TCGA data analysis was performed using the publicly accessible TCGA database (https://tcga-data.nci.nih.gov/tcga/). The mRNA values of Ntn-4 and Qki were collected from the samples of 483 patients with LUAD and 59 control samples, and from the samples of 486 patients with LUSC and 50 control samples, which were subsequently quantified according to TCGA normalization protocol. The mRNA levels of Qki and Ntn-4 that were associated with the survival data for the patients with LUAD and LUSC were also analysed according to TCGA datasets.

For the Ntn-4 and Qki-5 mRNA expression level analysis, a total of 30 patients diagnosed with LUAD, and 30 patients diagnosed with LUSC, at the Fourth Affiliated Hospital of China Medical University were enrolled in the present study. The Ethics Committee of the Fourth Hospital of China Medical University approved the present study (approval no. EC-2018-HX-013), and all the patients involved in this research signed informed consent forms. Tumour tissues, and the corresponding adjacent normal tissues were dissected and stored in liquid nitrogen for periods up to 90 days prior to performing the reverse transcription-quantitative PCR (RT-qPCR) experiments.

The tumour samples from patients with LUAD or LUSC were collected between July, 2018 and August, 2019. For the Ntn-4 and Qki-5 IHC staining of the human LUAD and LUSC tissues, slides were prepared using an ultrathin semiautomatic microtome, deparaffinized in 100% xylene and rehydrated in a graded ethanol series (70, 80, 90, 95 and 100%, Beijing InnoChem Science & Technology Co.,Ltd.). For antigen retrieval, the slides were treated with sodium citrate solution (pH 6.0) (Beijing Solarbio Science & Technology Co., Ltd.) for 6 min at 100°C. After incubating the slides in 1% bovine serum albumin in phosphate-buffered saline (PBS) for 30 min at room temperature, the slides were incubated with anti-Ntn-4 (rabbit, 1:1,000, cat. no. NBP191343, Novus Biologicals, LLC) and anti-Qki-5 (rabbit, 1:1,000, cat. no. AB9904, MilliporeSigma) antibodies overnight at 4°C. After washing the slides several times with PBS, they were incubated with a goat anti-rabbit IgG secondary antibody conjugated to horseradish peroxidase (HRP; 1:300, cat. no. BF03008, Biodragon Immunotech) for 30 min at room temperature. Staining was visualized using a DAB Horseradish Peroxidase Color Development kit (p0203, Beyotime Institute of Biotechnology). An appropriate amount of DAB dyeing solution was then added, and the tissues were fully covered and incubated at room temperature for 30 min away from light. The nuclei were counterstained for 30 sec with haematoxylin at room temperature. (C0107, Beyotime Institute of Biotechnology) For the negative control experiments, sections were incubated with PBS instead of the primary antibodies. Positive staining controls were performed using para-carcinoma tissues of breast cancer and renal cancer. The slides were finally covered with neutral balata and cover glass slips. The qualitative scoring system used for evaluation of the intensity of staining was as follows: Negative (no staining, 0); weakly positive (+); moderately positive (++); and strongly positive (+++). Images of the stained tissues were obtained using a positive fluorescence microscope (BX53/DP73, Olympus Corporation).

The human LUAD cell lines, A549 (cat. no. TCHu150) and H23 (cat. no. SCSP-5002) were purchased from The Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The LUSC cell line, NCI-H2107 (cat. no. CRL-5983_FL), and the human bronchial epithelial cell line, 16HBE (cat. no. PCS-300-010) were purchased from Jennino Biological Technology (Guangzhou, China). The cells were cultured in Gibco^® Roswell Park Memorial Institute (RPMI)-1640 medium (Thermo Fisher Scientific, Inc.) supplemented with 10% (v/v) Gibco^® foetal bovine serum (FBS; Thermo Fisher Scientific, Inc.) and maintained in a humidified atmosphere with 5% CO₂ at 37°C.

To generate the Ntn-4- and Qki-5-overexpressing cell lines (oe-Ntn-4 and oe-Qki-5), the Ntn-4 or Qki-5 coding sequence, respectively, was cloned into the GV492 (11.9 kb) lentiviral vector (Shanghai Genechem Co., Ltd.) to form recombinant plasmids. Subsequently, the recombinant GV492, Helper 1.0 (12.0 kb) and Helper 2.0 (5.8 kb) plasmids were co-transfected into 293T cells (Wuhan Procell Life Technology Co., Ltd.). After 48 or 72 h, the culture supernatant was harvested, concentrated and purified (82,700 g for 2 h at 4°C). The concentrations of lentivirus were measured using Immunostaining Plaque Assay (26), and the data obtained revealed that the titres of LV-Ntn-4 and LV-Qki-5 were 1.5E⁺⁰⁹ and 1.2E⁺⁰⁹ (TU/ml), respectively.

