METTL3-mediated m6A modification of circRNF220 modulates miR-330-5p/survivin axis to promote osteosarcoma progression

We obtained 32 pairs of OS and healthy para-carcinoma tissue samples from Renmin Hospital of Wuhan University. The aforementioned hospital’s Ethics Committee approved and reviewed investigations involving human samples and animals. All the patients signed an informed consent form.

OS cell lines (SaOS-2, HOS) and healthy osteoblasts (hFOB1.19) were cultured in 10% fetal bovine serum (FBS)-containing Dulbecco’s modified eagle’s media (DMEM). The circRNF220 shRNA, METTL3 shRNA and inhibitor of miR-330-5p were purchased from Servicebio (Wuhan, China). The pcDNA3.1-survivin plasmid was synthesized by Genepharma (Shanghai, China). Lipofectamine™ 3000 (Invitrogen, Shanghai, China) was used as the transfection vehicle, and the transfection efficiency was determined by qRT-PCR.

Total RNA extraction was carried out in accordance with the TRIzol (Invitrogen) protocol. PrimeScriptTM RT reagent Kit was used to synthesize cDNA, followed by qRT-PCR using TB Green Premix Ex TaqTM. We normalized the circRNA and mRNA levels against GAPDH, whereas the miRNA levels were normalized against U6. Supplementary Table 1 summarizes the primer information.

Briefly, 2 mg/mL of actinomycin D and 3 U/µL ribonuclease R (RNase R) were used to analyze the stability of circRNF220. After the given duration of culture, the circRNF220 and RNF220 mRNA expressions were assessed by qRT-PCR.

CircRNF220’s 3′-UTR and survivin having miR-330-5p’s possible binding site were inserted into the pGL6-miR vector (Genepharma, Shanghai, China). Followed by co-transfection of 293T cells with miR-330-5p mimic and reporter vectors. The luciferase activity was determined by qRT-PCR.

The Magna RIP kit was used to assess the enrichment of circRNF220. The lysate was incubated in RIP buffer containing anti-Ago2 or anti-IgG antibody-bound magnetic beads. Following RNA purification, qRT-PCR was utilized to determine the levels of miR-330-5p and circRNF220.

MiR-330-5p and circRNF220 probes were synthesized and acquired from GenePharma (Shanghai, China). The OS cells were hybridized with the above probes overnight at 37 °C per the manufacturer’s instructions. The images were then captured with immunofluorescence microscopy.

Following transfection, 96-well microplates were used to inoculate the OS cells, which were then treated with the CCK-8 solution (10 µl). The optical density (OD) at 450 nm was determined by a microplate reader. In addition, the cellular proliferative potential was evaluated using an EdU assay (Beyotime, China). Following a 30 min incubation of OS cells in EdU solution (100 µL), nuclei were stained with 300 µL DAPI solution for 5 min. Finally, cell images were taken by fluorescence microscopy (Zeiss, Germany).

After transfection, the transfected OS cells were inoculated into each well of a 6-well microplate. Then using a 10 µL pipette tip, a linear wound is created in each well. The cell’s migratory potential was microscopically evaluated at different time points.

The six-well microplates were inoculated with the post-transfected OS cells. After 2 weeks of incubation, colonies were fixed using methanol and stained with 0.1% crystal violet. Under microscopic observation, we enumerated the colonies and manually counted those consisting of at least 50 cells.

In brief, the upper chamber of a six-well Transwell plate was seeded with 5 × 10⁵ cells in serum-free medium, while a medium containing 20% FBS was dispensed into the bottom chamber. After that, the cells were immobilized in both chambers with paraformaldehyde (4%) and stained with crystal violet (0.1%). Finally, cells were quantified using a microscope.

RIPA buffer was utilized to extract the total cellular proteins. Equal amounts of proteins from different samples were run on an SDS-PAGE gel, followed by transferring them onto the polyvinylidene difluoride (PVDF) membrane. The PVDF membrane was then incubated overnight with a blocking buffer followed by incubation at 4 °C with 1:1000 dilution of primary antibodies, anti-GAPDH, and anti-survivin (Proteintech). Finally, after secondary antibody incubation, Image Lab (Bio-Rad) was used to photograph images.

Total RNA was extracted using TRIzol reagent, and anti-m6A antibody or immunoglobulin (IgG) bound protein A/G magnetic beads were used. The total RNA was washed with elution buffer to isolate m6A-modified RNA. Subsequently, qRT-PCR analysis was conducted to assess the m6A accumulation in circRNF220.

Murine experimentation was approved by the Renmin Hospital ethics committee of Wuhan University. Briefly, mice were injected with 3 × 106 HOS cells having sh-NC or sh-circRNF220. Five weeks later, the mice were killed and their tumor weights were recorded. Following that the tumors were subjected to immunohistochemistry and immunofluorescence staining.

The GEO database was used to search for circRNAs that expressed differentially in OS versus healthy tissues. Starbase 2.0 was used to predict potential binding sites for miR-330–5p, circRNF220, and survivin.

In contrast to healthy controls, circRNA array analysis in GSE140256 indicated a high OS tissue level of circRNF220 (P = 0.0005, Fig. 1A). Sanger sequencing was then used to confirm the back-spliced region sequence (Fig. 1B). After the transcription inhibition by Actinomycin D, the half-life of circRNF220 was longer than that of linear RNF220 in HOS (P = 0.0065) and SaOS-2 (P = 0.0073) cells (Fig. 1C). Furthermore, circRNF220 was more RNase R resistant than RNF220 (Fig. 1D). The qRT-PCR data confirmed elevated circRNF220 levels in the OS cells and tissue (Fig. 1E, F). Moreover, FISH assays revealed that circRNF220 was predominantly localized in the cytoplasm (Fig. 1G). In addition, clinicopathological information demonstrated that high expression of circRNF220 was associated with TNM stage and tumor size (Supplementary Table 2). Overall, the results showed that circRNF220 is stable, and its expression gets upregulated in OS.

