The MYC axis in advanced prostate cancer is impacted through concurrent targeting of ERβ and AR using a novel ERβ-selective ligand alongside Enzalutamide

Abstract

We have dissected the role of Estrogen receptor beta (ERβ) in prostate cancer (PCa) with a novel ERβ ligand, OSU-ERb-12. Drug screens revealed additive interactions between OSU-ERB-12 and either epigenetic inhibitors or the androgen receptor antagonist, Enzalutamide (Enza). Clonogenic and cell biolody studies supported the potent additive effects of OSU-ERB-12 (100nM) and Enza (1μM). The cooperative behavior was in PCa cell lines treated with either OSU-ERB-12 plus Enza or combinations involving 17β-estradiol (E2). OSU-ERb-12 plus Enza uniquely impacted the transcriptiome, accessible chromatin, and the AR, MYC and H3K27ac cistromes. This included skewed transcriptional responses including suppression of the androgen and MYC transcriptomes, and repressed MYC protein. OSU-ERb-12 plus Enza uniquely impacted chromatin accessibility at approximately 3000 nucleosome-free sites, enriched at enhancers, enriched for basic Helix-Loop-Helix motifs. CUT&RUN experiments revealed combination treatment targeting of MYC, AR, and H3K27ac again shaping enhancer accessibility. Specifically, it repressed MYC interactions at enhancer regions enriched for bHLH motifs, and overlapped with publicly-available bHLH cistromes. Finally, cistrome-transcriptome analyses identified ~200 genes that distinguished advanced PCa tumors in the SU2C cohort with high androgen and low neuroendocrine scores.

Figure 1.. Cooperative actions of OSU-ERb-12 with a range of drugs in LNCaP, PC-3, DU 145 and 22Rv1 cells.

A. Cells in quadruplicates were treated with a range of OSU-ERb-12 concentrations and a panel of other drugs (Supplementary Table 1) also at a range of doses. Inhibition of proliferation was measured by measuring cellular ATP. The observed combination of proliferation (Inhibition %) was plotted against the difference (delta) to the predicted inhibition; plot for 22Rv1 cells is shown B. The difference between the observed and predicted inhibition was measured with a t-test and the p values corrected for multiple testing, log10 transformed and visualized by heatmap. Representative plots are shown for nuclear receptor antagonists.

Figure 2.. Inhibition of colony formation by OSU-ERb-12 combinations and regulation of differentiation markers in 22Rv1 cells.

A. Representative images of 22Rv1 colonies treated with OSU-ERb-12 (100 nM) with either Enza (1 μM) or FK228 (1nM) or the combinations with quantification of colony number. B. Western immunoblot measurements of AR, CK5 and CK8 after treatment with OSU-ERb-12 plus Enza (time 7 days) with quantification (C).

Figure 3.. OSU-ERb-12 plus Enza induces a unique transcriptome associated with MYC repression in LNCaP, LNCaP C4-2 and 22 Rv1 cells.

A. LNCaP, LNCaP C4-2 and 22Rv1 cells in triplicate were treated with either vehicle, OSU-ERb-12 alone (100 nM) or E2 alone (***), Enza (1 μM) or the combinations for 12h and RNA-Seq was undertaken. FASTQ files were QC processed, aligned to hg38 (Rsubread), and processed with edgeR workflow to identify significant differentially enriched genes (DEGs) (logPV > 1 & absFC > .26) of the drug combination compared to the combination of the average of the individual treatments, and are illustrated by Volcano plots with top DEGs labelled. B. GSEA analyses (Hallmarks, and Chemical and Genetic Perturbations) was undertaken and enrichment terms were word frequency analyzed which identified that terms including Breast, Prostate, Nuclear Receptors and MYC were amongst the most frequent, and these terms were filtered for conditions where the NES was in the opposite direction, and the delta calculated, with top 40 NES terms illustrated. C. NES plots for the indicated treatments with either OSU-ERb-12 or E2 co-treatments. D. Western immunoblot measurements of MYC after treatment with OSU-ERb-12 plus Enza (time 24 h).

Figure 4.. OSU-ERb-12 plus Enza induces distinct patterns of chromatin accessibility in 22Rv1 cells.

