- Email: [email protected]
Optimization of isoamyl acetate production by using immobilized lipase from Mucor miehei by response surface methodology
Enzyme and Microbial Technology 26 (2000) 131–136
Optimization of isoamyl acetate production by using immobilized lipase from Mucor miehei by response surface methodology S. Hari Krishnaa, B. Manoharb, S. Divakara, S. G. Prapullaa, N. G. Karantha,* a
Fermentation Technology & Bio–Engineering Department, Central Food Technological Research Institute, Mysore 570 013, India b Food Engineering Department, Central Food Technological Research Institute, Mysore 570 013, India Received 7 December 1998; received in revised form 19 July 1999; accepted 20 July 1999
Abstract Immobilized lipase from Mucor miehei was employed for the esterification of isoamyl alcohol with acetic acid in n-heptane solvent. The important process variables studied were enzyme/substrate (E/S) ratio, alcohol (acid) concentration, and incubation period. Based on Box–Behnken design of experiments, a second order response function was developed. The percentage esterification increased with both E/S ratio and time and decreased with alcohol (acid) concentration. The model indicated optimum conditions for maximum esterification ranging from 20 to 99.6% in the alcohol (acid) concentration range of 0.031 to 0.3 M for a range of E/S ratios 8.33 to 50 g/mol, which were in good agreement with the experimental yields. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Isoamyl acetate; Response surface methodology; Esterification; Lipase; Lipozyme IM; Enzymatic synthesis; Mucor miehei
1. Introduction Isoamyl esters (in particular isoamyl acetate) are flavor esters of commercial importance. Traditionally, they have been produced by chemical synthesis or extracted from natural sources. Isoamyl acetate has a strong banana flavor and hence is desired very much in food industry [1]. Annual demand for isoamyl acetate in USA alone amounts to about 74 tonnes [1]. With increasing orientation toward ‘natural’ flavors, production of flavor esters by ‘natural’ means like employment of enzymes or through microbial sources is gaining importance industrially. Lipases (triacylglycerol hydrolases, EC 3.1.1.3) usually catalyze hydrolytic reactions. However, when employed in low water environment, they can perform the reverse reaction, namely, esterification also. Lipasecatalyzed biotransformations are gaining importance because of their regio-, stereo-, and substrate specificity’s, their milder reaction conditions and the relatively lower energy requirement. Enzymatic reactions can be efficiently accomplished by employing lipases in an adequate organic solvent, such as heptane and hexane, thereby shifting the * Corresponding author. Tel.: ⫹91-821-515-792; fax: ⫹91-821-517233. E-mail address: [email protected] (N.G. Karanth)
reaction equilibrium toward esterification rather than hydrolysis. Lipases have been employed for direct esterification and transesterification reactions in organic solvents to produce esters of glycerol [2], aliphatic alcohol’s [3], and terpene alcohol’s [4]. However the esterification of shortchain fatty acids and alcohol’s has not received much attention. Moreover, short-chain (⬍C5) acids and alcohol’s are found to exert inhibitory effects on the enzyme [5,6]. Data generation for esterification with such substrates is important because short-chain esters are vital components of many fruit flavors. Few attempts have been made to establish the feasibility of synthesizing isoamyl acetate by employing lipases from different sources [5–9]. However, use of high enzyme concentration and low yields have been the significant drawbacks. Also the lipase catalyzed esterification reaction has not been examined thoroughly. Reports on the use of high enzyme concentration leading to 80% yields are available [6,8,9]. Considering the high demand and benefits, an optimized process for high yield enzymatic synthesis of isoamyl esters, utilizing low enzyme contents, is important. The present work aims at a better understanding of the relationships between the important esterification variables (substrate and enzyme concentrations, and incubation period) and ester yield to determine optimum conditions for the synthesis of isoamyl acetate catalyzed by Lipozyme
0141-0229/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 1 - 0 2 2 9 ( 9 9 ) 0 0 1 4 9 - 0
132
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
Table 1 Levels of variables chosen for the study Variable
Levels
Xe Enzyme/substrate ratio, g/mol Xs Alcohol (acid) concentration, M Xt Incubation period, h
⫹1
0
⫺1
50.00 0.30 72
33.33 0.18 48
16.67 0.06 24
IM-20 (immobilized lipase from Mucor miehei). Response surface methodology (RSM), which is an efficient statistical technique for optimization of multiple variables to predict best performance conditions with minimum number of experiments, was employed, the results of which are discussed in the present work.
Equimolar concentrations of isoamyl alcohol and acetic acid were taken in n-heptane, followed by different amounts of the immobilized enzyme. The reaction mixture was agitated at 150 rev./min on an orbital shaker at 39 ⫾ 1°C. Enzyme/substrate (E/S) ratio, alcohol (acid) concentration and incubation period were at three levels namely ⫺1, 0, and 1 corresponding to minimum, middle and maximum (Table 1).