The qRT-PCR analysis showed that treating cells with sh-circRNF220 led to the knockdown of circRNF220 (Fig. 2A). Since sh-circRNF220#1 had higher knockdown efficiency, it was used for further experiments. CCK-8 and EdU assays demonstrated that circRNF220 silencing led to the suppression of OS cell multiplication (Fig. 2B, C). Further, upon the circRNF220 knockdown, OS cells’ capacity to form colonies was suppressed (Fig. 2D). sh-circRNF220 also decreased the invasive and migratory abilities of OS cells in wound-healing and Transwell assays (Fig. 2E, F). These results collectively confirmed that silencing circRNF220 repressed OS progression in vitro.

The expression of m6A writer enzymes (METTL3, METTL14, and WTAP) was detected using qRT-PCR. The results indicated that only METTL3 exhibited upregulation in OS cells compared to hFOB cells (Fig. 3A−C). Therefore, METTL3 was chosen for further investigation. In addition, the results revealed a significantly elevated expression of METTL3 in OS tissue samples (Fig. 3D). To investigate the potential mechanism by which METTL3 regulated circRNF220, we transfected sh-METTL3 to silence the mRNA and protein expression of METTL3 (Fig. 3E, F). The qRT-PCR results demonstrated a decrease in the expression of circRNF220 following the knockout of METTL3 (Fig. 3G). Furthermore, MeRIP-PCR detection revealed a significant reduction in the m6A level of circRNF220 after shMETTL3 transfection (Fig. 3H). Moreover, the m6A level of circRNF220 was increased by transfection with METTL3-wild type (wt). However, transfection with METTL3-mutant (mut) did not have any effect on the m6A level of circRNF220 (Fig. 3I). In addition, the expression of METTL3 was positively correlated to circRNF220 expression (Fig. 3J). These results revealed that METTL3 enhanced circRNF220 expression via m6A modification.

Circinteractome and starbase, two online databases, were utilized to predict the target of circRNF220. The analysis revealed that two miRNAs potentially bound to circRNF220 based on overlapping predicted results (Fig. 4A). Subsequently, qRT-PCR results confirmed that silencing circRNF220 only led to increased expression of miR-330-5p in OS cells (Fig. 4B). Therefore, miR-330-5p was selected for further investigation. The results revealed a binding site for miR-330-5p in the circRNF220 sequence (Fig. 4C). Subsequently, overexpression of miR-330-5p had a suppressive effect on the relative luciferase activity of the circRNF220-wt reporter but had no effect on the circRNF220-mut reporter (Fig. 4D). The circRNF220–miR-330-5p interaction was further confirmed using RIP assay (Fig. 4E). Furthermore, the FISH assay revealed the cytoplasmic enrichment of circRNF220 and miR-330-5p (Fig. 4F). MiR-330-5p was shown to be downregulated in OS cells and tissues as compared to healthy tissues and cells (Fig. 4G, H). According to Pearson’s correlation outcomes, circRNF220 was negatively correlated to miR-330–5p in the OS tissues (Fig. 4I). Furthermore, miR-330–5p levels were upregulated in OS cells upon silencing of circRNF220 (Fig. 4J). Thus, circRNF220 directly targets miR-330-5p.

The qRT-PCR results showed that the miR-330-5p inhibitor decreases the sh-circRNF220-induced upregulation of miR-330-5p (Fig. 5A). Consequently, the sh-circRNF220-mediated suppression of cell proliferation, invasion repression, and migratory retardation were all counterbalanced (Fig. 5B–H). The above findings confirmed that circRNF220 regulates OS progression through miR-330-5p sponging.

Bioinformatics analysis was used to forecast potential targets (survivin, KCNIP2, and PTBP1) for miR-330-5p (Fig. 6A). The expression of survivin was significantly reduced following the miR-330-5p mimic transfection in OS cells (Fig. 6B). Survivin and miR-330-5p were shown to share binding sites (Fig. 6C). The Luciferase reporter assay showed that compared to the survivin 3’UTR-mut reporter, the survivin 3’UTR-wt reporter exhibited upregulated luciferase inhibition in the OS cells following miR-330-5p transfection (Fig. 6D). In addition, survivin was upregulated in OS tissues (Fig. 6E). Western blotting unraveled that the survivin protein levels were lowered by miR-330-5p mimic (Fig. 6F), and survivin is negatively correlated to the expression of miR-330-5p while positively correlated to circRNF220 expression (Fig. 6G, H). In addition, the knockdown of circRNF220 evoked mRNA and protein downregulation of survivin, while miR-330-5p inhibitor antagonized this inhibition (Fig. 6I, J). These findings demonstrated that circRNF220 upregulated the level of survivin through miR-330-5p sponging.

Survivin overexpression and sh-circRNF220 vectors were transfected into OS cells to examine whether circRNF220 facilitated OS progression through the miR-330-5p/survivin axis. The results showed that survivin overexpression and miR-330-5p inhibition partially reversed the circRNF220 knockdown-induced suppression of cellular multiplication, migration, and invasion (Fig. 7A–G).

The in vivo effect of circRNF220 was investigated using a xenografted model. In the sh-circRNF220 group, the volume and weight of tumors were reduced compared to the control (Fig. 8A–C). In addition, the expression of circRNF220 and survivin were decreased, while miR-330-5p expression was increased (Fig. 8D–F). Similarly, immunofluorescence and western blot assays showed that the survivin protein level was lower in the sh-circRNF220 group (Fig. 8G). In addition, FISH revealed the colocation between circRNF220 and miR-330-5p (Fig. 8H, I).