- Email: [email protected]
Exploiting hysteresis for high activity enzymes in organic media
Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.
373
Exploiting hysteresis for high activity enzymes in organic media
Johann Partridge, Peter. J. Hailing and Barry D. Moore Department of Pure & Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street, Glasgow G1 I XL, Scotland, U.K.
Common methods for removing water from protein preparations such as freeze drying and air drying are shown here to be much more detrimental to enzyme activity than simple rapid dehydration. In a comparison of drying methods for cross-linked enzyme crystals of subtilisin Carlsberg, the highest rates of reaction in organic media were obtained with crystals dried by washing with anhydrous polar solvent. Crystals dried over molecular sieves in air or solvent showed significantly lower activity. In all cases, full activity could be recovered in aqueous buffer. When different solvents were used to dry the enzyme crystals, the catalytic rate was found to vary significantly- longer chain alcohols gave the highest rates while smaller "water-like" solvents such as methanol gave much lower rates. The study was extended to immobilised forms of subtilisin Carlsberg and ot-chymotrypsin. When these preparations were dried by solvent washing, high catalytic rates similar to those of crystals prepared by the same procedure were obtained. Catalytic efficiency in polar solvents was 1000 fold greater than that of freeze-dried powders. As with the crystals, subsequent drying over molecular sieves resulted in a substantial loss in enzymatic activity. The activities of solvent washed immobilised enzymes and of cross-linked crystals were found to vary significantly as a function of system water content.
1. Introduction In recent years it has been shown that enzymes exhibit novel properties in low water organic media, such as catalysis of reactions impossible in water, altered substrate specificity and increased thermostability [1]. This has led to an increased demand for enzyme preparations which exhibit high catalytic activity and reproducible selectivity [2-5]. The commonly used freeze-dried powders are known to exhibit low activity [6,7] but are often used because they can be obtained in one step from commercially supplied enzyme. Enzymes immobilised onto supports often show higher activity in organic media but the types of support material reported may not always be readily available and well controlled immobilisation procedures are often required [8-12]. Recently commercial cross-linked enzyme crystals (CLECs) have been introduced. As with enzymes immobilised on supports, these offer advantages over the conventional freeze dried preparations: access to individual enzyme molecules will be improved, whilst particle aggregation and diffusional limitation are considerably reduced. In addition, CLECs have been shown to exhibit high activity, good
374 stability and excellent reproducibility [13]. These characteristics make them potentially very attractive as catalysts for use in organic reaction mixtures. The method by which the biocatalyst is prepared for use in organic media often varies. With freeze dried powders, the enzyme which is essentially dry after lyophilisation, is added directly to the organic solvent which contains a known amount of water [14-16]. Similarly, enzymes immobilised from aqueous solution are collected by filtration, dried under vacuum and used directly in solvent with fixed water content. In other studies, the biocatalyst and solvent are dried exhaustively over water absorbents such as molecular sieves and then pre-equilibrated to well defined thermodynamic water activity (a,,), either together or separately prior to reaction [17,18]. This method offers advantages. Catalytic activity of a particular type of preparation is governed by the amount of water bound to the enzyme molecules, which is only a fraction of the total present. At equilibrium there will be a characteristic relationship between the amount of water bound and that dissolved in the organic phase. If the water level is expressed in terms of its aw, the relationship between catalytic activity and residual water level is generally similar for many different solvents [ 17,18]. There have been some literature reports that different catalytic activity can be found in systems with the same water content or a,,, but different hydration histories [19]. In this work, we set out to examine how different methods of enzyme pre-treatment effected their subsequent catalytic efficiency in organic solvent. Such studies on freeze dried enzymes are problematic. Random but significant differences in morphology between batches of powders is often found, probably due to variations in temperature and pressure during lyophilisation. Differences in catalytic rates can result, so that it can be difficult to make direct rate comparisons between laboratories and also to carry out meaningful studies of possible hysteresis effects arising from different treatments. We anticipated that using more homogeneous enzyme preparations these problems would be reduced. 2. Results and Discussion 2.1
The effect of different CLEC pre-treatments on catalytic efficiency Subtilisin Carlsberg CLECs (ChiroCLEC-BL) were supplied as a suspension in aqueous buffer. The manufacturer's literature reports that this suspension can be stored indefinitely at 4~ without any significant loss in catalytic activity. We were interested in studying the effect of different methods of preparing the crystals for use in organic media on their subsequent catalytic activity. In the present study, we routinely started with the fully active preparation in aqueous suspension and treated the samples by different methods for use in solvent as required. A number of routes can be envisaged for moving from an aqueous suspension to a "dry state". Whilst lyophilisation is necessary when drying dissolved enzymes for use in organic solvents, it is not the simplest and most obvious choice of water removal when enzymes are used as CLECs. Besides this, one of the main reasons for the low activity exhibited by freeze dried powders in organic media is known to be the deleterious effects of dehydration by this method. This process is therefore best avoided. Water can be removed from CLECs by more simple methods, including washing with anhydrous polar solvent and drying over molecular sieves in air or in solvent. As a starting point we explored the effect of washing CLEC with either propanol or acetonitrile. The solvent washed enzyme was then used directly in a reaction, or preequilibrated over molecular sieves in air or solvent (methods commonly applied to