2.3. E/S ratio To rationalize the enzyme concentration with respect to substrate (acid), a term called E/S ratio was used: E/S ratio (g/mol) ⫽ enzyme (g/liter)/substrate (mol/liter). 2.4. Analysis
2. Materials and methods 2.1. Materials Lipozyme IM-20 (Mucor miehei lipase immobilized on Duolite weak anion-exchange resin) was a gift from M/s Novo Nordisk, Denmark, which exhibited a specific hydrolytic activity of 20 mol/min/mg of protein at pH 7.0 and 30°C, with tributyrin as the substrate. All the chemicals employed were of analytical reagent grade. Isoamyl alcohol was from M/s Sarabhai M. Chemicals Ltd. (Baroda, India) and heptane and acetic acid were from M/s Sisco Research Laboratories (Bombay, India). Solvents were distilled once before use. 2.2. Esterification method Isoamyl acetate synthesis was carried out in 100 ml stoppered conical flasks with 10 ml working volume.
The extent of esterification was estimated at regular intervals of time by titrating aliquots (0.5 ml) of the reaction mixture against 0.01 N NaOH by using phenolphthalein as the indicator. The percentage esterification was determined from the difference in the volume of alkali consumed between aliquots of the reaction mixture withdrawn at the beginning of the reaction and after certain period of time. The analysis of the reaction mixture was also carried out by subjecting the samples to gas chromatographic analysis by using a Shimadzu gas chromatograph (GC 15-A) equipped with a Carbowax 20 M column (3 m length, 3.175 mm internal diameter [i.d.]) and a flame ionization detector. Nitrogen was used as a carrier gas with a flow rate of 30 ml/min. Column oven, injection port, and detector temperatures were at 100°C, 200°C, and 250°C respectively. The percentage of esterification determined by both GC analysis (which showed product formation) and titration (which showed acid consumption) were found to be in good agree-
Table 2 Actual values of variables as per Box-Behnken design of experiments and percentage esterification obtained SI. No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Enzyme/ substrate ratio g/mol
Alcohol (acid) concentration M
Incubation period h
Esterification, % Experimental
Predicted
50.00 50.00 16.67 16.67 50.00 50.00 16.67 16.67 33.33 33.33 33.33 33.33 33.33 33.33 33.33
0.30 0.06 0.30 0.06 0.18 0.18 0.18 0.18 0.30 0.30 0.06 0.06 0.18 0.18 0.18
48 48 48 48 72 24 72 24 72 24 72 24 48 48 48
26.8 89.3 10.9 59.4 35.9 31.3 27.1 17.1 21.5 16.8 90.0 40.0 25.1 26.1 28.6
27.6 78.3 10.4 61.1 44.6 27.3 27.4 10.1 16.3 21.7 89.7 49.7 27.3 27.3 27.3
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136 Table 3 Model validation experiments Enzyme/ substrate ratio g/mol
Alcohol (acid) concentration M
Incubation period h
Esterification, % Experimental
Predicted
28.00 50.00 25.00 41.67 37.23 43.68 50.00 46.67 33.99 16.67 33.33 50.00 16.67 33.33 50.00 16.67 33.33 50.00 16.67 50.00 16.67 50.00 40.43 8.33 15.00 8.33 15.00
0.125 0.060 0.060 0.060 0.033 0.060 0.066 0.071 0.060 0.062 0.077 0.093 0.098 0.116 0.136 0.119 0.139 0.162 0.143 0.195 0.171 0.241 0.300 0.039 0.044 0.031 0.036
72 72 24 72 56.6 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72
50.0 96.4 38.7 93.1 93.2 89.9 95.4 91.6 85.7 76.5 77.9 81.5 55.7 58.6 60.7 42.5 48.5 51.2 38.2 40.7 25.1 32.7 21.5 84.6 89.1 87.2 91.8
52.8 99.6 38.7 93.5 94.3 95 95 90 90 80 80 80 60 60 60 50 50 50 40 40 30 30 20 90 90 95 95
Average absolute relative deviation, % ⫽ 4.73.
