Process Biochemistry 43 (2008) 855860

Contents lists available at ScienceDirect

Process Biochemistry
journal homepage: www.elsevier.com/locate/procbio

Kinetic of the gibberellic acid and bikaverin production in an airlift bioreactor

Ma. del Carmen Chavez-Parga, Omar Gonzalez-Ortega, Ma. de la Luz X. Negrete-Rodrguez, Irma Galindo Vallarino, Guillermo Gonzalez Alatorre, Eleazar M. Escamilla-Silva *
gico de Celaya, Av. Tecnolo gico y Garca Cubas s/n, CP 38010, Laboratorio de Biotecnologa y Bioingeniera, Departamento de Ingeniera Qumica, Instituto Tecnolo Celaya, Guanajuato, Mexico

A R T I C L E I N F O

A B S T R A C T

Article history: Received 9 March 2007 Received in revised form 14 February 2008 Accepted 7 April 2008 Keywords: Kinetic modelling Airlift bioreactor Gibberellic acid Bikaverin Gibberella fujikuroi

A mathematical description of the principal kinetics involved in gibberellic acid and bikaverin production in an airlift bioreactor was attained using a non-structured model. Growth, dextrose and nitrogen consumption, and gibberellic acid and bikaverin production kinetics were considered for this purpose. Experimental data were obtained from submerged batch fermentation of the fungus Gibberella fujikuroi (CDBB H-984) and were tted to three different sigmoidal models: 2- and 3-parameter Gompertz models and a logistic model. The most appropriate model was determined by means of an F test using growth kinetic tting. The 2-parameter Gompertz model was found to be the most suitable and was used for both dextrose and nitrogen consumption, and gibberellic acid and bikaverin production tting. A sensitivity analysis was performed to determine the effect of the parameters in the global model. The parameters of the 2-parameter Gompertz model were transformed in order to give them a biological meaning. 2008 Elsevier Ltd. All rights reserved.

1. Introduction Gibberellic acid and bikaverin are well known secondary metabolites of the fungus Gibberella fujikuroi. Gibberellic acid is a hormone present in higher plants that regulates different growth processes, with agricultural applications [18]. Bikaverin is a red pigment with anti-protozoal [2] and anti-tumour [10] activity. Acetyl-CoA is the common precursor of these metabolites. However, their biosynthesis is carried out in different sub-cellular compartments and used in independent substrate pools [7]. Bikaverin is synthesized via the polyketide route whereas gibberellic acid is synthesized through the isoprenoid pathway. Previous works have modelled bikaverin production in a uidised bed bioreactor [5], using sigmoidal functions to explain growth kinetics, and in a stirred tank bioreactor [16] using Monodtype functions to explain growth kinetics. Gibberellic acid production modelling in solid-state fermentation is also described [11]. Sigmoidal models that are not coupled with substrate consumption are easy to employ and predict the growth kinetic of G. fujikuroi under different culture conditions [5,6,8,15] and [13]. These types of models allow the growth kinetic to be

* Corresponding author. Tel.: +52 461 175 75x130; fax: +52 461 177 44. E-mail addresses: eleazar@iqcelaya.itc.mx, eleazar@itc.mx (E.M. Escamilla-Silva). 1359-5113/$ see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.procbio.2008.04.007

integrated separately from the substrate and production kinetic. This enables the production kinetic to be integrated from the point at which the nitrogen is exhausted. The point at which the gibberellic acid or bikaverin starts to be biosynthesised can then be established. Monod-type models are coupled with substrate consumption. More specically, they are coupled with limiting substrate consumption. Since nitrogen is the limiting substrate, whether the fermentation is for gibberellic acid or bikaverin production, the model will predict no more growth once the nitrogen is totally consumed by the fungus. Due to the storage of fat and carbohydrate in the fungus after nitrogen is exhausted, the biomass will continue increasing. Moreover, if the model predicts a satisfactory nitrogen exhaustion time, this will affect the global model. The model will be simplied because of the reduction of the number of adjustable parameters, and the tting capacity affected. Both sigmoidal or Monod-type models facilitate fermentation analysis and enable information to be obtained in a practical way. Furthermore, once these models are robust, they can be used to describe the production process under different fermentation conditions such as temperature, pH or aeration, among others. Eventually, the models enable us to understand, design and control the fermentation process of these metabolites. The mathematical description formulated in the present work addresses the process of gibberellic acid and bikaverin production in an airlift bioreactor, considering growth, substrate consumption and metabolite production kinetics.

