Human gastric cancer cell lines AGS and MKN-7 were purchased from IMMOCELL (Xiamen, Fujian, China). Both cell lines were maintained in Eagle’s minimum essential medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 U/ml streptomycin. All the reagents were purchased from Gibco. Actinomycin D (1 μM; Sigma) was used to check the stability of RNA transcripts by treating cells for the indicated time points.

To check the localization and expression of circ0007360, we used the RNA FISH method. The whole procedure was performed according to the instructions from the Fluorescent in situ Hybridization Kit (Rabobio; C10910). The probes were also purchased from Rabobio as follows: FISH Probe Mix (Red; Rabobio; C10920), h-U6 FISH Probe Mix (Red) (Rabobio; LNC110101), and h-18S FISH probe mix (red; Rabobio; LNC110201).

Total gDNA was isolated from AGS cells using the DNeasy Blood and Tissue Kit (Qiagen; 69504). Divergent and convergent primers were used for polymerase chain reaction (PCR) amplification of circ007360 and GAPDH, respectively. The PCR products from cDNA or gDNA were separated using agarose gel electrophoresis. All primers used for the PCR are listed in Supplementary Table S1.

The circ0007360 ectopic expression construct was generated using PCR amplification and ligated to the pCDNA3.1 backbone. Circ0007360 and IRF7 shRNA were generated by inserting the annealed shRNA oligos into the pLKO.1-TRC backbone. The reporter constructs for wild type circ0007360, mutant circ0007360, wild type IRF7 3′UTR, or mutant IRF7 3′UTR were amplified and inserted into the pmirGLO (Promega) backbone. All primers used for cloning are listed in Supplementary Table S1.

Total RNA was isolated using the Total RNA Extraction Kit (Vazyme Biotech, Nanjing; China). Subsequently, 1 μg of RNA was reverse transcribed using the HiScript II First Strand cDNA Synthesis Kit (Vazyme Biotech). Next, the cDNA was subjected to quantitative PCR using the ChamQ SYBR Master Mix kit (Vazyme Biotech) and iQ5 qPCR machine (Bio-Rad) to detect the relative expression levels of target genes. The relative levels of various target genes were quantified using the 2^−ΔΔCt formula, with 18S RNA as the reference transcript for normalization. All primers used in this study are listed in Supplementary Table S1. RNase R treatment was performed as follows: 1 μg RNA was diluted in 40 μl water supplemented with 4 U RNase R (Thermo Fisher), after which the mixture was incubated for 15 min at 37°C.

To detect the localization of circ0007360, the PARIS^™ Kit (Thermo Fisher; AM 1921) was used to extract total RNA from the cytoplasm or nucleus. RT-qPCR was performed as described above.

Cells were plated in wells of six-well plates and transfected with 4 μg plasmids or 200 pmol miRNA mimics or inhibitors using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions.

AGS cells were co-transfected with the indicated luciferase reporter plasmids and the miR-762 mimic or mimic NC. At 48 h post-transfection, the cells were lysed and luciferase activity was measured using the Dual-Glo Luciferase Assay System (Promega). Firefly luciferase activity was normalized to Renilla luciferase activity.

Briefly, 24 h after transfection, the cells were plated at 1 × 10³ cells/well in 96-well plates. MTT solution (20 μl, 5 mg/ml; Qiancheng Biotech, China) dissolved in PBS was directly added to each well at the indicated time points. After incubation for 4 h at 37°C, the medium was replaced with 100 μl DMSO, and the absorbance at 490 nm was measured using a SpectraMax Absorbance Reader (Molecular Devices). The results are presented as the mean ± SD (n = 6).

At 24 h post-transfection, 500 cells were plated in the wells of a six-well plate. After culturing for 2 weeks at 37°C, the cells were fixed with 4% paraformaldehyde for 10 min. Staining was performed using 0.5% crystal violet for 10 min. The stained cell colonies were photographed and counted after three washes with water.

To quantify the frequencies of cells in different cell cycle phases, at 24 h post-transfection, the cells were fixed with 70% ethanol overnight at 4°C. Subsequently, the cells were washed twice with PBS before permeabilization in PBS buffer supplemented with Triton X-100 (0.2%) and RNase (10 μg/ml) for 30 min. Subsequently, the cells were stained with propidium iodide (PI; 20 μg/ml; A211-02; Vazyme Biotech) solution in the dark for 30 min at room temperature (RT), after which 1 × 10⁴ cells were quantified using a NovoCyte flow cytometer (ACEA Biosciences, San Diego, CA, United States).

