468

Biocatalyst behaviour in lowwater systems

George Bell, Peter J. Hailing, Barry D. Moore, Johann Partridge and D. Gareth Rees
An improving understanding of the parameters that affect biocatalyst activity,

specificity and stability in low-water non-aqueous media make reliable predictions

about the behaviour of such systems increasingly feasible. Here, we discuss some of the key factors, such as control of water activity, and the effects of solvent on Km and protein-ionization state, that must be addressed in order to obtain

predictable results.

The factors affecting activity, specificity and stability o f enzymes in aqueous media are well documented, and this allows experiments to be designed to answer very specific questions in process optimization. A similar level o f understanding is now being developed for biocatalyst behaviour in organic and other low-water systems. Many predictions ofbiocatalyst behaviour are based on relatively simple physical or chemical interpretations, sometimes combined with knowledge from biocatalysis in aqueous systems. T h e r m o d y n a m i c approaches that consider the distribution o f c o m ponents among the various phases, and their solvation in the bulk (i.e. organic) phase have proved to be particularly useful. These advances should be useful to those developing applications (see tLefs 1 and 2 for examples, and Ref. 3 for an example of a recent c o m mercial process). (For other recent reviews about enzymes in non-aqueous media, see tLe 4-10.) This article addresses those aspects of biocatalyst behaviour for which there is now a reasonably reliable predictive explanation. We focus on behaviour in non-aqueous media (NAM), but do not cover micelles/microemulsions 11 and water-rich aqueousorganic mixtures. T h e use of N A M for enzyme-catalysed reactions has several advantages, including: (1) the catalysis o f reactions that are unfavourable in water (e.g. reversal o f hydrolysis reactions in favour o f synthesis); (2) the solubilization o f hydrophobic substrates; (3) ease o f recovery for some products; (4) ease of recovery o f the insoluble biocatalyst; and (5) increased biocatalyst thermostability (at low water levels). N A M are also useful for fundamental studies of the effects of water on enzyme dynamics and catalysis.
G. Bell, P. J. Hailing, B. D. Moore,J. Pamidge and D. G. Rees are in tke Departments of Ckemical & ProcessEngineering, Bioscience & Biotecknology, and Pure & Applied Chemistry, University of Strathdyde, Glasgow, UK G1 1XW.
TIBTECHNOVEMBER1995 (VOL13)

The importance o f water and the control o f water activity (aw) The limited amount o f water present in N A M greatly influences the dynamic and catalytic properties of enzymes. Much of the work on the behaviour o f biocatalysts in such systems has been carried out with total water content fixed, and other parameters varied*. However, there are inherent problems with this approach, and these can limit useful interpretation o f experimental results. Zaks and Klibanov d e m o n strated that it is the water bound to the enzyme that determines the catalytic activity, rather than the total water content o f the system 12. If the same quantity of water were added to systems containing various organic solvents, the amounts o f water associated with the enzyme would vary. In general, the more polar the solvent, the more water will be held in solution and, therefore, unable to bind to the enzyme. It is increasingly accepted that the water relationships in N A M are best described by the t h e r m o dynamic water activity (aw; ILef. 7); (Boxes 1 and 2). Conventionally, the aw o f pure water is taken as 1. This is convenient when various solvents or media are to be compared. O n this basis, aw values vary from 1, down to 0 for a completely dry system. This means that aw is not usually close in numerical value to the water concentration. The aw o f all phases in the reaction mixture will be the same at equilibrium, even though their water concentrations will usually be different. If enzymatic reactions are carried out in different media at fixed aw, the amount of water associated with the biocatalyst will tend to remain the same, thus simplifying interpretation and prediction o f changes in enzyme performance. In systems with a constant water
*The terms 'content' and 'concentration' are often used interchangeably in relation to water. In this article wc make a clear distinction: 'water concentration' refers to that in a homogeneous single phase, norreally in true solution in an orgamc liquid; 'water content' refers to the total quantity in the whole multi-phase system, with its distribution between the phases often unknown.

