NSCLC tissues and matched peritumoral tissues were collected from patients with NSCLC who underwent surgical resection in Tangdu Hospital affiliated with Air Force Military University (Xi’an, China). All the patients met the criteria of the WHO 2004 classification (21). Tumors were assigned stages according to the eighth TNM edition (22). Frozen NSCLC tissues and matched peritumor tissues were randomly selected for Western blot analysis. The study was approved by the Ethics Committee of Tangdu Hospital (Xi’an, China).

Human lung cancer cell lines (A549) were purchased from the Type Culture Collection (Chinese Academy of Sciences, Shanghai, China). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and penicillin–streptomycin (Gibco) and incubated at 37°C with 5% CO₂.

The QKI expression pattern was studied using NSCLC tissue microarrays, which was purchased from Outdo Biotech Company (HLugA180su05, Shanghai, China). Immunohistochemistry (IHC) staining was performed using the QKI antibody (A7043, ABclonal) as previously described (23). IHC images were obtained using PANNORAMIC (3DHISTECH, Hungary), and the intensity of IHC staining was acquired using Image-Pro Plus 6.0 image software (Media Cybernetics, USA). The multiplied scores of IHC staining results were constructed based on intensity (0 = absent, 1 = faint yellow, 2 = reddish, 3 = brown) and percentage of stained positive cells (0 = 0%–5%, 1 = 6%–25%, 2 = 26%–50%, 3 = 51%–75%, 4 = 76%–100%). Scores of 0 to 5 were considered as low QKI expression, and scores of 6 to 12 were considered to indicate high QKI expression.

Plasmids containing QKI-6 cDNA and plasmids expressing negative control shRNA (sh-NC) and QKI-6 shRNA (sh-QKI-6) were produced (QKI6-ShRNA-Forward: 5′-GATCCGCTGCTCCAAGGATCATTACTTTCAAGAGAAGTAATGATCCTTGGAGCAGCTTTTTTCTCGAGG-3′; QKI6-ShRNA-Reverse: 5′-AATTCCTCGAGAAAAAAGCTGCTCCAAGGATCATTACTTCTCTTGAAAGTAATGATCCTTGGAGCAGCG-3′), and the lentivirus was generated in 239T cells by Biowit Technologies (Shenzhen, China).

To obtain stable QKI-6 overexpression or knockdown, 1 × 10⁵ A549 cells were seeded per well in six-well plates (Corning) and grown to 75% confluence. After changing the culture medium, lentiviral particles containing QKI-6 cDNA, negative control, QKI-6 shRNA, or negative control shRNA were added. Cells were then cultured at 37°C with 5% CO₂. The culture medium was removed and cells were isolated using 5 μg/ml puromycin for 24 h. Infection efficiency was evaluated by Western blot.

Total cellular protein from the A549 cell lines, NSCLC, and peritumoral tissues was extracted using RIPA buffer and quantified using the BCA assay. Sixty micrograms of protein from each sample was resolved by 10% SDS-PAGE, transferred to PVDF membranes (Millipore), which were then blocked in 5% skim milk in TBST, and incubated with primary antibody overnight at 4°C. The primary antibodies used were as follows: QKI (A7043, ABclonal), epidermal growth factor receptor (EGFR) (ab52894, ABclonal), AGR2 (1A8A8, Proteintech), E-cadherin (#14472, CST), N-cadherin (#13116, CST), p-SRC Y418 (ab40660, Abcam), and p-STAT3 Y705 (ab76315, Abcam). After washing with TBST, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 h. Protein bands were visualized using a Tanon 5200 chemiluminescent system (Shanghai, China).

Cell Counting Kit-8 (Beyotime, China, #c0037) was used to assess cell proliferation. A549 cells after gene transfection were plated in 96-well plates (1 × 10³ cells per well) for 1 to 5 days. The cells were treated with 20 μl sterile CCK-8 reagent per well and cultured for 4 h at 37°C. Absorbance was measured at 450 nm.

A549 cell lines were plated in triplicate at a density of 800 cells in 60 mm dishes. On day 14, the colonies were stained with 0.1% crystal violet, and visible colonies were counted.

