Sequence Similarity Network Guided Discovery of a Dehydrogenase for Asymmetric Carbonyl Dehydrogenation

Carbonyl dehydrogenation is one of the most valuable transformations in modern synthetic chemistry. As compared to traditional chemical synthesis methods, enzymatic dehydrogenation offers a greener and more selective alternative. However, except for a few rare natural dehydrogenases for desaturation, current enzymatic methods predominantly rely on enzyme promiscuity, which often suffers from lower efficiency and limited reaction controllability. Herein, we employed sequence similarity networks to mine natural dehydrogenases from a vast array of sequences with potential dehydrogenation activity. This approach led to the discovery of an uncharacterized FAD-dependent enzyme capable of efficiently performing the desymmetrizing desaturation of cyclohexanones, thereby generating diverse cyclohexenones bearing remote γ-quaternary stereocenters. The current method has enhanced the turnover frequency (TOF) by approximately 178-fold as compared to the best existing biocatalytic strategies and displayed almost no overoxidation reactions. Through a combination of experimental assays and computational studies, we elucidated that this enzyme enhances its dehydrogenation capability through an unconventional proton relay system, absent in previously reported enzyme promiscuity systems. Additionally, this streamlined enzymatic process demonstrated scalability to gram-scale synthesis with maintained efficiency and selectivity, offering robust and sustainable alternatives for the synthesis of chiral cyclohexenones with high optical purity.