The Tumor Suppressive MicroRNA miR-218 Targets the mTOR Component Rictor and Inhibits AKT Phosphorylation in Oral Cancer

MicroRNAs (miRNAs) are an abundant class of endogenous, small, non-coding RNAs, the products of which are small single-stranded RNAs of 19 to 22 nucleotides with a primary role in posttranscriptional silencing generally through binding to the 3′-UTR of protein-coding transcripts, in turn triggering mRNA degradation or translational repression (1). In human cancers, the expression of miRNAs is generally downregulated in malignant tissues compared with the corresponding nonmalignant tissues (2, 3), suggesting the deregulation of miRNA expression and the contribution of miRNAs to the multistep processes of carcinogenesis either as oncogenes or as tumor-suppressor genes (TSG; refs. 4, 5). Among various epigenetic mechanisms of cancer-related gene-silencing, DNA hypermethylation of CpG sites within CpG-islands is known to lead to the inactivation of many TSGs (6) and several tumor-suppressive miRNAs (TS-miRNAs; ref. 7). In fact, DNA methylation-mediated downregulation of miRNAs by proximal CpG-islands has been described by a number of groups including ours (8–10).

Oral cancer, predominantly oral squamous cell carcinoma (OSCC), is the most common head and neck neoplasm, affecting approximately 270,000 people worldwide in 2002 (11). In Japan, OSCC is relatively common, accounting for more than 5,500 deaths in 2003 (12). The carcinogenesis, including OSCC (13), is considered to arise through the progressive accumulation of multiple genetic abnormalities, which may impair the functions of oncogenes or TSGs that play a crucial role in the development of this disease. In addition, evidence has emerged that epigenetic mechanisms, such as altered DNA methylation patterns, play a significant role in the silencing of TSGs and contribute to malignant transformation during oral carcinogenesis (14).

In this study, we identified and characterized TS-miRNAs frequently silenced by DNA hypermethylation in OSCC and their targets using function-based screening with a cell proliferation assay for 327 synthetic miRNAs and a series of sequential analyses of DNA methylation and expression in OSCC cell lines and primary tumors. The function-based screening presented in this paper is a powerful tool for exploring a large number of double-stranded RNAs (dsRNAs) having tumor-suppressive effects, including TS-miRNAs and siRNAs, as therapeutic agents for several types of cancer cells directly (15, 16). Through this approach, 2 epigenetically silenced TS-miRNAs in OSCC, miR-218 and miR-585, which are located within the intron of SLIT2 and SLIT3, respectively, and a novel target of miR-218, Rictor, were identified in our study. This study is the first to show clearly that the tumor-suppressive activity of miR-585 is epigenetically silenced in OSCC and that miR-218 targets Rictor inducing the activation of a TOR-Akt signaling pathway.

Derivations of and culture conditions for cell lines were reported previously (17). OSCC cell lines were authenticated in our previous studies of array-CGH analyses (17). RT7, human oral keratinocytes immortalized by TERT and human papillomavirus (HPV) type 16 E6/E7 oncogenes, was kindly provided by Dr N. Kamata (Hiroshima University Faculty of Dentistry, Japan). To analyze the restoration of genes, cells were cultured with or without 10 μmol/L of 5-aza 2′-deoxycytidine (5-aza-dCyd) for 5 days. A total of 15 frozen primary samples were obtained from OSCC patients treated at Tokyo Medical and Dental University with written consent from each patient in the formal style and after approval by the local ethics committee.

A total of 10 nmol/L of dsRNA mimicking human mature miRNAs or control nonspecific miRNA (Ambion; Thermo Scientific Dharmacon), and 20 nmol/L of target-specific siRNAs for PIK3CA (Dharmacon) and Rictor (Invitrogen) or their control nonspecific siRNA were transfected individually into OSCC cells using Lipofectamine RNAiMAX (Invitrogen). The numbers of viable cells were assessed by the colorimetric water-soluble tetrazolium salt (WST) assay (Cell counting kit-8; Dojindo Laboratories). The cell cycle was evaluated by a fluorescence-activated cell sorting (FACS) analysis as described elsewhere (17).

