versão On-line ISSN 0104-6632
Braz. J. Chem. Eng. v.17 n.4-7 São Paulo dez. 2000
INFLUENCE OF SUBSTRATE PARTITION COEFFICIENT ON THE PERFORMANCE OF LIPASE CATALYZED SYNTHESIS OF CITRONELLYL ACETATE BY ALCOHOLYSIS
H.F.Castro*, P.C.Oliveira and E.B.Pereira
Department of Chemical Engineering,
School of Chemical Engineering of Lorena
P.O.Box 116, 12600-000, Phone + (55) (12) 552-6473, Lorena - SP, Brazil
(Received: August 27, 1999 ; Accepted: April 18, 2000)
Abstract - The enzymatic synthesis of selected low molecular weight esters such as acetate esters by direct esterification using acetic acid as acyl donor usually display low yields. The acetic acid changes the polarity of the reaction medium, which in this turn modifies the partitioning of water between the solid phase (enzyme preparation) and the liquid phase (substrate), resulting in its accumulation on the enzyme solid phase. This may reduce the enzyme´s local pH. Therefore, the enzyme active site is modified and the reaction became nearly impossible. Our previous work showed that there is a negative relationship between enzyme activity and substrate partition coefficient (Ps); that is, the higher the substrate partition coefficient the lower the amount of product formed. This work investigated the feasibility of minimizing this inhibition by replacing the esterification reaction for alcoholysis reaction using several acetate esters. This approach enhanced the reaction yields to 46%, which is about 3 times higher than that one obtained in the esterification route.
Keywords: lipase, alcoholysis, citronellyl acetate, partition coefficient.
Esters are important components of natural aromas, contributing to the flavor in most fruits and many other foods (Gatfield, 1995). With the increasing demand for natural products, the food industry is interested in the use of biotechnological route to produce ester flavors (Vulfson, 1993). In this context, it is not surprising that a great deal of research has been focused on the application of enzymes to catalyze various reactions (Welsh et al., 1989; de Castro and Anderson, 1995; Yahya et al., 1998; Oliveira and Alves, 1999).
Among these, bioprocessing using lipases is a well-established useful method for the preparation of esters either by esterification or interesterification reactions. The lipase-catalyzed synthesis of more than 50 flavoring esters have been described to date (Vulfson, 1993) and, in principle, the reaction can be carried out in a mixture of alcohol and carboxylic acid with or without solvents, resulting in high productivities and almost quantitative yields (Yahya et al., 1998). An exception is found when low molecular weight acid (water soluble, C< 4) is used as acyl donor. In such systems, biocatalyst deactivation may occur, presumably by either the action of a toxic organic substance or by the potential deleterious effects of liquid-liquid interfaces on the structure of the biocatalyst (Scott et al., 1996). The actual mechanism of the enzyme inhibition caused by some organic media is not well understood, although it has been suggested that the reduction of the catalytic activity could be associated with conformational changes on the enzyme structure (Welsh et al., 1989; Scott et al., 1996). This phenomenon appears to be related to the protein-solvent interaction and/ or to the removal of water and dehydration of enzyme protein.
In agreement with these remarks, the literature has indicated that the enzymatic synthesis of selected low molecular weight esters, such as acetate esters, by direct esterification using acetic acid as acyl donor usually display very low yields (de Castro et al., 1997a). According to several researchers (Welsh et al., 1989; Chulalaksananukul et al., 1992; Scott et al., 1996) the acetic acid changes the polarity of the reaction medium, which in this turn modifies the partitioning of water between the solid phase (enzyme preparation) and the liquid phase (substrate), resulting in its accumulation on the enzyme solid phase. This may reduce the enzymes local pH. Therefore, the enzyme active site is modified and the reaction became nearly impossible. This is also supported by more recent investigations of enzyme activity dependence on substrate concentration in the aqueous layer around the catalyst where enzymatic reaction occurs (Yang and Roob, 1994). According to these authors, a correlation between enzyme activity and substrate partition coefficient could be determined. Our previous results showed that there is a negative relationship between enzyme activity and substrate partition coefficient (Ps); that is, the higher the substrate partition coefficient the lower the amount of product formed (de Castro et al., 1997b). For this particular case, based on the substrate partition coefficient (Ps) value of 9 was estimated, a value favoring the migration of acetic acid to the enzyme preparation.