133
ment. The product has also been characterized by recording the 1H NMR spectra of the compound on a Bru¨ker-DRX 500 NMR instrument operating at 20°C. 2.5. Design of experiments The experimental design chosen for the study was a Box–Behnken design [10] that helps in investigating linear, quadratic, and cross-product effects of three factors, each varied at three levels and also includes three center points for replication. The crucial factors that are involved in the study are E/S ratio, alcohol (acid) concentration, and incubation period. The factors and the levels at which the experiments were carried out are given in Table 1. The design of experiments employed is presented in Table 2. The response function to study the effects of the variables on the percentage esterification is a second order polynomial of the form: Y ⫽ A0 ⫹ A1 䡠 Xe ⫹ A2 䡠 Xs ⫹ A3 䡠 Xt ⫹ A4 䡠 X2e ⫹ A5 䡠 Xs2 ⫹ A6 䡠 X2t ⫹ A7 䡠 Xe 䡠 Xs ⫹ A8 䡠 Xs 䡠 Xt ⫹ A9 䡠 Xt 䡠 Xe
(1)
where Y ⫽ Esterification, %; Xe ⫽ E/S ratio; Xs ⫽ Alcohol (acid) concentration; Xt ⫽ Incubation period; A0 ⫽ Constant; A1,A2,A3 ⫽ Linear coefficients; A4,A5,A6 ⫽ Quadratic coefficients; A7,A8,A9 ⫽ Cross-product coefficients. The coefficients of the response function, their statistical significance and process conditions for maximum percentage esterification were evaluated by the method of least squares by using the Microsoft Excel—97 software (version
Table 4. Coefficients of the response function to predict esterification percentage and analysis of variance Coefficient A0 A1 A2 A3 A5 A8
Values
Standard error R2 Average absolute relative deviation, %
27.314 8.600 ⫺25.338 8.663 17.023 ⫺11.325 6.823 0.95 13.17
Variance analysis: Source
Degrees of freedom
Sum of square
Mean of square
F-test
Significance F
Regression Residual Lack of fit Pure error Total
5 9 7 2 14
7922.81 418.96 410.93 8.03 8341.77
1584.56 46.55 58.71 4.02
34.04*
1.39E-05
14.62
N.S.**
* P ⬍ 0.01. ** P ⬍ 0.05.
134
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
Fig. 1. Response surface predicting esterification at E/S ratio ⫽ 50 g/mol.
5.0; Microsoft Corp., Redmond, WA, USA). Canonical form of the response function was developed following the method outlined by Montgomery [11]. Contour plot representing identical esterification lines under different conditions of E/S ratios and alcohol (acid) concentrations and optimum search were constructed by using the SOLVER program in the Microsoft Excel—97 software (version 5.0). SOLVER program uses the generalized reduced gradient (GRG2) nonlinear optimization code. The conditions represented by contour plot for different extents of esterification were tested for validation by conducting the experiments under these conditions and the results are shown in Table 3.
3. Results and discussion Coefficients of the full model Eq. (1) were evaluated by regression analysis and tested for their significance. The insignificant coefficients were eliminated on the basis of p values after examining the coefficients, and the model was finally refined. It was observed that all the linear coefficients, one quadratic term (Xs2) and one cross-product term (Xs 䡠 Xt) were only significant, the p values being very small (p ⬍ 0.01). The final response function to predict the
percentage esterification after eliminating the insignificant terms was as follows: Y ⫽ 27.314 ⫹ 8.6 Xe ⫺ 25.338 Xs ⫹ 8.663 Xt ⫹ 17.023 XS2 ⫺ 11.325 XS 䡠 Xt
(2)
The coefficients and the analysis of variance (ANOVA) are presented in Table 4, which indicate that the predictability of the model is at 95% confidence level. A coefficient of determination (R2) value of 0.949 showed that the equation is highly reliable. Besides, the computed F value (34.04) is much greater than the tabular F0.01(5,9) value (6.06) at 1% level, indicating that the treatment is highly significant. The model also showed statistically insignificant lack of fit, as is evident from the lower computed F value (14.62) than the tabular F0.05(7,2) value (19.35) at 5% level. The model was found to be adequate for prediction within the range of variables employed. With respect to canonical analysis, the predicted esterification is 14.46% at the stationary point of (Xe,Xs,Xt) ⫽ (⫺0.936, 0.7683, 0.374) for which the canonical form is Y ⫽ 14.46 ⫹ 18.832 W21 ⫹ 2.947 W22 ⫺ 3.417 W23. Because the lamda (canonical form coefficient) values are not of the same sign, the stationary point is a saddle point.
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
135
Fig. 2. Predicted esterification as a function of alcohol (acid) concentration at E/S ratio ⫽ 16.67 g/mol.
Fig. 3. Predicted esterification as a function of incubation period at E/S ratio ⫽ 33.33 g/mol.
