Enzyme immobilization using a

cellulose-binding domain: Properties of

a fl-glucosidase fusion protein
Edgar Ong, Neil R. Gilkes, Robert C. Miller, Jr., R. Antony J. Warren and
Douglas G. Kilburn
D e p a r t m e n t o f Microbiology, University o f British Columbia, Vancouver, B.C., Canada

Using molecular genetic techniques, a fusion protein has been produced which contains the cellulose-
binding domain (CBD) of an exoglucanase (Cex)from Cellulomonas fimi fused to a fl-glucosidase (Abg)
from Agrobacterium sp. The CBD functions as an affinity tag for the simultaneous purification and
immobilization of the enzyme on cellulose. Binding to celhdose was stable for prolonged periods at
temperatures from 4C to at least 50C, at ionic strengths from 10 mM to greater than 1 M, and at pH
values below 8. The fusion protein ('an be desorbed from celhdose with distilled water or at pH greater
than 8. Immobil&ed enzyme columns of the fusion protein bound to cotton fibers exhibited stable fl-
glucosidase activity for at least 10 days of continuous operation at temperatures up to 37C. At higher
temperatures, the bound enzyme lost activity. The thermal stability of the Jusion protein was greatly
improved by immobilization. Immobilization did not alter the pH stability. Except for its ability to bind
to cellulose, the properties of the fusion protein were virtually the same as those of the native enzyme.

Keywords:/3-Glucosidase;cellulose:fusion protein: adsorption; enzyme immobilization;stability

Introduction ciency. We have shown previously that Celhtlomonas

fimi cellulases can be purified by affinity chromatogra-
Enzyme immobilization is an important technique in
phy on unmodified cellulose. L~'~3 Fusion of the cellu-
biotechnology. ~ Many enzymes of industrial, analyti-
lose-binding domains (CBD) of C. fimi cellulases to
cal, and clinical importance have been immobilized in
heterologous proteins gives hybrid proteins which are
or on various types of supports. 2-5 Enzymes can be
active and which bind to cellulose. ~4.~5The adsorption
physically adsorbed or covalently attached to a support
of such fusion proteins to cellulose through the binding
or entrapped within a matrix or membrane. Immobi-
domain provides a simple technique for enzyme immo-
lized enzymes may exhibit enhanced pH, temperature, bilization. ~5
storage, and operational stabilities, and altered kinetic
A fusion protein (Abg-CBDcex) comprising an Agro-
properties .6-8They may be less sensitive than the native
bacterium sp./3-glucosidase at the N-terminus and the
enzymes to inhibition by substrate and/or product. 9
CBD of C. fimi exoglucanase (Cex) at the C-terminus
Cellulose is an ideal support for immobilizing pro-
can be purified by adsorption to and elution from cellu-
teins because it is cheap, inert, stable, and readily avail-
lose. Its binding to cellulose is stable enough and strong
able in different forms. Reactive sites can be added to
enough for the adsorbed form to be used as an immobi-
cellulose, e.g., with CNBr, for the immobilization of
lized enzyme. J5 This paper describes the properties of
enzymes. 5']'1~ The principal disadvantages of this ap-
Abg-CBDc Xwhen bound to cellulose or in solution and
proach are the expense of producing such carriers and
documents the utility of this generic method for enzyme
the potential for conformational change in the immobi- immobilization.
lized enzyme, resulting in decreased catalytic effi-
Materials and methods
Strains and plasmids
Address reprint requests to Dr. Kilburnat the Departmentof Micro-
biology, Universityof British Columbia,Vancouver, B.C., Canada, Escherichia coli strains JMI09, CAG440 j6 and
V6T lW5 CAG456 j6, and the plasmid pEO115 have been de-
Received 5 February 1990; revised 23 April 1990 scribed before. For this study, E. coli strains CAG440

