MicroRNAs in Prostate Cancer: Small RNAs with Big Roles

MicroRNAs (miRNAs) are small non-coding RNAs that regulate gene expression either by mediating translational repression or reducing the stability of a target mRNA. Deregulated expression of miRNAs is a common feature of human cancers and a growing body of evidence demonstrates the role of miRNAs as either oncogenes or tumor suppressors. Many miRNAs have been reported to be associated with the pathogenesis of primary prostate cancer (PCa) and the development of castration resistant prostate cancer (CRPC). PCa is the most common cancer and second leading cause of cancer death in American men. Although patients with primary PCa can be treated with chemotherapy and hormone therapy, many of them will develop resistance to conventional therapies and progress to a more sever condition called CRPC, which remains one of the most difficult cancers to treat. Since emerging evidence suggests miRNAs’ significant roles in the tumorigenesis of primary PCa and CRPC, the potential of using miRNAs as drug targets and biomarkers for primary PCa and CRPC has been gaining more attention. The aim of this review is to summarize recent studies on the involvements and mechanisms of the actions of several miRNAs in the development and progression of primary PCa and CRPC. Additionally, the potentail applications of using miRNAs as biomarkers and drug targets are briefly discussed.


Introduction
MicroRNA and mechanism of action MicroRNAs (miRNAs) are small non-coding RNAs which are approximately 19-23 nucleotides in length and regulate gene expression at the post-transcriptional level [1]. More than two thousand miRNAs have been identified in human genomes since their initial identification in the nematode Caenorhabditis elegans [2,3]. miRNAs have been shown to be important regulators of gene expression [4], impacting cell proliferation, apoptosis, differentiation and many other physiological functions. The biogenesis of miRNA starts from its transcription, by RNA Polymerase II, of a longer precursor called pri-miRNA [5]. This pri-miRNA is then cleaved by the Drosha and DGCR8 complex in the nucleus to generate a 70nucleotide stem-and-loop precursor called pre-miRNA [6][7][8]. After being exported to the cytoplasm by Exportin-5 [9], pre-miRNA is then cleaved by the RNAse Dicer to yield a 20 base pair miRNA/miRNA * duplex. This duplex is further loaded onto Argonaute (Ago) proteins to form a RNA-induced silencing complex (RISC) [10]. miRNAs suppress their target genes either by mediating translational repression or reducing the stability of a target mRNA via induction of its deadenylation and degradation [11][12][13][14][15][16]. miRNAs bind to the 3'-untranslated region (3' UTR) of target mRNAs through partial sequence complementarity [2,3,17] and the critical determinants of binding specificity are the nucleotides from position 2 to position 7, comprising the miRNA "seed sequence" [4,14,18,19]. Thus, miRNAs that have the same "seed sequence" are predicted to have highly overlapping sets of mRNA targets and thus similar biological functions [20].

MicroRNA and carcinogenesis
Deregulated expression of miRNAs is a common feature of human cancers and a growing body of evidence demonstrated the role of miRNAs as either oncogenes or tumor suppressors. miRNAs modulate tumorigenesis through multiple mechanisms. For example, miRNAs have been shown to function as oncogenes by inhibiting key molecules in anti-oncogenic pathways. Deletion of a miRNA that targets and represses oncogenes can lead to the activation of oncogenic pathways [21]. Furthermore, recent reports show that chromosomal translocations can sometimes cause truncation of the 3' UTRs of oncogenes, leading to loss of miRNA-mediated repression and promotion of oncogenic transformation [22]. Several miRNAs have been frequently reported to act as oncogenic miRNAs in various human cancers, especially in human prostate cancer (PCa), including miR-21, the miR-125 family and the miR-221~222 cluster. Their functions in the carcinogenesis of various human cancers, especially prostate cancer will be discussed in more detail in the next sections.
On the other hand, several miRNAs have been implicated as tumor suppressors based on their frequent deletion or reduced expression in multiple human cancers [23]. Several tumor suppressor miRNAs suppress oncogenic pathways by repressing the expression of critical oncogenes. For example, members of the let-7 family are reported to repress the RAS oncogene in lung cancer [24], decrease the mRNA of MYC, and lead to reduced proliferation in Burkitt lymphoma cells [25]. Several miRNAs have been indicated to act as tumor suppressors in PCa, such as the miR-15a~miR-16-1 cluster, the miR-34 family and miR-200 family. Their roles in modulating carcinogenesis of prostate cancer and other human cancers will be discussed more thoroughly later.

