Identification of Circ_0030998
By analyzing the microarray data GSE138589, which compared six pairs of CRC tissues and matched neighboring normal tissues. A total of 155 upregulated circRNAs and 29 downregulated circRNAs with p value < 0.05 and |fold change| > 1 were identified (Supplementary Table 2). The differently expressed circRNAs were exhibited by volcano plots as shown in Figure 1A. Among these 184 circRNAs, the top 30 ones (21 upregulated and 9 downregulated) with the most significant differences were selected for further study and shown by hierarchical clustering analysis in Figure 1B. To screen out the circRNA that may be related to the progression of CRC cells, the expression of 21 upregulated circRNAs were tested in CRC tissues, and it was shown that Circ_0030998 had the highest level than other 20 circRNAs (data not shown). Therefore, we chose Circ_0030998 for further analysis.
According to the circBase (http://www.circbase.org/) and UCSC Genome Browser Home (http://genome.ucsc.edu/), we found that Circ_0030998 was 220 base pairs (bp) in length located at chr13:113963957-113964177 and it was derived from the exon 3 of host gene LAMP1, which acted as an oncogene in the progression of several cancers19, 20.
Circ_0030998 was significantly upregulated in CRC tissues and associated with poor prognosis of CRC patients
Firstly, qRT-PCR was performed to identify the expression levels of Circ_0030998 in 90 pairs of CRC tissues and adjacent normal tissues. As shown in Figure 1C, Circ_0030998 was significantly upregulated in CRC tissues compared to the adjacent normal tissues. Moreover, we divided 90 CRC patients into two groups according to the median level (cutoff value = 2.961) of the relative Circ_0030998 expression in tumor tissues: the high group (n = 45) and the low group (n = 45). Then Pearson chi-square tests were used to analyze the relationship between Circ_0030998 expression level and patients’ clinicopathologic features. It was demonstrated that high expression of Circ_0030998 was related with lymph node metastasis and TNM stage but not gender, age, tumor location, CEA level and tumor size of CRC patients (Table 1).
Furthermore, Kaplan-Meier survival analysis was conducted to analyze the relationship between Circ_0030998 expression level and patients’ survival. It showed that patients with high Circ_0030998 levels had a shorter survival compared to those with low levels (Figure 1D). Univariate survival analysis showed that lymph node metastasis, TNM stage and Circ_0030998 expression were prognostic factors. Multivariate Cox regression analysis demonstrated that only TNM stage and Circ_0030998 expression were independent prognostic factors for CRC patients (Table 2).
Circ_0030998 was upregulated in CRC cell lines and promoted tumor proliferation and angiogenesis in vitro
The relationship of Circ_0030998 expression and prognosis of CRC patients suggested that Circ_0030998 was correlated with tumor malignancy. Thus, we investigated the functions of Circ_0030998 in CRC cell lines. Fist, we examined the expression of Circ_0030998 in six CRC cell lines (HT29, SW620, DLD-1, HCT116, SW480, LoVo) and a human non-tumorigenic colorectal epithelial cell line (NCM460) by qRT-PCR. The results showed that Circ_0030998 was significantly upregulated in CRC cell lines than NCM460 (Figure 2A). Then, the SW480 cell line which had the relatively highest expression of Circ_0030998 and HCT116 cell line which had the relatively lowest expression of Circ_0030998 were chosen for further study. Two siRNAs targeting Circ_0030998 were transfected into SW480, and si-Circ_0030998-1 which was chosen for subsequent experiments exhibited better knockdown efficiency (Figure 2B). Meanwhile, Circ_0030998 overexpression plasmid significantly upregulated the expression of Circ_0030998 in HCT116 cells (Figure 2C).