375 conventional freeze dried and immobilised enzymes). We followed the rate of a standard transesterification in anhydrous acetonitrile using CLECs pre-treated in different ways: the initial reaction rates are shown in Table 1. The highest rates were obtained with CLEC placed directly in the reaction mixture atter solvent washing, whilst equilibration at fixed water activity invariably reduced the rate. Although CLEC pre-equilibrated in air gave by far the lowest rate, this preparation is still more active than the conventional freeze dried preparation of subtilisin (<0.01nmol/mg/min). Interestingly, CLEC pre-equilibrated in anhydrous acetonitrile over molecular sieves gave higher transesterification rates. Such an observation has been previously reported in the literature [21]. Freeze dried subtilisin powder in hexane was shown to give higher rates when equilibrated in solvent as opposed to in air. However, the difference between the two treatments is much greater for the CLECs in acetonitrile. Table 1. Effect of enzyme treatment on catalytic activity of subtilisin Carlsberg CLEC in anhydrous acetonitrile. Washin B Solvent" Further Enzyme Treatment ~ PrOH none PrOH 1 PrOH 2 ACN none ACN 1 ACN 2 PrOH b none PrOH, AcN none AcN, PrOH none AcN, aqueous buffe,r, PrOH .. none
Rate (nmol/m~min) 224 • 7 34 + 4 2.9 + 0.1 112 + 4 33 + 3 2.3 + 0.2 187 + 2 84 • 4 102 + 3 190 + 14
Data from Partridge et al [20] " CLEC was rinsed 3 times with 1 ml of each solvent shown respectively (with the exception of b which was rinsed 6 times). Atter each wash, the solvent was removed by centrifugation. c (1) Solvent washed CLEC was suspended in anhydrous AcN and then equilibrated over molecular sieves for 3 days, 20~ (2) Solvent washed CLEC was equilibrated over molecular sieves for 3 days, 20~ After this time it was suspended in anhydrous AcN. From Table 1 it is apparent that the difference in catalytic rate for CLEC dried over molecular sieves through the vapour phase is small, regardless of the solvent used to wash the crystals prior to equilibration. This is also the case for CLEC equilibrated in the reaction solvent prior to use. However, it is interesting to note that CLEC washed in propanol and used directly in the reaction gives a rate approximately double that of CLEC which was washed in acetonitrile. When the crystals were washed consecutively with propanol (x3) and acetonitrile (x3), regardless of the order in which the solvents were used, the rates of catalysis were less than half that for CLEC washed solely in propanol, but were close to that of acetonitrile washed CLEC. When the number of washes is increased from 3 to 6 for propanol washed CLEC, the decrease in rate is barely significant. It is apparent that the effect of acetonitrile is dominant on the crystals and the catalytic efficiency of an acetonitrile washed
376 CLEC can not be increased by subsequent washing with propanol. To ascertain whether or not acetonitrile washing had a long term detrimental effect on the crystals, we measured the catalytic activity of CLEC which was rinsed with acetonitrile, buffer and propanol respectively. The majority of catalytic activity was recovered, indicating that acetonitrile does not appear to damage the CLEC irreversibly. It might also be inferred that pre-equilibration through the vapour phase results in structural damage to the enzyme crystals, so that most of their catalytic activity is irreversibly lost. We attempted to eliminate this possibility experimentally. Biocatalyst which had been solvent washed and leit to equilibrate through the vapour phase for 3 days, was returned to an aqueous environment. After this, it was solvent washed to remove excess water, and used immediately to catalyse a reaction in anhydrous acetonitrile. CLEC was found to regain high catalytic activity. Furthermore, the rate observed is comparable to that of CLEC which has undergone solvent washing only. It is interesting to note that the catalytic efficiency of CLECs extensively dried over molecular sieves can not be improved significantly by further equilibration over a high aw saturated salt. CLEC must be returned to the aqueous buffer if high catalytic activity is to be recovered. The observations discussed above are not exclusive to reactions in acetonitrile. Experiments in propanol have confirmed that enzyme which has undergone solvent washing has a catalytic rate more than 12-fold higher than that which has been dried further using molecular sieves. Similarly, for reactions in propanol, enzyme pre-equilibrated through the vapour phase can regain high catalytic activity by exposure to an aqueous environment. The results discussed above clearly demonstrate that the catalytic behaviour of CLECs exhibits pronounced hysteresis: rates vary by 80-fold depending on previous hydration history. One might speculate that the enzyme or enzyme crystal has a 'memory' of how it has been treated, and this 'memory' can only be erased by returning the CLEC to an aqueous environment. Two possible explanations may account for this observation: (1) different dehydration protocols effect the amount of water left bound to the enzyme, or (2) the conformation of the dried enzyme is very sensitive to the method of water removal. Most probably these two effects are intimately related. Since washing the CLEC with solvent gives the most efficient rates of catalysis, we went on to investigate the effect of washing with other solvents. Table 2 shows how transesterification rate in anhydrous acetonitrile varies as much as 7-fold depending on the organic solvent used to dry the catalyst. The alcohols generally gave the fastest rates of catalysis, followed by acetonitrile and acetone which gave intermediate rates. Interestingly, the smaller "water-like" solvents, ethane diol and methanol gave the lowest rates of all solvents tested. With CLEC washed in methanol, only 33% of the maximum activity is regained after brief washing with buffer and propanol. Leaving in buffer overnight resulted in increased catalytic activity (63% of the original propanol washed CLEC). Nevertheless, the catalytic performance of all solvent washed CLECs was still higher than that which had been equilibrated in air or in solvent at aw < 0.01. A plausible explanation for this variation in catalytic rate when different solvents are used to dry the CLEC is that each solvent has a very specific capacity to displace the water within the crystal. Conformational changes may result on removal of more water from the crystal leading to a less active enzyme state and decreased rates of catalysis. On the basis of these data one can conclude that care must be taken when choosing a solvent to wash the crystals if maximum enzymatic activity is to be achieved.