1995, Elsevier Science Ltd

0167- 7799/95/$9.50

469

reviews
Box 1. What is water activity?
Concentrations of reactants are usually adequate to describe equilibria for enzyme reactions in aqueous systems. This is convenient, as such concentrations are easy to understand and measure. Unfortunately, the concentration of water is not an adequate measure of its effect in NAM, as the influence of water depends not only on the amount of water, but also on the interactions between water and other components in the same phase. It is better to consider the thermodynamic activity of water (aw). It is difficult to visualize aw, but an analogy with another thermodynamic concept, temperature, may be useful. Consider a box containing a piece of wood, a piece of metal and a beaker of water. Eventually, all the materials will reach thermal equilibrium: they will be at the same temperature although the amount of heat will be different in each material as a result of their different heat capacities. We have grown used to the difficult concept of temperature; it is just accepted that heat distributes itself throughout a system until all the temperatures are equal. Now, consider a closed flask with an organic solvent containing dissolved water, some enzyme powder suspended in the solvent, and a vapour space. Water will distribute itself throughout the system. After a time, the system will reach equilibrium. The aw will be the same in each phase - hydrated enzyme, solvent and headspace although the concentrations of water will vary. Thus, equilibrium effects such as enzyme hydration and the balance between hydrolysis and synthesis reactions depend on aw. More simply, aw may be considered as a measure of water's 'preference' to be in one environment rather than another. Water molecules in a medium at high aw 'want' to move to one of lower aw. The resulting movement reduces the difference between the aw values of the media, until they become equal at equilibrium. At equilibrium: temperatures the same - heat contents differ; water activities the same - water concentrations differ.

content, apparent effects o f changing the solvent may simply reflect competition for water between the enzyme and the rest o f the system. In such cases, interpreting observed water effects in terms o f a w allows them to be explained using fewer independent variables. There are two further advantages: (1) Since aw is the same in all phases, the water activity o f the w h o l e system can be determined in the phase that is most convenient. Reasonably g o o d water sensors are available, at least for gas phases. Details o f this and other methods for water measurement and control are available on requestt. (2) There are relatively simple methods for fixing a w that can be used if equipment to measure water activi t ' / o r concentration is unavailable. This can improve repeatability w h e n there is variability in the initial water contents o f components o f the reaction mixture, e.g. between batches o f biocatalyst, a c o m m o n problem with simple freeze drying. Methods already available for measuring, fixing or controlling a w in these systems have been reviewed recently 7, and a further m e t h o d for controlling a w has recently been described by Wehtje et al. (Ref. 13): a saturated salt solution is circulated through the reaction vial via silicone tubing that is submerged in the reaction medium. A n y water produced in the reaction is removed by diffusion through the tube walls, maintaining an equilibrium aw set by the salt solution used. T h e a m o u n t o f water b o u n d to enzymes in organic solvents is very similar to that b o u n d to enzymes in air at the same aw, at least for low values. This was c o n cluded from calculations carried out on published data originally expressed in terms o f water concentration in the organic phase TM, and has n o w been confirmed
A detailed methods' sheet can be supphed on request (preferably by e-mail, p.j.haUing@strath.ac.uk).

