The transcriptional regulation of miR-21, its multiple transcripts, and their implication in prostate cancer

Abstract

MicroRNAs (miRNAs) are a natural part of the most recently discovered and global regulatory pathway known as RNA interference. Functional studies have shown how specific miR-NAs can function as tumor suppressors or oncogenes and, correspondingly, deregulated miRNA profiles have been observed in prostate and other cancers. However, the upstream pathways which regulate miRNA expression are only currently being uncovered. The Androgen Receptor (AR) is a nuclear hormone receptor and transcription factor which plays a paramount role in prostate cancer (PCa) pathobiology. We performed high throughput miRNA microarray analysis on two AR-responsive cell lines to identified 16 candidate AR-regulated miRNAs.¹ One of the most androgen-induced candidates was a known oncogenic miRNA, miR-21. In a small study of early grade PCa samples we found that miR-21 levels were frequently elevated in comparison to adjacent normal tissue. This observation was supported in the literature^2,3 and suggests clinical relevance. We found that the activated AR directly interacts with miR-21 regulatory regions, indicating direct transcriptional induction. Furthermore, we provide new reporter studies supporting AR-regulation. Importantly, in functional studies, we found that a modest overexpression of miR-21 enhanced tumor xenograft growth and was sufficient to support androgen-independent proliferation following surgical castration. Thus, our studies suggest a model where miR-21 contributes to androgen-dependent and androgen-independent PCa growth. However, the AR is only one of many reported transcriptional regulators of miR-21. Here we review our recent discoveries and further analyze the reported miR-21 regulatory regions, inhibitory and stimulatory signaling pathways, and primary transcripts.

Introduction

MicroRNAs (miRNAs) recently emerged as important regulators of several physiological and pathological processes. Importantly, abnormal miRNA expression has been documented in most all common solid tumors and leukemias, including Prostate Cancer (PCa).^2–10 As with traditional oncogenes and tumor suppressors, miRNA genes are located in sites of frequent amplification, loss-of-heterozygosity, or chromosomal breakpoints.^11,12 Moreover, in several cancers well known oncogenic transcriptional and signaling pathways have been implicated in miRNA gene expression.^13–15 Importantly, many miRNAs themselves have also been shown to have oncogenic as well as tumor suppressor properties (reviewed in refs. 16 and 17). Consequently, there has been intense study to identify the upstream pathways which regulate aberrant miRNA expression in cancer as well as the downstream pathways which miRNAs themselves regulate. We recently began studies to interrogate the role of miRNAs in PCa.

Androgen-Regulated miRNAs in Prostate Cancer

Androgen signaling is a pivotal component in the biology of the normal prostate and PCa. Androgen action is mediated through the Androgen Receptor (AR), a ligand-dependent transcription factor and a member of the hormone nuclear receptor superfamily.¹⁸ In physiological conditions, AR drives the prostate epithelia to arrest proliferation and differentiate. However, in cancer AR-signaling is modified to facilitate growth and survival. The most common treatment for PCa patients with advanced and metastatic disease consists of androgen-ablation that, after a favorable response, invariably will result in relapse and androgen-independent disease.¹⁹ Such Castration Resistant Prostate Cancer (CRPC) has been shown to be unresponsive to further androgen-deprivation therapy and yet to have an active AR signaling pathway through aberrant AR-reactivation.²⁰ For example, it was recently discovered that alternative splicing can generate a constitutively active AR²⁰ or, alternatively, ligand independent variants.²¹ Thus, AR itself and its downstream pathways remain valid targets for therapeutic intervention either in early PCa or CRPC. In view of the elemental and broad role of AR in the illness, we surmised that AR likely regulates several miRNAs that can participate in PCa growth and castration resistance.

Using a miRNA microarray screen¹³ we recently identified 16 androgen-induced miRNAs, including miR-21. We further confirmed miR-21 as androgen responsive by northern blot in several AR-positive PCa cell lines.¹ To determine whether the AR directly regulated miR-21 gene transcription, Chromatin Immunoprecipitation (ChIP) was performed. Our results confirmed androgen-dependent binding of AR to the reported promoter of miR-21, miPPR-21.²² Other groups have also documented the existence of AR-regulated miRNAs in PCa. For instance miR-125b was found to be androgen-responsive in a study of androgen-dependent versus androgen-independent cell lines. MiR-125b functions by inhibiting the translation of the pro-apoptotic gene, BAK, and was capable of supporting androgen-independent growth in vitro.²³ A separate study found miR-338 to be the only miRNA induced after androgen-treatment of LNCaP.⁴ However, we were unable to detect androgen-induced expression of miR-125b or miR-338 in our PCa models. Additional and more detailed studies are required to further characterize the androgen regulation of these three miR-NAs and the clinical significance of these phenomena.