Each of gene and CpG-island was searched miRBase database (release April 17, 2011; ref. 18), UCSC Genome Browser on Human February 2009 Assembly (hg19; ref. 19) and PubMed (20). The combined bisulfite restriction analysis (COBRA) and the bisulfite-sequencing analysis, using primer sets designed to amplify regions of interest (Supplementary Table S1), were done as described elsewhere (8).

Real-time reverse transcription PCR (RT-PCR) was done as described elsewhere (8).

Predicted targets for miRNAs and their target sites were analyzed using Microcosm Targets (21), miRanda (22), and TargetScan (23). Anti-caspase-3, anti-cleaved caspase-3, anti-cleaved PARP, anti-Akt, anti-phospho-Akt (Ser-473), anti-Rictor (Cell Signaling Technology) and anti-PIK3CA (Upstate Biotechnology) rabbit polyclonal antibodies were used in Western blotting. Luciferase constructs were made by ligating oligonucleotides containing the 3′-UTR target sites downstream of the luciferase gene in the pMIR-REPORT luciferase vector (Ambion). Luciferase activity was measured as described elsewhere (8).

Differences between subgroups were tested with the Mann–Whitney U test.

To identify TS-miRNAs in OSCC, we first examined 327 synthetic miRNAs mimicking human mature miRNAs by function-based screening in 2 OSCC cell lines, NA and SKN3. The strategy used and partial results obtained are shown in Figure 1A. Since the proliferation-inhibitory effect was made an index for the rating of tumor-suppressive activity in our function-based screening, relative cell growth ratios in Figure 1B and Supplementary Table S2 indicate effects of each synthetic miRNA on cell growth 5 days after transfection. In this first screening, 110 miRNAs, including known TS-miRNAs, such as miR-34 (24), miR-124 (9, 25), miR-193a (8), and miR-491 (15), showed remarkable inhibitory effects on cell growth (relative growth ratio < 0.5). Then we selected novel miRNAs with CpG-islands in the region 5′ upstream or around these genes and excluded miRNAs previously reported to correlate with DNA hypermethylation and their tumor-suppressive activity among 110 miRNAs by means of database analyses. From the first screening, 25 mature sequences of miRNAs emerged as candidates for new TS-miRNAs silenced by DNA hypermethylation in OSCC cell lines (Table 1).

Next, we explored the DNA methylation status of 35 CpG-islands around 25 miRNAs located at 26 loci in a panel of 18 OSCC cell lines and a normal counterpart, RT7, an immortalized human oral keratinocyte line, by COBRA (Fig. 1C). Since multiple copies of some mature miRNAs, as listed in Table 1, are transcribed from different loci, the number of mature forms of miRNAs is smaller than the number of genomic loci. In COBRA, on the basis of a comparison of mean values of methylation frequencies in each region examined, frequent DNA hypermethylation (>50% of OSCC lines) was found in only 6 of 25 miRNAs, that is, miR-218, miR-375, miR-383, miR-483, miR-489, and miR-497, although CpG-island hypermethylation on/around these miRNAs was also detected in RT7. We next investigated the consistency in the correlation between DNA methylation status in each region of these 6 miRNA genes by COBRA and their expression patterns in 18 OSCC cell lines and RT7 by TaqMan real-time RT-PCR (Table 2), and found that values were higher in some regions of miR-218-1 (50.0%–77.8%) and miR-375 (100%) than other miRNA genes (0%–56.3%; Fig. 1A–C and Supplementary Fig. S1A–C). miR-218-1 and miR-218-2 are located at 4p15.31 and 5q35.1 within the intron of SLIT2 and SLIT3, respectively (Fig. 2A and B). Interestingly, miR-585, together with miR-218-2, is also located within the intron of SLIT3, and the expression patterns of miR-218 and miR-585 were similar to those of SLIT2 and SLIT3, respectively (Fig. 2B and C); note that the percentage of OSCC cell lines with significant downregulation of miR-218, miR-585, SLIT2, SLIT3, and miR-375 expression (<0.5-fold) was 44.4% (8/18), 55.6% (10/18), 61.1% (11/18), 77.8% (14/18), and 100% (18/18), respectively (Fig. 2C and Supplementary Fig. 1C). To determine the relationship between DNA hypermethylation in CpG-islands on/around miR-218-1, miR-218-2, and miR-375 and the downregulation of gene expression, we treated OSCC cell lines with 5-aza-dCyd, observing a remarkable restoration of the expression levels of these miRNAs and their host genes in 55.6% (10/18), 44.4% (8/18), 33.3% (6/18), 61.1% (11/18), and 100% (18/18) of OSCC cell lines, respectively (Fig. 2C and Supplementary Fig. 1D). Consistent with these results, CpG-island hypermethylation was also confirmed by bisulfite-sequencing in the OSCC lines lacking the expression of those candidate miRNAs, but not in RT7 and the cell lines expressing these miRNAs (Fig. 2A and B and Supplementary Fig. 1E). These results strongly suggest that DNA hypermethylation of CpG sites within CpG-islands on/around these genes deregulates their expressions in OSCC cell lines.