856 2. Materials and methods 2.1. Micro-organism

Ma. del Carmen Chavez-Parga et al. / Process Biochemistry 43 (2008) 855860 (g l1), Yx/n is the yield factor for nitrogen and mn is the maintenance coefcient for nitrogen (h1). dp dx a bx dt dt (6)

G. fujikuroi (Sawada) strain CDBB H-984 maintained on potato glucose slants at 4 8C and sub-cultured every 2 months was used in the present work (Culture collection of the Department of Biotechnology and Bioengineering, CINVESTAV-IPN, Mexico). 2.2. Inoculum preparation Fully developed mycelial material from a slant was removed by adding an isotonic solution (0.9% NaCl). The mycelium is used to inoculate 300 ml of fresh culture medium contained in an Erlenmeyer ask. The ask is placed in a gyratory shaker (200 rpm) for 38 h at 29 1 8C. Developed mycelia will be used to inoculate the culture medium contained in the airlift bioreactor. The culture medium employed for the inoculum preparation is reported by Barrow et al. [3]. 2.3. Batch culture in the airlift bioreactor An Applikon airlift bioreactor Netherlands (working volume; 3.5 l) was employed in the present work. It consists of two concentric tubes with a settler where the air enters the bioreactor through the inner tube. The bioreactor is surrounded by a jacket lled with water allowing temperature control. It is also equipped with pH and dissolved oxygen sensors to control these variables. Moreover, it allows material from the bioreactor to be fed or removed by means of peristaltic pumps. Typical culture medium contained glucose (50 g l1), NH4Cl (0.75 g l1), KH2PO4 (5 g l1), MgSO47H2O (1 g l1), trace elements (2 ml l1). Typical fermentation conditions were pH set to 3.0, temperature set to 29 8C and aeration rate set to 1.6 vvm. At pre-determined times a sub-sample was withdrawn from the bioreactor and was fully analysed. 2.4. Analytical methods The dry weight method was employed to determine biomass concentration. Reducing sugars were determined by the DNS method [14]. Nitrogen concentration was determined by a modied Kjeldahl method [1]. Gibberellic acid concentration was determined with an HPLC method developed by Castillo and Martinez [4] and bikaverin concentration was determined by measuring absorbance at 520 nm. 2.5. Mathematical models Three different sigmoidal models were considered for the kinetic modelling of gibberellic acid and bikaverin production: 2-parameter (Eq. (1)) and 3-parameter (Eq. (2)) Gompertz models and a logistic model (Eq. (3)). These models were not coupled with substrate consumption. dx kx emt ax dt dx kx emt dt dx kx ax2 dt (1)

where p is the product concentration (mg l1) and a and b are LeudekingPiret constants. For model tting, Eqs. (1)(3) were integrated via the Runge-Kutta technique and parameter values were optimised by means of GREG subroutine [17], minimizing the residual sum of squares (RSS) (as named in Microsoft Excel1) using a DLL created by Digital Visual Fortran1. Following model discrimination, the selected model was used in Eqs. (4)(6) and the integration and optimisation process was repeated. 13 different data from fermentations of gibberellic acid production and 14 different data from fermentations of bikaverin production were analysed and parameter estimation was carried out by means of the minimization of residual sum of squares. 2.6. Model discrimination Model discrimination was carried out by means of an F test under the assumption that the 3-parameter model exactly predicts the concentration of biomass at every moment, according to the method described by Zwietering et al. [19]. The following was calculated: f RSS2 RSS1 =DF2 DF1 RSS1 =DF1 (7)

where RSS2 is the residual sum of squares value from the 2-parameter model, RSS1 is the RSS value from 3-parameter Gompertz model, DF1 is the degrees of freedom number from the 3-parameter Gompertz model and DF2 is the degree of freedom number from the 2-parameter model. This analysis is an approximation due to comparison of non-linear models and was also used by Chavez-Parga et al. [5,6] for model discrimination.