Prostate cancer and castration resistance
Prostate cancer is the most common cancer among American men, with 220,800 estimated new cases in 2015. Typically, patients with localized prostate cancer are treated with radical prostatectomy or primary radiation therapy and around 70% of those patients are cured. However, approximately 30% of patients will relapse and require hormone therapy in the form of androgen deprivation, also called "chemical castration". Patients receiving this conventional hormone therapy will typically be treated with agents such as leuprolide, which lowers serum testosterone and is often combined with antiandrogens such as bicalutamide. Although these conventional agents are initially effective, most patients will eventually develop resistance to hormone therapy. This relapsed disease is called castration-resistant prostate cancer (CRPC) and is responsible for almost all prostate cancer deaths in the US (~30,000 annually). This makes prostate cancer the second leading cause of cancer death amongst American men. Although CRPC can be treated with taxane-based chemotherapy, this kind of treatment provides only a moderate survival benefit of 1-2 months [26,27]. Since none of the conventional therapies can provide significant clinical benefit, CRPC remains one of the most difficult cancers to treat.

Deregulation of miRNAs in PCa
The deregulation of miRNAs has been frequently reported in primary PCa and also CRPC samples [34][35][36]. Porkka and colleagues have profiled the miRNA expression in 6 PCa cell lines, 9 PCa xenografts samples, 4 benign prostatic hyperplasia (BPH), and 9 PCa tumor samples. In this study they demonstrated 51 individual miRNA have been differentially expressed and the miRNAs expression profile can be used to further cluster the tumor samples into hormone naïve versus hormone refractory PCa [35]. This data not only demonstrates the important role of miRNAs in mediating PCa carcinogenesis, but also suggests that miRNAs can be useful as novel therapeutic targets or biomarkers for diagnosis and classification of PCa. Recently, more than 50 miRNAs have been reported to be involved in PCa carcinogenesis and the functions of many have been identified as well, including miR-21, miR-32, miR148a, miR-221~222, miR-616, miR-15a~16-1, miR-200, miR-101, miR-126, miR-330, miR-34 family and miR-125 family. The function of these miRNAs in PCa is summarized in Table 1.

miRNAs and androgen receptor (AR) signaling
As discussed above, the signaling pathway related to androgen receptor (AR) plays crucial role not only in the carcinogenesis of primary PCa, but also in the development of CRPC. A deregulated level of AR expression is one of the primary mechanisms of resistance to antiandrogen therapy, which restores the AR signaling and promotes the growth of CRPC [29]. Another known mechanism of resistance to hormone therapy is mutations in AR. For example, F876L, an identified mutation in AR, can convert antiandrogen enzalutamide into an AR agonist and lead to resistance to therapy [37]. Because of the importance of AR signaling, the interactions of AR and several miRNAs play crucial roles in the carcinogenesis of PCa and development of CRPC.
A good example is miR-21, whose promoter has an androgenresponse element (ARE). Androgen-induced AR can directly bind to the promoter of miR-21 and cause the overexpression of miR-21, consequently leading to castration-resistant growth of the PCa cell line [38]. Another microRNA with an ARE in its promoter is miR-125b-1, which can be directly regulated by AR binding to its promoter region and recruiting chromatin remodelers EZH2 or JMJD3. Repression of miR-125b-1 by androgen actually protects some target genes, such as the prostate cancer growth factor IGF1R, from degradation, which rescues the androgen-induced cell proliferation in LNCaP cells [39]. miR-141, one member of the miR-200 family, is also regulated by AR in both PCa cell lines and xenografts. Up-regulation of AR leads to ectopic overexpression of miR-141 and enhanced LNCaP cell growth [40]. Two other important targets of miR-141 are PTEN and CDKN1B (p27), which are also important mediators of AR-signaling-regulated proliferation and apoptosis [41,42]. The expression level of miR-34a is negatively correlated with AR level and reduced miR-34a expression confers paclitaxel resistance via derepression of its target BCL2 [43]. Rokhlin et al. reported that cooperation between miR-34a and miR-34c plays a crucial role in AR-dependent p53-mediated apoptosis in PCa cell lines [44]. miR-221 and miR-222 are two other miRNAs being down-regulated by AR signaling in PCa. Stable overexpression of miR-221 in LNCaP cells reduces the transcription of a group of androgen-responsive genes and confers androgen independent growth [45]. Down-regulation of miR-15a and miR-16-1 is also very common in advanced PCa. Restoration of miR-15a and miR-16-1 in the 13q14defective LNCaP androgen-dependent cell line leads to growth arrest and apoptosis, possibly due to targeting BCL2 and CCND1 [46].
