A total of 88 NSCLC patients whose diagnoses were confirmed via biopsy were selected for this study. These patients received no prior radiotherapy or chemotherapy before surgery. This study was approved by the Research Ethics Committee at Cangzhou Central Hospital. All patients signed the informed consent for the use of their patient information and tissues. Tumor tissues and adjacent noncancerous tissues were collected and stored at 80°C until use.

We compared the levels of circ_0006677 in NSCLC tissues and adjacent normal tissues using the GEO dataset (GSE112214).

The potential miRNAs that might interact with circ_0006677 was predicted using two online tools (circBank and CircInteractome). The interaction between miRNA and its target genes was analyzed using the TargetScan database.

NSCLC cell lines (CALU3, CALU6, A549, H1229, and H1975) and human bronchial epithelial cells (HBE) were purchased from the Cell Bank of Type Culture Collection of Chinese Academy of Sciences, Shanghai, China). All these cell lines were cultured in RPMI-1640 medium (Solarbio, China) supplemented with 10% fetal bovine serum, streptomycin, and penicillin at 37°C in 5% CO2. HEK293T cells were cultured in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin (SH30010, Hyclone), and 100 mg/ml streptomycin.

To generate circ_0006677 overexpression lentiviral plasmid, circ_0006677 was constructed into a pLV plasmid. For virus packaging, pLV-circ_0006677, PsPAX2, and pMD2G plasmids were co-transfected into HEK293T cells. The viral supernatant was collected to construct stable circ_0006677–overexpressing cell lines.

RNase R (Epicentre, China) was used to eliminate the linear RNAs. The expression of GAPDH and circ_0006677 before and after RNase R treatment was assessed.

Cells growth was monitored with CCK-8 reagent every day following the manufacturer’s instruction (Dojindo Laboratories, Kumamoto, Japan). In brief, 1, 000 cells were seeded in 96-well plates and cultured overnight. CCK-8 reagent was added at the time point, and the absorbance at 450 nm was measured using a DNM-9602 Microplate Reader (Perlong, Beijing, China).

One thousand cells were seeded in six-well plates and cultured for 14 days. Cell colonies were fixed and stained with 0.5% crystal violet in methyl alcohol. Images of cell clones were taken, and colonies were counted with ImageJ.

For migration assay, cells in the logarithmic growth stage were starved for 24 h, and the cells were digested the next day, centrifuged, and resuspended for a final concentration of 1 × 10 (Wan et al., 2016)/ml. Cell suspension (0.2 ml) was added to the Transwell upper chamber, and 700 μl of precooled DMEM cell culture medium containing 10% FBS was added to the lower chamber. The cells were cultured in a cell incubator containing 5% CO2 at 37°C. After 24 h, the Transwell chamber was removed and the cells in the upper chamber and the basement membrane were swabbed with wet cotton swabs. The cells were fixed with methanol for 30 min and stained with 0.1% crystal violet dye for 20 min. Five fields (100×) were randomly selected to count the number of transmembrane cells. For invasion assay, Matrigel was loaded in Transwell upper chambers and curdled at 37°C before cells were added to the upper chambers. After culture for 24 h, cells on the lower surface of the membrane were fixed with 4% paraformaldehyde for 10 min and stained by 0.1% crystal violet for 30 min. Cells were counted using an inverted microscope (magnification, ×20).

Glucose consumption and lactic acid production of the cells were measured by glucose assay and lactate assay following the manufacturers’ instruction (Abcam, United States). In brief, cells were incubated with 2-deoxy-d-glucose (2-DG) for 20 min at 37°C. Cells were then washed with PBS to remove exogenous 2-DG. Cells were lysed with extraction buffer, heated at 85°C for 40 min, cooled on ice for 5 min, and centrifuged at 13,000g. The supernatant was transferred to new tubes and incubated with mix A for 1 h at 37°C. After mix B was added to the reaction mixture, the absorbance of each well was analyzed under the wavelength of 450 nm every 2–3 min on a microplate reader, set in the kinetic mode.

