothesized that lidocaine would stabilize the foam, or decrease the
bubble size. Since bubbles are the carriers of proteins, bubble size is
expected to have an influence on foam fractionation. By stabilizing the
foam, the film thickness of each bubble increases and therefore will not
rupture or coalesce with neighboring bubbles. Liquid flow or drainage and
coalescence cause the foam to be unstable. Bubble size characteristics are
outlined in ref. 3 as follows:
1. Bubble size increases as air flow rates increase in a column.
2. Increase in protein concentration leads to reduced bubble diameter.
3. Finer air stones produce finer bubbles, however the air stone pore
size loses significance with increased air flow.
Smaller bubbles are effective in transferring proteins because of high
surface area-to-volume ratio and slower rise in the column.
It was also found that the fractionating bubbles developed a "skin" as
the proteins and surfactants were removed from the water (3). These skins
were formed by the dissolution of the hydrophobic groups into the bubble
surface while the hydrophilic ends remain in the water. This leads to a
stable foam that acquires these skins as the foam rises. The skins are pushed
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Effect of Lidocaine on Foam Stability 907
up the column through accumulation and collected for removal. The sur-
face layer also has a high viscosity (3). According to Du et al. (1),
Foam fractionation has a drawback in that protein denaturation can occur
(the loss of bioactivity of the protein). Clarkson et a1. studied protein
denaturation mechanism in foams and found that protein damage is
mainly due to surface denaturation at the gas-liquid interface. Surface
tension measurement and changes in protein structure occurring at the
interface showed that conformational changes were induced but no frag-
mentation or dissociation of subunits occurred in these foams. Denatured
proteins may alter their foaming properties during the foaming process
because structural changes can change their adsorption properties and
surface properties, thus, the foam stability and foam ability.
The specific objective in the study is to measure the foam decay time
for various concentrations of lidocaine in ovalbumin aqueous solutions in
order to measure the stability of the foam as a function of the lidocaine
concentration.

Materials and Methods


Solutions and Equipment
Ovalbumin protein was purchased from Sigma (lot no. 19H7002;
St. Louis, MO). Egg albumin was purchased from Matheson Coleman &
Bell (lot no. CB 951). A 20 mg/mL lidocaine HCl solution from Abbot was
used. The glass column used was 52 cm high with a 13-cm id. The apparatus
was the same as previously used by Du et al. (1). In addition, a 1-L polypro-
pylene graduated cylinder with a 6-cm id was used as a foaming column.

Foam Oecay
An ovalbumin solution of 60 mg/L was made by dissolving 240 mg of
ovalbumin in 4000 mL of water. The ovalbumin solution was then added
to the glass column. The stopcock at the bottom of the column was kept
closed until air was fed to the system. Once the solution was added, the inlet
airflow rate was set at 85 L/ min and the stopcock was turned to the" open"
position. For consistency, the foam was generated for 3 min. At 3 min, the
height was measured with a ruler. With no aeration, height measurements
were then taken every minute as the foam collapsed. When the foam
reached approx 4 cm, the foam decay process was stopped. This procedure
was repeated for various levels of lidocaine over the range of 0-1.6 mL from
a 20 mg/mL stock solution over 0.125-mL increments for a constant oval-
bumin concentration (60 mg/L).
Foam decay experiments were then conducted in a 1-L polypropylene
graduated cylinder using egg albumin. The albumin solution of 60 mg/L
was made by dissolving 240 mg of egg albumin in 4000 mL of water. To
create a foam in the closed-bottom polypropylene cylinder, a sparger con-
nected to the air hose was inserted from the top of the cylinder (like in a fish
Applied Biochemistry and Biotechnology Vol. 105-108,2003
908 Burapatana et al.
80

70
• Egg Albumin (100 mgfL)
60 -------i
GI
E 50 - - - - - - - i • Egg-Albumin + 1ml
:::0
;§! Udocaine HCI

.
.!!
40
o Water + 1ml Udocaine
..
E 30
0
'"- 20
- - - - - - - i HCI added
- - - - - - - i • Water + 2mIUdocaine
HCladded
10

0
Solution Composit ion

Fig. 1. The bar graphs show the effect of adding lidocaine to 100 mg/L of albumin
solution before foam is formed .

tank). For each run, 200 mL of albumin solution was used. After foaming
up to the BOO-mL mark in the cylinder, the air was turned off. Foam height
measurements were taken at approx 1 min intervals to monitor the foam
decreases in height.
Measurement of Surface Tension
The surface tension of egg albumin was directly determined using a
KSV Sigma 70 surface tension/contact angle meter. This device is fully
computer controlled. The measurements made in this experimentation
were based on the Wilhelmy plate method (4).

Results and Discussion


When lidocaine was added, it caused an increase in the foam volume
(from a 100 mg/L initial albumin solution) from 9.5 to 69 mL, as shown
in Fig. 1. Figure 1 shows that lidocaine by itself did not foam. The great
increase in foam volume from the addition of lidocaine seems to result
from the interaction between lidocaine and egg albumin protein. The
addition of lidocaine to ovalbumin, a major protein in egg albumin,
reduced the surface tension of the solution (Fig. 2). A lower surface ten-
sion, in general, indicated a higher ability to foam. The foam decay rates
for different increments of lidocaine are shown in Fig. 3. The data were fit
to first-order height decay curves,
dH =-kH
dt
and the time constants were evaluated by fitting the data using the Excel
computer program. The rate constant (k) fluctuated as the lidocaine levels
decreased. Figure 4 displays the rate constants generated by Excel from
Fig. 3 as a function of lidocaine volume added to the 60 mg/ mL ovalbumin
solution. The rate constants also characterize the stability of the foam. The
Applied Biochemistry and Biotechnology Vol. 105-108,2003
Effect of Lidocaine on Foam Stability 909
70
:[ 65
z
S 60
~
c
0
'iii 55
c
.!
50
....
ID
u
m
't:
::J 45
III
40
0 0.5 1.5 2
Lidocaine added (mL)

Fig. 2. Change in surface tension with added lidocaine. Initially, 60 mg/L of oval-
bumin was present.

