Figure 2. PRMT5 is an ERG-dependent inhibitor of AR signaling.

(A) Heat map showing all genes upregulated (red) or downregulated (blue) by at least 1.5 fold following knockdown with PRMT5 shRNA1 (sh1) or shRNA2 (sh2) compared to NTC shRNA. Rows represent probe sets; columns represent individual samples (technical replicates are marked by 1 or 2). Table indicates pathways significantly upregulated by PRMT5 knockdown (see Materials and methods, and Supplementary file 3 for significantly downregulated pathways). (B) qRT-PCR of AR targets PSA, NKX3-1, and SLC45A3 in VCaP cells expressing the noted shRNA constructs. Expression levels were normalized as described in Materials and methods; bars represent + SEM of three biological replicates, each with three technical repeats. (C) qRT-PCR of PSA and NKX3-1 from VCaP cells expressing the noted shRNA constructs alongside cDNAs expressing vector control (Vector), wild-type (WT) PRMT5, or a catalytically dead PRMT5 mutant (G365A/R368A). Data and error bars represented as in (B). (D) Top panels: cartoons of the PSA and NKX3-1 loci. ERG and AR binding sites (and control regions) are noted and numbered relative to the transcription start site (TSS) as described in Materials and methods. Bottom panels: ERG, PRMT5, and AR ChIP qPCR for the noted regions of PSA (left) or NKX3-1 (right) in VCaP cells upon ERG or PRMT5 knockdown. Normalization to IgG control ChIP is as described in Materials and methods; error bars represent + SEM of three biological replicates, each with three technical repeats. (E) Heatmap visualization of AR binding from ChIP-sequencing data as determined by normalized reads across the AR Cistrome (Materials and methods) in replicate samples induced using AR ligands (DHT or R881 as indicated) and harboring inducible PRMT5 shRNA1 (sh1), shRNA2 (sh2), or shRNA3 (sh3) compared to NTC shRNA. 1659 peaks show differential binding with at least 1.5 fold difference (p-value of 0.01, q-value 0.151). The majority of differentially bound sites exhibit increased binding (6% of the total Cistrome) under PRMT5 knockdown conditions.

DOI: http://dx.doi.org/10.7554/eLife.13964.005

Figure 2—figure supplement 1. AR signature analysis.

Correlation between three independent androgen receptor activation gene signatures (Malik et al., 2015; Mendiratta et al., 2009; Nelson et al., 2002) in comparison with the top-ranked upregulated genes following PRMT5 knockdown by shRNA in VCaP cells (Figure 2A).Each vertical line (green, blue, or red) represents the highest expressed probe set for each gene in the gene signatures. Lines are elongated if the probe set was upregulated following PRMT5 knockdown by at least 1.5 fold with a nominal p value <0.05. The p values shown are based on a two-tailed Fisher’s exact test described in Materials and methods.

DOI: http://dx.doi.org/10.7554/eLife.13964.006

Figure 2—figure supplement 2. ERG and PRMT5 effects on AR target genes are specific.

(A) Western blot of VCaP cells expressing wild-type (WT) PRMT5, a catalytic dead PRMT5 mutant (PRMT5 G365A/R368A) or a vector control (Vector) in the background of two independent shRNA vectors (sh1 and sh2). Cells were maintained in culture for 10 days in presence of 100 ng/ml doxycycline (Dox). Western blot analysis shows expression levels of HA affinity tag (overexpressed PRMT5), total PRMT5, and GAPDH as a loading control. (B) qRT-PCR of the migration genes PLAT, PLAU and the neuroendocrine genes NTRK3, CHGA, GDAP1 in VCaP cells expressing either ERG or PRMT5 shRNA constructs. Expression levels were normalized as described in Materials and methods; bars represent + SEM of three replicates. (C) qRT-PCR of PSA, NKX3-1 and SLC45A3 in LNCaP and 22Rv1 cells expressing either NTC shRNA, PRMT5 shRNA 1 or PRMT5 shRNA 2. Data represent normalized expression of PSA, NKX3-1 and SLC45A3 mRNA relative to the B2M transcript. Error bars represent + SEM of three replicates. (D) qRT-PCR for PSA, NKX3-1 and SLC45A3 in parental (PAR) 22Rv1 cells and following expression of either ERG or ERG DNAx and expressing either NTC shRNA, PRMT5 shRNA 1 or PRMT5 shRNA 2 (see Materials and methods for description) normalized relative to the B2M transcript. Error bars represent + SEM of three replicates.

DOI: http://dx.doi.org/10.7554/eLife.13964.007

Figure 2—figure supplement 3. ERG, AR and PRMT5 recruitment.

(A) Top panel: cartoons of the PSA and NKX3-1 loci including the promoter and enhancer regions bound by ERG and AR (-4100 and -3800 on PSA; +10800 and +62100 on NKX3-1) along with other control regions. All distances shown are relative to the PSA and NKX3-1 transcription start site (TSS). Bottom panels: chromatin Immunoprecipitations (ChIP) for ERG, PRMT5, and AR followed by quantitative PCR using primers specific to each region as shown by the horizontal bars in top panels. Shown are the recruitments to the PSA promoter and enhancer regions (left) and to the NKX3-1 promoter and enhancer regions (right) in 22Rv1 cells in the absence and presence of ERG. All recruitments are normalized to the IgG control ChIP. Normalization and error bars are described in Materials and methods. (B) The top 1000 (MACS mfold rank) AR and ERG recruitment peaks, analyzed for evolutionary conservation in mammalian species using Phastcons, showing that ERG and AR samples have internally consistent performance. (C) Top-scoring motifs in all samples for AR ChIPseq (left panel) and ERG ChIP-seq (right panel). (D) z-scores for the motifs identified in (C).

DOI: http://dx.doi.org/10.7554/eLife.13964.008