miR-21 Contributes to Prostate Cancer Pathogenesis

In gain of function studies we found that elevated miR-21, in two androgen-dependent PCa cell lines, was sufficient to drive androgen-independent growth. Additionally, we observed that miR-21 was enhancing androgen-supported proliferation of LNCaP cells. In mouse models, elevated levels of miR-21 improved tumor establishment, increased tumor growth and induced a castration-resistant phenotype. Thus, our data supports a model whereby AR activation directly induces miR-21 expression resulting in enhanced PCa growth and castration resistance (Fig. 1). Studies of miR-21 function in various cancer models indicate diverse roles in proliferation, invasion, intravasation, anchorage-independent growth, motility, metastasis and inhibition of apoptosis.^24–30 However, differently from other studies, we were unable to establish any link between miR-21 overexpression and enhanced survival either when depleting androgens from the media or when challenging the cells with several pro-apoptotic stimuli.¹ Results from a different group, using the drug Staurosporine (STS), support that miR-21 does not affect the apoptotic death of LNCaP cells.²⁷ We therefore postulate that miR-21 potentiates PCa growth, rather than survival. Differently from what was observed with LNCaP, Li et al.²⁷ report that miR-21 inhibits the modest levels of STS-induced apoptosis in two AR-negative and androgen-independent metastatic cell lines, PC-3 and DU-145. In addition, their data support that miR-21 is regulating motility and invasion in these specific cell lines. Therefore, future experiments are needed to decipher the role of miR-21 in the oncogenesis of androgen-dependent as well as androgen-independent PCa.

Figure 1.

MiR-21 mediated pathways. Several signaling pathways, including AR-mediated, stimulate miR-21 expression. Elevated miR-21 levels contribute to androgen-dependent proliferation and are able to fully support castration resistance. Those pathways in red indicate novel pathways which we recently identi3ed PCa models.¹ Pathways in black refer to the previously reported pathways described in other models.

To interrogate if miR-21 was elevated in human PCa tumors, we quantified its expression in 10 tumor-normal matched samples from early grade radical prostatectomies. In support of our model, 6 of the 10 tumors tested showed a statistically significant increase in miR-21 expression. On average, miR-21 was upregulated 3.52 fold in PCa. miR-21 expression did not correlate with stage or grade; however this might be explained by the small size of the sample set.¹ Given the functional contribution of miR-21 to castration-resistant disease in our models, we anticipated that more advanced PCa specimens would express high levels of miR-21. To investigate this we have completed additional analysis of an independently-published data set from the literature where miRNA expression was directly compared in hormone naïve versus CRPC.² In this small data set, Porkka and colleagues present normalized miRNA expression array data from 3 benign hyperplastic prostates (BPH), 5 hormone naïve PCa specimens, and 4 hormone refractory cases (CRPC). Our analysis of this data set reveals that miR-21 expression was significantly higher in CRPC when compared to BPH or hormone naïve PCa. Clearly larger studies are necessary to verify the correlation of miR-21 with aggressive or CRPC cases. Given its connection with androgen-independence, we anticipate that miR-21 expression may be a prognostic marker for the early detection of aggressive PCa. Indeed, elevated miR-21 expression has been indicative of aggressive disease in other cancers.^31–37 It is not yet clear if the other known AR-induced miRNAs, miR-125b and miR-338, are similarly elevated in clinical or advanced disease. The study which found miR-338 to be androgen responsive did not indicate significant miR-338 elevation in PCa specimens. Two reports have also found decreased, rather than increased, levels of miR-125b in PCa tumors.^2,4

The miR-21 Gene Locus

Special attention has been paid to miR-21 since it was identified in 2005 as an apoptotic suppressor in the highly malignant glioblastoma.³⁸ More strikingly, one year later, miR-21 was found to be the only miRNA consistently upregulated in 6 solid tumors, including PCa.³ Multiple studies have followed indicating that miR-21 is a cancer-related miRNA with oncogenic potential.^24–30 Curiously enough, miR-21 maps to chromosome 17q23.2, an area reportedly amplified in several kinds of tumors as well as PCa.³⁹ Several reports have interrogated whether genome alterations could account for the observed changes in miRNA levels. In breast tumors, another cancer with a significant hormonal component, the miR-21 genomic locus was found to be amplified thus providing a non-transcriptional mechanism for increased miR-21 expression.⁴⁰ Albeit, in a previous study Blenkiron et al.⁴¹ found that differences in expression of many miRNAs, including miR-21, could not be explained by chromosomal gain or loss. No conclusive answer exists as to whether higher miR-21 expression in PCa can be partially explained by changes in the genome copy number. On the other hand, there is significant evidence that transcriptional and posttranscriptional mechanisms contribute to increased miR-21 levels. Naturally, there has been great interest in understanding the signaling pathways which regulate miR-21 expression. Above we have summarized our new findings that miR-21 expression is regulated by the activated AR. Below we further expand the discussion to review the miR-21 gene locus, regulatory regions, transcripts and regulatory pathways. We believe these complex networks provide numerous pathways, both AR-dependent and AR-independent, for elevated miR-21 expression in PCa and therefore potential novel mechanisms for development of the advanced PCa phenotype (Fig. 1).