377 Table 2. Effect of different solvent washes on the catalytic activity of subtilisin Carlsberg CLE C in anhydrous acetonitrile.
Washin~ Solvent' EtOH BuOH PrOH Me2CO ACN Ethane Diol CH3 OH MeOH, aqueous buffer ,. PrOH
Rate (nmol/mg/min) 282 + 32 243 • 23 224 • 7 121 + 11 112 + 4 50 + 12 39+ 3 ...... 74 + 1
Data from Partridge et al [20] ' CLEC was rinsed 3 times with 1 ml of each solvent shown respectively. After each wash, the solvent was removed by centrifugation. The enzyme was then suspended in anhydrous acetonitrile. 2.2
Extension of our findings to immobilised preparations With dissolved enzymes water must generally be removed by freeze-drying or similar prolonged dehydration methods. With CLECs, rapid drying by solvent rinsing is possible and this appears to lead to better specific activity. We therefore hypothesised that if non-crystalline but immobilised enzymes were treated using the same solvent washing procedure, very high activities might be obtained. We tested this hypothesis for an immobilised form of subtilisin Carlsberg. The enzyme was prepared using a simple adsorption procedure from aqueous buffer (pH 7.8) unto a standard silica gel support. Table 3 shows a comparison of the initial reaction rates obtained in acetonitrile using subtilisin Carlsberg prepared in different forms. The propanol rinsed immobilised enzyme preparation (termed PREP) gave fairly low rates in anhydrous acetonitrile. Table 3. Effect of preparation type on subtilisin Carlsber~; activity in acetonitrile. Rate (nmol / nag "/rain) 0% H20 v/v in ACN 1% H20 v/v in ACN Enzyme form and treatment (aw < 0.01) (a, = 0.22)
freeze dried powder, air dried b CLEC, PrOH washed CLEC, PrOH washed, air dried b immobilised form, PrOH washed immobilised form, PrOH washed, air drie db
< 0.01 226 2.94 0.82 ....
0.13 610 13.8 142 0.60
Data from Partridge et al [20,22] ' refers to weight of enzyme in preparation, b sample placed in sealed jar over molecular sieves for 3 days to give enzyme at aw<0.01; further 3 days equilibration over H20-saturated potassium acetate was carried out for enzyme at a, = 0.22.
378 However, on addition of 1% water v/v to the reaction solvent, the PREP was found to exhibit high activity. At this water level, rates for the PREP were found to be comparable to the CLEC, and over 1000 times greater than the commonly used freeze-dried powders. If PREP was subsequently dried in air prior to assaying in the organic solvent most of the activity was lost and the residual level approached that obtained for the lyophilised powder (see Table 3). However, aqueous suspensions of the silica adsorbed enzyme can be stored at 4~ for at least 3 weeks with negligible loss of activity and converted to the PREP as required. The effect of washing the immobilised enzyme with different anhydrous organic solvents was also tested. The same pattern of results as that observed for the CLEC emerged: ethanol washing gave high rates, acetonitrile gave intermediate rates and methanol produced a much less active preparation. Cross-linked enzyme crystals of chymotrypsin are not commercially available but under the same reaction conditions described in Table 3 the PREP of this enzyme exhibited high catalytic activity in acetonitrile with water levels of 1%v/v and above. Again these rates were two orders of magnitude better than the freeze-dried powder. It is perhaps surprising the method described here for preparing conventional biocatalysts for reactions in low water media has not been reported previously. However, until recently the large hydration hysteresis effects obtainable with enzymes had not been fully recognised. Our work with CLECs [20] and immobilised enzymes [22] has shown that different methods of water removal can dramatically affect the enzyme activity obtained. 2.3
The effect of system water content on catalytic activity Previous studies with biocatalysts in organic media have shown that the amount of water present in the system plays an important role in controlling factors such as rate, stability and hydrolytic equilibria [ 1, 23]. We therefore proceeded to carry out a more detailed study of the variation in catalytic activity as a function of water level in the actual reaction mixture for subtilisin CLEC and the immobilised form which had been propanol washed. Figure 1 shows the rate profiles for propanol washed CLEC and PREP in acetonitrile as a function of the thermodynamic water activity, aw. The profile for the freeze dried form of this enzyme is also shown for comparison. However, it should be noted that since the activity per unit weight of enzyme is much lower in the lyophilised powder, the rate profiles for the three preparations are shown as relative rates normalised to their maximum values. According to previous studies the amount of water bound to a protein in solvent would be expected to be controlled by a~ [23]. Under these conditions differences in the amount of residual bound water should be eliminated and hence similar rate vs aw profiles might be expected for all three forms of the enzyme. As can be seen in Figure 1 this is not the case. The activity of the lyophilised powder continues to increase even up to aw 0.76, while a maximal rate is obtained at aw of 0.11 with the CLEC and aw of 0.44 with the PREP. The different rate profiles could arise because the three preparations of the enzyme differ in either water binding or the water required for catalytic activity. A large change in water binding isotherms is unlikely, but kinetic factors may be significant. The water content of the lyophilised powder will be determined by its adsorption isotherm. In contrast, treatment of the CLEC and PREP imposes a water desorption process, for which the isotherm will be different due to hysteresis, and even the apparent equilibrium value may not be reached in the time employed. An alternative hypothesis is that CLEC and PREP require less water because the propanol dehydration process leaves a large proportion of the enzyme molecules in
379 a conformation close to the active fOrm. Only low levels of water may then be needed to promote catalysis and further increases in water availability provide no beneficial effect. With the freeze-dried powder the very low rates suggest most of the enzyme is initially inactive. In this case a water catalysed reorganisation process is probably required to convert the enzyme back to an active state since much greater water levels are needed to obtain high activity. The difference in profiles in Figure 1 would therefore reflect the fact that water plays different roles in promoting enzyme catalysis depending on the hydration history of the system.