Similarly, the cell apoptosis assay was initiated by treating the transfected cells with Annexin V-fluorescein isothiocyanate (200 μg/ml) and PI (30 μg/ml) in the dark for 10 min at RT. After several washes with PBS buffer, the NovoCyte flow cytometer was used to determine the frequency of apoptotic cells.

To check the expression of CD44 and EpCAM on the cell surface, cells were stained with isotope-FITC, anti-CD44-FITC, or anti-EpCAM-FITC (all from BioLegend) and the signal was detected using a NovoCyte flow cytometer. In all the assays mentioned above, unstained cells served as negative controls to establish background fluorescence thresholds.

For migration assays, at 24 h post-transfection, 5 × 10⁵ serum-starved cells were seeded into the upper chambers of 24-well Transwell plates (Corning, NY, United States). For the evaluation of cell invasion, a diluted Matrigel (BD) pre-coated membrane of the top chamber was used. For both assays, 10% serum medium was added to the bottom wells. After 24 h of incubation, 4% PFA was applied to fix the migrated or invasive cells on the bottom part of the chamber, and the cells were stained with 0.5% toluidine blue before imaging. Five independent fields were selected for statistical analysis.

Upon 24 h transfection, cells were trypsinized and 1 × 10³ cells were seeded into wells of 24-well plates containing DMEM/F12 medium (IMMOCELL, China) supplemented with B27 (Gibco), 20 ng/ml EGF (Gibco), 20 ng/ml bFGF (Nowoprotein, China), and 4 μg/ml insulin (Gibco). Fresh medium was added every 3 days and the cells were cultured for 12–18 days before being photographed.

RIPA buffer (Beyotime) supplemented with 1 × complete protease inhibitor cocktail (Roche) was used to lyse the cells. Protein concentrations were determined using a bicinchoninic acid protein assay kit (Thermo Fisher). Equal amounts of proteins were separated using SDS-PAGE and subsequently transferred onto a 45-μm polyvinylidene difluoride (PVDF) membrane (Bio-Rad). Next, the membranes were blocked with 5% non-fat dry milk in Tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h at RT. The resulting membranes were probed with primary antibodies overnight at 4°C. After three washes with TBST, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 h at RT. The signal was detected using Clarity^™ Western ECL Substrate (Thermo Fisher) and the ChemiDoc Imaging System (Bio-Rad). All antibodies used in this study are listed in Supplementary Table S2.

All mouse experiments were conducted in accordance with a protocol approved by the Animal Care and Use Committee of Nanchang University. For tumor growth assays, 5 × 10⁶ AGS cells with or without circ0007360 depletion were subcutaneously injected into the lower back regions of 6-week-old female BALB/c nude mice for 34 days (n = 6). Tumor volumes were monitored every 4 days with calipers from 18 days post-inoculation. Tumor volumes were calculated using the following formula: length ×  width²/2. The mice were euthanized 34 days after inoculation, after which the tumors were collected and photographed. Moreover, the tumor weight was determined 34 days post-inoculation.

First, the tumors from the mice were fixed and embedded. After sectioning, the paraffin was removed by placing the slides in xylene three times, followed by placing them in 100% ethanol twice. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 20 min, after which the slides were rehydrated in 96, 70, and 50% ethanol, respectively. After antigen retrieval and washing, the slides were slowly cooled to RT, followed by three times washing with PBST. Primary antibody against Ki67 (BD; 550609) diluted (1:100) in 1% BSA solution was used to incubate the slides overnight at 4°C. The slides were washed with PBST three times before incubating with 1:200 diluted biotinylated secondary antibody (DAKO; E0353) for 30 min at RT. After three times washing with PBST, the slides were incubated with Vectastain complex (Vector Laboratories; PK-6100) for 30 min. Subsequently, the slides were washed with PBST three times and developed using the DAB reagent. Next, slides were counterstained with Mayer hematoxylin (Sigma-Aldrich; MHS80) for 45 s and dehydrated. Finally, Entellan mounting medium (Merck; 107961) was used to mount the slides. Images were captured, and the rate of positive cells was analyzed using the AIpathwell software (Servicebio, China).

Statistical analyses were performed using the GraphPad Prism 8 software. All results are shown as the mean ± SD. Unpaired Student’s t-test was used for analysis, and p < 0.05 was considered to be significant (*0.01 < p < 0.05, **0.001 < p < 0.01, ***0.0001 < p < 0.001, ****p < 0.0001). NS, not significant.