A. 22Rv1 cells were treated with vehicle control, OSU-ERb-12 alone (100 nM), Enza (1 μM) or the combinations for 6 h in triplicate and ATAC-Seq was undertaken. FASTQ files were QC processed, aligned to hg38 (Rsubread), sorted and duplicates removed before further processing with ATACseqQC to generate nucleosome free (NF) and mono-nucleosome regions. Differential enrichment of regions was measured with csaw and the significantly different regions (p.adj < .1) were then intersected to generate the Venn diagrams of overlapping regions by a minimum of 1bp (ChIPpeakAnno). B. Motif enrichment in NF regions was undertaken with Homer and ranked by the percent changes in motifs and those most changing were visualized. C. NF regions were annotated to genes (ChIPpeakAnno) and those genes overlapped with the transcriptomic data for either OSU-ERb-12 plus Enza, or E2 plus Enza, from which genes were filtered that were regulated in the opposite direction and further filtered to those that were either coactivator (CoA), corepressor (CoR), mixed function coregulators (Mixed) or Transcription factors (TF) and visualized.

Figure 5.. AR, ARv7, MYC and H3K27ac cistromes respond to OSU-ERb-12 plus Enza.

22Rv1 cells were treated with vehicle control, OSU-ERb-12 alone (100 nM), Enza (1 μM) or the combinations for 6 h in triplicate and CUT&RUN undertaken with antibodies to AR, ARv7, MYC or H3K27ac. FASTQ files were QC processed, aligned to hg38 (Rsubread), sorted and duplicates and differential enrichment of regions of the drug combination compared to the average of the individual treatments with csaw. A. The significant differentialy enriched regions, both up and down) in the combined treatments for the AR, ARv7, MYC and H3K27ac cistromes were intersected to generate the Venn diagram of overlapping regions by a minimum of 1bp (ChIPpeakAnno). B. The H3K27ac cistromes of the individual and combined treatments, compared to IgG, were intersected (minimum of 1bp) and the genomic regions in each intersection (Supplementary Figure 7A) were annotated to genes within 100 kb, and filtered as to whether the genes were either coactivator (CoA), corepressor (CoR), mixed function coregulators (Mixed) or Transcription factors (TF). Enrichment was then determined by hypergeometric test and the group gene member closest to a given H3K27ac site is indicated. C. Motif analyses of MYC and AR sites that were significantly reduced or lost by the cotreatment and the most significant of which is illustrated by transcription class terms and the most significant class member indicated.

Figure 6.. Cistrome-transciptome relationships impacted by OSU-ERb-12 plus Enza.

A. Unique AR and MYC cistromes in response to OSU-ERb-12/Enza were overlapped using GIGGLE with the CistromeDB collection (> 10,000 total ChIP-seq datasets, across > 1100 factors). The logPV of the FDR-corrected pval is illustrated for prostate specific TFs and coregulators. B. Each peak was annotated to genes within 100 kb and filtered as to whether the gene was a DEG from the respective RNA-Seq and the score.absFC calculated for peak:gene relationships that were also DEGs, and a t-test used to assess whether these values were significantly different between the groups where the CUT&RUN peak enrichment was reduced (D) or increased (U) and the annotated gene expression what down-regulated (D) or up-regulated (u). C. IGV image of the AR, MYC and H3K27ac at the ZBTB16 gene. Triplicate BAM files were merged and converted to bigwig files and visualized at the ZBTB16 gene as an example of a gene that displays significantly reduced enrichment by OSU-ERb-12/Enza for H3K27ac and MYC, and is downregulated at the mRNA level.

Figure 7.. Unique OSU-ERb-12 plus Enza cistrome-transciptome dependent target genes significantly associate with high androgen and low neuroendocine score tumors in the SU2C cohort.

The Zscore data frame from the SU2C cohort was filtered for 192 unique genes that were defined by; a loss of AR binding and upregulation by OSU-ERb-12 plus Enza; a loss of MYC binding and downregulation by OSU-ERb-12 plus Enza; a loss of H3K27ac binding and downregulation by OSU-ERb-12 plus Enza. These relationships were visualized with a heatmap (pheatmap) and tumors clustered by expression, and the significance of the relationship between tumor cluster id and either neuroendocrine score (NE range) or androgen score (AR range), as define previously(45), was measured with a chi-sq test.