Purpose: African Americans (AA) exhibit higher rates of prostate cancer incidence and mortality compared with European American (EA) men. In addition to socioeconomic influences, biologic factors are believed to play a critical role in prostate cancer disparities. We investigated whether population-specific and -enriched miRNA–mRNA interactions might contribute to prostate cancer disparities.

Experimental Design: Integrative genomics was used, combining miRNA and mRNA profiling, miRNA target prediction, pathway analysis, and functional validation, to map miRNA–mRNA interactions associated with prostate cancer disparities.

Results: We identified 22 AA-specific and 18 EA-specific miRNAs in prostate cancer versus patient-matched normal prostate, and 10 “AA-enriched/-depleted” miRNAs in AA prostate cancer versus EA prostate cancer comparisons. Many of these population-specific/-enriched miRNAs could be paired with target mRNAs that exhibited an inverse pattern of differential expression. Pathway analysis revealed EGFR (or ERBB) signaling as a critical pathway significantly regulated by AA-specific/-enriched mRNAs and miRNA–mRNA pairings. Novel miRNA–mRNA pairings were validated by qRT-PCR, Western blot, and/or IHC analyses in prostate cancer specimens. Loss/gain of function assays performed in population-specific prostate cancer cell lines confirmed miR-133a/MCL1, miR-513c/STAT1, miR-96/FOXO3A, miR-145/ITPR2, and miR-34a/PPP2R2A as critical miRNA–mRNA pairings driving oncogenesis. Manipulating the balance of these pairings resulted in decreased proliferation and invasion, and enhanced sensitization to docetaxel-induced cytotoxicity in AA prostate cancer cells.

Conclusions: Our data suggest that AA-specific/-enriched miRNA–mRNA pairings may play a critical role in the activation of oncogenic pathways in AA prostate cancer. Our findings also suggest that miR-133a/MCL1, miR-513c/STAT1, and miR-96/FOXO3A may have clinical significance in the development of novel strategies for treating aggressive prostate cancer. Clin Cancer Res; 21(21); 4970–84. ©2015 AACR.

Translational Relevance

Prostate cancer tends to be more aggressive and lethal in African Americans (AA) compared with European Americans (EA). An understanding of the molecular mechanisms associated with prostate cancer disparities can aid in the development of innovative and improved therapeutic options for the AA population. Integrative functional genomics analysis of patient specimens and prostate cancer cell lines has identified novel AA-specific and -enriched miRNA–mRNA pairs, including miR-133a/MCL1, miR-513c/STAT1, miR-96/FOXO3A, miR-145/ITPR2, and miR-34a/PPP2R2A, that reside in key oncogenic signaling pathways. The presence of these miRNA–mRNA pairs is computationally predicted to augment activation of EGFR–PI3K–AKT signaling in AA compared with EA cancers. Specific manipulation of these pairs reduced cell proliferation/invasion and enhanced docetaxel-induced cytotoxicity in AA prostate cancer cell lines. Converse manipulation resulted in a more aggressive phenotype in EA cell lines. Thus, targeting these novel miRNA–mRNA pairs may provide a potential clinical strategy for reducing AA prostate cancer burden.

MiRNAs (miRNAs) are small regulatory RNAs of approximately 21 to 25 nucleotides in length that complementarily target mRNAs to inhibit translation and/or promote mRNA degradation. Recently, several reports have suggested that miRNA aberrations may be an important factor in cancer development (1, 2). The potential connection between miRNA regulation and cancer has been made at several levels, suggesting that miRNAs play critical roles in cellular growth and differentiation, which are two cellular processes commonly defective in tumor cells (3). Additional evidence for the involvement of miRNAs in human cancer comes from observations that approximately 50% of these small regulatory RNAs are transcribed from genomic regions associated with a loss of heterozygosity, minimal amplicons, or breakpoint cluster regions (4). Cancer-related miRNAs have been identified in various cancers (5). In general, oncogenic miRNAs upregulated in tumors act as oncogenes (repressing tumor-suppressor and apoptosis-associated genes), whereas tumor-suppressor miRNAs are downregulated (leading to derepression of oncogenes and proliferation-related genes; ref. 6). Although many miRNAs are differentially expressed in various cancers, the identity of the mRNAs specifically targeted by these miRNAs, functional consequences of miRNA–mRNA pairings and their contributions to cancer pathogenesis remain to be elucidated.

Prostate cancer is now the most frequently diagnosed cancer and the second most common cause of cancer-related deaths in men residing in the United States (7). AAs have among the highest incidence of prostate cancer and mortality attributable to this disease, being 1.6 times more likely to develop prostate cancer, and 2.4 times more likely to die from prostate cancer compared with their EA counterparts (8). Multiple socioeconomic and environmental factors have been postulated to explain the observed prostate cancer health disparities, such as access to care, attitudes toward health care, socioeconomic differences, diet, and differences in the type and aggressiveness of treatment (8). However, adjustment for these factors does not preclude the higher mortality and recurrence rate in AA men and suggests that intrinsic biologic differences exist (9). The application of epidemiology and genomics has revealed biologic factors implicated in prostate cancer health disparities between AA and EA, such as differences in the hormonal milieu of the tumor (10), oncogenic activation (11), and tumor immunobiology (12). More recently, our genomic analysis identified multiple signaling pathways converging on the androgen receptor (AR) to activate transcription of AR-target genes promoting prostate cancer progression and aggressiveness in AA patients (13).

Given the importance of miRNAs in cancer, studies have been forthcoming on the association of miRNAs in prostate cancer pathogenesis. Volinia and colleagues (1) performed large-scale analysis of miRNA expression profiles in 540 samples derived from six types of solid tumors, and demonstrated that 46 miRNAs were differentially expressed when comparing prostate cancer with patient-matched normal prostate (NP), including upregulated let-7d*, miR-17-5p, and miR-21, and downregulated miR-24, miR-29, and miR-128a. An miRNA profiling study by Ozen and colleagues (14) revealed that 76 of the 85 differentially expressed miRNAs were downregulated (such as let-7c, miR-145, and miR-125b) in the prostate cancer clinical samples compared with normal tissues. More recently, Wang and colleagues (15) identified a set of deregulated miRNAs associated with cell-cycle regulation in aggressive prostate cancer by combining miRNA expression profiling and coexpression network analysis. Although these profiling studies have begun to shed light on the involvement of miRNAs in prostate cancer development, questions on the role of miRNAs in prostate cancer disparities still remain. A recent study evaluated the impact of miRNAs contained in the region of 8q24, a genetic risk locus conferring prostate cancer in AAs. However, no empirical evidence of miRNA transcription was found within the 8q24 prostate cancer risk locus (16). In the present study, we applied a systems biology approach, by combining genome-wide miRNA and mRNA expression profiling in prostate cancer patient specimens, miRNA target predictions, and miRNA–mRNA pairing and pathway analyses, to identify the oncogenic signaling pathway most significantly regulated by AA-specific/-enriched mRNAs and miRNA–mRNA pairings. The AA-specific/-enriched miRNA and mRNA elements were also evaluated in AA and EA prostate cancer cell lines for their functional relevance in cell proliferation, invasion, and chemosensitivity to cytotoxic agents.

Acquisition and characteristics of prostate cancer clinical specimens

Tissues were procured from the George Washington University Medical Faculty Associates adhering to IRB approved protocols (IRB#020867), as detailed in Supplementary Materials and Methods. High-quality prostate cancer and patient-matched NP biopsy cores from each of 20 AA and 15 EA patients were collected and processed for the microarray analyses. Prostate cancer cores were determined by pathologist to have Gleason score of 6 to 7 (17 AA and 13 EA) or 8to 9 (3 AA and 2 EA), whereas NP cores were negative for cancer. There was no significant difference between the two racial groups with respect to age (average age for AAs was 62.3 ± 8.2, average age for EAs was 63.3 ± 9.2) and Gleason score (Supplementary Table S1A).

Prostate cancer cell lines

Prostate cancer cell lines were purchased from the ATCC and passaged less than 6 months after receipt/resuscitation. Cell lines were tested and authenticated at the ATCC by short tandem repeat profiling of multiple unique genetic loci (D5S818, D13S317, D7S820, D16S539, vWA, TH01, Amelogenin, TPOX, and CSF1PO).