1991 Butterworth-Heinemann Enzyme Microb. Technol., 1991, vol. 13, January 59

Papers
and CAG456 were transformed with plasmid pEO1 us-
ing standard recombinant DNA procedures. 17
Production and purification o f Abg-CBDce
Three strains were used as sources of recombinant
Abg-CBDc~x: E. coli JM109/pEO1, E. coli CAG440/
pEO1, and E. coli CAG456/pEO1. Cells were grown at
30C in LB medium supplemented with 100/xg ampicil-
lin ml i and 0.2 mM isopropylthiogalactoside. After
overnight growth, 50 ml of cells were centrifuged at
17,400g and 4C for 10 min. The cells were resuspended
in 5 ml of 50 mM potassium phosphate buffer, pH 7.0
(phosphate buffer), supplemented with 3 mM EDTA,
and ruptured in a small French pressure cell (Aminco)
at 2,000 psi. The lysates were centrifuged at 39,200g
and 4C for 30 min to remove insoluble debris. For
large-scale production of Abg-CBDce x , 40-1 cultures of
cells were grown in a 110-1 fermentor (L. H. Fermenta-
tion) using the medium and conditions described
above. For the preparation of cell extract and the puri-
fication of Abg-CBDcex by affinity chromatography on
cellulose, the protocol described by Ong et al. 15 was
followed. If necessary, Abg-CBDc~ x was further puri-
fied by Mono Q anion-exchange chromatography using
a Pharmacia FPLC system. Enzyme preparations were
stored at 4C in the presence of 0.02% NaN 3.
Immobilization o f Abg-CBDce x on cellulose
The following cellulosic adsorbents were used: de-
waxed, absorbent cotton (Fisher); CF1 cellulose
(Whatman); AvicelTM PH-101 (FMC International), and
CellufineTM (Amicon). Cotton is a highly crystalline cel-
lulose. Its crystallinity index varies from 70% to
8 4 % . 18"19 Avicel is a microcrystalline cellulose with a
crystallinity index ranging from 64% to 81%.~9'2
Cellulose was first resuspended in water and centri-
fuged or filtered under vacuum. This procedure was
repeated once with water and twice with phosphate
buffer. Finally, the cellulose was resuspended to a fixed
volume in phosphate buffer. Sufficient enzyme to satu-
rate the cellulose, determined as described, 15was then
added. After brief mixing, the suspension was incu-
bated at 4C for 3 h with gentle agitation, then centri-
fuged. The supernatant was assayed for/3-glucosidase
activity, and the amount of activity bound to cellulose
was determined from the difference between this value
and the original activity in the preparation. The pellet
was resuspended in phosphate buffer, the mixture cen-
trifuged, and the supernatant discarded. The moist pel-
let was then kept at 4C until required.
Estimation o f soluble and immobilized
[3-glucosidase activity and protein
concentration
Purified Abg-CBDc~ x was diluted in phosphate buffer
containing 0.2 mg bovine serum albumin ml -I as a
stabilizer. /3-Glucosidase activity was estimated by
measuring the initial rate of p-nitrophenolate (pNP)
formation from p-nitrophenyl-/3-o-glucopyranoside
60 Enzyme Microb. Technol., 1991, vol. 13, January
[pNPG (Sigma)]. One unit of/3-glucosidase or pNP-
Gase activity releases I /xmol of pNP min-~ from 1.2
mM pNPG in phosphate buffer at 37C (standard condi-
tions). Cex itself has no activity against pNPGfl 1
The activity of immobilized Abg-CBDc~x was deter-
mined as follows: Cellulose to which Abg-CBDc~x was
adsorbed was pelleted in a microcentrifuge at 14,800g
and 4C for 10 min. The supernatant was removed and,
if indicated, assayed for /3-glucosidase activity as
above. The pellet was washed once with phosphate
buffer, centrifuged, and resuspended to the initial vol-
ume in phosphate buffer, pNPG was added and the
cellulose was maintained in suspension by periodic agi-
tation of the reaction mixture. After an appropriate
time, the cellulose was pelleted and the absorbance of
the supernatant determined at 400 nm.
Protein concentrations were determined by dye
binding. 22
Determination o f the affinity o f
Abg-CBDce x for Avicel
