(b) Solubility of product. Theoretically it is possible to calculate that for systems of up to
two substrates and two products, 25 combinations of preferential solubility exist with
only four all aqueous and four all organic. This leaves 17 cases where two-liquid
phase biocatalysis is appropriate. However the preferential distribution of substrate(s)
and product(s) between the phases is difficult to manipulate. In many cases at least one
substrate and the desired product are both preferentially distributed to the organic
phase. Where the distribution is favourable it can be exploited downstream. For
example the hydroxylation of aromatics by Pseudomonas putida gives a product of
many fold higher aqueous solubility than the substrate to afford an easy product
recovery by phase separation (Collins and Woodley, 1993). Even in such cases the
poorly-soluble compound will be present in trace quantities in the product stream and
will need to be removed.
ORGANIC PHASE
Organic Phase SelectionEnzymes and whole-cell catalysts usually operate in an aqueous environment. Hence
organic solvents (whether dissolved in an aqueous phase or present as a discrete second
liquid phase) have long been known to be harmful to biocatalytic activity. The search for
an indicator characteristic of different organic solvents which could be used to predict
biocatalyst tolerance has led to various proposals. One of the most comprehensive early
studies examined the relationship between solvent molecular weight and the Hildebrand
solubility parameter with the maintenance of biocatalytic activity (Brink and Tramper,
1985). More recently (Laane et al., 1987) it was suggested that solvents with Log P
values (logarithm of the partition coefficient of an organic solvent in a standard octanol-
water system) greater than 4 were most suitable, while use of those with lower Log P
values, especially less than 2, resulted in considerable loss of activity (Bruce and
Daugulis, 1991).
However, further studies in our laboratory have revealed that while the trend of
increased activity with increased Log P value appears universal, the solvent Log P-
biocatalyst activity profile is specific to a particular biocatalyst and the extent of its
contact with the organic phase. For example differences have been observed between
two-liquid phase biotransformations carried out in shaken-flasks and stirred reactors and
between the use of Gram-positive (Arthrobacter simplex) and Gram-negative
(Pseudomonas putida) bacteria as catalysts as shown in Figure 5.2 (Harrop et al., 1992).
While immobilisation may protect Gram-positive bacteria from solvent damage this has
not been conclusively observed for Gram-negative bacteria. Such observations underline
the empirical nature of these rules and provide guidance for the further research required
to understand the way in which different organic solvents damage different types of
biocatalyst. These studies will almost certainly require examination of the precise role of
the frequency and duration of interfacial contact with the biocatalyst.
Advances in the selection and design of two-liquid phase biocatalytic reactors 127