Ping Liu; Refeya Jannatul; Juan Chen; Lihua Hou; Mingjuan Gao; Pengjie Wang; Lulu Wang; Dekui Jin; Hao Chen; Rong Liu; Ran Wang; Yinhua Zhu; Fang Bing; Lirong Jia; Yanan Sun; Yixin Zhang; Fazheng Ren; Weilin Lin
Angew. Chem. Int. Ed. 2024, e202409746
https://doi.org/10.1002/anie.202409746

Abstract

Non-natural building blocks (BBs) present a vast reservoir of chemical diversity for molecular recognition and drug discovery. However, leveraging evolutionary principles to efficiently generate bioactive molecules with a larger number of diverse BBs poses challenges within current laboratory evolution systems. Here, we introduce programmable chemical evolution (PCEvo) by integrating chemoinformatic classification and high-throughput array synthesis/screening. PCEvo initiates evolution by constructing a diversely combinatorial library to create ancestral molecules, streamlines the molecular evolution process and identifies high-affinity binders within 2-4 cycles. By employing PCEvo with 108 BBs and exploring >10^17 chemical space, we identify bicyclic peptidomimetic binders against targets SAR-CoV-2 RBD and Claudin18.2, achieving nanomolar affinity. Remarkably, Claudin18.2 binders selectively stain gastric adenocarcinoma cell lines and patient samples. PCEvo achieves expedited evolution in a few rounds, marking a significant advance in utilizing non-natural building blocks for rapid chemical evolution applicable to targets with or without prior structural information and ligand preference.