Purified Abg-CBDce x (0.01-0.54 mg ml-l) was mixed
with 5 mg of Avicel which had been washed previously
with water and equilibrated with phosphate buffer. The
mixtures were equilibrated for 3 h at 4C with constant
agitation and then centrifuged at 14,800g and 4C for
10 min to pellet the Avicel. The supernatant was re-
moved and assayed for/3-glucosidase activity. The pro-
tein adsorbed ([P]ad, mg protein mg-l) to the Avicel
was calculated by subtracting the amount of protein
found in the supernatant ([P], mg protein ml-1) from
that initially added ([P]o, mg protein ml-1). Nonlinear
Scatchard plots ([P]aj/[P] vs. [P]ad) were resolved into
linear components by a reiterative method described
by Spears et al. 23 and the adsorption constants calcu-
lated by nonweighted, linear regression analysis. The
derived adsorption parameters were substituted into
the Langmuir equation 24 to determine the closeness of
fit of the experimental data.
Stability o f Abg-CBDce x adsorbed to
cellulose
Stability of the soluble and adsorbed Abg-CBDce x as a
function of pH was tested by incubating in the appro-
priate bufferand pH. The buffers z5 (50 mM) used to test
pH stability were: KCI-HCI (pH 2); glycine-HCl (pH
3); sodium acetate-acetic acid (pH 4 and 5); ci-
trate-phosphate (pH 6 and 6.5); phosphate (mono- and
dibasic) (pH 7, 7.5, and 8); glycine-NaOH (pH 9 and
10) and phosphate-NaOH (pH 11). A saturating
amount of Abg-CBDce x was bound to Cellufine. The
immobilized enzyme was filtered to dryness on a glass
membrane filter. Cellulose carrying the immobilized
enzyme (200 mg) was placed in a vial and 4.8 ml of
buffer of the appropriate pH was added to it. The mix-
ture was resuspended with a vortex mixer and then
incubated for 3 days at 25C and 150 rev min -1 in a
shaker bath. A 0.5-ml sample of the suspension was
centrifuged; the bound enzymes washed in phosphate
buffer, resuspended in 0.5 ml of phosphate buffer, and
 A cellulose-binding fl-glucosidase fusion protein: E. Ong et al.
assayed for pNPGase activity under standard condi-
tions. Control Abg-CBDcex in solution (0.034 mg ml- 1)
was incubated and assayed as above. Bound enzyme
was also analyzed by SDS-PAGE.26 A uniform amount
of cellulose, after incubation at the indicated pH, was
extracted with SDS loading buffer. The solution ob-
tained was then loaded onto a 10% SDS-polyacryl-
amide gel. The pH stability was also tested by eluting
a column (HR 5/2 Pharmacia) containing 0.5 ml of
Cellufine suspension (50% v/v) adsorbed with 2 mg
of Abg-CBDce x with a pH gradient of the appropriate
constant ionic strength buffers 27 (50 mM sodium ace-
tate, pH 6.9 and 0.2 M HC1 + 50 mM KCI, pH 1.2; 50
mM triethanolamine, pH 7. I and 0.2 M NaOH + 50 mM
KCI, pH 12.6). Protein desorption was monitored by
absorption at 280 nm. Fractions (1 ml) were collected
and the pH of each measured at 25C.
Stability of soluble and bound Abg-CBDce x as a func-
tion of ionic strength was determined as follows: a 500-
mg sample of Abg-CBDc~x-Cellufine suspension was
placed in a vial to which 4.5 ml of deionized water
or phosphate buffer (pH 7.0) of the appropriate ionic
strength was added. The cellulose was resuspended
and incubated for 24 h at 25C and 150 rev min -~.
Subsequently, a 0.5-ml sample of the suspension was
taken and treated essentially as in the pH stability
study. Control Abg-CBDce x in solution, 0.081 mg ml-1,
was prepared in the same manner. Residual pNPGase
activities were determined under standard conditions.
Stability of the fusion protein during storage was
determined as follows: Abg-CBDc~ adsorbed to Avicel
(0.8 mg enzyme g- J of Avicel) was incubated in a final
volume of 10 ml of phosphate buffer for 7 weeks at 4C
and 37C. A 0.5-ml sample of the suspension was taken
weekly and centrifuged to pellet the Avicel. The Avicel
was washed once in phosphate buffer, then resus-
pended to 0.5 ml in phosphate buffer. The suspensions
