An Integrated Approach Combining Receptor-Based Ensemble Screening and In Vivo Phenotypic Assessment Improves Drug-Induced Liver Injury Prediction

Inhibition of ABCB11, also called the Bile Salt Export Pump (BSEP), is closely linked to cholestatic drug-induced liver injury (DILI). ABCB11 undergoes large-scale conformational changes during its catalytic turnover, making it challenging to incorporate this flexibility into large-scale structure-based screening. Here, we report an ensemble docking-based drug screening framework that incorporates multiple functional conformations of ABCB11, derived from its free-energy landscape during the transport cycle, to improve the annotation accuracy for most DILI-concern FDA-approved drugs in the DILIrank 1.0 dataset, outperforming single-receptor docking approaches. We have then integrated the ensemble-screening framework with cellular transporter assays and in vivo zebrafish phenotypic screens to stratify 67 tyrosine kinase inhibitors (TKIs). The developed computational-experimental framework predicts significant DILI-risk associations for sunitinib, sorafenib, regorafenib, and tivantinib, while tofacitinib is associated with the least DILI-risk propensity. The integrated computational-experimental pipeline ensures robust DILI-risk prediction, significantly reducing the risk of false positives.