25.00

20.00

e '~
". ..
." ,- ..
.e. 15.00 .. 0 ml lidocaine added
E
g
iii
~ "" .,.\ • 0 50 mllidocalno added
I'- ~\ • 1.00 mllidocaino added
o 10.00 ~ • • .. 1.50 ml lidocaloo added

~

.. ,
0>
~
~
,~
..
~

5.00

\. .. I- ... "lI.
I

0.00
o 200 400 600 800 1000 1200 1400 1600
nme(aeconda)

Fig. 3. Changes in height of foam at different lidocaine concentrations are shown.


Ovalbumin was kept constant at 60 mg/L.

smaller the rate constant, the longer the foam remained in the column and,
hence, the more stable the foam. The decay rate profile appears to have
three characterizing intervals. The first interval is from 0 to 0.4 mL of
lidocaine added. In this interval, the decay rate is constantly decreasing.
The second interval is from 0.4 to 1.2 mL of lidocaine added. Here, the decay
rate seems to stabilize and hold constant. In this interval, the solution
seemed to be "saturated" with the amount of lidocaine added. A different
effect is observed in the third interval (from 1.2 to 1.6 mL of lidocaine
added). The foam is most stable when 0.375 mL of lidocaine is added. The
decay rate was used instead of the leveling time because the leveling time
Applied Biochemistry and Biotechnology Vol. 105-108, 2003
910 Burapatana et at.
0.0060

~
0.0050

0.0040 \ rj \
.....

~
...
\
C
I:
!1!III
0.0030

1\ It
I:

~
0
0
>-

\ r ~... J
III
CJ
0.0020
«II ~~
0
0.0010 /
~ ~
0.0000
0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800
Lidocaine added (ml)

Fig. 4. Foam decay constants for increasing amounts of lidocaine added to a 60


mg/L ovalbumin solution. A lower decay constant represents a more stable foam.
The foam was most stable when 0.375 mL of lidocaine was added to the 60 mg/L
ovalbumin solution.

29

.
28 • w/o lidocaine
E • • I
-
27 • w/1 mllidocaine
26
,£. •
~ 25
Ai 24 • •
l: 23
E
~ 22 •
II.
21 •
20
o 5 10 15 20
Tlme(mln)

Fig. 5. Foam decay time in polypropylene column. The egg albumin concentration
was at 60 mg/L. One milliliter of lidocaine was added. The decay rate constant for the
foam without lidocaine addition was 0.0956/ s, and for the foam with 1 mL oflidocaine
added was 0.0125/s, indicating a 7.6-fold enhancement in foam stability.

was taken from only one data point. The leveling time is defined as the time
it takes for the foam height to decrease by 63% (1 - 1/ e). Because of the
random collapsing pattern of the foam, one point would not be a good
representation of the foam-decaying behavior. Figure 5 shows the effect of
lidocaine on foam decay in a polypropylene column. The foam with
lidocaine took about six times longer to reach the same height as the foam
without lidocaine.
Applied Biochemistry and Biotechnology Vol. 705-108, 2003
Effect of Lidocaine on Foam Stability 911

For the purposes of this experiment, the foam collapse rates were fit
to first-order decay curves. Each set of data fit first-order decay curves
with R2 values >0.95. In addition, a new variable, L, or foam-leveling time
was defined. In the glass column part of the experiment, it was observed
that after one duplicated run, the error in measuring foam decay time was
80%. In the plastic column, the data also had a lot of uncertainty. The decay
rate varied about 30% from the mean value. The plastic column diameter
was 2.7 times smaller than the glass column. In both columns, the foam-
collapsing pattern was random. The level of the foam initially decreased
at a constant rate. Then, the random pattern started to have a very strong
effect. Although the height might not change as fast, the number of bubbles
was decreasing. The liquid fraction of the foam was also an important
factor. Wetness of the foam is a function of liquid fraction of the foam.
From the experiment, the wet foam (high liquid fraction) tended to stick
less to the wall. The dry foam (low liquid fraction) created many problems
during the experiment because it would form a layer around the wall. The
addition of lidocaine increased the foam decay time. The foam from
60 mg/L egg albumin solution usually took about 3 min to decrease
200 mL (out of 800 mL initial foam). The foam from 60 mg/L initial albumin
solution with the addition of 1 mL of lidocaine took 18 min to decrease the
volume by 200 mL. The addition of lidocaine decreased the surface tension
of the solution; therefore, it was much easier to foam. At the same gas
velocity, the foam with lidocaine had smaller bubbles. Smaller bubbles
suggested that the foam contained more water. Wet foam was more stable
than dry foam.

Conclusions
The addition of lidocaine to ovalbumin and egg albumin aqueous
solutions not only created much more foam, but this wetter foam was more
stable. There appears to be a peak in foam stability at a lidocaine concentra-
tion (0.375 mL of lidocaine added) where the foam decay rate constant
reached a minimum. For both glass and polypropylene columns, the
lidocaine generally stabilizes the foam by creating smaller bubbles at cer-
tain concentrations (0.375 mL of lidocaine added) of lidocaine while larger
bubbles were seen at other concentrations.