1.0 .
0.8
~> 0.6 "~
0.4 0.2 0.0 0.0
0.2
0.4
0.6
0.8
water activity, a,
Figure 1. Relative rate as a function of aw for the transesterification reaction in acetonitrile catalysed by propanol washed subtilisin CLEC (&), subtilisin PREP (r-3), and freeze dried subtilisin (O). Rates for each preparation were normalised relative to the maximum value: 610 nmol/mg/min at a~ of 0.11 for the CLEC, 159.7 nmol/mg/min at aw of 0.44 for the PREP and 3.3 nmol/mg/min aw of 0.76 for the freeze dried powder. Data from Partridge et al [20,22]. 3. Conclusions We have demonstrated that the catalytic behaviour of subtilisin CLECs in polar solvents exhibits pronounced hysteresis. The activity of solvent washed crystals was as much as 80-fold higher than that of crystals dried in air or solvent over molecular sieves. The solvent washing procedure was exploited to obtain high activity immobilised enzyme preparations of subtilisin Carlsberg and ct-chymotrypsin, enzymes with very different secondary and tertiary protein structures. This suggests the procedure may find widespread application as a simple and economical way of preparing biocatalysts for reactions in organic media. 4. Acknowledgement We thank J. J. Lalonde for helpful discussion. We are grateful to the Biotechnology and Biological Science Research Council for financial support.
380 References
.
3. .
5. .
7. 8.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.
A. Koskinen and A. M. Klibanov (eds), Enzymatic Reactions in Organic Media, Chapman and Hall, Andover, 1995. C. R. Wescott and A. M. Klibanov, Biochim. Biophys. Acta 1206 (1994) 1. J. Broos, J. F. J. Engbersen, I. K. Sakodinskaya, W. Verboom and D. N. Reinhoudt, J. Chem. Sot., Perkin Trans. 1 (1995) 2899. J. O. Rich and J. S. Dordick, J. Am. Chem. Soc. 119 (1997) 3245. G. A. Hutcheon, M. C. Parker, A. James and B. D. Moore, Chem. Commun. (1997) 931. K. Dabulis and A. M. Klibanov, Biotechnol. Bioeng. 41 (1993) 566. O. Almarsson and A. M. Klibanov, Biotechnol. Bioeng. 49 (1996) 87. A. O. Triantafyllou, D. B. Wang, E. Wehtje and P. Adlercreutz, Biocatal. Biotransfor. 15 (1997) 185. Z. Yang, A. J. Mesiano, S. Venkatasubramanian, S. H. Gross, J. M. Harris and A. J. Russell, J. Am. Chem. Soc. 117 (1995) 4843 E. Wehtje, P. Adlercreutz and B. Mattiasson, Biotechnol. Bioeng. 41 (1993) 171. J. F. Diaz and K. J. Balkus, J. Mol. Catal. B: Enzymatic 2 (1996) 115. M. T. Reetz, A. Zonta and J. Simpelkamp, Biotechnol. Bioeng. 49 (1996) 527. Y. F. Wang, K. Yakovlevsky and A. L. Margolin, Tetrahedron Lett. 37 (1996) 5317. A. Zaks and A. M. Klibanov, J. Biol. Chem. 263 (1988) 3194 A. Zaks and A. M. Klibanov, J. Biol. Chem. 263 (1988) 8017 S. H. M. van Erp, E. O. Kamenskaya and Y. L. Khmelnitsky, Eur. J. Biochem. 202 (1991) 379. R. H. Valivety, P. J. Hailing and A. R. Macrae, Biochim. Biophys. Acta 1118 (1992) 218 I. Svensson, E. Wehtje, P. Adlercreutz and B. Mattiasson, Biotechnol. Bioeng. 44 (1994) 549 P. J. Hailing, Enzyme Microb. Technol. 16 (1994) 178 J. Partridge, G. A. Hutcheon, B. D. Moore and P. J. Hailing, J. Am. Chem. Soc. 118 (1996) 12873 M. C. Parker, B. D. Moore and A. J. Blacker, Biocatalysis, 10 (1994) 269 J. Partridge, P. J. HaUing and B. D. Moore, Chem Commun. (1998) 841 G. Bell, P. J. Hailing, B. D. Moore, J. Partridge and D.G. Rees TIBTECH 13 (1995) 468.