by direct measurement 1s,16, At higher a w values, polar solvents (e.g. ethanol) appear to cause a slight decrease in the a m o u n t o f water associated with the enzyme; in contrast to the earlier prediction TM,nonpolar solvents also have the same effect 16. In the case o f polar solvents in particular, this may be attributed to direct c o m petition o f the solvent with water for binding sites. T h e effects o f changing solvent are best studied at fixed a w values 17,18. T h e maximal catalytic activity o f each e n z y m e studied was f o u n d to occur at the same a w, regardless o f the solvent used. This suggests that the solvents used did not directly affect e n z y m e - w a t e r interactions, as the overall shape o f profiles o f catalytic activity versus aw were the same. However, the absolute magnitude o f the rates did differ from solvent to solvent. Variations in aw were also responsible for some apparent solvent effects on enantioselectivity: w h e n studies were carried out at fixed aw, the effects disappeared 19. Different supports will adsorb different amounts o f water. Thus, to observe the direct effect o f the support on catalytic activity, it is again important to carry out reactions at k n o w n a w values. Studies o f this t y p e 2,21 have shown that profiles o f catalytic activity versus a w have the same general shape for several supports (Fig. 1), although the absolute rates vary considerably. These results suggest that most supports do not directly affect the e n z y m e - w a t e r interaction. However, for some supports, the shape o f the profile did change and, in one case, there was a dramatic change. In addition, some apparent effects on enantioselectivity may be the result o f competition between the support and the enzyme for water 22. W h e r e the desired reaction is a reverse hydrolysis, the synthetic yield will be maximized at low aw. A compromise must be made with the requirements for biocatalyst activity (for recent examples, see KeN 23,24). TIBTECH NOVEMBER (VOL13] 1995

470

reviews
Box 2. How to determine the activity of water (aw) in organic solvents
When a liquid mixture is heated, it will reach a temperature at which a bubble of vapour is formed - its bubble point. Similarly when a vapour is cooled, it reaches a temperature at which a drop of liquid forms - its dew point. Consider a mixture of acetonitrile and water in equilibrium with its vapour at 760 mmHg. The liquid is at its bubble point, and the vapour at its dew point. Experimental data show that if the mole fraction of water (xw) in the liquid is 0.858, then at equilibrium the mole fraction in the vapour will be 0.441 and the temperature will be 80C. If the liquid and vapour are in equilibrium, their temperature and the thermodynamic activity of each chemical component must be equal. In this case, we shall consider only the aw: At low pressures, aw is the ratio of the partial pressure of water in the vapour to the vapour pressure that pure water would have at the same temperature. (At pressures above - 5 bar, the thermodynamics become more complicated for the vapour phase - fugacities have to be used instead of pressures.) The partial pressure of water is the total pressure multiplied by its mole fraction (760 x 0.441 mmHg). The vapour pressure of pure water at 80C is 355 mmHg. Hence aw under these conditions is: 760 0.441 - 0.944 355 In a given medium, the aw value will increase with water concentration. Quantitatively, it is normally preferable to express water concentration as a mole fraction, xw. We may then write an equation aw-- ywXw, with Yw known as the activity coefficient. Sometimes Tw is almost independent of concentration, at least over some ranges. Analysis over the full range of water concentrations, xw gives the graph:

0.8 _,~ 0.6 ._>

0.4
0.2 0
I I i 0 0.2 0.4 0.6 (Pure acetonitrile) Water mole fraction (xw) I

0.8 (Pure water'

If the experimenter knows the water content in the solvent, the aw value can be found. For NAM, only the initial section of the graph is of interest: note that the steeper the slope, the more non-ideal the system is. The effect o f water on enzyme activity is sufficiently pronounced that this effect has been used to develop a moisture sensor for N A M (Ref. 25). vent effects on substrate binding 28. In-depth discussions o f enzyme specificity in N A M have recently been presented 8,31. There appears to be a general tendency for K m values to increase with water concentration or aw in a given solvent 32-34.

Kinetics Most kinetic studies have shown that the behaviour of enzymes in N A M follow conventional models (such as the Michaelis-Menten or Ping-Pong26.27). Some individual rate-constants can also be determined by measurements o f partial reactions or the partitioning o f a cormnon intermediate 27,28. However, the values o f kinetic parameters may be very different from those for the same enzyme and substrate in aqueous media, primarily because ofsolvation effects (Box 3). A quantitative treatment of such effects (for examples, see Refs 7,29,30) can be used to predict effects on kinetic parameters, which may account for observed differences between solvents. Where possible, a determination of the true enzyme~substrate dissociation constant (Ks) gives a better indication of solTIBTECHNOVEMBER1995 (VOL 13)

