Biology and Clinical Implications of the 19q13 Aggressive Prostate Cancer Susceptibility Locus

Abstract

Genome-wide association studies (GWAS) have identified rs11672691 at 19q13 associated with aggressive prostate cancer (PCa). Here, we independently confirmed the finding in a cohort of 2,738 PCa patients and discovered the biological mechanism underlying this association. We found an association of the aggressive PCa-associated allele G of rs11672691 with elevated transcript levels of two biologically plausible candidate genes, PCAT19 and CEACAM21, implicated in PCa cell growth and tumor progression. Mechanistically, rs11672691 resides in an enhancer element and alters the binding site of HOXA2, a novel oncogenic transcription factor with prognostic potential in PCa. Remarkably, CRISPR/Cas9-mediated single-nucleotide editing showed the direct effect of rs11672691 on PCAT19 and CEACAM21 expression and PCa cellular aggressive phenotype. Clinical data demonstrated synergistic effects of rs11672691 genotype and PCAT19/CEACAM21 gene expression on PCa prognosis. These results provide a plausible mechanism for rs11672691 associated with aggressive PCa and thus lay the ground work for translating this finding to the clinic.

Keywords: CEACAM21; GWAS; HOXA2; PCAT19; aggressive prostate cancer; allele-specific DNA-binding of transcription factor; eQTL; risk stratification; rs11672691; single cell CRISPR/Cas9-mediated editing.

Figure 1.. Candidate gene identification by eQTL analysis at the 19q13 aggressive PCa risk locus.

(A-D) The aggressive PCa risk allele G at rs11672691 associates with increased expression of CEACAM21 (A-C) and PCAT19 (D) in prostate tissues. Linear model P values assessed by Matrix eQTL. (E-F) PCAT19 or CEACAM21 promotes PCa cell proliferation measured by XTT colorimetric assays (mean ± SD of triplicate experiments), and aggressiveness by migration and invasion assays (mean ± SEM of triplicate experiments) in 22Rv1 cells infected with control or gene-specific shRNAs. (G) Ectopic CEACAM21 expression promotes 22Rv1 cell proliferation (mean ± SD of triplicate experiments). (H) CEACAM21 overexpression enhances RWPE1 cell migration and invasion (mean ± SEM of triplicate experiments). In E-H, *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test. (I) GSEA plot testing the enrichment of differentially expressed genes between CEACAM21 overexpressing versus control RWPE1 cells. (J) PCAT19 or (K,L) CEACAM21 transcript levels upregulated in human primary or metastasis PCa. P values by Mann-Whitney U-tests. (M) Higher levels of CEACAM21 correlate with increased risk of biochemical recurrence in a cohort of PCa patients. P value assessed by a log-rank test. See also Figures S1-S3 and Table S2.

Figure 2.. The risk allele G of rs11672691 enhances HOXA2 chromatin binding.

(A) ChIP-seq tracks showing the enrichment of active enhancer marks, silent epigenetic signature, and transcription factors at the rs11672691 region. (B) Reporter assays showing enhancer activity of the pGL3 promoter vector inserted with the rs11672691-containing or control DNA fragments. (C) rs11672691 and rs887391 reside within HOXA2 DNA-binding motifs. (D) Binding affinity of HOX A family members, HOXB13, AR, FOXA1, and ERG to the rs11672691 surrounding sequence competed with the counterparts harboring G or A allele. (E) ChIP-qPCR confirmation of transcription factor binding at rs11672691 region. (F) ChIP-qPCR results showing HOXA2 binding at rs11672691 in LNCaP cells. (G) ChIP-AS-qPCR indicating allele-specific binding of HOXA2 at rs11672691 in 22Rv1 cells. (H) HOXA2 prefers the binding to G allele than A allele at rs1672691 confirmed by ChIP Sanger sequencing. (I) Western blot result of V5 tagged HOXA2 expression in 22Rv1. (J) ChIP-AS-qPCR for HOXA2 enrichment at the rs11672691 region in HOXA2 overexpressing and control cells. Error bars, SD (B, D-G, J), n = 3 technical replicates. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t tests. See also Figure S4, Tables S3-S6

Figure 3.. Effect of HOXA2 on PCa cell growth, metastasis and patient prognosis.