When direct esterification cannot serve as an efficient route for ester synthesis, the transesterification reaction may be a better possibility. The reaction mixture is generally composed of the enzyme in a low water content organic solvent, containing the ester (as acyl donor, instead of carboxylic acid in the case of direct esterification) and the substituting alcohol as the nucleophile. This synthesis route was found to be the best strategy for the production of citronellyl acetate by using Novo's Lipozyme (de Castro et al., 1997b). Here we reported additional work on this matter dealing with the study of factors affecting the production of citronellyl acetate by alcoholysis reaction. The chain length of the ester and water content were investigated in order to achieve optimum lipase performance.
MATERIAL AND METHODS
Immobilized lipase (E.C. 188.8.131.52, Lipozyme 24 BIU/g) from Mucor miehie, supported on macroporous anionic resin beads, was kindly provided by Novo Nordisk (Denmark) RS-citronellol (97%) was purchased from Sigma, USA. Acetate esters were prepared in our laboratory following methodology described by Costa et al. (1995). All substrates were dehydrated before use, with 0.32 cm molecular sieves (aluminum sodium silicate, type 13, X-BHD Chemicals, Canada). Dry n-heptane, dried over molecular sieves, was used as the solvent for all experiments. Batch runs were carried out at 30°C in capped 100 ml glass vials, with reciprocating agitation (150 rpm). The substrate consisted of 100 mM of the acyl donor and 100 mM of citronellol, except when the molar ratio was an experimental variable. To control and monitor the water level in the reaction media, syntheses were carried out in the presence of molecular sieves, as previously described (de Castro et al., 1995). The results were evaluated by calculating the molar conversion based on the initial and final citronellol concentrations as expressed by equation 1:
0= Initial citronellol concentration and = Concentration of citronellol at a given time.
Estimation of Partition Coefficients
The partition coefficients (Lipozyme/external organic solvent) of citronellol and acetate esters were estimated according to the following equation (de Castro et al., 1997b):
Partition experiments were conducted under the same conditions as their reaction counterparts. In order to estimate the Lipozyme volume (V-Vo), a calibration curve volume of Lipozyme vs. mass of the Lipozyme was established [volume of matrix (cm3) = 0.41 x mass of Lipozyme (g) - 0.025]. The equilibrium concentrations were attained 2 hours after the immobilized lipase was added to the organic medium. The relation between the partition coefficients of the starting materials was designated as substrate partitioning coefficient (PS) as previously described (de Castro et al., 1997b).
RESULTS AND DISCUSSION
Many works clearly demonstrated that in the case of direct esterification, the chain length of acid and alcohol affects the yield of ester. In the transesterification reaction, Hirata et al. (1990) showed that the reactivity of aliphatic primary alcohols was dependent on the chain length and that this dependence was related to the nature of the solvent. The effect of the ester chain length in transesterification was further investigated by Chulalaksananukul et al. (1992) and propyl acetate was found to be the best for the production of geranyl acetate esters. However, only low reactant concentrations (about 30 mM) were tested, and it may be supposed that the mechanism they deduced did not take into consideration the substrate polarity. In this work, we investigated the reactivity of ester substrates of various chain length, ranging from butyl acetate (C6) to octyl acetate (C10) in the transesterification of citronellol. The reaction medium consisted of 100 mM of both reactants and 100 mg Lipozyme in 20 ml heptane. The molar conversion of citronellol was calculated after 24 hours reaction, and values attained for the ester chain lengths from 6 to 10 were reported in Table 1 together with their respective substrate partition coefficient (Ps).
The results showed that citronellyl acetate alcoholysis with medium chain length such as butyl and amyl acetate resulted in similar reaction rates. By increasing the chain length to C10 (octyl acetate) a decrease on the reaction rate was observed. In addition to the straight chain length, the effect of branching of the carbon chain was also studied. The branching was found to decrease the rate even more, probably due to stearic hindrance effects. Better results were achieved with butyl and amyl acetate.
In this reaction route, it is difficult to predict the reaction rate based solely on the substrate partitioning coefficient, since there does not appear to be a direct correlation with the reaction rate. As shown in Table 1, although substrate partition coefficients showed similar values, different molar conversions were attained. This situation is quite different from that one reported for esterification reaction where the value for substrate partition coefficient (citronellol and acetic acid) of about 9.5 has shown to have a strong influence on the formation of citronellyl acetate (de Castro et al., 1997b). Since the substrate polarity depends on many parameters, including the solvent, enzyme support, starting materials and reaction conditions, the results indicate that the substrate partitioning coefficient per se is not the only controlling factor, but rather a combined effect of the medium polarity with ester chain length.
