Drug-like molecules targeting androgen receptor's allosteric binding sites selected by augmented AI and high-throughput screening as antitumor agents against prostate cancer.

Despite therapeutic advancements, metastatic and nonmetastatic castration-resistant prostate cancer (mCRPC and nmCRPC) remain incurable due to drug resistance, partly due to Androgen receptor (AR) gene abnormalities and splice variants. An Augmented Artificial Intelligence (AI) -driven virtual drug screening using AtomNet® technology explored an 8 million small molecule library targeting AR allosteric sites to address resistance from AR mutations and AR-V7 splice variant. Screening identified compounds effectively reducing cancer cell growth, with ten actives in hormone-resistant cells. Dose-response assays identified compounds 1, 8, and 77 as most efficacious, exhibiting pIC values from < 5.86-4.82. Docking and molecular dynamics (MD) simulations indicated that compound 77 targets both the allosteric site within the AF2 domain and the ligand binding domain of AR. Compounds 1, 8, and 77 were most active in 2D and 3D spheroid assay, with compound 77 showing significant activity in AR+ cells and modest effects in AR-null cells. These compounds surpassed treatments like enzalutamide and galeterone in inhibiting colony and sphere formation, indicating new therapeutic potential. These effects, partly due to AR inhibition, suggest additional mechanisms in inhibiting cancer growth and disrupting models, as seen in PC3 cells. Moreover, the docking analysis indicated that compounds 1, 8, and 77 exhibit binding affinity with the Glucocorticoid receptor (GR). Furthermore, immunofluorescence assay on PC3 cells, demonstrated that compound 77 inhibits the nuclear translocation of GR following dexamethasone treatment, which is used to induce GR nuclear translocation. Furthermore, these findings revealed that compound 77 significantly reduces the transcription of GR and its downstream gene FKBP5, a classical GR target gene. These results substantiate our hypothesis that compound 77, particularly engages in off-target interactions, thereby potentially disrupting cancer growth.