An expanded variant list and assembly annotation identifies multiple novel coding and noncoding genes for prostate cancer risk using a normal prostate tissue eQTL data set

Extensive androgen receptor enhancer heterogeneity in primary prostate cancers underlies transcriptional diversity and metastatic potential

Androgen receptor (AR) drives prostate cancer (PCa) development and progression. AR chromatin binding profiles are highly plastic and form recurrent programmatic changes that differentiate disease stages, subtypes and patient outcomes. While prior studies focused on concordance between patient subgroups, inter-tumor heterogeneity of AR enhancer selectivity remains unexplored. Here we report high levels of AR chromatin binding heterogeneity in human primary prostate tumors, that overlap with heterogeneity observed in healthy prostate epithelium. Such heterogeneity has functional consequences, as somatic mutations converge on commonly-shared AR sites in primary over metastatic tissues. In contrast, less-frequently shared AR sites associate strongly with AR-driven gene expression, while such heterogeneous AR enhancer usage also distinguishes patients' outcome. These findings indicate that epigenetic heterogeneity in primary disease is directly informative for risk of biochemical relapse. Cumulatively, our results illustrate a high level of AR enhancer heterogeneity in primary PCa driving differential expression and clinical impact.

Identifying genes targeted by disease-associated non-coding SNPs with a protein knowledge graph

Genome-wide association studies (GWAS) have identified many single nucleotide polymorphisms (SNPs) that play important roles in the genetic heritability of traits and diseases. With most of these SNPs located on the non-coding part of the genome, it is currently assumed that these SNPs influence the expression of nearby genes on the genome. However, identifying which genes are targeted by these disease-associated SNPs remains challenging. In the past, protein knowledge graphs have often been used to identify genes that are associated with disease, also referred to as “disease genes”. Here, we explore whether protein knowledge graphs can be used to identify genes that are targeted by disease-associated non-coding SNPs by testing and comparing the performance of six existing methods for a protein knowledge graph, four of which were developed for disease gene identification. We compare our performance against two baselines: (1) an existing state-of-the-art method that is based on guilt-by-association, and (2) the leading assumption that SNPs target the nearest gene on the genome. We test these methods with four reference sets, three of which were obtained by different means. Furthermore, we combine methods to investigate whether their combination improves performance. We find that protein knowledge graphs that include predicate information perform comparable to the current state of the art, achieving an area under the receiver operating characteristic curve (AUC) of 79.6% on average across all four reference sets. Protein knowledge graphs that lack predicate information perform comparable to our other baseline (genetic distance) which achieved an AUC of 75.7% across all four reference sets. Combining multiple methods improved performance to 84.9% AUC. We conclude that methods for a protein knowledge graph can be used to identify which genes are targeted by disease-associated non-coding SNPs.

A microRNA Transcriptome-wide Association Study of Prostate Cancer Risk

Large genome-wide association studies have identified hundreds of single-nucleotide polymorphisms associated with increased risk of prostate cancer (PrCa), and many of these risk loci is presumed to confer regulatory effects on gene expression. While eQTL studies of long RNAs has yielded many potential risk genes, the relationship between PrCa risk genetics and microRNA expression dysregulation is understudied. We performed an microRNA transcriptome-wide association study of PrCa risk using small RNA sequencing and genome-wide genotyping data from N = 441 normal prostate epithelium tissue samples along with N = 411 prostate adenocarcinoma tumor samples from the Cancer Genome Atlas (TCGA). Genetically regulated expression prediction models were trained for all expressed microRNAs using the FUSION TWAS software. TWAS for PrCa risk was performed with both sets of models using single-SNP summary statistics from the recent PRACTICAL consortium PrCa case-control OncoArray GWAS meta-analysis. A total of 613 and 571 distinct expressed microRNAs were identified in the normal and tumor tissue datasets, respectively (overlap: 480). Among these, 79 (13%) normal tissue microRNAs demonstrated significant cis-heritability (median cis-h2 = 0.15, range: 0.03–0.79) for model training. Similar results were obtained from TCGA tumor samples, with 48 (9%) microRNA expression models successfully trained (median cis-h2 = 0.14, range: 0.06–0.60). Using normal tissue models, we identified two significant TWAS microRNA associations with PrCa risk: over-expression of mir-941 family microRNAs (PTWAS = 2.9E-04) and reduced expression of miR-3617-5p (PTWAS = 1.0E-03). The TCGA tumor TWAS also identified a significant association with miR-941 overexpression (PTWAS = 9.7E-04). Subsequent finemapping of the TWAS results using a multi-tissue database indicated limited evidence of causal status for each microRNA with PrCa risk (posterior inclusion probabilities <0.05). Future work will examine downstream regulatory effects of microRNA dysregulation as well as microRNA-mediated risk mechanisms via competing endogenous RNA relationships.

The variant rs77559646 associated with aggressive prostate cancer disrupts ANO7 mRNA splicing and protein expression

Prostate cancer is among the most common cancers in men, with a large fraction of the individual risk attributable to heritable factors. A majority of the diagnosed cases does not lead to a lethal disease, and hence biological markers that can distinguish between indolent and fatal forms of the disease are of great importance for guiding treatment decisions. Although over 300 genetic variants are known to be associated with prostate cancer risk, few have been associated with the risk of an aggressive disease. One such variant is rs77559646 located in ANO7. This variant has a dual function. It constitutes a missense mutation in the short isoform of ANO7 and a splice region mutation in full-length ANO7. In this study, we have analyzed the impact of the variant allele of rs77559646 on ANO7 mRNA splicing using a minigene splicing assay and by performing splicing analysis with the tools IRFinder (intron retention finder), rMATS (replicate multivariate analysis of transcript splicing) and LeafCutter on RNA sequencing data from prostate tissue of six rs77559646 variant allele carriers and 43 non-carriers. The results revealed a severe disruption of ANO7 mRNA splicing in rs77559646 variant allele carriers. Immunohistochemical analysis of prostate samples from patients homozygous for the rs77559646 variant allele demonstrated a loss of apically localized ANO7 protein. Our study is the first to provide a mechanistic explanation for the impact of a prostate cancer risk SNP on ANO7 protein production. Furthermore, the rs77559646 variant is the first known germline loss-of-function mutation described for ANO7. We suggest that loss of ANO7 contributes to prostate cancer progression.

CRISPRi screens reveal a DNA methylation-mediated 3D genome dependent causal mechanism in prostate cancer

Prostate cancer (PCa) risk-associated SNPs are enriched in noncoding cis-regulatory elements (rCREs), yet their modi operandi and clinical impact remain elusive. Here, we perform CRISPRi screens of 260 rCREs in PCa cell lines. We find that rCREs harboring high risk SNPs are more essential for cell proliferation and H3K27ac occupancy is a strong indicator of essentiality. We also show that cell-line-specific essential rCREs are enriched in the 8q24.21 region, with the rs11986220-containing rCRE regulating MYC and PVT1 expression, cell proliferation and tumorigenesis in a cell-line-specific manner, depending on DNA methylation-orchestrated occupancy of a CTCF binding site in between this rCRE and the MYC promoter. We demonstrate that CTCF deposition at this site as measured by DNA methylation level is highly variable in prostate specimens, and observe the MYC eQTL in the 8q24.21 locus in individuals with low CTCF binding. Together our findings highlight a causal mechanism synergistically driven by a risk SNP and DNA methylation-mediated 3D genome architecture, advocating for the integration of genetics and epigenetics in assessing risks conferred by genetic predispositions.