circ-ANKS1B is overexpressed in OS tissues and cell lines.
Based on human reference genome GRCh37/hg19, circ_0007294 is located at chr12:100166699-100175875 and is assumed to derive from the gene ANKS1B. Therefore, we termed hsa_circ_0007294 as “circ-ANKS1B”. To detect the expression of circ-ANKS1B in OS, qRT-PCR was conducted in 30 OS tissues and paired adjacent non-tumor tissues. We observed that circ-ANKS1B was significantly overexpressed in OS tissues compared to matched nonneoplastic counterparts (Fig. 1A, 1B). Similarly, circ-ANKS1B level was also higher in OS cell lines, including MG63, U2OS and 143B cells than that in normal human osteoblast cell lines (HFOB and HOBC) (Fig. 1C). Subsequently, we verified circ-ANKS1B expression in OS TMA using the ISH assay. Based on the result of ISH staining, we divided circ-ANKS1B expression levels into five stratifications (Fig. 1D), then we found that circ-ANKS1B expression was overexpressed in OS tissues comparing with adjacent non-tumor tissues (Fig. 1E). Taken together, we proved that circ-ANKS1B expression is upregulated during OS tumorigenesis.
High expression of circ-ANKS1B predicts poor prognosis of OS patients
Then, we investigated the correlation between circ-ANKS1B expression and clinicopathological characteristics (Table 1). We found that circ-ANKS1B expression is significantly related to later TNM stage and large tumor size (Fig. 2A, 2B). Meanwhile, circ-ANKS1B expression was higher in metastatic tumors than in primary osteosarcoma tissues (Fig. 2C). Additionally, there was a remarkable tendency for increased circ-ANKS1B expression in recurrent OS tissues (Fig. 2D). Moreover, we found that OS patients who with overexpression of circ-ANKS1B always lead to poor overall survival and disease-free survival (Fig. 2E, 2F). Furthermore, circ-ANKS1B expression was found to be independent prognostic factors by multivariate analysis (Table 2). These findings indicated that circ-ANKS1B might contribute to OS progression and spired us to investigate the biological mechanism of circ-ANKS1B.
Knockdown of circ-ANKS1B suppresses OS proliferation and invasion in vitro
To analyze the potential function of circ-ANKS1B, we performed functional experiments. Because circ-ANKS1B level was the highest in U2OS and 143B cells, we knocked down the expression of circ-ANKS1B in these two cell lines using specific shRNAs targeting circ-ANKS1B (Fig. 3A, 3B). Notably, circ-ANKS1B silencing did not affect the expression of its linear mRNA ANKS1B (Fig. 3C). Subsequently, to evaluate the function of circRNA-000284 on cell proliferation, CCK-8 assay was conducted and the results showed that suppression of circ-ANKS1B could dramatically inhibit the cell proliferation in contrast to control cells (Fig. 3D). And we also observed that knockdown of circ-ANKS1B could suppress the colony formation and DNA synthesis rate compared with that in the negative control (Fig. 3E, 3F). Additionally, results of migration and invasion assays suggested that the invasive and migratory capacity was significantly inhibited by transfection of sh-circ-ANKS1B (Fig. 3G, 3H). In a word, these findings indicated that circ-ANKS1B acted as a important role during progression of OS cells.
Depletion of circ-ANKS1B suppresses OS growth in vivo
To further determine whether circ-ANKS1B could affect OS tumorigenesis in vivo, we conducted stable circ-ANKS1B-depleted U2OS cells and injected them subcutaneously into the flanks of nude mice (Fig. 4A). The circ-ANKS1B-depleted group (sh-circ-ANKS1B) mice developed weaker luciferase signal compared with the control group (NC) (Fig. 4B). At five weeks after inoculation, tumor volume and weight were noticeably decreased in the circ-ANKS1B silencing group compared with the NC group (Fig. 4C and 4D). Moreover, Ki-67 IHC staining intensive from the xenograft tumors also declined in the circ-ANKS1B-depleted group (Fig. 4E), suggesting that suppression of circ-ANKS1B could inhibit progression of OS in vivo.
circ-ANKS1B functions as a sponge for miR-149-5p
Given that it has been widely identified that circRNAs exerted biological function mainly through sponging to specific miRNAs, then by using Circinteractome tool (https://circinteractome.nia.nih.gov/), we identified that miR-149-5p, a known tumor suppressor in OS, contains complementary sequence to circ-ANKS1B (Fig. 5A). In addition, Pearson correlation analysis suggested that circ-ANKS1B was negatively correlated with miR-149-5p expression (Fig. 5B). Additionally, we observed that miR-149-5p was evidently low-expressed in OS tissues and cell lines (Fig. 5C, 5D). Moreover, as shown in Fig. 5E and 5F, overexpression circ-ANKS1B significantly inhibited miR-149-5p expression, while circ-ANKS1B-depleted dramatically increased miR-149-5p expression level. More importantly, luciferase reporter assay was performed, and results confirmed miR-149-5p could directly interact with circ-ANKS1B (Fig. 5G). Taken together, these findings suggest that circ-ANKS1B directly bind to miR-149-5p and inhibit its activity.
circ-ANKS1B affects proliferation and invasion abilities of OS cells via miR-149-5p
We further examined whether circ-ANKS1B functions as an oncogene by sponging with miR-149-5p, U2OS or 143B cells were transfected with miR-149-5p mimics or mock control, with/without circ-ANKS1B overexpression vector. Cell proliferation and colony formation assays indicated that overexpression of miR-149-5p could inhibit cell proliferation, while overexpression of circ-ANKS1B partially reversed the suppressive effect of miR-149p-5p in U2OS or 143B cells (Fig. 6A and 6B). Consistently, transwell assay showed that circ-ANKS1B reversed the inhibition effect of miR-149-5p on cell invasion ability (Fig. 6C). Based on these results, circ-ANKS1B affected the proliferation and invasion abilities of OS cells, at least partly by miR-149-5p.
circ-ANKS1B-mediated miR-149 negatively regulates Ki-67 in OS
Recently studies revealed that circRNA could exert biological function by involving in circRNA-miRNA-mRNA crosstalk. Based on this, we performed TargetScan to identify the potential gene targeting by miR-149-5p. We found the 3’UTR of Ki-67, a well-known tumor proliferation marker, bears one potential miR-149-5p binding sites (Fig. 7A). In addition, we found expression of Ki-67 was negative correlated with miR-149-5p in OS samples (Fig. 7B). Luciferase reporter assay was performed to prove the effect of miR-149-5p on Ki-67 mRNA level. And the results showed that miR-149-5p could obviously inhibit luciferase activity of Ki-67-3’UTR wild-type compared with negative control (Fig. 7C). Furthermore, we proved that miR-149-5p could significantly regulate the mRNA and protein levels of Ki-67 in OS cells (Fig. 7D and 7E). Co-transfected with circ-ANKS1B and miR-149 mimic could reverse the function of circ-ANKS1B on Ki-67 expression (Fig. 7E). Consistently, Ki-67 was positively correlated with expression of circ-ANKS1B in OS tissues (Fig. 7F).
Functional experiments showed that Ki-67 overexpression reversed the inhibition effect of circ-ANKS1B on cell proliferation (Fig. 8A, 8B), as well as invasion ability (Fig. 8C). These finding indicated that circ-ANKS1B sponges miR-149-5p, which subsequently allows for Ki-67 translation. circ-ANKS1B promotes the proliferation of OS partly via the circ-ANKS1B/miR-149/Ki-67 axis.