Some insight into the reason for such variation on the reaction performance may be gained by considering the enzyme mechanism. In a typical reaction sequence, the lipase may react with citronellol to yield a dead-end enzyme-citronellol complex or with acetate esters to yield the lipase-acetate complex. Then the lipase-acetate complex transforms to an acyl-enzyme intermediate and the alcohol formed is released. This is followed by interaction of acyl enzyme with the citronellol to form another binary, which then yields the citronellol acetate and the free lipase. The results described in this work suggested that ester acetate chain length plays an important rule in the reaction progress.
Similar results from Langrand et al. (1990), showed that lipase from M. miehie was more active in synthesizing geranyl esters by esterification in the presence of acids with carbon atom numbers 4 to 6. Furthermore, other authors have demonstrated that for esterification of octanol this enzyme was most active in the presence of an acid with a carbon atom number of 7. This effect seems to be related to the fitting of both substrates into the active site of the enzyme: an optimum value of 15 for the sum of carbon chain length of the two substrates may be deduced from these results. From this set of results, and particularly from the substrate partition coefficient, butyl acetate may be considered the best substrate for the synthesis of geraniol in heptane by M. miehei lipase.
Further study of factors affecting the citronellol acetate synthesis from citronellol and butyl acetate was carried by employing a full 23 factorial design. The molar ratio (X1) of butyl acetate (BA) and citronellol (CITRO), the water adsorbent concentration (X2) and initial lipase concentration (X3) were chosen as independent process in the range to cover the intervals commonly used (de Castro et al., 1997b). The experimental matrix for the factorial design is shown in Table 2 together with the data for the response factor. The first three columns of data give the factors levels (+) or (-) in the dimensionless scale. All runs were performed at random. Three experiments were carried out at the centerpoint level, coded as "0", for experimental error estimation. The molar conversion of citronellol was used for the response. The results were analyzed by employing the Statgraphic software version 6.0.
The formation of citronellyl acetate was found to be dependent on the reaction conditions. No ester was formed in absence of water adsorbent and at low level of molar ratio, independent of the lipase concentration (runs 1 and 3). The addition of water adsorbent on the reaction media (runs 2 and 4) rose the conversion rate to 12 and 35.8%, respectively. Better conversion rate was attained at high levels for all variables (run 8). The experiment results showed in Table 2 were used to estimate the main variable effects and their interactions. According to the Student´s t-test, all variables are significant for the conversion of citronellol within the experimental range studied (Table 3). The most significant effect was the water adsorbent concentration, corroborating that excess of water (high medium polarity) may change the reaction equilibrium by decreasing the citronellol molar conversion. Although in the alcoholysis reactions there is no water formation throughout the process, the control of water content in the reactor vessel by using molecular sieve is expected to be as important as in the esterification reactions due to the polarity of butyl acetate. Since the presence of excess of water in the alcoholysis reaction system will lead to a hydrolytic side reactions, it is recommended to work at high level of water adsorbent (presence of molecular sieves). The molar ratio between reactants also shows a strong influence and butyl butyrate limitation may inhibit the enzymatic reaction. Though the optimal molar ratio for this system was an equimolar amount of both reactants. Among all factors, the initial Lipozyme concentration was the less influent, although better molar conversions were attained when high level of Lipozyme was used. Several interactions were also found to influence the molar conversion of citronellol. The interaction between the concentration of water adsorbent and molar ratio has the highest value among the others, therefore satisfactory citronellol conversion rate was attained at interaction of low level of molar ratio (1: 5) with high level of water adsorbent (20%).
The main effects were fitted by multiple regression analysis (Table 4) to a linear model as no significant value was given by checking the curvature (p> 0.05) and the best fitting response function can be written by Equation (3).
where Y= citronellol molar conversion (%); X1 and X2 coded values for molar ratio and water adsorbent concentration, respectively.
The statistical significance of this model was evaluated by the F-test (Table 5), which revealed that this regression is statistical significant at 99% probability level. The model did not show lack of fit and the determination coefficient (R2=0.96) indicates that 96% of the variability can be explained by the model.
According to this study, the maximum conversion molar of citronellol (45.8%) can be obtained, working at the highest level for all variables, that is, molar ratio of 1 (alcohol: ester), high concentration of enzyme (30% w/w) and presence of molecular sieve (20%). A detailed presentation of the optimum value predicted from the results using the response surface model is given in Figure1.
Alcoholysis runs were performed twice under the optimum conditions predicted by the model, and the average value was about 46% which confirms the prediction made by the same model.
