Early-Stage Implementation of ω-Transaminase-Catalyzed Reactions

Biocatalysis is not a new phenomenon but has been big part of research for decades. However, the implementation of biocatalysis in pharmaceutical produc-tion is still lacking, particular for companies which are almost exclusively domi-nated by chemists. Currently, there is a gap between chemists and biochemists, and biocatalysis is often not found in the pharmaceutical development toolbox for such companies.

The aim of this thesis is to demystify biocatalysis for the chemistry community, with a thorough how-to-biocatalysis example with the ω-transaminase-catalyzed synthesis of the pharmaceutically relevant chiral amine (S)-1-(6-methoxypyridin-3-yl)ethanamine used as a case study.

Initially, a suitable ω-transaminase was chosen with a commercially available screening kit. Afterwards, process parameter development (temperature and pH) was commenced using a snap-shot screening method, which significantly de-creased the experimental burden required to get an indication of the best param-eters. Subsequently, the reaction thermodynamics and amine donor excess re-quired were estimated computationally by means of quantum chemistry. These conformed with the thermodynamics achieved experimentally, highlighting the usability of this method to further limit the experimental work required in the future.

Afterwards, real-time monitoring of the case study was investigated, given this is a crucial tool for industrial production. Here, Raman spectroscopy and one-dimensional neural network modelling was used in a batch setting to achieve good predictive capabilities, with R² values higher than 0.99 for the spectra in-cluded in the modelling and 0.964 for an independent dataset. However, the model had difficulties with extrapolation, and as such any given reaction should be set inside the concentrations used in the modelling.

Finally, a novel loop reactor for soluble enzyme was developed as an alterna-tive to the common packed-bed-reactor with immobilized enzyme. Here, a hol-low-fiber membrane was used for enzyme retention, with a residence time of 4 h. The longest run in the loop reactor was 72 hours corresponding to 18 cycles of 4 hours each. For this run, a decrease of only half the enzyme activity was seen after this extended period, with an overall specific yield three-times higher than for a single batch run. This system illustrates the potential use of soluble enzymes in flow.

It is believed that methods and tools presented in this thesis should be able to further increase the adoption of biocatalysis in pharmaceutical communities dom-inated by chemists. The hope is, that the availability of screening kits for phar-maceutically relevant reactions will be the first steppingstone in this direction.