- Email: [email protected]
Optimization of fermentation conditions for the production of ethanol from sago starch by co-immobilized amyloglucosidase and cells of Zymomonas mobilis using response surface methodology
Enzyme and Microbial Technology 38 (2006) 209–214
Optimization of fermentation conditions for the production of ethanol from sago starch by co-immobilized amyloglucosidase and cells of Zymomonas mobilis using response surface methodology Veera Venkata Ratnam Bandaru a,∗ , Subba Rao Somalanka b,1 , Damodara Rao Mendu c,2 , Narasimha Rao Madicherla b,3 , Ayyanna Chityala b,4 a
Department of Neurology, Johns Hopkins University School of Medicine, 600 North Wolfe Street, Pathology 233/Meyer 222, Baltimore, MD 21287, USA b Center for Biotechnology, Department of Chemical Engineering, College of Engineering, Andhra University, Visakhapatnam 530003, India c Department of Pediatrics, Division of Infectious Diseases, Johns Hopkins University School of Medicine, 720 Rutland Avenue, Ross 1135B, Baltimore, MD 21205, USA Received 15 May 2005; received in revised form 6 June 2005; accepted 8 June 2005
Abstract Statistical experimental design was used to optimize the conditions of simultaneous saccharification and fermentation (SSF), viz. temperature, pH and time of fermentation of ethanol from sago starch with co-immobilized amyloglucosidase (AMG) and Zymomonas mobilis MTCC 92 by submerged fermentation. Maximum ethanol concentration of 55.3 g/l was obtained using a starch concentration of 150 g/l. The optimum conditions were found to be a temperature of 32.4 ◦ C, pH of 4.93 and time of fermentation of 17.24 h. Thus, by using SSF process with co-immobilized AMG and Z. mobilis cells MTCC 92, the central composite design (CCD) was found to be the most favourable strategy investigated with respect to ethanol production and enzyme recovery. © 2005 Elsevier Inc. All rights reserved. Keywords: Co-immobilization; Amyloglucosidase; Sago starch; Ethanol; Response surface methodology
1. Introduction The new developments in biotechnology will play an important role in resolving part of the energy and food problems that lie ahead. One important development, which has stimulated worldwide interest is the utilization of renewable carbohydrate sources for the production of ethanol as a liquid fuel [1–3]. Sago starch is an agricultural material abundantly produced in India and other tropical coun∗
Corresponding author. Tel.: +1 443 287 3717; fax: +1 410 502 6736. E-mail addresses: [email protected], [email protected] (V.V.R. Bandaru), [email protected] (D.R. Mendu), mnrmd [email protected] (N.R. Madicherla), [email protected] (A. Chityala). 1 Tel.: +91 891 284 4890/2534032. 2 Tel: +1 410 614 0058. 3 Tel.: +91 891 284 4883/2537641, mobile: +91 9440403268. 4 Tel.: +91 893 3222959/891 2552405. 0141-0229/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.enzmictec.2005.06.002
tries [4] and is an alternative source of energy. The starch is a product extracted from the seeds or tubers, stems of palms and cycads such as Metroxylon sagu and is a mixture of 27% (w/w) amylose and 73% (w/w) amylopectin [5]. There are a variety of products that can be obtained from starch biomass via hydrolysis. Alcohol is one of the largest volume of products that can be produced from biomass. Recently, there has been active research aimed at attaining an increase in ethanol yield by immobilized techniques. Zymomonas mobilis cells and AMG were co-immobilized in the form of alginate beads and SSF was carried using the co-immobilized cells and enzyme. The SSF process combines enzymatic hydrolysis of starch to glucose and ethanol fermentation into a single operation. Consequently, this process offers a great potential of increased rate of hydrolysis, reduction of fermentation time, decreased capital cost [6] and removing end point inhibition as well as eliminating the need
210
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
for separate reactors. SSF process has attracted many investigators [7,8]. The traditional ‘one-factor at a time’ technique used for optimizing a multivariable system is not only time consuming but also often easily misses the alternative effects between components. Also, this method requires to carry out a number of experiments to determine the optimum levels, which are untrue. These drawbacks of single factor optimization process can be eliminated by optimizing all the affecting parameters collectively by CCD using response surface methodology (RSM). Recently, many statistical experimental design methods have been employed in bioprocess optimization. Among them, RSM is the one suitable for identifying the effect of individual variables and for seeking the optimum conditions for a multivariable system efficiently. This method has been successfully applied to optimize alcoholic fermentation and other fermentation media [9–16]. A detailed account of this technique has been outlined [17]. Basically, this optimization process involves three major steps: performing the statistically designed experiments, estimating the coefficients in a mathematical model and predicting the response and checking the adequacy of the model. Hence, the authors report the application of the RSM using the Box–Wilson design [18] of experiments to develop a mathematical correlation between the temperature, pH and time of fermentation and concentration of ethanol. In the present study, the optimal conditions of temperature, pH and time of fermentation for maximum ethanol yield have been quantified from the Box–Wilson CCD.
