MiR-644a Disrupts Oncogenic Transformation and Warburg Effect by Direct Modulation of Multiple Genes of Tumor-Promoting Pathways

Prostate cancer is the most common noncutaneous cancer affecting men in the United States and ranks second in cancer-related deaths (1). Prostate cancer carcinogenesis and metastasis encompass significant genetic heterogeneity (2, 3). Numerous alternations in prostate cells lead to prostate cancer progression from a castration-sensitive to a castration-resistant prostate cancer (CRPC), which relates to poor prognosis for patients (4, 5). However, the molecular mechanisms contributing to the development of CRPC are only beginning to emerge. Clinical and scientific evidence indicates that CRPC contains clones of both the androgen receptor (AR)–negative and AR-positive tumor cells (6, 7). AR is a ligand-dependent transcription factor, and androgen signaling plays a crucial homeostatic role in balancing the normal physiology and biology of the prostate. Paradoxically, AR and the androgen signaling axis is also a predominant causative factor in the development of prostate cancer and its transition to castration-resistant cancer (8).

The current therapeutic options for prostate cancer include androgen deprivation therapy (ADT) and AR inhibitors. However, the development of resistance to these drugs leads to a poor prognosis for CRPC patients (9). Interestingly, large-scale integrative genomic and proteomic analyses of CRPC tumors reveal that multiple intrinsic signaling pathways also play a causative role in the development of CRPC (6, 10). Hence, an in-depth understanding of the molecular and genetic changes that occur during the development of CRPC is necessary to identify novel therapeutic targets. Considering the molecular complexity of the AR signaling cascade and prostate cancer heterogeneity, it is logical to the cotarget expression of genes involved in multiple carcinogenic pathways to achieve a sustained and clinically significant response.

Regulatory noncoding miRNAs fine-tune posttranscriptional control of gene expression and thus regulatory networks. miRNAs regulate gene expression by guiding the association between the RNA-induced silencing complex and target mRNAs (11). Numerous miRNAs regulate expression of genes involved in cell cycle, organogenesis, energy balance, and development, thus affecting cell proliferation, differentiation, stem cell maintenance, and cell death, as well as many other cellular functions (12). Depending upon the differential expression of miRNAs in a given cell and tissue, miRNAs either promote oncogenesis (oncomiRs) or act as tumor suppressors (13). Multiple intrinsic factors, including androgen signaling, AR spliced variants, steroid metabolism, myelocytomatosis viral oncogene homolog (c-Myc), and miRNA dysregulation, play a significant role in the development of CRPC (14–19). The inherent property of miRNAs to fine-tune gene expression by suppressing the translation of multiple genes of signaling networks involved in disease-promoting activities makes them attractive natural molecules in miRNA replacement therapy (20).

In this study, we demonstrate the potential therapeutic efficacy of miR-644a by inhibiting the expression of genes implicated in crucial signaling and transcriptional regulatory pathways involved in prostate cancer development and its progression to CRPC. We show that miR-644a is a broad-range cellular pathways modulator and functions as a potent novel tumor suppressor.

Human prostate cancer cell lines (LNCaP, DU-145, 22RV1, and PC-3 with catalog numbers CRL-1740, HTB-81, CRL-2505, and CRL-1435, respectively) were purchased from the ATCC, and C4.2B was purchased from Viromed Laboratories. The cells were cultured in complete RPMI 1640 medium supplemented with 10% FBS, 2 mmol/L l-glutamine, and penicillin and streptomycin antibiotics. LNCaP cell line was used to represent a castration-sensitive prostate cancer model. DU-145 and PC-3 cells were utilized to represent castration-resistant disease model. Further, we utilized 22RV1 cells that express an alternatively spliced AR isoform AR-V7 to test the therapeutic potential of miR-644a during therapeutic resistance. CHO-K1 cells were cultured in DMEM supplemented with 5% FBS, 2 mmol/L l-glutamine, 1 mmol/L l-proline, 10 mmol/L 4-(2-Hydroxyethyl) piperazine-1-ethanesulfonic acid (HEPES), and antibiotics. All the cell lines were maintained in a humidified 5% CO₂ atmosphere at 37°C. All the experiments were repeated at least 3 times with two stocks of cells.