Gene-expression microarrays

Total RNA was isolated from prostate cancer and patient-matched NP biopsy cores. For mRNA profiling, total RNA (1 μg) from each biopsy core was purified using the RNeasy Micro Kit (Qiagen) and interrogated with the Affymetrix Human Exon 1.0 ST GeneChip. For miRNA profiling, 250 ng of RNA from each biopsy core was isolated using the miRNeasy Kit (Qiagen,) and interrogated with the Agilent Human miRNA microarray V3 (Agilent Technologies). High-quality RNA samples were confirmed on the Agilent 2100 Bioanalyzer (Agilent Technologies). Affymetrix exon array data were normalized by quantile normalization with GC-RMA background correction, and data visualization and statistical analysis were performed by Partek Genomics Suite 6.6 software (Partek) as previously described (13). Raw data from Agilent miRNA microarray analysis were quantile normalized and analyzed in GeneSpring GX program version 12.5 (Agilent Technologies). Identification of statistically significant, differentially expressed/regulated mRNAs and miRNAs was based on ANOVA or the paired t test with a 10% FDR criterion to correct for multiple testing (13). Microarray data can be assessed at GEO using accession numbers GSE64331 and GSE64318 for Affymetrix exon and Agilent miRNA arrays, respectively.

Principal component analysis (PCA) plots and hierarchical clustering of mRNA and miRNA data were performed using the Partek Genomics Suite 6.6. Two-dimensional (2D) hierarchical clustering analysis used average linkage and a Euclidean distance metric.

miRNA–miRNA pairings and pathway analysis

TargetScanHuman 6.2 was used to identify mRNAs predicted to be targets of the ANOVA-defined differentially expressed miRNAs. The list of predicted target mRNAs was intersected with the ANOVA-defined differentially expressed mRNAs to generate a catalog of experimental miRNA–mRNA pairings. Pairings were categorized as having reciprocal (e.g., miRNA up and mRNA down, or miRNA down and mRNA up), positive (i.e., miRNA up and mRNA up), or negative correlations (i.e., miRNA down and mRNA down) in AA prostate cancer vs. NP or AA prostate cancer vs. EA prostate cancer comparisons. The differentially expressed mRNAs not belonging to any pairings are herein referred to as unpaired mRNAs.

Global test (17) [and Gene Set Enrichment Analysis (GSEA; ref. 18) as a secondary confirmatory approach] was implemented to identify statistically significant canonical signaling pathways containing differentially regulated gene sets that may be associated with AA prostate cancer aggressiveness, based on AA prostate cancer versus AA NP, AA prostate cancer versus EA prostate cancer, and EA prostate cancer versus EA NP comparisons (detailed description in Supplementary Materials and Methods). Note that significant genes identified by the Global test and ANOVA may be mutually exclusive. Representative genes in different pathways identified by the Global test were chosen for validation if these genes were also identified by ANOVA and TargetScan prediction analyses as unpaired mRNAs or mRNAs belonging to miRNA–mRNA pairings. The underlying assumption was that genes fulfilling the above criteria would have a greater likelihood of validation success. Validation of differential gene expression was accomplished by quantitative RT-PCR (qRT-PCR) and immunohistochemistry (IHC) in cohorts of patient specimens separate from those used in microarray analysis (Supplementary Table S1B and S1C). Western analysis and functional assays in prostate cancer cell lines were performed to validate predicted reciprocal miRNA–mRNA pairings.

qRT-PCR validation of mRNAs and miRNAs

qRT-PCR validation was performed as previously described (19, 20). qRT-PCR determinations of mRNAs and miRNAs were performed in duplicate and normalized to levels of housekeeping genes EIF1AX and miR-103, respectively. EIF1AX and miR-103 are constitutively expressed and resistant to expression changes (19, 20). qRT-PCR primer pair sequences for mRNA and miRNA determinations are provided in Supplementary Tables S2 and S3, respectively. Sequences to entire mature miRNA are reported in miRBase database (21).

Tissue processing, IHC, and Western blot analysis

Serial sections of formalin-fixed, paraffin-embedded (FFPE) prostate cancer specimens from AA and EA patients with Gleason score 6 to 8 were immunolabeled. Western blot analysis, as previously described (13), was performed on AA and EA prostate cancer cell lines MDA PCa 2b, RC77T/E, VCaP, LNCaP, and PC-3. Details for tissue processing, IHC, image capturing/quantification, and cell line information can be found in Supplementary Materials and Methods.

Antibodies

Antibodies used in IHC assays and Western blotting analysis were rabbit monoclonal antibodies for STAT1 and pFOXO3A (Cell Signaling Technology), FOXO3A (Millipore), and AMACR (Dako), rabbit polyclonal antibody for MCL-1 (Santa Cruz Biotechnology), mouse monoclonal antibodies for p63 (Biocare Medical), and β-actin (Santa Cruz Biotechnology).

Functional analysis of prostate cancer cell lines following miRNA mimic or inhibitor transfections

Prostate cancer cells were transfected with either miRNA mimics or antagomirs using DharmaFECT4 transfection reagent (Dharmacon), according to the manufacturer's protocol. MiR-133a mimic, miR-513c mimic, miR-96 mimic, miR-34a mimic, miR-145 mimic, miR-133a antagomir, miR-513c antagomir, miR-96 antagomir, and nonsense miRNA mimic and antagomir controls were purchased from Life Technologies.

In vitro functional assays, including cell proliferation, apoptosis, and invasion assays were conducted following miRNA mimic/antagomir transfections. Cell proliferation and apoptosis assays were performed using the BrdUrd Cell Proliferation Assay Kit (Calbiochem) and the Apo-ONE Caspase-3/7 Assay Kit (Promega) as described by the manufacturers. Detailed experimental design and protocols can be found in Supplementary Materials and Methods. Matrigel invasion assays were performed as previously described (19, 20).

Microarray analysis reveals differentially expressed mRNAs and miRNAs in AA and EA prostate cancer patient specimens

In an earlier study (13), a total of 70 prostate biopsy cores (20 cancerous and 20 patient-matched NP from AA patients; 15 cancerous and 15 patient-matched NP from EA patients) were subjected to mRNA profiling, and a three-way comparison identified 2,908 significant (ANOVA, 10% FDR multiple test correction) differentially expressed mRNAs. In the present study, we have classified these mRNAs as follows, 433 mRNAs are “AA-enriched” (significantly overexpressed in AA) and 755 mRNAs are “AA-depleted” (significantly underexpressed in AA) based on the AA prostate cancer versus EA prostate cancer comparison (Supplementary Table S4). Another 980 mRNAs (up or down) are defined as “AA-specific” based on the AA prostate cancer versus AA NP comparison (and not significant in EA prostate cancer vs. EA NP), whereas 740 mRNAs are “EA-specific” based on EA prostate cancer versus EA NP (and not significant in AA prostate cancer vs. AA NP, Supplementary Table S4). PCA and 2D hierarchical clustering demonstrated clear separation and consistency of gene-expression profiles in the three separate comparisons (Fig. 1A).

Figure 1.

mRNA and miRNA expression profiling of prostate cancer (PCa) specimens and patient-matched normal tissues derived from AA and EA patients. A, prostate cancer plots and hierarchical 2D clustering of mRNA expression in AA prostate cancer versus EA prostate cancer, and prostate cancer versus patient-matched normal tissue. B, prostate cancer plots and hierarchical clustergrams of miRNA expression in AA prostate cancer versus EA prostate cancer, and prostate cancer versus patient-matched normal tissue. For both A and B, samples are in rows, and mRNAs or miRNAs are in columns. Plots demonstrated clear separation and consistency of mRNA and miRNA expression profiles in group comparisons. For mRNA profiling, n = 20, 20, 15, and 15 for AA prostate cancer, AA-matched normal, EA prostate cancer, and EA-matched normal, respectively. For miRNA profiling, n = 14, 14, 13, and 13 for AA prostate cancer, AA-matched normal, EA prostate cancer, and EA-matched normal, respectively.

Figure 1.

mRNA and miRNA expression profiling of prostate cancer (PCa) specimens and patient-matched normal tissues derived from AA and EA patients. A, prostate cancer plots and hierarchical 2D clustering of mRNA expression in AA prostate cancer versus EA prostate cancer, and prostate cancer versus patient-matched normal tissue. B, prostate cancer plots and hierarchical clustergrams of miRNA expression in AA prostate cancer versus EA prostate cancer, and prostate cancer versus patient-matched normal tissue. For both A and B, samples are in rows, and mRNAs or miRNAs are in columns. Plots demonstrated clear separation and consistency of mRNA and miRNA expression profiles in group comparisons. For mRNA profiling, n = 20, 20, 15, and 15 for AA prostate cancer, AA-matched normal, EA prostate cancer, and EA-matched normal, respectively. For miRNA profiling, n = 14, 14, 13, and 13 for AA prostate cancer, AA-matched normal, EA prostate cancer, and EA-matched normal, respectively.

