World Journal of Microbiology & Biotechnology (2005) 21:683688 DOI 10.

1007/s11274-004-3854-0

Springer 2005

Mathematical description of bikaverin production in a uidized bed bioreactor

vez-Parga, Omar Gonza lez-Ortega, Guadalupe Sa nchez-Cornejo, Ma. de la Luz X. NegreteMa. del Carmen Cha lez-Alatorre and Eleazar M. Escamilla-Silva* Rodr guez, Guillermo Gonza Laboratorio de Biotecnologa y Bioingeniera, Departamento de Ingeniera Qumica, Instituto Tecnologico de Celaya, Av. Tecnologico y A. Garca, Cubas s/n, 38010 Celaya, Gto. Mexico *Author for correspondence: Tel.: +52-461-61-175-75, Fax: 52-461-61-177-44, E-mail: eleazar@iqcelaya.itc.mx

Keywords: Bikaverin, uidized bioreactor, Gibberella fujikuroi, immobilized, mathematical model

Summary Growth of Gibberella fujikuroi in submerged cultures occurs as micelles or lamentous hyphae dispersed in uid and pellets or stable, spherical agglomerations. Gibberella fujikuroi growth, substrate consumption and bikaverin production kinetics obtained from submerged batch fermentation were tted to three dierent sigmoid models: two and three-parameter Gompertz models and one Logistic model. Growth tting was used to compare between models and select the best one by means of an F test. The best model for describing growth was the two-parameter Gompertz model and was used for glucose consumption and bikaverin production tting. Data from eight dierent schemes of fermentations were analysed and parameter estimation was carried out by means of minimization of residual sum of squares. Some characteristic values obtained with the two-parameter Gompertz model t are: l 0.028 h)1, Yx/s 0.1089 g substrate/g biomass, a 0.1384 g product/g biomass.

Introduction Bikaverin is a red pigment with specic anti-protozoal activity against Leishmania brasiliensis (Balan et al. 1970; Desjeux 1992) and anti-tumour activity (Fuska et al. 1975). Additionally, bikaverin and its derivatives have a cytotoxic eect on in vitro proliferating cells of Erlich ascites carcinoma, Sarcoma 37 and leukaemia L-5178 (Fuska et al. 1975) and is a fermentation product of Gibberella fujikuroi or Fusarium sp. (Cornforth & Ryback 1971; Kjaer et al. 1971). It has been chemically synthesized (Barton et al. 1976; Katagiri et al. 1981). Kjaer (1971), reported the chemical and spectroscopic characteristics of bikaverin, while the crystal and molecular structures were described by de Boer et al. (1971). The formation of bikaverin precedes that of gibberellin (Escamilla-Silva et al. 2000) and both secondary metabolites are produced from the primary metabolite acetyl-CoA. Bikaverin is synthesized via the polyketide route while gibberellin is synthesized through the isoprenoid pathway (Jones & Pharis 1987). The industrial production of these secondary metabolites is done with cultures of mycelia in liquid (submerged) or solid substrate fermentation. Models of mould growth and metabolic production based on characteristics of mycelial physiology are important to understand, design, and control those industrial fermentation processes (Fredickson et al. 1979; Nielsen

& Villadsen 1992; Bailey & Ollis 1998, Escamilla-Silva et al. 2001 a, b). In other words these models enable us to obtain information in a practical way, facilitating fermentation analysis, and can be used to solve problems that may appear during the fermentation process. Furthering our research work on metabolite production (Escamilla-Silva et al. 1999, 2000, 2001a, b), we now oer an unstructured model based on the Gompertz and Logistic function is now oered to describe the kinetic studies of biomass and bikaverin production using G. fujikuroi in submerged batch fermentation in a uidized bed bioreactor.

