Hybrid Rational Design and Artificial Intelligence Model-driven Enhancement of ω-Transaminase Catalytic Activity for the Production of 2-Amino-4-hydroxybutyric Acid.

Transaminases (TAs) play a crucial role in asymmetric synthesis, among which ω-TA exhibits a broad substrate spectrum and high enantioselectivity, albeit with limitations such as poor stability and low enzymatic activity. This study aimed to improve the catalytic efficiency of the rate-limiting transaminase in the cascade reaction for 2-amino-4-hydroxybutyric acid, using strategies such as semirational design and DLKcat large model prediction. After single-point and combinatorial mutations, ultimately, a combinatorial mutant, S62T/T68D/Y221Q (VfTA-M3), with a relative specific activity of 379.42% compared to the wild-type (VfTA-WT), was obtained, which also exhibited improved stability relative to the wild-type enzyme. Using this mutant in conjunction with aldolase RtALD, a cascade catalytic system was established for the synthesis of 2-amino-4-hydroxybutyric acid from inexpensive formaldehyde and pyruvate. After 20 h of reaction, the substrate conversion rate reached 80%, laying a solid foundation for the production of other chiral amino acids and chiral amines.