Acid, base and ion effects p H is a familiar concept in aqueous-medium enzymology. Where there is no longer a dilute aqueous phase (i.e. a,v<<l), the concept must be modified. In N A M , the p H o f the solution from which an enzyme is lyophilized exerts a strong influence on the subsequent behaviour o f the enzyme ('pH m e m o r y ' ; Ref. 35); the p H of the original solution undoubtedly controls the ionization states of enzyme functional groups. However, pre-adjustment o f p H is inflexible as further p H changes cannot be made after the reaction is initiated, and if a large number o f p H values need to be tested, this method becomes laborious. In addition, some reactions alter acid/base concentration,

471

reviews

100

60 40 rr 20

,0
a

02

04

06

018
Figure 1

Water activity (aw)

5 10 20 50 100 200 mg water per g dry catalyst

500

The use of water activity to simplify interpretation when support and enzyme compete for water. The effects on catalytic activity of immobilized Candida rugosa lipase as a function of (a) water activity (awl and (b) water content. Support was polyamide (crosses), polypropylene (open circles) or anion-exchangeresin (closed circles). The total water content of the immobilized enzyme was determined (the errors are large at the lowest water contents). The aw profile for a fourth support, hydrophobed porous glass, was somewhat different. (Data from Ref. 17.)

Box 3. How does solvent affect Kr.?

The first step of any enzyme reaction is the binding of the substrate to the active site. This involves a desolvation step in which the substrate is stripped of solvent molecules. The more strongly the solvent molecules are associated with the substrate, the less favourable binding will be. (Binding may also involve expulsion of solvent from within the active site but for simplicity we shall ignore that here.)

+
Substrate-solvent Enzyme

Substrate-enzyme

@ @

Solvent

Generally, an equilibrium is established between substrate bound to the enzyme and unbound substrate. The position of this equilibrium depends on the free energy of solvation, and in many cases can be usefully approximated by the Michaelis-Menten constant Kin. In such cases, Km is a measure of the relative affinity of the substrate for the enzyme active site compared with solvent molecules. substrate with low affinity for solvent ~binds strongly to enzyme, low Km substrate with high affinity for solvent ~binds weakly to enzyme, high Km In aqueous and organic media, the difference in Km between substrates is a major factor in determining relative reaction rates and hence enzyme specificity. In organic media it must also be remembered Krn will vary from solvent to solvent. To a crude approximation we can use the rule 'like prefers like' to describe solvation. Hence, polar substrates will tend to have higher Km in polar solvents than in nonpolar solvents. Changes in solvent affinity with substrate structure are determined by a wide range of intermolecular forces. Differences in Km can often be used to explain changes and even reversals in substrate specificity. Similarly, because of variations in Km with solvent, observable reaction rates can vary dramatically if the medium is changed but concentrations of reactants are kept constant. When studying enzyme catalysed reactions in organic solvents it is important to ensure that the variation in Km with solvent has been accounted for before drawing further conclusions about direct solvent effects on the kinetics and specificity.

thus influencing biocatalyst performance. The problem would be better solved by introducing buffering compounds into the organic phase. Buffering systems involving highly hydrophobic acids and their sodium salts, and highly hydrophobic bases together with their hydrochlorides, have been develope@6. The ratio of acid/amine to their salts can be varied to buffer changes in acidity/basicity. Such buffers

have been shown to affect the performance of two modal-end/me reactions in low-water systems, largely eliminating the need for pre-lyophilization pH-adjustment (Fig. 2). This is the first step towards the development of a range of compounds that, ultimately, will need to approach the diversity of buffering species that are available for conventional aqueous-medium enzymology.
TIBTECHNOVEMBER1995 (VOL 13)

472

reviews
120

100 7 -

E
t-

80
a0

O
t'-

~.
t-

40

20

pH of aqueous washing solution Figure 2 'pH memory' and its abolition by organic-phase buffers. Immobilized subtilisin was pH-adjusted and dried before catalytic activity was measured in organic medium with (open circles) or without (closed circles) the addition of Na+-triphenylacetatebuffer (1:1). (Data from Ref. 36.)

