The Cancer Genome Atlas (TCGA) dataset was employed to analyze the CPN2 transcription level and the relationship between CPN2 with the clinical characteristics. The level 3 RNA-seq data of 109 normal samples and 1,015 lung cancer samples were downloaded from UCSC website (https://xenabrowser.net/). The gene transcription level was estimated as log₂ transformed reads per kilobase per million mapped reads normalized count.

The Kaplan–Meier plotter (www.kmplot.com) was capable of assessing the prognostic effect of 54,000 genes across 21 cancer types. The portal includes gene chip and RNA-seq data source from Gene Expression Omnibus (GEO), TCGA, and European Genome–phenome Archive (EGA). The Kaplan–Meier plotter database was used to analyze the association between CPN2 expression by the validated probe (216223_at) and overall survival (OS) among lung cancer patients. The median cutoff value of CPN2 expression equal to 24 was employed to separate the patients into the high-expression group and low-expression group in each cohort. P <0.05 was regarded as statistically significant.

Co-expression genes about CPN2 (|Pearson coefficient| >0.3) in TCGA— (LUAD) were gathered from LinkedOmics database (http://www.linkedomics.org/) and visualized in Cytoscape 3.7.2. The unrelated genes in this network were removed, and degree scores were calculated through cytoHubba algorithm. The top 10 genes were considered the most related genes and shown in yellow dots. The corresponding Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment about these co-expression genes was also presented by R project (4.0.2).

The protein–protein interaction analysis of CPN2 was conducted by the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database (http://string-db.org/) using default parameters. Gene Ontology (GO) and KEGG analyses were performed by R project (4.0.2). The top 10 results are shown in a bubble plot, and the adjusted P value <0.05 was considered statistically significant.

The gene set enrichment analysis (GSEA; https://www.gsea-msigdb.org/) was conducted by GSEA software, version 4.0.3. The gene matrix of LUAD in TCGA was separated into two groups depending on CPN2 expression, and different pathways were performed by this analysis. False discovery rate (FDR) <0.05 was considered statistically significant. The top 10 enriched pathways were shown by R project (4.0.2).

The tissue microarray obtained from Shanghai Outdo Biotech Co., Ltd. of China contains 94 LUAD samples and 86 adjacent normal samples, which undergo surgical resection between September 2004 and April 2009. The follow-up time ranged from 1 month to 10 years. The use of clinical specimens for research purposes has been approved by the Research Ethics Committee of Shanghai Outdo Biotech Co., Ltd. The criteria for inclusion were as follows: diagnosed with LUAD by pathological method and has complete survival information. All clinical specimens were treated with conventional methods, and immunohistochemistry was performed to detect the expression of CPN2 protein through previously described protocols (14). About 4-μm-thickness sections were incubated with the anti-CPN2 antibody. The immunohistochemical results were read and interpreted independently by two senior pathologists who were blinded to the sample information. Tumor cells with partial or complete membrane staining were positive at any intensity. According to the ratio of chromogenic cells, the scoring criteria were as follows: 0: negative, 1: positive <25%, 2: positive at 25–49%, 3: positive at 50–74%, and 4: positive at 75–100%. According to the staining degree of cells, no staining was 0, light yellow was 1, brownish yellow was 2, and brown was 3. The immunohistochemical score was obtained by multiplying the two indexes (intensity of staining and number of positively staining cells). The expression of CPN2 was considered to be high expression if the multiplication score was more than 6.

The human lung cancer-derived cell line A549 was obtained from the American Type Culture Collection (Manassas, VA, USA). The cell line was recently authenticated and tested for mycoplasma contamination. The cells were routinely cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum in an incubator with a humidified atmosphere of 5% CO₂ at 37°C.

Total RNA was isolated from cells using TRIzol reagent (Invitrogen, USA) according to the manufacturer’s protocol. The cDNA was synthesized from 2 μg total RNA with PrimeScript^® RT Reagent Kit with gDNA Eraser (Takara, Japan). The mRNA expression was measured through quantitative reverse transcription–polymerase chain reaction (qRT-PCR) by using SYBR Premix Ex Taq (Takara, Japan). The primer sequences are shown in Supplementary Table S1. The relative expression levels among the different samples were calculated using the 2^-ΔΔCt method with normalization to actin. The experiments were performed at least three times.

