A total of 60 NSCLC (38 males and 22 females; age, 49–83 years; mean age, 61.7 years) and paired paratumor tissue specimens (2 cm from NSCLC tissue) were collected from patients following surgical resection between January 2015 and April 2018 at Liaoning Cancer Hospital and Institute (Shenyang, China). All patients provided written informed consent, and the study was performed in accordance with the Declaration of Helsinki. Approval of the study was granted by the Institute Research Medical Ethics Committee of Liaoning Cancer Hospital & Institute. All patients were histologically diagnosed by two pathologists, and classified as stage I, II, III or IV, using the Tumor-Node-Metastasis (TNM) staging system (version 8) (16).

The human NSCLC (A549, H460, H1299 and H1975) and normal bronchial epithelial cell line 16HBE were purchased from the Institute of Biochemistry and Cell Biology of the Chinese Academy of Sciences. The 16HBE cells were cultured in Airway Epithelial Cell Basal Medium (American Type Culture Collection); NSCLC cell lines were cultured in Dulbecco's modified Eagle medium (Gibco; Thermo Fisher Scientific, Inc.). All media were supplemented with 10% FBS (Thermo Fisher Scientific, Inc.). All cells were incubated at a constant temperature of 37°C in a humidified atmosphere containing 5% CO₂.

CircPrimer (17) was used to compare the nucleotide sequences of circTADA2A and TADA2A mRNA. The clinical value of miR-632-targeted genes in The Cancer Genome Atlas was analyzed using online software UALCAN (18). Data on differentially expressed circRNAs (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE101586), miRNAs (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE19945 and ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE74190) and mRNAs in NSCLC were downloaded from the GEO database (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE19188 and ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE30219). Differentially expressed circRNAs, miRNAs and mRNAs were analyzed using the online software GEO2R (19).

In order to detect the stability of circTADA2A, RNase R assay was performed as previously described (15). A total of 5 U/µg RNase R (Epicentre; Illumina, Inc.) was added to 2 mg total RNA (isolated from A549 or H1299 cells) and incubated for 30 min at 37°C. Controls were untreated. Subsequently, RNeasy MinElute Cleaning kit (Qiagen, Inc.) was applied to purify the RNAs according to the manufacturer's instructions. Next, reverse transcription-quantitative (RT-q)PCR was performed to detect circRNAs or mRNAs.

Actinomycin D assay was performed as previously reported (20). Briefly, 2 µg/ml actinomycin D (Sigma-Aldrich; Merck KGaA) was added to A549 and H1299 cells and incubated at 37°C for 4, 8, 12 and 24 h. Next, total RNA from A549 or H1299 cells was extracted, and the levels of circTADA2A and linear TADA2A mRNA were measured by RT-qPCR.

RNA extraction and RT-qPCR were performed as previously described (21). Total RNA from CRC tissue and cells was extracted using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. To qualify the level of circTADA2A, RNase R (3 U/1,000 ng; Epicentre; Illumina, Inc.) was added to digest linear RNAs during RNA extraction and incubated for 10 min at 37°C. A total of 500 ng RNA was reverse transcribed into cDNA using a Goscript™ Reverse Transcription system (Promega Corporation) according to the manufacturer's instructions. Stem-loop RT was performed for miR-638 using the TaqMan MicroRNA detection kit (Thermo Fisher Scientific, Inc.). Then, qPCR was performed using SYBR Premix Ex Taq II (Takara Bio, Inc.). β-actin and U6 were employed as endogenous controls for mRNA and miR-637, respectively. PCR amplification conditions were as follows: 2 min at 95°C for one cycle, followed by denaturation for 15 sec at 95°C and extension for 60 sec at 60°C for 38 cycles. The relative expression levels of the genes were determined using the 2^−ΔΔCq method (16). The primer sequences are listed in Table I.