were then assayed for pNPGase activity under stan-
dard conditions to determine residual activity. As a
control, 0.33 mg of Abg-CBDc~ x were incubated in 10
ml of phosphate buffer and assayed in the same manner
as the immobilized enzyme. In addition, all samples
were analyzed by SDS-PAGE. 26 Samples of the solu-
ble Abg-CBDce x were first concentrated by lyophiliza-
tion. After addition of SDS loading buffer to each sam-
ple, equivalent volumes were loaded onto a 10%
SDS-polyacrylamide gel.
Performance of Abg'CBDcex immobilized
enzyme columns
Cotton fibers were cut to lengths of 0.85 mm or less
with a Wiley cutting mill (A. Thomas Co.). One to two
grams of the cut cotton were washed twice with water
and once with phosphate buffer by filtration and resus-
pension. The fibers were mixed with Abg-CBDc~x in 50
ml phosphate buffer, incubated at 4C for 3 h with
shaking at 200 rev min-1, then pelleted at 17,400g and
4C for 15 min. The supernatant was removed and, in
some cases, assayed for/3-glucosidase activity. The
fibers were resuspended in phosphate buffer and
Enzyme Microb. Technol., 1991, vol. 13, January
packed into a jacketed column (Pharmacia XK 16/20).
The column was connected to a circulating water bath
(Forma) and equilibrated at the desired temperature.
Phosphate buffer was passed through the column to
elute unbound enzyme. Substrate (3 mM pNPG in phos-
phate buffer) was then passed through the column at a
constant flow rate using a peristaltic pump (Pharmacia
P3). The outflow from the column was collected in 7.5
ml fractions to which an equal volume of 1 M Na2CO 3
was added by the same peristaltic pump to inactivate
any enzyme that might have desorbed from the sup-
port. A sample of each fraction was transferred to an
ELISA plate and the pNP produced was measured by
absorbance at 405 nm. When CF1 cellulose was used
as a support instead of cotton, the same protocol was
followed, except that the CF1 cellulose was not pre-
treated mechanically.
Results
Improved yield of Abg-CBDce x in E. coli
CAG456/pEO1
E. coli CAG456 has an amber mutation in the htpR
locus and a temperature-sensitive suppressor t-RNA;
E. coli CAG440 is an htpR + isogenic variant of
CAG456.16 The htpR locus encodes the sigma factor
required for transcription of heat shock and other SOS
genes. 28 Overproduction of foreign proteins in E. coli
can stress the cells, which respond by producing prote-
ases (e.g. Lon protease) which may degrade the over-
produced proteins, z9 Some proteins are unstable when
produced in a wild-type strain, but exhibit significantly
longer half-lives when produced in an htpR strain, j6
There was a three-fold increase in total enzyme activity
(to 298 units) and a four-fold increase in the specific
activity (to 1.41 units mg -I) of Abg-CBDce x when it
was produced in CAG456 compared to JM109. The
specific activity of Abg-CBDc~x produced in either
CAG440 or JM109 was comparable. The CAG456 cells
appeared to contain more of the fusion protein as de-
tected by Western blot analysis lz (Figure 1). The low-
molecular-weight bands (<30 K) in the cell extracts (4,
5, and 6 in Figure 1) represent nonspecific antibody
binding. The same bands appeared in the cell extract
of JMl09 that did not carry the plasmid (3 in Figure 1).
The lower-molecular-weight bands in Cex (<35 K; 1 in
Figure 1) and in Abg-CBDc~ x (<20 K; 2 in Figure 1)
are degradation products formed during storage of the
enzymes. 12.15
Efficiency o f the adsorbed Abg-CBDce x
Although Avicel could be saturated with Abg-CBDcex,
the activity of the enzyme adsorbed to the Avicel was
less than expected from the activity actually removed
from solution by adsorption. The adsorbed enzyme was
about 35% as active as the enzyme in solution. The
activity of adsorbed enzyme was slightly higher at sub-
saturating levels (Figure 2). There was no significant
adsorption of Abg itself to cellulose (data not shown).
61
 Papers
12.0
0.040 e~
10.0 O
O 'o0