From the model, percentage esterification was predicted as a function of E/S ratio, alcohol (acid) concentration and incubation period in the range of variables studied. The average absolute deviation of the reduced model is 13.17%. The model was tested for validity and adequacy by carrying out additional independent experiments and the results (Table 3) show that the model predictions reasonably agree with experimental values (average absolute relative deviation ⫽ 4.73%). Fig. 1 shows the 3-D response surface plot at an E/S ratio of 50 g/mol. It shows the best yield at Xs ⫽ 0.06 and Xt ⫽ 72, which corresponded with the experimentally determined yield of 96.4% under these conditions. The surface shows a decrease in esterification with increase in alcohol (acid) concentrations. However, with increase in incubation period percentage esterification increased at all alcohol (acid) concentrations studied. Fig. 2 shows the percentage esterification as a function of alcohol (acid) concentration at different periods of incubation at Xe ⫽ 16.67. A minimum esterification of 6% was predicted at Xe ⫽ 16.67, Xs ⫽ 0.22, and Xt ⫽ 24. Percentage esterification decreased with alcohol (acid) concentration up to a Xs value of 0.22 to 0.26 and thereafter it increased up to Xs ⫽ 0.3. However, at 72 h percentage esterification decreased progressively with increase in alcohol (acid) concentration. Esterification as a function of incubation period at various alcohol (acid) concentrations at Xe ⫽ 33.33 is shown in Fig. 3. The percentage esterification observed after 72 h
steadily increases from 14% at Xs ⫽ 0.3 to 97% at Xs ⫽ 0.06, which indicates clearly that esterification predominates over hydrolysis (prevalent at Xs ⫽ 0.3). At Xe ⫽ 33.33 and Xs ⫽ 0.3, increase in incubation period resulted in a slight decrease in esterification, ranging from 21.5% after 24 h to 16.8% after 72 h. The decrease in esterification with increase in alcohol (acid) concentration with a minimum at Xs ⫽ 0.22 to 0.26 indicate probably competitive binding between alcohol and acid with the enzyme exhibiting a greater affinity for the alcohol than the acid resulting in reduced acyl transfer and hence reduction in the extent of esterification. For any given amount of enzyme, the esterification shows a marked deviation between Xs ⫽ 0.22 to 0.26, which may correspond to a saturation concentration of the alcohol. In an earlier report in the literature, by using a very high concentration of free lipase (50 g/liter, E/S ⫽ 200), at a substrate concentration of 0.25 M, 80% conversion in 24 h was observed [6]. Welsh et al. (1990) [8] have reported 75.8% conversion in 48 h with 2% w/v (20 g/liter) native lipase from Candida cylindracea with 0.05 M substrate. Razafindralambo et al. (1994) [9] have reported 80% conversion in 24 h at 4 : 1 alcohol : acid ratio with 50 g/liter of free lipase-esterase preparation from Mucor miehei. Apart from incomplete conversion, these studies have used high free enzyme concentrations that are disadvantageous from the process economics angle. But the present study shows that conversion over 95%
136
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
of E/S ratio and alcohol (acid) concentrations up to ⫺1.5 level. The experimental yields for these points also showed good correspondence with the predicted ones. Hence, the above mentioned studies have yielded good amount of information in achieving maximum yield by using minimum amount of the enzyme which is of importance commercially.
Acknowledgments We thank the Director, CFTRI, Mysore, India for encouragement and facilities. S.H.K. thanks the CSIR, New Delhi, India, for the award of a Senior Research Fellowship.
References
Fig. 4. Contour plot showing the ranges of E/S ratios and alcohol (acid) concentrations to obtain various extent of esterification. Other variable Xt (incubation period) is constant at 72 h (⫹1 level).
could be achieved at even very low enzyme concentrations as can be inferred from the contour plot (Fig. 4). The contour plot shows lines corresponding to identical extent of esterification under different conditions of E/S ratio and alcohol (acid) concentrations. Validation experiments were conducted by considering several points from the contour plot. The experimental values of the ester yield obtained from these points corresponded well with the predicted ones. Contour plot also shows the conditions outside the range employed for developing the model namely for values
[1] Welsh FW, Murray WD, Williams RE. Microbiological and enzymatic production of flavor and fragrance chemicals. CRC Crit Rev Biotechnol 1989;9:105– 69. [2] Akoh CC, Copper C, Nwosu CV. Lipase G-catalyzed synthesis of monoglycerides in organic solvent and analysis by HPLC. J Am Oil Chem Soc 1992;69:257– 60. [3] Welsh FW, Williams RE. Lipase-mediated production of ethyl butyrate and butyl butyrate in nonaqueous systems. Enzyme Microb Technol 1990;12:743– 8. [4] Claon PA, Akoh CC. Effect of reaction parameters on SP 435 lipase catalyzed synthesis of citronellyl acetate in organic solvent. Enzyme Microb Technol 1994;16:835– 8. [5] Gillies B, Yamazaki H, Armstrong DW. Production of flavor esters by immobilized lipase. Biotechnol Lett 1987;9:709 –14. [6] Langrand G, Triantaphylides C, Baratti J. Lipase catalyzed formation of flavor esters. Biotechnol Lett 1988;10:549 –54. [7] Langrand G, Rondot N, Triantaphylides C, Baratti J. Short-chain flavor esters synthesis by microbial lipases. Biotechnol Lett 1990;12: 581– 6. [8] Welsh FW, Williams RE, Dawson KH. Lipase-mediated synthesis of low molecular weight flavor esters. J Food Sci 1990;55:1679 – 82. [9] Razafindralambo H, Blecker C, Lognoy G, Marlier M, Wathlet JP, Severin M. Improvement of enzymatic synthesis yields of flavor acetates: the example of isoamyl acetate. Biotechnol Lett 1994;16: 247–50. [10] Box GEP, Behnken DW. Some new three level designs for the study of quantitative variables. Technometrics 1960;2:455–76. [11] Montgomery DC. Design and Analysis of Experiments, 3rd edition. New York: John Wiley & Sons, Inc., 1991.