For CPN2 knockdown, the control siRNA and knockdown siRNA of CPN2 gene were synthesized by Jima (Shanghai, China). The target sequences of CPN2 gene knockdown that were used for siRNA are listed in Supplementary Table S2. For CPN2 overexpression, the cDNA of CPN2 gene was cloned into mammalian expression vector pcDNA3.1-T2A-EGFP. The cells were transiently transfected with those plasmids by ViaFect Transfection Reagent (Promega) according to the manufacturer’s protocol. After 48 h of transfection, the cells were used for subsequent functional experiments.

For the cell proliferation assay, lung cancer cells at a density of 5,000 cells per well were seeded into 96-well plates. After culturing for 24 h, the cells were transfected with the control siRNA and knockdown siRNA of CPN2 gene. The optical density value was detected by measuring the absorbance at 450 nm after 1, 2, 3, and 4 days by using a CCK-8 kit (Dojindo, Japan) according to the manufacturer’s guidelines. All assays were carried out at least in triplicate.

For the cell colony formation assay, after transfection for 24 h, about 500 lung cancer cells were seeded into 6-well plates and cultured. About 3 weeks later, the cell colonies (clones which contained >50 cells were included in the statistics) were stained with crystal violet (0.1%), counted, and photographed. All assays were carried out at least in triplicate.

For the wound healing assay, transfected cells were cultured in six-well plates until confluent. After scratching the monolayer, the cells were photographed at 0, and 48 h. Images were taken from five random optical fields on each filter. For the Transwell assay, a Transwell chamber (Corning, USA) which was coated with or without Matrigel mix (Corning, USA) was used to assess cell invasion and migration, respectively. After being transfected, 2 × 10⁴ cells were plated in the top chamber with serum-free medium, and a medium containing 10% fetal bovine serum was used in the lower chamber as a chemoattractant. After incubation for 24 h, the cells located on the bottom of the chamber were fixed with 4% paraformaldehyde for 15 min, stained with crystal violet (0.1%) for 15 min, and photographed under a microscope. The migrated or invaded cells were counted in five randomly selected fields in each well. Each sample was assayed in triplicate.

SPSS 20.0 software (SPSS, Inc., Chicago, IL, USA) was used to perform the statistical analyses. The difference in gene expression between normal and tumor tissues was analyzed using t-test. The survival plot was analyzed by the Kaplan-Meier method. Different clinical pathological factors and CPN2 expression were considered single factors and performed through univariate and multivariate Cox regression. P <0.05 was considered statistically significant for all groups.

To assess the impact of CPN2 expression on lung cancer, we assembled gene expression datasets from TCGA cohort. CPN2 expression was significantly upregulated in lung cancer tissues compared with adjacent normal tissues (P < 0.01; Figure 1A). Then, we divided all lung cancer samples into two groups: adenocarcinoma and squamous cell carcinoma (LUSC). We found that CPN2 expression was increased both in LUAD (P < 0.01; Figure 1B) and LUSC (P < 0.01; Figure 1C). In LUAD patients, CPN2 expression was significantly upregulated in tumor tissues both in paired and unpaired samples compared with adjacent normal tissues (P < 0.01; Supplementary Figure S1).

In order to evaluate the diagnostic efficacy of CPN2 in lung cancer patients, the receiver operating characteristic (ROC) curve was used to test the hypothesis. Compared with pathological analysis, the area under ROC curve (AUC) of CPN2 gene expression in TCGA database for diagnosing LUAD was 0.86 (95% CI: 0.78–0.92), the sensitivity was 80.2%, and the specificity was 78.9%. The AUC of CPN2 for diagnosing LUSC was 0.62 (95% CI: 0.54–0.70), the sensitivity was 62.9%, and the specificity was 85.25% (Figure 1D). It suggested that CPN2 gene expression was a promising candidate biomarker of LUAD, which has potential clinical diagnosis prospect and application value.