Since miR-21, miR-221~222, miR-15a~16-1, miR-200 family (miR-141), miR-34 family and miR-125 family are among the most thoroughly studied miRNAs in PCa and their functions are closely regulated by AR signaling, we will discuss their function in PCa in more detail in next section. Moreover, their relationships with AR signaling and related cellular function are summarized in Figure 1. Limited by the scope of this manuscript, other PCa related miRNAs will only be briefly summarized in Table 1 as well.
Oncogenic miRNAs in PCa miR-21: Among the oncogenic miRNAs in PCa, miR-21 is one of the most thoroughly studied. The oncogenic role of miR-21 is strongly suggested by the observation that it is overexpressed in many different human cancers, including diffuse large B cell lymphoma, acute myeloid leukaemia, chronic lymphocytic leukaemia and Hodgkin's lymphoma [47][48][49][50]. Overexpression of miR-21 in a transgenic mouse model leads to a pre-B malignant lymphoid-like phenotype and the tumors regress completely when miR-21 is inactivated [51]. In addition, overexpression of miR-21 in another transgenic mouse model (CAG-miR-21; K-ras LA2 ) leads to enhanced tumorigenesis of non-small-cell lung cancer (NSCLC), by repressing multiple negative regulators (Spry1, Spry2, Btg2, Pdcd4) of the Ras/MEK/ERK pathway. In contrast, genetic deletion of miR-21 results in impaired Ras-driven tumor development [52].
In prostate cancer, overexpression of miR-21 has been demonstrated in androgen-independent, more malignant, and metastatic PCa cell lines, including PC-3 and DU-145. Interestingly, its expression is barely detectable in the androgen-dependent LNCaP cell line. In this study, Li et al. also showed that suppression of miR-21 decreases the motility and invasion of PCa cell lines, driving them to undergo apoptosis through modulating the target genes such as MARCKS, TPM1 and PDCD4 [53]. Ribas et al. reported that AR directly binds to the miR-21 promoter miPPR-21 and leads to miR-21 overexpression, thus elevating the level of miR-21 that is required for androgen-dependent PCa growth. More interestingly, overexpression of miR-21 alone is sufficient to confer castration resistance to androgen-dependent PCa tumors [38]. miR-21's involvement in mediating androgen-independent growth of PCa was shown in yet another study, in which reduced expression of miR-21 in an androgen-independent PC-3M-MM2 cell line lead to reduced cell viability, migration and invasiveness. The expression and function of miR-21 in this cell line is dependent on Akt signaling [54]. Notably, miR-21's potential application as both a diagnostic and prognostic biomarker in PCa has also been demonstrated by Yaman Agaoglu et al., who found that the miR-21 level was not only significantly higher in the blood plasma of 26 patients with localized PCa compared to 20 healthy controls, but also higher in the serum of advanced prostate cancer patients compared to localized PCa patients [55]. Furthermore, serum level of miR-21 is highest in patients with CRPC compared to patients with androgen-dependent PCa, which suggests its potential to be utilized as a predictive biomarker as well [56]. More recently, high stromal expression of miR-21 has been proved to be associated with poor biochemical recurrence-free survival after radical prostatectomy in PCa patients with high Gleason score [57].