Optimization of isoamyl acetate production by using immobilized lipase from Mucor miehei by response surface methodology S. Hari Krishnaa, B. Manoharb, S. Divakara, S. G. Prapullaa, N. G. Karantha,* a
Fermentation Technology & Bio–Engineering Department, Central Food Technological Research Institute, Mysore 570 013, India b Food Engineering Department, Central Food Technological Research Institute, Mysore 570 013, India Received 7 December 1998; received in revised form 19 July 1999; accepted 20 July 1999
Abstract Immobilized lipase from Mucor miehei was employed for the esterification of isoamyl alcohol with acetic acid in n-heptane solvent. The important process variables studied were enzyme/substrate (E/S) ratio, alcohol (acid) concentration, and incubation period. Based on Box–Behnken design of experiments, a second order response function was developed. The percentage esterification increased with both E/S ratio and time and decreased with alcohol (acid) concentration. The model indicated optimum conditions for maximum esterification ranging from 20 to 99.6% in the alcohol (acid) concentration range of 0.031 to 0.3 M for a range of E/S ratios 8.33 to 50 g/mol, which were in good agreement with the experimental yields. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Isoamyl acetate; Response surface methodology; Esterification; Lipase; Lipozyme IM; Enzymatic synthesis; Mucor miehei
1. Introduction Isoamyl esters (in particular isoamyl acetate) are flavor esters of commercial importance. Traditionally, they have been produced by chemical synthesis or extracted from natural sources. Isoamyl acetate has a strong banana flavor and hence is desired very much in food industry [1]. Annual demand for isoamyl acetate in USA alone amounts to about 74 tonnes [1]. With increasing orientation toward ‘natural’ flavors, production of flavor esters by ‘natural’ means like employment of enzymes or through microbial sources is gaining importance industrially. Lipases (triacylglycerol hydrolases, EC 3.1.1.3) usually catalyze hydrolytic reactions. However, when employed in low water environment, they can perform the reverse reaction, namely, esterification also. Lipasecatalyzed biotransformations are gaining importance because of their regio-, stereo-, and substrate specificity’s, their milder reaction conditions and the relatively lower energy requirement. Enzymatic reactions can be efficiently accomplished by employing lipases in an adequate organic solvent, such as heptane and hexane, thereby shifting the * Corresponding author. Tel.: ⫹91-821-515-792; fax: ⫹91-821-517233. E-mail address: [email protected] (N.G. Karanth)
reaction equilibrium toward esterification rather than hydrolysis. Lipases have been employed for direct esterification and transesterification reactions in organic solvents to produce esters of glycerol [2], aliphatic alcohol’s [3], and terpene alcohol’s [4]. However the esterification of shortchain fatty acids and alcohol’s has not received much attention. Moreover, short-chain (⬍C5) acids and alcohol’s are found to exert inhibitory effects on the enzyme [5,6]. Data generation for esterification with such substrates is important because short-chain esters are vital components of many fruit flavors. Few attempts have been made to establish the feasibility of synthesizing isoamyl acetate by employing lipases from different sources [5–9]. However, use of high enzyme concentration and low yields have been the significant drawbacks. Also the lipase catalyzed esterification reaction has not been examined thoroughly. Reports on the use of high enzyme concentration leading to 80% yields are available [6,8,9]. Considering the high demand and benefits, an optimized process for high yield enzymatic synthesis of isoamyl esters, utilizing low enzyme contents, is important. The present work aims at a better understanding of the relationships between the important esterification variables (substrate and enzyme concentrations, and incubation period) and ester yield to determine optimum conditions for the synthesis of isoamyl acetate catalyzed by Lipozyme
0141-0229/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 1 - 0 2 2 9 ( 9 9 ) 0 0 1 4 9 - 0
132
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
Table 1 Levels of variables chosen for the study Variable
Levels
Xe Enzyme/substrate ratio, g/mol Xs Alcohol (acid) concentration, M Xt Incubation period, h
⫹1
0
⫺1
50.00 0.30 72
33.33 0.18 48
16.67 0.06 24
IM-20 (immobilized lipase from Mucor miehei). Response surface methodology (RSM), which is an efficient statistical technique for optimization of multiple variables to predict best performance conditions with minimum number of experiments, was employed, the results of which are discussed in the present work.
Equimolar concentrations of isoamyl alcohol and acetic acid were taken in n-heptane, followed by different amounts of the immobilized enzyme. The reaction mixture was agitated at 150 rev./min on an orbital shaker at 39 ⫾ 1°C. Enzyme/substrate (E/S) ratio, alcohol (acid) concentration and incubation period were at three levels namely ⫺1, 0, and 1 corresponding to minimum, middle and maximum (Table 1).
