摘要
Protein–polymer conjugates are typically limited to polyethylene glycol derivatives, which restricts functional innovation. Random heteropolymers (RHPs) can mimic molecular chaperones but have only been physically mixed with proteins, leaving covalent conjugation unexplored. Here, we report the first synthesis of protein–RHP conjugates using an aqueous, oxygen-tolerant polymerization technique combined with a BayesianKAN machine learning framework. With glucose oxidase as a model enzyme, we constructed a conjugate library and, within four iterative rounds requiring one-tenth of the conventional sample size, identified an optimized conjugate exhibiting a 59.52% enhancement in thermal stability. Mechanistic characterization shows that RHP conjugation preserves the native protein structure and provides chaperone-like protection against thermal unfolding and aggregation. This work establishes a broadly applicable, data-efficient platform for discovering functional protein–polymer conjugates, with significant implications for biocatalysis and biopharmaceutical development.
危险品化学品经营许可证(带存储)