Data-Driven Design of Random Heteropolypeptides as Synthetic Polyclonal Antibodies

Abstract

Antibodies are indispensable in biomedicine. However, conventional antibody development faces challenges, including high costs and lengthy production timelines spanning months. Here, we present a data-driven workflow for engineering random heteropolypeptides (RHPs) as synthetic polyclonal antibodies (SpAbs) with programmable binding properties. By combining high-throughput synthesis of selenopolypeptide derivatives with algorithm-assisted optimization, we rapidly identified SpAbs targeting human interferon-α (IFN) and tumor necrosis factor-α (TNF-α) within 2 weeks. The SpAbs exhibited binding affinities comparable to natural antibodies, with the top candidate achieving a dissociation constant of 7.9 nM for TNF-α and 418-fold selectivity over human serum albumin, effectively neutralizing TNF-α-induced cytotoxicity. Liquid-phase electron microscopy revealed flexible, intrinsically disordered protein-like conformations and folding-upon-binding dynamics. This study establishes a robust framework for SpAb discovery, demonstrating that sequence-independent RHPs can serve as functional antibody mimics with tunable binding properties, rapid optimization, and broad potential in diagnostics and therapeutics.