Mathematical models for bikaverin production A great variety of mathematical models for batch fermentations are currently available (Fredickson et al. 1979; Humphrey 1979; Gutke 1980; Kossen & Oosterhuis 1985; Aynsley & Ward 1990; Nielsen & Villadsen 1992; Larralde-Corona et al. 1994). In general, models used for describing growth may or may not be coupled with substrate consumption and have two or three adjustable parameters in the simplest cases. Equations (1)(3) present models studied that are independent of substrate consumption. dx kxel t ax dt 1

684 dx kxel t dt dx kx ax2 dt

Chavez-Parga et al. where RSS2 is the RSS value from two-parameter model, RSS1 is the RSS value from the three-parameter of the Gompertz model, DF1 is the number of degrees of freedom from the three-parameter Gompertz model and DF2 is the number of degrees of freedom from the twoparameter model. This f value will be compared with a value from an F table values. If f is lower than F no extra parameter is required. If f is greater than F, the threeparameter model must be selected. This analysis is an approximation, due to the comparison of non-linear models and was used by Zwietering et al. (1990) for model discrimination. According to the number of experiments accepted using the F test, a model for describing growth kinetics will be selected. This model will be used for substrate consumption and production formation kinetics tting.

where x is the biomass concentration (g l)1), a and k are kinetic constants, l, the specic growth rate (h)1) and t, the time (h). Equations (1) and (2) correspond to the Gompertz model and Equation (3) describes the Logistic model. Equation (1) contains three adjustable parameters (k, l and a) whereas Equations (2) and (3) contain two adjustable parameters (l and k). Equations (1) and (2) consider a self-limited growth where the rate decreases exponentially with time. Equation (1) has an extra term that takes account of inhibition of growth (a) and has been used by Escamilla-Silva (1997) and Negrete (2002) in gibberellic acid production by Gibberella fujikuroi. Equation (3) describes exponential growth with an inhibiting factor proportional to x2. Many studies carried out on growth kinetics have used the Logistic model and they have several weaknesses and the drawback of the logistic equation is its failure to predict a lag and decline phases after the stationary population has exhausted all the available resources (Weiss & Ollis 1980; Gouveia et al. 2000). Equations (4) and (5) describe substrate consumption and bikaverin production kinetics, respectively. Equation (4) considers parallel consumption of substrate for growth and for maintenance requirements. Equation (5) is the classic LeudekingPiret model that combines growth and non-growth associated contributions to product formation. ds 1 dx mx dt Yx=s dt 4

Materials and methods Micro-organism Gibberella fujikuroi (Sawada) strain CDBB H-984 maintained on potato glucose slants at 4 C and sub-cultured every 2 months was used in this experiment (Culture collection of the Department of Biotechnology and xico). Bioengineering, CINVESTAV-IPN, Me Preparation of inoculum The mycelial material from a fully developed slant was removed by adding 5 ml physiological saline (0.9% NaCl), and homogenizing the mixture in a Waring blender for 30 s at 50% of its maximum speed, 15 s at 60% and 10 s at 80%. The homogenate was inoculated in a 500 ml Erlenmeyer ask with 300 ml of culture medium and placed on a gyratory shaker (180 rev min)1, 2.5 cm/stroke) at 29 1 C for 38 h (Escamilla-Silva et al. 1999). The developed mycelia were then used in the immobilization process. Preparation of immobilized mycelium Inside a sterilized laminar ow chamber, mycelia were immobilized in autoclaved Ca-polygalacturonate (Escamilla-Silva et al. 2000). Briey, a suspension of G. fujikuroi mycelium was mixed with an equal volume of 8% (w/v) polygalacturonic acid solution. The mixture was homogenized in a Mixer (Osterizer junior of 230 ml at 100 rev min)1,) for 10 min. The solution was adjusted with polygalacturonic acid to a nal concentration of 4%. The mixture was forced through a multi-needle template (gauge 18 for 5 mm beads and 21 for 3 mm xico) owing ones) with a peristaltic pump (CRODE, Me at 10 ml min)1 and the droplets were collected in a sterile 3.5% CaCl2 solution. After soaking for 3 h, the liquid was decanted and the spherical beads with a wet weight of approximately 1.4 10)4 g mycelium were

where s is the substrate concentration (g l)1) and x, the biomass concentration (g l)1); Yx=s , the yield factor, m, the maintenance factor and t(h), the time through the fermentation. dp dx a bx dt dt 5

where a and b are LeudekingPiret parameters, p is the product concentration (g l)1). First, Equations (1)(3) are integrated via Runge Kutta technique and parameter values are optimized by means of GREG subroutine (Stewart et al. 1990) minimizing the residual sum of squares (RSS). Model comparison will be performed by an F test under the assumption that the three-parameter model exactly predicts the biomass at every moment. To do this the following will be calculated: f RSS2 RSS1 =DF2 DF1 RSS1 =DF1 6