The miR-221~222 cluster: Another member of the PCa related oncogenic miRNA is the miR-221~222 cluster. The miR-221~222 cluster is frequently overexpressed in a variety of human cancers, including prostate carcinoma, glioblastoma, thyroid papillary carcinoma, non-small cell lung cancer, breast cancer and diffuse large B cell lymphoma [58][59][60][61][62][63]. Notably, the fact that the ectopic expression of the miR-221~222 cluster is at a much higher level in the androgen- the dihydrotestosterone (DHT) induced up-regulation of prostatespecific antigen (PSA) and promoted androgen-independent growth of this androgen-dependent PCa cell line. Moreover, this androgenindependent growth is largely dependent on an ectopic level of miR-221~222 because knock-down of this cluster restored the response to DHT induction of PSA [65]. One of the known mechanisms underlying miR-221~222 mediated castration resistance is their direct binding to and suppression of the p27 Kip1 gene [58][59][60][61][62][63][64][65][66], consequently rescuing PCa cells from cell cycle arrest at the G1 phase. miR-221~222 has also been reported to suppress HECTD2 and RAB1A genes and the down-regulation of HECTD2 significantly enhances AR mediated transcription and androgen-independent PCa cell growth [45]. Similar results have been reported in vivo. Mercatelli et al showed that anti-miR-221-222 antagomir treatment in the castration resistant PC-3 xenografting model led to reduced tumor growth by removing the inhibition on p27 [66]. Yang et al., also reported that down-regulation of miR-221 and miR-222 leads to decreased cell proliferation, migration and increased apoptosis in androgen-independent PCa cell line [67].
Similar to miR-21, miR-221 has been reported to be upregulated in the blood plasma in patients with localized PCa compared to healthy individuals, suggesting its potential to be used as a diagnostic biomarker for localized PCa. miR-221 is also more highly expressed in advanced PCa patients compared to localized PCa patients, suggesting its potential to be a prognostic biomarker as well [55]. miR-221 is recently reported as an independent predictor for cancer-related death of high-risk PCa patients, by partially regulating JAK/STAT signaling pathway [68].
The miR-125 family: In contrast to miR-21 and miR-221~222, the relationship between the miR-125 family and carcinogenesis is much more complicated because its members have been reported to function as either tumor suppressor miRNAs or oncogenic miRNAs in different types of tumors [69,70]. The miR-125 family is highly conserved and consists of three miRNAs: miR-125a, miR-125b-1 and miR-125-2. Considering the oncogenic properties of this family, miR-125b-1 has been reported primarily to promote carcinogenesis in various types of tumors and cell lines. miR-125b-1 have been previously reported to be overexpressed in pancreatic cancer, oligodendroglia cancer and especially in prostate cancer [70][71][72][73]. White et al. has reported that miR-125b-1 is overexpressed in all the five AR-positive (CWR22R, PC-346C, LNCaP, cds1, cds2) and four AR-negative (DU145, PC-3, pRNS-1-1, RWPE-1) prostate cancer cell lines [74]. Moreover, the level of miR-125b-1 in PC-3 cell lines is regulated directly by androgen-AR signaling and transfection of synthetic miR-125b-1 promotes androgen-independent growth of LNCaP cells, partially by targeting one of its target genes, BAK1. Since knockdown of BAK1 alone cannot fully restore the effect of overexpressed miR-125b-1, more targets are expected to be involved in mediating miR-125b-1's function in PCa [72]. Ozen and colleagues examined the miRNA expression profile as well as their potential targets in PCa tumor samples and found that EIF4EBP1 is another miR-125b-1 target involved in mediating PCa carcinogenesis [75]. The other reported miR-125b-1 targets in PCa are TP53 and PUMA [72][73][74]. More recently, Amir et al. reported that miR-125b-1 represses p14 ARF , which releases the suppression on Mdm2 and inhibits p53. Consequently, this inhibits p21 and Puma, leading to suppression of apoptosis in PCa cell lines. Treatment of PCa cells with a miR-125b-1 inhibitor can reverse its action by increasing p14, decreasing Mdm2 and inducing apoptosis [76]. Knock-down of miR-125b-1 using RNA interference in PC-3 prostate cancer cells leads to reduced proliferation, which can be restored by the reintroduction of synthetic mature miR-125b-1 [77]. Taken together, these results confirm the role of miR-125b-1 in conferring castration resistance in prostate cancer.