2. Materials and methods
2.5. Co-immobilization The enzyme was immobilized on powdered chitin using the procedures of Stanley et al. [19]. Exponentially growing Z. mobilis cells (8 g dry cell weight) were centrifuged and resuspended with the immobilized glucoamylase on chitin in 50 ml physiological saline. The suspension was carefully mixed with 50 ml 4% sodium alginate solution. The slurry was then added drop wise to a 0.05 M CaCl2 solution with continuous stirring using a 5 ml disposable pipette tip. Beads of 3–4 mm diameter were formed in this solution. The total volume of beads was approximately 68 ml. The concentrations of the chitin immobilized glucoamylase and Z. mobilis cells in the beads were 77.2 and 93.2 g dry weight/l beads, respectively. 2.6. Production media and fermentation Starch liquefaction was carried out by adding 0.2% (v/w) ␣-amylase to the slurry at pH 6.5 and heating at 95 ◦ C for 1 h. No problem was faced during the solubilization of starch because of the reduction in viscosity of fermentation mashes by enzymes. Fermentation media were composed of 150 g sago starch, 10 g yeast extract and co-immobilized enzyme and cell beads (120 ml bead volume) in 1 l water. Fermentation was carried out in a Biostat M fermentor supplied by B. Braun Co., Germany, with all necessary controls. The reactor was of 2 l capacity and the working volume was 1 l. The operating conditions were maintained at a temperature of 30 ◦ C and pH 5.0. The reactor was maintained under anaerobic conditions.
2.1. Substrate
2.7. Analytical methods
Sago starch was collected from cultivators, East Godavari District, Andhra Pradesh, India.
Ethanol was estimated by GLC in which a flame ionization detector and stainless steel column (2.0 m length, 3.0 mm i.d.) packed with Porapak-Q (50–80 mesh, manufactured by Nucon Engineers, India) were used. The column oven was operated isothermally at 150 ◦ C and the detector and injection ports were kept at 170 ◦ C. Nitrogen was used as carrier gas at a flow rate of 30 cm3 /min and the combustion gas was a mixture of hydrogen and air [13]. Sugars were determined using Miller’s method [20].
2.2. Organism Z. mobilis MTCC 92 obtained from IMTECH, Chandigarh, India, was used throughout this study. 2.3. Enzymes ␣-Amylase from Bacillus licheniformis and AMG from Aspergillus niger were obtained from Sigma Chemical Co. (St Louis, MO). The activities of the two enzymes were 60 KNU/g and 200 AGU/ml, respectively. 2.4. Growth conditions The Z. mobilis MTCC 92 was maintained on agar slants having composition (g/l): glucose, 100; yeast extract, 10; KH2 PO4 , 1; (NH4 )2 SO4 , 1; MgSO4 ·7H2 O, 0.5 and the cells were grown at a temperature of 35 ◦ C and pH of 5.5.
2.8. Experimental design and optimization CCD [18] was used in the optimization of ethanol production. Temperature (X1 , ◦ C), pH (X2 ) and time of fermentation (X3 , h) were chosen for the independent variables shown in Tables 1 and 2. Ethanol concentration (Yi , g/l) was used as the dependent output variable. For statistical calculations the variables Xi were coded as xi according to Eq. (1): xi =
Xi − x¯ i , xj
i = 1, 2, 3, . . . , k
(1)
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
211
Table 1 Independent variables in the experimental plan
Table 3 Experimental and the predicted yields for ethanol
Variables
Run no.
Temperature, X1 pH, X2 Time, X3 (h)
Coded levels (◦ C)
−1.682
−1
0
1
1.682
13.18 3.318 6.59
20 4 10
30 5 15
40 6 20
46.82 6.682 23.41
Table 2 The central composite design matrix employed for three independent variables (actual values given in Table 1) Run no. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
X1
X2
X3
−1 1 −1 1 0 0 −1 1 −1 1 0 0 −1.682 1.682 0 0 0 0 0 0
−1 −1 1 1 0 0 −1 −1 1 1 0 0 0 0 −1.682 1.682 0 0 0 0
−1 1 1 −1 0 0 1 −1 −1 1 0 0 0 0 0 0 −1.682 1.682 0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
X1
20 40 20 40 30 30 20 40 20 40 30 30 13.18 46.82 30 30 30 30 30 30
X2
4 4 6 6 5 5 4 4 6 6 5 5 5 5 3.318 6.682 5 5 5 5
X3
10 20 20 10 15 15 20 10 10 20 15 15 15 15 15 15 6.59 23.41 15 15
Ethanol yield (g/l) Experimental
Predicted
25.92 43.54 32.57 19.48 52.10 51.92 26.43 30.11 24.16 40.64 51.01 52.40 11.45 22.62 37.42 34.60 32.57 39.37 52.00 51.89
26.2817 41.2643 30.3534 20.2888 51.8215 51.8215 24.0027 30.7082 24.8173 38.6599 51.8215 51.8215 12.8248 23.5334 38.865 35.4431 30.34735 43.8805 51.8215 51.8215
where xi is dimensionless value of an independent variable, Xi the real value of an independent variable, x¯ i the real value of the independent variable at the center point and xj is the step change. A 23 -factorial central √composite experimental design, with six axial points (α = 3)and six replications at the center points (n0 = 6) leading to a total number of 20 experiments was employed (Table 2) for the optimization of the parameters. The second degree polynomials (Eq. (2)) were calculated with the statistical package (Stat-Ease Inc., Minneapolis, MN, USA) to estimate the response of the dependent variable:
