1. hsa_circ_0001829 is upregulated in GC cell lines and tissues, and is a stable circular RNA.
To investigate the potential role of circRNAs in MNNG-induced malignant transformation of GC cells, we analyzed the expression profile of circRNAs in GES-1-T and GES-1-N cells using ribosomal RNA depleted and RNase R treated RNA-seq, and found a total of 4,650 differentially expressed circRNAs with a criterion of fold change (FC) more than 2.0 or less than 0.5 and P < 0.05, of which 1,509 circRNAs were upregulated and 3,141 circRNAs were downregulated in GES-1-T cells (Fig. 1A). The heat map expression of the top 10 upregulated circRNAs, the top 10 downregulated circRNAs and 40 randomly selected differentially expressed circRNAs were shown in Fig. 1B. For further analysis, we focused on the upregulated circRNAs and validated the induced expression of the top 10 upregulated circRNAs in GES-1-T cells by qRT-PCR assay. Consistent with the results of RNA-seq, the expressions of the 10 circRNAs were significantly increased in GES-1-T cells, with hsa_circ_0001829 showing the greatest increase (Fig. 1C). Furthermore, hsa_circ_0001829 exhibited upregulated expression in four gastric cancer cell lines (MKN-28, MGC-803, BGC-823 and SGC-7901) (Fig. 1D). Moreover, qRT-PCR analysis with 20 pairs of matched GC and adjacent tissue samples revealed that hsa_circ_0001829 expression was upregulated in GC tissues (Fig. 1E).
With a backspliced length of 641 bp, hsa_circ_0001829 is derived from exon 1 within solute carrier family 45 member 4 (SLC45A4) gene locus (Fig. 1F) and is located at chr8:142264087–142264728. Sanger sequencing of RT-PCR product confirmed the head-to-tail splicing of hsa_circ_0001829 (Fig. 1F). Then, the stability of hsa_circ_0001829 was evaluated. Compared with linear SLC45A4, hsa_circ_0001829 was resistant to digestion induced by RNase R exonuclease (Fig. 1G), indicating that hsa_circ_0001829 harbors a loop structure. Following treatment with Actinomycin D (an inhibitor of transcription), the half-life of hsa_circ_0001829 exceeded 24 h, while that of linear SLC45A4 displayed only about 4.3 h, revealing that hsa_circ_0001829 is highly stable (Fig. 1H).
These results demonstrated that hsa_circ_0001829 was upregulated in GC cell lines and tissues, and was a stable circular RNA.
2. Silencing of hsa_circ_0001829 inhibits GC cells proliferation, migration and invasion in vitro.
To investigate the biological function of hsa_circ_0001829, three small interfering RNAs (siRNAs) targeting the junction sites of hsa_circ_0001829 were designed to silence hsa_circ_0001829 expressions. These siRNAs significantly decreased hsa_circ_0002819 expression level with no effect on SLC45A4 linear isoform in both GES-1-T and MKN-28 cell lines (Fig. 2A). And we selected si-circRNA-1# and si-circRNA-2# for the subsequent experiments due to the relative higher interference efficiency. Following transfection of GES-1-T and MKN-28 cells with si-circRNA-1# and si-circRNA-2#, CCK-8 and EdU assays were employed to measure the cell proliferation. As shown, silencing of hsa_circ_0001829 markedly inhibited cell proliferation in both cell lines (Fig. 2B, C). Furthermore, colony-forming ability of these two cell lines stably transfected with sh-circRNA-1# and sh-circRNA-2# by lentiviral infection were significantly reduced after silencing of hsa_circ_0001829 (Fig. 2D). Additionally, knockdown of hsa_circ_0001829 impaired migration and invasion ability of GES-1-T and MKN-28 cells, as demonstrated by transwell and wound healing assays (Fig. 2E, F).
Then assays were conducted to evaluate whether hsa_circ_0001829 silencing had effect on the cell cycle and apoptosis. As shown in Fig. 4A, more cells were distributed in G2/M phase upon silencing hsa_circ_0001829, suggesting that knockdown of hsa_circ_0001829 induced cell cycle arrest at G2/M. Moreover, the apoptosis rates of GES-1-T and MKN-28 cells upon hsa_circ_0001829 knockdown were obviously increased (Fig. 4B). These results demonstrated that hsa_circ_0001829 silencing induced G2/M cell cycle arrest and increased apoptosis rates.