The miR-21 Transcriptional and Post-Transcriptional Regulation

Early studies of miR-21 gene regulation revealed that the pairs IL-6/STAT3,⁴² and phorbol 12-myristate 13-acetate (PMA)/ AP-1,²² were effective inducers of miR-21 expression. Additional reports have identified alternative regulators for this miRNA, for instance induction through Ras,⁴³ ERK1/2,⁴⁴ EGFR⁴⁵ and Estrogen Receptor.⁴⁶ Oppositely, miR-21 expression has also been reported to be suppressed by NFI,²² C/EBP,²² Gfi1,⁴⁷ and Estrogen Receptor⁴⁸ (Fig. 2). Some of these pathways have been fine mapped through ChIP and promoter/reporter studies while others have simply reported altered miR-21 expression levels. Before discussing these details, we will first introduce miRNA transcripts and highlight a few points that make miR-21 unique.

Figure 2.

Genomic organization and transcripts of miR-21 gene Locus. The miR-21 hairpin is located on chromosome 17q23.1 immediately downstream of the coding gene, TMEM49. While TMEM49 and miR-21 transcripts overlap, they are independently regulated. White boxes represent terminal TMEM49 exons 11 and 12. Three previously identi3ed and functional promoter regions for pri-miR-21 are represented at the bottom of the 3gure. Promoter elements and positive (blue) and negative (red) transcriptional regulators are mapped above. Transcription Start Sites (TSS) are indicated by arrows. TMEM49 and miR-21 poly(A) signals are indicated by A.

Most miRNAs are organized in conventional RNA polymerase II transcripts which are further post-transcriptionally modified by 5′-methyl-G-capping, splicing and polyadenylation.⁴⁹ These units are known as primary-miRNAs (pri-miR-NAs) and are further processed by cellular machinery to release the functional or mature miRNA from a precursor hairpin (pre-miRNA). Pri-miRNAs can be coding or non-coding genes and the pre-miRNA hairpins are often located in non-coding or intronic gene regions. Interestingly, there are very few miRNAs which have been mapped in the 3′-Untranslated Regions (UTRs) of coding genes. This is not surprising as one might anticipate that miRNA processing and mRNA polyadenylation may interfere with each other, or with transcript stability. miR-21 is one of these rare miRNAs and is located immediately adjacent to the UTR of TMEM49. It is plausible that miR-21 may therefore be expressed as part of the TMEM49 transcript; however, many studies have found that miR-21 and TMEM49 are independently regulated and that local promoter regions initiate miR-21 expression through long, non-spliced and non-coding pri-miR-21 transcripts.^22,42,49 Due to its unique location, all pri-miR-21 transcripts contain the last exon of TMEM49 (Fig. 2). While there have been several studies of miR-21 gene promoters and transcripts, there are many contradictions regarding the actual size of the pri-miR-21 transcript, its minimal promoter, and Transcription Start Site (TSS). For practical reasons, in this review we will designate the promoters and transcripts using the name of the first author describing them.

