Asymmetric reduction of ketones and β-keto esters by (S)-1-phenylethanol dehydrogenase from denitrifying bacterium Aromatoleum aromaticum

Enzyme-catalyzed enantioselective reductions of ketones and keto esters have become popular for the production of homochiral building blocks which are valuable synthons for the preparation of biologically active compounds at industrial scale. Among many kinds of biocatalysts, dehydrogenases/reductases from various microorganisms have been used to prepare optically pure enantiomers from carbonyl compounds. (S)-1-phenylethanol dehydrogenase (PEDH) was found in the denitrifying bacterium Aromatoleum aromaticum (strain EbN1) and belongs to the short-chain dehydrogenase/reductase family. It catalyzes the stereospecific oxidation of (S)-1-phenylethanol to acetophenone during anaerobic ethylbenzene mineralization, but also the reverse reaction, i.e., NADH-dependent enantioselective reduction of acetophenone to (S)-1-phenylethanol. In this work, we present the application of PEDH for asymmetric reduction of 42 prochiral ketones and 11 β-keto esters to enantiopure secondary alcohols. The high enantioselectivity of the reaction is explained by docking experiments and analysis of the interaction and binding energies of the theoretical enzyme-substrate complexes leading to the respective (S)- or (R)-alcohols. The conversions were carried out in a batch reactor using Escherichia coli cells with heterologously produced PEDH as whole-cell catalysts and isopropanol as reaction solvent and cosubstrate for NADH recovery. Ketones were converted to the respective secondary alcohols with excellent enantiomeric excesses and high productivities. Moreover, the progress of product formation was studied for nine para-substituted acetophenone derivatives and described by neural network models, which allow to predict reactor behavior and provides insight on enzyme reactivity. Finally, equilibrium constants for conversion of these substrates were derived from the progress curves of the reactions. The obtained values matched very well with theoretical predictions.