Close modal

We also sought to investigate the relationship between miRNA and mRNA profiles in the same cohort of patients. Of the original 70 biopsy cores used for mRNA expression analysis, 54 provided sufficient material for miRNA expression profiling (14 cancerous and 14 patient-matched NP from AA patients; 13 cancerous and 13 patient-matched NP from EA patients). MiRNA profiling revealed 10, 33, and 29 miRNAs that were differentially expressed (ANOVA or paired t test, 10% FDR, fold change ≥ 1.5) between AA prostate cancer versus EA prostate cancer, AA prostate cancer versus AA NP and EA prostate cancer versus EA NP, respectively. Eleven of these miRNAs represent race-independent noncoding RNAs (miRNAs found significant in both AA prostate cancer vs. AA NP and EA prostate cancer vs. EA NP comparisons), along with 2 AA-enriched, 8 AA-depleted, 22 AA-specific and 18 EA-specific miRNAs (Supplementary Table S5). Prostate cancer and 2D hierarchical clustering demonstrated clear separation of miRNA profiles (Fig. 1B). In summary, we postulate that AA-enriched, AA-depleted, and race-specific miRNAs and mRNAs (but not race-independent mRNAs and miRNAs) may be associated with the biologic component of prostate cancer disparities.

Novel reciprocal miRNA–mRNA pairings and dysregulated-unpaired mRNAs in oncogenic signaling pathways promoting prostate cancer disparities

AA-enriched/-depleted, AA-specific and EA-specific miRNAs were analyzed by TargetScanHuman 6.2 (implemented in IPA miRNA Target Filter), resulting in the identification of 3,153, 5,244, and 3,812 predicted target mRNAs, respectively. We focused attention on those miRNA–mRNA pairings with the following criteria: (i) the predicted target mRNA was also differentially expressed in our microarray analysis (13), and (ii) the miRNA exhibited a reciprocal expression relationship with its target mRNA (“up–down” or “down–up”). Using these criteria, we have compiled 150 reciprocal miRNA–mRNA pairings in AA prostate cancer versus EA prostate cancer, 103 pairings in AA prostate cancer versus AA-matched NP and 137 pairings in EA prostate cancer versus EA-matched NP (Supplementary Table S6).

In a separate analysis to identify biologic pathways most significantly associated with AA prostate cancer aggressiveness, we applied the Global test to our gene-expression data from prostate biopsy cores. The Global test is a permutation-based approach, coupled with a penalized logistic regression model, to identify gene sets in pathways most significantly associated to clinical phenotypes/outcomes (17). Using this approach, we identified 124, 106, and 137 significant KEGG-annotated signaling pathways (FDR < 0.05) in AA prostate cancer versus EA prostate cancer, AA prostate cancer versus AA NP, and EA prostate cancer versus EA NP comparisons, respectively (Supplementary Table S7). Among the significant KEGG oncogenic pathways associated with AA prostate cancer were ERBB, MTOR, WNT, JAK-STAT, TGFβ, P53, and VEGF. Noteworthy was the ERBB pathway in AA prostate cancer, where a great majority of pathway genes (mRNAs) identified as significant by the Global test were upregulated in AA prostate cancer versus EA prostate cancer and AA prostate cancer versus AA NP comparisons (Fig. 2A; Supplementary Table S7). Conversely, the vast majority of significant genes in the ERBB pathway of EA prostate cancer were downregulated according to Global testing of EA prostate cancer versus AA prostate cancer and EA prostate cancer versus EA NP comparisons (Fig. 2B; Supplementary Table S7). Similar findings were obtained when analyzing our gene-expression data by the GSEA approach (Supplementary Table S8; ref. 18). Collectively, our pathway analysis suggests that differential gene regulation of ERBB signaling components in AA versus EA prostate cancer may play a critical role toward promoting prostate cancer disparities. A finding that may be particularly relevant given the well-developed targeted therapies for this critical oncogenic pathway (22, 23).

Figure 2.

The ERBB signaling pathway is highly activated in AA prostate cancer (PCa) specimens. Differentially expressed mRNAs [identified by Global test or Global test plus ANOVA (indicated by asterisk)] and miRNAs (identified by ANOVA or paired t test) populating the ERBB signaling pathway in AA prostate cancer (A) and EA prostate cancer (B). Upregulated (red) and downregulated (green) miRNAs with underline representing population-specific miRNAs, whereas miRNAs not underlined represent population-enriched (red) or -depleted (green) miRNAs. The same coloring and underlining scheme is used for differentially expressed mRNAs. The ERBB pathway in AA prostate cancer (A) is more highly activated compared with EA prostate cancer (B) as determined by GO-Elite. Eight novel reciprocal miRNA–mRNA pairings are highlighted, including miR-133a/MCL1, miR-96/FOXO3A, miR-513c/STAT1, miR-34a/PPP2R2A, miR-145/ITPR2, miR-145/MKK4, miR-634/MKK4, and miR-129*/MKK4. MiRNAs listed in boxes represent the population-specific (underlined) or -enriched/-depleted miRNAs predicted to target genes in the ERBB signaling pathway belonging to positively or negatively correlated pairings or nondifferentially expressed targets (see Supplementary Table S6).

Figure 2.

The ERBB signaling pathway is highly activated in AA prostate cancer (PCa) specimens. Differentially expressed mRNAs [identified by Global test or Global test plus ANOVA (indicated by asterisk)] and miRNAs (identified by ANOVA or paired t test) populating the ERBB signaling pathway in AA prostate cancer (A) and EA prostate cancer (B). Upregulated (red) and downregulated (green) miRNAs with underline representing population-specific miRNAs, whereas miRNAs not underlined represent population-enriched (red) or -depleted (green) miRNAs. The same coloring and underlining scheme is used for differentially expressed mRNAs. The ERBB pathway in AA prostate cancer (A) is more highly activated compared with EA prostate cancer (B) as determined by GO-Elite. Eight novel reciprocal miRNA–mRNA pairings are highlighted, including miR-133a/MCL1, miR-96/FOXO3A, miR-513c/STAT1, miR-34a/PPP2R2A, miR-145/ITPR2, miR-145/MKK4, miR-634/MKK4, and miR-129*/MKK4. MiRNAs listed in boxes represent the population-specific (underlined) or -enriched/-depleted miRNAs predicted to target genes in the ERBB signaling pathway belonging to positively or negatively correlated pairings or nondifferentially expressed targets (see Supplementary Table S6).

Close modal

Next, we mapped the population-associated miRNAs and miRNA–mRNA pairings (Supplementary Tables S4, S5, and S6) onto the ERBB signaling pathway (Fig. 2). Altogether, 17 AA-specific miRNAs (miR-15b, miR-20a, miR-25, miR-148a, miR-203, miR-129*, miR-659, miR-125-3p, miR-513c, miR-671-3p, miR-887, miR-145, miR-130b, miR-634, miR-767-3p, miR-1225-3p, and miR-197-3p), 2 AA-enriched miRNAs (miR-96 and miR-130b) and 4 AA-depleted miRNAs (miR-133a, miR-758, miR-34a, and miR-99b) were predicted to target 56 of 85 signaling genes of the ERBB pathway in AA prostate cancer (Fig. 2A; Supplementary Table S6), leading to a projected overall activation of oncogenic signaling based on GO-Elite analysis (24). Of the reciprocal miRNA–mRNA pairings in the ERBB pathway of AA prostate cancer (Fig. 2A), 14 were novel (i.e., predicted miRNA targeting of mRNA not validated in literature), namely miR-133a/MCL1 (down–up), miR-96/PPP2R3A (up–down), miR-133a/PPP2R2D (down–up), miR-767-3p/MTOR (down–up), miR-1225-3p/MTOR (down–up), miR-129*/MTOR (down–up), miR-129*/PIK3AP1 (down–up), miR-96/COL5A1 (up–down), miR-34a/IKBKE (down–up), miR-129*/IKBKB (down–up), mi-933/IKBKB (down–up), miR-145/MKK4 (down–up), miR-634/MKK4 (down–up), and miR-129*/MKK4 (down–up; Supplementary Table S6).