Collectively, these results indicated that silencing of hsa_circ_0001829 inhibits GC cells proliferation, migration and invasion, and the decreased cell cycle progression and increased apoptosis may account for the inhibitory effect of hsa_circ_0001829 knockdown on GC proliferation.
3. Overexpression of hsa_circ_0001829 promotes GC cells proliferation, migration and invasion in vitro.
To further verify the functional role of hsa_circ_0001829, hsa_circ_0001829-overexpressing plasmid (pcDNA-circ_0001829) was constructed. The increase of hsa_circ_0001829 expressions induced by pcDNA-circ_0001829 was confirmed by qRT-PCR analysis, which showed approximately tenfold and fifteenfold overexpression in GES-1-T and MKN-28 cells, respectively; meanwhile, no significant change in SLC45A4 mRNA level was observed (Fig. 3A). Subsequent CCK-8 and EdU cell proliferation assays revealed that overexpression of hsa_circ_0001829 obviously increased the growth of both GES-1-T and MKN-28 cells (Fig. 3B, C). Additionally, the migration and invasion abilities of GES-1-T and MKN-28 cells were enhanced by transfection with hsa_circ_0001829 overexpression vector, as indicated by the results of transwell and wound healing assays (Fig. 3D, E).
Next, the effect of hsa_circ_0001829 overexpression on cell cycle and apoptosis rates of GES-1-T and MKN-28 cells were also analyzed. Contrary to the effect induced by hsa_circ_0001829 knockdown, a significant reduction of cells in the G2/M phase and an increase of cells in the G0/G1 phase were observed upon hsa_circ_0001829 overexpression (Fig. 4C). Furthermore, hsa_circ_0001829 overexpression reduced the apoptotic rates of GES-1-T and MKN28 cells (Fig. 4D). These results demonstrated that hsa_circ_0001829 overexpression stimulated cell cycle progression and decreased apoptosis rates.
Taken together, these data indicated that overexpression of hsa_circ_0001829 promotes GC cells proliferation, migration and invasion, and the promoting cell cycle progression and decreased apoptosis may account for the promotional effect of hsa_circ_0001829 overexpression on GC proliferation.
4. Hsa_circ_0001829 acts as a molecular sponge for miR-155-5p.
To explore the regulatory mechanism of hsa_circ_0001829 in GES-1-T cells, we first evaluated its subcellular distribution. The subcellular localization was first evaluated via qRT–PCR in nuclear or cytoplasmic fractions; as shown in Fig. 5A, hsa_circ_0001829 was mainly distributed in the cytoplasm of GES-1-T cells. Then FISH assay was employed, and the red fluorescent distribution also indicated that hsa_circ_0001829 was mainly distributed in the cytoplasm of GES-1-T cells (Fig. 5B).
Given that circRNAs in the cytoplasm have been widely reported to serve as miRNA sponges, we subsequently explored whether hsa_circ_0001829 could act as a miRNA sponge. TargetScan(22) and miRanda(23) were applied to predict the potential absorbing miRNAs of hsa_circ_0001829. Then, 13 overlapped miRNAs of the two databases were found to potentially interact with hsa_circ_0001829. Among the 13 candidate miRNAs, we mainly focused on five miRNAs (miR-153-3p, miR-144-5p, miR-194-5p, miR-155-5p and miR-564-5p), which have been reported to be closely related to cancer development(24–28). For further investigation, dual luciferase reporter assays were performed and demonstrated that relative luciferase activity was reduced by more than 30% when cells were co-transfected with hsa_circ_0001829 wild-type vector and miR-155-5p mimics, whereas co-transfection with the other four miRNAs mimics induced non-significant differences, suggesting that hsa_circ_0001829 interacts directly with miR-155-5p (Fig. 5C). To further confirm this finding, luciferase reporter plasmids with a wild type (WT) of hsa_circ_0001829 sequence and a mutant (Mut) hsa_circ_0001829 sequence in the binding sites of miR-155-5p were generated. We found that co-transfection of hsa_circ_0001829 (WT) and miR-155-5p mimic significantly reduced the luciferase activity, while co-transfection of hsa_circ_0001829 (Mut) and miR-155-5p mimic had no effect on luciferase activity (Fig. 5D).
These results indicated that hsa_circ_0001829 directly binds to miR-155-5p, acting as a sponge for miR-155-5p.