To produce the short hairpin RNAs (shRNAs) of human Ntn-4 and Qki-5, the target genes for Ntn-4 (sh-Ntn-4) (sequence, 5′-CCGGGCGCTATTTGTACTTCTAAAT-3′), Qki-5 (sh-Qki-5) (sequence, 5′-CCGGCTATTAACCCACAGCATTTT-3′) and the negative control (NC) scramble (sh-NC, the scramble sequences of shNtn-4 and shQki-5 are TTCTCCGAACGTGTCACGT and TATTAACCACAATGCGCTTTGCCC, respectively) were inserted into the GV118 vector (7.4 kb) (Shanghai Genechem Co., Ltd.). The recombinant plasmids Helper 1.0 (12.0 kb) and Helper 2.0 (5.8 kb) were again co-transfected into 293T cells. The production and quantification of LV-Ntn-4-RNAi were performed following the protocol described above (26). The titre of LV-Ntn-4-RNAi was determined to be 2.00E⁺⁰⁸ TU/ml. Lentiviral production was performed by Shanghai Genechem Co., Ltd. Finally, the cells (~1×10⁵ cells/ml) were seeded in cell culture dishes and subsequently transfected with the lentivirus, following the manufacturer's instructions.

The transfected A549 and H2107 cells were seeded into 96-well plates at a density of 4×10³ cells/well, and allowed to grow for 2 days. Cell proliferation was determined using a Cell Counting Kit-8 (CCK-8) (Beyotime Institute of Biotechnology) assay. The absorbance at 450 nm was read using a microplate reader (BioTek Instruments, Inc.).

Total RNA was extracted using Invitrogen^® TRIzol™ reagent (Thermo Fisher Scientific, Inc.), and first-strand cDNA was generated using an RT system in a 20 µl reaction mixture [SYBR-Green Master Mix (2X) (No ROX) 10 µl, PCR forward primer (10 µM) 0.4 µl, PCR reverse primer (10 µM) 0.4 µl, DNA 2 µl and ddH₂O 7.2 µl] containing 1 µg total RNA. Aliquots (0.5 µl) of cDNA were amplified using Fast SYBR-Green PCR Master Mix (Vazyme Biotech Co., Ltd.) in each 20 µl reaction. PCRs were run on a Roche Light Cycler 480 II with the following primers: Qki-5 forward, 5′-TCCGAGGCAAAGGCTCAATGAG-3′ and reverse, 5′-GCTCTGTTCTGAGCATCTTCCAC-3′; Ntn-4 forward, 5′-GTACTTTGCGACTAACTGCTCC-3′ and reverse, 5′-TCCAGTGCATGGAAAAGGACT-3′; and GAPDH forward, 5′-GAAGGTGAAGGTCGGAGT-3′ and reverse, 5′-GAAGATGGTGATGGGATTTC-3′. The PCR thermocycling conditions were pre-denaturation for 30 sec at 95°C, denaturation for 20 sec at 95°C, annealing for 15 sec at 58°C, and extension for 15 sec at 72°C. The relative expression values of Qki-5 and Ntn-4 were calculated and normalized to those of GAPDH in each sample using the 2^−ΔΔCq method (27).

Following transfection of the cells for 48 h, the cells were collected, washed and lysed with cell lysis buffer (P0013B, Beyotime Institute of Biotechnology) containing protease inhibitors phenylmethylsulfonyl fluoride (PMSF; ST506-2, Beyotime Institute of Biotechnology), and the protein concentrations were quantified using an enhanced BCA protein assay kit (Biosharp Life Sciences). Protein electrophoresis was performed using 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE; kgb113k, Nanjing KeyGen Biotech Co., Ltd.), and western blotting was performed on a nitrocellulose transfer membrane. After using 5% skimmed milk powder closed 2 h at room temperature, the nitrocellulose filter membranes (66485, Pall Life Sciences) were incubated at 4°C overnight with the following primary antibodies: Anti-β-actin (mouse, 1:5,000; cat. no. 66009-1-Ig, Proteintech Group, Inc.), anti-Ntn-4 (rabbit, 1:1,000; cat. no. NBP191343, Novus Biologicals, LLC), anti-Qki-5 (rabbit, 1:1,000; cat. no. AB9904, MilliporeSigma), anti-E-cadherin (rabbit, 1:5,000; cat. no. 20874-1-AP), anti-N-cadherin (rabbit, 1:5,000; cat. no. 22018-1-AP), anti-vimentin (rabbit, 1:5,000; cat. no. 13066-1-AP) and anti-Snail (rabbit, 1:500; 13099-1-AP) (all from Proteintech Group, Inc.). After washing three times with Tris-buffered saline containing Tween (TBST; cat. no. t917680; Beijing Innochem Technology Co., Ltd), the membranes were incubated with HRP-conjugated secondary antibody (Goat anti-Rabbit, 1:5000; cat. no. A21020; Abbkine Scientific Co., Ltd.) at room temperature for 2 h. The protein bands were scanned using an Odyssey infrared imager (LI-COR Biosciences). The relative grey values of the proteins of interest were analysed using ImageJ software (Version 1.51j8, National Institutes of Health) and normalized against those of β-actin.