To confirm the growth inhibitory effects of miR-218, miR-585, and miR-375 in the function-based screening, we used 2 synthetic miRNAs purchased from Ambion and Thermo Scientific Dharmacon to take into consideration the off-target effects of dsRNAs. The ectopic expression of these miRNAs in OSCC cell lines lacking their expressions by the transient transfection of dsRNA was evaluated by TaqMan real-time RT-PCR analysis (data not shown). Consistent with the results of the function-based screening (Supplementary Table S2), restoration of miR-218 expression significantly reduced cell growth in the OSCC cell lines tested (Fig. 2D). In addition, a growth inhibitory effect of miR-585 was clearly detected in Dharmacon product, whereas Ambion product of dsRNA mimicking miR-585 was not on the market and could not be examined. In our in vitro analysis, tumor-suppressive functions of miR-218 and miR-585 were confirmed. Ambion product of dsRNA mimicking miR-375 induced growth inhibition whereas Dharmacon product facilitated cell proliferation in both OSCC cell lines, and so miR-375 was excluded from candidates in this study.

To determine whether the CpG-island hypermethylation of miR-218-1/SLIT2 and miR-218-2/miR-585/SLIT3 also occurs in primary OSCCs in a tumor-specific manner, the correlation between DNA methylation status and the expression patterns of these genes in 15 primary OSCCs and corresponding noncancerous oral mucosae was examined using COBRA and TaqMan real-time RT-PCR assay, respectively. In COBRA, the frequency of tumor-specific DNA hypermethylation in regions of miR-218-1 and miR-218-2/miR-585 was 6.7% to 93.4% and 6.7% to 13.3%, respectively, in primary OSCCs (Table 2, Supplementary Fig. 2A). Although the frequency of aberrant methylation at these CpG-islands was relatively low in primary tumors, expression levels of miR-218 and miR-585 in tumors as compared with paired normal oral mucosae were markedly reduced in 73.3% (11/15) and 66.7% (10/15) of primary OSCCs, respectively (< 0.5-fold expression, Supplementary Fig. 2B), and the consistency was comparatively high in these regions of miR-218-1 (75.0%–100%) and miR-218-2/miR-585 (100%). Although miR-218 expression had been reported to be specifically reduced in HPV-positive cell lines and tissues (26), all of OSCC cell lines and primary samples examined in this study were confirmed the negative HPV-status by genomic PCR (Supplementary Fig. S3). Thus, our results suggest that miR-218 and miR-585 are most likely TS-miRNAs frequently silenced through tumor-specific DNA hypermethylation in OSCC.