373
Exploiting hysteresis for high activity enzymes in organic media
Johann Partridge, Peter. J. Hailing and Barry D. Moore Department of Pure & Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street, Glasgow G1 I XL, Scotland, U.K.
Common methods for removing water from protein preparations such as freeze drying and air drying are shown here to be much more detrimental to enzyme activity than simple rapid dehydration. In a comparison of drying methods for cross-linked enzyme crystals of subtilisin Carlsberg, the highest rates of reaction in organic media were obtained with crystals dried by washing with anhydrous polar solvent. Crystals dried over molecular sieves in air or solvent showed significantly lower activity. In all cases, full activity could be recovered in aqueous buffer. When different solvents were used to dry the enzyme crystals, the catalytic rate was found to vary significantly- longer chain alcohols gave the highest rates while smaller "water-like" solvents such as methanol gave much lower rates. The study was extended to immobilised forms of subtilisin Carlsberg and ot-chymotrypsin. When these preparations were dried by solvent washing, high catalytic rates similar to those of crystals prepared by the same procedure were obtained. Catalytic efficiency in polar solvents was 1000 fold greater than that of freeze-dried powders. As with the crystals, subsequent drying over molecular sieves resulted in a substantial loss in enzymatic activity. The activities of solvent washed immobilised enzymes and of cross-linked crystals were found to vary significantly as a function of system water content.
1. Introduction In recent years it has been shown that enzymes exhibit novel properties in low water organic media, such as catalysis of reactions impossible in water, altered substrate specificity and increased thermostability [1]. This has led to an increased demand for enzyme preparations which exhibit high catalytic activity and reproducible selectivity [2-5]. The commonly used freeze-dried powders are known to exhibit low activity [6,7] but are often used because they can be obtained in one step from commercially supplied enzyme. Enzymes immobilised onto supports often show higher activity in organic media but the types of support material reported may not always be readily available and well controlled immobilisation procedures are often required [8-12]. Recently commercial cross-linked enzyme crystals (CLECs) have been introduced. As with enzymes immobilised on supports, these offer advantages over the conventional freeze dried preparations: access to individual enzyme molecules will be improved, whilst particle aggregation and diffusional limitation are considerably reduced. In addition, CLECs have been shown to exhibit high activity, good
374 stability and excellent reproducibility [13]. These characteristics make them potentially very attractive as catalysts for use in organic reaction mixtures. The method by which the biocatalyst is prepared for use in organic media often varies. With freeze dried powders, the enzyme which is essentially dry after lyophilisation, is added directly to the organic solvent which contains a known amount of water [14-16]. Similarly, enzymes immobilised from aqueous solution are collected by filtration, dried under vacuum and used directly in solvent with fixed water content. In other studies, the biocatalyst and solvent are dried exhaustively over water absorbents such as molecular sieves and then pre-equilibrated to well defined thermodynamic water activity (a,,), either together or separately prior to reaction [17,18]. This method offers advantages. Catalytic activity of a particular type of preparation is governed by the amount of water bound to the enzyme molecules, which is only a fraction of the total present. At equilibrium there will be a characteristic relationship between the amount of water bound and that dissolved in the organic phase. If the water level is expressed in terms of its aw, the relationship between catalytic activity and residual water level is generally similar for many different solvents [ 17,18]. There have been some literature reports that different catalytic activity can be found in systems with the same water content or a,,, but different hydration histories [19]. In this work, we set out to examine how different methods of enzyme pre-treatment effected their subsequent catalytic efficiency in organic solvent. Such studies on freeze dried enzymes are problematic. Random but significant differences in morphology between batches of powders is often found, probably due to variations in temperature and pressure during lyophilisation. Differences in catalytic rates can result, so that it can be difficult to make direct rate comparisons between laboratories and also to carry out meaningful studies of possible hysteresis effects arising from different treatments. We anticipated that using more homogeneous enzyme preparations these problems would be reduced. 2. Results and Discussion 2.1
The effect of different CLEC pre-treatments on catalytic efficiency Subtilisin Carlsberg CLECs (ChiroCLEC-BL) were supplied as a suspension in aqueous buffer. The manufacturer's literature reports that this suspension can be stored indefinitely at 4~ without any significant loss in catalytic activity. We were interested in studying the effect of different methods of preparing the crystals for use in organic media on their subsequent catalytic activity. In the present study, we routinely started with the fully active preparation in aqueous suspension and treated the samples by different methods for use in solvent as required. A number of routes can be envisaged for moving from an aqueous suspension to a "dry state". Whilst lyophilisation is necessary when drying dissolved enzymes for use in organic solvents, it is not the simplest and most obvious choice of water removal when enzymes are used as CLECs. Besides this, one of the main reasons for the low activity exhibited by freeze dried powders in organic media is known to be the deleterious effects of dehydration by this method. This process is therefore best avoided. Water can be removed from CLECs by more simple methods, including washing with anhydrous polar solvent and drying over molecular sieves in air or in solvent. As a starting point we explored the effect of washing CLEC with either propanol or acetonitrile. The solvent washed enzyme was then used directly in a reaction, or preequilibrated over molecular sieves in air or solvent (methods commonly applied to
