The enantioselective reduction of tetrahydrothiophene-3-one by an alcohol dehydrogenase (ADH) was targeted as it yields an important API (active pharmaceutical ingredient) building block. A set of mutant libraries was generated based on random mutagenesis, random deletions, and semi-rational simultaneous mutagenesis of 3-6 amino acids to study different evolutionary pathways. The best variants were identified using a previously established assay upon addition of required adjustments and validating control experiments. Testing of 2*106 mutants resulted in 3.000 promising hits, which were further narrowed down using conventional screening techniques to give 200 final hits with improved catalytic properties compared to the wild-type (WT) under the applied assay conditions. We were pleased to find that 20% of those hits also showed improvement under industrial reaction conditions. Our results highlight that – regardless of compromises required for uHTP applications – the used technique allows identification of industrially relevant enzymes. Fortunately, the semi-rationally designed mutants and those obtained through random mutagenesis showed comparable improvements, which makes our approach broadly applicable to different goals. A random combination of identified beneficial mutations gave a second set of mutant libraries, which are currently under investigation. We expect this experiment to give a highly improved biocatalyst after only 2 rounds of evolution; compared to traditional strategies which require up to 20 rounds. In collaboration with a PhD student in the Hollfelder lab, the crystal structure of the wild-type enzyme was solved and analytics for a detailed characterization of the enzyme kinetics have been established; both will be applied to explain the impact of identified mutations. The obtained data provides a unique basis to investigate epistatic interactions and their – usually neglected - importance within enzyme engineering campaigns. Remaining experiments to finish the project are currently being performed under my supervision.
The initially proposed project turned out to be a ‘text book’ example for successful project management, which allowed time to go a step further and investigate how the field of functional metagenomics could benefit from microfluidic assays. In line with the main objective, our efforts focused at the identification of novel ADHs. Regardless of the initially highlighted potential of colorimetric assay for this purpose, the direct transfer of known applications into the microfluidic format is hampered by the unique system requirements. We have identified a highly water-soluble fluorescent dye as a suitable label for fluorogenic substrates for droplet-based applications. Constituting an example, a set of substrates was synthesized and can be expanded to many different enzyme classes in the future. Assay evaluation was performed using a positive control obtained from the JM enzyme collection. Additional studies indicated that the assay can give access to enzymes from different enzyme families without sequence bias (only 20% sequence identity between hits within the JM enzyme collection) and thus, is likely to allow identification of biocatalysts within new sequence space. To our satisfaction, the generation and screening of a metagenomic soil library (sample taken just outside the JM laboratories in Cambridge) resulted in the identification of a new ADH. Most similar proteins in the database show around 70% sequence identity and are mainly annotated as amidases with a minor amount expected to be oxidoreductases, which potentially suggests a novel enzyme family or reaction mechanism. More detailed investigations are required and will be performed by the Hollfelder group. Further efforts to generate additional (meta)genomic libraries to identify further biocatalysts are ongoing.
Research is still ongoing to raise the impact of the envisaged publications; however, we believe that the obtained results will lead to ≥2 publications and reveal general principles helping microfluidics assays to flourish in the future. In addition, this work significantly extends targetable goals using microfluidics, which will be highlighted in a review providing a complete overview of biacatlyst-related droplet screens to date.