We examined tumor-suppressive effects of miR-218 and miR-585 on HSC-2 and NA (Fig. 3A and B). Restoration of the expression of these miRNAs significantly reduced cell growth in both cell lines tested. Since a large number of transfectants were rounded and floating compared with the control counterpart at 5 days after transfection, we conducted a FACS analysis and a Western blotting for caspase-mediated apoptosis using both cell lines 72 hours after the transfection. In the FACS analysis, the effect of miR-218 on the cell cycle in HSC-2 cells was weak while the accumulation of NA cells in the G₂/M phase was observed in miR-218-transfectants. Overexpression of miR-585 induced the accumulation of cells in the G₂/M phase and sub-G₁ phase in the HSC-2 and NA cell lines, respectively. However, the Western blotting showed that ectopic expression of miR-218 and miR-585 remarkably increased protein levels of caspase3, cleaved caspase3 and cleaved PARP in both cell lines (Fig. 3C), miR-218- and miR-585-induced reductions of cell growth in these OSCC cell lines were not inhibited by caspase inhibitors (Supplementary Fig. S4A). In contrast, the G₂/M cell-cycle arrest was not confirmed by Western blotting for the phosphorylation status of Cdc2 and Chk1 in these transfectants (Supplementary Fig. S4B). These results, consistent with the previous report of miR-218-inducing apoptosis in vitro (27), suggest that miR-218 and miR-585 induce a reduction in cell growth at least in part through caspase-mediated apoptosis, whereas the cause of the cell death except the apoptosis in these transfectants remains unclear.

While 5 genes of EGFR-coamplified and overexpressed protein (ECOP), IκBs kinase (IKK-β), LIM and SH3 protein 1 (LASP1), paxillin (PXN), and Robo1 have also been reported as direct targets of miR-218 (28–32), the percentage of OSCC cell lines with notable upregulation of these genes (>2-fold expression) was 50.0% (9/18), 38.9% (7/18), 44.4% (8/18), 16.7% (3/18), and 11.1% (2/18), respectively (Fig. 4A). In addition, we also confirmed that their protein levels were decreased in miR-218-transfectants compared with their control counterparts (Fig. 4B).

To explore novel oncogenic targets of miR-218 and miR-585 in OSCC cell lines, we used the algorithms, MicrocosmTargets, miRanda, and TargetScan, and selected rapamycin-insensitive companion of mTOR (Rictor) as miR-218-targets and jun B proto-oncogene (JunB) as miR-585-targets. The expression of Rictor or JunB was frequently upregulated in 38.9% (7/18) and 44.4% (8/18) of OSCC cell lines, respectively, compared with RT7 (>2-fold expression; Fig. 4A and Supplementary Fig. S5A). In Western blotting of Rictor and JunB, their protein levels were markedly reduced in miR-218- and miR-585-transfectants, respectively, compared with their control counterparts (Fig. 4B and Supplementary Fig. S5B). To further determine whether the predicted target sites of miR-218 and miR-585 in the 3′-UTR of mRNAs of Rictor and JunB, respectively (Fig. 4B and Supplementary Fig. S5C), were responsible for the translational regulation by dsRNA mimicking these miRNAs, we next conducted luciferase assays with vectors containing these 3′-UTR target sites downstream of the luciferase reporter gene. In this analysis, we observed a statistically significant reduction of luciferase activity in a vector containing the target site of Rictor, but not JunB (Fig. 4B and Supplementary Fig. S5D). These findings, together with the results of Western blotting, suggest Rictor and JunB to be a novel direct target of miR-218 and an indirect target of miR-585, respectively. Therefore, we focused on Rictor as the most likely target of miR-218, and conducted further analyses to explore the underlying molecular mechanisms of oral carcinogenesis.

Rictor, together with the mammalian target of rapamycin (mTOR) kinase, forms mTOR complex 2 (mTORC2), and the Rictor-mTOR complex directly regulates the phosphorylation of Akt at Ser-473, resulting in cell growth (33). We have previously reported that the phosphorylation of Akt at Ser-473 and Thr-308 was markedly increased in a HSC-2 cell line harboring a missense mutation in the PIK3CA gene, A3140G in exon 20, corresponding to the amino acid change H1047R (34). Interestingly, although NA cells without a genetic alteration of PIK3CA, EGFR, or PTEN also showed phosphorylated Akt, it has been unknown why the phosphorylation increased in this cell line. The phosphorylation status of Akt at Ser-473 in 18 OSCC cell lines and RT7 were shown in Figure 4A. To investigate whether Rictor might be associated with growth inhibitory effects of miR-218 in OSCC cells, we analyzed the phosphorylation of Akt and cell proliferation in HSC-2 and NA cells transfected with or without dsRNAs mimicking miR-218 or specific siRNAs for Rictor and PIK3CA (Fig. 4C). The treatment with PIK3CA- or Rictor-specific siRNA significantly inhibited cell growth in HSC-2 and NA cells whereas we found no noticeable differences in cell growth ratio of each transfectant in both cell lines (Supplementary Fig. S6). In Western blotting, the marked inhibition of the phosphorylation of Akt at Ser-473 in HSC-2 cells was observed in the treatment of RNA interference of PIK3CA, but not in restoration of miR-218 expression and knockdown of Rictor expression, suggesting that phosphatidylinositol 3-kinase (PI3K) seems mainly to regulate the phospho-Akt activation in HSC-2 cells with a mutant form of PIK3CA. On the other hand, a notable reduction in the phosphorylated Akt was detected in NA cells at 48 hours after the transient transfection of dsRNAs mimicking miR-218 or specific siRNA for Rictor compared with specific siRNA for PIK3CA, suggesting Akt to be significantly activated and phosphorylated by the TOR-Akt signaling pathway, not PI3K-Akt signaling pathway in NA cells without a mutant form of PIK3CA, EGFR, or PTEN. Our results clearly showed that miR-218 acts as a suppressor of the TOR-Akt signaling pathway, independently of the PI3K-Akt signaling pathway, in OSCC (Fig. 4D).

