Fig. 2: Epiregulon predicts the responses of AR-modulating drugs.

a Mechanisms of action of 3 AR-modulating agents. b Six prostate cancer cell lines were treated and profiled for changes in their gene expression and chromatin accessibility by paired scATAC-seq and scRNAseq. Shown is the UMAP representation (5028 VCaP cells, 5958 LNCaP cells, 3568 22Rv1 cells, 945 MDA-PCa-2b cells, 3639 NCI-H660 cells and 3980 DU145 cells). Cells were merged from 2 technical replicates. Cells were treated for 24 h at 1 µM of enzalutamide or ARV-110 or 0.1 µM of SMARCA2_4.1. c Immunoblotting of AR and HDAC as a loading control after 24 h of treatment. This is a representative result from 2 biological replicates. d Chemical structure of SMARCA2_4.1. e Prostate cancer cell lines were treated for 5 days and cell viability was measured by CellTiter-Glo. Dotted line indicates the concentrations used in the scATAC-seq/scRNA-seq experiment. Data are presented as mean ± standard deviation (s.d.) based on 4 biological replicates. f Shown are the AR gene expression and the AR activity computed by Epiregulon (co-occurrence weight estimation method). Numbers of cells are as follows: VCaP (DMSO 1392 cells, Enza 1266 cells, ARV-110 1377 cells, SMARCA2_4.1 993 cells); LNCaP (DMSO 1499 cells, Enza 1966 cells, ARV-110 758 cells, SMARCA2_4.1 1735 cells); 22Rv1 (DMSO 970 cells, Enza 1002 cells, ARV-110 554 cells, SMARCA2_4.1 1042 cells); MDA-PCa-2b (DMSO 306 cells, Enza 76 cells, ARV-110 188 cells, SMARCA2_4.1 375 cells). Boxplots presented as median values ± 25%. Lower whisker is the smallest observation ≥25% quantile −1.5 × interquantile range (IQR). Upper whisker represents the largest observation ≤75% + 1.5 × IQR. g Each cell was identified by the HTO tag corresponding to treatment. A cell was classified into either the DMSO or AR inhibitor-treated group based on AR activity. Bar plots show the median receiver operating characteristic curve (AUROC) in the two sensitive cell lines, LNCaP and VCaP, treated with enzalutamide or ARV-110 for a total of 4 samples. Source data are provided as a Source Data file. Created in BioRender. Yao, X. (2025) https://BioRender.com/x50fdft.