However, the relationship between miR-125b-1 and AR illustrates a good example of the complexity of miRNA-mediated regulation in PCa carcinogenesis. Sun et al. demonstrated that AR directly binds to the promoter region of miR-125b-1 and down-regulates its expression to protect multiple AR target genes from degradation, such as IGF1R, which plays an important role in carcinogenesis of metastatic CRPC [39]. These results are controversial to the previously demonstrated role of miR-125b-1 in promoting PCa carcinogenesis and conferring castration resistance growth.

Tumor Suppressor miRNAs in PCa
The miR-34 family: MicroRNAs belonging to the miR-34 family, miR-34a, miR-34b and miR-34c, have also been reported to be tumor suppressors and modulators of p53 function in regulating cell cycle arrest and apoptosis [78]. The genomic loci of miR-34a or miR-34b~c are frequently deleted in various human cancers including pancreatic cancer, neuroblastoma, breast cancer, lung cancer and prostate cancer [79][80][81][82]. Ectopic expression of miR-34a in the p53 mutant colon cancer cell line or miR-34a~c in human lung fibroblast cells significantly induces cell cycle arrest [83,84] by repressing targets involved in cell cycle regulation, including Cyclin-E2, CDK6 and E2F5 [78,83]. miR-34a can also negatively regulate cell cycle machinery by repressing E2F1 and E2F3 levels [85]. Moreover, the miR-34 family may modulate p53-mediated apoptosis by repressing the expression of the anti-apoptotic protein Bcl2 [83,86].
In prostate cancer, doxorubicin (DOX) treatment led to 3-fold increase of miR-34a and apoptosis, which can be counterbalanced by knocking down AR. Interestingly, this DOX induced miR-34 upregulation does not exist in AR-negative PCa cell lines such as DU-145 and PC-3, or AR-positive LNCaP cells being cultured in androgen free medium. Moreover, the AR-dependent inhibition of p53 suppressed the expression level of both miR-34a and miR-34c. On the other hand, simultaneous inhibition or forced overexpression of both miR-34a and miR-34c modulated the apoptosis induced by DOX [44]. It is reported that prostate cancer stem cells (CSCs) are enriched in the CD44+ population and that miR-34a is under-expressed in these CSCs isolated from xenograft and primary tumors. Enforced expression of miR-34a in these prostate CSCs cells inhibited tumor regeneration and metastasis, while the expression of antagomirs against miR-34a promoted carcinogenesis and metastasis. miR-34a's function in modulating prostate CSCs survival is due to its targeting of CD44 because knockdown of CD44 can recapitulate miR-34a overexpression in prostate CSCs [87]. Members of the miR-34 family have potential to be diagnostic markers since they are significantly overexpressed in the serum of patients with stages 3 and 4 PCa compared to normal control cohorts.
As discussed above, since one miRNA can simultaneously regulate hundreds of target genes and its regulation is frequently modest [11,15], the relationship between miRNAs and their targets can be very complicated. Although many previous in vitro studies have proposed miR-34 family as key downstream effectors of p53 and that they have crucial roles in regulating PCa carcinogenesis, miR-34 is not required for p53-mediated cell cycle arrest or apoptosis in a miR-34-deficient mouse model. Furthermore, in contrast to previous in vitro studies, genetic ablation of miR-34 is not sufficient to accelerate tumorigenesis in different in vivo settings [88]. More recently, Cheng et al. reported that prostate epithelium-specific co-inactivation of both miR-34 and p53 significantly expands the prostate stem cell compartment and develop adenocarcinomas, while inactivation of either one has no such effect [89]. These results suggest that the function of certain miRNA in carcinogenesis is highly contextual and condition dependent.
