MiR-218 is frequently downregulated in gliomas
To investigate miR-218 function in glioma tumorigenesis, miR-218-1 and miR-218-2 expression was analyzed in gliomas and normal brain tissues (control subjects) using The Cancer Genome Atlas (TCGA) dataset. As shown in Fig. 1a, compared to control subjects, both miR-218-1 and miR-218-2 was significantly downregulated. Moreover, miR-218-2 expression levels was found significantly higher than miR-218-1 expression levels in gliomas (4.99 ± 1.95 vs. 0.25 ± 0.43, P <0.001), indicating that mature miR-218 in gliomas was mostly constituted by miR-218-2, which was consistent with a previous study in thyroid cancers . We further analyzed miR-218-1 and miR-218-2 expression in gliomas with different histologic grades. As shown in Fig. 1b (left panel), miR-218-1 expression were not significantly different between gliomas with histologic grade 2 (G2) and grade 3 (G3) (P =0.71). However, the gliomas with histologic G3 had a significant lower miR-218-2 expression than those with histologic G2 (P =0.002) (Fig. 1b, right panel).
Next, a large cohort of gliomas in TCGA dataset was analyzed by Kaplan-Meier method. As shown in Fig. 1c, the expression of miR-218-1 or miR-218-2 almost did not affect the survival of glioma patients when their survival time were less than 2000 days. However, downregulation of miR-218-2 but not miR-218-1 was significantly correlated with poor patient survival when their survival time more than 2000 days (Fig. 1d). The above findings suggest that miR-218-2 may be a potential biomarker to predict glioma patients’ long-term survival.
MiR-218 inhibits glioma cell growth
To explore biological function of miR-218 in glioma, we performed a series of in vitro experiments with miR-218 gain-of-function in glioma cells (Fig. 2a). Our data showed that miR-218 mimics significantly suppressed the proliferation of U251 and SHG44 cells compared to the control (Fig. 2b). Next, we assessed the effect of miR-218 mimics on cell growth using soft agar colony formation assay. The results showed that, compared with control cells, the colonies formed by miR-218 overexpressing cells were fewer (Fig. 2c). In vivo tumor-suppressing effect of miR-218 was also evaluated in nude mice. The results showed that, the U251 cells with miR-218 stably expressing induced tumors having significantly smaller mean tumor volume and longer latency relative to control (Fig. 2d). We isolated and weighed the xenograft tumors at the end of the experiments. As shown in Fig. 2e, compared with control tumors, weight of the tumors with miR-218 stably expressing was significantly less (P =0.0009). As expected, miR-218 stably expressing significantly decreased the percentage of Ki-67 positive cells in the tumors (Fig. 2f).
Then, the effect of miR-218 mimics on cell cycle contributions and apoptosis in U251 and SHG44 cells was tested. We found that, compared to control cells, the cell cycle of miR-218 overexpressing cells was arrested at the G0/G1 phase (Fig. 3a). The percentage of G0/G1 phase was respectively rose from 51.7 ± 2.4% to 62.3± 2.0% in U251 cells (P =0.004) and from 52.3± 2.7% to 66.6 ± 3.7 % in SHG44 cells (P =0.005). In addition, compared with the control, miR-218 mimics transfection showed elevation in both early and late apoptosis (20.5 ± 1.1% vs. 28.9 ± 1.8 % in U251 cells, P <0.002; 7.2 ± 1.3 % vs. 16.0 ± 2.1 % in SHG44 cells, P =0.003) (Fig. 3b). Collectively, our results further support miR-218 as a tumor suppressor in glioma cells.
MiR-218 inhibits glioma cell migration and invasion
The effect of miR-218 on migration and invasion potential was assessed in U251 and SHG44 cells. We found that the miR-218 overexpressing cells which migrated were significantly less than migrated control cells (Fig. 4). Moreover, miR-218 mimics were found to clearly downregulate the ability of cells passing through the matrigel-coated membrane (Fig. 4). These data indicate that miR-218 expression was closely associated with metastatic phenotypes of glioma cells.
TNC is identified as a new target of miR-218
A panel of candidate genes, which potentially targeted by miR-218, were identified by target predicting tools such as miRanda, TargetScan and miRDB. Among them, genes involved in vital signal pathways were selected to accept further detection, including IKBKB, TNC and WNT2B. As shown in Fig. 5a, b; Supplementary Fig. 1, among these three genes, only TNC was dramatically downregulated by miR-218 mimics in these two cell lines at both mRNA and protein levels. We further determined that miR-218 modulated TNC by direct interacting with it. Two TNC 3’ UTR (attached to luciferase coding region) luciferase reporter plasmid, which containing putative miR-218 binding sites: wild type (WT) 5’-AAGCACA-3’ or mutated (MUT) 5’-ACGAATA-3’, were constructed (Fig. 5c). We found that, in U251 and SHG44 cells transfected with WT luciferase reporter plasmid, luciferase activity was significantly suppressed by miR-218 mimics (Fig. 5d). While, in these cells transfected with mutated MUT luciferase reporter plasmid, the luciferase activity almost wasn’t affected (Fig. 5d). Our results demonstrated TNC as a direct target of miR-218.
