Over the last decade, enzymes have attracted much attention as they are efficient and extremely specific catalysts in many synthetic chemical applications. Baeyer-Villiger-monooxygenases (BVMOs) represent a notable example of a group of enzymes that have emerged as powerful biocatalysts [1
]. BVMOs incorporate one atom of molecular oxygen into a carbon-carbon bond of an organic substrate next to a carbonyl group while the other oxygen atom is reduced to water. Most characterized BVMOs are NADPH-dependent flavoproteins and belong to a sequence-related family, called Type I BVMOs [2
]. Phenylacetone monooxygenase (PAMO) from Thermobifida fusca
represents a prototype Type I BVMO, and its characterization by us showed that it is a soluble, monomeric protein of about 65 kDa and is well expressed in Escherichia coli
]. Substrate profiling revealed that it is mainly active towards small aromatic ketones and sulfides [5
]. However, PAMO is also able to convert larger substrates, albeit with a poor activity and selectivity [8
]. In addition, PAMO is remarkably thermostable and tolerant towards organic solvents [5
]. The determination of its atomic structure showed that PAMO comprises two domains; an FAD and NADPH-binding domain with the active site sandwiched in between at the domain interface [11
]. Moreover, a recent study, using complementary biochemical and structural experiments, revealed that PAMO and related enzymes function mainly as oxygen-activating enzymes. These can react with any appropriate substrate that is able to reach the catalytic center within the active site [12
]. The detailed structural and mechanistic understanding of PAMO as well as its remarkable stability make this enzyme an attractive target for potential biocatalytic applications.
The reproducible expression of BVMOs and other biotechnologically relevant enzymes has become a pressing matter. Not only because of their growing use in a variety applications, but also in the design of novel screening methods for directed-evolution experiments to identify and isolate novel enzyme variants with the desired properties. Common strategies to optimize this typically rely on small scale reactions, using either purified enzyme, or whole cells expressing the enzyme of interest. Various studies on cyclohexanone monooxygenase (CHMO), a well-characterized BVMO from Acinetobacter
demonstrate that whole cell biocatalytic systems are particularly well-suited for this purpose. Different whole cell biocatalytic systems, using Saccharomyces cerevisiae
or E. coli
, have been employed successfully to investigate and improve critical parameters for its expression as well as conditions for CHMO-catalyzed biotransformations [13
]. Specifically, these systems were used either in microscale or bench-scale reactions for substrate profiling, analysis of substrate or product inhibition, comparison of different expression hosts, assessment of biocatalyst stability, analysis of oxygen supply, investigation of substrate uptake, quantification of kinetic data, and the detailed analysis of different microwell formats [15
]. Combined, these studies emphasize the importance of a robust host organism in combination with a powerful expression system, and highlight the relevance of different factors governing the expression of the target enzyme, such as expression temperature, time and period of induction. Furthermore, they provide insight into conditions that control the efficiency of biotransformation, including the source of reducing power for in vivo
coenzyme regeneration as well as substrate and product inhibition.
Although valuable, the overall picture provided by these studies is blurred because of the variety of host organisms, different expression systems, various model substrates and differing reaction conditions employed in several studies for the same biocatalyst. To provide a clear picture on this issue, we present in this study a rational and systematic approach to optimize the expression of a biocatalyst in a reproducible fashion. To this end, we have used PAMO as a model BVMO and followed a stepwise strategy to improve the biotransformation performance of recombinant E. coli expressing PAMO. Using a microscale approach, the best expression conditions for PAMO were investigated first, including different host strains, temperature as well as time and induction period for PAMO expression. Next, this optimized system was used to improve conditions of the biotransformation step, the PAMO-catalyzed conversion of phenylacetone, by evaluating the best electron donor, substrate concentration, and the temperature and length of biotransformation. This resulted in an efficient and highly reproducible PAMO whole cell biocatalyst and, moreover, the optimized procedure was successfully adapted for mutant screening. The strategy presented in this study provides a valuable tool for the reproducible optimization of bioconversions and in the design of novel activity-based screening procedures suitable for BVMOs and probably other NAD(P)H-dependent enzymes as well.