References
1. Du, L. P., Prokop, A., and Tanner, R. D. (2002), Appl. Biochem. Biotechnol. 98, 1075-1O9l.
2. Goldstein, A., Aronow, L., and Kalman, S. M. (1974), in Principles of Drug Action: The
Basis of Pharmacology, 2nd ed., John Wiley & Sons, New York, NY.
3. SMCM (2002) Foam Fractionation: The Ins and Outs, St. Mary's College of Maryland, St.
Mary'S City, MD; (Website: www.smcm.edu/academics/smp/archive/plauger/
berlin.htm). Accessed December 17, 2002.
4. (1997) KSV Sigma 70 Experiments Reference, August I, 1997, p. 17. (Also at Website:
http://www.KSVLTD.Fi/Products/Surface%20chemistry /Sigma70 / sigma70.html)

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Copyright © 2003 by Humana Press Inc.
All rights of any nature whatsoever reserved.
0273-2289/03/105-108/0913 /$20.00

Estimation of Solubility Effect


on the Herbicide Controlled-Release Kinetics
from Lignin-Based Formulations

FELIX M. PEREIRA, ADILSON R. GONc;ALVES, ANDRE FERRAl,


FLAvlO T. SILVA, AND SAMUEL C. DLIVEIRA*

Departamento de Biotecnologia,
Faculdade de Engenharia Qufmica de Lorena, CP 116, 12.600-000,
Lorena, SP, Brazil, E-mail: scoliveira@debiq.faenquil.br

Abstract
Understanding the main phenomena involved in the controlled-release
kinetics of herbicides in a water bath is a very important requisite for dif-
fusive-parameter estimation, because, some mathematical models based
on Pick's second law for diffusion have been developed to describe the
controlled-release kinetic data. However, the validity of these models is
restricted to the following assumptions: (1) the formulation is an isother-
mal slab; (2) the release occurs through the two faces of the slab; (3) the
herbicide is dissolved in the water contained in the slab pores at a concen-
tration less than the saturation concentration (Cis); (4) the total sum of the
individual volumes of the pores is EAL (E is the slab porosity, A is the slab
area, and L is the slab thickness); and (5) the initial concentration of herbi-
cide in the pores is Mol EAL (Mo is the initial amount of herbicide in the
matrix). The fourth assumption may be invalid for mathematical descrip-
tion of systems in which the herbicide concentration in the slab may be
above the saturation concentration. If this were true, the final assumption
would also be invalid, because the initial concentration of herbicide in the
pores is Cjs in this case. This work presents a study of the solubility effect on
the controlled-release kinetics of herbicides from lignin matrices.
Index Entries: Mathematical modeling; controlled-release; lignin; herbi-
cide; diffusion; solubility.

Introduction
The technology of controlled-release of herbicides is proving to be
very attractive as a solution to the problems with defensive application and

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 913 Vol. 105-108,2003


914 Pereira et al.

contamination. Polymeric and macromolecular matrices are often utilized


as supports for the herbicide, and among the various materials used as
support, lignin is an interesting alternative since it can be obtained from
most agroindustrial residues such as sugarcane bagasse, kraft liquor, rice,
and wheat straws (1).
To understand the main phenomena involved in controlled-release
systems, mathematical models for describing the transport of herbicide
through the matrix and the soil, and the herbicide degradation by chemical
and biologic processes have been developed.
Some mathematical models based on Pick's second law for diffusion
(Eq. 1) have been presented in order to describe the controlled-release
kinetics data in water bath experiments (2-5).
2
iJc iw iJ Ciw
-=D -- (1)
at eff ax 2
in which Ciw is the herbicide concentration into the matrix (g/ cm3), Deff is the
effective diffusion coefficient (g/[cm2 day]), t is the time (d), and x is the
spatial coordinate (cm).
To solve the partial differential equation, two alternative boundary
conditions were assumed for the matrix surface: (1) a herbicide concentra-
tion equal to zero (2,5), and (2) the existence of a stagnant unstirred layer
of herbicide solution (2-4). The second boundary provided a good fit of the
model to the experimental data (2-4). However, if the herbicide concentra-
tion in the matrix is above the herbicide solubility, Eq. 1 is invalid for
describing the release kinetics. Since, in some cases, the herbicide solubil-
ity in the water is relatively low, and the concentration in the matrix is
high, a model considering the solubility effect is more realistic.

Materials and Methods


Herbicides Formulations
The formulations of the herbicides (2,4-D) dichlorophenoxyacetic acid,
ametryn (2 -ethylamino-4-isopropylamino-6-methyIthio-l,3 ,S-triazine),
anddiuron(3-[3,4-dichlorophenyl]-1,1-dimethylurea)withligninextracted
from several sources were prepared by a melting process. After melting,
the final formulation was ground and sieved to select a fraction granule size
range from 0.71 to 1.00 mm. The main characteristics of controlled-release
formulations (CRFs) obtained are presented in Table l.
Release Experiments in Static Water Bath System
The formulation granules were weighed and immersed in 30 mL of
deionized water. The flasks were closed and maintained at 30°C. In the
first 10 d sampling was performed by changing all the water contents in
the flasks. The concentration of herbicide in the samples was determined
by high-performance liquid chromatography (HPLC). Released herbi-