375 conventional freeze dried and immobilised enzymes). We followed the rate of a standard transesterification in anhydrous acetonitrile using CLECs pre-treated in different ways: the initial reaction rates are shown in Table 1. The highest rates were obtained with CLEC placed directly in the reaction mixture atter solvent washing, whilst equilibration at fixed water activity invariably reduced the rate. Although CLEC pre-equilibrated in air gave by far the lowest rate, this preparation is still more active than the conventional freeze dried preparation of subtilisin (<0.01nmol/mg/min). Interestingly, CLEC pre-equilibrated in anhydrous acetonitrile over molecular sieves gave higher transesterification rates. Such an observation has been previously reported in the literature [21]. Freeze dried subtilisin powder in hexane was shown to give higher rates when equilibrated in solvent as opposed to in air. However, the difference between the two treatments is much greater for the CLECs in acetonitrile. Table 1. Effect of enzyme treatment on catalytic activity of subtilisin Carlsberg CLEC in anhydrous acetonitrile. Washin B Solvent" Further Enzyme Treatment ~ PrOH none PrOH 1 PrOH 2 ACN none ACN 1 ACN 2 PrOH b none PrOH, AcN none AcN, PrOH none AcN, aqueous buffe,r, PrOH .. none
Rate (nmol/m~min) 224 • 7 34 + 4 2.9 + 0.1 112 + 4 33 + 3 2.3 + 0.2 187 + 2 84 • 4 102 + 3 190 + 14
Data from Partridge et al [20] " CLEC was rinsed 3 times with 1 ml of each solvent shown respectively (with the exception of b which was rinsed 6 times). Atter each wash, the solvent was removed by centrifugation. c (1) Solvent washed CLEC was suspended in anhydrous AcN and then equilibrated over molecular sieves for 3 days, 20~ (2) Solvent washed CLEC was equilibrated over molecular sieves for 3 days, 20~ After this time it was suspended in anhydrous AcN. From Table 1 it is apparent that the difference in catalytic rate for CLEC dried over molecular sieves through the vapour phase is small, regardless of the solvent used to wash the crystals prior to equilibration. This is also the case for CLEC equilibrated in the reaction solvent prior to use. However, it is interesting to note that CLEC washed in propanol and used directly in the reaction gives a rate approximately double that of CLEC which was washed in acetonitrile. When the crystals were washed consecutively with propanol (x3) and acetonitrile (x3), regardless of the order in which the solvents were used, the rates of catalysis were less than half that for CLEC washed solely in propanol, but were close to that of acetonitrile washed CLEC. When the number of washes is increased from 3 to 6 for propanol washed CLEC, the decrease in rate is barely significant. It is apparent that the effect of acetonitrile is dominant on the crystals and the catalytic efficiency of an acetonitrile washed
376 CLEC can not be increased by subsequent washing with propanol. To ascertain whether or not acetonitrile washing had a long term detrimental effect on the crystals, we measured the catalytic activity of CLEC which was rinsed with acetonitrile, buffer and propanol respectively. The majority of catalytic activity was recovered, indicating that acetonitrile does not appear to damage the CLEC irreversibly. It might also be inferred that pre-equilibration through the vapour phase results in structural damage to the enzyme crystals, so that most of their catalytic activity is irreversibly lost. We attempted to eliminate this possibility experimentally. Biocatalyst which had been solvent washed and leit to equilibrate through the vapour phase for 3 days, was returned to an aqueous environment. After this, it was solvent washed to remove excess water, and used immediately to catalyse a reaction in anhydrous acetonitrile. CLEC was found to regain high catalytic activity. Furthermore, the rate observed is comparable to that of CLEC which has undergone solvent washing only. It is interesting to note that the catalytic efficiency of CLECs extensively dried over molecular sieves can not be improved significantly by further equilibration over a high aw saturated salt. CLEC must be returned to the aqueous buffer if high catalytic activity is to be recovered. The observations discussed above are not exclusive to reactions in acetonitrile. Experiments in propanol have confirmed that enzyme which has undergone solvent washing has a catalytic rate more than 12-fold higher than that which has been dried further using molecular sieves. Similarly, for reactions in propanol, enzyme pre-equilibrated through the vapour phase can regain high catalytic activity by exposure to an aqueous environment. The results discussed above clearly demonstrate that the catalytic behaviour of CLECs exhibits pronounced hysteresis: rates vary by 80-fold depending on previous hydration history. One might speculate that the enzyme or enzyme crystal has a 'memory' of how it has been treated, and this 'memory' can only be erased by returning the CLEC to an aqueous environment. Two possible explanations may account for this observation: (1) different dehydration protocols effect the amount of water left bound to the enzyme, or (2) the conformation of the dried enzyme is very sensitive to the method of water removal. Most probably these two effects are intimately related. Since washing the CLEC with solvent gives the most efficient rates of catalysis, we went on to investigate the effect of washing with other solvents. Table 2 shows how transesterification rate in anhydrous acetonitrile varies as much as 7-fold depending on the organic solvent used to dry the catalyst. The alcohols generally gave the fastest rates of catalysis, followed by acetonitrile and acetone which gave intermediate rates. Interestingly, the smaller "water-like" solvents, ethane diol and methanol gave the lowest rates of all solvents tested. With CLEC washed in methanol, only 33% of the maximum activity is regained after brief washing with buffer and propanol. Leaving in buffer overnight resulted in increased catalytic activity (63% of the original propanol washed CLEC). Nevertheless, the catalytic performance of all solvent washed CLECs was still higher than that which had been equilibrated in air or in solvent at aw < 0.01. A plausible explanation for this variation in catalytic rate when different solvents are used to dry the CLEC is that each solvent has a very specific capacity to displace the water within the crystal. Conformational changes may result on removal of more water from the crystal leading to a less active enzyme state and decreased rates of catalysis. On the basis of these data one can conclude that care must be taken when choosing a solvent to wash the crystals if maximum enzymatic activity is to be achieved.