A 24-well Transwell chamber was used for the migration assay. Cells (1 × 10⁵) were plated in the top chamber of an 8.0-μm membrane (Corning) Transwell in 200 μl DMEM without FBS. The bottom chamber contained 700 μl DMEM with 20% FBS. After 24 h, the migrated cells were fixed and stained with 0.1% crystal violet. Cells in six fields were counted randomly.

A549 (5 × 10⁵) cells were plated in six-well plates and cultured for 20 h. After vortexing in PBS, the cell samples were fixed with 70% ethanol for 1 h. The cells were isolated by centrifugation and resuspended in PBS. The samples were incubated with propidium iodide (PI)/RNase (Thermo) for 30 min in the dark at room temperature and detected using a FACS Calibur (BD).

Sh-NC and sh-QKI-6 cells were cultured in 100 mm dishes. All the sample cells were treated using the Mhelix Biotech kit according to the manufacturer’s instructions (Shanghai, China). Label-free quantification and LC/MS proteomics analysis were performed by the Mhelix Biotech Company (Shanghai, China). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (24) partner repository with the dataset identifier PXD027013.

Trizol reagent (Invitrogen) was used to isolate total RNA. According to the instruction of the M-MLV assay kit (Invitrogen), RNA was reverse-transcribed into cDNA. The primer sequences were as follows: AGR2-Forward: 5′-TTGTCCTCCTCAATCTGGTT-3′; AGR2-Reverse: 5′-ATCGGCTCTAACTGTCAGAGA-3′; GAPDH-Forward: 5′-GGGAAACTGTGGCGTGAT-3′; GAPDH-Reverse: 5′-GAGTGGGTGTCGCTGTTGA-3′. SYBR Green qPCR SuperMix for quantitative PCR was purchased from Vazyme Company. Quantitative PCR was performed using the ABI PRISM^® 7500 Sequence Detection System. Amplification conditions were set to 40 cycles at 95°C for 5 s and 65°C for 30 s.

We used GraphPad Prism version 8 (GraphPad software) for statistical analyses. Data were presented as mean ± standard deviation (SD). Student’s t-test, two-way ANOVA, or log-rank Cox was used for the statistical analyses. P <0.05 was considered statistically significant.

We first studied the putative functions of QKI-6 in NSCLC by comparing QKI mRNA expression in tumor and normal tissues. QKI-6 expression was lower in the NSCLC group than in the normal control group (GEPIA, Gene Expression Profiling Interactive Analysis, Figure 1A ). Furthermore, QKI protein expression was higher in normal tissue than in NSCLC CPTAC (Clinical Proteomic Tumor Analysis Consortium) samples (n = 222, Figure 1B ). QKI protein levels were significantly lower in tumor tissues than in normal tissues in four matched pairs of NSCLC and normal tissues ( Figure 1C ). Next, we analyzed the survival of 270 cases of NSCLC. Low QKI expression was positively correlated with worse NSCLC survival in the TCGA samples (HR = 1.677, 95% CI = 1.124–2.503, P = 0.011, Figure 1D ). Kaplan–Meier analysis results based on 97 NSCLC cases with lymph node metastases N1–N3 showed that low QKI levels were correlated with poor prognosis (HR = 1.931, 95% CI = 1.154–3.232, P = 0.015, Figure 1E ). Furthermore, Cox univariate analysis indicated that lymphatic invasion and distant metastasis were risk factors for overall survival in the TCGA samples (P < 0.05, Figure 1F ). In the multivariate survival analysis, when other risk factors (P < 0.05, HR > 1 in univariate analysis) such as clinical stage, tumor invasion, lymphatic invasion, distant metastasis, and QKI expression were taken into the multivariate mode, QKI expression, lymphatic invasion, and distant metastasis had predictive value for OS in patients with NSCLC ( Figure 1G ).

We performed IHC analysis to study the QKI level on the NSCLC tissue microarray containing 180 dots (86 pairs of tumor–normal tissues and 8 tumor dots, Figure 2A , Table S1 ). Among the tissue pairs, 54.7% (47/86) of the normal tissue samples and 25.5% (22/86) of the corresponding tumor tissue samples were identified as QKI positive. The QKI-6 expression levels in tumor tissues were significantly lower than those in normal tissues (staining scores: 4.51 vs. 7.14, P < 0.0001, Figure 2B ). Moreover, downregulation of QKI expression in the NSCLC tissue microarray was correlated with poor prognosis, as seen in the results of the TCGA sample analysis (HR = 2.24; 95% CI: 1.35–3.71; P = 0.007, log-rank test, Figure 2C ). When analyzing QKI expression with clinicopathological variables, we found that QKI level was significantly associated with NSCLC tumor invasion and clinical stage ( Table 1 ).

