2.1 Tissue samples

Thirty NSCLC tissues and 30 matched para-cancerous tissues were collected from NSCLC cases subjected to excision in The Affiliated Huai'an Hospital of Xuzhou Medical University and The Second People's Hospital of Huai'an. Immediately after collection, liquid nitrogen was used to freeze all the samples prior to the extraction of total RNA. All participants signed the informed consent for scientific investigations. This research was conducted in line with the Helsinki Declaration, and it was monitored and approved by the Ethics Committee of The Affiliated Huai'an Hospital of Xuzhou Medical University and The Second People's Hospital of Huai'an.

2.2 Cell lines and transfection

The Cell Bank of Chinese Academy of Sciences provided five human NSCLC cell lines (CAL-12T, HCC827, A549, H460 and H1299) and one human bronchial epithelioid cell line (16HBE). They were cultured in RPMI 1640 medium (HyClone) with 1% penicillin/streptomycin (Gibco) and 10% foetal bovine serum (FBS; Pan Life Sciences) and maintained under a humid circumstance with 5% CO₂ at 37°C.

Subsequently, 1 × 10⁶ recombinant lentivirus-transducing units and 6 μg/mL Polybrene (Sigma) were utilized to infect cells in the case of the fusion of 60%-80%. Treatment with 2 μg/mL puromycin was adopted for the selection of cells with 2 weeks of stable transfection, and these cells were chosen through flow cytometry for later experiments. GenePharma provided lentiviruses and plasmids applied in this research.

2.3 Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

In line with the regimen of the manufacturer, the total RNA extraction from tissues and cells was performed using TRIzol reagent (Invitrogen), followed by qRT-PCR mentioned above.²¹ Specific primers used in this research were shown below:

GAPDH: forward: 5′-TGACTTCAACAGCGACACCCA-3′,

reverse: 5′-CACCCTGTTGCTGTAGCCAAA-3′;

miR-488: forward: 5′-ACACTCCAGCTGGGTTGAAAGGCTATTTC-3′,

reverse: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAG

TTGAGGACCAAGA -3′

circ-0008003: forward: 5′-TGGGGCTAACAAACTTCACC-3′,

reverse: 5′-GTCCCACCAGAGGTTGTCAC-3′.

2.4 Cell proliferation assay

Based on the regimen of the manufacturer, cell proliferation was examined via Cell Counting Kit-8 (CCK8) (Beyotime) assay. Following 24 hours of culture in a 96-well plate, the cells were subjected to 1 hour of incubation with CCK8 in an incubator for cell culture. A TECAN infinite M200 Multimode microplate reader (TECAN) was employed to test the cell optical density at 450 nm.

For EDU assay, EDU labelled solution (KeyGen Biotech) was used. Nucleus staining was performed by adding DAPI.

2.5 Cell invasion assay

By reference to the regimen of the manufacturer, the invasion capacity of cells was assessed by Transwell assay. Transwell chambers (Millipore Corporation) pre-coated with matrigel were then used. During invasion assessment, the upper chamber was inoculated with 2 × 10⁵ cells in medium with no serum, whereas the lower chamber was added with 600 μL of medium with 10% FBS as a chemoattractant. Following 24 hours of incubation, cells subjected to migration were fixed and dyed with a 5% crystal violet solution (Beyotime), and the images of cells undergoing migration were captured by an optical microscope (Olympus Corporation). Moreover, Image-pro Plus 6.0 (Media Cybernetics) was utilized to count the migrated cells.

2.6 Luciferase reporter assay

Wild-type plasmids circ-0008003-WT and ZNF281-WT containing binding sites for miR-488 and mutant plasmids circ-0008003-MUT and ZNF281-MUT were consolidated into the pGL3 promoter vector (GenePharma). After the seeding of NSCLC cells into a 24-well plate, the cells underwent cotransfection with 80 ng plasmids + 5 ng pRL-SV40 and 50 nmol/L miR-488 mimics or negative control (NC) with the help of Lipofectamine 2000 reagent. After that, the luciferase activity was assessed using the dual-luciferase reporter assay kit (Promega). All experiments were carried out thrice, respectively.

