Differentially expressed circRNAs and circRNA/miRNA/mRNA regulatory network
Additionally, we have also identified the differentially expressed circRNAs between OS and corresponding adjacent nontumor tissue samples. Compared with nontumor samples, there were totally 1088 significantly differentially expressed circRNAs in OS samples, including 1052 upregulated circRNAs and 36 downregulated circRNAs (Figure 1a). The expression levels of differentially expressed circRNAs were significantly different (Figure 1b).
Furthermore, we have estimated the circRNA/miRNA interactions using miRanda and TargetScan. After the cross analysis of predicted results of two software, there were 36 differentially expressed circRNAs and their corresponding miRNAs in regulatory network (Supplementary Figure 2a). Among them, circ_0001722 with small expression standard deviations in three OS samples and three nontumor samples (standard deviation values were 3.1 and 0.7, respectively) was selected for subsequent analysis. Regarding circ_0001722, there were 11 overlapped regulatory pairs between miRanda and TargetScan (Supplementary Figure 2b), including miR-365b-5p, miR-3122, miR-211-5p, miR-208b-5p, miR-3913-5p, miR-204-5p, miR-365a-5p, miR-5006-3p, miR-208a-5p, miR-3137 and miR-6875-3p. Then, the target genes of these 11 miRNAs were also estimated utilizing miRTarBase database. We found that miR-204-5p and its corresponding 76 target genes, comprising RUNX2, were supported by most experimental validation (469 experimental results). The circRNA/miRNA/mRNA regulatory network, based on circ_0001722, 11 miRNAs, and target mRNAs, was shown in Figure 1c.
Expression of circ_0001722 is significantly upregulated inOS tissues and cell lines
Firstly, we determined whether circ_0001722 is a closed circular RNA that is resistant to RNase R digestion. We investigated its expression level in OS tissues and cell lines. Our result showed that RNase R digestion could decrease the RNA level of linear GAPDH, but could not affect the level of circ_0001722 significantly (Figure 1d), indicating that circ_0001722 was resistant to RNase R digestion. To confirm circ_0001722 is a closed circRNA, we ran PCR amplification of circ_0001722 with specific divergent primers. PCR products amplified by divergent primers were detected by Sanger sequencing to confirm the circular structure of circ_0001722. Our data showed that the back splicing junction sites of circ_0001722 were consistent with the sequence from circBase (Figure 1e). Then, we measured the expression level of circ_0001722 in a series of 20 surgically removed fresh OS tissues and their corresponding adjacent nontumor tissues by qRT-PCR. Comparing with nontumor tissues, the expression level of circ_0001722 in OS tissues was significantly upregulated (P <0.01; Figure 1f). Similarly, the expression of circ_0001722 in OS cell lines (MG63 and U2OS) was significantly higher than in normal human fetal osteoblastic cell line (hFOB 1.19) (P <0.01; Figure 1g).
Downregulation ofcirc_0001722 suppresses cell proliferation of human OS cells in vitro and in vivo.
We noticed that the expression of circ_0001722 was higher in MG63 cells and U2OS cells. Based on the observation, exogenous shRNA (sh-circ_0001722) was used to knock down circ_0001722 expression in two OS cell lines (Figure 2a). Based on the high silencing efficiency, we carried out cell proliferation assay and matrigel invasion assay. The results showed that the viability level of MG63 cells and U2OS cells transfected with sh-circ_0001722 was lower than that of the cells transfected with sh-Mock (Figure 2b). Consistently, sh-circ_0001722 significantly decreased the number of OS cells invaded through the matrigel. Quantitative analysis of cell numbers revealed that, in the negative control (sh-Mock) and blank control groups, the number of cells invaded through the matrigel was almost 5 times higher than sh-circ_0001722 group (figure 2c). The above experiments verified to some extent that silencing circ_0001722 can inhibit the proliferation and invasion of OS cells in vitro . To investigate whether circ_0001722 regulates the tumorigenesis of OS in vivo , we established OS xenograft mouse models. MG63 cells transfected with sh-circ_0001722 or sh-Mock were inoculated subcutaneously in the right flank of athymic nude mice (5 × 106 cells per mouse, 5 mice per group). After 4 weeks, all experimental mice were euthanized and tumor tissues were collected (Figure 2d). Comparing with sh-Mock group, smaller tumor volume and lower tumor weight were observed in sh-circ_0001722 group. The same conclusion was reached with thein vitro experimental results. Downregulation of circ_0001722 can suppress tumorigenesis of OS cells in vivo . In addition, compare with the control group, knocking down of circ_0001722 led to the decrease of Ki67 expression, which implied the cell proliferation of sh-circ_0001722 group is slower than sh-Mock group. Taken altogether, downregulation of circ_0001722 can restrain the growth of OS cellsin vitro and in vivo .