~ 0.020
8.0
E

g
-i
E 6.0
~ _~D 0.000
0,00
. . . .
0.20
[P] (mg.mk"1)
0.40
4.0
m
2.0
x,o
........ O O , ~n~... I
0.0 ~ , I , D n ,, , ' '
0.000 0.010 0.020 0,030 0.040
[P]ad (rngmg-1)

Figure 3 Adsorption of Abg-CBDcex to Avicel. Main panel shows

Figure 1 Western blot analysis of Abg-CBDc~x production in
various strains of E. c o i l The primary antibody used was rabbit
anti-Cex. A r r o w indicates Abg-CBDcex . Lanes 1 and 2 are purified
Cex and Abg-CBDcex, respectively; lanes 3, 4, 5, and 6 are extracts
from JM109, JM109/pEO1, CAG 440/pEO1, and CAG 456/pEO1,
respectively
a Scatchard analysis of data for Abg-CBDcex bound to Avicel (5
mg). The experimental data (open circles) were resolved as t w o
classes of binding interactions, of high and low affinity (straight
lines). Closed circles show plot given by substitution of the de-
rived parameters into the Langmuir equation. Inset shows data
plotted as adsorption isotherms: open and closed circles as
above
obvious that fusion of the CBDce x to Abg did not change
its binding characteristics significantly.

