Exome Sequencing of African-American Prostate Cancer Reveals Loss-of-Function ERF Mutations

Abstract

African-American men have the highest incidence of and mortality from prostate cancer. Whether a biological basis exists for this disparity remains unclear. Exome sequencing (n = 102) and targeted validation (n = 90) of localized primary hormone-naïve prostate cancer in African-American men identified several gene mutations not previously observed in this context, including recurrent loss-of-function mutations in ERF, an ETS transcriptional repressor, in 5% of cases. Analysis of existing prostate cancer cohorts revealed ERF deletions in 3% of primary prostate cancers and mutations or deletions in ERF in 3% to 5% of lethal castration-resistant prostate cancers. Knockdown of ERF confers increased anchorage-independent growth and generates a gene expression signature associated with oncogenic ETS activation and androgen signaling. Together, these results suggest that ERF is a prostate cancer tumor-suppressor gene. More generally, our findings support the application of systematic cancer genomic characterization in settings of broader ancestral diversity to enhance discovery and, eventually, therapeutic applications.Significance: Systematic genomic sequencing of prostate cancer in African-American men revealed new insights into prostate cancer, including the identification of ERF as a prostate cancer gene; somatic copy-number alteration differences; and uncommon PIK3CA and PTEN alterations. This study highlights the importance of inclusion of underrepresented minorities in cancer sequencing studies. Cancer Discov; 7(9); 973-83. ©2017 AACR.This article is highlighted in the In This Issue feature, p. 920.

Figure 1. Exome mutation plot of 102 primary prostate cancers from African-American men

Significantly mutated genes (SPOP, ERF, FOXA1) determined by MutSigCV (q < 0.1) and other selected prostate cancer genes are shown with mutation frequencies and types of mutations. Mean mutation rate per sample is shown in the top panel. Pathologic features of the prostate tumors including Gleason score, pathologic stage, and ERG by FISH, PTEN deletion by immunohistochemistry, and SPINK1 overexpression are shown in the second panel for available samples. Somatic copy number events across the samples at recurrently affected loci with noted prostate cancer genes in parentheses are shown in the bottom panel.

Figure 2. ERF RNA expression is significantly lower in ERF mutated prostate cancer

A) Histology of prostatic adenocarcinoma, Gleason score 3+4=7 (case STID21598), (left). This representative ERF mutated tumor demonstrates only few ERF RNA signals as highlighted by arrowheads (center). RNA signals are seen as scattered yellow dots during image fragmentation for automated analysis (right). B) Histology of prostatic adenocarcinoma, Gleason score 4+3=7 (case STID21603), (left). In contrast to ERF mutated cases, this representative ERF wt prostate cancer demonstrates numerous signals by RNA in situ hybridization (asterisk, center). These are highlighted as yellow dots for automated image analysis (right). C) Automated image analysis of three ERF mutant cases vs. six ERF wt prostate cancer cases demonstrates a significant difference (p<0.05) in RNA expression by this in situ assay. D) IGV screenshot of deletions spanning the ERF locus in the AAPC exome cohort E) Validation of ERF-deleted prostate cancer by fluorescence in situ hybridization (FISH). Top panel shows histology of prostatic adenocarcinoma, Gleason score 4+3=7 (case STID12329). F) In the bottom panel, a FISH assay shows hemizygous ERF deletion in the tumor (left image; blue square in panel F), as demonstrated by only one target (red) signal and two centromeric (green) reference signals. In contrast, ERF wild type status (right image; yellow square in panel F) is seen in nuclei of adjacent benign glands. (Original magnification: H&E at 40×; FISH images at 100× under oil immersion).

Figure 3. ERF is mutated in the AAPC cohort and can act as a tumor suppressor

A) Types and locations of mutations in ERF found in the discovery (black) and extension (green) AAPC cohorts. Mutations were validated by Fluidigm Access Array except those indicated by #. B) Mutation significance p-values for genes from the AAPC cohort (n=102) and TCGA (n=333) were plotted. Dotted lines are drawn at the q-value = 0.1 for significance within each cohort. C) Anchorage-independent growth assays of ERF knockdown in PC-3 prostate cancer cell lines were performed in triplicate. Colonies were quantified with CellProfiler, * p<.05, student’s T test. Data shown is representative of two independent experiments. Western blot results are depicted for knockdown of ERF with two short hairpins and control hairpin.

Figure 4. The association of ERF knockdown signatures with the single-sample GSEA enrichment profiles of ETS and AR Gene sets

A) An ERF knockdown signature profile in the prostate epithelial cell line LHS-AR associates with the enrichment profiles of ETV1 gene sets in the TCGA and CRPC expression datasets B) A composite ERF knockdown signature profile, i.e. the enrichment profile of the overlap between ERF knockdown signatures in the VCAP and LNCaP cell lines, associates with the enrichment profile of an ERG gene set across the TCGA and CRPC expression datasets C) The same composite ERF knockdown signature association with enrichment profile of AR gene sets across the CCLE, TCGA, and CRPC datasets. D) Focus formation assay of LNCaP-sgERF cell lines in the context of charcoal-stripped serum (CSS) and androgen treatment (0.1nM R1881). Crystal violet staining was quantitated and compared to control. Mean and standard error of three replicates is shown. *p <0.05 by Student’s t-Test, fold-change sg4 ERF vs sg5 GFP, sg5 ERF vs sg5 GFP. Data shown are representative of two independent experiments.