rial CCD and regression analysis. Also this method evaluating the effective factors and building models to study interaction and select optimum conditions of variables for a desirable response. The full CCD, based on three basic principles of an ideal experimental design, primarily consists of (1) a complete 2n factorial design, where n is the number of test variables, (2) n0 center points (n0 ≥ 1) and (3) two axial points on the axis of each design variable at a distance of 1.682 (2n/4 = 1.682 for n = 3) from the design center. Hence, the total number of design points (20) is N = 2n + 2n + n0 . The suitable temperature, initial pH and time of fermentation were also determined using statistical CCD. The experimental design matrix is given in Tables 1 and 2. Twenty experiments were performed using different combinations of the variables as per the CCD. Using the results of the experiments the following second order polynomial equation giving the ethanol as a function of temperature (X1 , ◦ C), pH (X2 ) and time of fermentation (X3 , h) was obtained:
Yi = b0 + b1 X1 + b2 X2 + b3 X3 + b11 X12 + b22 X22
Yi = −223.529 + 7.609949X1 + 51.6821X2 + 3.16389X3
+ b33 X32 + b12 X1 X2 + b23 X2 X3 + b13 X1 X3
(2)
where Yi is predicted response, X1 , X2 , X3 the independent variables, b0 the offset term, b1 , b2 , b3 the linear effects, b11 , b22 , b33 the squared effects and b12 , b23 , b13 are interaction terms.
3. Results and discussion The most important physical factors, which affect the fermentative production of ethanol are the temperature, initial pH and time of fermentation. The RSM includes full facto-
− 0.11891X12 − 5.18443X22 − 0.20794X32 − 0.22387X1 X2 + 0.39075X2 X3 + 0.064175X1 X3 (3) The predicted levels of ethanol using the above equation were given along with experimental data in Table 3. The coefficients of the regression model (Eq. (3)) calculated are listed in Table 4, in which they contain three linear, three quadratic and three interaction terms and one block term. The effects of all three parameters, i.e. temperature, pH and time of fermentation and their interactions with each other on percent ethanol concentration were found to be significant
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
212
Table 4 Coefficients, t-statistics and significance probability of the model Term
Coefficient
Value
Standard error of coefficient
t-Value
p-Value
Constant Temperature pH Time Temperature × temperature pH × pH Time × time Temperature × pH pH × time Temperature × time
b0 b1 b2 b3 b11 b22 b33 b12 b23 b13
−223.529 7.609949 51.6821 3.16389 −0.11891 −5.18443 −0.20794 −0.22387 0.39075 0.064175
26.01494 0.603539 7.032349 1.207079 0.006087 0.608655 0.024346 0.081707 0.163413 0.016341
−8.59232 12.60887 7.349195 2.621113 −19.5373 −8.51784 −8.54107 −2.73998 2.391174 3.927156
6.27E−06 1.83E−07 2.46E−05 0.025549 2.7E−09 6.77E−06 6.61E−06 0.020833 0.037883 0.002833
Fig. 1. Parity plot showing the distribution of experimental vs. predicted values of ethanol yield.
(p ≤ 0.05). The parity plot showed a satisfactory correlation between the values of experimental values and predictive values (Fig. 1), wherein, the points cluster around the diagonal line which indicates the good fit of the model, since the deviation between the experimental and predictive values was less. And also the goodness of the model could be checked by different criteria. The coefficient of determination, R2 is 0.9824 which implies that 98.24% of the sample variation in the ester yield is attributed to the independent variables. The R2 value also indicates that only 1% of the variation is not explained by the model. The value of R is 0.99. The corresponding analysis of variance (ANOVA) was presented in Table 5. Chi-square test was also carried out to check the best 2 < χ2 , since fit of the model. The model was a good fit. χcal tab 2 2 χcal is 1.6052 and χtab is 30.14. The predicted optimum levels of temperature pH and time of fermentation were obtained by applying the regression analysis to the Eq. (3). The predicted
Fig. 2. Response surface and contour plot of temperatures vs. pH on ethanol production (time was kept constant at 15 h).
and experimental ethanol concentration at the optimum levels of fermentation conditions were also determined by using Eq. (3). Figs. 2–4 represent the isoresponse contour and surface plots for the optimization of fermentation conditions of ethanol. The effects of the pH and temperature on the ethanol production showed in Fig. 2. An increase in the temperature with pH up to the optimum point increased the ethanol production to a maximum level and a further increase in the temperature with pH the trend is reversed. The interaction effect of the time and pH on the ethanol production in Fig. 3 clearly indicates a proper combination for production of ethanol. An increase in the pH with time increased the ethanol production gradually but at a higher pH and time the trend is reversed. The optimum for maximum ethanol production lies near the centre point of the pH and time. A similar
Table 5 ANOVA for the entire quadratic model Source of variation
Sum of squares (SS)
Degrees of freedom (d.f.)
