MiR-193a-5p is upregulated in osteosarcoma
The expression of miR-193a-5p was investigated in 25 pairs of primary osteosarcoma and corresponding adjacent noncancerous tissues by qRT-PCR. The results showed that miR-193a-5p was expressed at significantly higher levels in the tumor tissues than in the corresponding adjacent noncancerous tissues (Figure 1A). Moreover, the level of miR-193a-5p in osteosarcoma cell lines (MG63, SJSA-1, SaOS-2 and U-2OS) was significantly higher than that in the normal human osteoblast hFOB1.19 cell line (Figure 1B).
MiR-193a-5p promotes colony formation, migration, invasion, and EMT of osteosarcoma cells
To explore the role of miR-193a-5p in osteosarcoma, we used a lentivirus-mediated expression system to interfere with miR-193a-5p expression in SaOS-2 cells, which have high levels of endogenous miR-193a-5p, and to overexpress or interfere with miR-193a-5p in U-2OS cells, which have moderate levels of endogenous miR-193a-5p (Figure 1C). Compared with the cells infected with negative control lentivirus, the cells infected with anti-miR-193a-5p lentivirus formed much fewer colonies, while the cells infected with pre-miR-193a-5p had much more colonies (Figure 1D). Wound healing and transwell invasion assays showed that overexpressing miR-193a-5p promoted, but silencing miR-193a-5p decreased, cell migration and invasion (Figure 1E and 1F).
EMT plays a key role in malignant progression[30]. Thus, we investigated the effects of miR-193a-5p on several EMT markers. Overexpression of miR-193a-5p reduced the expression of the epithelial marker CDH1, but augmented the expression of the mesenchymal cell markers MMP9, Slug and Snail. Conversely, knockdown of miR-193a-5p increased CDH1 expression, but decreased the expression of MMP9, Slug and Snail (Figure 1G). These results indicate that miR-193a-5p promote EMT process.
MiR-193a-5p promotes osteosarcoma metastasis in a xenograft mouse model
To investigate the effects of miR-193a-5p on metastasis, U-2OS cells stably overexpressing pre-miR-193a-5p, ant-miR-193a-5p or nontarget sequence were implanted into nude mice by tail-vein injection. At 37 days after injection, all mice were euthanized, the metastatic nodules on the surface of the liver and lung were counted, and metastatic lesions were detected by HE staining. Compared with those in mice injected with control cells, the liver metastatic nodules were more and larger in the mice injected with U-2OS-193UP, but fewer and smaller in the mice injected with U-2OS-193DN (Figure 2A, 2C and 2D). There were no visible metastatic nodules on the surface of the lung, but HE staining showed that there were more and larger metastatic nodules in the lung of the pre-miR-193a-group than in the anti-miR-193a-5p group or control group (Figure 2B and 2E). IHC staining showed that miR-193a-5p overexpression decreased CDH1 expression but increased the expression of MMP9, Snail and Slug, whereas miR-193a-5p knockdown had completely opposite effects in both the liver and lung (Figure 2F and 2G).
Additionally, we constructed a xenograft mouse model using SaOS-2 cells stably overexpressing pre-miR-193a-5p, ant-miR-193a-5p or nontarget sequence and obtained results (Supplementary Figure S1) consistent with those observed in the U-2OS-derived xenograft mouse model.
Quantitative proteomics analysis identifies NCX2 as a potential target of miR-193a-5p
To identify the potential target genes of miR-193a-5p, we used TMT-based quantitative proteomics to analyse the differential expression of proteins following overexpression or knockdown of miR-193a-5p in U-2OS. By analysing the results of three biological triplicates, we found that 90 proteins were downregulated (fold change ˂0.833, p˂0.05) and 140 proteins were upregulated (fold change > 1.20, p˂0.05) after ectopic expression of pre-miR-193a (Supplementary Table S2), while 303 proteins were downregulated and 375 proteins were upregulated following knockdown of miR-193a-5p (Supplementary Table S3). Among the differentially regulated proteins, only two proteins (NCX2 and HLA-B) were both downregulated by pre-miR-193a overexpression and upregulated by knockdown of miR-193a-5p, but no protein was both upregulated by pre-miR-193a overexpression and downregulated by knockdown of miR-193a-5p. Western blotting further confirmed that overexpression of miR-193a-5p in U-2OS decreased, while knockdown of miR-193a-5p in U-2OS or SaOS-2 increased, the levels of NCX2 and HLA-B proteins (Figure 3A). These results suggest that NCX2 and HLA-B are potential targets of miR-193a-5p.
MiR-193a-5p directly targets NCX2 in osteosarcoma
We then search the potential binding site (miRNA response element, MRE) of miR-193a-5p in the NCX2 and HLA-B genes by miRWalker, a comprehensive database for the prediction of MREs [31]. An MRE for miR-193a-5p was identified in the 3′- UTR of NCX2 (Figure 3B), but not in the 3′-UTR, 5′-UTR or coding sequence of the HLA-B gene. To confirm whether the predicted MRE in NCX2 could be recognized by miR-193a-5p, the wildtype or mutant MRE was cloned downstream of the firefly luciferase gene in the pmirGLO vector and cotransfected with miR-193a-5p mimics. Luciferase activity assays showed that miR-193a-5 significantly suppressed the luciferase activity of pmirGLO-MRE, but not that of pmirGLO-mtMRE (Figure 3C). These results suggest that NCX2 is a direct target of miR-193a-5p.