2.3. E/S ratio To rationalize the enzyme concentration with respect to substrate (acid), a term called E/S ratio was used: E/S ratio (g/mol) ⫽ enzyme (g/liter)/substrate (mol/liter). 2.4. Analysis
2. Materials and methods 2.1. Materials Lipozyme IM-20 (Mucor miehei lipase immobilized on Duolite weak anion-exchange resin) was a gift from M/s Novo Nordisk, Denmark, which exhibited a specific hydrolytic activity of 20 mol/min/mg of protein at pH 7.0 and 30°C, with tributyrin as the substrate. All the chemicals employed were of analytical reagent grade. Isoamyl alcohol was from M/s Sarabhai M. Chemicals Ltd. (Baroda, India) and heptane and acetic acid were from M/s Sisco Research Laboratories (Bombay, India). Solvents were distilled once before use. 2.2. Esterification method Isoamyl acetate synthesis was carried out in 100 ml stoppered conical flasks with 10 ml working volume.
The extent of esterification was estimated at regular intervals of time by titrating aliquots (0.5 ml) of the reaction mixture against 0.01 N NaOH by using phenolphthalein as the indicator. The percentage esterification was determined from the difference in the volume of alkali consumed between aliquots of the reaction mixture withdrawn at the beginning of the reaction and after certain period of time. The analysis of the reaction mixture was also carried out by subjecting the samples to gas chromatographic analysis by using a Shimadzu gas chromatograph (GC 15-A) equipped with a Carbowax 20 M column (3 m length, 3.175 mm internal diameter [i.d.]) and a flame ionization detector. Nitrogen was used as a carrier gas with a flow rate of 30 ml/min. Column oven, injection port, and detector temperatures were at 100°C, 200°C, and 250°C respectively. The percentage of esterification determined by both GC analysis (which showed product formation) and titration (which showed acid consumption) were found to be in good agree-
Table 2 Actual values of variables as per Box-Behnken design of experiments and percentage esterification obtained SI. No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Enzyme/ substrate ratio g/mol
Alcohol (acid) concentration M
Incubation period h
Esterification, % Experimental
Predicted
50.00 50.00 16.67 16.67 50.00 50.00 16.67 16.67 33.33 33.33 33.33 33.33 33.33 33.33 33.33
0.30 0.06 0.30 0.06 0.18 0.18 0.18 0.18 0.30 0.30 0.06 0.06 0.18 0.18 0.18
48 48 48 48 72 24 72 24 72 24 72 24 48 48 48
26.8 89.3 10.9 59.4 35.9 31.3 27.1 17.1 21.5 16.8 90.0 40.0 25.1 26.1 28.6
27.6 78.3 10.4 61.1 44.6 27.3 27.4 10.1 16.3 21.7 89.7 49.7 27.3 27.3 27.3
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136 Table 3 Model validation experiments Enzyme/ substrate ratio g/mol
Alcohol (acid) concentration M
Incubation period h
Esterification, % Experimental
Predicted
28.00 50.00 25.00 41.67 37.23 43.68 50.00 46.67 33.99 16.67 33.33 50.00 16.67 33.33 50.00 16.67 33.33 50.00 16.67 50.00 16.67 50.00 40.43 8.33 15.00 8.33 15.00
0.125 0.060 0.060 0.060 0.033 0.060 0.066 0.071 0.060 0.062 0.077 0.093 0.098 0.116 0.136 0.119 0.139 0.162 0.143 0.195 0.171 0.241 0.300 0.039 0.044 0.031 0.036
72 72 24 72 56.6 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72 72
50.0 96.4 38.7 93.1 93.2 89.9 95.4 91.6 85.7 76.5 77.9 81.5 55.7 58.6 60.7 42.5 48.5 51.2 38.2 40.7 25.1 32.7 21.5 84.6 89.1 87.2 91.8
52.8 99.6 38.7 93.5 94.3 95 95 90 90 80 80 80 60 60 60 50 50 50 40 40 30 30 20 90 90 95 95
Average absolute relative deviation, % ⫽ 4.73.