Mathematical model of bikaverin production washed with sterile distilled H2O and stored at 4 C for 24 h. The diameter of the beads was measured in a microscope (Leica, LMDS) with a micrometer grid and pellets of 3 mm (0.15) were used in shake asks as preliminary experiment whereas beads of 3 and 5 mm were used in the uidized bed bioreactor to obtain information about mass transfer eects. Culture media in shake asks The effect of glucose, rice our, NH4Cl and KH2PO4 on bikaverin production was studied in 500 ml shakeasks using a factorial experimental design 24 in triplicate (Montgomery, 1991). The eects of glucose (80 and 100 g l)1), inorganic nitrogen (NH4Cl 1 or 3 g l)1), phosphate (KH2PO4 3 or 5 g l)1), and vegetal source (rice our 2 or 5 g l)1) were investigated while the MgSO47H2O concentration was 2.5 g l)1. Rice our was used as G. fujikuroi easily infects and grows abundantly in rice plants (Coolbaugh 1983). Erlenmeyer 500-ml asks were added with 300 ml of culture medium as given in the experimental design and 1980 biocatalyst beads with a wet weight of 1.4 10)4 g mycelium each, placed on an orbital bench top shaker 300 5 rev min)1, in an incubation hood and incubated at 29 1 C for 168 h. Batch culture in uidized bioreactor Results and discussion An orthogonal experimental design L8 (25) (Peace 1981; Ross 1989) was used to investigate eects of [A] glucose (80 or 100 g l)1), [B] inorganic nitrogen (NH4Cl, 1 or 3 g l)1), [C] inoculum ratio (10% or 15% v/v equal to an initial biomass concentration of 0.2 or 0.3 g biomass dry weight l)1, respectively), [D] air-ow rate (3 or 5 volumes of air per volume of medium min)1 (v/v/m)), and [E] bead size (3 or 5 mm) on bikaverin production in two 3.5 l (work volume 2 l) bubble column bioreactors especially designed for this purpose. They consisted of two concentrically tubes; the outer tube (15 cm diameter; 76 cm height) was of the jacket type to recycle water and control temperature, and the inner tube (10.16 cm of diameter, 76 cm height) was the fermentation system (CRODE, Mexico) equipped to control pH, air ow rate, culture, broth feed and temperature. The experimental design allowed to measure eects of each factor and interactions of the factors AB and AC. Additionally, media containing 60 g glucose dm)3 or 120 g glucose l)1, 1 g NH4Cl l)1, 10% v/v inoculum and beads of 3 mm size were used with air owing at 3 v/v/m. The fermentation was done for 168 h at 29 C and sterile H2O was added at 10 or 12 h intervals to compensate for evaporation loss while the pH was not adjusted. Each day, a sub-sample of 30 ml was taken from the culture medium and analysed for dry weight of biomass, reductive sugars, NH4+-N, pH and bikaverin. Analytical methods

685

Ten beads were taken from the sub-sample, disintegrated by adding 25 ml of a 0.5 M EDTAphosphate solution and stirred for 25 min. The released and external mycelium was ltered through a pre-weighed Millipore 0.45 lm membrane and free mycelial biomass was measured by drying to constant weight in an oven at 90 C. NH4+-N in the medium was determined by Kjeldahl (A.O.A.C. 1990) and reducing sugars, i.e. glucose, by the dinitrosalicylic acid method (Miller 1959). Magnesium and potassium were determined directly in ltered samples by atomic absorption spectrophotometry (Perkin-Elmer 2882, USA). Phosphates (P-PO4)3) were determined by the molybdenum-blue method on a UVVIS spectrophotometer at 690 nm (Perkin-Elmer model Lambda 2, USA) (A.O.A.C. 1990). A sub-sample of ca. 2.5 g moist mycelium was washed twice with 100 ml distilled H2O and 30 ml of methanol. The sample was centrifuged, acidied with 0.15 ml 6 M HCl and extracted four times with 50 ml chloroform. Extracts were pooled, heated and boiled for 30 s, ltered through 0.45 lm Millipore membrane with chloroform, adjusted to 200 ml and analysed for bikaverin on a UV VIS spectrophotometer at 544 nm (Perkin-Elmer model Lambda 2, USA).

