(63a) Computational Protein Design Enables Efficient Regeneration of a Biomimetic Cofactor to Support Diverse Redox Chemistries

Enzymatic biotransformation has been regarded as a feasible solution to manufacture chiral chemicals in an affordable and environmentally friendly manner. In such processes, natural redox cofactors NAD(P)H are regenerated in situ by coupled enzymatic reactions. Recent studies have explored opportunities to replace the expensive and unstable natural redox cofactors with their simpler analogs. However, the widespread application of these biomimetics is impeded by the difficulty in engineering oxidoreductases, especially the commonly used redox cofactor regeneration enzymes, to accept them efficiently. Here we report the successful engineering of the Bacillus subtilis glucose dehydrogenase (Bs GDH) to utilize a simpler biomimetic redox cofactor nicotinamide mononucleotide (NMN+) ~1000-fold more efficiently than the wild type. This was enabled by computationally redesigning the enzyme-cofactor interaction network to improve the shape and electrostatic complementary of the binding pocket towards NMN+. Coupled with an enoate reductase, the engineered Bs GDH showed high activity (~0.10 s-1 initial turnover frequency) and robustness (TTN ~39,000), which are within the range of what's generally accepted for industrial catalysts. We have further demonstrated the compatibility of the NMN(H) cycling system to a diverse range of biotransformation chemistries involving reduction of activated C=C double bonds, C≡C triple bonds, and nitro groups, as well as supplying electrons to cytochrome P450. Overall, these results suggest that the biomimetic cofactors may have potential as viable alternatives in enzymatic biotransformation. The design principles discovered here may shed light on future protein engineering efforts to expand the application of non-natural redox cofactors.