2.7 RNA reduction assay

RNAs were labelled with SP6/T7 RNA polymerase (Roche) and the Biotin RNA labelling mix (Roche Diagnostics) in vitro and treated with RNase-free DNase I (Roche). Then, RNeasy Mini Kit (Qiagen) was utilized for the purification of RNAs labelled with biotin (Bio-miR-488-Mut, Bio-miR-488-WT and Bio-NC). Thereafter, the mixed solution (50 pmol of biotinylated RNA and 1 mg NSCLC cell) was added with washed beads (6 mL) containing streptavidin agarose (Life Technologies), and 1 hour of incubation was carried out at room temperature. After the rinsing of beads, the amplification of eluted RNAs was conducted, followed by measurement of them via qRT-PCR.

2.8 RNA immunoprecipitation (RIP) assay

By reference to the guidance of the manufacturer, RIP assay was carried out with the use of the Magna RIP kit (Millipore). Specifically, lysis buffer (Gibco) containing RNase inhibitor was used to lyse H460 and H1299 cells. After that, protein A/G agarose beads with AGO2 antibody were added to incubate cell lysates, with normal rabbit IgG as NC. Then, the RNAs that were co-precipitated underwent washing, purification and examination using qRT-PCR.

2.9 RNA-Fluorescence In Situ Hybridization (RNA-FISH)

RNA-FISH was conducted using circ-0008003 probes conjugated with fluorescence by reference to the former research.²² Then, DNA probe sets (Biosearch Technologies) were employed for hybridization based on the regimen of Biosearch Technologies. Ultimately, a FV1000 confocal laser microscope (Olympus) was utilized for probing cells.

2.10 In vivo assays

Shanghai SLAC Laboratory Animal Co. Ltd. provided female nude mice aged 4 weeks, which were randomly allocated into two groups: 1 × 10⁷ circ-0008003 shRNA H460 cells in 100 μL of medium with no serum were injected at the left flank in one group (n = 4), while control cells treated with NC were injected in the other group (n = 4). Afterwards, a calliper was applied to test and calculate the tumour size every other day on the basis of the formula: size = length×width²/2. Prior to tumour excision and photography, the mice were killed after 3 weeks. Then immunohistochemistry (IHC) staining was carried out with the fixed tumours. Based on the Care and Use of Laboratory Animals of the National Institutes of Health, assays were conducted.

2.11 Immunohistochemistry staining

Ki-67 antibody and PCNA antibody (Abcam) were used for IHC staining by reference to the regimen mentioned above.²³

2.12 Statistical analysis

Experimental data (obtained from experiments conducted thrice) were expressed as the mean ± standard deviation (SD). Intergroup comparison was carried out using the Student's t test, and Kaplan-Meier (K-M) curves were plotted for the comparison of overall survival via log-rank test. To figure out the correlation between clinicopathological characteristics and circ-0008003 expression, Fisher's exact test or chi-square test was employed. Besides, miR-488's association with circ-0008003/ZNF281 expression was determined by Pearson's correlation analysis. P < .05 meant a statistical significance.

3.1 Abnormal circ-0008003 expression was observed in NSCLC tissues and cells

GSE112214 microarray analysis demonstrated that circ-0008003 displayed a high expression in NSCLC tissues. On the basis of the criteria of P < .05 and fold change >4, the different expression of circRNAs including circ-0008003 was confirmed in NSCLC tissues (Figure 1A). To verify the role of circ-0008003 in NSCLC, qRT-PCR was adopted to test its expression in NSCLC tissues and matched non-tumour tissues. According to the findings, NSCLC tissues had remarkably raised circ-0008003 expression relative to non-tumour tissues (Figure 1B). And circ-0008003 level in TNM stage Ⅰ-Ⅱ was evidently lower than that in TNM stage Ⅲ-Ⅳ (Figure 1C). Besides, tissues undergoing lymphatic metastasis had higher circ-0008003 expression than those undergoing no lymphatic metastasis (Figure 1D). Additionally, circ-0008003 relative expression in NSCLC cells was evidently raised relative to human bronchial epithelioid cell line 16HBE (Figure 1E). Because of the highest or lowest circ-0008003 expressions in H460 and H1299 cells, they were picked for later assays.^{24, 25} Circ-0008003 has resistance to RNase R digestion, so it was considered to possess circRNA features (Figure 1F). These results suggest the elevated aberrant expression of circ-0008003 and its role in NSCLC as an oncogene.