The first description of a miR-21 regulatory region was by Cai and colleagues⁴⁹ (Fig. 2). This promoter, mapping −3403 to −2395 Kb upstream of the miR-21 hairpin, is constitutively active in 293 embryonic kidney cell reporter studies.⁴⁹ Sequence analysis revealed the presence of a “CCAAT” box transcription control element but did not identify a canonical “TATA” box at the expected location ~25 bp from the TSS. Loffler and colleagues assayed a very similar promoter region extending from −3565 to −2415 upstream of miR-21 hairpin (Fig. 2). They confirmed its activity in reporter assays and found IL-6 enhancement by STAT3.⁴² In a third and more comprehensive study, Fujita et al.²² used an informatics algorithm to identify an independent and conserved promoter within intron 10 of TMEM49. This promoter was specifically termed as miPPR-21. This bioinformatics approach was restricted to 100 Kb upstream the miRNA and positively scored sequence conservation among vertebrates as well as the presence of core promoter elements for RNA polymerase II (TATA, CCAAT and GC box). Fujita and colleagues applied TRANSFAC matrices to analyze miPPR-21 and mapped several conserved enhancer elements including binding sites for activation protein 1 (AP-1) (composed of Fos and Jun family proteins), Ets/PU.1, C/EBPα, NFI, SRF, p53 and STAT3. The Fujita promoter, miPPR-21 (−3770 to −3337 relative to the hairpin, Fig. 2), shows minimal overlap with the Cai promoter and maps inside of TMEM49 intron 10. Interestingly, given the very small overlap of ~60 bases, the Fujita and Cai promoter regions function separately suggesting that each is an independent promoter. In a series of experiments using reporter plasmids with the full miPPR-21 or several truncated forms, Fujita and colleagues confirmed the region as a promoter and demonstrated the activity of three AP-1 and two Ets/PU.1 elements after treatment with PMA. In yet a fourth study, Ozsolak and colleagues combined nucleosome mapping with chromatin signatures to decipher the position of miRNA promoters.⁵⁰ They identified and assayed the transcriptional activity of 4 regions upstream of the miR-21 hairpin. Two of these were very similar to Cai and Fujita promoters. When comparing these promoters, they observed a superior induction with the region identified by Fujita (miPPR-21) when compared to Cai. Interestingly, Ozsolak and colleagues found that the Cai promoter was active in HeLa cells, although the same promoter was not able to induce transcription in a melanoma cell line. Thus, there may be cell lineage specificity for the usage of one or other promoter.

From the previous promoters, only miPPR-21 presented a highly conserved androgen response element (ARE) in its sequence. In ChIP studies we proved the androgen-dependent recruitment of AR to miRPP-21 at levels similar to that of the classical androgen-regulated gene, PSA.¹ Since these experiments were completed, Wang and colleagues published a high throughput study of AR-bound cistrons in androgen activated LNCaP cells compared to the androgen-independent LNCaP-abl cell line (in androgen-depleted conditions). The main objective of their study was to detect a change in the AR-mediated transcriptional program between androgen-dependent and androgen-independent PCa. However, in the case of miRPP-21, their results showed the presence of AR-bound ChIP amplicons within the promoter region in both androgen-dependent and androgen-independent PCa cells.⁵¹ These new findings altogether corroborate our original ChIP results using LNCaP, extend its validity to LNCaP-abl cells and support that the AR/miR-21 signaling pathway might be active in androgen-dependent and androgen-independent stages of PCa. To further study the miPPR-21 promoter in an AR-responsive context, we completed a series of reporter studies using luciferase under the regulation of miPPR-21. First, mimicking experiments from Fujita and coworkers, we corroborated the promoter responsiveness to low concentrations of PMA in the PCa cell line C4-2 (Fig. 3A). This is relevant because PMA-regulated AP-1 transcription factors have been reported to be associated with advanced PCa.⁵² To further support AR-regulation of the miPPR-21 region, we quantified the effect of the AR-specific inhibitor, Casodex, on miPPR-21 reporter activity. We were able to observe Casodex-repressed promoter activity further supporting AR-regulation of miR-21 via the miPPR-21 promoter (Fig. 3B). In summary, this suggests that miPPR-21 is active in PCa and remains inducible by PMA and androgens.

Figure 3.

miPPR-21 promoter activity in PCa. We reproduced the strategy reported by Fujita et al.²² to clone miPPR-21 in a PGL3 basic vector (InVitro-gen) further modi3ed to accommodate an Internal Ribosome Entry Site (IRES) upstream the Luciferase ORF. C4-2 cells were plated in 96 well plates 24 hours prior to Lipofectamine 2000 transfection with the indicated vectors and Renilla luciferase control. Fire3y luciferase activity was measured and normalized to a Renilla luciferase activity after lysing the cells. Results are expressed as Luciferase arbitrary units (a.u). The data represent 3 independent measurements. (A) PMA induction. Transfected C4-2 cells were challenged for 5 hours with vehicle (DMSO, white columns) or 2 nM PMA (black columns). (B) AR Inhibition. Transfected C4-2 were maintained for 48 hours in RPMI complete media plus 10% FBS either in combination with Vehicle (DMSO, white columns) or 10 μM of AR-antagonist (Casodex, black columns) prior to lysis.