The target validation of miR-644a target sites was performed as described previously (21). In brief, wild-type (WT) untranslated region (UTR) reporter plasmids were constructed by cloning fragments of 3′UTR spanning the predicted target site for miR-644 downstream of the firefly luciferase coding region in pMIR-REPORT vector (Ambion). The primers used to amplify the 3′UTR fragments and the fragment sizes are provided in the Supplementary Material and Methods. Site-directed mutagenesis of the putative target site for miR-644 in WT-UTR construct was carried out to generate the MUT-UTR constructs. Nucleotide sequences of the constructs were confirmed by DNA sequencing. For luciferase assays, CHO-K1 cells (30,000 cells/well) were plated in 24-well plates one day before transfection. For all miRNA target validation experiments, cells were transfected using Lipofectamine 2000 (Invitrogen), with 100 ng of WT-UTR or MUT-UTR firefly luciferase reporter construct, 0.5 ng of Renilla luciferase reporter plasmid (Promega), and either miR-644 mimic (10 nmol/L) or NC mimic (10 nmol/L). Cell lysates were assayed for firefly and Renilla luciferase activities 48 hours after transfection using the Dual-Luciferase Reporter Assay System (Promega) in Victor 3 Multilabel Counter 1420 (PerkinElmer). Renilla luciferase activity served as a control for transfection efficiency. Data are represented as a ratio of firefly luciferase activity to Renilla luciferase activity.

Total RNA from normal and cancerous tissues was obtained from Prostate Cancer Biorepository Network site hosted at the University of Washington, Seattle, after Institutional Review Board and Material Transfer Agreement approvals. First-strand cDNA was synthesized from 100 ng of total RNA from prostate cancer cells or patient samples using stem-loop primers specific for human mature miR-644 and snoRNA (RNU66). Reverse transcription and RT-qPCR was carried out using the TaqMan MicroRNA Reverse Transcription Kit and TaqMan MicroRNA Assays (Applied Biosystems) as described previously. The relative expression of miR-644a was calculated as 2^−ΔC_t, where ΔC_t = C_t value of miR-644a in a sample – C_t value of RNU66 in that sample. Mean miR expression ± SE was calculated from three independent experiments.

In this study, to detect and quantify the expression of proteins in all prostate cancer cells we transiently transfected the cells with miR-644a mimic or NC mimic in 50 nmol/L concentration or as indicated for a particular experiment. For apoptosis detection, in addition to miR-644a treatments, the LNCaP cells were treated with a known AR antagonist, bicalutamide (100 μmol/L), and apoptosis was assayed 4 days after transfection. Apoptotis was determined by using the Cell Death Detection ELISA^PLUS Kit (Roche Applied Science) according to the manufacturer's protocols.

Prostate cancer cells were seeded in 6-well plates at 200,000 cells per well 48 hours before transfection. The cells were maintained in charcoal/dextran-stripped FBS and transfected with the miR-644 mimics using Lipofectamine 3000 reagent. The cells were treated with enzalutamide at the concentration of 1.0 μmol/L. After 24 hours of treatment with miR-mimics and enzalutamide, the cells were induced for AR activation using DHT (10 nmol/L) or DMSO as a control. Twenty-four hours after DHT induction, the cells were processed for further analysis.

22RV1 human prostate cell xenografts were established in male athymic nude male mice 4 weeks of age by s.c. injecting 2.0 × 10⁶ cells suspended in 100 μL Matrigel (BD Biosciences) into both flanks. When palpable tumors established at approximately 5 mm diameter (66 mm³), intratumoral treatment with siPORT amine transfection reagent (Ambion) complexed miR-644a (3.15 μg of synthetic mimics every 3 days injection) was done in treatment groups (n = 6 animals). Mice in enzalutamide treatment group was fed with the drug through oral gavage every 3 days (10 mg/kg/day). Antitumor effects of miR-644a with or without enzalutamide treatment were assessed thrice a week by measuring tumor volume. The tumor growth was monitored for 21 days from the start of treatment. Tumors and organs were harvested for further analyses. The animal xenograft study was approved by the Institutional Animal Care and Use Committee, and experiments were assisted by a board-certified veterinarian according to institutional guidelines.

The prostate cancer patient gene expression data was obtained from Catalogue of Somatic Mutations in Cancer (COSMIC) Browser database from adenocarcinoma samples with high-level amplifications (n = 498). The samples were collated, and the genes displaying overexpression pattern were selected for further analysis. The miR-644a target pool was selected from prediction algorithms including Targetscan 6.2, miRanda, miRtarget2, and RNAhybrid algorithms. Next, the target pool was enriched using g:Profiler software (http://biit.cs.ut.ee/gprofiler/index.cgi) using the options “significant only” and “ordered query” with a P value <0.05 and Benjamini–Hochberg FDR significance threshold. The results were visualized using GeneMania plugin for Cytoscape 3.1 (http://www.genemania.org/). Further analysis was performed using Reactome (http://www.reactome.org/) and EnrichR (http://amp.pharm.mssm.edu/Enrichr/) pathway databases using P value < 0.05.