The transfected A549 and H2107 cells were grown until they had reached ~90% confluency in a 12-well plate, and a cell wound was created by scratching the cells with a sterile 200-µl pipette tip. After washing the cells with PBS twice, they were cultured in RPMI-1640 medium supplemented with 0.5% FBS for 48 h. Cell images were captured under a microscope (IX73, Olympus Corporation). The gap sizes were measured using ImageJ software (version 1.51J8) and calculated as the percentages after 48 h relative to the gap sizes at 0 h.

The cells were collected and resuspended in RPMI-1640 medium (Thermo Fisher Scientific, Inc.). Aliquots (200 µl) of cell suspension were placed in the upper chamber of a Transwell insert (Corning, Inc.). Subsequently, the Transwell chamber was inserted into a 24-well plate. Following incubation of the cells in RPMI-1640 medium containing 10% (v/v) FBS for 24 h at 37°C, the cells on the upper membrane of the insert were gently removed using a cotton swab, whereas those cells that had migrated to the bottom surface were fixed with cold methanol (Wuhan Servicebio Co.,Ltd) and stained with crystal violet (Wuhan Servicebio Co., Ltd.) for 30 min at room temperature. Finally, the cell numbers on the bottom surface were photographed and quantified using an inverted microscope system (Olympus IX73; Olympus Corporation).

RIP experiments were performed according to the manufacturer's instructions using the BersinBio™ RIP kit (Guangzhou Bersinbio Co., Ltd.). Briefly, the A549 cells were lysed using RIP lysis buffer, and the cell extract was incubated with protein A/G agarose beads (Guangzhou Bersinbio Co., Ltd.) conjugated with either Qki-5 antibody (rabbit, 1:400; cat. no. AB9904, MilliporeSigma) or control IgG (rabbit, 1:4000; cat. no. abs20035, Absin) for 2 h. To purify RNA from the immunoprecipitation, the beads were washed, and the proteins were removed using Proteinase K (0.5 mg/ml). Subsequently, RT-qPCR experiments were performed to analyse the mRNA expression level of Ntn-4.

BALB/c (nu/nu) nude female mice (6 weeks old, weighing 20–23 g, n=20 for one trial) were randomly divided into four groups with 5 mice in each group as follows: The oe-NC, oe-Qki-5 + sh-NC, oe-Qki-5 + sh-Ntn-4 and oe-Qki-5 + oe-Ntn-4. The A549 cells (1×10⁷ cells/ml, 0.2 ml) stably transfected with the vectors overexpression (oe)-Qki-5, oe-Qki-5 and oe-Ntn-4, oe-Qki-5 and sh-Ntn-4, or empty vector were subcutaneously injected into the flanks of mice. During the procedure, the mice were anaesthetised using 4% isoflurane (Shandong Keyuan Pharmaceutical Co.; in oxygen), with the subsequent maintenance of anaesthesia with 2.5% isoflurane. Tumour diameters were measured using callipers, and volumes were calculated based on the formula V= (a × b²)/2, where a and b represent the longest and the shortest diameter of the tumour, respectively. To minimize animal suffering, the mice were housed under standard conditions and cared for (23°C; 50% humidity; 12:12 light/dark cycle; food and water were available ad libitum) according to institutional guidelines. At the end of the experiment, the animals were euthanized with 5% isoflurane inhalation for 5–10 min; the vital signs were checked to ensure their disappearance, and cervical dislocation was finally performed to avoid the possibility of recovery. Finally, the tumours were isolated, weighed and photographed. The animal experiments were approved by the Animal Research Ethics Committee of the Animal Centre of Shengjing Hospital affiliated with China Medical University (No. KT201945).