Then, RNA fluorescence in situ hybridization assays with specific Cy3-labeled probe for Circ_0030998 were performed to identify the subcellular localization of Circ_0030998, and the results demonstrated that Circ_0030998 was mainly localized in the cytoplasm of SW480 and HCT116 cells (Figure 2D). Next, CCK-8 and colony formation assays showed that downregulation of Circ_0030998 significantly inhibited the proliferation and the cloning ability of SW480 cells; whereas, overexpression of Circ_0030998 promoted these functions in HCT116 cells (Figure 3A and 3B). Moreover, flow cytometry analyses showed that Circ_0030998 downregulation led to a significant G1/G0 phase arrest in SW480 cells and vice versa in HCT116 cells when Circ_0030998 was overexpressed (Figure 3C). Furthermore, HUVECs were used to examine the effect of Circ_0030998 on angiogenesis, and the results showed that the formation of tube-like structures were significantly inhibited when Circ_0030998 was downregulated and vice versa when Circ_0030998 was overexpressed (Figure 3D). These results suggested that Circ_0030998 acted as an oncogene that promoted the CRC cells proliferation and angiogenesis.
Downregulation of Circ_0030998 inhibited CRC growth in vivo
To further confirm the roles of Circ_0030998 on tumorigenesis in vivo, shRNA targeting Circ_0030998 was constructed, and the significant knockdown efficiency of sh-Circ_0030998 was verified by qRT-PCR as shown in Figure 4A. Then, SW480 cells transfected with either sh-NC or sh-Circ_0030998 were injected subcutaneously in the left flank of nude mice. As shown in Figure 4B and 4C, the tumor volumes in the sh-Circ_0030998 group were obviously smaller than those in the sh-NC group. Meanwhile, the Circ_0030998 expression in tissues from sh-Circ_0030998 group was significantly lower than that from sh-NC group (Figure 4D). These data suggested that knockdown of Circ_0030998 could inhibit CRC cells proliferation in vivo.
Circ_0030998 facilitated CRC cells proliferation and angiogenesis by spongingmiR-567
Many studies have revealed that circRNAs functioned as “miRNA sponges”, namely, competing endogenous RNAs (ceRNAs), to block the formation of Ago2-mediated silencing complex. Because Circ_0030998 was mainly localized in the cytoplasm of CRC cells, we supposed that Circ_0030998 may also function as a ceRNA. Then two bioinformatics databases, circBank (http://www.circbank.cn/index.html) and Circular RNA Interactome (https://circinteractome.nia.nih.gov/), were used to screen for miRNAs that could bind with Circ_0030998. As shown in the Supplementary Table 3, only miR-567 and miR-556-5p were both predicted by two databases. QRT-PCR showed that only miR-567 was upregulated in SW480 cells when Circ_0030998 was knockdown (data not shown); furthermore, we examined miR-567 expression in 90 CRC tissues, and a significant inverse correlation was found between miR-567 and Circ_0030998 as shown in Figure 5A; so we hypothesized that miR-567 was the miRNA sponged by Circ_0030998.
To further validate whether Circ_0030998 could interact with miR-567 directly in CRC cells, the wildtype and mutated putative binding sites of Circ_0030998 were cloned and inserted into luciferase reporter vectors respectively for luciferase reporter assays (Figure 5B). The results showed that the luciferase activity of the wildtype but not mutant Circ_0030998 was significantly inhibited by miR-567 mimics in both SW480 and HCT116 cells (Figure 5C). Furthermore, the RIP assays demonstrated that miR-567 significantly increased the enrichment of Circ_0030998 by Ago2 RIP in both SW480 and HCT116 cells compared to miR-NC (Figure 5D). These results suggested that Circ_0030998 functioned as a ceRNA for miR-567 in CRC cells.
Next, we examined the effect of miR-567 on CRC cells proliferation and angiogenesis. CCK-8 and colony formation assays showed that downregulation of miR-567 by miR-567 inhibitor increased HCT116 cells proliferation (Figure 5E and 5F). Flow cytometry analyses demonstrated that miR-567 inhibitor promoted HCT116 cells cycle progression (Figure 5G). Moreover, inhibition of miR-567 increased the tube-like structures formation of HUVECs (Figure 5H).