All analyses, unless otherwise indicated, were undertaken using the R platform for statistical computing [R version 3.3.1 (2016-06-21), Platform: x86_64-apple-darwin13.4.0 (64-bit), Running under: OS X 10.11.6 (El Capitan) (R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/], and a range of library packages were implemented in Bioconductor as indicated.

Expression levels of the miR-644a genes were measured in the PRAD cohort. For the neuroendocrine prostate cancer (NEPC) cohort, the cgdsr tool was used to gain ITCH CNA and miR-644a target gene expression. In the first instance, the cross-correlation of ITCH CNA and miR-644a target expression was undertaken in R corrplot. In the PRAD cohort, the TCGA_PRAD_HiSeqV2 RNA-Seq data and associated clinical data were downloaded from University of California Santa Cruz database and tumor:normal Z scores calculated to yield the tumor expression of all detectable genes, in at least 80%, relative to normal tissue. The tumor-associated expression alterations of all detectable miR-644a target genes were measured (Z-scores), and genefilter was used to select for genes that were commonly and significantly altered. Specifically, a threshold for filtering was taken for selected genes that were altered by more than 2 Z scores in 30% of tumors. The expression of genes was used to cluster tumors and visualized with a heat map. The association of patient cluster membership and clinical outcome (either categorical data or continuous data that were categorized) was then tested using a χ² test and regression models.

Statistical analyses were performed using SPSS software or GraphPad Prism. Data were presented as mean ± SE from at least three independent experiments and two independent stocks of cell lines purchased from the ATCC. Independent samples t test was used to assess statistically significant differences. Statistical significance was accepted for P < 0.05.

Because miRNAs can fine-tune the expression of coordinated cellular pathways, we surmised to identify an miRNA that can potentially target the expression of AR as well as critical genes implicated in CRPC. By computational prediction, we identified a handful of miRNAs that have the potential to target the AR 3′UTR (Supplementary Fig. S1A). Next, we tested if these predicted miRNAs can negatively modulate the posttranscriptional AR expression in prostate cancer cells. We transfected a panel of predicted AR-targeting miRNA mimics in LNCaP and C4-2B cells and found that miR-644a maximally reduced AR expression as compared with other predicted AR-targeting miRNA mimics (Lane 8, Supplementary Fig. S1B and S1C). Earlier, we have shown that miR-644a targets β-actin and GAPDH gene expression, two proteins critical to cytoskeletal dynamics and glycolysis, respectively (21). To test the prospect of miR-644a to regulate posttranscriptional expression of multiple other genes involved in disease-promoting activities, we in silico determined the putative targets of miR-644a potentially associated with prostate carcinogenesis–promoting pathways. We used miRecords, which predicts miRNA targets by considering at least four prominent algorithms including Targetscan 6.2, miRanda, miRtarget2, and RNAhybrid (22). MiR-644a–predicted target genes were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology pathways analysis by comparing with gene set comprehensive libraries from multiple sources including KEGG, Wikipathways, BioCarta, and Reactome (23).

In addition, expression data analysis from 498 adenocarcinoma patients from Cosmic V78 revealed that miR-644a is predicted to target genes involved in pathways regulating apoptosis and hypoxia as well as NOTCH, c-Myc, and androgen signaling pathways (Supplementary Table S1; ref. 24).

Furthermore, our bioinformatics analysis also predicted a potential role of miR-644a in other cancers including small cell lung cancer, pancreatic cancer, chronic myeloid leukemia, and colorectal cancers (Supplementary Table S1). The bioinformatics analysis of experimentally validated gene expression datasets including EnrichR, Reactome, and Cytoscape predicted many genes including AR, Bcl2, Bcl-xl, c-Myc, SRC, Snail family zinc finger 1 (Snail), EZH2, E2F1, and Zinc Finger E-Box Binding Homeobox 1 (ZEB1) as miR-644a targets (Supplementary Table S2). These genes have also been associated with prostate cancer and development of resistance to androgen signaling inhibitors including enzalutamide and abiraterone acetate (25).