3. Results and discussion All data analysed in this work were reported by Chavez-Parga et al. [5,6]. For gibberellic acid production 13 different kinetic data were considered and for bikaverin production, fourteen. All experiments were performed under different conditions of nutrients and air-ow rate. Typical growth kinetics for gibberellic acid and bikaverin production are shown in Figs. 1 and 2, respectively. As can be seen, there is no lag phase. This is due to the similarity between the inoculum and fermentation medium. Therefore, the curve starts with a rapid growth phase where the dry weight increases exponentially. Once the limiting nutrient starts to decline, the dry weight stops increasing exponentially but continues to increase due to fat and carbohydrate accumulation in the fungus. Eventually, a stationary phase is reached. In some

(2)

(3)

where x is the biomass concentration (g l1), k is a kinetic parameter related to initial specic growth rate (h1), a is a kinetic parameter related to growth inhibition, m is the specic growth rate (h1) and t is the time (h) Eq. (1) contains three adjustable parameters while Eqs. (2) and (3) contain two adjustable parameters. Eqs. (1) and (2) consider a self-limited growth where the rate decreases exponentially with time. Eq. (1) has an extra term that takes into account inhibition, and has been used by Escamilla-Silva et al.[8] and Negrete [15] in gibberellic acid production modelling by G. fujikuroi. Eq. (3) describes exponential growth with an inhibition factor proportional to x2. The Logistic model has been used by Machado et al. [13] in describing the growth kinetic of G. fujikuroi in solid-state fermentation. Eq. (4) describes dextrose consumption and Eq. (5) describes nitrogen consumption. These last two equations consider parallel consumption of the substrate for growth and for maintenance requirements. Eq. (6) is the classic LeudekingPiret model which combines growth- and non-growth associated contributions to product formation. Eq. (6) will be used to describe both gibberellic acid and bikaverin production kinetics. ds 1 dx ms x dt Y x=s dt dn 1 dx mn x dt Y x=n dt
1

(4)

(5)

where s is the dextrose concentration (g l ), Yx/s is the yield factor for dextrose, ms is the maintenance coefcient for dextrose (h1), n is the nitrogen concentration

Fig. 1. Growth kinetic obtained during gibberellic acid production. Experimental and simulated data from the eleventh experiment (glucose, 50 g l1; NH4Cl, 0.75 g l1; KH2PO4, 5 g l1; MgSO47H2O, 1 g l1; trace elements, 2 ml l1; temperature, 29 8C; pH 3.0; aeration rate, 1.6 vvm; inoculum ratio, 10%).

Ma. del Carmen Chavez-Parga et al. / Process Biochemistry 43 (2008) 855860 Table 2 F test results for model comparison during bikaverin production Experiment f Gompertz 2 Logistic 1 2 3 4 5 6 7 8 9 10 11 12 13 14 3.8899 1.4129 2.2258 3.8884 0.0005 0.9910 0.0054 0.1203 0.0853 0.0048 4.4007 3.0924 11.9926 0.0315 5.3342 3.4193 3.0575 5.7927 0.5296 1.1592 2.2743 2.5504 3.4845 5.8873 5.4912 0.3012 9.8611 2.1331 4.75 4.6 4.96 4.6 4.54 4.75 4.84 4.84 4.96 4.96 5.59 5.12 5.99 5.32 F table RSS

857

Gompertz 3 Gompertz 2 Logistic 1.6138 4.4624 8.8877 1.7819 3.3382 4.4638 2.4572 2.3164 3.7221 1.3843 0.5506 2.0595 3.1838 5.1446 2.1369 4.9127 10.8660 2.2769 3.3383 4.8325 2.4584 2.3418 3.7539 1.3849 0.8968 2.7672 9.5475 5.1649 2.3311 5.5522 11.6052 2.5192 3.4561 4.0326 1.9492 1.7793 5.0191 2.1992 0.9825 2.1285 8.4164 6.5164

Fig. 2. Growth kinetic obtained during bikaverin production. Experimental and simulated data from the fourth experiment (glucose, 50 g l1; NH4Cl, 0.75 g l1; KH2PO4, 5 g l1; MgSO47H2O, 1 g l1; trace elements, 2 ml l1; temperature, 29 8C; pH 3.0; aeration rate, 0.5 vvm; inoculum ratio, 10%).