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Herbicide Controlled-Release Kinetics 915
Table 1
Main Characteristics of Formulation Used in Runs Conducted in Water Bath
Lignin production Precipitant
Symbol Herbicide Lignin source process agent
CRFl 2,4-0 Bagasse Steam explosion HCI
CRF2 2,4-0 Bagasse Steam explosion H 2SO4
CRF3 2,4-0 Eucalyptus' Kraft H 2SO4
CRF4 2,4-0 Eucalyptusb Kraft H 2SO4
CRF5 2,4-0 Eucalyptusb Kraft HCl
CRF6 2,4-0 Pinus Kraft
CRF7 Oiuron Bagasse Steam explosion HCI
CRF8 Ametryn Bagasse Steam explosion HCl

"From kraft liquor containing 14% solids.


bFrom kraft liquor containing 37% solids.
CPinus kraft lignin (INDULIN AT; Westvaco).

cide was expressed as a percentage of the initial amount added to the


formulation: MJ Mo. The release experiments in a static water bath sys-
tem were conducted with the formulations CRF1, CRF2, CRF3, CRF4,
CRF5, CRF6, and CRF7.
Release Experiments in Dynamic Water Bath System
The CRF samples were weighed and allocated in 40-mm-Iong by 18-
mm-diameter glass columns containing sintered disks in the top and bot-
tom. The system was maintained at 30°C and fed from the bottom with
deionized water at a flow rate of 2 mL/min. The effluent was collected on
the top and stored in glass flasks and was changed periodically. The herbi-
cide content in the flasks was determined by HPLC. The release experi-
ments in the dynamic water bath system were conducted with the
formulations CRF7d and CRF8d, where the subscript d indicates that the
formulation was assayed in the dynamic water bath system.
Mathematical Modeling
When the herbicide concentration in the matrix is above the herbicide
solubility, two different mechanisms occur: diffusion and dissolution. The
dissolution rate is proportional to the difference between the herbicide
solubility (Cis) and its local concentration (c iw ) (5). Thus, the release kinetics
is described by Eq. 2.
2
ac iw a Ciw (
- - = D eJJ - - 2 + k Cis - Ciw
)
(2)
at ax
in which k is the dissolution rate constant.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


916 Pereira et a/.
Table 2
Results of Parameter Estimation

CRF D' (d-1) x 105 k (d-1) X 103 ex 103a Fealea Ftabb R2

CRF1 2.1 ± 0.8 2±1 8±1 0.9968


CRF2 2.3 ± 0.9 0.0001 ± 0.03 0.07 ± 0.01 0.9966
CRF3 0.72 ± 0.2 28±9 1.4 ± 0.1 0.9952
CRF4 0.9 ± 0.4 50 ±20 2.5 ± 0.4 0.9920
CRF5 0.4± 0.3 70±40 30 ±20 0.9939
CRF6 0.3 ± 0.2 34 ± 21 1.3 ± 0.7 0.9950
CRF7 0.17 ± 0.03 0.56 ± 0.08 0.085 ± 0.003 0.9986
CRF7d 0.17 ± 0.02 0.1 ± 0.06 0.106 ± 0.001 1.055 1.583 0.9990
CRF8d 0.14 ± 0.05 40±8 2.0 ± 0.7 0.73 1.739 0.9990
a Feale is the calculated value of F for the lack of fit of the model to experimental data.
bFlab is the tabulated value of F for a confidence level of 95%.
C Experiments without replicates.

The initial and boundary conditions are given by Eqs. 3 and 4, respectively.
Ciw (x,t =0) =Cis (3)

Ciw (O,t) =Ciw (L,t) =0 (4)


The solution of Eq. 2 using the initial and boundary conditions is
given by (6)
Mt

MO(=
c" VD"k tanh[I n;-]+~ f
2D "2,2 (1-exp{-[k+{2n+l)2D"]t}) ]l (5)
[{2n+l}
V D" Jt n=O {k+{2n + 1)2D }
in which C* =2eAL Cis/(3tMO) and D' =3t2Deff/U. The D', k, and C* parameters
were estimated using the Marquardt's (7) method.

Results and Discussion


The results of parameter estimation are presented in Table 2, which
Table 2 shows inaccurate parameter estimates for 2,4-D CRFI-CRF6
according to Student's t-test. This result can be explained by the high solu-
bility of the 2,4-D in water (620 ppm). For high solubility, the herbicide
concentration in the matrix is below saturation, and the estimation of the
dissolution rate constant (k) is inaccurate, because, in this case, the rate-
limiting mechanism is diffusion.
For the formulation containing the herbicide diuron (low solubility in
water--42 ppm), assayed in static (CRF7) and dynamic (CRF7d) bath sys-
tems, the mathematical model fit was very good for CRF7d according to
Fisher's test providing significant parameter estimates according to

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Herbicide Controlled-Release Kinetics 917
0.015
0.9 +- +
0.8
0.01 + oj.

0.7 +
0.005

..
0.6 1& +- +-
i
1 +-
::I
~
0.5 r
::ii 11.4 0 l'
0.3
-0.005 • +-.... + ++
l' oj.
0.2 -0.01
0.1 1'+
0 -0.015
0 100 200 300 400 500 0 100 200 300 400 500

line (days) lime (days)

oj. CRF7 (exp) _CRF7 (mod)

Fig. 1. Plot of experimental (exp) data vs model (mod) predictions and residual
values (residual = experimental value - predicted value) for CRF7.