Mean squares (MS)
F-value
Probe > F
Regression Residual
2994.509 53.40789
9 10
332.7233 5.340789
62.29852
0
Total
3047.917
19
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
Fig. 3. Response surface and contour plot of pH vs. time on ethanol production (temperature was kept constant at 30 ◦ C).
213
at fixed levels is laborious and time consuming. This method requires a complete series of experiments for every factor of interest. Moreover, such a method does not provide means of observing possible factors interactions. In contrast, CCD offer a number of important advantages. For instance, the researchers could easily determine factor effects with considerably less experimental effort, identify factors, find optima, offer greater precision and facilitate system modeling. Thus, the present study using the RSM with CCD enables to find the importance of factors at different levels. A high similarity was observed between the predicted and experimental results, which reflected the accuracy and applicability of RSM to optimize the process for ethanol production. The results of this study have clearly indicated RSM is an effective method for maximum production of ethanol using SSF with AMG and Z. mobilis MTCC 92.
References
Fig. 4. Response surface and contour plot of temperature vs. time on ethanol production (pH was kept constant at 15).
effect on the response was observed for the temperature at any level of the time an increase in the temperature with time up to the optimum point increased the ethanol production to maximum level and a further increase in the temperature with time decreased the ethanol production is shown in Fig. 4. Therefore, an optimum was observed near the central value of pH, temperature and time. The optimum conditions for maximum ethanol concentrations were obtained at a temperature of 32.4 ◦ C, pH of 4.93 and time of fermentation of 17.24 h. A maximum ethanol concentration of 55.3 g/l was obtained at these optimum parameters.
4. Conclusion The one factor at a time is the most frequently used operation in optimization process. This technique is based on changing one parameter at a time, while keeping the others
[1] Scheller WA. The clean fuel from biomass and wastes. In: Proceedings of the institute of technology. 1977. [2] Prescott SC, Dunn CG. Industrial microbiology. 3rd ed. New York: McGraw-Hill; 1959. [3] Lindeman LR, Rocchiccioli C. Ethanol in Brazil: brief summary of the state of industry in 1977. Biotechnol Bioeng 1979;31: 1107–19. [4] Tan K. Production of sago starch. In: Stanton WR, Flach M, editors. Proceedings of the second international sago symposium. The Hague: Martinus Nijho; 1980. p. 56. ISBN 90-2472470-8. [5] Swinkels JJM. In: Van Beynum GMA, Roels JA, editors. Starch conversion technology. New York: Marcel Dekker; 1985. p. 23. ISBN 0-82477194-X. [6] Kosaric N, Wieczorek A, Cosentino G, Magee R. In: Dellweg H, editor. Biotechnology, vol. 3. Weinheim: Verlag Chemie; 1983. p. 315. ISBN 3-52725765-9. [7] Deshpande V, Raman HS, Rao M. Simultaneous saccharification and fermentation of cellulose to ethanol using penicillium funiculosum cellulose and free or immobilized Saccharomyces uvarum cells. Biotechnol Bioeng 1983;25:1679–84. [8] Spangler DJ, Emert GH. Simultaneous saccharification and fermentation with Z. mobilis. Biotechnol Bioeng 1986;28:115–8. [9] Chen SL. Optimization of batch alcohol fermentation of glucose syrup substrate. Biotechnol Bioeng 1981;23:1827–36. [10] Bowman L, Geiger E. Optimization of fermentation conditions for alcohol production. Biotechnol Bioeng 1984;26:1492–7. [11] Zertuche L, Zall RR. Optimizing alcohol production from whey using computer technology. Biotechnol Bioeng 1985;27:547–54. [12] Ratnam BVV. Studies on physico-chemical and nutritional parameters for the production of ethanol from palmyra jaggery by submerged fermentation using Saccharomyces cerevisiae. Ph.D. Thesis, Andhra University, Visakhapatnam, AP, India; 2001. [13] Ratnam BVV, Narasimha Rao M, Damodara Rao M, Subba Rao S, Ayyanna C. Optimization of fermentation conditions for the production of ethanol from sago starch using response methodology. World J Microbiol Biotechnol 2003;19:523–6. [14] Sunitha I, Subba Rao MV, Ayyanna C. Optimization of medium constituents and fermentation conditions for the production of L-glutamic acid by the coimmobilized whole cells of Micrococus glutanicus and Pseudomonas reptilivora. Bioprocess Eng 1998;18:353–9. [15] Ambati P, Ayyanna C. Optimizing medium constituents and fermentation conditions for citric acid production from palmyra jag-
214
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
gery using response surface method. World J Microbiol Biotechnol 2001;17:331–5. [16] Maddox IS, Reichert SH. Use of response surface methodology for the rapid optimization of microbiological media. J Appl Bacteriol 1977;43:197–204. [17] Cochran NG, Cox GM. Experimental designs. John Wiley and Sons Inc.; 1968. p. 611.
[18] Box GEP, Wilson KB. On the experimental attainment of optimum conditions. J R Stat Soc (Ser B) 1951;13:1–45. [19] Stanley WL, Watters GG, Kelly SH, Olson AC. Glucoamylase immobilized on chitin with Glutaraldehyde. Biotechnol Bioeng 1978;20:135–40. [20] Miller GL. Use of DNS reagent for determination of reducing sugars. Anal Chem 1959;31:426–8.