To determine whether QKI-6 could exert anticancer effects, QKI-6 overexpression and knockdown were performed in A549 cells ( Figures 3A, B ). Overexpression of QKI-6 inhibited tumor cell proliferation, and knockdown of QKI-6 promoted cell proliferation (P < 0.0001, Figures 3C, D ). Furthermore, overexpression of QKI-6 significantly reduced cell colony formation ability (P < 0.01, Figures 3E, F ) and induced a block in the G0/G1 phase. Conversely, knockdown of QKI-6 in A549 cells increased the proliferation and the proportion of cells in the S and G2 phases ( Figures 3G, H ). Moreover, overexpression of QKI-6 significantly inhibited A549 cell migration, and knockdown of QKI-6 promoted A549 cell migration ( Figures 3I, J ).

To elucidate the mechanisms by which QKI-6 could promote A549 cell migration, label-free proteomics technology was used to detect changes in protein expression levels in sh-QKI-6 and sh-NC cells. We identified 4,373 differentially expressed proteins by using Proteome Discoverer 1.4 between the sh-QKI-6 and sh-NC cells ( Figure 4A ). The label-free quantitation analysis of sh-QKI-6 and sh-NC proteins was performed using the Progenesis LC-MS software to identify 316 differentially expressed proteins (>2-fold). Of these, 145 were upregulated and 171 were downregulated (sh-QKI-6/sh-NC >2-fold or <0.5-fold, P < 0.05, Figure 4B , Table S1 ). All the differentially expressed proteins were analyzed using the Gene Ontology (GO, www.geneontology.org) database, which provided networks of molecular function (MF, Figure 4C ), biological process (BP, Figure 4D ), and cellular component (CC, Figure 4E ) to describe the gene product attributes (25). As shown in Figures 4C–E , the differentially expressed proteins were involved in various MFs and CCs. The classification of these proteins according to their molecular function showed that they were mainly involved in oxidoreductase activity, cytoskeletal protein binding, and actin binding ( Figure 4C , Figure S1 ). We also found that QKI-6 played a significant role in BPs including lipid metabolism, cell migration, and cell motility ( Figure 4D , Figure S2 ). The cellular component repartition analysis showed that the identified proteins were mainly presented in the cell periphery and plasma membrane ( Figure 4E , Figure S3 ).

Among the differentially expressed proteins, AGR2 was significantly upregulated in sh-QKI-6 A549 cells compared with sh-NC. We represented the result of AGR2 OS in the TCGA sample (n = 504) by Kaplan–Meier analysis. Higher AGR2 mRNA expression levels were correlated with poor prognosis (HR = 1.41, 95% CI = 1.05–1.89, P = 0.021, log-rank test, Figure 5A ). Furthermore, in the TCGA and GTEx databases, AGR2 mRNA levels in NSCLC were significantly upregulated compared with those in normal tissues (P < 0.05, Figure 5B ), and AGR2 mRNA levels were negatively correlated with QKI-6 mRNA levels (n = 513, r = −0.32, 95% CI = −0.39 to −0.23, P < 0.001, Spearman’s test, Figure 5C ). In addition, Western blot analysis showed that AGR2 was significantly upregulated in sh-QKI-6 A549 cells ( Figure 5D ). In this study, AGR2 mRNA expression was analyzed in A549 cells to confirm the effectiveness of si-RNA-AGR2 ( Figure 5E ). EGFR, p-SRC, and p-STAT3 simultaneously increased in QKI-6 knockdown A549 cells and decreased in AGR2 knockdown A549 cells. QKI-6 expression also affected the expression of E-cadherin and N-cadherin ( Figure 5F ). Furthermore, siRNA-mediated AGR2 knockdown significantly decreased migration and invasion in control and QKI-6 knockdown A549 cells ( Figures 5G–J , P < 0.01). Thus, QKI-6 is likely an inhibitor of NSCLC metastasis and suppresses EMT and metastasis by regulating AGR2 via EGFR–SRC–STAT3 signaling.