Student's t-test. The estimated values of D* (1.70 x 1O~ d-1) and C* (approx
1.0 x 10-4) were close for diuron formulation assayed in both static and
dynamic systems. This result indicates that the model has a physical mean-
ing, because these parameters correlate only with the formulation, inde-
pendently of the assay type. This result was not observed in previous works
using the models based on Eq. 1 (2-5).
The k parameter estimated for CRF7 (5.6 x 10-4 d-1) is lesser than for
CRF7d (1.0 x 10-3 d-1). This can be explained by the fact that in the static-bath
system, the volume of water changed daily was less than in the dynamic
bath system. Thus, the concentration of herbicide in the static bath medium
was higher than that in the dynamic bath, and the dissolution rate in the
static system was less than that in the dynamic system.
For the CRF containing the herbicide ametryn (intermediate solubil-
ity in water-18S ppm), the fit of the mathematical model was very good
according to Fisher test and provided significant parameters according to
student's t-test.
Figures 1-3 present plots of the experimental data versus model pre-
dictions, and the distribution of the residual values for CRF7, CRF7d' and
CRF8d • A good fit of the model to the experimental data was observed for
these formulations. Moreover, the residuals distribution obtained for CRF7,
CRF7d, and CRF8d indicate a randomized behavior.

Conclusion
The model considering the solubility effect on the controlled-release
of herbicide in a water bath was able to describe the experimental data for
formulations containing herbicides with intermediate (ametryn) and low
(diuron) solubilities in water. One important result obtained in this work
was the estimation of an equal diffusivity parameter for the diuron formu-
lation assayed in two different water bath systems.

Applied Biochemistry and Biotechnology Vol. 705-108,2003


918 Pereira et al.
0.025
0.9 l'
0.02
0.8 0.015
0.7 0.01 ...+ t"
o 0.6
i
0.005 .... f"
1"1" *=*=+
~ 0.5 0 + ....
~
'"
~.'"
:iii 0.4 -0.005 f" ....
0.3 ·0.01
+ ;+....
0.2 -0.015 f"
0.1 ·0.02
O~~~~~~-r-r-r-T-T~ -0.025
o 15 30 45 60 75 90 105 120 135 150 165 0 15 30 45 60 75 90105120135150165
lime (days) lime (clays)

f" CRF7d (exp) _CRF7d (mod)

Fig. 2. Plot of experimental (exp) data vs model (mod) predictions and residual
values (residual = experimental value - predicted value) for CRF7d•

1 0.02
0.9 0.015 ....

'+" '"
0.8
0.01 1'1'
0.7
0.6 0.005 f"
.... 1"+
~ 0.5
0.4 -0.005
0

l'
0.3
-0.01
0.2
0.1 -0.015
0 -0.02
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
lime (clays) lime (days)

.... CRFSd (exp) _CRF8d (mod)

Fig. 3. Plot of experimental (exp) data vs model (mod) predictions and residual
values (residual = experimental value - predicted value) for CRF8d •

For the herbicide with high solubility in water (2,4-D), the parameter
estimates obtained were not statistically significant according to student's
t-test. This fact may be ascribed to the high solubility of 2,4-D, rendering a
herbicide concentration in the matrix less than its solubility in water.
The next step in the validation of the current model is determination
of the volume of the matrix pores (EAL), in order to obtain an experimental
value for C. With the use of an experimental value for C in the model, the
estimation of D* and k will be more accurate.

Acknowledgments
This work was supported by Funda<;ao de Amparo a Pesquisa do
Estado de Sao Paulo, under process no. 99/05002-2 and by Conselho
Nacional de Desenvolvimento CientHico e Tecnol6gico.

Applied Biochemistry and Biotechnology Vol. 105-108,2003


Herbicide Controlled-Release Kinetics 919

References
1. Wilkins, R. M. (1990), in Biodegradable Polymer Methods, Controlled Delivery of Crop-
Protection Agents, Wilkins, R M., ed., Taylor & Francis, London, UK, pp. 149-165.
2. Pereira, F. M. (2001), MS thesis, Faculdade de Engenharia Quimica de Lorena,
Departamento de Biotecnologia, Lorena, Brazil.
3. Pereira, F. M., Gon"alves, A R, Ferraz, A, Silva, F. T., and Oliveira, S. C. (2002),Appl.
Biochem. Biotechnol. 98-100, 101-107.
4. Pereira, F. M., Gon"alves, A R, Ferraz, A, Silva, F. T., and Oliveira, S. C. (2001),Appl.
Biochem. Biotechnol. 91-93, 563-572.
5. Oliveira, S. c., Pereira, F. M., Ferraz, A, Silva, F. T., and Gon"alves, A R (2000),Appl.
Biochem. Biotechnol. 84-86,595-615.
6. Siegel, R A (1989), in Controlled-Release of Drugs: Polymers and Aggregate Systems,
Rosoff, M., ed., VCH, New York, NY, pp. 9-11
7. Marquardt, D. W. (1963), J. Soc. Ind. Appl. Math. 11(2),431-441.

