Abstract

Genistein, a bioactive isoflavone with therapeutic potential in treating oxidative stress, cardiovascular diseases, and cancer, faces production limitations during microbial biosynthesis due to product feedback inhibition of isoflavone synthase (IFS). To overcome this bottleneck, a biosensor-assisted continuous directed evolution platform was developed. Specifically, a genistein-specific biosensor was engineered using the Bradyrhizobium japonicum transcription factor FrrA, capable of distinguishing genistein from its precursor, (2 S )-naringenin. The biosensor was systematically optimized to eliminate background fluorescence at inhibitory genistein concentrations while maintaining strong responses to elevated product levels, enabling precise detection during high-throughput screening. By coupling this biosensor with droplet microfluidic sorting, a high-quality mutant library of Trifolium pratense IFS (TpIFS) generated through deaminase-T7 RNA polymerase fusion-mediated continuous evolution, was screened. This approach successfully identified TpIFS M6 , a feedback-resistant mutant exhibiting a 6.6-fold increase in the product inhibition constant and a 3.8-fold reduced binding affinity for genistein. Consequently, the genistein yield obtained with TpIFS M6 was 3.1 times that of the wild-type. Molecular dynamics simulations revealed that mutations I187R and F303A prevented inhibitor-induced conformational displacement of the catalytic I-helix. This work establishes a generalizable high-throughput screening strategy to mitigate enzyme feedback inhibition, facilitating the robust biosynthesis of plant-derived natural products.