Samples of patient tissue and NSCLC cell lines were lysed with TRIzol reagent (Invitrogen Carlsbad, CA, United States) and total RNAs were isolated following the manufacturer’s instructions. The RNA sample was reverse transcribed by the Reverse Transcriptase Kit (Takara, Beijing, China). Real-time PCR was performed with the CFX Connect Real-time PCR Machine (Bio-Rad) and SYBR green reagent, as described previously (Zou et al., 2020). GAPDH was used as internal controls. Primer sequences were as follows: MIR578-Forward: CTTCTTGTGCTCTAGGAT and MIR578-Reverse: GAA​CAT​GTC​TGC​GTA​TCT​C; SOCS2-Forward: GGT​CGG​CGG​AGG​AGC​CAT​CC and SOCS2-Reverse: GAA​AGT​TCC​TTC​TGG​TGC​CTC​TT; circ_0006677-Forward: ACT​TGT​GAT​GCC​CTG​ACT​G and circ_0006677-Reverse: ACT​TGG​ATC​CGT​CAC​GTT​G; and GAPDH-Forward: GAA​GGT​GAA​GGT​CGG​AGT​C and GAPDH-Reverse: GAA​GAT​GGT​GAT​GGG​ATT​TC.

The wild-type (WT) or mutated (MUT) sequences at 3′-UTR of SOCS2 mRNA were synthesized and cloned into the pMIR-Reporter Vector (Promega Corporation, Madison, WI, United States). Using Lipofectamine 2000 reagent, A549 and H1299 cells were co-transfected with SOCS2 wild-type 3′UTR or mutant-type SOCS2 3′-UTR with miR-578 mimic or control mimic (miR-NC). 48 h after transfection, cells were harvested, and the luciferase activity was measured using a dual luciferase assay (Promega Corporation, Madison, WI, United States). Each experiment was repeated three times.

Biotin-labeled RNAs were transcribed using the Biotin RNA Labeling Mix (Promega Corporation, Madison, WI, United States) and T7 RNA polymerase (Promega Corporation, Madison, WI, United States), treated with RNase-free DNase I (Promega, United States), and then purified with the RNeasy Mini Kit (Promega, United States). Transcriptional production of a biotinylated circ_0006677 probe or control probe (NC-probe) was heated at 95°C for 5 min, placed on ice for 5 min, and placed at room temperature for 20 min to form the secondary structure. Folded RNA was mixed with cell extracts for 2 h. Then, 50 µl of streptavidin agarose beads (Invitrogen, United States) were added to each binding reaction and incubated for 1.5 h. After washing the beads, the samples and inputs were digested by Proteinase K, and RNA was isolated and quantified by qRT-PCR assay.

Cells were lysed by RIPA lysis buffer (Beyotime, Shanghai, China). Protein concentration was determined using BCA assay (Beyotime). Proteins were subjected to 10% SDS–polyacrylamide gel electrophoresis. Separated proteins were transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, United States) and immunoblotted with primary antibodies: anti-SOCS2 (ab71676, Abcam), anti-GAPDH (ab76523, Abcam), anti-MMP2 (ab97779, Abcam), and anti-MMP9 (ab38898, Abcam), and secondary antibody: goat anti-rabbit (ab205719; Abcam). Bands were visualized using the EasyBlot ECL Kit (Sangon Biotech, China). The intensity of the blots was quantified by ImageJ (Rawak Software, Germany).

Fifty four-week-old BALB/C^nu/nu nude mice (20 g) were purchased from Beijing Vital River Laboratory Animal Technology (Beijing, China) and employed to establish the mouse models, as reported previously (Zhu et al., 2019). In brief, 2 × 10⁶ A549 cells were subcutaneously injected into the nude mice. Tumor’s volume of each mouse was measured every 7 days until 35 days. All procedures and animal experiments were approved by the Animal Ethics Committee of Cangzhou Central Hospital.

Statistical analyses were performed using SPSS 21.0 software (IBM, United States). All experiments were repeated three times. All the data were presented as mean ± standard deviation (SD). Differences between two groups compared with unpaired two-tailed t-test, while ANOVA was used for multiple comparisons. The Kaplan–Meier method was used for survival analysis. All correlations used in this analysis were Pearson’s correlation coefficient. p < 0.05 was considered as statistically significant.