To evaluate the clinical significance of CPN2, we used the Kaplan–Meier public database to analyze the relationship between CPN2 expression and the clinical outcomes of lung cancer patients. Survival plot showed that a higher CPN2 was associated with poor overall survival (P < 0.001, Figure 2A). The subgroup analysis showed that a higher CPN2 predicted a worse survival outcome in the adenocarcinoma group (P < 0.001, Figure 2B), but not in the squamous cell carcinoma group (P > 0.05, Figure 2C). The results showed that the expression level of CPN2 gene was significantly related to the prognosis of LUAD. In patients with stages I and II, the OS time of patients with a low expression of CPN2 gene was significantly longer than that of patients with a high expression of CPN2 gene (P < 0.01) (Figures 2D, E). In patients with stage III, there was no significant correlation between the level of CPN2 gene expression and the OS time of patients (P > 0.05) (Figure 2F). It suggested that CPN2 can be used as a potential tumor marker for the prognosis of lung adenocarcinoma in early stage.

To further clarify the clinical significance of CPN2, immunohistochemical analysis was conducted in a tissue microarray of 94 LUAD tissues and 86 adjacent normal tissues. The results showed that CPN2 was expressed at lower levels in adjacent normal tissues; on the contrary, CPN2 was expressed at higher levels in LUAD tissues (Figure 3A). Compared with adjacent normal tissues, the expression of CPN2 was significantly upregulated in both paired and unpaired (P < 0.01, Figure 3B) LUAD tissues. In order to evaluate the value of CPN2 protein expression in the diagnosis of lung cancer patients, the ROC curve was used to test the hypothesis. The AUC of CPN2 protein expression for diagnosing LUAD was 0.88 (95% CI: 0.83–0.93, Figure 3C). It suggested that CPN2 protein expression was a promising candidate biomarker for LUAD.

Then, we calculated the correlation between the expression of CPN2 and the clinical pathologic parameters in LUAD patients. By dividing the cohort into two groups through the cutoff value of CPN2 protein expression as mentioned above, we found that the CPN2 expression status was significantly associated with lymph node status (P = 0.006), pathological N stage (P = 0.011), and American Joint Committee on Cancer (AJCC) stage (P = 0.035), but not with age, gender, grade, tumor size, total lymph nodes, pathological T stage, or pathological M stage (Table 1).

To evaluate whether CPN2 protein expression correlated with the prognosis of LUAD patients, survival plots were addressed to determine OS in LUAD patients. The results showed that LUAD patients with a higher CPN2 expression had a significantly shorter OS than those with a lower CPN2 expression (P < 0.0001, Figure 4A). The subgroup analysis by AJCC stage revealed that this trend was significant only in stages I and II but not in stages III and IV (Figures 4B, C).

Furthermore, univariate and multivariate Cox regression were conducted to explore the prognostic value of CPN2 between multiple clinical characteristics. All samples separated by the median expression cutoff value of CPN2 were collected to perform this analysis. The results showed that upregulation of CPN2 significantly predicted a poor outcome in LUAD (HR = 3.753, 95% CI = 2.063–6.828, P = 0.001, Table 2). These results indicated that CPN2 possesses the potential capability to be an independent prognostic factor for LUAD.

To further explore the meaningful function of CPN2 in LUAD, co-expression genes were gained from LinkedOmics web tool. A total of 2,615 positively correlated genes (dark red dots) and 957 negatively correlated genes (dark green dots) were gathered and shown in a volcano plot (FDR < 0.05, Supplementary Figure S2A). The top 50 significant genes associated with CPN2 were drawn in the heat map (Supplementary Figures S2B, C).

We gathered 242 co-expression genes using LinkedOmics analysis to speculate the similarity mechanism of CPN2 in LUAD. The criteria were as follows: |Pearson coefficient| >0.3 and FDR < 0.05. The result is shown in Figure 5A. The top 10 hub genes were COL1A2, COL1A1, COL3A1, COL11A1, COL5A3, COL5A1, COL5A2, COL6A1, COL10A1, and COL8A2. Furthermore, the KEGG pathway enrichment analysis revealed the functional annotation of these genes. The results showed that the KEGG pathways were mainly focused on the PI3K-Akt signaling pathway, focal adhesion, proteoglycans in cancer, regulation of actin cytoskeleton, axon guidance, ribosome, phagosome, osteoclast differentiation, platelet activation, and extracellular matrix–receptor interaction (adjusted P < 0.05; Figure 5B). These enrichment results also validated the role of CPN2 which acts as a potential oncogene in LUAD. In addition, these co-expression genes were validated in TCGA-LUAD database (Figure 5C) and presented significant correlations with r > 0.4.