Western blotting was performed as previously reported (21). Total protein from A549 and H1299 cells was extracted using radio immunoprecipitation assay lysis buffer (Sigma-Aldrich; Merck KGaA) and quantified by BCA kit (Santa Cruz Biotechnology, Inc.). Total proteins (20 µg) were separated by 10% SDS-PAGE and transferred onto PVDF membranes (EMD Millipore), which was blocked with 5% BSA (Sigma-Aldrich; Merck KGaA) for 1 h at room temperature. Primary anti bodies against KIAA0101 (1:1,000; cat.no. ab226255; Abcam) and GAPDH (1:500; cat.no. ab8245; Abcam) were then added to the membranes and incubated at 4°C overnight. Next, a secondary antibody [horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG; 1:2,000; cat. no. ab205719; Abcam] was added to the membranes and incubated at 24°C for 1 h. Signals of the targeted proteins were detected using an ECL Western Blotting Substrate kit (cat. no. ab65623; Abcam) and bands were analyzed with Image J software version 2 (National Institutes of Health).

Specific small interfering (si)RNAs targeting circTADA2A and KIAA0101 and scramble control siRNAs (siSCRs) were purchased from Shanghai GenePharma Co., Ltd. miR-638 mimics and miR-negative control (NC), as well as miR-638 inhibitor (inh) and inh-NC were purchased from Shanghai GenePharma Co., Ltd. circTADA2A overexpression plasmids containing miR-638 binding sequences (oecircTADA2A) were also synthesized by Shanghai GenePharma Co., Ltd. When cells reached 80–90% confluence, 5 nm miR-638 mimics/inhibitors and 50 nM siRNAs were transfected into A549 and H1299 cells using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions at room temperature. Subsequent experiments were performed 24 h after transfection. Sequences of siRNAs, miR-638 mimics, miR-NC, miR-638 inhibitor and inh-NC are listed in Table I.

CCK-8 assay was performed as previously described (20). In brief, 2×10³ A549 and H1299 cells were seeded into 96-well plates, and 200 µl culture medium was added to each well. The cells were then incubated at 37°C with 5% CO₂. Following incubation for 1–5 days, each well was supplemented with 20 µl CCK-8 solution (Dojindo Molecular Technologies, Inc.). Following 2 h incubation, a microplate reader (Bio-Rad Laboratories, Inc.) was used to measure the absorbance at 450 nm. The assays were repeated three times.

Transwell assay was conducted as previously reported (22). For invasion assays, upper chambers (Corning, Inc.) were pre-coated with Matrigel (1:8; BD Biosciences) for 1 h at room temperature. A549 and H1299 cells (5×10⁴) were incubated in the upper chambers. Serum-free DMEM was added to the upper chamber; medium in the lower chamber was filled with DMEM supplemented with 10% FBS and the cells were incubated at 37°C for 12 h. After 12 h, non-migratory cells were removed and while migratory or invasive A549 and H1299 cells were fixed with 4% paraformaldehyde at room temperature for 30 min, stained with crystal violet at 25°C for 1 min and counted using an inverted microscope (Olympus Corporation).

The assay was performed as previously described (20). Specific probes targeting circTADA2A or miR-638 were synthesized by Shanghai GenePharma Co., Ltd. The assay was conducted using a FISH kit (Shanghai GenePharma Co., Ltd.) according to the manufacturer's instructions. Cy3-labeled circTADA2A and Dig-labeled locked nucleic acid miR-638 probes (Shanghai GenePharma Co., Ltd.) were measured using the aforementioned FISH kit, followed by visualization with a confocal microscope (Carl Zeiss AG).