Applied Biochemistry and Biotechnology Vol. 105-108, 2003


Author Index
A de Albuquerque, C. N., 853
de Azeredo, L. A 1., 749
Ackermann, Jr., D. M., 659 de Castro, H. F., 547
Adney, W. S., 689 Decker, S. R., 689
Agblevor, F. A, 219 de Morase, F. F., 809
Aiello, c., 715 de Sousa, F., 615
Almeida, R. M. R. G., 867 de Vrije, T., 557
Antunes, O. A c., 649 DiCosimo, R., 675
Aristiguuieta, M., 155
Arruda, E. J., 829
Atukorale, P. V., 659
E
Aulich, T. R., 843 Eiteman, M. A., 317
Eklund, R., 451
B Ezeji, T. c., 375
Bakri, Y., 737
Ballesteros, 1., 87, 141
F
Ballesteros, M., 87, 141 Farmer, J., 69
Barboza, M., 705, 867 Ferraz, A, 913
Barreto, J. F. A, 881 Ferrer, A, 155, 715
Batz, S., 219 Franden, M. A, 255
Blaschek, H. P., 375 Freire, D. M. G., 749, 757
Blum, D. L., 775 Furlan, S. A., 547
Bocchini, D. A., 393
Bon, E. P. S., 799 G
Bothast, R. J., 319 Galbe, M., 127
Budde, M., 557 Gamblale, W., 403
Bura, R., 319 Giordano, R. c., 413
Burapatana, V., 659, 905 Giordano, R. L. c., 413, 705
Burns, R. E., 437 Gisushi, A, 821
Butler, E. E., 905 Gomes, E., 393
Gon<;alves, A R., 195, 769, 913
C Gong, C. S., 471
Cabanas, A., 87 Gonzalez, A, 141
Carier, D. J., 881, 891 Gorton, L., 615
Castilho, L. R., 749 Gottschalk, L. M. F., 799
Chauhan, G., 905 Graham, P. J., 231
Chen, H., 775 Grant, D. R., 43
Cherry, J. R., 675 Gray, M. c., 179
Chiarini, E., 481 Gregg, D. J., 231
Chua, H., 581 Greenbaum, E., 303
Claassen, P., 557 Groberg, M., 375
Clausen, E. c., 881, 891 Giivenilir, Y., 677
Cockrem, M., 827
Coelho, R. R. R., 749 H
Colina, A, 715
Hames, B. R., 5
Converse, A 0., 179,515
Hamilton, J. E., 27
Hanley, T. R., 353, 383, 593
D Hartig, S., 905
Dale, B. E., 155 Hasan, S. D. M., 403
Damiano, V. B., 393 Henk, L. L., 115
da Costa, A c., 437 Heo, N. H., 567
da Silva, 1. 1., 649 Hess, J. R., 43, 423
Da Silva, M. AM., 757 Himmel, M. E., 689
Da Silva, R., 393 Ho, N. W. Y., 245

Applied Biochemistry and Biotechnology 921 Vols. 705-708, 2003


922 Index
Hokka, C.O., 867 McLaren, J. 5., 631
Holtzapple, M. T., 523 McMillan, J. D., 69
Hoskinson, R L., 43 Medeiros, V. c., 757
Hotta, T., 821 Moon, N. K., 365
Houghton, T. P., 423 Mori, M., 437
Huang, W.-c., 471 Mostafa, N. A A, 523
Hwang, B., 493
N
Negro, M. J., 87, 141
Ishii, M., 287 Newman, M., 69, 165
Ng, L. M., 581
J Nguyen, Q. A, 27, 165
Ni, H, 265
Jacques, P., 737 Nilvebrant, N.-O., 615
Jeffries, T. W., 265, 277, 631 Nitschke, M., 295
Jennings, E., 689 Nobrega, R, 799
Jeong, G.-T., 493 Novoa, V. Z., 725
Jewell, D. N., 659
Jimenez, H 5., 725
Jin, Y.-S., 277
o
Jonsson, L. J., 615 Oh,H., 603
Joshi, C. P., 17 Ohata, Hi 247
Ojeda-de-Rodriguez, G., 155
K Oliva, J. M., 87,141
Oliveira, G. G. c., 787
Kadar, Z., 557
Oliveira, RD., 809
Kang, H., 567
Oliveira, S. c., 547, 913
Kang, M. H., 265
Oliveira-Esquerre, K. P., 437
Karim, M. N., 115
Olivo, J. E., 809
Keller, F. A, 27
Olson, E. 5., 843
Kim, B. K., 365
Ortiz, R c., 725
Kim, E. K., 637
Kim, J. 5., 337 p
Kim, K. H., 165
Kim, K. I., 637 Park, D.-H., 493, 567
Kim,W.K.,637 Park, s. c., 567
Kincaid, H, 905 Pastore, G. M., 295
Penna, T. C. V., 287, 481
L Pereira, F. M., 913
Pereira, Jr., N., 649
Lacey, J. A, 205, 423 Persson, P., 615
Langone, M. A P., 757 Pessoa Jr., A, 481, 787, 853
Lawford, H. G., 457 Pettersson, P.O., 451, 505
Lee, J. 5.,567 Pilcher, L., 127
Lee, J. W., 303 Pimenova, N. V., 383
Lee, Y. Y., 337,451,505 Pinson, M. L., 659
Leite, S. G. F., 749 Priddy, S. A, 353
Li, X., 515
Li, X.-L., 775
Ljungdahl, L. G., 775 Q
Lloyd, T., 53 Qureshi, N., 375
Lo, W.,581
R
M Reczey, K., 557
Machoshvili, I. A, 287 Reimann, A, 615
Mansfield, S. D., 319 Resende, M. M., 413
Manzanares, P., 87, 141 Roberto, F. F., 245
Matsumoto, M., 247 Roberto, I. c., 787
Matsunaga, T., 247 Rodrigues, R 0., 809