All data analysed in this work were reported by Sanchez (1996) and Escamilla-Silva et al. (1999, 2000, 2001a, b) and correspond to eight dierent experiments. These experiments were performed under dierent conditions of nutrients and airow rate in a uidized bed bioreactor. The three models gave good ttings for Gibberella fujikuroi growth data (Figure 1). As can be seen in the growth curve there is no lag phase because the medium

12 10 Biomass, gL-1 8 6 4 2 0 0 24 48 72 96 120 144 168 Time, h

Figure 1. Experimental data from eighth experiment (glucose, 80 g l)1; NH4Cl, 3 g l)1; KH2PO4, 5 g l)1; MgSO4, 2.5 g l)1; temperature, 29 C; pH, free; aeration rate, 3 v/v/m; inoculum ratio, 10%; bead diameter, 3 mm) for the bikaverin production tted with studied models.
Exp data Gomp2 Logistic Gomp3

686 is limited by nitrogen. Since the strain requires little or no acclimatization, and the growth in the bioreactor starts quickly due to the use of mycelial cells with vigorous growth as inoculum (Escamilla-Silva et al. 1999). A rapid growth phase and a phase of nutrient consumption (balanced phase) can be observed. The dry weight of the cells increased exponentially and the glucose, nitrogen, phosphates, magnesium and potassium consumption remained almost constant per unit increase of dry weight. This is the point at which the bikaverin production began. Most of kinetic growth studies take this phase for the mathematical modelling growth e.g. Monod function one which does not include all the curve growth and this is one of the useful aspects of the model presented in this study. Borrow et al. (1961) indicate that there is a transition phase that corresponds to the time where the limiting nutrient is exhausted and it can be the glucose or nitrogen. As the culture media designed for this study are not limited by phosphate or magnesium, the limitation by nitrogen or glucose occurs after the initial oxygen restriction and this permits greater cell proliferation. Therefore the mycelial composition is dierent from the balanced growth phase and the dry weight increases proportionally, though the growth and nutrient consumption rates decreased. It is worth mentioning that in this continuous phase there was an increase in the carbohydrate and fat content of the cells, until the depletion of the glucose or the carbon source. Various authors reported that this phase marks the end of bikaverin production. However, in this study we can see that the production of bikaverin continued (Figure 2), when the carbon source began to be exhausted (Figure 3). The eects of oxygen and bikaverin production are now being studied in our laboratory. A mathematical description of Figure 1 is presented below: The three-parameter Gompertz model gave some problems with tting due to its high nonlinearity. Two-parameter models posed no problems
1.4 1.2 Bikaverine, g L-1 1 0.8 0.6 0.4 0.2 0 0 24 48 72 96 120 144 168
Time, h
Experimental Gompertz2

Chavez-Parga et al.
100 Dextrose, g L-1 80 60 40 20 0 0 24 48 72 96 120 144 168 Time, h
Experimental Gompertz2

Figure 3. Substrate consumption of the eighth (glucose, 80 g l)1; NH4Cl, 3 g l)1; KH2PO4, 5 g l)1; MgSO4, 2.5 g l)1; temperature, 29 C; pH, free; aeration rate, 3 v/v/m; inoculum ratio, 10%; bead diameter, 3 mm) experiment tted with two-parameter Gompertz model for bikaverin production.

with tting and can be used for estimating the kinetics parameters. Analysis showed that residuals were not left in any model. F test results are presented in Table 1. As an example consider results from the second experiment. The F value is lower than the logistic model f value and greater than the two-parameter Gompertz model f value. Therefore, the two-parameter Gompertz is better than the logistic model. There is no advantage in using the three-parameter Gompertz model even though its RSS value is lower than the other two-parameter models. The logistic model was accepted in 37.5% of the experiments and the two-parameter Gompertz model in 75% of the cases. In only one experiment (Exp. 5) did the three-parameter Gompertz model t the data better than two-parameter models. This was due an apparent death phase presented in this experiment (Figure 4) that the three-parameter Gompertz is capable of predicting. In the two experiments where both two-parameter models are better than the three-parameter model, the difference between the two-parameters models is minimal. On the basis of these results it was decided to choose the two-parameter Gompertz model to perform kinetic tting for substrate consumption and product formation. Optimized parameter values are presented in Table 2 for the two-parameter Gompertz model.