Here, we clearly identified miR-218 and miR-585 as TS-miRNAs silenced through tumor-specific DNA hypermethylation in oral cancer and characterized miR-218 targeting Rictor and inhibiting the phosphorylation of Akt at Ser-473 in a OSCC cell line without a mutant form of PIK3CA, EGFR, or PTEN. In this study, we carried out function-based screening using 327 dsRNAs mimicking mature miRNAs to identify TS-miRNAs having remarkable inhibitory effects on the growth of OSCC cell lines, although expression-based and DNA methylation-based screening had been successfully done previously in OSCC (8) and hepatocellular carcinoma (9), respectively. Since several known TS-miRNAs, such as miR-34 (24), miR-124 (9, 25), miR-193a (8), and miR-491 (15), were actually identified through this approach, function-based screening would be suitable for exploring dsRNAs, including miRNAs and siRNAs, as therapeutic agents for several types of cancer cells. The tumor-suppressive function of candidate miRNAs eventually identified in this screening was reevaluated using 2 kinds of dsRNA purchased independently to take account of off-target effects, known to complicate the interpretation of phenotypic effects in gene-silencing experiments using siRNAs (35).

We consider that the hypermethylation of CpG-islands on/around miRNA genes a good marker to explore novel epigenetically silenced TS-miRNAs, similar to classic TSGs in several types of cancer, and have already reported miR-137, miR-193a, miR-124, and miR-203 as epigenetically silenced TS-miRNAs (8, 9). In this study, a second screening, combining DNA methylation and expression analyses in a panel of OSCC cell lines, resulted in the identification of miR-218 and miR-585 as prime candidates for TS-miRNAs silenced by DNA hypermethylation in OSCC. Our study is the first to show that miR-218 and miR-585 were frequently silenced by DNA hypermethylation in OSCC. This miRNA gene is located at 4p15.31 (miR-218-1) and 5q35.1 (miR-218-2) in introns of 2 host genes, SLIT2 and SLIT3, respectively, and miR-585, together with miR-218-2, is also located within the intron of SLIT3 at 5q35.1. Copy number losses at these loci and the downregulated expressions of miR-218, SLIT2, and SLIT3 were reported in some types of cancer (36) although we found no homozygous loss in these regions in our previous studies of genomewide copy-number aberrations in 39 OSCC cell lines by array-CGH analyses (17, 37). Frequent epigenetic inactivation of SLIT2 and SLIT3 was also described in several cancers (38, 39). Although, in our study, the expression levels of these miRNAs and their host genes were remarkably restored in OSCC cell lines treated with 5-aza-dCyd, it was reported that the treatment of this inhibitor of DNA methylation did not reactivate the downregulated expression of SLITs in cervical cancer (40), suggesting a complex mechanism of inactivation in Slits in some types of cancers. Very recently, the downreguration of miR-218 and its inhibitory effects on cell proliferation and invasion have been shown in several types of cancer, including HPV-positive cell lines and tissues (26–29, 31, 32, 36, 41), and in the present study, we confirmed the tumor-suppressive functions and the mechanism of action of this miRNA in OSCC. Regarding miR-585, however, this is the first report that this is a novel TS-miRNA frequently silenced by tumor-specific DNA hypermethylation in OSCC, suggesting a crucial role, similar to miR-218, in human cancers.