experiments, a slight decline in dry weight can be observed due to consumption of reserves by the fungus given the environmental conditions at the end of the fermentation process. The 3-parameter Gompertz model can describe this decline-phase whereas the other 2 models are not able to do this. The 3 studied models gave good ttings of growth kinetics. The 3-parameter Gompertz model gave some problems during tting due to its high non-linearity while the other 2 models did not present these problems. It is important to mention that the residuals obtained from the 3 studied models did not show any kind of trend. F test results are presented in Table 1 for kinetics belonging to gibberellic acid production and Table 2 show the same for bikaverin production. In the case of growth kinetics during gibberellic acid production, the logistic model was accepted in 69% of the experiments and the 2-parameter Gompertz model was accepted in 77% of the experiments. In only two experiments (15%) in Table 1 (Experiments. 4 and 6) the 3-parameter Gompertz model was accepted by the F test. A decline in biomass concentration in these two experiments is present at the end of the fermentation period, which can be modelled by the 3-parameter Gompertz model.

For the case of growth kinetics during bikaverin production, the results are similar to those obtained during gibberellic acid production. The Logistic model was accepted in 71% of the experiments and the 2-parameter Gompertz model was accepted in 93% of the experiments. In only one experiment (7%) of Table 2 (Experiment 13) the 3-parameter Gompertz model was accepted by the F test. It is worthwhile mentioning that in three experiments in Table 2 (Experiments 68), the logistic model tted experimental data better than the 3-parameter Gompertz model. This can be explained in terms of a low initial specic growth rate present in these experiments and can be modelled by the logistic model better than the Gompertz-type models. Given the above results, the 2-parameter Gompertz model was chosen to perform kinetic tting for substrate consumption and product formation. The same result was obtained by Chavez-Parga et al. [5,6]. Figs. 3 and 4 present experimental and tted data obtained during gibberellic acid production and bikaverin production, respectively. From Fig. 3 it is evident that gibberellic acid production begins after nitrogen is exhausted in the culture medium. In the case of bikaverin production, Fig. 4, production begins a few hours before total exhaustion of nitrogen. This conrms that the production of these secondary metabolites is dependent on the nitrogen concentration in the culture medium, as stated by Giordano et al. [12]. Nonetheless, their regulatory mechanism differs,

Table 1 F test results for model comparison during Gibberellic acid production Experiment f Gompertz 2 Logistic 1 2 3 4 5 6 7 8 9 10 11 12 13 6.0628 0.4676 0.9861 18.0081 1.0881 12.2952 1.0733 0.1642 0.0058 0.6583 0.2364 2.7034 0.0679 4.8127 0.1506 3.0541 29.7135 3.0482 26.2446 9.4806 1.7126 2.8752 3.4012 2.8293 26.7302 1.6475 5.12 5.99 5.99 5.32 5.32 4.84 4.67 5.12 5.59 5.59 5.32 7.71 5.12 F table RSS Gompertz 3 Gompertz 2 Logistic 1.9982 5.3887 10.6941 5.8580 4.9742 1.8974 11.5592 6.2321 7.9964 9.3722 3.8783 0.1510 9.2478 3.3443 5.8087 12.4516 19.0444 5.6508 4.0181 12.5135 6.3458 8.0030 10.2536 3.9930 0.2531 9.3176 3.0668 5.5240 16.1375 27.6157 6.8695 6.4242 19.9890 7.4180 11.2808 13.9260 5.2499 1.1604 10.9407

Fig. 3. Substrate consumption and gibberellic acid production kinetics. Experimental (symbols) and simulated (lines) data from the eighth experiment (glucose, 80 g l1; NH4Cl, 0.75 g l1; KH2PO4, 5 g l1; MgSO47H2O, 1 g l1; trace elements, 2 ml l1; temperature, 29 8C; pH 3.0; aeration rate, 1.6 vvm; inoculum ratio, 10%).

