Direct reprogramming of ARPC to NEPC with functional bypass of a dependence on AR signaling.

(A) Experimental schema of the reprogramming of ARPC cells to NEPC through the introduction of a candidate factor lentiviral (LV) pool and propagation in neural stem cell media (N-SCM) permissive to NEPC growth. (B) Immunoblot analysis of the C4-2B and LNCaP cell lines transduced with either green fluorescent protein (GFP) or the LV pool and subjected to the reprogramming assay. Lysates were collected on day 14. (C) Immunocytochemical analysis of the C4-2B and LNCaP lines at day 14. (D) Experimental schema for the stringent negative selection of AR+ cell populations based on the AR-dependent expression of an inducible FKBP-Casp8 fusion protein that dimerizes and induces caspase activation and apoptosis in the presence of the FK506 analog AP20817. (E) Relative cell viability determined by CellTiter-Glo assay of treated versus untreated groups after five days (n=4 each). P-value was assessed by Student’s t-test. **** denotes p <0.0001.

The pioneer neural transcription factor ASCL1 suppresses AR expression and drives NE transdifferentiation in prostate cancer.

Immunoblot analysis of (A) leave-one-out conditions and (B) factor reconstitution conditions in reprogramming studies using the C4-2B cell line. PRNB represents dominant-negative TP53 H175R, shRB1, MYCN, and BCL2. Lysates were collected on day 14. (C) Relative cell viability over time determined by CellTiter-Glo assay of C4-2B cells reprogrammed with various factor combinations (n=4 per condition). * denotes p <0.05, ** denotes p <0.01.

ASCL1 and NeuroD1 are competent to induce NE lineage reprogramming of prostate cancer.

(A) Immunoblot analysis of C4-2B and LNCaP cells reprogrammed with GFP or PRNB, SRRM4, and ASCL1 and/or NeuroD1. Lysates were collected on day 14. (B) Heatmap of RNA-seq gene expression from reprogrammed C4-2B cell line conditions showing 22-gene AR and NE signature scores and select genes associated with the AR program and NE programs (NEURO I and NEURO II). UQ: upper quartile normalization. (C) Partial least squares-discriminant analysis (PLS-DA) plot based on RNA-seq gene expression of reprogrammed C4-2B cell line conditions (black shapes) projected onto human ARPC (red dots) and NEPC (blue dots) samples from Beltran et al., 2016. Ellipses represent 95% confidence level for multivariate t-distributions defined by ARPC (red) and NEPC (blue) data.

Dynamic changes in cancer phenotype and transcriptional and epigenetic landscapes during acute reprogramming of ARPC to NEPC.

(A) Experimental schema for the temporal investigation of the reprogramming of ARPC to NEPC by immunoblot, RNA-seq, and CUT&RUN analyses. (B) Immunoblot analysis of C4-2B cells modified with PRNB and SRRM4 (PRNBS), PRNBS and NeuroD1 (PRNBSN), or PRNBS and ASCL1 (PRNBSA) that were collected at various timepoints during the 14-day reprogramming period. Heatmaps of RNA-seq gene expression from C4-2B cells reprogrammed with (C) PRNBSA or (D) PRNBSN over the 14-day reprogramming period showing 22-gene NE signature scores and select genes associated with the AR program and NE programs (NEURO I and NEURO II). (E) PLS-DA plot based on RNA-seq gene expression of C4-2B cells reprogrammed with PRNBSA (top) or PRNBSN (bottom) at different timepoints (black shapes) projected onto human ARPC (red dots) and NEPC (blue dots) samples from Beltran et al., 2016. Ellipses represent 95% confidence of t-distribution defined by ARPC (red) and NEPC (blue) data. (F) Principal component analysis (PCA) of CUT&RUN enrichment for H3K4me3, H3K4me1 and H3K27Ac signal in C4-2B cells reprogrammed with PRNBSA over time.

Time-dependent changes in ASCL1-regulated gene programs and super-enhancer organization during NE transdifferentiation of prostate cancer.