We started our investigation by checking whether circ0007360 is a real circular RNA. To this end, a convergent primer pair was used to amplify the junction site of circ0007360. Sanger sequencing showed that the junction site could be detected, as predicted (Figure 1A). However, as expected, the PCR products generated by the divergent primer pair could not be detected when genomic DNA (gDNA) was used as a template (Figure 1B). Of note, the convergent primer pair for circ0007360 or GAPDH, which served as a positive control, amplified the expected fragment in both cDNA and gDNA (Figure 1B). Since resistance to exonuclease RNase R is also a gold standard for testing whether an RNA transcript is circular (Jeck and Sharpless, 2014), we treated the cDNA from two gastric cancer cell lines, AGS and MKN-7, with RNase R. RT-qPCR results showed that the levels of ATF6 mRNA, which is the host mRNA of circ0007360, but not circ0007360, were reduced upon RNase R challenge (Figure 1C). Furthermore, a time-course experiment directed by actinomycin D, an inhibitor of transcription, revealed that the stability of circ0007360 was significantly higher than ATF6 mRNA in both gastric cancer cell lines (Figure 1D). Collectively, these results indicate that circ0007360 is a real circular RNA.

As the localization of circRNAs determines their functions (Hsiao et al., 2017), we next checked the localization of circ0007360 using an RNA FISH assay. Circ0007360 was more abundant in the cytoplasm, where the 18S RNA was localized, than that in the nucleus, where another positive control U6 RNA was localized (Figure 2A). Moreover, subcellular fractionation analysis confirmed the abundant cytoplasmic location of circ0007360 (Figure 2B). Since the most common mechanism of cytosolic circRNAs is acting as sponges for matching miRNAs, we performed in silico prediction for target miRNAs of circ0007360. After analyzing the intersection result from three databases (Targetscan (Agarwal et al., 2015), miRDB (Chen and Wang, 2020) and circMine (Zhang et al., 2021)), six candidate miRNAs were enriched (miR-762, miR-4427, miR-4656, miR-3135b, miR-4680-3p and miR-8077; Figure 2C). Given the fact that miR-762 is a well-annotated miRNA whose promoting role in cancer progression has been reported by many research groups (Li et al., 2015; Ge et al., 2019; Chen et al., 2020), we focused on miR-762 for further experimental validation. Two putative binding sites of miR-762 on circ0007360 was found using the CircMir 1.0 platform (Xu et al., 2021) (Figures 2D,E). To prove the interaction between these two molecules, reporter assays were performed. We found that miR-762 attenuated the luciferase activity of both predicted interacting fragments from circ0007360 (WT), whereas the corresponding mutant circ0007360 (MUT) fragments were not affected by miR-762 overexpression (Figure 2F).

To investigate the effects of circ0007360 and miR-762 on the survival of gastric cancer cells, we misexpressed them by transfecting circ0007360 ectopic expressing construct, shRNA against circ0007360, or miRNA inhibitors for miR-762 into cells (Figure 3A). MTT evaluation suggested that overexpression of circ0007360 or depletion of miR-762 mitigated, while inhibition of circ0007360 expression augmented the proliferation of gastric cancer cells (Figure 3B). Moreover, colony formation analysis confirmed the negative or positive effects of circ0007360 and miR-762 on gastric cancer growth, respectively (Figures 3C,D). We then tested whether the cell cycle was altered upon the misexpression of these two RNAs. Results from propidium iodide (PI)-based flow cytometry assay revealed that overexpression of circ0007360 or knockdown of miR-762 alleviated, whereas circ0007360 depletion potentiated the cell cycle progression of gastric cancer cells (Figures 3E,F). In contrast, we found that the apoptosis of cells was enhanced by circ0007360 overexpression or the absence of miR-762 but was inhibited by silencing circ0007360 expression (Figures 3G,H). Furthermore, we examined the migration and invasion of gastric cancer cells in the aforementioned genetic settings. Surprisingly, the migratory and invasive abilities of cells decreased significantly when circ0007360 was ectopically expressed or miR-762 was depleted, while the migration and invasion increased upon knockdown of circ0007360 (Figures 4A,B). Taken together, our results demonstrate the tumor-suppressive effects of circ0007360 and the tumor-promoting effects of miR-762 in two gastric cancer cell lines.

