Accelerated de novo design of xanthine oxidase-inhibitory peptides via physics-informed computational approaches

Xanthine oxidase (XO) dysregulation causes hyperuricemia and gout, but current synthetic inhibitors are limited by off-target effects and associated risks. Although peptide inhibitors exhibit inherent advantages in target specificity, their discovery remains constrained by the low throughput and inefficiency of traditional experimental approaches. In this study, we developed a computational framework for the de novo design of enzyme-inhibitory peptides. Our approach combines backbone design based on tertiary motifs (TERMs) with ProteinMPNN to generate candidate peptides, and employs a hybrid screening process that combines computational models with physics-informed methods and molecular dynamics simulations. This strategy successfully identified the candidate peptide 1121_4 exhibiting a twofold reduction in IC₅₀ value compared to previously reported XO peptide inhibitors. Our results demonstrate the feasibility of designing enzyme-inhibitory peptides based on pocket structure to obstruct substrate entry, thereby presenting a novel strategy for the efficient design of XO inhibitors.