Applied Biochemistry and Biotechnology Vols. 105-108,2003


Index 923

Roth, C. J., 5 Tsao, G. T., 471


Rousseau, J. 0.,457 Tucker, M. P., 165
Ruzene, D. S., 195, 769
Ryu, H.-W., 603 U
Urn, B.-H., 115
S
Saddler, J. N., 231, 319 V
Saez, F., 141 Vasavada, A, 317
Samsudin, S., 905 Vieira, M. F., 705
Santana, C. c., 827, 829 Vinzant, T. B., 689
Santana, M. H. A, 403 Vitolo, M., 787, 853
Santos, AS., 649
Sarquis, M. I., 649 W
Schell, D. J., 69 Wallace, S. N., 881, 891
Seo, D. K., 637 Wan, Y.,593
Sharma, R. K., 843 Wang, T., 905
Shaw, P. G., 43, 205, 423 Wardle, A E., 659
Shoemaker, S. P.,3 Wee, Y.-J., 603
Silva, D. P., 787 Wong, P.-K., 581
Silva, F. T., 913 Woo, J.-c., 493
Sin, S. N., 581 Wright, L. L., 3
Singh, A, 255 Wyman, C. E., 53, 101, 179, 515
Sluiter, A 0.,5
Sobral, K. A, 809 X
Soderstrom, J., 127
Sousa, Jr., R., 413 Xiang, Q., 337, 451, 505
Stambuk, B. u., 255 Ximenes, E. A, 775
Stedman, M. L., 659 Xu, Y.,853
Stuhler, S. L., 101
Sulbaran-de-Ferrer, B., 155, 715 y
Suzuki, N., 247 Yano, M.,821
Szengyel, Z., 557 Yenigiin, B., 677
Yokouchi, H., 247
T Yoo, I. S., 637
Yoon, H. H., 637
Tanner, R. D., 659,905
Yu, P. H. F., 581
Templeton, D. W., 5
Yun, J.-S., 603
Thanakoses, P.,523
Thomas, S. R., 5
Thompson, D. N., 43, 205, 423 Z
Thonart, P., 737 Zacchi, G., 127,451
Timpe, R. c., 843 Zanin, G. M., 809
Torget, R. W., 451, 505 Zazueta-Sandoval, R., 725
Toyama, H., 821 Zhang, M., 255
Toyama, N., 821 Zhu, W.,659
Trumbo, J., 219 Zollner, R. L., 403

Applied Biochemistry and Biotechnology Vols. 105-108,2003


Subject Index
A biorefinery, 843
bioremediation, 581
acetic acid, 457 biosurfactant, 295
acetone, 891 biotransformation, 649
acetonitrile, 891 breakthrough, 659
acid bubble bioreactor, 493
hydrolysis, 165, 337, 505 butanol, 375, 757
phosphatase, 677
activated
carbon, 353
c
sludge, 581 Caldicellulosiruptor saccharolyticus, 557
activation volume, 195 Candida
adsorption, 353,705 boidinii, 265
aeration, 799 succiphila, 255
affinity partitioning, 853 canister, 659
afforestation, 231 capric acid, 757
airflow, 659 carnohydrate-binding molecule, 775
alcohol, 843 carboxylic acids, 523
aldose reductase, 265 cassava flour wastewater, 295
alkali detoxification, 615 CbXYL1,265
alkalinity,567 cell recycle, 603
allergenic extract, 403 cellulase, 689, 775, 821
alumina, 809 cellulose, 337, 505, 689, 821
ammonia, 365 -binding molecule, 775
nitrogen, 567 biosynthesis, 17
treatment, 155 synthase, 17
ammonium, 843 characterization, 219
ampicillin, 705 cheese whey proteolysis,413
amylase, 247,829 chloride, 205
amyloglucosidase, 637 chitosan, 809, 829
anaerobic Cladosporium sphaerospermum, 649
codigestion, 567 clavulanic acid, 867
fungi,775 Clostridium beijerinckii BAlO1, 375
aqueous two-phase systems, 787, 853 colchicine, 821
arabinose, 457 combustion, 43,205
artificial neural networks, 413 computational fluid dynamicS, 593
aspen, 17 cone-and-plate impeller, 383
autohydrolysis, 515 continuous adsorption, 867
controlled
B -pore silica, 809
release, 913
Bacillus sp., 295 copper, 581
circulans, 393 corn
stearothermophilus, 287 fiber,318
bacteria, 581 oil,799
bagasse, 523 stover, 5, 27, 69, 115, 165,383
baker's yeast, 853 cotton gin waste, 219
benomyl, 821 crystallinity,505
biobleaching, 393 cyclodextringlycosyltransferase, 809
biochemical oxygen demand, 437
biodegradation, hydrocarbon, 725
bioenergy,43,205 D
bioethanol, 615 depolymerization, 53
biomass,247,403 desorption, 581
cellulosic,593 detoxification, 353
hydrolysate, 457 diffusion, 913