377 Table 2. Effect of different solvent washes on the catalytic activity of subtilisin Carlsberg CLE C in anhydrous acetonitrile.
Washin~ Solvent' EtOH BuOH PrOH Me2CO ACN Ethane Diol CH3 OH MeOH, aqueous buffer ,. PrOH
Rate (nmol/mg/min) 282 + 32 243 • 23 224 • 7 121 + 11 112 + 4 50 + 12 39+ 3 ...... 74 + 1
Data from Partridge et al [20] ' CLEC was rinsed 3 times with 1 ml of each solvent shown respectively. After each wash, the solvent was removed by centrifugation. The enzyme was then suspended in anhydrous acetonitrile. 2.2
Extension of our findings to immobilised preparations With dissolved enzymes water must generally be removed by freeze-drying or similar prolonged dehydration methods. With CLECs, rapid drying by solvent rinsing is possible and this appears to lead to better specific activity. We therefore hypothesised that if non-crystalline but immobilised enzymes were treated using the same solvent washing procedure, very high activities might be obtained. We tested this hypothesis for an immobilised form of subtilisin Carlsberg. The enzyme was prepared using a simple adsorption procedure from aqueous buffer (pH 7.8) unto a standard silica gel support. Table 3 shows a comparison of the initial reaction rates obtained in acetonitrile using subtilisin Carlsberg prepared in different forms. The propanol rinsed immobilised enzyme preparation (termed PREP) gave fairly low rates in anhydrous acetonitrile. Table 3. Effect of preparation type on subtilisin Carlsber~; activity in acetonitrile. Rate (nmol / nag "/rain) 0% H20 v/v in ACN 1% H20 v/v in ACN Enzyme form and treatment (aw < 0.01) (a, = 0.22)
freeze dried powder, air dried b CLEC, PrOH washed CLEC, PrOH washed, air dried b immobilised form, PrOH washed immobilised form, PrOH washed, air drie db
< 0.01 226 2.94 0.82 ....
0.13 610 13.8 142 0.60
Data from Partridge et al [20,22] ' refers to weight of enzyme in preparation, b sample placed in sealed jar over molecular sieves for 3 days to give enzyme at aw<0.01; further 3 days equilibration over H20-saturated potassium acetate was carried out for enzyme at a, = 0.22.
378 However, on addition of 1% water v/v to the reaction solvent, the PREP was found to exhibit high activity. At this water level, rates for the PREP were found to be comparable to the CLEC, and over 1000 times greater than the commonly used freeze-dried powders. If PREP was subsequently dried in air prior to assaying in the organic solvent most of the activity was lost and the residual level approached that obtained for the lyophilised powder (see Table 3). However, aqueous suspensions of the silica adsorbed enzyme can be stored at 4~ for at least 3 weeks with negligible loss of activity and converted to the PREP as required. The effect of washing the immobilised enzyme with different anhydrous organic solvents was also tested. The same pattern of results as that observed for the CLEC emerged: ethanol washing gave high rates, acetonitrile gave intermediate rates and methanol produced a much less active preparation. Cross-linked enzyme crystals of chymotrypsin are not commercially available but under the same reaction conditions described in Table 3 the PREP of this enzyme exhibited high catalytic activity in acetonitrile with water levels of 1%v/v and above. Again these rates were two orders of magnitude better than the freeze-dried powder. It is perhaps surprising the method described here for preparing conventional biocatalysts for reactions in low water media has not been reported previously. However, until recently the large hydration hysteresis effects obtainable with enzymes had not been fully recognised. Our work with CLECs [20] and immobilised enzymes [22] has shown that different methods of water removal can dramatically affect the enzyme activity obtained. 2.3
The effect of system water content on catalytic activity Previous studies with biocatalysts in organic media have shown that the amount of water present in the system plays an important role in controlling factors such as rate, stability and hydrolytic equilibria [ 1, 23]. We therefore proceeded to carry out a more detailed study of the variation in catalytic activity as a function of water level in the actual reaction mixture for subtilisin CLEC and the immobilised form which had been propanol washed. Figure 1 shows the rate profiles for propanol washed CLEC and PREP in acetonitrile as a function of the thermodynamic water activity, aw. The profile for the freeze dried form of this enzyme is also shown for comparison. However, it should be noted that since the activity per unit weight of enzyme is much lower in the lyophilised powder, the rate profiles for the three preparations are shown as relative rates normalised to their maximum values. According to previous studies the amount of water bound to a protein in solvent would be expected to be controlled by a~ [23]. Under these conditions differences in the amount of residual bound water should be eliminated and hence similar rate vs aw profiles might be expected for all three forms of the enzyme. As can be seen in Figure 1 this is not the case. The activity of the lyophilised powder continues to increase even up to aw 0.76, while a maximal rate is obtained at aw of 0.11 with the CLEC and aw of 0.44 with the PREP. The different rate profiles could arise because the three preparations of the enzyme differ in either water binding or the water required for catalytic activity. A large change in water binding isotherms is unlikely, but kinetic factors may be significant. The water content of the lyophilised powder will be determined by its adsorption isotherm. In contrast, treatment of the CLEC and PREP imposes a water desorption process, for which the isotherm will be different due to hysteresis, and even the apparent equilibrium value may not be reached in the time employed. An alternative hypothesis is that CLEC and PREP require less water because the propanol dehydration process leaves a large proportion of the enzyme molecules in
379 a conformation close to the active fOrm. Only low levels of water may then be needed to promote catalysis and further increases in water availability provide no beneficial effect. With the freeze-dried powder the very low rates suggest most of the enzyme is initially inactive. In this case a water catalysed reorganisation process is probably required to convert the enzyme back to an active state since much greater water levels are needed to obtain high activity. The difference in profiles in Figure 1 would therefore reflect the fact that water plays different roles in promoting enzyme catalysis depending on the hydration history of the system.