The miR-15a~miR-16-1 cluster: Another notable member of the family of tumor suppressor miRNAs is the miR-15a~miR-16-1 cluster. The genomic region of the miR-15a~miR-16-1 cluster (13q14) is deleted in more than half of B cell chronic lymphocytic leukemias (B-CLL) [90]. Deletion of miR-15a~miR-16-1 promotes the development of clonal lymphocytosis in the peripheral blood and leads to B-CLL in mice [91]. Furthermore, altered expression of miR-16-1 is associated with the development of autoimmune and B lymphoproliferative disease (B-LPD) in New Zealand black (NZB) mice, a mouse model for human chronic lymphocytic leukemia (CLL) [92]. In prostate cancer, miR-15a and miR-16-1 levels are significantly down-regulated likely due to its targeting of CDK1 and CDK2 associated with cell cycle control and cellular proliferation [93]. Although the miR-15a~miR-16-1 cluster is also reported to target multiple oncogenes including BCL2, CCND1 (encoding cyclin D1) and WNT3A in prostate cancer [46,94], the identities of the critical targets of these miRNAs remains unclear [21]. Moreover, serum level of miR-16 has been reported to be up-regulated in patients with stages 3 and 4 PCa compared to normal donors, suggesting its potential to be used as a diagnostic marker of PCa [95]. and metastasis potential of PC-3 cells [102]. Similar results were also confirmed in another EMT model of DU145-LN4 that decreased expression of miR-200 family induced and EMT phenotype [103].
Since the level of miR-141 was found to be correlated with serum PSA level, the miR-200 family also has promising potential to be used as a diagnostic biomarker for PCa [104]. Serum levels of miR-141 and miR-200b have been found to be significantly higher in advanced PCa patients compared to localized PCa patients, and their levels are elevated with increasing tumor stage as well [105,106]. These results strongly suggest the potential application of using the miR-200 family as prognostic biomarkers.

Conclusion and Future Challenges
Although the study of miRNAs' function in prostate cancer carcinogenesis only began less than a decade ago, huge progress has been made in demonstrating the critical role of miRNA in primary PCa carcinogenesis, as well as in the development of castration resistant prostate cancer. More than 50 miRNAs have been reported to be involved in PCa carcinogenesis and the functions of many have been identified as well, including miR-21, miR-32, miR148a, miR-221~222, miR616, miR-15a~16-1, miR-200, miR-101, miR-126, miR-330, miR-34 family and miR-125 family. Limited by the scope of this review, we only discussed those most thoroughly studied in PCa (others were also summarized in Table-1). As discussed above, these miRNAs play important roles in modulating the progression of PCa tumors and are also involved in the development of castration resistance in CRPC. Because of the high genetic heterogeneity, CRPC is still one of the most difficult cancers to treat with conventional chemotherapy and hormone therapy, underscoring the importance of developing novel biomarkers and drug targets. Therefore, miRNAs' potential for use as treatment targets and biomarkers in PCa has attracted continued attention in recent years. However, there are still many challenges and more work need to be done to address these challenges.
One of the future challenges is effectively identifying the key targets of miRNAs in PCa carcinogenesis, because one miRNA can simultaneously regulate hundreds of target genes and its regulation is frequently modest [11,15]. Many new approaches have been developed in recent years for this purpose, such as HITS-CLIP (high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation) and SILAC (stable isotope labeling with amino acids in cell culture) [11,107,108]. Combined with RNA-Seq and functional library screening, the identification of function relevant miRNA targets in PCa is much more effective. Another challenge in utilizing miRNAs as novel drug targets for PCa is developing antagonist and delivery system with high efficiency and specificity. Various types of chemically modified anti-miRs have been utilized to repress miRNAs, such as locked nucleic acid oligonucleotides (LNAs), polylysine-conjugated peptide nucleic acids (PNAs) and phosphorodiamidate morpholino oligomers (PMOs) [109][110][111][112]. However, because the cellular internalization of these hydrophobic, large anti-miRs is often ineffective, a more efficient delivery system needs to be developed. Some attempts have been made to enhance the cellular uptake of anti-miRs by conjugating the antisense oligonucleotides with cellpenetrating peptides (CPPs), such as Tat, Ant, MPG and more recently, Polymer nanoparticles (NPs) [109,113,114].
Overall, because of the complicated relationship between miRNAs and their targets in different cell contexts, there is still a lot of work to be done to determine the key mechanisms by which miRNAs regulate PCa carcinogenesis and confer castration resistance, and design novel agents and effective delivery approaches for miRNA targeting therapy.