In contrast with the projected activation of ERBB signaling in AA prostate cancer, EA prostate cancer was comprised mostly of downregulated oncogenes and upregulated EA-specific/-enriched miRNAs (predicted to target oncogenes) that were projected by GO-Elite to restrain ERBB pathway activity (Fig. 2B). Note that AA- and EA-specific miRNAs do not overlap by definition. Hence, the inverse expression pattern of AA- and EA-specific/enriched/depleted miRNAs targeting different components of the ERBB signaling pathway likely plays a critical role in the differential aggressiveness of prostate cancer progression in the two racial populations.

qRT-PCR validation in AA and EA prostate cancer biopsy specimens

qRT-PCR validation assays were performed in a second cohort of prostate cancer biopsy specimens from patients to validate our microarray analysis (Supplementary Table S1B). We specifically reassessed a combination of 30 differentially expressed miRNAs and mRNAs (identified as significant by both Global test and ANOVA; the exception being BCL2L11 that was identified as significant by ANOVA only) residing in the ERBB signaling pathway, as well as four additional signaling pathways [i.e., non–small cell lung cancer (NSCLC) signaling, the JAK–STAT pathway, tight junction signaling, phosphatidylinositol signaling]. A comparison of the microarray and qRT-PCR results revealed high concordance (28 of 30) in our expression measurements. Successful validations included AA-enriched and -depleted mRNAs (Fig. 3A), AA-enriched and -depleted miRNAs (Fig. 3B), population-specific mRNAs (Fig. 3C) and population-specific miRNAs (Fig. 3D). Encompassed within the validations were the novel reciprocal miRNA–mRNA pairings miR-133a/MCL1 (down–up; target mRNA significant in the ERBB pathway by the Global test), miR-96/FOXO3A (up–down; NSCLC signaling), miR-513c/STAT1 (down–up; JAK–STAT pathway), miR-34a/PPP2R2A (down–up; tight junction signaling), miR-145/ITPR2 (down–up; phosphatidylinositol signaling) and miR-145/MKK4 (down–up; ERBB pathway; Figs. 2A and 3). Interestingly, four of the target mRNAs (FOXO3A, STAT1, PPP2R2A, and ITPR2) in these pairings are also known to participate downstream of ERBB signaling, and hence included in Fig. 2A for illustration (25–29).

Figure 3.

qRT-PCR validation of population-enriched/-depleted and -specific mRNAs and miRNAs in AA and EA prostate cancer (PCa). A, qRT-PCR validation of differentially expressed mRNAs in AA prostate cancer versus EA prostate cancer. B, qRT-PCR validation of differentially expressed miRNAs in AA prostate cancer versus EA prostate cancer. C, qRT-PCR validation of population-specific mRNAs. D, qRT-PCR validation of population-specific miRNAs. The expression levels of mRNA or miRNAs from AA and EA patients are presented as Box-and-Whiskers plots (in A–D). Box: top quantile, median and bottom quantile. Whiskers: top extreme (90 percentile of the dataset) and bottom extreme (10 percentile of the dataset). Dot plots represent the relative expression levels of mRNA or miRNA from individual patient samples; *, P < 0.05 using the Student t test (n = 6–9 independent experiments in A and B), or a paired Student t test (n = 5–8 independent experiments in C and D).

Figure 3.

qRT-PCR validation of population-enriched/-depleted and -specific mRNAs and miRNAs in AA and EA prostate cancer (PCa). A, qRT-PCR validation of differentially expressed mRNAs in AA prostate cancer versus EA prostate cancer. B, qRT-PCR validation of differentially expressed miRNAs in AA prostate cancer versus EA prostate cancer. C, qRT-PCR validation of population-specific mRNAs. D, qRT-PCR validation of population-specific miRNAs. The expression levels of mRNA or miRNAs from AA and EA patients are presented as Box-and-Whiskers plots (in A–D). Box: top quantile, median and bottom quantile. Whiskers: top extreme (90 percentile of the dataset) and bottom extreme (10 percentile of the dataset). Dot plots represent the relative expression levels of mRNA or miRNA from individual patient samples; *, P < 0.05 using the Student t test (n = 6–9 independent experiments in A and B), or a paired Student t test (n = 5–8 independent experiments in C and D).

Close modal

qRT-PCR assessment of population-specific prostate cancer cell lines

We also assessed the expression of AA-enriched and -depleted miRNAs and mRNAs (depicted in Fig. 3) in a panel of prostate cancer cell lines derived from AA (MDA PCa 2b, RC77T/E) and EA patients (VCaP, LNCaP, and PC-3; see Supplementary Materials and Methods). There was strong overall agreement between the microarray data from patient specimens and qRT-PCR results of prostate cancer cell lines. Specifically, AA-depleted mRNAs (FOXO3A and BCL2L11) tended to be underexpressed in AA versus EA prostate cancer cell lines, and AA-enriched mRNAs (PIK3CB, PPP2R2A, MCL1, 14-3-3ϵ, ITGB5, and STAT1) tended to be overexpressed in AA versus EA prostate cancer cell lines (Fig. 4A). An analogous consistency was observed for the miRNAs (Fig. 4B). Again, contained within these validations were the novel reciprocal miRNA–mRNA pairings miR-133a/MCL1 (down–up), miR-96/FOXO3A (up–down), and miR-513c/STAT1 (down–up; Fig. 4C). As a final consistency check, miRNA–mRNA pairings were found to be consistent with Western blot analysis where FOXO3A was underexpressed, whereas MCL-1 and STAT1 were overexpressed in AA versus EA prostate cancer cell lines (Fig. 4D).

Figure 4.

Population-specific prostate cancer (PCa) cell lines are in vitro cell models for prostate cancer disparities. A, qRT-PCR validation of microarray mRNA data in population-specific prostate cancer cell lines. B, qRT-PCR validation of microarray miRNA data in population-specific prostate cancer cell lines. C, heat maps demonstrating inverse correlation between expression of miRNAs and mRNAs in AA prostate cancer versus EA prostate cancer comparisons. D, Western blot analysis reveals that protein expression correlates with mRNA expression in population-specific cell lines. Relative protein level was normalized to β-actin. Representative blots of 4 to 6 independent determinations. Data (in A, B, and D), mean ± SEM, with *, P < 0.05 using an unpaired Student t test, n = 4–6 independent experiments for each cell line. Means were derived by combining results from AA cell lines versus EA cell lines.

Figure 4.

Population-specific prostate cancer (PCa) cell lines are in vitro cell models for prostate cancer disparities. A, qRT-PCR validation of microarray mRNA data in population-specific prostate cancer cell lines. B, qRT-PCR validation of microarray miRNA data in population-specific prostate cancer cell lines. C, heat maps demonstrating inverse correlation between expression of miRNAs and mRNAs in AA prostate cancer versus EA prostate cancer comparisons. D, Western blot analysis reveals that protein expression correlates with mRNA expression in population-specific cell lines. Relative protein level was normalized to β-actin. Representative blots of 4 to 6 independent determinations. Data (in A, B, and D), mean ± SEM, with *, P < 0.05 using an unpaired Student t test, n = 4–6 independent experiments for each cell line. Means were derived by combining results from AA cell lines versus EA cell lines.

Close modal

Immunohistochemical assessment of MCL-1, STAT1, and FOXO3A in AA and EA prostate cancer specimens

Next, we examined protein expression of MCL-1, STAT1, and FOXO3A by immunohistochemical examination of archived FFPE prostate cancer specimens from AA and EA patients, representing a third cohort with associated Gleason scores ranging from 6 to 9 (Fig. 5A and B; Supplementary Table S1C and Fig. S1). To ensure that MCL-1, STAT1, and FOXO3A protein expression was indeed present in cancerous cells, another series of IHC was performed where our proteins of interest were examined along with alpha-methylacyl CoA racemase (AMACR; positive control for cancer cells) and p63 (marker for NP basal cells) in serial sections (30). IHC results demonstrated overexpression of MCL-1 and STAT1 in the cytoplasm of AA versus EA cancerous cells, and that the equivalent regions in adjacent sections stained strongly for AMACR but negative for p63 (Fig. 5B). For FOXO3A, staining was greater in the nuclei of EA versus AA cancerous cells, and in the equivalent regions of adjacent sections there was strong cytoplasmic staining for AMACR and negative staining for p63 in cancerous cells (Fig. 5B). In summary, our IHC findings in patient specimens perfectly match the Western results from prostate cancer cell lines.

Figure 5.

IHC reveals differential protein expression in AA prostate cancer (PCa) versus EA prostate cancer. A, paraffin-embedded tissue sections of human prostate cancer specimens show strong MCL-1 and STAT1 expression in the cytoplasm of cancer cells of AA specimens, whereas FOXO3A immunoreactivity was detected in cancer cell nuclei of EA specimens. Images shown are representative of 13 AA and 13 EA specimens from different patients. B, the intensities of cytoplasmic MCL-1 and STAT1, and nuclear FOXO3A were quantified by using the ratio of total intensity of immunoreactive MCL-1, STAT1, or FOXO3A over the total area of cells in the images. Data, box plots of n = 13 AA or EA samples, with *, P < 0.05 using the Student t test. C, serial FFPE sections derived from AA and EA prostate cancer patients were immunostained for AMACR (a prostate cancer marker), p63 (a normal basal cell marker) and the protein of interest (MCL-1, STAT1, or FOXO3A). Enlarged pictures (rectangles as indicated) enhance the nuclear or cytoplasmic distribution of these proteins at the cellular level.