(A) Depletion of HOXA2 reduces PCa cell proliferation, mean ± SD of triplicate experiments. (B) Genome-wide loss-of-function screening of the essential genes for cell survival. Lower ATARiS scores demonstrate elevated dependency of cell viability on given genes. AR, HOXB13, MYC and BRD4 are known to be important for PCa cell growth and survival, and TP53 vice versa. (C) The number of 22Rv1 cells infected with control or HOXA2 shRNAs in invasion assay. Mean ± SEM of triplicate experiments. In A and C, *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test. (D and E) Elevation of HOXA2 mRNA levels in human primary or metastasis PCa. P values examined by Mann-Whitney U-tests. (F and G) Higher levels of HOXA2 correlate with increased risk of biochemical recurrence (F) and reduced time for overall survival (G) in PCa patient cohorts. (H) Higher levels of HOXA2 shows predictive values for biochemical recurrence in patient group with Gleason Score 7 (intermediate risk). P values determined by a log-rank test (F-H), and Cox regression analysis (F and H). See also Figure S5.

Figure 4.. Direct effect of rs11672691 on PCAT19 and CEACAM21 expression.

(A) Depletion of HOXA2 diminishes the mRNA levels of PCAT19 and CEACAM21. (B and C) Scatter plots showing an expression correlation between HOXA2 and CEACAM21 or PCAT19 in prostate tissues. (D) PCAT19 knockdown results in decreased expression of CEACAM21. (E) Scatter plot displaying a correlation between PCAT19 and CEACAM21 expression in human prostate. (F) Reporter assay showing contribution of the G compared with the A allele at rs11672691 to an increased enhancer (E) activity for CEACAM21 promoter (P). Error bars, SD from five technical replicates. (G) 3C analysis of chromatin interactions between CEACAM21 and rs11672691 locus within nearly 100 kb region (chr19:41982282–42079092). (H) Sanger sequencing of CRISPR/Cas9-modified and parental 22Rv1 cells. (I) Analysis of PCAT19 and CEACAM21 expression in mutated and parental 22Rv1 cells. (J) Chromatin enrichment of HOXA2 at the rs11672691 site measured by ChIP-qPCR. (K) 3C measurement of chromatin interactions between CEACAM21 promoter and rs11672691 locus in the CRISPR/Cas9 modified and parental 22Rv1 cells. NS, non-significant. In A,D,G,I,J, and K, data shown are mean ± SD of triplicate experiments, **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test. See also Figure S6 and Table S7.

Figure 5.. Effect of rs11672691 genotype on PCa cell growth, aggressive behavior and patient prognosis.

(A) The phenotype of each 22Rv1 cell lines analyzed by microscopy under culture for two days. (B) Cell proliferation analysis of CRISPR/Cas9-modified and parental 22Rv1 cells, mean ± SD of triplicate experiments. (C) Representative images showing wound healing assay for migration ability of 22Rv1 cells with different genotypes of rs11672691. (D) Quantification of percentage fraction of original wound closure in triplicate plates. Migration was assessed every 10 h. Error bars, SD from three biological replicates. In B, D, *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test. (E and F) Association analyses show that the patient group with rs11672691 GG genotype correlates with increased risk for biochemical recurrence in two independent cohorts of PCa patients. (G and H) Survival analyses show that the patients carrying rs11672691 GG genotype are more likely to get metastasis disease earlier. In E-H, P values examined by a log-rank test.

Figure 6.. Synergistic effect of rs11672691 genotype and CEACAM21 or PCAT19 expression on PCa patient prognosis.

(A and B) CEACAM21 indicates strong predictive value in the patient group with rs11672691 GG genotype (A), but not in the group with rs11672691 AA or GA genotype (B). (C and D) PCAT19 shows no prognostic value in a cohort of PCa patients (C), but the PCa patient group carrying rs11672691 GG genotype with higher PCAT19 expression tumors indicates a significant association with increased risk of biochemical relapse (D). The P values were assessed by a log-rank test (A-D) and Cox regression analysis (A and D). See also Figure S7.