IHC assay was performed using a specific HRP/DAB (ABC) Detection IHC kit (Abcam) as previously described (23). NSCLC tissues were sliced following 4% paraformaldehyde fixation for 4 h at room temperature, 10% EDTA (pH, 7.3) decalcification, dehydration and paraffin embedding. Then, the tissue sections (4-µm thick) were incubated with anti-KIAA0101 antibody (1:500; cat.no. ab226255; Abcam) at 4°C overnight. The tissue sections were incubated with secondary antibodies (1:1,000; cat. no. E043201; Dako; Agilent Technologies, Inc.) at 37°C for 30 min. Subsequently, the tissue sections were incubated with streptavidin HRP for another 30 min at room temperature (Labeled Streptavidin Biotin kit; cat. no. K060911-8; Dako; Agilent Technologies, Inc) at room temperature and stained with 0.05% 3,3-diaminobenzidine for 60 sec at room temperature. Finally, the slides were counterstained with 10% hematoxylin for 3 min at room temperature, dehydrated in a graded ethanol series (absolute ethyl alcohol for 3 min, 95% ethanol for 3 min and 85% ethanol for 3 min), then observed under a light microscope.

Reporter plasmids containing wild-type (wt) or mutant (mut) circTADA2A sequences (luc-circTADA2A-wt and luc-circTADA2A-mut), as well as reporter plasmids containing wt or mut KIAA0101 sequences (luc-KIAA0101-wt and luc-KIAA0101-mut) were synthesized by Shanghai GenePharma Co., Ltd. The procedures were performed as previously described (15). When A549 and H1299 cells reached 70% confluence, reporter plasmids and miR-638 mimics or miR-NC were co-transfected into A549 and H1299 cells using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h, luciferase activity was measured and compared with Renilla luciferase activity using the Dual-Luciferase Reporter Assay system (Promega Corporation) according to the manufacturer's protocol.

RIP assay was performed as previously reported (24) using a Magna RIP RNA-Binding Protein Immunoprecipitation kit (EMD Millipore). Argonaute (Ago)2 plasmid or vector were transfected into A549 and H1299 cells. Subsequently, RIP Lysis Buffer combined with a protease inhibitor cocktail and RNase inhibitors was used to lyse A549 and H1299 cells. Cell lysates (200 µl) were incubated with 5 µg antibody against Ago2 (EMD Millipore) or rabbit IgG-coated beads at 4°C overnight with rotation. RNeasy MinElute Cleanup kit (Qiagen, Inc.) was used to extract the immunoprecipitated RNA and extracted RNAs were reverse transcribed and subjected to qPCR to detect the abundance of circTADA2A.

The targeted binding effect between circTADA2A and miR-638 was verified by RNA pull-down assay as previously reported (25). Biotinylated miR-638 probe was purchased from Shanghai GenePharma Co., Ltd. A549 and H1299 cells (1×10⁶) were harvested, lysed by 200 µl Pierce^® IP lysis buffer (Thermo Fisher Scientific, Inc.) and sonicated. The pull-down assay was performed using a Pierce™ Magnetic RNA-Protein Pull-Down kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. In brief, 50 µl C-1 magnetic beads (Thermo Fisher Scientific, Inc.) were co-incubated with 50 pmol biotinylated miR-638 or NC probes at 25°C for 30 min to generate probe-coated beads. The beads were harvested by a magnetic stand (Thermo Fisher Scientific, Inc.), then incubated with cell lysate at 4°C for 60 min. The beads were harvested using a magnetic separation stand (Thermo Fisher Scientific, Inc.). The pulled down RNA complexes were eluted and extracted using a RNeasy Mini kit (Qiagen, Inc.), Pulled down circTADA2A and circular nucleolar protein 10 (NOL10) were determined by RT-qPCR, as aforementioned.

All data are presented as the mean ± SD (n=3). Statistical analysis was performed with GraphPad Prism v5.0 (GraphPad Software, Inc.) and SPSS 19.0 statistical software (IBM Corp.). Survival analysis was performed by Kaplan-Meier analysis using log-rank test. The clinical value of circTADA2A was measured by receiver operating characteristic (ROC) curve analysis. The association between circTADA2A and miR-638 was analyzed by Spearman correlation analysis. For comparison of two groups, χ² or paired Student's t test were performed. For the comparison of multiple groups, one-way ANOVA followed by post hoc Dunnett's or Tukey's multiple comparisons test were performed, as appropriate. P<0.05 was considered to indicate a statistically significant difference.