In addition to the multiple regulatory regions, there have been several reported miR-21 transcripts and TSS’s. Cai and coworkers were the first laboratory to describe a heterogeneous TSS, more concretely by mapping two TSS separated by 27 bp (Fig. 2). The TSS2 was proved to be the major site in 293T cells and produced an ~3.4-kb pri-miR-21 (Genbank number AY699265). One year later, a new Genbank annotation depicted a similar size transcript for miR-21 (BC053563.1), yet with a different TSS. The annotation was the result of an ambitious project launched by the NIH to provide the sequence for a library of full length cDNAs.⁵³ Differently from the former two studies, Fujita and colleagues mapped a novel TSS for pri-miR-21 located 900 bp upstream of the Cai TSS’s. In an attempt to study whether the former TSS’s described by Cai and coworkers were active at the same time, Fujita and colleagues used primer extension experiments and were unable to detect any transcripts initiating at the Cai TSS’s in their models (experiments not shown by the authors). Addressing the question of the actual size of the transcript, Fujita and colleagues utilized northern blots to identify an ~4.3 Kb pri-miR-21 transcript initiated in intron 10 of TMEM49.²² In addition to the multiple putative promoter regions and TSS’s, two poly(A) signals exist downstream of the miR-21 hairpin and are separated by ~700 bp, thus potentially producing transcripts which differ by ~700 bp. Finally the peculiar location of the miR-21 hairpin, immediately downstream of TMEM49 UTR, raises the question of how the three poly(A) sites inside of the body of the pri-miR-21 (readily used by TMEM49) are skipped in order to produce the non-spliced 3.4–4.3 Kb pri-miR-21 transcripts (Fig. 2). Additional studies are clearly required to fine map the minimal pri-miR-21 promoter and to characterize the 5′- and 3′-termini of the dominant miR-21 transcripts. Previous studies indicate that these may vary in different cell and disease models.

In addition to transcriptional regulation, post-transcriptional regulation of pri-miR-21 has been observed. David and colleagues reported that miR-21 was induced after bone morphogenetic protein (BMP) and TGFβ treatment. These studies support that elevated miR-21 levels were due to an increase in the Drosha processing of the pri-miR-21 transcript (mediated by Smad proteins) rather than induced transcription.⁵⁴ In addition, other reports described an increase of several elements of the miRNA processing pathway in PCa which may also affect miR-21 levels. Using tissue gene microarray Chiosea and coworkers reported higher levels of several members of the miRNA machinery in PCa specimens. Interestingly, their analysis revealed that increased Dicer levels correlated with clinical stage, lymph node status and Gleason score.⁵⁵ Alternatively, an independent study by Ambs and colleagues documented increased levels of Drosha, Dicer and DGCR8 (which encodes an essential cofactor for Drosha) in PCa tumors when compared to normal tissue.⁴ Hence, increased levels of mature miR-21 in PCa could be similarly explained by a superior performance of the miRNA processing pathway.

Concluding Remarks

We recently uncovered that miR-21 can be directly induced by the activated AR, thus placing miR-21 as a downstream effector in the androgen-signaling pathway. As a result of miR-21 enhanced expression, LNCaP cells were more proliferative in the presence of androgens. Pilot studies of tumor samples indicate that miR-21 is already elevated in early grade PCa patients. We also found that elevated miR-21 is sufficient to fully support androgen-independence. This could be a new mechanism contributing to the androgen-resistant phenotype. Supporting our observations, elevated miR-21 has been reported in advanced castration resistant disease. In addition to AR-induction of miR-21 several other known pathways control miR-21 expression. Intriguingly, contradictory reports on the actual location of the miR-21 promoter, primary transcript, and coordinating TSS’s have been published. Here we reviewed several publications addressing these and other matters and found that the promoter initially described by Fujita et al.²² is experimentally supported over the other ones. Subsequently we used the same promoter to construct a reporter vector and corroborated its response to PMA and androgens in PCa. Further studies fine mapping the minimal promoter as well as other regulatory regions, either placed in close proximity or distant to the promoter, are required. Additionally, other posttranscriptional regulatory pathways may be considered. Special interest to Smad proteins and their regulatory function on miR-21 has to be awarded. This fact highlights the existence of specific ways to modulate singular miR-NAs. A better understanding of miR-21, its regulation, and the downstream pathways it regulates will potentially uncover novel tools for cancer diagnosis, prognosis and therapy.

Abbreviations

AR

androgen receptor

PCa

prostate cancer

miRNA

microR-NA

PMA

phorbol 12-myristate 13-acetate

ARE

androgen response element

CRPC

castration resistant prostate cancer

ChIP

chromatin immunoprecipitation

STS

staurosporine

BPH

benign hyperplastic prostate

TSS

transcription start site

AP-1

activation protein 1

UTR

untranslated region

pri-miRNA

primary-miRNA

IRES

internal ribosome entry site