Genes that interacted with CPN2 were evaluated using STRING analysis and visualized by Cytoscape software. The 10 interaction genes were carboxypeptidase N, polypeptide 1 (CPN1), complement component 5 (C5), carboxypeptidase B1 (CPB1), histidine-rich glycoprotein (HRG), latrophilin 2 (LPHN2), latrophilin 1 (LPHN1), cerebellin 1 precursor (CBLN1), complement component 3 (C3), phosphatidylethanolamine binding protein 1 (PEBP1), and latrophilin 3 (LPHN3), and the interaction network is shown in Figure 6A. In addition, GO and KEGG enrichments of these interacted genes were also performed. The GO enrichment presented that regulation of complement activation, humoral immune response, regulation of humoral immune response, complement activation, negative regulation of endopeptidase activity, negative regulation of peptidase activity, negative regulation of proteolysis, regulation of endopeptidase activity, complement activation, alternative pathway, and regulation of peptidase activity were mainly enriched biological processes (Figure 6B). Consistently, pertussis, complement and coagulation cascades, Staphylococcus aureus infection, systemic lupus erythematosus, neuroactive ligand–receptor interaction, herpes simplex virus 1 infection, prion diseases, and legionellosis were mainly enriched KEGG pathways (Figure 6C).

GSEA was performed to further explore the functional role of CPN2 in LUAD progression. We separated TCGA gene matrix into two groups based on CPN2 expression and performed GSEA analysis. Many tumor-related pathways were significantly enriched between the two groups. From the results of the KEGG pathway analysis, we found that a high expression of CPN2 was significantly enriched in the mTOR signaling pathway (Figure 7A), TGF-BETA signaling pathway (Figure 7B), NOTCH signaling pathway (Figure 7C), TOLL-like-receptor signaling pathway (Figure 7D), WNT signaling pathway (Figure 7E), and MAPK signaling pathway (Figure 7F). FDR <0.25 was considered statistically significant. Furthermore, the top 10 enriched pathways according to the normalized enrichment score are shown in Supplementary Figure S3. FDR <0.25 was considered statistically significant. Integrating the results of GSEA and co-expression gene enrichment, we propose that CPN2 may play a vital role through cancer-related pathways.

For the cellular functional experiments, the loss of function assay was performed using CPN2 silencing in lung cancer cell lines. Firstly, we knocked down the CPN2 expression using siRNA vector in the A549 cell line used in our study. As shown by qRT-PCR, CPN2 expression was significantly reduced after the transfection (Figure 8A). Then, the CCK-8 assay results showed that CPN2 knockdown remarkably suppressed the proliferative ability of A549 cells (Figure 8B). The colony formation assay showed that the CPN2 knockdown dramatically inhibited the clones’ number of lung cancer cells (Figure 8C). Then, the wound healing assay was used to detect the effect of CPN2 knockdown on cell migration ability. As shown in Figure 8D, compared with the control group, the migration distance of cells was significantly reduced after the knockdown of CPN2 (P < 0.01). At the same time, a Transwell assay was used to detect the migration and invasion ability of cells after CPN2 knockdown. Transwell migration and Matrigel invasion assays showed that CPN2 downregulation significantly inhibited the migratory and invasive capabilities of lung cancer cells, respectively (Figure 8E). In order to solidly establish the role of CPN2 in lung cancer, we performed the rescue experiment with CPN2 cDNA after the siRNA transfections to rescue the phenotype. The CCK-8 assay (Figure 8F) and Transwell assay (Figure 8G and Supplementary Figure S4) results showed that CPN2 overexpression significantly promoted the cell growth, migration, and invasion ability after CPN2 knockdown. In conclusion, these above-mentioned findings supported the conclusion that CPN2 exerts an oncogenic role in lung cancer cells.