By analyzing the profile of circRNA expression in five lung adenocarcinoma and paired normal lung tissue samples from a genome-wide study (GSE101586), it was found that 22 upregulated circRNAs and 17 downregulated circRNAs were significantly differentially expressed in lung cancer compared with normal tissue (cut-off criteria: P<0.05 and |logFC|≥1.2; Fig. 1A; Table SI). circTADA2A (probe ID: ASCRP002384; circBase ID: hsa_circ_0043278) was selected for further study since it was the most significantly upregulated circRNA (Fig. 1B; Table SI). The expression of circTADA2A was determined in 60 NSCLC tissue specimens and cell lines. circTADA2A was upregulated in NSCLC tissue specimens and cell lines (Fig. 1C and D). Additionally, it was uncovered that the changes of circTADA2A in A549 and H1299 were more significantly than in H460 and H1975 cells. A549 and H1299 were selected for use in subsequent experiments. In comparison with its parental gene, circTADA2A was found to form a covalently closed continuous loop, and comprised exons 5 and 6 of the TADA2A gene, using the online software CircPrimer (17) (Fig. 1E). RNase R and actinomycin D assays were performed to detect the stability of circTADA2A. circTADA2A was more stable than its parental TADA2A mRNA following treatment with RNase R and actinomycin D (Fig. 1F-I).

In order to determine the clinical role of circTADA2A, the association between circTADA2A expression levels and clinicopathological characteristics of patients with NSCLC was analyzed. Higher levels of circTADA2A were more frequently observed in patients with lymph node metastasis and advanced stage disease (Fig. 2A and B). Furthermore, high expression of circTADA2A was associated with clinicopathological characteristics of patients with NSCLC, particularly TNM stage and lymph node metastasis. (Table II). Kaplan-Meier analysis indicated that patients with NSCLC who had a higher circTADA2A level had poorer overall survival (Fig. 2C). circTADA2A may be a valuable biomarker in NSCLC according to the results of ROC curve analysis (area under the curve=0.8606; Fig. 2D).

In order to evaluate the role of circTADA2A on proliferation and migration of NSCLC cells, siRNAs targeting the back-splice sites were used to knock down expression of circTADA2A in A549 and H1299 cells. circTADA2A was significantly knocked down, while TADA2A mRNA was not changed following transfection with circTADA2A siRNAs (Fig. 3A and B). sicircTADA2A-1 was selected as the silencing tool for subsequent experiments as it exhibited the greatest efficacy. Next, the proliferation, migration and invasion of A549 and H1299 cells were detected following knockdown of circTADA2A. The results indicated a significant delay in the proliferation, migration and invasion of A549 and H1299 following knockdown of circTADA2A compared with cells transfected with siSCR (Fig. 3C-F).