858

Ma. del Carmen Chavez-Parga et al. / Process Biochemistry 43 (2008) 855860

Fig. 4. Substrate consumption and bikaverin production kinetics. Experimental (symbols) and simulated (lines) data from the eleventh experiment (glucose, 50 g l1; NH4Cl, 0.75 g l1; KH2PO4, 5 g l1; MgSO47H2O, 1 g l1; trace elements, 2 ml l1; temperature, 29 8C; pH, 3.0; aeration rate, 1.6 vvm; inoculum ratio, 10%).

because bikaverin production stops after a few days of fermentation while gibberellic acid continues to be biosynthesised. It also conrms, as stated by Escamilla-Silva et al. [9], that bikaverin production precedes that of gibberellic acid. As can be seen in Figs. 3 and 4, the global model tted experimental data in an acceptable way for both gibberellic acid and bikaverin production. Tables 3 and 4 present optimised parameters for gibberellic acid and bikaverin production.

As can be seen in Tables 3 and 4, the value of mn in most of the experiments is zero. This implies that this parameter must not be present in the global model. This was to be expected since nitrogen is present in low quantities and is used for growth requirements. Since the global model is able to satisfactorily predict the dextrose consumption kinetic, it enables the establishment of a fed-batch scheme to maintain dextrose concentration at a value favourable to gibberellic acid production. a Values obtained for gibberellic acid and bikaverin production should be zero due to the secondary metabolite nature of these substances. However, the growth-associated term in the LuedeckingPiret equation must be present because of the accumulation of fat and carbohydrate by the fungus. This makes the dry weight increase further, even though there is no more nitrogen source in the culture medium. As can be seen in Table 4, zero values for b are present. According to results from experiments not shown in this work, b values tend to be negative because of evident bikaverin degradation. This means that bikaverin degradation could be considered rst order with respect to biomass concentration. In the present experiments the fermentation period ended before bikaverin degradation started to occur. A sensitivity analysis was performed to assess the individual effect of the parameters in the mathematical model. To do this, optimised parameter values were varied 5 and 10% and their effect on the model was evaluated by the change in RSS relative to the RSS obtained with optimised parameters. Fig. 5 shows the results of the sensitivity analysis where it is clear that the model is most

Table 3 Optimised parameters values obtained during gibberellic acid production Experiment 1 2 3 4 5 6 7 8 9 10 11 12 13

m (h1)
0.0340 0.0406 0.0354 0.0239 0.0306 0.0496 0.0276 0.0534 0.0390 0.0451 0.0241 0.0179 0.0377

k (h1) 0.0694 0.0441 0.0998 0.0563 0.0732 0.1094 0.0690 0.0972 0.0782 0.1086 0.0525 0.0333 0.0600

Yx/s (g x (g d)1) 1.8652 0.2158 ND 0.2468 1.0988 0.4173 ND 0.4588 0.3023 0.4642 ND 0.2822 0.4750

ms (g d (g x h)1) 0.0446 0.0183 ND 0.0209 0.0703 0.0050 ND 0.0210 0.0341 0.0311 ND 0.0056 0.0535

Yx/n (g x (g n)1) 42.9868 33.2860 13.6791 24.4159 28.7200 33.2672 12.2258 49.3878 16.9196 28.3609 19.4043 ND ND

mn (g n (g x h)1) 0 0 0 0 0 0 0 0.0001 0 0 0 ND ND

a (mg g (g x)1))
6.0616 8.8358 7.9276 ND 2.5420 0.4092 ND 4.0849 ND 3.8957 9.6709 5.8646 4.6053

b (mg g (g x h)1)
0.0351 0.0282 0 ND 0.0139 0.0352 ND 0.0317 ND 0.0576 0.0442 0.0246 0.2329

ND, no experimental data.