Determination o f binding interactions and Stability o f adsorbed Abg-CBDce x

adsorption parameters
The adsorption isotherms for Cex (Gilkes et al., manu-
script in preparation) and Abg-CBDce , are comparable.
At 5C, Avicel PH-101 was saturated by approximately
0.04 mg of either enzyme mg ~ of Avicel (Figure 3).
The binding of both enzymes involved a low- and a
high-affinity component. The high- and low-affinity dis-
sociation constants and the saturation constants were
very similar for Cex and Abg-CBDce~ (Table 1). It is
5,0f
4.0
0 0
f F o o
~ 0C
e 2,0
There are two aspects to the stability of an immobilized
enzyme: the intrinsic stability of the enzyme and the
maintenance of the immobilized state. The latter is of
particular interest for Abg-CBDcex because it is immo-
bilized by simple adsorption to the cellulose surface.
Abg-CBDce x was inactivated by prolonged incubation
at or below pH 5.0, either in solution or adsorbed to
cellulose (Figure 4a). SDS-PAGE analysis showed no
significant desorption from cellulose at low pH (Figure
4b). However, incubation at pH 7.5 to 9.0 resulted
in loss of adsorbed activity, although soluble enzyme
remained active under these conditions. Subsequent
analysis by SDS-PAGE showed that the loss of activity
100
was likely the result of desorption of the enzyme (Fig-
00 ure 4b). Above pH 9, the soluble enzyme was inacti-
o vated.
o
~0 These observations were confirmed by elution of a
column containing Abg-CBDce Xadsorbed to cellulose,
._~
40

1,0 20
Table 1 Adsorption parameters for interaction of Cex and
Abg-CBDcex with Avicel
0.0 ' ' ll0 210 310 40 510 6

Abg-CBDCex EquilibriumConcentration(U mL"1) Kda [PJad,mb

Enzyme High Low High Low
Figure 2 Adsorption equilibria and immobilization yield of Abg-
CBDcex adsorbed to Avicel. The a m o u n t of Abg-CBDcex used was
from 0.05 to 0.70 mg. The Avicel with adsorbed Abg-CBDcex was Cex 7,3 10 -3 12.1 10 -2 1.0 10 2 3.4 10 -2
assayed for pNPGase activity. This was the bound activity (M). Abg-CBDcex 1.5 10 -3 5.2 10 -2 1.3 10 -2 2.7 10 -2
The theoretical bound activity (T) was determined by the differ-
ence in activity found in the supernatant and the activity added a Equilibrium dissociation constant (rag protein m1-1)
initially. Immobilization yield (closed circles) is (M/T)100 b Saturation constant (rag protein mg -1 adsorbent)

62 Enzyme Microb. Technol., 1991, vol. 13, January

 A cellulose-binding [3-glucosidase fusion protein: E. Ong et al.
1.4 a 6.0
[] Solubleenzyme I
1,2 Immobilizedenzyme
I
10.0
2,0
~ 0.8 8.0

~
0.0
~o. 0.6
1.0 1 I 6.0
N
~ 0.4 60
4.0
0.2 40
2O
0.0 : : :__: : : . : m: I1: : .
2 3 4 5 6 6.5 7 7.5 8 9 10 11 20
20 40
Fraction Number
60 80 oo
A pH
Figure 5 Desorption of Abg-CBDcex from cellulose with a pH
gradient. Open circles indicate absorption at 280 nm. Closed
circles indicate pH. Experimental details are given in Materials
4 5 6 7 8 9 10 11 and methods

ImmobilizedfusionI
+ SOlublefusion I

_~,.0,~
N 05
'c1
m
~ 00
Z-~_ I 37CI

Figure 4 pH Stability of Abg-CBDcex adsorbed to Cellufine. (a)

