AI-Guided Discovery of Natural-Product-Inspired Scaffolds as Dual P-gp/BCRP Inhibitors to Overcome Cancer Multidrug Resistance.
TL;DR
The utility of AI-guided molecular design for overcoming MDR is demonstrated and a promising natural-product-inspired chemotype for dual P-gp/BCRP inhibitor is established.
Abstract
ATP-binding cassette (ABC) transporters ABCB1/P-glycoprotein (P-gp) and ABCG2/breast cancer resistance protein (BCRP) drive multidrug resistance (MDR) by limiting intracellular chemotherapy accumulation, and are coexpressed in cancers with overlapping substrates. Here, we combine marine natural product fragment mining, artificial intelligence (AI)-based molecular generation, and structure-guided prioritization to design a natural-product-inspired library. Two optimized analogs, Ib18 and It12, reversed P-gp/BCRP-mediated resistance in overexpressing cells, outperforming reference inhibitors. Mechanistic studies confirmed target engagement without expression downregulation. Docking and molecular dynamics simulations provided the structural rationales for dual transporter engagement. Transcriptomics showed Ib18 avoided stress responses triggered by reference inhibitors, supporting an expression-independent MDR-sensitizing mechanism with limited cytotoxicity. In xenografts, Ib18 enhanced the antitumor activity and tumor accumulation of mitoxantrone without detectably increasing systemic toxicity. Together, these findings demonstrate the utility of AI-guided molecular design for overcoming MDR and establish a promising natural-product-inspired chemotype for dual P-gp/BCRP inhibitor.