Table 4 Optimised parameter values obtained during bikaverin production Experiment 1 2 3 4 5 6 7 8 9 10 11 12 13 14

m (h1)
0.0507 0.0655 0.0328 0.0612 0.0499 0.0475 0.0449 0.0369 0.0348 0.0358 0.0501 0.0474 0.0613 0.0575

k (h1) 0.0589 0.1062 0.0611 0.1053 0.0914 0.0875 0.0802 0.0581 0.0623 0.0673 0.0745 0.0952 0.1061 0.1029

Yx/s (g x (g d)1) 0.2531 0.4624 0.3241 0.4714 0.2257 0.3804 0.4851 0.4190 0.4883 0.4301 0.4795 0.7278 0.6328 1.4516

ms (g d (g x h)1) 0.0191 0.0551 0.0006 0.0243 0.0193 0.0193 0.0259 0.0240 0.0167 0.0163 0.0183 0.0262 0.0093 0.0263

Yx/n (g x (g n)1) 18.2285 16.8396 10.9847 22.0172 18.8756 15.0377 14.8137 14.3470 15.5046 20.2204 14.8765 ND 16.3166 13.9746

mn (g n (g x h)1) 0 0 0 0 0 0 0 0 0 0 0 ND 0 0

a (mg b (g x)1)
18.7315 9.7985 5.7520 15.9750 26.2261 8.7086 11.6856 12.7818 4.5009 2.7749 4.0910 2.2235 1.0684 0.8987

b (mg b (g x h)1)
0.0058 0.0949 0 0.0129 0.0114 0 0.0074 0.0012 0.0047 0.0126 0 0 0.0046 0

ND, no experimental data.

Ma. del Carmen Chavez-Parga et al. / Process Biochemistry 43 (2008) 855860

859

Fig. 5. Sensitivity analysis for global model employing 2-parameter Gompertz model as a basis.

sensitive to change in the values of k and m. (DRSS is the RSS variation due to the change in parameter value, RSS0 is the RSS with the optimised parameter value, DP is the increase in parameter value and P0 is the optimised parameter). Since k and m belong to exponential functions in the growth kinetic model this was to be expected. This simplied analysis tells us that care must be taken when determining k and m from experimental data. Another problem arises since we are only able to determine k and m from experimental data. The next step is to give these parameters a biological meaning, which helps to establish which experimental data is very important. It can be easily demonstrated that for the 2-parameter Gompertz model: k m0 (8)

Following the F test, the 2-parameter Gompertz model was selected to describe the principal kinetics involved during gibberellic acid or bikaverin production in an airlift bioreactor employing G. fujikuroi. Experimental data were tted satisfactorily by the global model, with the 2-parameter Gompertz equation as a basis, for experimental data obtained during gibberellic acid or bikaverin production. The Logistic model, according to the F test, could also be used to describe principal kinetics. However, this equation should be used when the initial specic growth rate is slow due to an acclimatisation of the fungus to the new environment. It should also be used when the initial rates of substrate consumption are slow (the latter due to the shape of curves described by the global model employing a logistic equation as a basis). A sensitivity analysis was performed on the global model and was found to be more sensitive to changes in values of k and m. A biological meaning was given to those parameters suggesting that they are dependent on an initial specic growth rate and biomass concentration. The latter helps to establish which part of the fermentation period is critical to obtaining as much experimental data as possible and allows the establishment of initial parameter values during the optimisation process (tting) of experimental data. Acknowledgements The research was funded by Consejo Nacional de Ciencia y Tecnologia (Project: 33973-B) and Consejo del Sistema Nacional de Educacion Tecnologica (Project: 648-P), SAGARPA-CONACYT (Project: 2004-C01-77/A-1) References
[1] A.O.A.C.. Ofcial Methods of Analysis. Washington, DC: 15th Association of Ofcial Chemists; 1990, ISBN: 0-935584-42-0. [2] Balan J, Fuska J, Kuhr I, Kuhrova V. Bikaverin, an antibiotic from Gibberella fujikuroi effective against Leishmania brasiliensis. Folia Microbiologica 1970; 15:47984. [3] Barrow A, Jefferys EG, Nixon IS. Process for the production of gibberellic acid. Patent GB 838,032; 1960. [4] Castillo G, Martinez S. Reversed-phase C18 high-performance liquid chromatography of gibberellins GA3 and GA1. Journal of Chromatography A 1997;782: 1379. [5] Chavez-Parga, Ma. del C. Produccion de acido giberelico en un biorreactor airlift. Ph.D. Thesis, Instituto Tecnologico de Celaya, Guanajuato, Mexico; 2005. [6] Chavez-Parga Ma del C, Gonzalez-Ortega O, Sanchez-Cornejo G, NegreteRodriguez Ma de la LX, Gonzalez-Alatorre G, Escamilla-Silva EM. Mathematical description of bikaverin production in a uidized bed bioreactor. World Journal Microbiology and Biotechnology 2005;21:6838. [7] Domenech CE, Giordano W, Avalos J, Cerda-Olmedo E. Separate compartments for the production of sterols, carotenoids and gibberellins in Gibberella fujikuroi. FEBS European Journal of Biochemistry 1996;239:7205. [8] Escamilla-Silva E, Torre M de la, Ciurlizza A. Aplicacion del modelo de Gom pertz para el estudio cinetico de crecimiento del hongo Gibberella fujikuroi en cultivo sumergido. In: VII Congreso Nacional de Biotecnologia y Bioingenieria; 1997. [9] Escamilla-Silva EM, Poggi-Varaldo H, Torre-Martinez M de la, Sanchez-Cornejo MG, Dendooven L. Selective production of bikaverin in a uidized bioreactor with immobilized Gibberella fujikuroi. World Journal of Microbiology and Biotechnology 2001;17:46974. [10] Fuska J, Proksa B, Fuskova A. New potential cytotoxic and antitumor substances. Neoplasma 1975;22:3358. [11] Gelmi C, Perez-Correa R, Agosin E. Modelling Gibberella fujikuroi growth and GA3 production in solid-state fermentation. Process Biochemistry 2002;37: 103340. [12] Giordano W, Avalos J, Cerda-Olmedo E, Domenech EC. Nitrogen availability and production of bikaverin and gibberellins in Gibberella fujikuroi. FEMS Microbiology Letters 1999;173:38993. [13] Machado MMC, Oishi OB, Pandey A, Soccol RC. Kinetics of Gibberella fujikuroi and Gibberellic acid production by solid-state fermentation in a packed-bed column bioreactor. Biotechnology Progress 2004;20:144953. [14] Miller GL. Use of dinitrosalicylic acid reagent for determination of reducing sugar. Analytical Chemistry 1959;31:4268.