Residual pNPGase activity of soluble Abg-CBDcex and adsorbed
Abg-CBDce X expressed relative to the value obtained at day 3
using phosphate buffer, pH 7.0. (b) SDS-PAGE of adsorbed Abg-
CBDcex after incubation at different pH values. Numbers refer to 00 20 4.0 6.0 8.0
pH. Arrow refers to Abg-CRDce x. Experimental details are given Time (weeks)
in Materials and methods
A

0 1 2~34 567 0 1 2 3 4 5 6 7
using an increasing (Figure 5a) or decreasing (Figure
i ~ B
5b) pH gradient. Protein was eluted above pH 9 and
was inactive. N o desorption was seen at low pH.
Abg-CBDce x in solution was stable for 24 h at 25C,
p H 7.0, and ionic strengths from 0 to 1 M (data not
shown). Immobilized Abg-CBDce x was also stable un-
der these conditions but was largely desorbed from C
the cellulose in distilled water (data not shown). Abg-
CBDce x adsorbed to Avicel was stable and remained
bound for up to 7 weeks at 4C and 37C. In solu-
tion Abg-CBDcex was stable at 4C but not at 37C
(Figure 6). Figure 6 Stability of Abg-CBDcex adsorbed to Avicel. (a) Residual
pNPGase activity of bound Abg-CBDce x expressed relative to
initial activity at week 0. (b) SDS-PAGE of adsorbed Abg-CBDcex.
Immobilized enzyme columns Numbers refer to time in weeks. Letters refer to different sets of
samples: A, Bound fusion at 4C; B, soluble fusion at 4C; C,
It appears that Cex binds with high affinity to the crys- bound fusion at 37C; D, soluble fusion at 37C. Experimental
talline regions of cellulose and with low affinity to the details are given in Materials and methods

E n z y m e M i c r o b . Technol., 1991, vol. 13, J a n u a r y 63

Papers
100

.0 o ....
N 6O

" [ ~o ~ 10C

g_
~ 20 i

0 i i ,
0 5 10 15 20
Time (days)

Figure 7 Performance of Abg-CBDcex immobilized enzyme col-

umns: effect of cellulose structure. A m o u n t of cellulose used
was 1 g. Enzyme loading was from 0.064 to 0.155 mg protein g-1
cellulose. Temperature of column run was 4C; flow rate was 7.8
ml h -1. Circles refer to cotton and squares to CF1 cellulose. 50CO
Phosphate buffer only was passed through the CF1 cellulose
column from days 6 to 12. Further details are given in Materials
and methods O5

00 ' ~ ~ '
0 2 4 6 8 10 ~2
T ~ e (c~ys)

amorphous regions (Gilkes et al., manuscript in prepa-

ration). Abg-CBDc~x exhibits similar behavior when
bound to CF1 cellulose and to cotton.
Abg-CBDce x was adsorbed to CF1 cellulose, which
was then packed into a column and equilibrated at 4C.
A solution of pNPG was passed through the column,
and the amount of pNP released in the effluent was
monitored. The conversion of pNPG to pNP dropped
from 70% initially to 25% during the first 13 days of
operation, but remained nearly constant during the fol-
lowing 8 days (Figure 7). When adsorbed to cotton
fibers, enzyme activity decreased only 12% in the 15-
day period (Figure 7). This suggests that, in the case
of Abg-CBDc~, adsorbed on the CF1 cellulose, the en-
zyme bound to the amorphous regions of the cellulose
was released gradually during the first 13 days of opera-
tion of the column. The enzyme tightly bound to the
crystalline regions of the CF1 cellulose was not re-
leased and accounted for the long-term stable activity.
Abg-CBDcex adsorbed to cotton was stable for at
least 10 days during continuous operation at 4C, 10C,
and 37C (Figure 8). It was inactivated completely in 3
days at 50C, but protein was not released from the
column (data not shown). Abg-CBDc~x in solution was
inactivated after 3 days at 50C. The activity of ad-
sorbed Abg-CBDc~, was highest at temperatures be-
tween 20C and 30C (Table 2). At this temperature
range, conversion of pNPG to pNP was approximately
50% even after 24 h of operation.
Discussion
The adsorption parameters derived from the Scatchard
analysis of Abg-CBDce x adsorbed to Avicel were simi-
lar to those for Cex. The affinity of the CBD for cellu-
lose is not reduced significantly when it is fused to
another protein. Furthermore, the activity of the fusion
protein adsorbed to cellulose is virtually indistinguish-
able from that of Abg itself, indicating that the Abg part
Figure 8 Performance of Abg-CBDcex immobilized enzyme col-
umns: temperature stability. A m o u n t of cotton used was 2 g.
Enzyme loading was from 0.25 to 1.5 mg protein g 1 cellulose.
Flow rate was 7.8 ml h 1. Values are expressed relative to the
initial, highest p-nitrophenolate produced at that temperature.
Further details are given in Materials and methods
of Abg-CBDcex has the same conformation as native
Abg. When a saturating amount of Abg-CBDcex was
adsorbed to Avicel and the bound activity measured,
only 35% of the bound protein remained active. This
loss of activity could result from steric hindrance. Al-
ternatively, the activity of the fusion protein might be
reduced by interaction of the active site with the fl-1,4-
glucosidic bonds in the cellulose, to the exclusion of
pNPG, i.e. cellulose could act as a competitive inhibi-
tor of Abg-CBDc~x under these conditions. When sub-
saturating levels of the fusion protein were adsorbed
to Avicel, the immobilization yield increased. Under
Table 2 Effect of temperature on the activity of Abg-CBDcex
immobilized enzyme column
Percent conversion to pNP
Temperature
(C) 5 h 20 h
4 30 30
10 38 38
20 50 50
30 48 48
40 45 25
A m o u n t of cotton used was 2 g. Enzyme loading was 0.85 mg
protein g-1 cellulose. Flow rate was 7.8 ml h -1