To characterize miR-644a–mediated regulation of AR expression and transactivation function, we ectopically overexpressed miR-644a mimics in multiple AR-positive prostate cancer cell lines, including LNCaP, LAPC4, and 22RV1 cell lines, which express both AR mRNA and AR protein and represent castration-sensitive and -resistant cell models. Ectopic expression of miR-644a mimics in LNCaP and LAPC4 cells significantly reduced AR protein levels as compared with mock and negative control (NC) mimic–transfected cells. Similarly, reduction in both full-length and AR-V7 protein levels by miR-644a was also detected in 22RV1 cells (Fig. 1A), supporting the potential of miR-644a in targeting both isoforms of AR implicated in the development of drug resistance and CRPC (18, 26). Hsp70 expression was used as a loading control because it is not a predicted target of miR-644a. Furthermore, RT-qPCR experiments confirmed significant downregulation of AR mRNA in miR-644a–transfected LNCaP, LAPC4, and 22RV1 cells (Fig. 1B). These results demonstrate that miR-644a effectively suppresses AR expression in both castration-sensitive and -resistant cellular models of prostate cancer. Furthermore, miRNA:mRNA firefly luciferase target validation experiments confirm a direct miR-644a interaction with the 3′UTR of AR from nucleotides 340 to 358 (Fig. 1C and D). The above data indicate that miR-644a targets AR gene expression and reduces both mRNA and protein levels in prostate cancer cells by directly interacting with the 3′UTR of AR mRNA.

Relapsed AR expression and its transactivation function is a key event in prostate carcinogenesis and a major cause of castration-resistant disease. AR coregulator proteins play an important regulatory role in AR transcriptional activities, and many of these coregulators are highly upregulated in CRPC (27). Several AR coregulators were predicted targets of miR-644a, suggesting a fine-tuning paradigm (Supplementary Table S2). Among AR coregulators, steroid receptor coactivator-1 (SRC1), steroid receptor coactivator-2 (SRC2), CREB-binding protein (CBP), steroid receptor coactivator-3 (SRC3), cyclin D1 (CCND1), and AR associated protein-24 (ARA24) were predicted targets of miR-644a (Supplementary Table S3). Next, we validated the expression of predicted AR coregulators by RT-qPCR and immunoblotting in miR-644a–overexpressing prostate cancer cells. miR-644a overexpression resulted in significant downregulation of the expression of AR coregulators including SRC-1, SRC-2, SRC-3, CCND1, CBP, and ARA24 at mRNA level (Fig. 2A) as well as the protein level in LNCaP and PC-3 cells (Fig. 2B). The AR-null PC-3 cell line is an established model to study androgen signaling–independent miR-644a biological activities. In addition, we confirmed the inhibitory function of miR-644a on AR transactivation and signaling by analyzing the PSA protein and mRNA levels in AR-positive LNCaP and 22RV1 cells (Fig. 2C). Above data demonstrate that by the direct interaction with AR and its coactivators, miR-644a contributes to the negative regulation of AR transactivation activities.

To evaluate plausible tumor suppressor potential of miR-644a, we subcutaneously injected 22RV1 cells in NCR nu/nu mice on both flanks (n = 5) and treated mice with intratumoral injections of miR-644a mimic. We compared treatment groups that received enzalutamide through oral gavage and the group that received a combination of miR-644a with enzalutamide. Tumor volumes of miR-644a–injected mice were significantly lower than those receiving enzalutamide alone demonstrating potentiation of enzalutamide therapy by miR-644a (Fig. 3A–C). Intratumoral delivery of miR-644a reduced tumor growth by >50% compared with that of NC mimics. We confirmed the expression of miR-644a by RT-qPCR in harvested tumors to determine the delivery and expression of mimics (Supplementary Fig. S2A). Immunoblotting of tumor lysates confirmed that miR-644a treatment correlates with downregulation of AR WT (110 kDa) as well as AR-V7 (80 kDa) isoforms (Supplementary Fig. S2B). We also validated the AR transactivation function by PSA expression analysis of serum collected from the animals (Supplementary Fig. S2C). Furthermore, mRNASeq data obtained from the tumors show several genes were downregulated in miR-644a–treated animals (Fig. 3D). Functional analysis of the downregulated genes using KEGG indicates that miR-644a is predicted to target genes involved in proliferation, apoptosis, epithelial–mesenchymal transition (EMT), and aerobic glycolysis. Many of these genes are also associated with multiple cancer pathways supporting a tumor-suppressor role of miR-644a (Supplementary Table S4). AR, SRC-1, SRC-1, GAPDH, and ACTB among many others were downregulated in tumors instilling the confidence in the mRNASeq dataset for further exploration of other downregulated genes as potential targets of miR-644a. In summary, the xenograft experiment determined a potent tumor-suppressor function of miR-644a, presumably by fine-tuning the expression of androgen signaling–dependent as well as –independent cellular pathways.

