Abstract 4601: Rational design of potent androgen receptor degraders via AI-assisted molecular modeling and validation of anti-tumor efficacy in prostate cancer

Abstract The androgen receptor (AR) plays a central role in the progression of prostate cancer, and its sustained signaling remains a major challenge in castration-resistant prostate cancer (CRPC). Targeted protein degradation of the AR offers a promising therapeutic strategy to overcome therapeutic resistance that limits the efficacy of current AR antagonists. To rationally design novel AR degraders, we employed AnHorn’s proprietary AIMCADD (Artificial Intelligence-bio-Informatic-MedChem-Computer-Aided Drug Design) platform to predict the three-dimensional binding conformations between the AR ligand-binding domain (AR-LBD) and CRBN E3 ligase. Molecular dynamics (MD) simulations were used to calculate binding free energies and identify the most stable AR-CRBN ternary complex. Based on this complex, the binding conformations of AR-LBD and CRBN warheads were obtained via molecular docking, followed by AI-assisted linker generation using AIMLinker, a deep neural network-based algorithm, to construct a virtual library of degrader candidates. Molecular docking and dynamic simulations were further applied to evaluate ternary complex stability and binding affinity, from which a subset of top-ranked candidates were selected for experimental validation. Lead candidates effectively induced AR degradation in LNCaP, 22Rv1, and VCaP cells. Notably, these degraders also retained activity against clinically relevant AR mutants frequently observed in CRPC patients, including L702H, T878A, H875Y, W742C, and F877L. The degradation was abolished by co-treatment with the proteasome inhibitor MG132, confirming a ubiquitin-proteasome-dependent mechanism. Direct binding of the lead compound to AR was further verified by drug affinity responsive target stability (DARTS) assay. Quantitative RT-PCR analysis revealed a marked downregulation of AR downstream targets, KLK3 (PSA), suggesting the loss of AR transcriptional activity. In cell viability assays, the most potent compound selectively suppressed proliferation of AR-positive prostate cancer cells, with minimal cytotoxicity in normal cells. In xenograft animal models, once daily administration significantly reduced tumor volume and prostate specific antigen (PSA) level in plasma without observable systemic toxicity. In summary, through AI-assisted molecular design and computational screening, we developed a potent AR degrader that efficiently induces proteasome-dependent AR degradation, suppresses AR transcriptional activity, and exhibits robust anti-tumor efficacy in vivo. These results highlight a promising therapeutic approach for advanced prostate cancer through targeted AR degradation. Citation Format: Chih-Chang Chou,You-Sheng Lin,Cheng-Li Chou,Kuan-Hung Chen,Shu-Jen Chen,Chu-Chiang Lin. Rational design of potent androgen receptor degraders via AI-assisted molecular modeling and validation of anti-tumor efficacy in prostate cancer abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4601.