MiR-21 Is Induced by Hypoxia and Down-Regulates RHOB in Prostate Cancer

Abstract

Tumour hypoxia is a well-established contributor to prostate cancer progression and is also known to alter the expression of several microRNAs. The over-expression of microRNA-21 (miR-21) has been consistently linked with many cancers, but its role in the hypoxic prostate tumour environment has not been well studied. In this paper, the link between hypoxia and miR-21 in prostate cancer is investigated. A bioinformatic analysis of The Cancer Genome Atlas (TCGA) prostate biopsy datasets shows the up-regulation of miR-21 is significantly associated with prostate cancer and clinical markers of disease progression. This up-regulation of miR-21 expression was shown to be caused by hypoxia in the LNCaP prostate cancer cell line in vitro and in an in vivo prostate tumour xenograft model. A functional enrichment analysis also revealed a significant association of miR-21 and its target genes with processes related to cellular hypoxia. The over-expression of miR-21 increased the migration and colony-forming ability of RWPE-1 normal prostate cells. In vitro and in silico analyses demonstrated that miR-21 down-regulates the tumour suppressor gene Ras Homolog Family Member B (RHOB) in prostate cancer. Further a TCGA analysis illustrated that miR-21 can distinguish between different patient outcomes following therapy. This study presents evidence that hypoxia is a key contributor to the over-expression of miR-21 in prostate tumours, which can subsequently promote prostate cancer progression by suppressing RHOB expression. We propose that miR-21 has good potential as a clinically useful diagnostic and prognostic biomarker of hypoxia and prostate cancer.

Keywords: RHOB; biomarker; hypoxia; miR-21; microRNA; prostate cancer.

Figure 1

Up-regulation of miR-21 is associated with prostate cancer. (A) UCSC Xena analysis of TCGA-PRAD samples shows miR-21 expression is significantly increased in prostate tumour tissue (n = 494) compared to that of normal prostate tissue (n = 52). (Welch’s t-test *** p < 0.001). (B) miR-21 is significantly elevated in the serum of prostate cancer patients compared to that of healthy, non-cancer control patients. Data are from GEO datasets GSE112264 (n, Non-cancer = 41, PCa = 809) and GSE113486 (n, Non-cancer = 100, PCa = 40). (One-way ANOVA with multiple comparison tests, ** p < 0.01, *** p < 0.001.) UCSC Xena analysis of TCGA-PRAD samples shows expression of miR-21 is significantly associated with (C) Gleason grade (D) pathological T stage and (E) number of positive lymph nodes. (One-way ANOVA with multiple comparison tests, * p < 0.05, ** p < 0.01, *** p < 0.001). (F) Firebrowse analyses of TCGA-PRAD data (n > 400) confirm miR-21 expression had significant positive correlation with Gleason score, number of positive lymph nodes and pathological T stage. (p- and Q-values were generated by Spearman’s correlation with multiple hypothesis correction.) All boxplots show mean and Tukey whiskers.

Figure 2

miR-21 is up-regulated by hypoxia in vitro and in vivo. (A) After 24 and 48 h (h) of hypoxia (0.1% oxygen), miR-21 expression is significantly increased relative to atmospheric oxygen levels (20%) in LNCaP cells. (B) In LNCaP spheroids, miR-21 expression is significantly elevated as spheroid size (and associated hypoxia) increases. (Both paired t-test, * p < 0.05, *** p < 0.001). In bicalutamide-treated (BCA) mice with xenograft LNCaP tumours, miR-21 is increased in the (C) tumours and (D) serum at Day 7 and Day 28, relative to vehicle-treated (Veh) animals (n = 4 mice per group). (E) Regulome Explorer analysis revealed significant positive correlation between expression of hypoxia-induced miR-210 and miR-21 (Pearson correlation, p < 0.001). All bar graphs show mean ± SEM of at least three biological replicates.

Figure 3

RHOB expression is inversely correlated with miR-21 in prostate cancer. (A) Overexpression of miR-21 results in significant reduction in the levels of RHOB mRNA levels in LNCaP cells, as well as reduction in the levels of PTEN, a well-established miR-21 target. (Paired t-test, * p < 0.05). (B) Representative Western blot shows over-expression of miR-21 causes down-regulation of RhoB protein in LNCaP cells (Figure S7). (C) In bicalutamide-treated LNCaP tumours (BCA), RHOB is significantly reduced by Day 28 relative to that of vehicle-treated tumours (Veh). (Paired t-test, ** p < 0.01). (D) CancerMIRNome analysis of TCGA-PRAD samples, including normal (n = 52) and tumour (n = 491) tissue samples, shows the expressions of miR-21 and RHOB are significantly negatively correlated (Pearson correlation, p < 0.001). UCSC Xena analysis of TCGA-PRAD samples shows (E) RHOB expression is significantly reduced in tumour (n = 498) tissues relative to normal (n = 52) tissue and significantly decreases with (F) Gleason score and (G) pathological T stage. (All Welch’s t-test, *** p < 0.001.) (H) Firebrowse analyses of TCGA-PRAD data (n > 400) confirm RHOB expression has significant negative correlation with Gleason score, number of positive lymph nodes and pathological T stage. (p- and Q-values were generated by Spearman’s correlation with multiple hypothesis correction.) All bar graphs show mean ± SEM of three biological replicates. All boxplots show mean and Tukey whiskers.

Figure 4

miR-21 over-expression increases migration and colony-forming ability of RWPE-1 cells. (A) Confirmed over-expression of miR-21 in RWPE-1 cells resulted in (B) significantly increased capability for migration of the pre-miR-21 transfected cells, relative to untreated and the scrambled negative control (pre-miR-neg). (One-way ANOVA, *** p < 0.001.) Mean ± SEM of three biological replicates is shown. (C) Cells transfected with miR-21 have significantly higher colony-forming ability compared to cells transfected with pre-miR-neg as shown by quantified colony forming assays (n = 3) and (D) representative images of crystal violet colony staining. (E) Confirmation that RHOB mRNA levels are reduced in RWPE-1 cells following overexpression of pre-miR-21. (All bar charts; paired t-test, ** p < 0.01.)

Figure 5

Potential of miR-21 as a biomarker of prostate cancer. (A) ROC curve analysis demonstrating that miR-21 shows high potential for distinguishing between tumour and normal tissue. Analysis performed using CancerMIRNome based on PRAD cohort in TCGA database. (B) Biochemical recurrence is associated with significantly high levels of miR-21 (n, no recurrence = 406, recurrence = 61). (Welch’s t-test, *** p < 0.001.) (C) Significant difference in miR-21 levels between patient remission response after primary therapy (n, complete = 380, partial = 41, none (stable or progressive disease) = 58.) (One-way ANOVA with multiple comparison tests, * p < 0.05, *** p < 0.001.) KM survival curves show high levels of miR-21 significantly associated with (D) decreased overall survival and (E) reduced disease-free interval. (Both log-rank (Mantel–Cox) test, * p < 0.05, ** p < 0.01.) Data analysis for B to E was performed using UCSC Xena based on PRAD cohort in TCGA database. All boxplots show mean and Tukey whiskers.