Next, by analyzing TNC expression in gliomas and normal brain tissues using TCGA dataset, we confirmed TNC to be significantly elevated in gliomas (Fig. 5e), which was consistent with a previous study . In addition, the relationship between miR-218-1/miR-218-2 and TNC expression in gliomas was also investigated. As shown in Fig. 5f, TNC expression was not significantly correlated with miR-218-1 expression (P =0.08, r =0.08; Pearson’s correlation coefficient) (left panel), while was strongly correlated with miR-218-2 expression (P <0.0001, r =0.18; Pearson’s correlation coefficient) (right panel).
MiR-218 functions as a tumor suppressor in gliomas cells by inhibiting TNC/AKT/AP-1/TGFβ1 positive feedback loop
Then, the mechanism of malignant phenotypes of glioma cells inhibited by miR-218 was explored. There is evidence revealing that TNC can increase phosphorylation of AKT at Ser 473 by interacting with intergrins, thereby activating the PI3K/AKT pathway [23-25]. Thus, we speculated that miR-218 inhibited the PI3K/AKT signaling via targeting TNC. It was demonstrated that miR-218 mimics expectedly downregulated TNC expression, and clearly inhibited phosphorylation of AKT at Ser 473, while almost did not affect phosphorylation of AKT at Thr 308 in U251 and SHG44 cells (Fig. 6a).
Evidently, targeted by PI3K/AKT signaling pathway, transcription factor AP-1 is constitutively activated in glioma and important in cell proliferation [26-29]. AP-1, which can bind to a common DNA binding sequence, is heterodimers composed primarily by the FOS and JUN families. AP-1 activation contains complex process, such as increased expression or phosphorylation of FOS and JUN . As shown in Fig. 6a, we found that miR-218 mimics strongly inhibited JNK phosphorylation, while almost did not affect FOS and JUN expression in U251 and SHG44 cells. Considering that TGFβ1 is a well-known target of AP-1 [31-33], thus we speculated that miR-218 could downregulate TGFβ1 expression by suppressing AP-1 activity. As shown in Fig. 6a, compared to the control, miR-218 mimics expectedly decreased TGFβ1 expression in U251 and SHG44 cells. These results suggest that transcriptional activity of AP-1 can be inhibited by miR-218, as supported by the AP-1 luciferase reporter assay (Fig. 6b).
Next, to confirm the above observations in vivo, we performed western blot analysis to detect the indicated gene expression in the xenograft tumors. Our data demonstrated that, compared with control tumors, TNC expression was significantly decreased in miR-218-overexpression tumors (Fig. 6c). As expected, phosphorylation of Akt at Ser 473 and JNK and TGFβ1 expression in miR-218-overexpression tumors were dramatically lower than that of control tumors, but phosphorylation of Akt at Thr 308 and the expression of FOS and JUN were not different between two kinds of tumors, further supporting the in vitro results.
It should be noted that TGFβ1 has been reported to, in turn, induce TNC expression involving Smad3/4, Sp1, Ets1 and CBP300 . Thus, we suppose that TGFβ1 is able to activate AKT/AP-1 signaling axis by increasing TNC expression, thereby forming a positive feedback loop. To prove this, we treated U251 and SHG44 cells with recombinant human TGFβ1 proteins. The results showed that TGFβ1 treatment markedly induced TNC expression and subsequently increased phosphorylation of Akt at Ser 473 and JNK expression, while this effect could be reversed by miR-218 mimics (Fig. 6d). This was also supported by the AP-1 luciferase reporter assay (Fig. 6e). Altogether, those results indicate that miR-218 exerts tumor suppressor roles in glioma cells by blocking the TNC/AKT/AP-1/TGFβ1 positive feedback loop.
Give the above, a model is proposed to explore the mechanism of miR-218 inhibiting malignant progression of glioma (Fig. 6f). Briefly, miR-218 represses TNC expression by binding to its 3’ UTR. This will reduce AKT phosphorylation and subsequently suppress transcriptional activity of AP-1 by decreasing JNK phosphorylation, thereby downregulating the expression of TGFβ1, which be able to, in turn, activate the TNC/AKT/AP-1 signaling axis. Thus, miR-218 acts as a potent tumor suppressor in glioma by blocking the TNC/AKT/AP-1/TGFβ1 positive feedback loop.