Table 1. F test results for model comparison in the Bikaverin production. Exp. f F RSS

Gompertz2 Logistic Table Gompertz2 Logistic Gompertz3 1 2 3 4 5 6 7 8 1.9364 3.9265 0.8643 0.2101 27.7701 5.6308 3.2518 2.5857 1.3026 7.1646 6.8316 10.1431 16.5492 4.67 4.54 4.54 4.54 4.54 4.54 2.8110 4.54 15.8933 4.54 4.8086 2.5442 1.3907 1.9931 8.3054 0.8349 1.4388 0.8244 4.6045 2.9795 1.9138 3.2946 6.1265 0.5732 1.4040 1.4482 4.1852 2.0164 1.3149 1.9655 2.9128 0.6070 1.1824 0.7032

Figure 2. Bikaverin production data from the eighth (glucose, 80 g l)1; NH4Cl, 3 g l)1; KH2PO4, 5 g l)1; MgSO4, 2.5 g l)1; temperature, 29 C; pH, free; aeration rate, 3 v/v/m; inoculum ratio, 10%; bead diameter, 3 mm) experiment tted with two-parameter Gompertz model.

Mathematical model of bikaverin production

10 9 8 7 6 5 4 3 2 1 0 0 24 48 72 96 Time, h

687 associated with growth, as it is with most secondary metabolites.

Biomass, gL-1

Conclusions
Exp data Gomp2 Logistic Gomp3

120

144

168

Figure 4. Experimental data from fth (glucose, 60 g l)1; NH4Cl, 3 g l)1; KH2PO4, 5 g l)1; MgSO4, 2.5 g l)1; temperature, 29 C; pH, free; aeration rate, 3 v/v/m; inoculum ratio, 10%; bead diameter, 3 mm) experiment for the bikaverin production tted with studied models.

Table 2. Parameter kinetics estimated with two-parameter Gompertz model for bikaverin production. Exp. Parameter l 1 2 3 4 5 6 7 8 0.0132 0.1026 0.0146 0.0298 0.0280 0.0269 0.0306 0.0271 a 0.0262 0.1591 0.0235 0.0614 0.0441 0.0484 0.0472 0.0574 Yx/s 0.0955 0.3183 0.1076 0.1568 0.1089 0.0760 0.0821 0.1163 m 1E-20 0.0681 1E-20 0.0307 0.0385 0.0156 0.0240 0.0083 a 1.0718 0.0105 0.2617 0.1114 0.1384 0.0076 0.8651 0.2458 b )2.1132E-03 2.7560E-03 )3.1404E-04 6.1003E-04 4.7206E-04 2.9655E-03 1.9758E-03 3.7703E-03

Experimental data obtained from bikaverin production by the fungus Gibberella fujikuroi were tted to sigmoid models. Results from the F test showed that it is not necessary to have a three-parameter model, but a twoparameter model for prediction of growth kinetics. The two-parameter Gompertz model was accepted in 75% of the analysed experiments while the logistic model was accepted in 37.5% of the experiments. The three-parameter Gompertz model was eliminated by the F test and the convergence problems presented during parameter estimation. The Yx/s value in most of the analysed experiments is near to 0.1 g substrate/g biomass. In some cases, the maintenance coecient is not important, while in others it approximates a value of 0.02 g substrate/g biomass h. b values of LeudeckingPiret are positive and negative. Negative values appear when the model is approaching a maximum value of bikaverin due to its degradation. This suggests that bikaverin degradation could be considered as a rst order reaction due to biomass.

Acknowledgment The research was funded by Consejo Nacional de Ciencia y Tecnolog a (Project: 33973-B) and the Consejo n Tecnolo gica (Prodel Sistema Nacional de Educacio ject: 903.03-P).