3.2 The promotive effect of circ-0008003 on NSCLC cellular processes

Because of the confirmed increase in circ-0008003 expression in NSCLC, its influences in NSCLC cell biological processes were tested by silencing or raising circ-0008003 expressions in H460 and H1299 cells. QRT-PCR findings illustrated that circ-0008003 was remarkably pulled down or raised in NSCLC cells after the addition of circ-0008003 shRNA or overexpressing vector (Figure 2A). According to CCK8 and 5-Ethynyl-2'-deoxyuridine (EDU) assays, the proliferation ability of NSCLC cells was decreased after silencing circ-0008003 and increased after up-regulating circ-0008003 (Figure 2B-C). Besides, the invasion abilities of the two cell lines became weak after silencing circ-0008003 and strong after up-regulating circ-0008003 unfolded by Transwell assay (Figure 2D). These data reveal that circ-0008003 promotes NSCLC cell invasion and proliferation in vitro.

3.3 Tumour growth in vivo was slowed down by reduced circ-0008003 expression

H460 cells treated with circ-0008003 shRNA or NC were used for in vivo experiments to continuously prove the effect of circ-0008003 in tumour formation of NSCLC. In comparison with those in NC group, repressed circ-0008003 resulted in relatively small xenografts, identical to the results in vitro (Figure 3A). Moreover, relative to those in control group, tumours derived from H460 cells with repressed circ-0008003 had limited weight and size (Figure 3B,C). And IHC staining was adopted to examine the expressions of Ki67 and PCNA, the markers of cell proliferation.^{26, 27} Consequently, there were substantially less stained PCNA and Ki67 in tumours in circ-0008003-repressed cell group relative to those in NC-treated H460 cell group (Figure 3D). It can be deduced that circ-0008003 repression inhibits tumour formation in vivo.

3.4 Circ-0008003 was considered to be an endogenous sponge for miR-488 in NSCLC

CircRNAs expressed in the cytoplasm have been demonstrated to substantially affect pathological and physiological processes by sponging miRNAs to some extent.^{28, 29} To ascertain the cellular localization of circ-0008003, cells were isolated into the nucleus and cytoplasm fractions, in which the controls, U6 and GAPDH, could be seen. QRT-PCR validated that circ-0008003 was monitored mainly in the cytoplasm fractions in NSCLC cells (Figure S1A). Identically, a substantial staining of circ-0008003 in the cytoplasm of H1299 cells was also determined in RNA-FISH (Figure S1B). It can be seen that circ-0008003 probably functions in NSCLC at the post-transcriptional level. To forecast circ-0008003 miRNAs with underlying interaction, the online bioinformatics tool Circinteractome was employed. As a result, circ-0008003 was unveiled to bind to miR-488 with the highest potential, and the sequences of wild-type circ-0008003 (with forecasted binding sites), mutant circ-0008003 (with mutant sites) and miR-488 were displayed in Figure 4A. Subsequently, circ-0008003's interaction with miR-488 primarily appeared in the cytoplasm rather than the nucleus of H460 and H1299 cells (Figure S1C,D), thus validating the above forecasting.