To investigate the expression of circ_0006677 in NSCLC tissues, we compared the levels of circ_0006677 in three pairs of NSCLC tissues and adjacent normal tissues in the GEO dataset (GSE112214). We found that circ_0006677 was downregulated in NSCLC tissues compared with adjacent normal tissues (Figure 1A). By performing qRT-PCR analysis, we further confirmed the lower expression of circ_0006677 in NSCLC tissues (n = 88) than the adjacent normal tissues (n = 88) (Figure 1B). Moreover, according to the median of the expression value, 88 patients were divided into the circ_0006677-high group and the circ_0006677-low group. We found that higher levels of circ_000667 expression were significantly associated with longer overall survival (Figure 1C). Moreover, the expression of circ_0006677 was negatively related to the tumor size, TNM stage, and lymph node metastasis (Table 1). The qRT-PCR assays revealed that the expression of circ_0006677 was significantly lower in a series of NSCLC cell lines (especially, A549 and H1299 cells) than the HBE (Figure 1D). Furthermore, A549 and H1299 cells were treated with RNase R, and the expression of WDR78 and circ_0006677 was detected using qRT-PCR assays. The levels of WDR78 were significantly decreased after RNase R treatment, while the expression of circ_0006677 was not altered (Figure 1E). Nuclear/cytoplasmic fractionation of NSCLC cells was performed to explore the subcellular location of circ_0006677 in NSCLC cells. Our qRT-PCR assays suggested that circ_0006677 was mainly located and enriched in the cytoplasm (Figure 1F). These results indicated that circ_0006677 might play a tumor-suppressive role in NSCLC.

To validate the function of circ_0006677 in NSCLC, we sought to ectopically express circ_0006677 in A549 and H1299 cells, which have the lowest levels of circ_0006677. To dos so, we infected A549 and H1299 cells with lentivirus to construct stable cell lines overexpressing circ_0006677 (designated A549-circ_000667 and H1299-circ_000667). Compared with control cells, the circ_0006677 expression level was significantly increased in A549-circ_000667 and H1299-circ_000667 cells (Figure 2A).

Overexpression of circ_0006677 significantly inhibited the proliferation (Figure 2B) and colony-forming capacity (Figure 2C) of A549 and H1299 cells. We also evaluated the impact of circ_0006677 overexpression on cell migration and invasion ability. Upregulation of circ_0006677 could significantly inhibit the migration and invasion ability of A549 and H1299 cells (Figures 2D–F). To evaluate whether circ_0006677 influences NSCLC cell metabolism, we performed glucose assays and lactate assays. Overexpression of circ_0006677 significantly decreased the glucose consumption and lactic acid production in A549 and H1299 cells (Figures 2G,H). Taken together, our data suggested that circ_0006677 plays tumor-suppressive roles in NSCLC cells.

To investigate the downstream regulatory mechanism of circ_0006677 in NSCLC cells, we predicted the potential miRNAs using online prediction tools. As a result, we identified 5 miRNAs (including hsa-miR-640, hsa-miR-578, hsa-miR-557, hsa-miR-1248, and hsa-miR-630) that might bind to circ_0006677 (Figure 3A). Subsequent qRT-PCR assays showed that the expression of miR-578 (but not the remaining miRNAs) was significantly downregulated after overexpression of circ_0006677 in A549 and H1299 cells (Figure 3B). Luciferase reporter assay also confirmed that miR-578 overexpression could decrease the luciferase activity of wild-type circ_0006677 in A549 and H1299 cells. However, the inhibitory effect disappeared after mutation of the predicted miR-578 binding site was generated (Figure 3C). RNA pull-down assays further confirmed that miR-578 could bind directly to the endogenous circ_0006677 (Figure 3D). To examine the role of miR-578 in NSCLC, we first detected the expression of miR-578 in 88 pairs of NSCLC tissues and normal tissues using qRT-PCR assays. Our results demonstrated that the levels of miR-578 were significantly upregulated in NSCLC tissues compared with adjacent normal tissues (Figure 3E). Correlation analysis suggested that the expression of circ_0006677 was negatively correlated with miR-578 expression (Figure 3F). Taken together, our results suggested that circ_0006677 is a direct target of miR-578 in NSCLC cells.