0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00 0 24 48 72 96 Time, h 120 144

1.2

0.050 0.049 0.048 -ds/dt dp/dt 0.047 0.046 0.045 0.044 0.043

-ds/dt dp/dt

1.0 0.8 0.6 0.4 0.2 0.0 168

References
A.O.A.C. 1990 Ocial Methods of Analysis. Washington, DC: 15th Association of Ocial Chemists, ISBN 0-935584-42-0. Aynsley, M. & Ward, A.C. 1990 A mathematical model for the growth of mycelial fungi in submerged culture. Biotechnology and Bioengineering 35, 820830. Bailey, J.E. & Ollis, D.F.1998 Biochemical Engineering Fundamentals. NY: McGraw-Hill International Editions, Chemical Edition Series. ISBN 0-07-066601-6. Balan, J., Fuska, J., Kuhr, I. & Kuhrova, V. 1970 Bikaverin, an antibiotic from Gibberella fujikuroi, eective against Leishmania brasiliensis. Folia Microbiologica 15, 479484. Barton, D.H., Cottier, L., Freund, K., Luini, F., Magnus, P.D. & Salazar I. 1976 Total synthesis of bikaverin. Journal of the Chemical Society[Perkin 1] 5, 499503. Borrow, A., Jefferys, E.G., Kessel, R.H.J, Lloyd, E.C., Lloyd, P.B. & Nixon, I.S. 1961 Metabolism of Gibberella fujikuroi in stirrer culture. Canadian Journal of Microbiology 7, 227276. Coolbaugh, R.C. 1983 Early stages of gibberellin biosynthesis. In The Biochemistry and Physiology of Gibberellins, Vol. 1. ed. A. Crozier. New York: Praeger. pp. 5398. ISBN 0275909638. Cornforth, J.W. & Ryback, G. 1971 Isolation and characterization of a fungal vacuolation factor (bikaverin) Journal of the Chemical Society C: Organic 27862788.

Figure 5. Specic , substrate  growth rate (l) for Gibberella  fujikuroi  ds and Bikaverin production dp rates from the dt dt eighth experiment (glucose, 80 g l)1; NH4Cl, 3 g l)1; KH2PO4, 5 g l)1; )1 MgSO4, 2.5 g l ; Temperature, 29 C; pH, free; aeration rate, 3 v/v/ m; inoculum ratio, 10%; bead diameter, 3 mm). consumption

Figures 2 and 3 show tted data from the eighth experiment. Figure 5 presents l variations through the time and substrate consumption. It also presents product formation rates of the eighth experiment in which it is shown clearly that the bikaverin production is not

688
De Boer, J.J., Bright, D., Dalling, G. & Hewitt, T.G. 1971 Crystal and molecular structure of the chloroform solvate of bikaverin. Journal of the Chemical Society C: Organic, 27882791. Desjeux, P. 1992 Human Leishmaniases: epidemiology and public health aspects. World Health Statistical Quarterly 45, 267275. n del modelo de Gompertz para Escamilla-Silva, E.M. 1997 Aplicacio tico de crecimiento del hongo Gibberella fujikuroi en el estudio cine cultivo sumergido. VII Congreso Nacional de Biotecnologia y n Sin, Me xico. Bioingeniera, Mazatla Escamilla-Silva, E.M., Dendooven, L., Uscanga-Reynell, J.A., Mon lez-Alatorre, G. & de la Torre-Mart nez roy-Ram rez, A.I., Gonza M. 1999 Morphological development and gibberellin production by different strains of Gibberella fujikuroi in shake asks and bioreactor. World Journal of Microbiology and Biotechnology 15, 753755. Escamilla-Silva, E.M., Dendooven, L., Magan a-Plaza, I., Parra, Saldivar. R. & de la Torre-Martinez, M. 2000 Optimization of gibberellic acid production by immobilized Gibberella fujikuroi. Journal of Biotechnology 76, 147155. Escamilla-Silva, E.M., J menez,-Islas, H., Dendooven, L., Gutierrez lez, G.F. & Ochoa-Tapia J.A. 2001a A method to evaluate Gonza the isothermal effectiveness factor for dynamic of oxygen in mycelial pellets in submerged cultures. Biotechnology Progress 17, 95103. Escamilla-Silva, E.M., Poggi-Varaldo, H., de la Torre-Mart nez, M., nchez-Cornejo, G. & Dendooven, L. 2001b Selective production Sa of bikaverin in a uidized bed bioreactor with immobilized Gibberella fujikuroi mycelia. World Journal Microbiology and Biotechnology 17, 469474. Fredickson, A.G., Magee, R.D. & Tsuchiya, H.M. 1979 Mathematical models for fermentation processes. Advance in Applied Microbiology 13, 419. Fuska, J., Proksa, B. & Fuskova, A. 1975 New potential cytotoxic and antitumor substances. Neoplasma 22, 335338. Gouveia, E.R., Baptista-Neto, A., Hokka, C.O. & Badino Jr., A.C. 2000 Studies on the rheology and oxygen mass transfer in the clavulanic acid production by Streptomyces clavuligerus. Brazilian Journal of Chemical Engineering 17, 827834. Gutke, R. 1980 Theoretical basis of kinetics growth, Metabolism and product formation of microorganisms. In: UNEP/UNESCO/ ICRO Training Course. Vol. 1, pp. 30112. Jena, Science Academy of East Germany, Central Institute for Microbiology and Experimental Therapy (ZIMET). Humphrey, A.E. 1979 Fermentation process modelling. An overview. Annals of the New York Academy of Sciences 32, 1733.