Reaction efficiency is dictated by environmental parameters such as enzyme hydration, reaction temperature, substrate concentration, substrate size and chemical structure. The transesterification reaction was proved to be an efficient route to produce ester that cannot be achieved by direct esterification. The use of acetate esters as acyl donors minimize the inhibition of lipase in the synthesis of citronellyl acetate. The carbon chain length of the ester markedly affected the reaction rate of citronellyl acetate. Butyl acetate served as an excellent acyl donor. By adopting such procedure, it was possible to attain citronellol conversion rates of about 46% which is about 3 times higher than that one obtained in the esterification route.
We acknowledge Dr. I.C. Roberto for helping in the statistic analyses. This work was financially supported by CNPq and FAPESP.
Box, G.E.P., Hunter, W.G. and Hunter, J.S. Statistics for Experimenters: An Introduction to Design, Data Analysis and Model Building, Wiley & Sons Inc., New York (1978). [ Links ]
Chulalaksananukul, W.; Condoret, J.S.; Combes, D. Kinetics of Geranyl Acetate Synthesis by Lipase-catalysed Transesterification in n-Hexane. Enzyme and Microbial Technology, 14, 293- 298 (1992). [ Links ]
de Castro, H.F. and Anderson, W.A. Fine Chemicals by Biotransformation Using Lipases. Química Nova., 18, No.5, 544-554 (1995). [ Links ]
de Castro, H.F. and Jacques, S.S., 1995. Influência da Adição de Agentes Dessecantes no Desempenho da Reação de Butirato de Citronila utilizando Lipase Imobilizada. Arquivos de Biologia e Tecnologia, 38, N. 2, 339-344 (1995). [ Links ]
de Castro, H.F.; Soares, C.M.F. and Oliveira, P.C. Síntese de Ésteres Terpenóides por Via Enzimática: Influência do Tamanho da Cadeia Alifática do Ácido Graxo e da Estrutura do Álcool de Terpeno. Ciência e Tecnologia de Alimentos, 17, 224-228 (1997a). [ Links ]
de Castro, H.F.; Oliveira, P.C. and Pereira, E.B. Evaluation of Different Strategies for Lipase Catalysed Synthesis of Citronellyl Acetate. Biotechnology Letters, 9, 229-232 (1997b). [ Links ]
Costa, S.A.; Cortez, E.V. and de Castro, H.F. Comparação do Desempenho da Síntese de Butirato de Citronila por Via Química e Enzimática. Anais do 1º Congresso Brasileiro de Engenharia Química, p.81-83, São Carlos, SP (1995). [ Links ]
Gatfield, I.F. Enzymatic and Microbial Generation of Flavors. Perfume & Flavorist, 20, 5-14 (1995). [ Links ]
Hirata, H.; Higuchi, K. and Yamashina, T. Lipase-catalyzed Transesterification in Organic Solvent: Effcts of Water and Solvent, Thermal-stability and some Applications. Journal of Biotechnology, 14, 157-167 (1990) [ Links ]
Langrand, G.; Rondon, N.; Triantaphylides, C. and Baratti, J. Short Chain Flavor Esters Synthesis by Microbial Lipases. Biotechnology Letters, 12, 581-596 (1990). [ Links ]
Oliveira, D. and Alves, T.L.M. Enzymatic Alcoholysis of Palm and Palm Kernel Oils. Optimization by Statistical Methods. Applied Biochemistry and Biotechnology, 77-79, 835-844 (1999). [ Links ]
Scott, C.D.; Scott, T.C.; Blanch, H.W.; Klibanov, A.M. and Russell, A.J., in: Bioprocessing in Nonaqueous Media: Critical Needs and Opportunities. Oak Ridge National Laboratory, Report ORNL/TM -12849 (1996). [ Links ]
Vulfson, E.N Enzymatic Synthesis of Food Ingredients in Low Water Media Trends in Food Science & Technology, 4, 209-215 (1993). [ Links ]
Welsh, F.H.; Murray, W.D and Williams, R.E. Microbiological and Enzymatic Production of Flavour and Fragrance Chemicals. Critical Reviews in Biotechnology, 9, 105-169, (1989). [ Links ]
Yahya, A.R.M.; Anderson, W.A. and Moo-Young, M. Ester Synthesis in Lipase-catalyzed Reactions. Enzyme and Microbial Technology, 23, 438-450 (1998). [ Links ]
Yang, Z. and Robb, D.A. Partition Coefficients of Substrates and Products and Solvent Selection for Biocatalysis under Nearly Anhydrous Conditions. Biotechnology and Bioengineering, 43, 365-370 (1994). [ Links ]
*To whom correspondence should be addressed