5. SMAD2 was validated as a target gene of miR-155-5p.
Having investigated the interaction between hsa_circ_0001829 and miR-155-5p, we next used TargetScan to predict the target genes of miR-155-5p and 556 target genes were obtained. Then KEGG pathway analysis by David(29) was applied to predict the involvement pathways of 556 genes, and 14 significantly enriched pathways were acquired (Fig. 5E). Importantly, “pathway in the cancer” ranked first and SMAD2 was one of genes in this pathway.
We then overexpressed or silenced miR-155-5p expression and measured the expression levels of SMAD2. Results showed that transfection of miR-155-5p mimic decreased the expression of SMAD2 mRNA, while transfection of miR-155-5p inhibitor increased SMAD2 mRNA expression in both GES-1-T and MKN-28 cells (Fig. 5F). Thus, SMAD2 is a putative target gene of miR-155-5p. To confirm the interaction between miR-155-5p and SMAD2, we conducted dual luciferase reporter assays and found that cells co-transfected with the plasmid containing 3′-UTR-WT regions of SMAD2 (SMAD2 WT) and miR-155-5p mimic had significantly less luciferase activity than the controls, while mutation of the potential miR-155-5p binding sites in the SMAD2 3′-UTR (SMAD2 Mut) abolished this effect (Fig. 5G).
Collectively, these results indicated that miR-155-5p directly targets SMAD2.
6. Hsa_circ_0001829 promotes GC progression via the miR-155-5p-SMAD2 pathway.
We further determined whether hsa_circ_0001829 regulates the expression of SMAD2 by sponging miR-155-5p. First, we tested the expression of SMAD2 mRNA in 20 pairs of matched GC and adjacent tissue samples. Compared with adjacent normal tissue, SMAD2 mRNA was highly expressed in cancer tissues (Fig. 6A). And we found that the level of hsa_circ_0001829 was positively correlated with the mRNA level of SMAD2 (Fig. 6B). Furthermore, hsa_circ_0001829 silencing decreased SMAD2 expression and hsa_circ_0001829 overexpression significantly increased SMAD2 expression, in both GES-1-T and MKN-28 cells (Fig. 6C). Next, GES-1-T and MKN-28 cells were co-transfected with miR-155-5p mimics and hsa_circ_0001829 overexpression vectors. The qRT-PCR assay showed that the promoting effect of hsa_circ_0001829 overexpression on SMAD2 expression was reversed by miR-155-5p mimics (Fig. 6D). These findings indicated that hsa_circ_0001829 sponges miR-155-5p to regulate SMAD2 expression.
To determine whether hsa_circ_0001829 promotes GC progression by sponging miR-155-5p, rescue experiments were performed by co-transfecting hsa_circ_0001829 vectors and miR-155-5p mimics into GES-1-T or MKN-28 cells. The CCK-8 and EdU assay showed that miR-155-5p mimics effectively reversed the hsa_circ_0001829-induced promotion of proliferation in GES-1-T and MKN-28 cells (Fig. 6E, F). Likewise, transwell assay indicated that miR-155-5p mimics attenuated the ability of hsa_circ_0001829 to promote invasion of GES-1-T and MKN-28 cells (Fig. 6G). Taken together, these results demonstrated that hsa_circ_0001829 promotes GC progression through the miR-155-5p-SMAD2 pathway.
7. Suppression of hsa_circ_0001829 expression inhibits subcutaneous tumor growth in vivo.
To further clarify whether hsa_circ_0001829 influences tumor growth in vivo, sh-cirRNA-1# or sh-NC were stably infected by lentiviral infection into GES-1-T cells, and the cells were subcutaneously injected into the flank of NOG mice. The tumors of mice in the sh-circRNA-1# group were considerably decreased in size and volume compared to those in the sh-NC group (Fig. 7A, B). The tumor weights were much lower in the sh-circRNA-1# group than in the sh-NC group (Fig. 7C). IHC analysis demonstrated that the expression of PCNA and MMP2 was decreased by hsa_circ_000289 knockdown (Fig. 7D), which suggest that hsa_circ_0001829 knockdown inhibited the proliferative and invasion abilities of subcutaneous tumors. Also, the SMAD2 expression in the tumors of the sh-circRNA-1# group was decreased, which was consistent with above observations in vitro (Fig. 7D).
These results indicated that hsa_circ_0001829 knockdown inhibited subcutaneous tumor growth in vivo, and the decreased proliferative and invasion abilities may account for the inhibitory effect.