value. There can also be very specific interactions between enzyme and solvent molecules, as demonstrated by the differential effects o f using enantiomers o f a chiral solvent42. As these will solvate water and the achiral substrate identically, the differences must result from interactions with the chiral enzyme. Direct effects o f solvent on enzyme selectivity have been discussed more fully in Ref. 8. The choice of organic solvent for a given reaction should be determined by three factors: (1) the effects o f solvent on the reaction [including solubility o f the substrates, reaction equilibrium position 43,44, kinetics and enzyme specificity]; (2).the toxicity o f the solvent (important for food and pharmaceutical-based processes where compliance with safety and solventdisposal legislation will be a major consideration); and (3) the effect of the solvent on biocatalyst stability. The following, commonly used solvents offer a wide range ofphysicochemical characteristics for pilot experiments: hexane, toluene, isopropyl ether or methyl tertiary butyl ether, methyl isobutyl ketone, ethyl acetate, n-butanol, ethanol, dimethylformamide, tetrahydrofuran and acetonitrile.

Biocatalyst form and preparation

Without special treatment, proteins are largely insoluble in most organic solvents, including anhydrous systems, and those with only a few percent o f dissolved water. (A few solvents, such as dimethylsnlphoxide and 2,2,2-trifluoroethanol, will dissolve proteins.) Consequently, enzymes are usually present as solid-liquid biphasic suspensions. This has the advantage that the insoluble enzyme is readily separated, making it easy to reuse and simplifying the downstream processing of the product. A disadvantage is that reaction rates can exhibit the same diffusional limitations that are observed when solid-phase immobilized biocatalysts are used in aqueous reaction media. The need to construct a model o f diffhsional constraints o f solid enzymes 4S has prompted studies o f the microscopic morphology o f lyophilized enzyme powders. Lyophilized particles o f subtilisin Carlsberg have had their fine structure characterized at varying hydration/water activity by environmental scanning electron microscopy 46. The effect o f increasing hydration of enzyme powders (achieved by altering the temperature and pressure in the instrument) studied directly after lyophilization [10% H 2 0 (w/w)] was to convert flake/sheets (100 0.01-0.1 txm thick) to a swollen branched structure at an aw o f about 0.85. Several different approaches have been used for the preparation o f solid-phase enzyme catalysts for use in NAM. Three commonly used methods provide some opportunity to adjust key parameters such as p H and associated ions: (1) lyophilization (freeze drying) - this is the most widely used method; (2) precipitation, usually by the classical technique o f acetone addition; and (3) immobilization, most often by adsorption. In aqueous systems, immobilization often limits diffusion. As even free enzymes form solid suspensions in NAM, immobilizing them may actually reduce the

Temperature
Increasing the temperature produces the same pattern o f effects on biocatalysts in N A M as it does in water: faster rates, but faster inactivation. However, the thermostability o f a given enzyme may increase strikingly at low water-content. Reduced thermoinactivation in NAM is probably a result of the low aw, rather than any direct influence o f the solvent. The mounting evidence for this has been derived from differential-scanning calorimetry and other spectroscopic measurements 3749, and also from standard inactivation studies 4. Raising the amount of water usually increases the inactivation rate: enzyme molecules with small amounts o f associated water are believed to become more rigid 41, reducing the rate of inactivation. Enzymes that have greater stability in aqueous solution also tend to show higher stability in NAM.

Choosing an appropriate solvent

The effects o f various solvents on enzyme behaviour have been studied in detail. Attempts are often made to correlate the effects with solvent hydrophobicity (generally measured by logP). However, the solvent is frequently not directly responsible: sometimes changes in biocatalyst behaviour are actually due to water (when solvents have been compared at equal water content rather than aw - see above); and other effects may reflect substrate solvation. The solvent may also denature or inhibit enzymes, and logP values may be a useful guide when this is the case. With a bulk organic phase, activity and stability are generally better in solvents with a higher logP
TIBTECHNOVEMBER1995 (VOL13)