Next, we determined the functional significance of miR-644a expression in the growth and viability of prostate cancer cells. The clonogenic soft-agar assay in 22RV1 cells demonstrated that the overexpression of miR-644a affected growth and survival of prostate cancer cells as apparent by a decrease in the number of colonies as well as the decrease in the size of the colonies compared with the NC mimic (Fig. 4A). Also, cell viability assays demonstrate a significant growth inhibition by miR-644a in both LNCaP and PC-3 cells (Fig. 4B). We reasoned that the growth-inhibitory effect of miR-644a on cancer cells could be attributed to its proapoptotic activities. Hence, we determined PARP cleavage and downregulation of antiapoptotic biomarkers in miR-644a–overexpressing LNCaP and PC-3 cells. Indeed, in miR-644a–transfected cells, PARP proteolytic cleavage was prominent when compared with NC-transfected cells as determined by immunoblotting (Fig. 4C). We further verified the proapoptosis activity of miR-644a by using a quantitative sandwich immunoassay, which demonstrated that miR-644a–expressing cells have higher levels of apoptosis as compared with NC mimic–transfected cells, as well as AR antagonist enzalutamide-treated cells (Fig. 4D). Moreover, the level of apoptosis in the miR-644a–transfected cells approaches the level of apoptosis induced by treatment with enzalutamide.

Further, characterization of miR-644a–treated cells for apoptotic biomarkers using a protein microarray revealed miR-644a–induced changes consistent with apoptosis in LNCaP and PC3 cells (Supplementary Fig. S3A and S3B). Multiple proapoptotic posttranslational modifications of proteins including cleaved caspase-3, cleaved caspase-7, Bad Ser136, SAPK/JNK Thr183/Tyr185, cleaved PARP, and p53 Ser15 were upregulated. In addition, antiapoptotic markers AKT Ser473 phosphorylation and Survivin were downregulated (Supplementary Fig. S3B and Supplementary Table S5). We further confirmed that the proapoptotic function of miR-644a is mediated by the direct targeting of two antiapoptotic genes, Bcl-xl and Bcl2 (Fig. 4E and F) in prostate cancer cells.

The effect of miR-644a transfection on the proliferation of prostate cancer cells was also analyzed by BrdU incorporation in miR-644a–treated LNCaP and PC-3 cells. MiR-644a substantially decreased proliferation in both cell lines (Supplementary Fig. S4A and S4B), indicating a significant role of miR-644a in regulating growth, survival, and proliferation of prostate cancer cells. Collectively, these results demonstrate that miR-644a can suppress the growth and proliferation of prostate cancer cells.

To further delineate the underlying mechanism for the antitumor activity of miR-644a, we sought to determine the expression of known key genes implicated in prostate cancer tumor metabolism. The cellular pathway analysis predicted the proto-oncogene c-Myc as a potential target of miR-644a, among many others; hence, we performed c-Myc gene expression analysis. Immunoblots and RT-qPCR determined the downregulation of c-Myc both at protein and mRNA levels, respectively, in miR-644a–overexpressing cells (Fig. 5A and B).

Further, we confirmed that c-Myc is a direct target of miR-644a using a firefly luciferase reporter. Overexpression of miR-644a mimics repressed firefly luciferase expression containing c-Myc 3′UTR, whereas the seed region mutations (Supplementary Fig. S5A, lower sequence plot) in the miR-644a target sequence rescued the repression of luciferase (Supplementary Fig. S5B). Above experiments determined that miR-644a–mediated downregulation of c-Myc expression may be of therapeutic significance in castration-resistant disease (28).

In addition, AKT and IGF1R contribute to aerobic glycolysis and tumor growth in prostate cancer (29). Essentially, insulin-like growth factor (IGF1) signaling is mediated by the activation of insulin-like growth factor receptor (IGF1R), which promotes AR signaling in CRPC via activation of AR-V7 splice variant (30). Increased IGF1R levels in prostate cancer cells are associated with the development and progression of the disease and a valuable target in CRPC treated with IGF1R antagonists (31). Because IGF1R/AKT signaling contributes to tumor energy metabolism and castration-resistant AR signaling, we validated the expression of these factors in miR-644a–overexpressing prostate cancer cells. Immunoblotting of both LNCaP and PC3 cell lysates revealed downregulation of IGF1R and AKT expression in both castration-sensitive and -resistant cell lines (Fig. 5A).