Statistical analyses were performed using GraphPad Prism version 8.0.0 for Windows by observers blinded to the experimental design. The data are presented as the mean ± standard error of the mean (SEM). The data between two groups were compared using an unpaired ttest, whereas oneway analysis of variance (ANOVA) followed by the post-hoc Tukey's Student Range (HSD) test was adopted for data comparisons among multiple groups. The correlation between the mRNA expression levels of Qki-5 and Ntn-4 was analysed using Pearson's correlation (R² correlation). The optical density values and tumour volume at different time points were compared by twofactor ANOVA. A value of P<0.05 was considered to indicate a statistically significant difference.

The NSCLCassociated expression dataset was retrieved from TCGA database (https://portal.gdc.cancer.gov/). A total of 1,078 samples were obtained, including 483 LUAD samples and 59 control samples, as well as 486 LUSC samples and 50 control samples. The Ntn-4 expression data were subsequently obtained and analysed, and the results confirmed that the mRNA expression level of Ntn-4 was lower in both the LUAD and LUSC samples compared with the control samples (Fig. 1A). No statistically significant differences in the overall survival of the patients with LUSC were observed when comparing between the high and low Ntn-4 mRNA level groups (Fig. 1B); however, the patients with LUAD with a low mRNA expression level of Ntn-4 exhibited poorer overall survival rates compared with the patients with high Ntn-4 mRNA expression levels (Fig. 1C). In addition, the results of RT-qPCR revealed that the expression of Ntn-4 in NSCLC tissues was lower compared with that in the control tissues (Fig. 1D). Moreover, IHC staining analysis also revealed comparatively lower expression levels of Ntn-4 in NSCLC tissues relative to the control tissues (Fig. 1E). With respect to the in vitro experiments, the expression level of Ntn-4 was detected in the human NSCLC cell lines, H23, A549 and H2170, and Ntn-4 was found to be expressed at low levels in the human NSCLC cell lines compared with the 16HBE control cell line (Fig. 1F). Moreover, the western blot analysis revealed that, compared with that in the control cells, the Ntn-4 protein expression levels in the NSCLC cells were lower (Fig. 1G).

To further examine the effects of Ntn-4 on NSCLC cells, the A549 and H2170 cells were transfected with Ntn-4 shRNA or Ntn-4 overexpression lentivirus. The results from the ensuing RT-qPCR and western blot analysis experiments confirmed that the levels of Ntn-4 were increased in the cells with the lentivirus-mediated overexpression of Ntn-4 compared with the control cells (Fig. 2A and B), whereas the levels were decreased in both the A549 and H2170 cells with Ntn-4 shRNA lentivirus transfection (Fig. 2C and D). The OD values of both A549 (Fig. 2E) and H2170 (Fig. 2F) cells were found to be significantly increased in the sh-Ntn-4 group compared with the sh-NC group, although they were decreased in the cells transfected with the oe-Ntn-4 lentivirus compared with the cells transfected with oe-NC lentivirus, indicating that Ntn-4 may exert its effect through the inhibition of NSCLC cell proliferation. Cell migration and invasion were subsequently assessed using wound healing and Transwell assays. The quantification of the resultant data revealed that the efficiency of scratch wound healing was lower in the cells transfected with oe-Ntn-4 compared with the cells transfected with oe-NC, whereas Ntn-4 silencing led to the opposite findings in terms of the scratch wound healing ability of the cells (Fig. 2G and H). Moreover, the numbers of invading cells were also reduced upon transfection with the oe-Ntn-4 vector compared with sh-NC transfection (Fig. 2I). Collectively, these data suggested that Ntn-4 can reduce the migratory and invasive capabilities of the NSCLC cells.