Figure 5.

IHC reveals differential protein expression in AA prostate cancer (PCa) versus EA prostate cancer. A, paraffin-embedded tissue sections of human prostate cancer specimens show strong MCL-1 and STAT1 expression in the cytoplasm of cancer cells of AA specimens, whereas FOXO3A immunoreactivity was detected in cancer cell nuclei of EA specimens. Images shown are representative of 13 AA and 13 EA specimens from different patients. B, the intensities of cytoplasmic MCL-1 and STAT1, and nuclear FOXO3A were quantified by using the ratio of total intensity of immunoreactive MCL-1, STAT1, or FOXO3A over the total area of cells in the images. Data, box plots of n = 13 AA or EA samples, with *, P < 0.05 using the Student t test. C, serial FFPE sections derived from AA and EA prostate cancer patients were immunostained for AMACR (a prostate cancer marker), p63 (a normal basal cell marker) and the protein of interest (MCL-1, STAT1, or FOXO3A). Enlarged pictures (rectangles as indicated) enhance the nuclear or cytoplasmic distribution of these proteins at the cellular level.

Close modal

Disruption of AA-specific and -enriched reciprocal miRNA–mRNA pairings affect cell proliferation, antiapoptosis, and invasion

To more firmly establish a causal link among our reciprocal miRNA–mRNA pairings, a series of miRNA mimics and antagomirs were transfected into population-specific prostate cancer cell lines and the protein products of predicted target mRNAs were measured by Western blot analysis. Two AA prostate cancer lines (RC77T/E, MDA PCa 2b) and 2 EA prostate cancer lines (LNCaP and PC-3) were chosen for in vitro functional assays on the basis of congruent qRT-PCR, Western and immunohistochemical findings (Figs. 4 and 5). Transfection of a miR-133a mimetic into AA and EA lines led to a downregulation of MCL-1 protein compared with cells transfected with nonsense control RNA (Fig. 6A, left). Conversely, miR-133a antagomir transfection into AA and EA lines led to an upregulation of MCL-1 protein compared with nonsense control (Fig. 6A, right). This antagomir-mediated upregulation in prostate cancer cells was anticipated given the “converse” mimetic-induced downregulation in prostate cancer cells. Analogous confirmatory findings were also demonstrated for AA-enriched miR-96 (predicted target FOXO3A) and downregulated AA-specific miR-513c (predicted target STAT1; Fig. 6A). Taken together, our in vitro mimic/antagomir manipulation of population-specific prostate cancer cell lines was consistent with observations in patient specimens (see Figs. 3 and 5), providing strong evidence of a causal link between our reciprocal miRNA–mRNA pairings.

Figure 6.

Functional validation of miR-133a/MCL1, miR-513c/STAT1, and miR-96/FOXO3A pairs in prostate cancer (PCa) aggressiveness. A, overexpression of miR-133a mimic, miR-513c mimic, or miR-96 antagomir in prostate cancer cell lines resulted in downregulation of MCL-1 and STAT1, and upregulation of FOXO3A, respectively. In contrast, inhibition of miR-133a or miR-513c with antagomirs or overexpression of miR-96 mimic resulted in upregulation of MCL-1 and STAT1, and downregulation of FOXO3A. AA lines are MDA PCa 2B, RC77T/E, and EA cell lines are LNCaP and PC-3. Representative Western blots from three to six independent experiments. B, BrdUrd-labeled cell proliferation assays of prostate cancer cell lines transfected with miR-133a mimic, miR-513c mimic or miR-96 antagomir, or “converse” antagomir/mimic treatments (miR-133a antagomir, miR-513c antagomir, or miR-96 mimic) were compared with cells treated with NS (nonsense scrambled negative control). Data, mean ± SEM of n = 3–4 independent experiments, with *, P < 0.05 using ANOVA and Tukey post hoc test. C, apoptosis assays in population-specific prostate cancer cell lines transfected with mimics or antagomirs. Apoptosis activity was assayed by measuring caspase-3/7 activity using the Apo-ONE Kit (Promega), and the data were normalized to caspase-3/7 level of vehicle-treated NS control. Data, mean ± SEM for n = 3–6 independent experiments, with P < 0.05 using ANOVA and Tukey post hoc test comparing mimic or antagomir transfection plus vehicle treatment to NS transfection plus vehicle (*), or mimic or antagomir transfection plus vehicle to mimic or antagomir transfection plus docetaxel (#). D, prostate cancer cells transfected with miR-96 antagomir or miR-513c mimic were significantly less invasive compared with NS control-treated cells. Data, mean ± SEM of n = 4–6 independent experiments, with *, P < 0.05 using ANOVA and Holmes post hoc test; Antag, antagomir.

Figure 6.

Functional validation of miR-133a/MCL1, miR-513c/STAT1, and miR-96/FOXO3A pairs in prostate cancer (PCa) aggressiveness. A, overexpression of miR-133a mimic, miR-513c mimic, or miR-96 antagomir in prostate cancer cell lines resulted in downregulation of MCL-1 and STAT1, and upregulation of FOXO3A, respectively. In contrast, inhibition of miR-133a or miR-513c with antagomirs or overexpression of miR-96 mimic resulted in upregulation of MCL-1 and STAT1, and downregulation of FOXO3A. AA lines are MDA PCa 2B, RC77T/E, and EA cell lines are LNCaP and PC-3. Representative Western blots from three to six independent experiments. B, BrdUrd-labeled cell proliferation assays of prostate cancer cell lines transfected with miR-133a mimic, miR-513c mimic or miR-96 antagomir, or “converse” antagomir/mimic treatments (miR-133a antagomir, miR-513c antagomir, or miR-96 mimic) were compared with cells treated with NS (nonsense scrambled negative control). Data, mean ± SEM of n = 3–4 independent experiments, with *, P < 0.05 using ANOVA and Tukey post hoc test. C, apoptosis assays in population-specific prostate cancer cell lines transfected with mimics or antagomirs. Apoptosis activity was assayed by measuring caspase-3/7 activity using the Apo-ONE Kit (Promega), and the data were normalized to caspase-3/7 level of vehicle-treated NS control. Data, mean ± SEM for n = 3–6 independent experiments, with P < 0.05 using ANOVA and Tukey post hoc test comparing mimic or antagomir transfection plus vehicle treatment to NS transfection plus vehicle (*), or mimic or antagomir transfection plus vehicle to mimic or antagomir transfection plus docetaxel (#). D, prostate cancer cells transfected with miR-96 antagomir or miR-513c mimic were significantly less invasive compared with NS control-treated cells. Data, mean ± SEM of n = 4–6 independent experiments, with *, P < 0.05 using ANOVA and Holmes post hoc test; Antag, antagomir.

Close modal

The oncogenic consequences of disrupting steady-state expression of our prototype reciprocal pairings were assessed in AA lines RC77T/E and MDA PCa 2b, and EA lines LNCaP and PC-3. In the first set of functional assays, prostate cancer lines were transfected with a series of mimics, antagomirs, or nonsense control RNA and tested for proliferative activity using a bromodeoxyuridine (BrdUrd)-labeling assay. In each case, the miR-133a mimic, miR-513c mimic, and miR-96 antagomir significantly suppressed proliferation of the AA and EA prostate cancer cell lines compared with nonsense control (Fig. 6B, top). Conversely, the majority of “converse” antagomir/mimic treatments (miR-133a antagomir, miR-513c antagomir, and miR-96 mimic) significantly enhanced proliferation in both AA lines and EA line LNCaP, as anticipated (Fig. 6B, bottom). Interestingly, EA line PC-3 was completely resistant to the proliferation-inducing effects of all three “converse” antagomir/mimic treatments (Fig. 6B).

Next, apoptotic sensitivity in the absence and presence of 11 nmol/L docetaxel, a cytotoxic agent used in prostate cancer chemotherapy (31), was assessed in prostate cancer cell lines by caspase-3/7 activity assay. In the absence of any antagomir or mimic treatment, AA lines RC77T/E and MDA PCa 2b were chemoresistant to docetaxel-induced apoptosis (see nonsense control transfected cells in Fig. 6C, top). In contrast, docetaxel treatment alone significantly induced apoptosis in EA lines LNCaP and PC-3 (see nonsense control transfected cells in Fig. 6C, top). In the absence of docetaxel treatment, transfection of AA and EA cell lines with the miR-133a mimic, miR-513c mimic, or miR-96 antagomir precipitated a generalized (exception being miR-513c mimic-transfected RC77T/E and LNCaP cells) and significant increase in apoptosis compared with nonsense control transfected cells (Fig. 6C, top). Strikingly in AA prostate cancer cells (but not in EA cells), the combination of a mimic or antagomir treatment with docetaxel treatment resulted in apoptotic activity that was greater than either treatment alone, suggesting that disruption of key miRNAs sensitized cells to docetaxel (Fig. 6C, top). Interestingly, the “converse” antagomir/mimic treatments (miR-133a antagomir, miR-513c antagomir, and miR-96 mimic) in the absence of docetaxel had the effect of rendering AA lines, but not EA lines, more resistant to apoptosis (Fig. 6C, bottom). On the basis of the proliferative and apoptotic findings, EA prostate cancer lines compared with AA lines appear to be less susceptible to the oncogenic-promoting effects of the reciprocal pairs miR-133a/MCL1, miR-96/FOXO3A, and miR-513c/STAT1.