m0
lnxmax =x0

(9)

where m0 is the initial specic growth rate, x0 is the initial biomass concentration and xmax is the maximum biomass concentration. This implies that we must have as many points as possible from the early stages of fermentation since k and m are related to initial specic growth rate. Moreover, initial biomass concentration must be measured as accurately as possible since it affects the values of m and k. Maximum biomass concentration is determined with relative ease due to the amount of experimental data measured in the nal stage of the fermentation. 4. Conclusions It is evident that both sigmoidal and Monod-type models have advantages and disadvantages, and given our experimental data either can be chosen. However, the use of a non-substrate coupled model is recommended. This is due to the secondary metabolite nature of the substances studied and to the accumulation of fat and carbohydrate by the fungus after nitrogen exhaustion. Ultimately, given the amount of assumptions that need to be considered, these types of models continue to be limited in describing what happens under different culture conditions. The next step is to explore the effect of variables such as pH or aeration on the model parameters. The global model can then be used to predict fermentation behaviour and can eventually be used to perform an optimisation or a control scheme.

860

Ma. del Carmen Chavez-Parga et al. / Process Biochemistry 43 (2008) 855860 [17] Stewart W, Caracotsios M, Sorensen J. GREG Subroutine. Madison: Department of Chemical Engineering, University of Wisconsin; 1990. [18] Tudzynski B. Biosynthesis of gibberellins in Gibberella fujikuroi: biomolecular aspects. Applied Microbiology and Biotechnology 1999;52:298310. [19] Zwietering MH, Jongenburger I, Rombouts FM, Vant Riet K. Modeling of the bacterial growth curve. Applied and Environmental Microbiology 1990;56: 187581.

[15] Negrete, RX. Produccion de acido giberelico (GA3) en biorreactor de tanque agitado con Gibberella fujikuroi empleando aceite de maiz como fuente de carbono. M.Sc. Thesis. Insituto Tecnologico de Celaya, Guanajuato, Mexico; 2002. [16] Shukla R, Chand S, Srivastava AK. Batch kinetics and modeling of gibberellic acid production by Gibberella fujikuroi. Enzyme and Microbial Technology 2005;36:4927.