Then, luciferase reporter assay was conducted to ascertain the above-mentioned interaction. It was illustrated that the luciferase activity of circ-0008003-WT could be repressed by miR-488 mimics, but that of circ-0008003-MUT remained unchanged (Figure 4B). According to the RNA reduction experiment, circ-0008003 only gathered in complexes repressed not by Bio-miR-488-Mut or Bio-NC but by Bio-miR-488-WT (Figure 4C). And repressing circ-0008003 expression led to the elevated level of miR-488 in H460 and H1299 cells (Figure 4D). Further, miR-488 expression in NSCLC cells and tissues was investigated through qRT-PCR, the results of which ascertained that miR-488 was markedly pulled down in NSCLC cells and tissues relative to the controls (Figure 4E,F). Eventually, the inverse relationship between the level of miR-488 and circ-0008003 expression was uncovered by Pearson's correlation analysis in NSCLC patients (Figure 4G). These findings uncover that circ-0008003 may promote NSCLC progression through the inverse adjustment of miR-488 expression.

3.5 Aberrant miR-488 expression slowed down NSCLC progression

Figure S2A illustrated that the expression of miR-488 was greatly inhibited or enhanced in NSCLC cells treated with miR-488 inhibitor or mimics relative to that in cells treated with NC. Besides, miR-488 impeded cell invasion and proliferation (Figure S2B-D). Collectively, the results indicate that miR-488 influences NSCLC evolution as a tumour suppressor.

3.6 Circ-0008003 accelerated NSCLC development by repressing miR-488

To ascertain that the role of circ-0008003 required miR-488 reduction in a direct manner, miR-488 inhibitor was transfected into circ-0008003-repressed H460 and H1299 cells, and the observably reduced miR-488 was discovered by qRT-PCR (Figure 5A). A foregone conclusion was obtained from CCK8 and EDU assays that the proliferation capacities of circ-0008003-repressed H460 and H1299 cells were promoted by miR-488 suppression (Figure 5B,C). Lastly, silencing miR-488 was proved by Transwell assay to evidently boost the invasion capacities of circ-0008003-repressed H460 and H1299 cells (Figure 5D). The above discovery implies that miR-488 participates in the carcinogenic process regulated by circ-0008003 in NSCLC cells.

3.7 ZNF281 was a direct target of miR-488

Next, the specific potential mechanism of circ-0008003/miR-488 axis in stimulating NSCLC evolution was discussed. With bioinformatics prediction tools (TargetScan, miRDB and Starbase), ZNF281, a carcinogene in cancers, was found to be a target of miR-488 (Figure 6A). Besides, the direct binding of miR-488 to ZNF281 3'UTR at the forecasting sites was ascertained in NSCLC cells (Figure 6B). From immunoprecipitate conjugated with Ago2 in H460 and H1299 cells, miR-488 and ZNF281 mRNA were obtained, validating the existence of the interaction between miR-488 and ZNF281 mRNA in RNA-induced silencing complex (RISC) adjusted by miR-488 (Figure 6C). Next, the expression of ZNF281 was disclosed to notably rise in NSCLC cells and tissues relative to controls (Figure 6D-E). What's interesting was that miR-488 level was shown as an inverse correlator while circ-0008003 expression as a positive correlator of ZNF281 expression in NSCLC tissues (Figure 6F,G). As a result, ZNF281 may affect circ-0008003/miR-488 axis in NSCLC evolution.

3.8 ZNF281 positively adjusted circ-0008003 in NSCLC evolution

To figure out the adjustment of ZNF281 on the role of circ-0008003 in NSCLC evolution, the function of ZNF281 overexpression in the biological processes of circ-0008003-repressed H460 cells was analysed. In the first place, it was revealed that mRNA and protein expression levels of ZNF281 were remarkably impeded after the repression of circ-0008003 and became normal after cotransfection with ZNF281 overexpression vector (Figure 7A,B). Afterwards, circ-0008003-repressed H460 cells predominantly regained their repressed proliferation abilities following cotransfection with ZNF281 overexpression vector (Figure 7C,D). Additionally, ZNF281 overexpression alleviated the hampering of circ-0008003-repressed invasion processes in H460 cells (Figure 7E). Basically, circ-0008003 facilitates NSCLC evolution via circ-0008003/miR-488/ZNF281 axis.