Finally, the consequences of disrupting steady-state expression of our prototype reciprocal pairings on the invasive activity of prostate cancer cell lines were assessed by Matrigel assay. Both miR-513c mimic and miR-96 antagomir treatments in AA lines RC77T/E and MDA PCa 2b, and EA lines LNCaP and PC-3 resulted in a significant decrease in invasive activity (Fig. 6D), though we cannot discount the possibility that this decrease may be due in part to decreased proliferative activity (Fig. 6B, top). In an attempt to identify reciprocal pairings that modulate invasion without affecting proliferation, we tested two additional down–up pairings (miR-145/ITPR2 and miR-34a/PPP2R2A in the EGFR–PI3K–AKT pathway) in the AA prostate cancer lines. Western blot analysis confirmed a causal link for these two reciprocal pairings, as transfection with either the miR-145 mimic or miR-34a mimic in AA lines resulted in a reduction of ITPR2 or PPP2R2A protein levels, respectively (Supplementary Fig. S2A). As shown in Supplementary Fig. S2B, the miR-145 mimic affected both proliferation and invasion, whereas the miR-34a mimic was associated with a significant decrease in invasion and had no effect on proliferation in both AA prostate cancer cell lines. Taken together, these findings support the notion that depletion of miR-133a (leading to upregulation of MCL1), miR-513c (upregulation of STAT1), miR145 (upregulation of ITPR2), and miR-34a (upregulation of PPP2R2A), coupled with enrichment of miR-96 (downregulation of FOXO3A) collectively drives proliferation, chemoresistance and/or invasion in AA prostate cancer cells.

In this study, we performed an integrated analysis of differential miRNA and mRNA expression profiles in prostate cancer and NP specimens derived from AA and EA patients. Our goal was to identify significant oncogenic signaling pathways that are populated with AA-specific/-enriched reciprocal miRNA–mRNA pairs. Emphasis was placed on cataloging novel reciprocal pairs (i.e., predicted miRNA targeting of the mRNA has yet to be experimentally validated). The underlying hypothesis being that these novel reciprocal pairs may play a mechanistic role in prostate cancer disparities (i.e., more aggressive nature of AA prostate cancer), which could be assessed by systematically disrupting reciprocal pairs with mimic/antagomir treatment of population-specific prostate cancer cell lines and testing for a loss (or gain) of oncogenic function. To date, the integrated analysis of miRNA–mRNA pairs has been limited to a handful of prostate cancer studies (32, 33) and none have been related to prostate cancer disparities.

There are a number of available miRNA-target mRNA prediction algorithms (34). However, it is estimated that up to 40% of all miRNA-target mRNA predictions are false positives (35), representing a major obstacle in the identification of true miRNA–mRNA interacting partnerships with functional consequences in cancer. An approach exploited by this study was to incorporate both a sequence-based algorithm for miRNA target predictions and focusing on miRNA–mRNA predictions exhibiting reciprocal differential expression profiles (up–down and down–up). Such a strategy has been demonstrated to provide more accurate predictions (35). A total of 390 reciprocal pairings were identified in prostate cancer and NP specimens from AA and EA patients. These pairs (along with unpaired differentially expressed miRNAs and mRNAs) were found populated in 19 and 18 significant cancer signaling pathways from the perspective of AA and EA prostate cancer, respectively.

ERBB signaling pathway in prostate cancer disparities

The ERBB signaling pathway is regarded as a critical oncogenic signaling pathway in cancer, as mutations and/or overexpression of the EGFR and mutations in multiple PI3K isoforms are frequently detected in various types of cancers, including prostate, head and neck, renal, lung, breast, colon, ovarian, glioma, pancreas, and bladder cancers (22, 23). In terms of prostate cancer disparities, EGFR overexpression has been shown to be significantly associated with AA patients (11). Our findings suggest that 18 reciprocal miRNA–mRNA pairs populating the EGFR–PI3K–AKT signaling pathway in AA prostate cancer, and likely working in concert with overexpressed EGFR (11), drive AA prostate cancer.

MiR-513c/STAT1 (down–up) represented a novel predicted pairing, and miR-513c has previously been shown to be downregulated in neuroendocrine lung tumors (36). However, the role of miR-513c in cancer and the identification of its target mRNA(s) have remained undetermined. Our results demonstrate for the first time that STAT1 serves as a target of miR-513c. The STAT1 protein is a transcription factor and its overexpression in prostate cancer cells has been associated with docetaxel resistance (37). Interestingly, the AA prostate cancer cell lines investigated in this study were resistant to docetaxel-induced apoptosis, but became sensitized upon treatment with a miR-513c mimic that downregulated STAT1. Additional functions of the miR-513c/STAT1 pair in AA prostate cancer cells include proliferation and invasion, as disruption of this pairing with a miR-513c mimic resulted in a loss of proliferative and invasive activities. The role of miR-513c/STAT1 in driving AA prostate cancer was further supported by experiments using a “converse” targeting approach (i.e., miR-513c antagomir) in EA prostate cancer cell lines, resulting in STAT1 upregulation and a more aggressive phenotype reminiscent of the AA prostate cancer lines (i.e., increased proliferation and chemoresistance).

Downregulation of miR-133a has been observed in various cancers (38), acting as a tumor suppressor by targeting multiple oncogenes, such as FSCN1, MMP14, LASP1, EGFR, IGF1R, and GSTP1 (39). In our study, MCL1 was identified as a novel target of miR-133a, and overexpression of a miR-133a mimic in prostate cancer cell lines led to a downregulation of MCL-1 protein and a corresponding decrease in proliferative activity, as well as loss of chemoresistance to docetaxel. MCL-1 has been demonstrated to be overexpressed in prostate cancer and is linked to higher Gleason scores and increased bone metastasis in prostate cancer patients (29). As was the case for miR-513c/STAT1, we demonstrated a role of miR-133a/MCL1 in driving AA prostate cancer by using a “converse” targeting approach (i.e., miR-133a antagomir) in EA prostate cancer cell lines, resulting in MCL-1 upregulation and a more aggressive phenotype, again reminiscent of the AA prostate cancer lines.

Upregulation of miR-96 has been observed in lung, breast, bladder, and colorectal cancers (40). MiR-96 promotes cell proliferation by targeting the FOXO1 gene, encoding a transcription factor, in breast and prostate cancer (41, 42); and enhances proliferative, invasive, and migratory activity by targeting FOXO1 and RECK in breast cancer, bladder, and lung cancers (43, 44). In this study, we further demonstrated that FOXO3A targeted by miR-96 in prostate cancer, confirming a previous observation in breast cancer (45). Disruption of miR-96/FOXO3A (up–down) in AA prostate cancer cell lines with a miR-96 antagomir resulted in FOXO3A protein upregulation and a corresponding decrease in proliferative, invasive, and chemoresistant activities. Conversely, introduction of a miR-96 mimic into EA prostate cancer cell lines had the opposite effect by downregulating FOXO3A protein and promoting proliferation and chemoresistance. In essence, the EA prostate cancer cell lines transformed into a more aggressive AA prostate cancer–like phenotype. Taken together, these findings are consistent with the known tumor-suppressor effect of FOXO3A in prostate cancer (46).

Another intriguing miRNA–mRNA pair residing in the ERBB signaling pathway of AA prostate cancer is miR-145/ITPR2 (down–up). Recent genome-wide association studies have implicated the inositol 1,4,5-triphosphate receptor type 2 (ITPR2) gene as a novel risk locus for renal cell carcinoma (47, 48). MiR-145 has been implicated as a tumor-suppressive miRNA as it is downregulated in different cancers and its expression has been associated with an inhibition of prostate cancer cell invasion and migration in vitro (49). Our findings link miR-145 and ITPR2 for the first time as a functional reciprocal pair that promotes invasion and proliferation in AA prostate cancer.