133
ment. The product has also been characterized by recording the 1H NMR spectra of the compound on a Bru¨ker-DRX 500 NMR instrument operating at 20°C. 2.5. Design of experiments The experimental design chosen for the study was a Box–Behnken design [10] that helps in investigating linear, quadratic, and cross-product effects of three factors, each varied at three levels and also includes three center points for replication. The crucial factors that are involved in the study are E/S ratio, alcohol (acid) concentration, and incubation period. The factors and the levels at which the experiments were carried out are given in Table 1. The design of experiments employed is presented in Table 2. The response function to study the effects of the variables on the percentage esterification is a second order polynomial of the form: Y ⫽ A0 ⫹ A1 䡠 Xe ⫹ A2 䡠 Xs ⫹ A3 䡠 Xt ⫹ A4 䡠 X2e ⫹ A5 䡠 Xs2 ⫹ A6 䡠 X2t ⫹ A7 䡠 Xe 䡠 Xs ⫹ A8 䡠 Xs 䡠 Xt ⫹ A9 䡠 Xt 䡠 Xe
(1)
where Y ⫽ Esterification, %; Xe ⫽ E/S ratio; Xs ⫽ Alcohol (acid) concentration; Xt ⫽ Incubation period; A0 ⫽ Constant; A1,A2,A3 ⫽ Linear coefficients; A4,A5,A6 ⫽ Quadratic coefficients; A7,A8,A9 ⫽ Cross-product coefficients. The coefficients of the response function, their statistical significance and process conditions for maximum percentage esterification were evaluated by the method of least squares by using the Microsoft Excel—97 software (version
Table 4. Coefficients of the response function to predict esterification percentage and analysis of variance Coefficient A0 A1 A2 A3 A5 A8
Values
Standard error R2 Average absolute relative deviation, %
27.314 8.600 ⫺25.338 8.663 17.023 ⫺11.325 6.823 0.95 13.17
Variance analysis: Source
Degrees of freedom
Sum of square
Mean of square
F-test
Significance F
Regression Residual Lack of fit Pure error Total
5 9 7 2 14
7922.81 418.96 410.93 8.03 8341.77
1584.56 46.55 58.71 4.02
34.04*
1.39E-05
14.62
N.S.**
* P ⬍ 0.01. ** P ⬍ 0.05.
134
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
Fig. 1. Response surface predicting esterification at E/S ratio ⫽ 50 g/mol.
5.0; Microsoft Corp., Redmond, WA, USA). Canonical form of the response function was developed following the method outlined by Montgomery [11]. Contour plot representing identical esterification lines under different conditions of E/S ratios and alcohol (acid) concentrations and optimum search were constructed by using the SOLVER program in the Microsoft Excel—97 software (version 5.0). SOLVER program uses the generalized reduced gradient (GRG2) nonlinear optimization code. The conditions represented by contour plot for different extents of esterification were tested for validation by conducting the experiments under these conditions and the results are shown in Table 3.
3. Results and discussion Coefficients of the full model Eq. (1) were evaluated by regression analysis and tested for their significance. The insignificant coefficients were eliminated on the basis of p values after examining the coefficients, and the model was finally refined. It was observed that all the linear coefficients, one quadratic term (Xs2) and one cross-product term (Xs 䡠 Xt) were only significant, the p values being very small (p ⬍ 0.01). The final response function to predict the
percentage esterification after eliminating the insignificant terms was as follows: Y ⫽ 27.314 ⫹ 8.6 Xe ⫺ 25.338 Xs ⫹ 8.663 Xt ⫹ 17.023 XS2 ⫺ 11.325 XS 䡠 Xt
(2)
The coefficients and the analysis of variance (ANOVA) are presented in Table 4, which indicate that the predictability of the model is at 95% confidence level. A coefficient of determination (R2) value of 0.949 showed that the equation is highly reliable. Besides, the computed F value (34.04) is much greater than the tabular F0.01(5,9) value (6.06) at 1% level, indicating that the treatment is highly significant. The model also showed statistically insignificant lack of fit, as is evident from the lower computed F value (14.62) than the tabular F0.05(7,2) value (19.35) at 5% level. The model was found to be adequate for prediction within the range of variables employed. With respect to canonical analysis, the predicted esterification is 14.46% at the stationary point of (Xe,Xs,Xt) ⫽ (⫺0.936, 0.7683, 0.374) for which the canonical form is Y ⫽ 14.46 ⫹ 18.832 W21 ⫹ 2.947 W22 ⫺ 3.417 W23. Because the lamda (canonical form coefficient) values are not of the same sign, the stationary point is a saddle point.
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
135
Fig. 2. Predicted esterification as a function of alcohol (acid) concentration at E/S ratio ⫽ 16.67 g/mol.
Fig. 3. Predicted esterification as a function of incubation period at E/S ratio ⫽ 33.33 g/mol.