473

reviews
diffusional limitation. When dispersed over a solid surface, access to individual enzyme molecules is improved, and particle aggregation is reduced. Furthermore, enzymes that work optimally at an interface will benefit from increased interracial area. The use o f enzymes in crystalline form is also possible, but this precludes the adjustment o f key parameters. However, this is possible for cross-linked enzyme crystals47, which may prove to be useful, although they are expensive at present.
13 Wehtje, E., Svensson, I., Adlercreutz, P. and Mattiasson, B. (1993) Biotechnol. Tech. 7, 873-878 14 Hailing, P.J. (1990) Biochim. Biophys. Acta 1040, 225-228 15 McMinn, J. H., Sowa, M. J., Charnick, S. B. and Paulaitis, M. E. (1993) Biopolymers 33, 1213-1224 16 Parker, M. C., Moore, B. D. and Blacker, A. J. (1995) Biotecknol. Bioeng. 46, 452-458 17 Valivety, R. H., Halling, P. J. and Macrae, A. R. (1992) Biochim. Biophys. Acta 1118, 218-222 18 Svensson, I., Wehtje, E., Adlercreutz, P. and Mattiasson, B. (1994) Biotechnol. Bioeng. 44, 549-556 19 Berglund, P., Vorde, C. and Hogberg, H. E. (1994) Biocatalysis 9, 123-130 20 Adlercreutz, P. (1991) Eur.J. Biochem. 199, 609-614 21 Valivety, R. H., Halling, P. J., Peilow, A. D. and Macrae, A. R. (1994) Eur.J. Biochem. 222, 461-466 22 Orsat, B., Drtina, G.J., Williams, M. G. and Klibanov, A. M. (1994) Biotechnol. Bioeng. 44, 1265-1269 23 Ljunger, G., Adlercreutz, P. and Mattiasson, B. (1994) Enzyme Microb. Technol. 16, 751-755 24 Dudal, Y. and Lortie, R. (1995) Biotechnol. Bioeng. 45, 129-134 25 Manmno, 8., Cosio, M. S. and Wang, J. (1994) Analyst 119, 2001-2003 26 Mart'/, A., Chulalaksananukul, W., Willemot, R. M. and Condoret, J. S. (1992) Biotechnol. Bioeng. 39, 273-280 27 ChatteIjee, S. and Russell, A. J. (1993) Enzyme Microb. Technol. 15, 1022-1029 28 Chatterjee, 8. and Russell, A. J. (1992) Biotechnol. Bioeng. 40, 1069-1077 29 Parida, S. and Dordick, J. S. (1993) J. Org. Chem. 58, 3238-3244 30 Wescott, C. R. and Klibanov, A. M. (1993)J. Am. Chem. Soc. 115, 1629-1631 31 Wescott, C. R. and Klibanov, A. M. (1994) Biochim. Biophys. Acta 1206, 1-9 32 van Erp, S. H. M., Kamenskaya, E. O. and Khmelnitsky, Y. L. (1991) Eur. J. Biochem. 202, 379-384 33 Bovara, R., Carrea, G., Ottolina, G. and 1kiva, 8. (1993) Biotechnol. Lett. 15, 937-942 34 Valivety, R. H., Hailing, P. J. and Macrae, A. R. (1993) Biotechnol. Lett. 15, 1133-1138 35 Yang, Z., Zacherl, D. and Russell, A.J. (1993) J. Am. Chem. Soc. 115, 12251-12257 36 Blackwood, A. D., Curran, L.J., Moore, B. D. and Halling, P. J. (1994) Biochim. Biophys. Acta 1206, 161-165 37 Volkin, D. B., 8taubli, A., Langer, R. and Klibanov, A. M. (1991) Biotechnol. Bioeng. 37, 843-853 38 Battistel, E. and Bianchi, D. (1994) J. Phys. Chem. 98, 5368-5375 39 Turner, N. A., Duchateau, D. B. and Vulfson, E. N. (1995) Biotechnol. Lett. 17, 371-376 41) Toscano, G., Pirozzi, D., Maremonti, M. and Greco, G. (1994) Biotechnol. Bioeng. 44, 682-689 41 Rupley, J. A. and Careri, G. (1991) Adv. Protein. Chem. 41, 37-172 42 Ottolina, G., Bovara, A., Riva, S. and Carrea, G. (1994) Biotechn& Lett. 16, 923-928 43 Tewari, Y. B., Schantz, M. M., Pandey, P. C., Rekharsky, M. V. and Goldberg, P,,. N. (1995)J. Phys. Chem. 99, 1594-1601 44 Panintrarux, C., Adachi, S., Araki, Y., Kimura, Y. and Matsuno, R. (1995) Enzyme Microb. Technol. 17, 32-40 45 Kamat, S., Beckman, E. J. and Russell, A.J. (1992) Enzyme Microb. Technol. 14, 265-271 46 Roziewski, K. and Russell, A. J. (1992) Biotechnol. Bioeng. 39, 1171-1175 47 Persichetti, R. A., St Clair, N. L., Griffith, J. P., Navia, M. A. and Margolin, A. L. (1995)J. Am. Chem. Soc. 117, 2732-2737 48 Mozhaev, V. V., Pohevsky, K. J., Slepnev, V. I., Badun, G. A. and Levashov, A. V. (1991) FEBS Lett. 292, 159-161 49 Paradkar, V. M. and Dordick, J. 8. (1994)J. Am. Chem. Soc. 116, 5009-5010
TIBTECNNOVEMBER1995 (VOL 131