It should also be noted that AA prostate cancer was associated with a large number of upregulated oncogenes (such as ITGA5, PIK3CB, PIK3AP, ITPR2, STAT1, CSNK2A1, MKK4, 14-3-3ϵ, MTOR, and MCL1) as well as dysregulated unpaired miRNAs that are unique to AA prostate cancer (e.g., AA-specific/depleted miRNAs) and computationally predicted to target EGFR–PI3K–AKT signaling components (such as EGFR, AKT3, GSK3, JAK1, JUN, and KRAS) leading to pathway activation. Conversely, our analysis identified an equally large number of dysregulated oncogenes plus unpaired miRNAs that were specific to EA prostate cancer and computationally predicted to target a different set of EGFR–PI3K–AKT signaling components leading to pathway suppression. Also noteworthy, unpaired AA-specific miR-767-3p (downregulated in AA prostate cancer vs. AA NP) and unpaired EA-specific miR-195 (upregulated in EA prostate cancer vs. EA NP) were both predicted to target the EGFR mRNA, resulting in an anticipated up- and downregulation of the EGFR protein, respectively. This finding would be consistent with the observed racial disparity of EGFR overexpression in AA prostate cancer (11). Although our analysis has focused on five reciprocal miRNA–mRNA pairings, it is important to stress that the miRNAs in these pairings would be expected to coordinately target other mRNAs (i.e., MCL1, FSCN1, MMP14, LASP1, EGFR, IGF1R and GSTP1 by miR-133a, and FOXO3A, FOXO1 and RECK by miR-96), presumably leading to the aggressive phenotypic features found in AA prostate cancer. Finally, our findings suggest that these deregulated miRNA–mRNA pairs, uniquely found in AA prostate cancer, appear to target the EGFR–PI3K–AKT axis, thus driving prostate cancer aggressiveness in the AA population.

Understanding the origins and etiology of cancer disparities is a complex endeavor and it is imperative that such disparities be addressed at all levels of intervention, both social and biologic. Evidence exists, indicating that one component of the disparity may be related to biologic differences in the molecular etiology of the disease resulting in tumor aggressiveness. We have used a population-based comparative approach in an attempt to discern potential drivers of prostate cancer aggressiveness and have identified novel pathway alterations in miRNA–mRNA pairs that may contribute to prostate cancer disparities. Given the projected use of miRNA mimics and antagomirs as potential cancer therapeutics (50), our study serves as a first pass catalog of dysregulated miRNA–mRNA pairs residing in key oncogenic signaling pathways in AA prostate cancer.

No potential conflicts of interest were disclosed.

Conception and design: B.-D. Wang, Y. Ji, J. Rhim, S.R. Patierno, N.H. Lee

Development of methodology: B.-D. Wang, V. Patel, A. Popratiloff, P. Latham, N.H. Lee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.-D. Wang, K. Ceniccola, Q. Yang, R. Andrawis, A. Popratiloff, P. Latham, N.H. Lee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.-D. Wang, K. Ceniccola, J. Olender, A. Popratiloff, P. Latham, Y. Lai, N.H. Lee

Writing, review, and/or revision of the manuscript: B.-D. Wang, K. Ceniccola, Y. Ji, J. Rhim, J. Olender, A. Popratiloff, P. Latham, N.H. Lee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.-D. Wang, R. Andrawis, J. Olender, N.H. Lee