64 Enzyme Microb. Technol., 1 9 9 1 , v o l . 13, J a n u a r y

 A cellulose-binding [3-glucosidase fusion protein: E. Ong et al.
these conditions, the fusion protein would presumably
bind to the high-affinity sites on the cellulose fibers.
The pH-stability profile of Abg-CBDce x in solution
(Figure 4a) was similar to that reported for Abg. 3 Ad-
sorbed Abg-CBDcex remained active and bound to cel-
lulose during prolonged incubation at pH 6 to 7 (Figure
4a). The enzyme remained bound at low pH values,
but, like Abg itself, lost activity at pH 5 or below.
Desorption of Abg-CBDce x from cellulose was evident
at alkaline pH values, as also reported for Trichoderma
reesei cellulase. 31 Abg-CBDce x remained active and
bound to cellulose over a wide range of ionic strengths.
Like Cex, it could be readily desorbed from cellulose
with water. This suggests that hydrogen bonding and/
or hydrophobic interaction rather than ionic interaction
are responsible for the binding of the CBD to cellu-
lose. 32 The stability of Abg-CBDc~ x during long-term
storage at 37C was greatly improved when adsorbed
to Avicel, indicating that binding to cellulose protects
the fusion protein from denaturation. Although Abg-
CBDce x remains bound at 50C, the Abg part of the
fusion protein remains catalytically active only at tem-
peratures of 20C to 30C. This is not surprising, since
Abg comes from a mesophilic soil bacterium. It is possi-
ble that a thermophilic enzyme-CBD fusion protein
could be adsorbed to cellulose for substrate conversion
at elevated temperatures.
In conclusion, we believe that the cellulose-binding
domain has ideal properties for the immobilization of
enzymes. The simple stable coupling of any en-
zyme-CBD fusion to a readily available inexpensive
cellulosic support opens the possibility of using a ge-
neric method for immobilizing a whole range of en-
zymes. The absence of or minimal change in the confor-
mation of an enzyme fused to the CBD results in the
maintenance of catalytic efficiency of the enzyme. The
enhanced stability of the fusion protein under a wide
variety of operating conditions increases the volumet-
ric productivity of the catalytic reaction. The con-
venience of operating a reaction continuously is an
advantage over a batch reaction, wherein enzymes are
replenished periodically and products removed contin-
uously to prevent inhibition of the reaction. The Abg-
CBDc~x model shows that these benefits can be
achieved by optimizing the operating parameters and
by choosing the appropriate cellulosic matrix.
Acknowledgements
This work was supported by the Natural Sciences and
Engineering Research Council of Canada. We thank
C. Gross for providing E. coli strains CAG440 and
CAG456, S. Withers for helpful discussions, and J.
Wong for technical assistance.
Enzyme Microb. Technol., 1991, vol. 13, January
References
1 Hartmeier, W. lmmobilizedBiocatalysts Springer-Verlag, Ber-
lin, 1988
2 Stanley, W. L., Watters, G. G., Kelly, S. H. and Olson, A. C.
Biotechnol. Bioeng. 1978, 20, 135-140
3 Kucera, J. Biotechnol. Bioeng. 1983, 25, 599-602
4 Fujikawa, S., Yokota, T. and Koga, K. Appl. Microbiol. Bio-
technol. 1988, 28, 440-441
5 Jin, F. and Toda, K. Biotechnol. Lett. 1988, 10, 221-223
6 Gianfreda, L., Modafferi, M. and Greco, G., Jr. Enzyme Mi-
crob. Technol. 1985, 7, 78-82
7 Kimura, T., Ogata, M., Yoshida, M. and Nakakuki, T. Biotech-
nol. Bioeng. 1989, 33, 845-855
8 Baker, J. O., Oh, K. K., Grohmann, K. and Himmel, M. E.
Biotechnol. Lett. 1988, 10, 325-330
9 Webster, 1. A., Schwier, C. E. and Bates, F. S. Enzyme Mi-
crob. Technol. 1985, 7, 266-274
10 Tomar, M. and Prabhu, K. A. Enzyme Microb. Technol. 1985,
7, 557-560
I1 Dumitriu, S., Popa, M., Artenie, V. and Dan, F. Biotechnol.
Bioeng. 1989, 34, 283-290
12 Gilkes, N. R., Warren, R. A. J., Miller, R. C., Jr. and Kilburn,
D. G. J. Biol. Chem. 1988, 263, 10401-10407
13 Owolabi, J. B., Beguin, P., Kilburn, D. G., Miller, R. C.,
Jr. and Warren, R. A. J. Appl. Environ. Microbiol. 1988, 54,
518-523
14 Greenwood, J. M., Gilkes, N. R., Kilburn, D. G., Miller,
R. C., Jr. and Warren, R. A. J. FEBS Lett. 1989, 244, 127-131
15 Ong, E., Gilkes, N. R., Warren, R. A. J., Miller, R. C., Jr. and
Kilburn, D. G. Bio/Technology 1989, 7, 604-607
16 Baker, T. A., Grossman, A. D. and Gross, C. A. Proc. Natl.
Acad. Sci. USA 1984, 81, 6779-6783
17 Maniatis, T., Fritsch, E. F. and Sambrook, J. Molecular Clon-
ing: A Laboratory Manual Cold Spring Harbor Laboratory,
N.Y., 1982
18 Wood, T. M. Methods Enzymol. 1988, 160, 19-25
19 Lee, S. B., Shin, H. S., Ryu, D. D. Y. and Mandels, M.
Biotech. Bioeng. 1982, 24, 2137-2153
20 Ooshima, H., Sakata, M. and Harano, Y. Biotech. Bioeng.
1983, 25, 3103-3114
21 Gilkes, N. F., Langsford, M. L., Kilburn, D. G., Miller, R. C.,
Jr. and Warren, R. A. J. J. Biol. Chem. 1984, 259, 10455-10459
22 Bradford, M. Anal. Biochem. 1976, 72, 248-254
23 Spears, G., Sneyd, J. G. T. and Loten, E. G. Biochem. J. 1971,
125, 1149-1151
24 Peitersen, N., Medeiros, J. and Mandels, M. Biotechnol.
Bioeng. 1977, 19, 1091-1094
25 Gomori, G. Methods Enzymol. 1955, 1, 138-146
26 Laemmli, K. Nature 1970, 227, 680-685
27 Elving, P. J., Markowitz, J. M. and Rosenthal, I. Anal. Chem.
1956, 28, 1179-1180
28 Grossman, A. D., Erickson, J. W. and Gross, C. A. Cell 1984,
38, 383-390
29 Goldberg, A. L. and Goff, S. A. Maximizing Gene Expression
(Reznikoff, W. and Gold, L., eds.) Butterworths, Boston, 1986,
pp. 287-313
30 Han, Y. W. and Srinivasan, V. R. J. Bacteriol. 1969, 100,
1355-1363
31 Otter, D. E., Munro, P. A., Scott, G. K. and Geddes, R.
Biotechnol. Bioeng. 1989, 34, 291-298
32 Ong, E., Greenwood, J. M., Gilkes, N. R., Kilburn, D. G.,
Miller, R. C., Jr. and Warren, R. A. J. Trends Bioteehnol. 1989,
7, 239-243
65