Applied Biochemistry and Biotechnology 925 Vols. 705-708,2003


926 Index
dilute gene
acid, 53, 165 cloning, 265
hydrolysis, 451 expression, 265
-sulfuric acid, 69 genomic integration, 457
dilution rate, 375 ginseng polysaccharide, 493
Drechslera (Helminthosporium) monoceras, glucose-6-phosphate dehydrogenase, 853
403 glycerol, 757
greenhouse gas, 231
E
economics, 231 H
effective diffusivity, 353 H2 production, 303
egg albumin, 659,905 pathways, 303
enzymatic photosynthetic, 303
conversion, 69 Hl04,115
digestibility, 365 H 2SO4, 115
hydrolysis, 27, 115, 127, 165, 319 heat transfer, 101
reactor, 413 heavy metal, 581
enzyme, 677 helical impeller, 383
immobilized, 809 hemicellulose, 53
purification, 677
herbicide, 913
Escherichia coli, 481
heterogeneous model, 451
ester, 843
n-hexane, 757
esterification, 843
ethanol, 87, 231, 247, 319,457, 547, 891 hexokinase, 787
bio-,615 hollow-fiber module, 603
production, 141 hybrid model, 413
/water pulping, 195, 769 hydrocarbon biodegradation, 725
experimental design, 749 hydrogenase, 303
external mass transfer, 353 hydrogen
extraction, 881, 891 bond, 337, 505
peroxide, 365
F production, 557
hydrolysis, 53, 101, 179, 515, 593
feedstock, 5 dilute acid, 451
fermentation, 141,255, 319, 523
continuous, 375
inhibitors, 615
solid state, 403 immobilized enzyme, 809
filamentous fungi, 725 inhibitors, 141
filter paper ion-moderated partition chromatography,
assay, 689 515
unit, 689 isosafrole, 649
flask-to-liquid volume ratio, 493
flavanolignans, 881 K
flocculation, 547
flow field, 593 kinetic model, 53, 69, 101,337,451,505
fluidized bed, 205 Kluyveromyces marxianus, 141, 255
foam, 905 kraft pulp, 393
stability, 905
food waste, 567, 637 L
functional link neural networks, 437 j3-lactamics, 705
fungal lactic acid, 603, 637
pretreatment, 27 Lactobacillus delbrueckii, 637
upgrading, 423 lidocaine, 905
lignin, 337, 913
G peroxidase production, 799
garlic seedlings, 677 lignocellulose hydrolysates, 615
gasification, 205 lime pretreatment, 523

Applied Biochemistry and Biotechnology Vols. 105-108,2003


Index 927
lipase, 757 Pleurotus ostreatus, 423
liquid polypropylene sheet, 659
hot water, 87 poplar, 87
-liquid extraction, 787 biomass, 141
posthydrolysis, 319
M potassium, 205
power law parameters, 383
marine prehydrolysate, 457
algae, 247 biomass, 457
bacteria, 247 pressure, 195
mass balance equations, 413 pulsation, 471
mathematical modeling, 913 pretreatment, 69, 87,101,115,165,179,
metabolic 365
engineering, 277 lime,523
flux, 277 steam,319
metal removal, 581 principal components regression, 437
methane, 567 productivity, 799
methanol, 891 projection to latent structures, 5
microbial pretreatment, 27 protease production, 749
milk thistle, 881, 891 protein, 403
molar ratio, 757 foams, 659
monocaprin, 757 purification, 481
monoglycerides, 757 purification, 705, 829, 867
multiple linear regression, 437
multivariate analysis, 5
R
N reaction volume, 195
reactor, 375
NADH,265 design, 101
NADPH,265 recombinant
near infrared spectroscopy, 5 green fluorescent protein, 481
neural networks
Zymomonas AX101, 457
artificial,413 repeated batch, 603
functionallink,437 rheological properties, 383
newspaper,365 rice straw, 155, 715
nitrogen source, 603

o s
oligomers, 179, 337, 515 saccharification, 247
Organosolv pulping, 769 Saccharomyces cerevisiae, 265,787
Orpinomyces, 775 selective harvest, 43
ovalbumin, 905 separate hydrolysis and fermentation, 127
oxygen shrinking-bed reactor, 593
sensitivity,303 silica, 43, 205
transfer, 471 controlled-pore, 809
oxygenases, 725 silymarin, 881, 891
simultaneous saccharification and
p fermentation, 127, 637
slagging, 205
Panax ginseng, 493 softwood, 127
paper sludge, 557 solubility, 179,913
partial least squares, 437 soybean, 829
Penicillium canescens, 737 sporulation, 287
petroleum contamination, 725 starch-based packing peanuts, 375
phenylboronate, 829 statistical experimental design, 403
phenylpropanoids, 649 steam explosion, 87, 219
Pichia stipitis, 547 straw composite, 423
pilot scale, 69 Streptomyces, 749
piperonal, 649 viridosporus, 799

Applied Biochemistry and Biotechnology Vols. 105-108,2003


928 Index
sugarcane wastewater treatment process, 581
bagasse, 195,523,547, 769 water, 881
molasses, 749 wheat
sugars, 155, 179 bran,403
degradation, 615 straw, 43, 423
mixture, 547 white-rot fungi, 423
sulfite oxidation, 471 whole crop utilization, 43
summative composition, 219 wood
development, 17
T waste, 231
thermochemistry, 515 X
thermophilic, 393
thermoresistance, 287 xylanase, 393, 715, 737
Thermotoga elfii, 557 bleaching, 769
three-phase-partitioning method, 481 xylan, 515
transformed hairy root, 493 conversion, 69
transpost kinetics, 255 xylitol dehydrogenase, 277
trees, 17 xylose, 255,277,457
triazine dyes, 853 xylose reductase, 265, 277
Trichoderma reesei, 689, 715, 821 y
V yeast, 457
baker's, 853
volatile fatty acids, 567 extract, 603

W z
waste activated sludge, 567 Zymomonas mobilis, 457

Applied Biochemistry and Biotechnology Vols. 105-108,2003

Vous aimerez peut-être aussi