A Large-Scale Analysis of Genetic Variants within Putative miRNA Binding Sites in Prostate Cancer

Abstract

Prostate cancer is the second most common malignancy among men worldwide. Genome-wide association studies have identified 100 risk variants for prostate cancer, which can explain approximately 33% of the familial risk of the disease. We hypothesized that a comprehensive analysis of genetic variations found within the 3' untranslated region of genes predicted to affect miRNA binding (miRSNP) can identify additional prostate cancer risk variants. We investigated the association between 2,169 miRSNPs and prostate cancer risk in a large-scale analysis of 22,301 cases and 22,320 controls of European ancestry from 23 participating studies. Twenty-two miRSNPs were associated (P<2.3×10(-5)) with risk of prostate cancer, 10 of which were within 7 genes previously not mapped by GWAS studies. Further, using miRNA mimics and reporter gene assays, we showed that miR-3162-5p has specific affinity for the KLK3 rs1058205 miRSNP T-allele, whereas miR-370 has greater affinity for the VAMP8 rs1010 miRSNP A-allele, validating their functional role.

Significance: Findings from this large association study suggest that a focus on miRSNPs, including functional evaluation, can identify candidate risk loci below currently accepted statistical levels of genome-wide significance. Studies of miRNAs and their interactions with SNPs could provide further insights into the mechanisms of prostate cancer risk.

Figure 1. miRSNP association with prostate cancer risk

Manhattan Plot with −log p-values adjusted for study site and principal components. 2,169 miRSNPs were assessed for association with prostate cancer risk. 22 SNPs representing 16 genes were found to be associated with risk of prostate cancer after correction for multiple testing (p<2.3×10⁻⁵). (The 10 most significant SNPs are labelled). KLK3 rs1058205 and VAMP8 rs1010 (in bold) were selected for further functional validation studies. (OR = odds ratio).

Figure 2. Expression levels of 16 genes harbouring 22 significant miRSNPs in cancerous and normal tissue from prostate cancer patients

Oncomine analysis of the Grasso dataset (26) of 59 tumor and 28 non-tumor (normal) samples, shows the expression of 7 genes (PDLIM5 – VAMP8) to be deregulated (p<0.05) in prostate cancer.

Figure 3. miR-3162-5p directly targets the KLK3 rs1058205 SNP T allele and miR-370-5p targets the VAMP8 rs1010 SNP A allele with greater affinity

Following over-expression with miR-3162-5p, reporter vector assays demonstrated a ~29% decrease in luciferase levels (p = 0.048) for the KLK3 rs1058205 SNP T-allele compared to the C-allele (A). Over-expression of miR-370-5p resulted in a change in luciferase activity for both VAMP8 rs1010 SNP variants with the decrease for the A-allele ~2 fold (p = 0.0067) stronger than for the G-allele (B). Mean +/− SD, n = 3. (* = p < 0.05) (** = p < 0.01)

Figure 4. miR-3162-5p induces a reduction in KLK3 mRNA and KLK3 protein expression in LNCaP cells homozygous for the rs1058205 T SNP-allele

(A) qPCR analyses in LNCaP cells revealed a 25% decrease in KLK3 mRNA (p = 0.016) following over-expression of miR-3162-5p compared to the negative control miRNA mimic treatment. (B) Western blot analyses in LNCaP cells revealed a 32% decrease in cellular KLK3 protein (p = 0.007) following over-expression of miR-3162-5p compared to the negative control. (C) Representative Western blot. Mean +/− SD, n = 3. (* = p < 0.05) (** = p < 0.01)