To explore whether miR-578 exhibits a tumor-promoting role in NSCLC cells, we analyzed the expression of miR-578 in NSCLC cell lines and HBE. We validated an increased expression of miR-578 in NSCLC cells compared to HBE (Figure 4A). Relatively higher levels of miR-578 were found in A549 and H1299 cells, where circ_0006677 was expressed at low levels (Figure 1D).

Our qRT-PCR experiments showed that the transfection with miR-578 inhibitor significantly decreased the level of miR-578 in A549 and H1299 cells (Figure 4B). Cellular functional assays demonstrated that downregulation of miR-578 significantly inhibited the proliferation of A549 and H1299 cells (Figure 4C). Similarly, the colony formation, migration, and invasive abilities were also remarkably reduced by miR-578 silencing in A549 and H1299 cells (Figures 4D–F). Furthermore, transfection with miR-578 inhibitor significantly attenuated glucose consumption and lactate production in A549 and H1299 cells (Figures 4G,H). These results suggested that miR-578 promotes malignant properties and glucose metabolism in NSCLC cells.

SOCS2, a member of the SOCS family, participates in a classical negative feedback system to inhibit cytokine signal transduction (Ricobautista et al., 2006; Das et al., 2017; Chhabra et al., 2018; Tong et al., 2019). Using the TargetScan database, we identified a binding site of miR-578 in the 3′-untranslated region (3′-UTR) of SOCS2 mRNA. Thus, we explored whether the tumor-promoting role of miR-578 was mediated by the suppression of SOSC2. Overexpression of miR-578 significantly decreased the luciferase reporter activity of wild-type SOCS2 3′-UTR in A549 and H1299 cells (Figure 5A). However, these effects were abolished by mutations in the putative miR-578-binding site within the SOCS2 3′-UTR sequence (Figure 5A).

Western blot analysis demonstrated the inhibitory effects of miR-578 on SOCS2 protein expression in A549 and H1299 cells (Figure 5B). In addition, overexpression of circ_0006677 significantly increased the protein expression level of SOCS2 in A549 and H1299 cells, while co-transfection with miR-578 mimic reversed the effects of circ_0006677 overexpression on SOCS2 expression (Figure 5C). These findings suggested that circ_0006677 upregulates SOCS2 expression in NSCLC cells through sponging miR-578.

To further explore whether circ_0006677 regulates NSCLC progression via the miR-578/SOCS2 axis, we conducted rescue experiments by transfecting NSCLC cells with the circ_0006677 expression vector, along with miR-578 mimic (or SOCS2 siRNA). As expected, overexpression of circ_0006677 significantly inhibited the proliferation, colony formation, migration, and invasion of A549 and H1299 cells; however, either miR-578 overexpression or knockdown of SOCS2, could partly restore these phenotypes of A549 and H1299 cells (Figures 6A–E). Similarly, either miR-578 overexpression or SOCS2 silencing reversed the inhibitory effects of circ_0006677 overexpression on glucose consumption and lactate production (Figures 6F,G). Collectively, our finding provided the first evidence that circ_0006677 regulates SOCS2 expression through miR-578, thereby inhibiting the progression and glucose metabolism of NSCLC.

To investigate the role of circ_0006677 in vivo, we established the xenograft mouse models by subcutaneously inoculating the stable A549 cells overexpressing circ_0006677 or control cells into nude mice. The volume and weight of tumors in mice bearing A549 cells overexpressing circ_0006677 were markedly lower than those in mice bearing control A549 cells (Figures 7A,B). The qRT-PCR analysis validated that the levels of circ_0006677 were significantly higher, whereas miR-578 expression was significantly lower in circ_0006677–overexpressing tumors than in the control group (Figure 7C). Immunohistochemistry staining revealed that Ki-67 expression was decreased, while the SOCS2 level was increased in the circ_0006677–overexpressing group compared with the control group (Figure 7D), confirming the inhibitory effects of circ_0006677 on NSCLC growth in vivo.