Considering the contribution of cancer stem cells to tumor recurrence and chemotherapy resistance (Bekaii-Saab and El-Rayes, 2017), we investigated whether circ0007360 and miR-762 affect the stemness of gastric cancer cells. Data from tumor sphere formation assays showed that circ0007360 overexpression or miR-762 depletion reduced the size of tumor spheres formed by gastric cancer cells (Figure 5A). However, the absence of circ0007360 promoted the size of the tumor spheres (Figure 5A). We then tested the cell surface expression of stem cell markers to further evaluate the stemness of cells at the molecular level. Flow cytometry results revealed that the levels of CD44 and EpCAM were downregulated upon circ0007360 ectopic expression or miR-762 depletion, but the levels of the two markers were upregulated when circ0007360 was silenced (Figure 5B). Furthermore, protein levels of other stemness markers, such as MYC, Nanog, ALDH1, SOX2, and CD133, were also reduced by the overexpression of circ0007360 or miR-762 depletion (Figure 5C). However, knockdown of circ0007360 potentiated the protein levels of these markers (Figure 5C). In conclusion, circ0007360 hinders to, while miR-762 contributes to, the stemness of gastric cancer cells.

To validate the significance of miR-762 in the effects exerted by circ0007360, we proceeded by depleting miR-762 in gastric cancer cells with circ0007360 knockdown (Figure 6A). Surprisingly, we observed that the enhancement of cell proliferation, colony formation abilities, cell cycle progression, cell migration, invasion, and stemness of cells, and the inhibition of cell apoptosis mediated by the absence of circ0007360 were significantly restored when miR-762 expression was silenced by the specific miRNA inhibitors (Figures 6A–I and Supplementary Figure S1). Taken together, the rescue experiments demonstrate that the biological effects triggered by the loss of circ0007360 are highly dependent on the upregulation of miR-762, which serves as a pivotal downstream target miRNA of circ0007360.

To screen the target mRNA of miR-762, we mined three well-established databases, miRDIP (Tokar et al., 2018), TargetScan (Agarwal et al., 2015), and miRTarBase (Huang et al., 2020) (Figure 7A). Of note, IRF7 was the only mRNA hit shared by these three databases (Figures 7A,B). Next, we performed a luciferase reporter assay to validate the interaction between miR-762 and the IRF7 3′UTR. As expected, miR-762 dramatically inhibited the activity of the wild type, but not the mutant IRF7 3′UTR (Figure 7C). Consistently, both the mRNA and protein levels of IRF7 increased upon depletion of miR-762 (Figures 7D–F). Moreover, as the upstream modulator of miR-762, circ0007360 positively regulated the levels of IRF7 (Figures 7D–F). Thus, our results demonstrate that circ0007360 may indirectly enhance the expression of IRF7 by decreasing miR-762 levels.

Since we validated that IRF7 is a downstream target gene of miR-762, the following investigations were performed to confirm the significance of IRF7 in miR-762-induced gastric cancer progression. First, IRF7 was silenced by shRNA in cells depleted of miR-762 (Figures 8A,B). Results from MTT and colony formation assays indicated that inhibition of cell survival resulting from miR-762 knockdown was alleviated when IRF7 was depleted (Figures 8C,D). In line with these data, the effects of miR-762 depletion on cell cycle arrest, apoptosis, migration, invasion, and stemness were rescued upon silencing IRF7 (Figures 8E–J). Thus, our data prove that IRF7 inhibition is responsible for the tumor-promoting function directed by miR-762. Moreover, taking all these in vitro data together, we indicate that the progression of gastric cancer may be attenuated by the circ0007360/miR-762/IRF7 axis.

To further consolidate the aforementioned in vitro results, we performed in vivo experiments to determine the effects of circ0007360 on gastric cancer cells. Therefore, we inoculated AGS cells with or without circ0007360 depletion in mice through subcutaneous injection and found that the tumor volume and tumor weight increased significantly in the circ0007360 knockdown group compared with the control group (Figures 9A–C). Moreover, RT-qPCR results confirmed that miR-762 levels were enhanced, while IRF7 levels were reduced, upon depletion of circ0007360 in the isolated tumors formed by the human gastric cancer cells (Figures 9D–F). Furthermore, we found that absence of circ0007360 resulted in the upregulation of stemness markers expressed, as observed in in vitro assays (Figures 9G,H). Importantly, knockdown of circ0007360 led to an increase in the percentage of Ki67-positive cells in tumors (Figures 9I,J). Collectively, the in vivo mouse experiments confirm the inhibitory role of circ0007360 on tumorigenesis and cell stemness and the existence of the circ0007360/miR-762/IRF7 axis in tumors formed in the mouse xenograft model.