Random Hybrid Polypeptides Regulate Immune Signaling through Promiscuous and Dynamic Interaction Landscapes

Abstract We developed a high-throughput synthesis and screening workflow combined with machine-learning-assisted optimization to identify functional compositions for immunomodulation in random hybrid polypeptides (RHPs). Using a selenium-containing polypeptide platform, we constructed a nucleoside-functionalized RHP library incorporating nucleoside, cationic, and anionic interaction motifs. Machine-learning-assisted optimization of this library identified a lead RHP, termed IDP-IMP (intrinsically disordered protein-inspired immunomodulatory polypeptide), that suppresses CpG DNA–Toll-like receptor 9 (TLR9) signaling with limited cytotoxicity. Rather than acting as a conventional strong nucleic acid scavenger or TLR9 antagonist, IDP-IMP showed moderate binding/association with both CpG and TLR9 through different side chain compositions, leading to reduced intracellular CpG–TLR9 colocalization. Across the polymer library, stronger CpG binding did not necessarily correspond to greater inhibitory activity, indicating that CpG binding alone did not predict functional activity. In a collagen-induced arthritis rat model, IDP-IMP attenuated inflammation and joint pathology without detectable systemic toxicity. These findings demonstrate a high-throughput approach for identifying statistical polymer compositions perturbing CpG–TLR9 signaling. The results highlight the importance of IDP-like weak, multivalent, and dynamic interactions in immunomodulation.