Optimization of fermentation conditions for the production of ethanol from sago starch by co-immobilized amyloglucosidase and cells of Zymomonas mobilis using response surface methodology Veera Venkata Ratnam Bandaru a,∗ , Subba Rao Somalanka b,1 , Damodara Rao Mendu c,2 , Narasimha Rao Madicherla b,3 , Ayyanna Chityala b,4 a
Department of Neurology, Johns Hopkins University School of Medicine, 600 North Wolfe Street, Pathology 233/Meyer 222, Baltimore, MD 21287, USA b Center for Biotechnology, Department of Chemical Engineering, College of Engineering, Andhra University, Visakhapatnam 530003, India c Department of Pediatrics, Division of Infectious Diseases, Johns Hopkins University School of Medicine, 720 Rutland Avenue, Ross 1135B, Baltimore, MD 21205, USA Received 15 May 2005; received in revised form 6 June 2005; accepted 8 June 2005
Abstract Statistical experimental design was used to optimize the conditions of simultaneous saccharification and fermentation (SSF), viz. temperature, pH and time of fermentation of ethanol from sago starch with co-immobilized amyloglucosidase (AMG) and Zymomonas mobilis MTCC 92 by submerged fermentation. Maximum ethanol concentration of 55.3 g/l was obtained using a starch concentration of 150 g/l. The optimum conditions were found to be a temperature of 32.4 ◦ C, pH of 4.93 and time of fermentation of 17.24 h. Thus, by using SSF process with co-immobilized AMG and Z. mobilis cells MTCC 92, the central composite design (CCD) was found to be the most favourable strategy investigated with respect to ethanol production and enzyme recovery. © 2005 Elsevier Inc. All rights reserved. Keywords: Co-immobilization; Amyloglucosidase; Sago starch; Ethanol; Response surface methodology
1. Introduction The new developments in biotechnology will play an important role in resolving part of the energy and food problems that lie ahead. One important development, which has stimulated worldwide interest is the utilization of renewable carbohydrate sources for the production of ethanol as a liquid fuel [1–3]. Sago starch is an agricultural material abundantly produced in India and other tropical coun∗
Corresponding author. Tel.: +1 443 287 3717; fax: +1 410 502 6736. E-mail addresses: [email protected], [email protected] (V.V.R. Bandaru), [email protected] (D.R. Mendu), mnrmd [email protected] (N.R. Madicherla), [email protected] (A. Chityala). 1 Tel.: +91 891 284 4890/2534032. 2 Tel: +1 410 614 0058. 3 Tel.: +91 891 284 4883/2537641, mobile: +91 9440403268. 4 Tel.: +91 893 3222959/891 2552405. 0141-0229/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.enzmictec.2005.06.002
tries [4] and is an alternative source of energy. The starch is a product extracted from the seeds or tubers, stems of palms and cycads such as Metroxylon sagu and is a mixture of 27% (w/w) amylose and 73% (w/w) amylopectin [5]. There are a variety of products that can be obtained from starch biomass via hydrolysis. Alcohol is one of the largest volume of products that can be produced from biomass. Recently, there has been active research aimed at attaining an increase in ethanol yield by immobilized techniques. Zymomonas mobilis cells and AMG were co-immobilized in the form of alginate beads and SSF was carried using the co-immobilized cells and enzyme. The SSF process combines enzymatic hydrolysis of starch to glucose and ethanol fermentation into a single operation. Consequently, this process offers a great potential of increased rate of hydrolysis, reduction of fermentation time, decreased capital cost [6] and removing end point inhibition as well as eliminating the need
210
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
for separate reactors. SSF process has attracted many investigators [7,8]. The traditional ‘one-factor at a time’ technique used for optimizing a multivariable system is not only time consuming but also often easily misses the alternative effects between components. Also, this method requires to carry out a number of experiments to determine the optimum levels, which are untrue. These drawbacks of single factor optimization process can be eliminated by optimizing all the affecting parameters collectively by CCD using response surface methodology (RSM). Recently, many statistical experimental design methods have been employed in bioprocess optimization. Among them, RSM is the one suitable for identifying the effect of individual variables and for seeking the optimum conditions for a multivariable system efficiently. This method has been successfully applied to optimize alcoholic fermentation and other fermentation media [9–16]. A detailed account of this technique has been outlined [17]. Basically, this optimization process involves three major steps: performing the statistically designed experiments, estimating the coefficients in a mathematical model and predicting the response and checking the adequacy of the model. Hence, the authors report the application of the RSM using the Box–Wilson design [18] of experiments to develop a mathematical correlation between the temperature, pH and time of fermentation and concentration of ethanol. In the present study, the optimal conditions of temperature, pH and time of fermentation for maximum ethanol yield have been quantified from the Box–Wilson CCD.