Tumor energy metabolism is mainly dependent on the glycolytic pathway rather than oxidative phosphorylation, and GAPDH is a central player in glycolysis-dependent energy supply forming the core of cancer cell survival. Upregulated expression of GAPDH in high-grade tumors underlines the importance of metabolism in prostate cancer (32). Because GAPDH expression is downregulated by miR-644a by direct interaction, we examined its enzymatic activity in prostate cancer cells as well as in the xenografts. The downregulated GAPDH activity in both 22RV1 xenograft tumors and PC-3 cells (Supplementary Fig. S5C and S5D) indicates miR-644a antagonist role in tumor energy metabolism. Further glycolytic extracellular flux analysis revealed that miR-644a significantly decreased glycolysis and glycolytic capacity in 22RV1 and PC-3 cells (Fig. 5C and D). Collective repression of c-Myc, Akt, IGF1R, and GAPDH expression by miR-644a that suppresses Warburg effect and proliferation underlines that miR-644a is a potent prostate tumor suppressor.

Our in silico analysis also predicted EMT genes as potential targets of miR-644a, including ZEB1, cyclin-dependent kinase 6 (cdk6), and snail. Therefore, we analyzed if miR-644a exerts any influence on EMT in prostate cancer cells. Our results indicated that miR-644a suppresses invasion of 22RV1 (Fig. 6A and B) as well as PC-3 cells. Such repression of EMT phenotype is likely to be mediated via the targeting of ZEB1, cdk6, and snail in prostate cancer. We confirmed the downregulation of ZEB1, cdk6, and snail, both at protein and mRNA levels in miR-644a–overexpressing cells (Fig. 6C and D). Furthermore, Firefly luciferase reporter target assays determined that ZEB1, cdk6, and snail genes are direct targets of miR-644a (Supplementary Fig. S6A–S6C). Also, the reversal of EMT was confirmed by immunoblotting for EMT markers vimentin, E-cadherin, and β-catenin in PC-3 cells (Fig. 6C). Wound-healing assay in LNCaP and PC-3 cells overexpressing miR-644a confirmed the inhibition of migration of prostate cancer cells (Fig. 6E). Together, the above data highlight the significance of fine-tuning of the expression of multiple regulators of EMT by miR-644a for improved prostate cancer therapeutics.

Using the cBioPortal (PMID 23550210, PMID: 22588877) to access The Cancer Genome Atlas prostate cohorts, we examined the expression and copy-number variation (CNV) of the miR-644a host gene, ITCH. Strikingly, ITCH was amplified in the neuroendocrine cohort of tumors (PMID: 26855148; Supplementary Fig. S7A). We, therefore, examined the correlation of expression of the miR-644a target genes with the CNV of ITCH in the NEPC cohort. The cross-correlation plots (Supplementary Fig. S7B) reveal that in the NEPC cohort there was a significant negative correlation between ITCH amplification and the expression of 11 miR-644a target genes including NCOR2.

We also reasoned that in tumor cohorts even when clear amplification of ITCH was not apparent, miR-644a targets may be associated with the disease that is more aggressive. For example, we examined the expression of the most altered miR-644a targets in the PRAD cohort (PMID: 26544944). By filtering these genes to those that were most altered in expression, tumors were clustered into two groups (Supplementary Fig. S7C). A χ² test established that this grouping of tumor significantly stratified patients by whether they experienced tumor recurrence or not.

To determine the differential expression of mature miR-644a, we analyzed its expression in prostate cancer tissue derived from patients with or without ADT. MiR-644a expression is downregulated in patients after ADT treatment and is inversely correlated with AR expression in the same tissue (Fig. 7A). This inverse correlation implicates the regulation of AR gene expression by miR-644a. Further, comparisons of patient samples from different stages of prostate cancer and benign tissues revealed that miR-644a expression was highly downregulated in metastatic prostate cancer samples (Fig. 7B), supporting a tumor-suppressor role for miR-644a in prostate cancer. This observation was also validated in prostate cancer cell lines demonstrating loss of miR-644a expression in multiple prostate cancer cell lines compared with benign epithelial cells (Supplementary Fig. S8).

In summary, our results support the role of miR-644a in tumor-suppressing biological activities, potentially by fine-tuning the intrinsic expression of critical pathways involved in the development of CRPC.