(A) Heatmap of MSigDB gene sets enriched over time with ASCL1-regulated genes identified by integrating CUT&RUN and RNA-seq data from C4-2B cells reprogrammed with PRNBSA over the 14-day reprogramming period. (B) Heatmap showing dynamic changes in super-enhancer regions in C4-2B cells reprogrammed with PRNBSA over the 14-day reprogramming period. Genomic regions shown were obtained by setting a p-value cutoff of 1e-03 on Kendall correlation between day and peak activity. RLE: Relative Log Expression normalization. (C) Plot showing Gene Ontology (GO) enrichment of Hallmark gene sets associated with decreasing super enhancer signals from D2 to D14 in C4-2B cells reprogrammed with PRNBSA.

ASCL1/NeuroD1 inhibit AR expression by remodeling chromatin accessibility at the somatically acquired AR enhancer and global AR binding sites with enhancer activity.

(A) Tracks showing chromatin accessibility peaks at the AR locus and AR gene expression (violin plot) from pseudo-bulk analysis of scATAC-seq and scRNA-seq data from C4-2B cells reprogrammed with PRNB, PRNBS, PRNBSA, and PRNBSN. Tracks in the region of the somatically acquired AR enhancer are magnified. (B) Immunofluorescence analysis of a human prostate cancer hepatic metastasis with ASCL1 staining (top), AR staining (middle), and an overlay of ASCL1 and AR staining (bottom) shown. DAPI was used as a nuclear counterstain. (C) Tracks showing chromatin accessibility at the AR locus by pseudo-bulk analysis of scATAC-seq data from the AR-/ASCL1+ and AR+/ASCL1- tumor cell populations present in the human prostate cancer hepatic metastasis shown in B. Tracks in the region of the somatically acquired AR enhancer are magnified. (D) Schematic representation of different classes of AR enhancers (top) is shown. The distribution of chromatin accessibility at AR bindings sites (ARBS) of different enhancer activity and at random genomic regions of the same interval are shown. Black dots represent the means. ns denotes p >0.05, ** denotes p <0.01, *** denotes p <0.001.

ASCL1/NeuroD1 induce the expression and activity of NE-associated transcription factors.

(A) Volcano plots showing the differential expression of genes encoding transcriptions factors in pairwise comparisons of RNA-seq gene expression data of C4-2B cells reprogrammed with PRNBS vs. PRNB, PRNBSA vs. PRNBS, and PRNBSN vs. PRNBS. Red dots represent genes with a −log10(FDR) > 2. FC represents fold-change. (B) Plots of differential transcription factor motif activity from footprinting analysis of pseudo-bulk scATAC-seq data from C4-2B cells reprogrammed with PRNBS vs. PRNB, PRNBSA vs. PRNBS, and PRNBSN vs. PRNBS. Note that labelled motifs in the second and third panel do not show up in the first panel. (C) PCA of chromatin accessibility signals from pseudo-bulk scATAC-seq of reprogrammed C4-2B cell line conditions projected on previously published ATAC-seq data from NEPC and ARPC PDX models. (D) Heatmap of RNA-seq gene expression data showing cell surface targets relevant to ARPC and NEPC in reprogrammed C4-2B cell conditions.

Downregulation of MHC class I antigen processing and presentation genes in NEPC is functionally attributable to ASCL1/NeuroD1.

(A) Heatmap of RNA-seq gene expression data showing the NE signature and MHC class I pathway scores (top) and select genes (bottom) in ARPC and NEPC from Beltran et al., 2016. (B) Heatmap of RNA-seq gene expression data showing the AR and NE signature scores and select genes in AR+/NE- and AR-/NE+ metastatic prostate cancers from the University of Washington Tissue Acquisition Necropsy (UW TAN) program. (C) Heatmap of RNA-seq gene expression data from reprogrammed C4-2B cell line conditions showing the NE signature and MHC class I pathway scores (top) and select MHC class I genes including B2M (bottom). UQ: upper quartile normalization. (D) UMAP analysis of scRNA-seq and scATAC-seq data from C4-2B cells reprogrammed with PRNB, PRNBS, PRNBSA or PRNBSN using the Seurat package. UMAP plots of reprogrammed cells colored based on (E) NEPC score and (F) B2M gene expression. (G) Flow cytometry histogram plots showing the percentage of B2M positive cells in reprogrammed C4-2B cells.