Organic-solvent-soluble enzymes
Attempts to solubilize proteins in various types of low-water solvent systems have attracted, and continue to attract, much interest: Mozhaev et al. have reported that it is possible to prepare catalytically active dispersions of unmodified enzymes in fairly hydrophobic organic solvents48; several groups have reported enzyme-surfactant complexes that dissolve in low-water organic solvents; and Paradkar et al. have shown that, in one such complex, the enzyme maintains its native structure and exhibits catalytic activity approaching that of an aqueous environment 49. Fully dispersed or 'soluble' enzymes can be useful for more fundamental work. In these cases, homogeneous solutions o f proteins in organic media are desirable or absolutely necessary, as is the case for some kinetic analysis and physical spectroscopic techniques.

Conclusion
It is now possible to make reliable predictions about some aspects of biocatalyst behaviour in NAM. This is particularly true for the effects o f water, including competition for water between the various components of the system. Once aw has been fixed, solvent choice can be based on Substrate solvation and likely effects on enzyme stability. Other topics, including solvent-protein interactions, and the effects o f ionization and counter-ions, require further fundamental study before the basis for process development is as sound as in aqueous enzymology.

Acknowledgement
We thank Biotechnology and Biological Sciences Research Council for financial support.

References
1 Vulson, E. N. (1993) Trends Food Sci. Technol. 4, 209-215 2 Tramper, J., Vemme, M. H., Beeftink, H. H. and von Stockar, U., eds (1992) Biocatalysis in Non-Conventional Media, Elsevier 3 Matsumae, H., Furui, M., 8hibatani, T. and Tosa, T. (1994) J. Ferment. Bioeng. 78, 59-63 4 Lamare, S. and Legoy, M. D. (1993) Trends Biotechnol. 11,413-418 5 Gill, I. and Vulfson, E. (1994) Trends Biotechnol. 12, 118-122 6 Kvittingen, L. (1994) Tetrahedron 50, 8253-8274 7 Halling, P.J. (1994) Enzyme Microb. Technol. 16, 178-206 8 Carrea, G., Ottolina, G. and Riva, S. (1995) Trends Biotechnol. 13, 63-70 9 Kamat, S. V., Beckman, E. J. and Russell, A. J. (1995) Cr/t. Rev. Biotechnol. 15, 41-71 10 Vermue, M. H. and Tramper, J. (1995) PureAppl. Chem. 67,345-373 11 Holmberg, K. (1994) Adv. Colloid Interface Sci. 51,137-174 12 Zaks, A. and Klibanov, A. M. (1988)J. Biol. Chem. 263, 8017-8021