Clinical genomics research has developed a better understanding of the genetic alterations in CRPC in recent years. However, translational application of these findings to develop more durable and effective treatments of CRPC is significantly hampered by our inability to identify functionally and clinically relevant drivers of CRPC and to develop an effective common control of their fine-tuned expression for homeostatic cellular controls. Besides, development of drug resistance in CRPC driven by enhanced invasion as well as metastasis remains a significant challenge in the clinical management and treatment of the disease. The critical deterrent in prostate cancer diagnosis and prognosis is the molecular and clonal heterogeneity of the disease, which leads to major concerns of overtreatment of indolent tumors and undertreatment of high-risk disease, leading to the development of metastatic disease (33). Due to the inherent heterogeneous nature of prostate cancer involving pathways including androgen signaling, PI3K/AKT, Wnt/β-catenin, EMT, AR coactivators, c-Myc, and intratumoral steroidogenesis, development of a common therapeutic to target or at least fine-tune the expression of multiple cellular pathways has not been attempted in the clinical setting due to apparent reasons. Indeed, the development of next-generation androgen signaling inhibitors including enzalutamide and Abiraterone acetate has been the most noticeable advancement in prostate cancer therapeutics. In advanced prostate tumors, however, this therapy fails in the majority of patients and drug resistance is a daunting challenge. Herein, we have identified an miRNA, miR-644a, that can fine-tune the expression of major causative pathways, including androgen signaling, c-Myc, EMT, and tumor energy metabolism via glycolysis, and potentiate the therapeutic effects of CRPC drug enzalutamide. An earlier study has demonstrated that miR-644a could target the expression of AR in LNCaP cells by measuring the protein levels only (34). A recent report confirmed the role of miR-644a in suppressing drug resistance by inhibition of cell survival and EMT in breast cancer complementing our findings (35). Another study recently demonstrated that miR-644a inhibits gastric cancer cell proliferation and invasion as well as promotes apoptosis in liver cancer cells (36, 37).

Our in silico analysis revealed the implication of miR-644a target pool in multiple cancer-related signaling pathways including metabolism, cell cycle, apoptosis, and signal transduction. In particular, clinically relevant oncogenes include c-Myc, SRC, and metabolic oncogenes FASN and GLS. Other notable targets of miR-644a include AKT, β-catenin, and SRC kinase, which collectively play a significant role in metastatic prostate cancer disease (15, 38).

In addition, many coactivators, which alter the transcriptional activity of AR including cdk6, SRC1, SRC2, SRC3, CBP, CCND1, and ARA24, have been implicated in prostate cancer (39, 40). In particular, targeting the expression of p160 group of coactivators SRC1, SRC2, and SRC3 by miR-644a would not only reduce the activation of the AR but also interfere with the ligand-independent activation of AR by IGF1 and Akt pathways. Further, cdk6 interacts with AR and enhances its transcriptional activity in prostate cancer (41). CBP is part of the p300–CBP complex, which is a transcriptional integrator that modulates a large number of transcription factors. Elevated expression of CBP has been associated with androgen ablation and involved in proliferation and agonistic activity of antiandrogens (42). MiR-644a not only targets AR and c-Myc expression in prostate cancer cells, but it also fine-tunes the expression of AR coactivators offering an advantage over conventional AR-targeted therapies.

Elevated expression of Bcl2 and Bcl-xl and the resulting dysregulation of apoptosis pathway are linked to CRPC and poor prognosis leading to metastasis (43, 44). The role of miR-644a in enhancing apoptosis in both castration-sensitive and -resistant cells is strengthened by validating antiapoptotic genes Bcl2 and Bcl-xl as its direct targets in addition to c-Myc.

MiR-644a may modulate cell proliferation by regulating critical effectors in cell cycle including c-Myc, cdk6, and CCND1 (45–47). In addition to targeting CCND1, a regulator of cell proliferation, miR-644a also downregulates cdk6 which is a binding partner of CCND1 (48). Furthermore, cdk6 plays a significant role in AR signaling by activating AR and is implicated in cell-cycle progression (41). Elevated expression of cdk6 in androgen-sensitive prostate cancer cells in response to androgen is involved in cell cycle and enhanced AR activity (49).

Most importantly, we have revealed a novel role of miR-644a in EMT, a process that plays a crucial role in the development of metastatic CRPC. EMT in prostate cancer is mediated by activation of Snail and ZEB-1 by AR, which leads to repression of E-cadherin and increased expression of β-catenin (50–52). Cdk6 is elevated in prostate cancer cells and correlates with tumor grade, and further cdk6 interacts with AR and enhances AR activity (41, 53). Likewise, both ZEB1 and snail expressions are upregulated in prostate cancer and are implicated in invasion and metastasis (54). By targeting Snail, ZEB-1, SRC-1, and cdk6 together, miR-644a leads to downregulation of EMT markers including β-catenin, vimentin, and increased expression of upregulation of E-cadherin. Our results validate the effect of miR-644a on EMT in both androgen-sensitive and castration-resistant cells. Taken together, miR-644a may modulate EMT in prostate cancer cells in an AR-independent manner by targeting the critical regulators in prostate cancer.