In this study, we successfully identified a possible direct target of miR-218, Rictor, although ECOP, IKK-β, LASP1, PXN, and Robo1 have also been reported as direct targets of this miRNA. Since ECOP and IKK-β are components of the NF-κB pathway, our findings of caspase-mediated apoptosis in cells overexpressing miR-218 strongly support early reports of a correlation between miR-218 and targets (28, 29). LASP1, PXN, and Robo1 are known to be associated with cell adhesion, mobility, and migration, and their overexpression induced by miR-218 suppression seem to enhance tumor invasion and metastasis (30–32). Interestingly, although Slit-Robo signaling has been described to facilitate tumor cell migration, Robo1 targeted by miR-218 is known to be a receptor for Slit which is the host gene of miR-218 and shown to form a negative feedback loop involving Slit, miR-218, and Robo1 (27, 32). In our expression analysis, the percentage of OSCC cell lines with remarkable downregulation of miR-218, miR-585, SLIT2, and SLIT3 (< 0.5-fold), expression was 44.4% (8/18), 55.6% (10/18), 61.1% (11/18), and 77.8% (14/18), respectively, whereas that with notable upregulation of these genes (>2-fold expression) was 38.9% (7/18), 11.1% (2/18), 33.3% (6/18), and 0% (0/18), respectively. Since the biological and functional significances of Slits-miR-218-Robo signaling in cancer remains largely unknown, further study is needed for the characterization of this signaling.

Rictor is a component of the mTOR-containing complex mTORC2, which directly regulates the phosphorylation of Akt at Ser-473 (33). Knockdown of Rictor or mTOR in colon cancer was shown to inhibit cancer cell proliferation in vitro and tumor formation in vivo (42), and the selective requirement of Rictor for the development of prostate cancer induced by a loss of Pten in mice was shown (43). These results including our findings strongly support the notion that the silencing of miR-218 through the hypermethylation of CpG-islands around this miRNA is likely to be an important mechanism of carcinogenesis and cancer progression at least partly involving the activation of mTORC2-Akt signaling in OSCC. In this study, we analyzed the effects of miR-218 on the phosphorylation of Akt using 2 representative OSCC cell lines; HSC-2, with a missense mutation in the PIK3CA gene and activated PI3K-Akt signaling pathway, and NA, without genetic alterations of PIK3CA, EGFR, or PTEN (34). Although it has been unknown why the phosphorylation of Akt is increased in NA cells, our findings clearly showed miR-218 to act as a suppressor of the TOR-Akt signaling pathway, independently of the PI3K-Akt signaling pathway, in OSCC, and that the activation of this signaling pathway through methylation-mediated silencing of miR-218 may contribute to the pathogenesis of OSCC.

In conclusion, we described here the identification of 2 TS-miRNAs, miR-218 and miR-585, frequently silenced by DNA hypermethylation in OSCC, using function-based screening and a series of sequential analyses. Moreover, we identified Rictor as a potential target of miR-218, suggesting that the epigenetic silencing of miR-218 and consequent activation of the TOR-Akt signaling pathway induced by the overexpression of Rictor contribute to oral carcinogenesis, and that targeting miR-218 may provide a novel strategy for the treatment of OSCC, although further studies in vivo will be needed to confirm that dsRNA mimicking miR-218 can act as a TS-miRNA.

No potential conflicts of interest were disclosed.

We thank Ayako Takahashi and Rumi Mori for technical assistance.

This study was supported in part by Grant-in-Aid for Scientific Research (A), (B), and (C), and Scientific Research on Priority Areas and Innovative Areas, and a Global Center of Excellence (GCOE) Program for International Research Center for Molecular Science in Tooth and Bone Diseases from the Ministry of Education, Culture, Sports, Science, and Technology, Japan; a Health and Labour Sciences Research Grant by the Ministry of Health, Labour and Welfare, Japan; and a grant from the New Energy and Industrial Technology Development Organization (NEDO).

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.