Study supervision: S.R. Patierno, N.H. Lee

This work was supported by NCI grant R01 CA120316 (to N.H. Lee), DOD grant PC121975 (to N.H. Lee), Affymetrix Collaborations in Cancer Research Award (to N.H. Lee), NCI supplement grant U01 CA116937 (to S.R. Patierno), and American Cancer Society grant IRG-08-091-01 (to B.-D. Wang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Volinia
S
,
Calin
GA
,
Liu
CG
,
Ambs
S
,
Cimmino
A
,
Petrocca
F
, et al
A microRNA expression signature of human solid tumors defines cancer gene targets
.
Proc Natl Acad Sci U S A
2006
;
103
:
2257
61
.
2.
Esquela-Kerscher
A
,
Slack
FJ
. 
Oncomirs - microRNAs with a role in cancer
.
Nat Rev Cancer
2006
;
6
:
259
69
.
3.
Cummins
JM
,
Velculescu
VE
. 
Implications of micro-RNA profiling for cancer diagnosis
.
Oncogene
2006
;
25
:
6220
7
.
4.
Calin
GA
,
Liu
CG
,
Sevignani
C
,
Ferracin
M
,
Felli
N
,
Dumitru
CD
, et al
MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias
.
Proc Natl Acad Sci U S A
2004
;
101
:
11755
60
.
5.
Lee
YS
,
Dutta
A
. 
MicroRNAs in cancer
.
Annu Rev Pathol
2009
;
4
:
199
227
.
6.
Shi
XB
,
Tepper
CG
,
White
RW
. 
MicroRNAs and prostate cancer
.
J Cell Mol Med
2008
;
12
:
1456
65
.
7.
Jemal
A
,
Siegel
R
,
Ward
E
,
Murray
T
,
Xu
J
,
Thun
MJ
. 
Cancer statistics, 2007
.
CA Cancer J Clin
2007
;
57
:
43
66
.
8.
Powell
IJ
. 
Epidemiology and pathophysiology of prostate cancer in African-American men
.
J Urol
2007
;
177
:
444
9
.
9.
Evans
S
,
Metcalfe
C
,
Ibrahim
F
,
Persad
R
,
Ben-Shlomo
Y
. 
Investigating Black-White differences in prostate cancer prognosis: a systematic review and meta-analysis
.
Int J Cancer
2008
;
123
:
430
5
.
10.
Devgan
SA
,
Henderson
BE
,
Yu
MC
,
Shi
CY
,
Pike
MC
,
Ross
RK
, et al
Genetic variation of 3 beta-hydroxysteroid dehydrogenase type II in three racial/ethnic groups: implications for prostate cancer risk
.
Prostate
1997
;
33
:
9
12
.
11.
Shuch
B
,
Mikhail
M
,
Satagopan
J
,
Lee
P
,
Yee
H
,
Chang
C
, et al
Racial disparity of epidermal growth factor receptor expression in prostate cancer
.
J Clin Oncol
2004
;
22
:
4725
9
.
12.
Wallace
TA
,
Prueitt
RL
,
Yi
M
,
Howe
TM
,
Gillespie
JW
,
Yfantis
HG
, et al
Tumor immunobiological differences in prostate cancer between African-American and European-American men
.
Cancer Res
2008
;
68
:
927
36
.
13.
Wang
BD
,
Yang
Q
,
Ceniccola
K
,
Bianco
F
,
Andrawis
R
,
Jarrett
T
, et al
Androgen receptor-target genes in African American prostate cancer disparities
.
Prostate Cancer
2013
;
2013
:
763569
.
14.
Ozen
M
,
Creighton
CJ
,
Ozdemir
M
,
Ittmann
M
. 
Widespread deregulation of microRNA expression in human prostate cancer
.
Oncogene
2008
;
27
:
1788
93
.
15.
Wang
L
,
Tang
H
,
Thayanithy
V
,
Subramanian
S
,
Oberg
AL
,
Cunningham
JM
, et al
Gene networks and microRNAs implicated in aggressive prostate cancer
.
Cancer Res
2009
;
69
:
9490
7
.
16.
Pomerantz
MM
,
Beckwith
CA
,
Regan
MM
,
Wyman
SK
,
Petrovics
G
,
Chen
Y
, et al
Evaluation of the 8q24 prostate cancer risk locus and MYC expression
.
Cancer Res
2009
;
69
:
5568
74
.
17.
Goeman
JJ
,
van de Geer
SA
,
de Kort
F
,
van Houwelingen
HC
. 
A global test for groups of genes: testing association with a clinical outcome
.
Bioinformatics
2004
;
20
:
93
9
.
18.
Tian
L
,
Greenberg
SA
,
Kong
SW
,
Altschuler
J
,
Kohane
IS
,
Park
PJ
. 
Discovering statistically significant pathways in expression profiling studies
.
Proc Natl Acad Sci U S A
2005
;
102
:
13544
9
.
19.
House
CD
,
Vaske
CJ
,
Schwartz
AM
,
Obias
V
,
Frank
B
,
Luu
T
, et al
Voltage-gated Na+ channel SCN5A is a key regulator of a gene transcriptional network that controls colon cancer invasion
.
Cancer Res
2010
;
70
:
6957
67
.
20.
Wang
BD
,
Kline
CL
,
Pastor
DM
,
Olson
TL
,
Frank
B
,
Luu
T
, et al
Prostate apoptosis response protein 4 sensitizes human colon cancer cells to chemotherapeutic 5-FU through mediation of an NF kappaB and microRNA network
.
Mol Cancer
2010
;
9
:
98
.
21.
Griffiths-Jones
S
,
Saini
HK
,
van Dongen
S
,
Enright
AJ
. 
miRBase: tools for microRNA genomics
.
Nucleic Acids Res
2008
;
36
:
D154
8
.
22.
Wong
KK
,
Engelman
JA
,
Cantley
LC
. 
Targeting the PI3K signaling pathway in cancer
.
Curr Opin Genet Dev
2010
;
20
:
87
90
.
23.
Seshacharyulu
P
,
Ponnusamy
MP
,
Haridas
D
,
Jain
M
,
Ganti
AK
,
Batra
SK
. 
Targeting the EGFR signaling pathway in cancer therapy
.
Expert Opin Ther Targets
2012
;
16
:
15
31
.
24.
Zambon
AC
,
Gaj
S
,
Ho
I
,
Hanspers
K
,
Vranizan
K
,
Evelo
CT
, et al
GO-Elite: a flexible solution for pathway and ontology over-representation
.
Bioinformatics
2012
;
28
:
2209
10
.
25.
Zhu
G
,
Fan
Z
,
Ding
M
,
Zhang
H
,
Mu
L
,
Ding
Y
, et al
An EGFR/PI3K/AKT axis promotes accumulation of the Rac1-GEF Tiam1 that is critical in EGFR-driven tumorigenesis
.
Oncogene
. 
2015 Mar 9
. [
Epub ahead of print
].
26.
Han
W
,
Carpenter
RL
,
Cao
X
,
Lo
HW
. 
STAT1 gene expression is enhanced by nuclear EGFR and HER2 via cooperation with STAT3
.
Mol Carcinog
2013
;
52
:
959
69
.
27.
Cronshaw
DG
,
Kouroumalis
A
,
Parry
R
,
Webb
A
,
Brown
Z
,
Ward
SG
. 
Evidence that phospholipase-C-dependent, calcium-independent mechanisms are required for directional migration of T-lymphocytes in response to the CCR4 ligands CCL17 and CCL22
.
J Leukoc Biol
2006
;
79
:
1369
80
.
28.
Krol
J
,
Francis
RE
,
Albergaria
A
,
Sunters
A
,
Polychronis
A
,
Coombes
RC
, et al
The transcription factor FOXO3a is a crucial cellular target of gefitinib (Iressa) in breast cancer cells
.
Mol Cancer Ther
2007
;
6
:
3169
79
.
29.
Zhang
S
,
Zhau
HE
,
Osunkoya
AO
,
Iqbal
S
,
Yang
X
,
Fan
S
, et al
Vascular endothelial growth factor regulates myeloid cell leukemia-1 expression through neuropilin-1-dependent activation of c-MET signaling in human prostate cancer cells
.
Mol Cancer
2010
;
9
:
9
.
30.
Signoretti
S
,
Waltregny
D
,
Dilks
J
,
Isaac
B
,
Lin
D
,
Garraway
L
, et al
p63 is a prostate basal cell marker and is required for prostate development
.
Am J Pathol
2000
;
157
:
1769
75
.
31.
Tannock
IF
,
de Wit
R
,
Berry
WR
,
Horti
J
,
Pluzanska
A
,
Chi
KN
, et al
Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer
.
N Engl J Med
2004
;
351
:
1502
12
.
32.
Gade
S
,
Porzelius
C
,
Falth
M
,
Brase
JC
,
Wuttig
D
,
Kuner
R
, et al
Graph based fusion of miRNA and mRNA expression data improves clinical outcome prediction in prostate cancer
.
BMC Bioinformatics
2011
;
12
:
488
.
33.
Feng
J
,
Huang
C
,
Diao
X
,
Fan
M
,
Wang
P
,
Xiao
Y
, et al
Screening biomarkers of prostate cancer by integrating microRNA and mRNA microarrays
.
Genet Test Mol Biomarkers
2013
;
17
:
807
13
.
34.
Zheng
H
,
Fu
R
,
Wang
JT
,
Liu
Q
,
Chen
H
,
Jiang
SW
. 
Advances in the techniques for the prediction of microRNA targets
.
Int J Mol Sci
2013
;
14
:
8179
87
.
35.
Huang
JC
,
Babak
T
,
Corson
TW
,
Chua
G
,
Khan
S
,
Gallie
BL
, et al
Using expression profiling data to identify human microRNA targets
.
Nat Methods
2007
;
4
:
1045
9
.
36.
Mairinger
FD
,
Ting
S
,
Werner
R
,
Walter
RF
,
Hager
T
,
Vollbrecht
C
, et al
Different micro-RNA expression profiles distinguish subtypes of neuroendocrine tumors of the lung: results of a profiling study
.
Mod Pathol
2014
;
27
:
1632
40
.
37.
Patterson
SG
,
Wei
S
,
Chen
X
,
Sallman
DA
,
Gilvary
DL
,
Zhong
B
, et al
Novel role of Stat1 in the development of docetaxel resistance in prostate tumor cells
.
Oncogene
2006
;
25
:
6113
22
.
38.
Nohata
N
,
Hanazawa
T
,
Enokida
H
,
Seki
N
. 
microRNA-1/133a and microRNA-206/133b clusters: dysregulation and functional roles in human cancers
.
Oncotarget
2012
;
3
:
9
21
.
39.
Uchida
Y
,
Chiyomaru
T
,
Enokida
H
,
Kawakami
K
,
Tatarano
S
,
Kawahara
K
, et al
MiR-133a induces apoptosis through direct regulation of GSTP1 in bladder cancer cell lines
.
Urol Oncol
2013
;
31
:
115
23
.
40.
Xu
XM
,
Qian
JC
,
Deng
ZL
,
Cai
Z
,
Tang
T
,
Wang
P
, et al
Expression of miR-21, miR-31, miR-96 and miR-135b is correlated with the clinical parameters of colorectal cancer
.
Oncol Lett
2012
;
4
:
339
45
.
41.
Guttilla
IK
,
White
BA
. 
Coordinate regulation of FOXO1 by miR-27a, miR-96, and miR-182 in breast cancer cells
.
J Biol Chem
2009
;
284
:
23204
16
.
42.
Haflidadottir
BS
,
Larne
O
,
Martin
M
,
Persson
M
,
Edsjo
A
,
Bjartell
A
, et al
Upregulation of miR-96 enhances cellular proliferation of prostate cancer cells through FOXO1
.
PLoS ONE
2013
;
8
:
e72400
.
43.
Guo
H
,
Li
Q
,
Li
W
,
Zheng
T
,
Zhao
S
,
Liu
Z
. 
miR-96 downregulates RECK to promote growth and motility of non–small cell lung cancer cells
.
Mol Cell Biochem
2014
;
390
:
155
60
.
44.
Zhang
J
,
Kong
X
,
Li
J
,
Luo
Q
,
Li
X
,
Shen
L
, et al
miR-96 promotes tumor proliferation and invasion by targeting RECK in breast cancer
.
Oncol Rep
2014
;
31
:
1357
63
.
45.
Lin
H
,
Dai
T
,
Xiong
H
,
Zhao
X
,
Chen
X
,
Yu
C
, et al
Unregulated miR-96 induces cell proliferation in human breast cancer by downregulating transcriptional factor FOXO3a
.
PLoS ONE
2010
;
5
:
e15797
.
46.
Shukla
S
,
Bhaskaran
N
,
Maclennan
GT
,
Gupta
S
. 
Deregulation of FoxO3a accelerates prostate cancer progression in TRAMP mice
.
Prostate
2013
;
73
:
1507
17
.
47.
Audenet
F
,
Cancel-Tassin
G
,
Bigot
P
,
Audouin
M
,
Gaffory
C
,
Ondet
V
, et al
Germline genetic variations at 11q13 and 12p11 locus modulate age at onset for renal cell carcinoma
.
J Urol
2014
;
191
:
487
92
.
48.
Wu
X
,
Scelo
G
,
Purdue
MP
,
Rothman
N
,
Johansson
M
,
Ye
Y
, et al
A genome-wide association study identifies a novel susceptibility locus for renal cell carcinoma on 12p11.23
.
Hum Mol Genet
2012
;
21
:
456
62
.
49.
Kojima
S
,
Enokida
H
,
Yoshino
H
,
Itesako
T
,
Chiyomaru
T
,
Kinoshita
T
, et al
The tumor-suppressive microRNA-143/145 cluster inhibits cell migration and invasion by targeting GOLM1 in prostate cancer
.
J Hum Genet
2014
;
59
:
78
87
.
50.
Melo
SA
,
Kalluri
R
. 
Molecular pathways: microRNAs as cancer therapeutics
.
Clin Cancer Res
2012
;
18
:
4234
9
.

Supplementary data