Chavez-Parga et al.
Jones, A. & Pharis, R.P. 1987 Production of gibberellins and bikaverin by cells of Gibberella fujikuroi immobilized in carrageenan. Journal of Fermentation Technology 65, 717722. Katagiri, N., Nakano, J. & Kato, T. 1981 Synthesis of bikaverin. Journal of the Chemical Society [Perkin 1] 22, 27102716. Kjaer, D., Kjaer, A., Pedersen, C., Bu lock, J.D. & Smith J.R. 1971 Bikaverin and norbikaverin, benzoxanthentrione pigments of Gibberella fujikuroi. Journal of the Chemical Society C: Organic 27922797. Kossen, N.W.F. & Oosterhuis, N.M.G. 1985 Modelling and scaling-up of bioreactors. In Biotechnology, Vol. 2. eds. Rehm HJ, Reed G, pp. 571605. Wenheim: VCH. ISBN 0895730421. lez-Blanco, P.C. & Viniegra-Gonza lez, Larralde-Corona, C.P., Gonza G. 1994 Comparison of alternative kinetic models for estimating the specic growth rate of Gibberella fujikuroi by image analysis techniques. Biotechnology Techniques 8, 261266. Miller, G.L. 1959 Use of dinitrosalicylic acid reagent for determination of reducing sugars. Analytical Chemistry 31, 426428. Montgomery, D.C. 1991 Design and Analysis of Experiments, 2nd edn. New York: John Wiley & Sons. ISBN 0-471-52000-4. Negrete, R.X. 2002 Produccion de acido giberelico (GA3) en biorreactor de tanque agitado con Gibberella fujikuroi empleando aceite de maiz como fuente de carbono. Master-Science Thesis, Instituto Tecnologico de Celaya, Gto., Mexico. Nielsen, J. & Villadsen, J. 1992 Modelling of microbial kinetics. Chemical Engineering Science 47, 42254270. Peace, G.S. 1981 Taguchi Methods: A Hands-on Approach. New York: Addison-Wesley Publication Incorporated. ISBN 0-201-56311-8. Ross, P.J. 1989 Taguchi Techniques for Quality Engineering: Loss Function, Orthogonal Experiments, Parameters and Tolerance Design. NY: McGraw Hill. ISBN 0070539588. Sanchez, M.G. 1996 Obtencion de bikaverina mediante celulas inmovilizadas de Gibberella fujikuroi en biorreactor de lecho uidizado. Master-Science Thesis, Universidad Autonoma de Queretaro, Qro., Mexico. Stewart, W., Caracotsios, M. & Sorensen, J. 1990 GREG Subroutine. Madison: Department of Chemical Engineering, University of Wisconsin. Weiss, R. & Ollis, D. 1980 Extra cellular microbial polysaccharide kinetics. Biotechnology and Bioengineering 22, 859873. t Riet, K. Zwietering, M.H., Jongenburger, I., Rombouts, F.M. & Van 1990 Modelling of the bacterial growth curve. Applied and Environmental Microbiology 56, 18751881.