circRNAs have been reported to exert their functions by acting as miRNAs sponges in various cancer types (26). It was hypothesized that circTADA2A functions via a similar mechanism in NSCLC. circTADA2A was primarily located in the cytoplasm of A549 and H1299 cells (Fig. 2E and F). RIP assay showed that circTADA2A was significantly enriched in Ago2 protein (Fig. 4A). Online prediction software Circular RNA Interactome (27), was used to predict potential miRNAs that may interact with circTADA2A. A total of 8 potential miRNAs that could bind to circTADA2A were found (Table SII). Among them, only miR-638 was expressed at low levels in NSCLC tissue according to genome-wide studies using GSE74190 and GSE19945 datasets (Fig. 4B-D, and Tables SIII and SIV). Furthermore, miR-638 was expressed at low levels in NSCLC tissue and cells (Fig. 4E and F). Additionally, an inverse correlation between circTADA2A and miR-638 was observed (Fig. 4G). RNA FISH assay indicated that circTADA2A and miR-638 co-localized in the cytoplasm (Fig. 4H). A luciferase reporter assay was conducted to confirm the targeted binding effect between circTADA2A and miR-638. Compared with that of miR-NC, miR-638 mimic significantly decreased luciferase reporter activity (Fig. 4I and J). When the predictive target sites for miR-638 were mutated, the change in luciferase activity was insignificant. Ago RISC catalytic component 2 (AGO2) is a key RNA binding protein for the ceRNA theory of RNA-RNA interactions among miRNAs, long non-coding RNAs and circRNAs (28–30). In order to determine the mechanism of circTADA2A, RIP assay was performed, using IgG as a control. The enriched circTADA2A and circNOL10 [a circRNA reported not to bind to Ago2 (31)] were analyzed. The results indicated that, compared with the NC probe, circTADA2A but not circNOL10 was pulled down by the miR-638 probe (Fig. 4K and L). Lastly, it was observed that circTADA2A was not affected by up- or downregulation of miR-638. Similarly, the level of miR-638 was not affected either by up- or downregulation of circTADA2A (Fig. 4M and N).

In order to identify the downstream genes of miR-638, the online prediction results of TargetScan (32) and upregulated genes according to two genome-wide studies, GSE19188 and GSE30219 (cut-off criteria: P<0.05 and logFC value ≥2), were combined. It was found that four mRNAs, matrix metalloproteinase 1, BUB1 mitotic checkpoint serine/threonine kinase, KIAA0101 and steroid 5 α-reductase 1, overlapped (Fig. 5A; Table SV,Table VI,Table SVII). KIAA0101 has been reported to be implicated in NSCLC proliferation and metastasis (8). Among the aforementioned four genes, KIAA0101 was more closely correlated with shorter OS rate in lung cancer according to Kaplan-Meier analysis using the online software UALCAN (18) (Fig. S1). Therefore, KIAA0101 was selected for subsequent experiments. KIAA0101 was upregulated in NSCLC according to the results of IHC staining and genome-wide studies using GSE19188 and GSE30219 datasets (Fig. 5B-D). Furthermore, up- and downregulation of miR-638 inversely regulated KIAA0101 protein expression (Fig. 5E and F). Functionally, upregulation of miR-638 suppressed A549 and H1299 cell proliferation and metastasis. When the level of KIAA0101 was elevated (following co-transfection of miR-638 mimic and specific KIAA0101 overexpression plasmids), the suppressive effect of miR-638 was abolished (Fig. 5G-J). Lastly, it was verified that KIAA0101 was a direct target of miR-638 via six base pair binding sites (positions 1,473-1,479; Fig. 5K-M).

The present study investigated whether the oncogenic role of circTADA2A in A549 and H1299 cell proliferation and migration was achieved, at least partially, via the miR-638/KIAA0101 signaling pathway. The protein expression of KIAA0101 following overexpression or knockdown of circTADA2A was analyzed by western blotting. Overexpression of circTADA2A positively regulated KIAA0101 protein expression and vice versa (Fig. 6A and B). circTADA2A sponged miR-638, and KIAA0101 was a downstream target of miR-638. The present study investigated whether circTADA2A regulated KIAA0101 via miR-638 sponging. circTADA2A promoted KIAA0101 protein expression; this effect was abolished by miR-638 mimics (Fig. S2A and B). CCK-8 assay demonstrated that overexpression of circTADA2A promoted the proliferation of A549 and H1299 cells; this effect was abrogated by miR-638 mimics (upregulation of miR-638; Fig. 6C and D). Similarly, overexpression of circTADA2A promoted migration and invasion of A549 and H1299 cells but this effect was attenuated by miR-638 mimics (Fig. 6E and F). Overall, the present study indicated that circTADA2A promoted cell proliferation and migration, at least partially, via the miR-638/KIAA0101 pathway in A549 and H1299 cells.