According to the LUAD and LUSC datasets, the mRNA expression level of Qki was lower in the LUAD and LUSC tissue samples compared with the control samples (Fig. 3A). No statistically significant differences in the overall survival of patients with LUSC were observed comparing between the high and low Qki mRNA level groups; however, patients with LUAD with a low mRNA expression level of Qki exhibited poorer overall survival rates compared with those with high Qki mRNA expression levels (Fig. S1). RT-qPCR experiments were subsequently performed to further confirm that the mRNA expression level of Qki-5 was lower in NSCLC tissues compared with that in the adjacent tissues (Fig. 3B). Furthermore, the IHC staining experiments also revealed a low level of Qki-5 in NSCLC tissues compared with the adjacent normal tissues (Fig. 3C). Relative to the control cells, the mRNA expression level of Qki-5 was also decreased in NSCLC cell lines (Fig. 3D). To clarify the association between Qki-5 and Ntn-4, the NSCLC cells were subsequently transfected with oe-Qki-5 lentivirus. The mRNA and protein levels of Qki-5 were found to be markedly upregulated in the A549 and H2170 cells transfected with oe-Qki-5 compared with the cells that were transfected with oe-NC lentivirus (Fig. 3E and F). As was anticipated, the expression of Ntn-4 was upregulated in the oe-Qki-5 cells (Fig. 3G and H). Furthermore, the degree of correlation between Qki-5 and Ntn-4 in the LUAD and LUSC datasets was analysed, and the level of Qki-5 was found to be positively correlated with the level of Ntn-4 (Fig. 3I and J). Pearson's correlation analysis also confirmed that the level of Qki-5 positively correlated with that of Ntn-4 in clinical samples of NSCLC (Fig. 3K and L). To provide further evidence that Qki-5 may be responsible for regulating Ntn-4, Qki-5 was silenced in 16HBE cells, where both Qki-5 and Ntn-4 had been overexpressed. In these experiments, the Ntn-4 protein levels were found to be downregulated in the cells in which Qki-5 was knocked down, even though Qki-5 and Ntn-4 had been overexpressed (Fig. 3M and N). Taken together, these findings indicated that Qki-5 is able to upregulate the expression level of Ntn-4 in NSCLC cells.

To investigate whether Qki-5 inhibits NSCLC progression by upregulating the expression of Ntn-4, Qki-5 was first overexpressed in A549 cells. The level of Ntn-4 was found to be increased in the cells following Qki-5 and/or Ntn-4 overexpression, whereas the level was decreased in cells following the silencing of Ntn-4 (Fig. 4A and B). Moreover, the OD values were found to be significantly decreased in the oe-Qki-5 + sh-NC group compared with the oe-NC group, although, by contrast, they were increased in the cells co-transfected with oe-Qki-5 and sh-Ntn-4. In addition, the OD values of the cells co-transfected with oe-Qki-5 and oe-Ntn-4 were found to be lower compared with those of the oe-Qki-5 + sh-NC group (Fig. 4C). The results from the scratch wound healing and Transwell assays revealed that Ntn-4 silencing was able to reverse the inhibitory effects of Qki-5 on cell migration and invasion, and both the efficiency of wound healing and the numbers of invaded cells were increased in the cells co-transfected with oe-Qki-5 and sh-Ntn-4. By contrast, the migratory and invasive capabilities of the cells were further decreased in the cells co-transfected with oe-Qki-5 and oe-Ntn-4 compared with those co-transfected with oe-Qki-5 and sh-NC (Fig. 4D and E). To further determine the role of Qki-5 and Ntn-4 in EMT, the levels of EMT marker molecules were detected, and it was found that the overexpression of Qki-5 and/or Ntn-4 led to an increase in the level of E-cadherin (Fig. 4F and G), whereas the levels of N-cadherin, vimentin and the EMT transcription factor, Snail, in A549 cells were decreased compared with the control group (Fig. 4F-J). However, by overexpressing Qki-5 and inhibiting the expression of Ntn-4 in the A549 cells, the suppressive effects of Qki-5 on EMT were found to be partly reversed (Fig. 4F-J). Subsequently, RIP assay was performed to determine whether Qki-5 is able to bind to Ntn-4 mRNA. The data obtained revealed that Ntn-4 mRNA was highly enriched in the Qki-5 antibody-precipitated RNA fraction (Fig. 4K), suggesting that Qki-5 could inhibit NSCLC progression partly by upregulating the expression of Ntn-4.

To confirm the effects of Ntn-4 on tumour growth inhibition in vivo, the A549 cells were initially transfected with the oe-NC, oe-Qki-5 + sh-NC, oe-Qki-5 + sh-Ntn-4 or oe-Qki-5 + oe-Ntn-4 lentiviruses, and subsequently the cells were collected and subcutaneously injected into nude mice. The average volumes of the growing tumours were measured each week, and 5 weeks later, the tumours were separated and weighed. Decreases in both the volume and weight of tumours were noted in the mice injected with oe-Qki-5 + sh-NC cells compared with those in mice injected with oe-NC cells. However, the volume and weight of the tumours were increased in the mice injected with oe-Qki-5 + sh-Ntn-4 cells, whereas the cells transfected with oe-Ntn-4 and oe-Qki-5 together led to the most prominent repressive effects on tumour growth (Fig. 5).