From the model, percentage esterification was predicted as a function of E/S ratio, alcohol (acid) concentration and incubation period in the range of variables studied. The average absolute deviation of the reduced model is 13.17%. The model was tested for validity and adequacy by carrying out additional independent experiments and the results (Table 3) show that the model predictions reasonably agree with experimental values (average absolute relative deviation ⫽ 4.73%). Fig. 1 shows the 3-D response surface plot at an E/S ratio of 50 g/mol. It shows the best yield at Xs ⫽ 0.06 and Xt ⫽ 72, which corresponded with the experimentally determined yield of 96.4% under these conditions. The surface shows a decrease in esterification with increase in alcohol (acid) concentrations. However, with increase in incubation period percentage esterification increased at all alcohol (acid) concentrations studied. Fig. 2 shows the percentage esterification as a function of alcohol (acid) concentration at different periods of incubation at Xe ⫽ 16.67. A minimum esterification of 6% was predicted at Xe ⫽ 16.67, Xs ⫽ 0.22, and Xt ⫽ 24. Percentage esterification decreased with alcohol (acid) concentration up to a Xs value of 0.22 to 0.26 and thereafter it increased up to Xs ⫽ 0.3. However, at 72 h percentage esterification decreased progressively with increase in alcohol (acid) concentration. Esterification as a function of incubation period at various alcohol (acid) concentrations at Xe ⫽ 33.33 is shown in Fig. 3. The percentage esterification observed after 72 h
steadily increases from 14% at Xs ⫽ 0.3 to 97% at Xs ⫽ 0.06, which indicates clearly that esterification predominates over hydrolysis (prevalent at Xs ⫽ 0.3). At Xe ⫽ 33.33 and Xs ⫽ 0.3, increase in incubation period resulted in a slight decrease in esterification, ranging from 21.5% after 24 h to 16.8% after 72 h. The decrease in esterification with increase in alcohol (acid) concentration with a minimum at Xs ⫽ 0.22 to 0.26 indicate probably competitive binding between alcohol and acid with the enzyme exhibiting a greater affinity for the alcohol than the acid resulting in reduced acyl transfer and hence reduction in the extent of esterification. For any given amount of enzyme, the esterification shows a marked deviation between Xs ⫽ 0.22 to 0.26, which may correspond to a saturation concentration of the alcohol. In an earlier report in the literature, by using a very high concentration of free lipase (50 g/liter, E/S ⫽ 200), at a substrate concentration of 0.25 M, 80% conversion in 24 h was observed [6]. Welsh et al. (1990) [8] have reported 75.8% conversion in 48 h with 2% w/v (20 g/liter) native lipase from Candida cylindracea with 0.05 M substrate. Razafindralambo et al. (1994) [9] have reported 80% conversion in 24 h at 4 : 1 alcohol : acid ratio with 50 g/liter of free lipase-esterase preparation from Mucor miehei. Apart from incomplete conversion, these studies have used high free enzyme concentrations that are disadvantageous from the process economics angle. But the present study shows that conversion over 95%
136
S. Hari Krishna et al. / Enzyme and Microbial Technology 26 (2000) 131–136
of E/S ratio and alcohol (acid) concentrations up to ⫺1.5 level. The experimental yields for these points also showed good correspondence with the predicted ones. Hence, the above mentioned studies have yielded good amount of information in achieving maximum yield by using minimum amount of the enzyme which is of importance commercially.
Acknowledgments We thank the Director, CFTRI, Mysore, India for encouragement and facilities. S.H.K. thanks the CSIR, New Delhi, India, for the award of a Senior Research Fellowship.
References
Fig. 4. Contour plot showing the ranges of E/S ratios and alcohol (acid) concentrations to obtain various extent of esterification. Other variable Xt (incubation period) is constant at 72 h (⫹1 level).
could be achieved at even very low enzyme concentrations as can be inferred from the contour plot (Fig. 4). The contour plot shows lines corresponding to identical extent of esterification under different conditions of E/S ratio and alcohol (acid) concentrations. Validation experiments were conducted by considering several points from the contour plot. The experimental values of the ester yield obtained from these points corresponded well with the predicted ones. Contour plot also shows the conditions outside the range employed for developing the model namely for values
[1] Welsh FW, Murray WD, Williams RE. Microbiological and enzymatic production of flavor and fragrance chemicals. CRC Crit Rev Biotechnol 1989;9:105– 69. [2] Akoh CC, Copper C, Nwosu CV. Lipase G-catalyzed synthesis of monoglycerides in organic solvent and analysis by HPLC. J Am Oil Chem Soc 1992;69:257– 60. [3] Welsh FW, Williams RE. Lipase-mediated production of ethyl butyrate and butyl butyrate in nonaqueous systems. Enzyme Microb Technol 1990;12:743– 8. [4] Claon PA, Akoh CC. Effect of reaction parameters on SP 435 lipase catalyzed synthesis of citronellyl acetate in organic solvent. Enzyme Microb Technol 1994;16:835– 8. [5] Gillies B, Yamazaki H, Armstrong DW. Production of flavor esters by immobilized lipase. Biotechnol Lett 1987;9:709 –14. [6] Langrand G, Triantaphylides C, Baratti J. Lipase catalyzed formation of flavor esters. Biotechnol Lett 1988;10:549 –54. [7] Langrand G, Rondot N, Triantaphylides C, Baratti J. Short-chain flavor esters synthesis by microbial lipases. Biotechnol Lett 1990;12: 581– 6. [8] Welsh FW, Williams RE, Dawson KH. Lipase-mediated synthesis of low molecular weight flavor esters. J Food Sci 1990;55:1679 – 82. [9] Razafindralambo H, Blecker C, Lognoy G, Marlier M, Wathlet JP, Severin M. Improvement of enzymatic synthesis yields of flavor acetates: the example of isoamyl acetate. Biotechnol Lett 1994;16: 247–50. [10] Box GEP, Behnken DW. Some new three level designs for the study of quantitative variables. Technometrics 1960;2:455–76. [11] Montgomery DC. Design and Analysis of Experiments, 3rd edition. New York: John Wiley & Sons, Inc., 1991.