2. Materials and methods
2.5. Co-immobilization The enzyme was immobilized on powdered chitin using the procedures of Stanley et al. [19]. Exponentially growing Z. mobilis cells (8 g dry cell weight) were centrifuged and resuspended with the immobilized glucoamylase on chitin in 50 ml physiological saline. The suspension was carefully mixed with 50 ml 4% sodium alginate solution. The slurry was then added drop wise to a 0.05 M CaCl2 solution with continuous stirring using a 5 ml disposable pipette tip. Beads of 3–4 mm diameter were formed in this solution. The total volume of beads was approximately 68 ml. The concentrations of the chitin immobilized glucoamylase and Z. mobilis cells in the beads were 77.2 and 93.2 g dry weight/l beads, respectively. 2.6. Production media and fermentation Starch liquefaction was carried out by adding 0.2% (v/w) ␣-amylase to the slurry at pH 6.5 and heating at 95 ◦ C for 1 h. No problem was faced during the solubilization of starch because of the reduction in viscosity of fermentation mashes by enzymes. Fermentation media were composed of 150 g sago starch, 10 g yeast extract and co-immobilized enzyme and cell beads (120 ml bead volume) in 1 l water. Fermentation was carried out in a Biostat M fermentor supplied by B. Braun Co., Germany, with all necessary controls. The reactor was of 2 l capacity and the working volume was 1 l. The operating conditions were maintained at a temperature of 30 ◦ C and pH 5.0. The reactor was maintained under anaerobic conditions.
2.1. Substrate
2.7. Analytical methods
Sago starch was collected from cultivators, East Godavari District, Andhra Pradesh, India.
Ethanol was estimated by GLC in which a flame ionization detector and stainless steel column (2.0 m length, 3.0 mm i.d.) packed with Porapak-Q (50–80 mesh, manufactured by Nucon Engineers, India) were used. The column oven was operated isothermally at 150 ◦ C and the detector and injection ports were kept at 170 ◦ C. Nitrogen was used as carrier gas at a flow rate of 30 cm3 /min and the combustion gas was a mixture of hydrogen and air [13]. Sugars were determined using Miller’s method [20].
2.2. Organism Z. mobilis MTCC 92 obtained from IMTECH, Chandigarh, India, was used throughout this study. 2.3. Enzymes ␣-Amylase from Bacillus licheniformis and AMG from Aspergillus niger were obtained from Sigma Chemical Co. (St Louis, MO). The activities of the two enzymes were 60 KNU/g and 200 AGU/ml, respectively. 2.4. Growth conditions The Z. mobilis MTCC 92 was maintained on agar slants having composition (g/l): glucose, 100; yeast extract, 10; KH2 PO4 , 1; (NH4 )2 SO4 , 1; MgSO4 ·7H2 O, 0.5 and the cells were grown at a temperature of 35 ◦ C and pH of 5.5.
2.8. Experimental design and optimization CCD [18] was used in the optimization of ethanol production. Temperature (X1 , ◦ C), pH (X2 ) and time of fermentation (X3 , h) were chosen for the independent variables shown in Tables 1 and 2. Ethanol concentration (Yi , g/l) was used as the dependent output variable. For statistical calculations the variables Xi were coded as xi according to Eq. (1): xi =
Xi − x¯ i , xj
i = 1, 2, 3, . . . , k
(1)
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
211
Table 1 Independent variables in the experimental plan
Table 3 Experimental and the predicted yields for ethanol
Variables
Run no.
Temperature, X1 pH, X2 Time, X3 (h)
Coded levels (◦ C)
−1.682
−1
0
1
1.682
13.18 3.318 6.59
20 4 10
30 5 15
40 6 20
46.82 6.682 23.41
Table 2 The central composite design matrix employed for three independent variables (actual values given in Table 1) Run no. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
X1
X2
X3
−1 1 −1 1 0 0 −1 1 −1 1 0 0 −1.682 1.682 0 0 0 0 0 0
−1 −1 1 1 0 0 −1 −1 1 1 0 0 0 0 −1.682 1.682 0 0 0 0
−1 1 1 −1 0 0 1 −1 −1 1 0 0 0 0 0 0 −1.682 1.682 0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
X1
20 40 20 40 30 30 20 40 20 40 30 30 13.18 46.82 30 30 30 30 30 30
X2
4 4 6 6 5 5 4 4 6 6 5 5 5 5 3.318 6.682 5 5 5 5
X3
10 20 20 10 15 15 20 10 10 20 15 15 15 15 15 15 6.59 23.41 15 15
Ethanol yield (g/l) Experimental
Predicted
25.92 43.54 32.57 19.48 52.10 51.92 26.43 30.11 24.16 40.64 51.01 52.40 11.45 22.62 37.42 34.60 32.57 39.37 52.00 51.89
26.2817 41.2643 30.3534 20.2888 51.8215 51.8215 24.0027 30.7082 24.8173 38.6599 51.8215 51.8215 12.8248 23.5334 38.865 35.4431 30.34735 43.8805 51.8215 51.8215
where xi is dimensionless value of an independent variable, Xi the real value of an independent variable, x¯ i the real value of the independent variable at the center point and xj is the step change. A 23 -factorial central √composite experimental design, with six axial points (α = 3)and six replications at the center points (n0 = 6) leading to a total number of 20 experiments was employed (Table 2) for the optimization of the parameters. The second degree polynomials (Eq. (2)) were calculated with the statistical package (Stat-Ease Inc., Minneapolis, MN, USA) to estimate the response of the dependent variable:
rial CCD and regression analysis. Also this method evaluating the effective factors and building models to study interaction and select optimum conditions of variables for a desirable response. The full CCD, based on three basic principles of an ideal experimental design, primarily consists of (1) a complete 2n factorial design, where n is the number of test variables, (2) n0 center points (n0 ≥ 1) and (3) two axial points on the axis of each design variable at a distance of 1.682 (2n/4 = 1.682 for n = 3) from the design center. Hence, the total number of design points (20) is N = 2n + 2n + n0 . The suitable temperature, initial pH and time of fermentation were also determined using statistical CCD. The experimental design matrix is given in Tables 1 and 2. Twenty experiments were performed using different combinations of the variables as per the CCD. Using the results of the experiments the following second order polynomial equation giving the ethanol as a function of temperature (X1 , ◦ C), pH (X2 ) and time of fermentation (X3 , h) was obtained:
Yi = b0 + b1 X1 + b2 X2 + b3 X3 + b11 X12 + b22 X22
Yi = −223.529 + 7.609949X1 + 51.6821X2 + 3.16389X3
+ b33 X32 + b12 X1 X2 + b23 X2 X3 + b13 X1 X3
(2)
where Yi is predicted response, X1 , X2 , X3 the independent variables, b0 the offset term, b1 , b2 , b3 the linear effects, b11 , b22 , b33 the squared effects and b12 , b23 , b13 are interaction terms.