We further detected NCX2 expression in osteosarcoma tissues by qRT-PCR and Western blotting and found that both NCX2 mRNA and protein levels were significantly lower in osteosarcoma tissues than in the corresponding normal tissues (Figure 3D and 3E). Correlation analysis showed that NCX2 mRNA was negatively corelated with the levels of miR-193a-5p and miR-193a-3p (Figure 3F and 3G). Compared with miR-193a-5p, miR-193a-3p had a weaker association with NCX2, possibly because NCX2 is mainly regulated by pomiR-193a-5p. The above results indicated that miR-193a-5p recognized NCX2 mRNA and negatively regulated the expression of NCX2.
MiR-193a-5p promotes colony formation, migration, invasion and EMT by targeting NCX2
To assess whether miR-193a-5p exerts oncogenic effects by targeting NCX2, we upregulated (or downregulated) NCX2 expression by adenovirus-mediated transfer of its cDNA (or shRNA) into cells and performed in vitro assays. As shown in Figure 4a-e, NCX2 overexpression reduced, while miR-193-5p overexpression increased, the colony formation, migration and invasion of U-2OS cells. Moreover, the increase in colony formation, migration and invasion induced by miR-193a-5p overexpression could be abrogated by NCX2 overexpression. Conversely, NCX2 knockdown increased, while miR-193-5p knockdown decreased, cell migration and invasion, and the effects of miR-193a-5p knockdown were counteracted by NCX2 knockdown (Figure 4). Additionally, the effects of NCX2 and miR-193a-5p on EMT-related proteins were investigated. Consistent with the above results (Figure 1G), miR-193a-5p overexpression promoted the EMT process, while NCX2 overexpression reduced the effects of miR-193a-5p overexpression on the expressions of EMT-related marker (Figure 4f). Moreover, the effects of miR-193a-5p knockdown were rescued by knockdown of NCX2 (Figure 4f). We repeated the above experiments in SaOS-2 cells and obtained similar results (Supplementary Figure S2). These results suggest miR-193a-5p plays oncogenic roles by targeting NCX2.
NCX2 suppresses the EMT process by inhibiting Ca2+-dependent Akt phosphorylation
NCXs normally extrude Ca2+ from the cell [16, 17]. The Ca2+ indicator Fluo-8 was used to measure the influence of NCX2 on intracellular free Ca2+ levels. The results showed that overexpression of NCX2 in SaOS-2 cells reduced, while knockdown of NCX2 in U2-OS cells increased, the level of intracellular free Ca2+ (Figure 5A and 5B) indicating that NCX2 promoted Ca2+ efflux. It has been reported that AKT is activated by Ca2+/calmodulin (CaM)-dependent protein kinase kinase 2 (CaMKK2) in a Ca2+/CaM-dependent manner[32]. Therefore, we detected the effects of NCX2 on AKT phosphorylation. As expected, NCX2 suppressed the phosphorylation of AKT1 (Figure 5C). Further experiments demonstrated that the suppression of AKT phosphorylation by NCX2 could be rescued by ionomycin-induced Ca2+ influx (Figure 5D). These results indicated that NCX2 suppressed AKT phosphorylation by reducing intracellular free Ca2+ levels.
Moreover, we investigated whether NCX2 plays a role in the EMT process by affecting AKT phosphorylation. As shown in Figure 5e, the effects of NCX2 overexpression on the expression of EMT-related markers were suppressed by the AKT activator SC-79, but enhanced by the AKT inhibitor afuresertib (Figure 5E). The above results suggest that NCX2 suppresses the EMT process by inactivating the Akt pathway.
Knockdown of miR-193a-5p sensitizes osteosarcomato afuresertib
Afuresertib is an oral AKT inhibitor with demonstrated cytotoxic and antiproliferative activities against various kinds of cancer cells. The above results indicated that miR-193a-5p activates AKT by targeting NCX2. Therefore, we explored whether downregulation of miR-193a-5p enhances the effects of afuresertib treatment. In vitro assays showed that knockdown of miR-193a-5p decreased, while overexpression of miR-193a-5p increased, the IC50 value of afuresertib in SaOS-2 and U-2OS cells (Figure 6A, 6B and 6C). Then, a subcutaneous tumor model was generated for in vivo evaluation of the efficacy of antagomir-193a-5p (a chemically-modified single-stranded miR-193a-5p inhibitor) and afuresertib. The results showed that treatment with antagomir-193a-5p or afuresertib alone could suppress the growth of tumor and metastasis to both liver and lung in osteosarcoma xenograft mice, while the combination treatment had stronger anti-tumor effect (Figure 6D, 6E, 6F, 6G, 6H, 6I and 6J).