1.0 .
0.8
~> 0.6 "~
0.4 0.2 0.0 0.0
0.2
0.4
0.6
0.8
water activity, a,
Figure 1. Relative rate as a function of aw for the transesterification reaction in acetonitrile catalysed by propanol washed subtilisin CLEC (&), subtilisin PREP (r-3), and freeze dried subtilisin (O). Rates for each preparation were normalised relative to the maximum value: 610 nmol/mg/min at a~ of 0.11 for the CLEC, 159.7 nmol/mg/min at aw of 0.44 for the PREP and 3.3 nmol/mg/min aw of 0.76 for the freeze dried powder. Data from Partridge et al [20,22]. 3. Conclusions We have demonstrated that the catalytic behaviour of subtilisin CLECs in polar solvents exhibits pronounced hysteresis. The activity of solvent washed crystals was as much as 80-fold higher than that of crystals dried in air or solvent over molecular sieves. The solvent washing procedure was exploited to obtain high activity immobilised enzyme preparations of subtilisin Carlsberg and ct-chymotrypsin, enzymes with very different secondary and tertiary protein structures. This suggests the procedure may find widespread application as a simple and economical way of preparing biocatalysts for reactions in organic media. 4. Acknowledgement We thank J. J. Lalonde for helpful discussion. We are grateful to the Biotechnology and Biological Science Research Council for financial support.
380 References
.
3. .
5. .
7. 8.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.
A. Koskinen and A. M. Klibanov (eds), Enzymatic Reactions in Organic Media, Chapman and Hall, Andover, 1995. C. R. Wescott and A. M. Klibanov, Biochim. Biophys. Acta 1206 (1994) 1. J. Broos, J. F. J. Engbersen, I. K. Sakodinskaya, W. Verboom and D. N. Reinhoudt, J. Chem. Sot., Perkin Trans. 1 (1995) 2899. J. O. Rich and J. S. Dordick, J. Am. Chem. Soc. 119 (1997) 3245. G. A. Hutcheon, M. C. Parker, A. James and B. D. Moore, Chem. Commun. (1997) 931. K. Dabulis and A. M. Klibanov, Biotechnol. Bioeng. 41 (1993) 566. O. Almarsson and A. M. Klibanov, Biotechnol. Bioeng. 49 (1996) 87. A. O. Triantafyllou, D. B. Wang, E. Wehtje and P. Adlercreutz, Biocatal. Biotransfor. 15 (1997) 185. Z. Yang, A. J. Mesiano, S. Venkatasubramanian, S. H. Gross, J. M. Harris and A. J. Russell, J. Am. Chem. Soc. 117 (1995) 4843 E. Wehtje, P. Adlercreutz and B. Mattiasson, Biotechnol. Bioeng. 41 (1993) 171. J. F. Diaz and K. J. Balkus, J. Mol. Catal. B: Enzymatic 2 (1996) 115. M. T. Reetz, A. Zonta and J. Simpelkamp, Biotechnol. Bioeng. 49 (1996) 527. Y. F. Wang, K. Yakovlevsky and A. L. Margolin, Tetrahedron Lett. 37 (1996) 5317. A. Zaks and A. M. Klibanov, J. Biol. Chem. 263 (1988) 3194 A. Zaks and A. M. Klibanov, J. Biol. Chem. 263 (1988) 8017 S. H. M. van Erp, E. O. Kamenskaya and Y. L. Khmelnitsky, Eur. J. Biochem. 202 (1991) 379. R. H. Valivety, P. J. Hailing and A. R. Macrae, Biochim. Biophys. Acta 1118 (1992) 218 I. Svensson, E. Wehtje, P. Adlercreutz and B. Mattiasson, Biotechnol. Bioeng. 44 (1994) 549 P. J. Hailing, Enzyme Microb. Technol. 16 (1994) 178 J. Partridge, G. A. Hutcheon, B. D. Moore and P. J. Hailing, J. Am. Chem. Soc. 118 (1996) 12873 M. C. Parker, B. D. Moore and A. J. Blacker, Biocatalysis, 10 (1994) 269 J. Partridge, P. J. HaUing and B. D. Moore, Chem Commun. (1998) 841 G. Bell, P. J. Hailing, B. D. Moore, J. Partridge and D.G. Rees TIBTECH 13 (1995) 468.