Resampling-Iteration-Evolution (RIE): A Rational Design System for Engineering High-Affinity Peptides against Small Molecules.

Abstract

Peptides are attractive molecular recognition elements because of their compact size, ease of modification, and structural tunability. The design of high-affinity peptides for small molecules lags far behind that of antibodies and aptamers due to the lack of general and effective discovery and optimization strategies. Computational design offers a promising avenue to overcome this bottleneck. Herein, we report a computational design pipeline, termed resampling-iteration-evolution (RIE), for the rapid discovery of high-affinity molecular recognition peptides (MRPs) from large structural and sequence databases. RIE mimics a top-down evolutionary pocket-narrowing process by resampling native three-dimensional binding architectures, preserving pocket cooperativity, and introducing rational mutations to generate affinity-enhanced variants. Using this strategy, we designed MRPs for folic acid (FA), triiodothyronine (T3), and cholic acid (CHD), highlighting its applicability to structurally diverse and clinically important small molecules. Despite being derived from 130-270-residue protein receptors, the resulting 18-mer MRPs exhibited micromolar-level apparent Kd values of 8.0-12.7 μM in DMSO-based systems. By exploiting the small size and facile functionalization of MRPs, highly sensitive sensors for the rapid detection of small molecules were developed through flexible fluorophore modification. These results establish RIE as a rapid and effective in silico platform for the development of peptide binders for small-molecule targets.