Furthermore, we performed rescue assays to confirm whether Circ_0030998 regulated CRC cells proliferation and angiogenesis via miR-567. CCK-8 and colony formation assays showed that miR-567 mimic could partially weaken the promotive effect of Circ_0030998 on HCT116 cells proliferation (Figure 6A and 6B). MiR-567 mimic also reversed the promotive effect of Circ_0030998 on HCT116 cells cycle progression and tube-like structures formation of HUVECs (Figure 6C and 6D). Taken together, these results suggested that Circ_0030998 promoted CRC cells proliferation and angiogenesis via miR-567.
MiR-567 inhibited CRC cells proliferation and angiogenesis via VEGFA
MiRNAs could regulate target mRNAs by binding with their 3’UTRs via complementary base pairing. Previous studies showed that miR-567 could regulate KPNA421, ATG522 and then inhibited tumor progression or chemoresistance. To explore the mechanism by which miR-567 inhibited CRC cells proliferation and angiogenesis, three miRNA databases were used to predict the potential target genes in the present study. As shown in Figure 7A and Supplementary Table 4, there were 10 potential target genes in the overlapped fraction of three databases, namely BNC2, BVES, CSRNP3, KANSL1L, MB21D2, MBNL2, NEUROD2, UBR3, VEGFA, ZSWIM6. Among these 10 candidate genes, only VEGFA could promote cancer proliferation and angiogenesis. So we hypothesized that VEGFA may be the downstream target gene of miR-567 in regulating CRC cells proliferation and angiogenesis.
Then, we examined miR-567 and VEGFA expression in 90 CRC tissues, and it showed that VEGFA expression was negatively correlated with miR-567 in CRC tissues (Figure 7B). To further confirm the interaction between miR-567 and VEGFA, we performed luciferase reporter assays with luciferase reporter vectors containing the wildtype or mutated putative binding sites of VEGFA (Figure 7C). The results showed that the luciferase activity of the wildtype VEGFA was significantly inhibited by miR-567 mimics in both SW480 and HCT116 cells compared with mutant VEGFA (Figure 7D).
Moreover, we designed pcDNA-VEGFA for ectopic expression to verify the effect of VEGFA on CRC cells proliferation and angiogenesis. The efficiency of pcDNA-VEGFA was examined by qRT-PCR as shown in Figure 7E. CCK-8 and colony formation assays showed that overexpression of VEGFA promoted HCT116 cells proliferation (Figure 7F and 7G). Flow cytometry analyses demonstrated that VEGFA promoted HCT116 cells cycle progression (Figure 7H). Meanwhile, VEGFA promoted the tube-like structures formation of HUVECs (Figure 7I). The effect of VEGFA on HCT116 cells and HUVECs coincided with that of miR-567 inhibitor. These findings suggested that miR-567 inhibited CRC cells proliferation and angiogenesis via VEGFA.
Circ_0030998 promoted CRC cell proliferation and angiogenesis via themiR-567/VEGFA axis
Furthermore, western blotting showed that VEGFA was decreased in SW480 transfected with si-Circ_0030998, and vice versa in HCT116 cells when Circ_0030998 was overexpressed (Figure 8A). We also designed si-VEGFA and conducted rescue assays to confirm whether Circ_0030998 functioned in CRC via VEGFA. CCK-8 and colony formation assays showed that the effect of Circ_0030998 on HCT116 cells proliferation was partially reversed by si-VEGFA (Figure 8B and 8 C). Si-VEGFA also weakened the promotive effect of Circ_0030998 on HCT116 cells cycle progression and tube-like structures formation of HUVECs (Figure 8D and 8 E). All these data suggested that Circ_0030998 promoted CRC cells proliferation and angiogenesis via the miR-567/VEGFA axis.