3. Results and discussion The most important physical factors, which affect the fermentative production of ethanol are the temperature, initial pH and time of fermentation. The RSM includes full facto-
− 0.11891X12 − 5.18443X22 − 0.20794X32 − 0.22387X1 X2 + 0.39075X2 X3 + 0.064175X1 X3 (3) The predicted levels of ethanol using the above equation were given along with experimental data in Table 3. The coefficients of the regression model (Eq. (3)) calculated are listed in Table 4, in which they contain three linear, three quadratic and three interaction terms and one block term. The effects of all three parameters, i.e. temperature, pH and time of fermentation and their interactions with each other on percent ethanol concentration were found to be significant
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
212
Table 4 Coefficients, t-statistics and significance probability of the model Term
Coefficient
Value
Standard error of coefficient
t-Value
p-Value
Constant Temperature pH Time Temperature × temperature pH × pH Time × time Temperature × pH pH × time Temperature × time
b0 b1 b2 b3 b11 b22 b33 b12 b23 b13
−223.529 7.609949 51.6821 3.16389 −0.11891 −5.18443 −0.20794 −0.22387 0.39075 0.064175
26.01494 0.603539 7.032349 1.207079 0.006087 0.608655 0.024346 0.081707 0.163413 0.016341
−8.59232 12.60887 7.349195 2.621113 −19.5373 −8.51784 −8.54107 −2.73998 2.391174 3.927156
6.27E−06 1.83E−07 2.46E−05 0.025549 2.7E−09 6.77E−06 6.61E−06 0.020833 0.037883 0.002833
Fig. 1. Parity plot showing the distribution of experimental vs. predicted values of ethanol yield.
(p ≤ 0.05). The parity plot showed a satisfactory correlation between the values of experimental values and predictive values (Fig. 1), wherein, the points cluster around the diagonal line which indicates the good fit of the model, since the deviation between the experimental and predictive values was less. And also the goodness of the model could be checked by different criteria. The coefficient of determination, R2 is 0.9824 which implies that 98.24% of the sample variation in the ester yield is attributed to the independent variables. The R2 value also indicates that only 1% of the variation is not explained by the model. The value of R is 0.99. The corresponding analysis of variance (ANOVA) was presented in Table 5. Chi-square test was also carried out to check the best 2 < χ2 , since fit of the model. The model was a good fit. χcal tab 2 2 χcal is 1.6052 and χtab is 30.14. The predicted optimum levels of temperature pH and time of fermentation were obtained by applying the regression analysis to the Eq. (3). The predicted
Fig. 2. Response surface and contour plot of temperatures vs. pH on ethanol production (time was kept constant at 15 h).
and experimental ethanol concentration at the optimum levels of fermentation conditions were also determined by using Eq. (3). Figs. 2–4 represent the isoresponse contour and surface plots for the optimization of fermentation conditions of ethanol. The effects of the pH and temperature on the ethanol production showed in Fig. 2. An increase in the temperature with pH up to the optimum point increased the ethanol production to a maximum level and a further increase in the temperature with pH the trend is reversed. The interaction effect of the time and pH on the ethanol production in Fig. 3 clearly indicates a proper combination for production of ethanol. An increase in the pH with time increased the ethanol production gradually but at a higher pH and time the trend is reversed. The optimum for maximum ethanol production lies near the centre point of the pH and time. A similar
Table 5 ANOVA for the entire quadratic model Source of variation
Sum of squares (SS)
Degrees of freedom (d.f.)
Mean squares (MS)
F-value
Probe > F
Regression Residual
2994.509 53.40789
9 10
332.7233 5.340789
62.29852
0
Total
3047.917
19
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
Fig. 3. Response surface and contour plot of pH vs. time on ethanol production (temperature was kept constant at 30 ◦ C).
