Identification of functionally active, low frequency copy number variants at 15q21.3 and 12q21.31 associated with prostate cancer risk

Abstract

Copy number variants (CNVs) are a recently recognized class of human germ line polymorphisms and are associated with a variety of human diseases, including cancer. Because of the strong genetic influence on prostate cancer, we sought to identify functionally active CNVs associated with susceptibility of this cancer type. We queried low-frequency biallelic CNVs from 1,903 men of Caucasian origin enrolled in the Tyrol Prostate Specific Antigen Screening Cohort and discovered two CNVs strongly associated with prostate cancer risk. The first risk locus (P = 7.7 × 10(-4), odds ratio = 2.78) maps to 15q21.3 and overlaps a noncoding enhancer element that contains multiple activator protein 1 (AP-1) transcription factor binding sites. Chromosome conformation capture (Hi-C) data suggested direct cis-interactions with distant genes. The second risk locus (P = 2.6 × 10(-3), odds ratio = 4.8) maps to the α-1,3-mannosyl-glycoprotein 4-β-N-acetylglucosaminyltransferase C (MGAT4C) gene on 12q21.31. In vitro cell-line assays found this gene to significantly modulate cell proliferation and migration in both benign and cancer prostate cells. Furthermore, MGAT4C was significantly overexpressed in metastatic versus localized prostate cancer. These two risk associations were replicated in an independent PSA-screened cohort of 800 men (15q21.3, combined P = 0.006; 12q21.31, combined P = 0.026). These findings establish noncoding and coding germ line CNVs as significant risk factors for prostate cancer susceptibility and implicate their role in disease development and progression.

Fig. 1.

Genomic characteristics of the transcriptionally active, low-frequency, deletion CNVs selected for prostate cancer risk association analysis. (A) Genomic location along the 22 autosomal chromosomes. (B) CNV size distribution.

Fig. 2.

Genomic region for the noncoding CNV on chromosome region 15q21.3 associated with aggressive prostate cancer risk. (A) DNA and RNA views of seven individuals at chr15q21.3locus. Paired DNA (Left) and RNA (Right) data are visualized for seven individuals with different numbers of DNA copies (two copies, one-copy loss, two-copy loss) at the locus of interest (27). No transcripts are present. (B) The tracks include the following information: Prostate cancer-associated CNV on 15q21.3 (dark red; coordinates include midpoint distance to neighboring marker on the genotyping platform; black vertical bars on platform indicate probe locations); the locus corresponding to Variation_66804 in Conrad et al. (24) (black); RefSeq track for validated transcripts; ENCODE DNasel hypersensitivity; ENCODE mono and trimethylation tracks, H3K4Me1 and H3K4Me3, consistent with enhancer activity (48); ENCODE transcription factor-binding signal (intensities are proportional to binding support); H3K4Me2 and c-Jun ChIP-seq tracks in RWPE cells; predicted androgen receptor (AR) (black) and estrogen receptor (ER) (blue) binding sites (28, 29); and sequence conservation tracks. The positions are to scale and adapted from UCSC genome browser, Feb2009 (GRCh37/hg19) assembly (http://genome.ucsc.edu/). (C) Summary of functional effects of the intergenic 15q21.3 prostate cancer risk locus. This CNV overlaps a potential regulatory element with AP-1 transcription factor binding sites. Our data support a trans- and cis-regulatory role for this region. Presence of deletion in this CNV region is associated with expression of target genes both in cis and trans, potentially leading to prostate cancer predisposition.

Fig. 3.

Tumorigenic effects of MGAT4C on prostate cell lines. (A) Proliferation rate at 24 h in RWPE1, LNCaP, and VCaP cells transfected with MGAT4C. (B) Proliferation rate at 24 and 48 h in RWPE1, LNCaP, and VCaP cells following knockdown of MGAT4C expression. (C) Migration of MGAT4C overexpressing RWPE1, LnCaP, and VCaP cells compared with vector control cells (Boyden Chamber Assay). (D) Representative images of migrated cells for each cell line. (E) MGAT4C expression levels in human localized and lethal prostate cancer samples.