As a master regulator of mRNA expression of multiple oncogenes, miR-644a may modify the cell's proteome and thus alter its phenotype and tumor growth potential. MiR-644a targets and downregulates factors involved in tumor energy metabolism of prostate cancer cells including GAPDH, a glycolytic enzyme implicated in tumor metabolism and proto-oncogene c-Myc, and overexpression of c-Myc contributes to the genesis of many human cancers including prostate cancer (45). Growth factor–mediated expression initiates signaling via c-Myc and PI3K/AKT/mTOR pathways, and c-Myc promotes the expression of genes involved in amino acid transport and protein synthesis, therefore unbalancing the energy homeostasis, in favor of tumor growth (55). Interestingly, aerobic glycolysis flux through GAPDH is the rate-limiting step in the pathway and control points in glycolysis and plays an essential role in tumor metabolism. We show that miR-644a targets c-Myc and GAPDH directly and downregulates the glycolysis and glycolytic capacity in prostate cancer cells.

In conclusion, our study shows that miR-644a directly regulates the expression of various CRPC-relevant targets including AR, AR coactivators including SRC-1, SRC-2, SRC-3, CBP, CCND1, and ARA24, antiapoptotic factors including Bcl2, Bcl-xl, oncogene c-Myc, EMT, and finally, drivers including Snail, cdk6, and ZEB1. Our study supports the concept that instead of targeting the function of one oncogenesis-promoting factor by the drug, e.g., AR targeting by AR antagonist, it would be potentially beneficial to fine-tune the expression of multiple pathways by miRNA replacement adjunctive therapy and to potentiate therapeutic effects of androgen signaling inhibitors enzalutamide and abiraterone acetate. By in vivo, in vitro, and animal experiments, we show that a strikingly large number of genes are posttranscriptionally regulated by miR-644a either directly or indirectly. Specifically, experiments in our animal model using 22RV1 cells containing AR-V7 show that miR-644a sensitizes the tumors to enzalutamide treatment. Although we see a marginal effect with miR-644a synergistically, adjunct therapy by targeting both isoforms of AR and the numerous oncogenesis signaling pathways might be beneficial for patients with advanced disease.

MiR-644a's antisurvival and antiproliferation biological properties appear to play a vital role in androgen signaling, apoptosis, and tumor energy metabolic pathways. We demonstrate that miR-644a inhibits or rather fine-tunes the expression of the molecular pathways, which promote tumor energy metabolism and Warburg effect, as well as genes that are drivers of CRPC development of metastasis, in a coordinated manner, support the strong potential of miRNA therapeutic.

No potential conflicts of interest were disclosed.

Conception and design: J.S. Ebron, C.M. Weyman, S. Gupta, G.C. Shukla

Development of methodology: J.S. Ebron, E. Shankar, C.M. Weyman, D.J. Lindner, G.C. Shukla

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.S. Ebron, E. Shankar, S. Gupta, D.J. Lindner, G.C. Shukla

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.S. Ebron, E. Shankar, C.M. Weyman, S. Gupta, M.J. Campbell, G.C. Shukla

Writing, review, and/or revision of the manuscript: J.S. Ebron, J. Singh, K. Sikand, C.M. Weyman, M.J. Campbell, G.C. Shukla

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.S. Ebron, G.C. Shukla

Study supervision: G.C. Shukla

Other (constructed different 3′UTR plasmids used for target validation and performed real-time gene expression experiments for various endogenous target genes): J. Singh

Other (performed the screening of miRNAs): K. Sikand

We are thankful to Drs. John Kirwan and Anny Mulya, Department of Pathology, Lerner Research Institute, Cleveland Clinic, for help with Seahorse flux analyzer. Research in G.C. Shukla lab is supported by the grants W81XWH-14-1-0508 and W81XWH-14-1-0509 from the Department of the Defense Prostate Cancer Research Program. We acknowledge the use of resources and services provided by the PCBN funded by the grants PCBN W81XWH-10-2-0056 and W81XWH-10-2-0046. G.C. Shukla lab also acknowledges the support of CSU Faculty Research Development funding and John Vitullo Bridge funding provided by the Center for Gene Regulation in Health and Disease. S. Gupta lab is supported by the Department of Defense W81XWH-15-1-0558, USPHS R21CA193080, R03CA186179, and VA Merit Review 1I01BX002494 grants. X. Liu acknowledges support from an R01 CA157429 grant from NIH.

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.