213
at fixed levels is laborious and time consuming. This method requires a complete series of experiments for every factor of interest. Moreover, such a method does not provide means of observing possible factors interactions. In contrast, CCD offer a number of important advantages. For instance, the researchers could easily determine factor effects with considerably less experimental effort, identify factors, find optima, offer greater precision and facilitate system modeling. Thus, the present study using the RSM with CCD enables to find the importance of factors at different levels. A high similarity was observed between the predicted and experimental results, which reflected the accuracy and applicability of RSM to optimize the process for ethanol production. The results of this study have clearly indicated RSM is an effective method for maximum production of ethanol using SSF with AMG and Z. mobilis MTCC 92.
References
Fig. 4. Response surface and contour plot of temperature vs. time on ethanol production (pH was kept constant at 15).
effect on the response was observed for the temperature at any level of the time an increase in the temperature with time up to the optimum point increased the ethanol production to maximum level and a further increase in the temperature with time decreased the ethanol production is shown in Fig. 4. Therefore, an optimum was observed near the central value of pH, temperature and time. The optimum conditions for maximum ethanol concentrations were obtained at a temperature of 32.4 ◦ C, pH of 4.93 and time of fermentation of 17.24 h. A maximum ethanol concentration of 55.3 g/l was obtained at these optimum parameters.
4. Conclusion The one factor at a time is the most frequently used operation in optimization process. This technique is based on changing one parameter at a time, while keeping the others
[1] Scheller WA. The clean fuel from biomass and wastes. In: Proceedings of the institute of technology. 1977. [2] Prescott SC, Dunn CG. Industrial microbiology. 3rd ed. New York: McGraw-Hill; 1959. [3] Lindeman LR, Rocchiccioli C. Ethanol in Brazil: brief summary of the state of industry in 1977. Biotechnol Bioeng 1979;31: 1107–19. [4] Tan K. Production of sago starch. In: Stanton WR, Flach M, editors. Proceedings of the second international sago symposium. The Hague: Martinus Nijho; 1980. p. 56. ISBN 90-2472470-8. [5] Swinkels JJM. In: Van Beynum GMA, Roels JA, editors. Starch conversion technology. New York: Marcel Dekker; 1985. p. 23. ISBN 0-82477194-X. [6] Kosaric N, Wieczorek A, Cosentino G, Magee R. In: Dellweg H, editor. Biotechnology, vol. 3. Weinheim: Verlag Chemie; 1983. p. 315. ISBN 3-52725765-9. [7] Deshpande V, Raman HS, Rao M. Simultaneous saccharification and fermentation of cellulose to ethanol using penicillium funiculosum cellulose and free or immobilized Saccharomyces uvarum cells. Biotechnol Bioeng 1983;25:1679–84. [8] Spangler DJ, Emert GH. Simultaneous saccharification and fermentation with Z. mobilis. Biotechnol Bioeng 1986;28:115–8. [9] Chen SL. Optimization of batch alcohol fermentation of glucose syrup substrate. Biotechnol Bioeng 1981;23:1827–36. [10] Bowman L, Geiger E. Optimization of fermentation conditions for alcohol production. Biotechnol Bioeng 1984;26:1492–7. [11] Zertuche L, Zall RR. Optimizing alcohol production from whey using computer technology. Biotechnol Bioeng 1985;27:547–54. [12] Ratnam BVV. Studies on physico-chemical and nutritional parameters for the production of ethanol from palmyra jaggery by submerged fermentation using Saccharomyces cerevisiae. Ph.D. Thesis, Andhra University, Visakhapatnam, AP, India; 2001. [13] Ratnam BVV, Narasimha Rao M, Damodara Rao M, Subba Rao S, Ayyanna C. Optimization of fermentation conditions for the production of ethanol from sago starch using response methodology. World J Microbiol Biotechnol 2003;19:523–6. [14] Sunitha I, Subba Rao MV, Ayyanna C. Optimization of medium constituents and fermentation conditions for the production of L-glutamic acid by the coimmobilized whole cells of Micrococus glutanicus and Pseudomonas reptilivora. Bioprocess Eng 1998;18:353–9. [15] Ambati P, Ayyanna C. Optimizing medium constituents and fermentation conditions for citric acid production from palmyra jag-
214
V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209–214
gery using response surface method. World J Microbiol Biotechnol 2001;17:331–5. [16] Maddox IS, Reichert SH. Use of response surface methodology for the rapid optimization of microbiological media. J Appl Bacteriol 1977;43:197–204. [17] Cochran NG, Cox GM. Experimental designs. John Wiley and Sons Inc.; 1968. p. 611.
[18] Box GEP, Wilson KB. On the experimental attainment of optimum conditions. J R Stat Soc (Ser B) 1951;13:1–45. [19] Stanley WL, Watters GG, Kelly SH, Olson AC. Glucoamylase immobilized on chitin with Glutaraldehyde. Biotechnol Bioeng 1978;20:135–40. [20] Miller GL. Use of DNS reagent for determination of reducing sugars. Anal Chem 1959;31:426–8.