LUAD samples with complete clinical information and gene expression data were obtained from The Cancer Genome Atlas Program (TCGA) database (https://www.genome.gov/Funded-Programs-Projects/Cancer-Genome-Atlas). To explore the underlying mechanisms related to CD146 expression, we conducted the functional enrichment analysis and signaling pathway enrichment analysis on the significant differentially expressed genes (DEGs) identified between high- and low-CD146 expression groups, as determined by the median CD146 expression value. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, gene set variation analysis (GSVA) as well as Gene Set Enrichment Analysis (GSEA) were all used to explore the underlying molecular mechanisms related to CD146 expression and the potential role of CD146 in NSCLC.

In addition, needle biopsy tissues were collected from 150 NSCLC patients treated at Zhengzhou University People’s Hospital from 2017 to 2021. These patients were followed up for five years after diagnosis. The CD146 expression level of each sample was evaluated by immunohistochemistry, with samples then assigned to either the high- or low-CD146 expression groups by the median value. Our study was evaluated and approved by the Ethics Review Committee of Zhengzhou University People’s Hospital and adhered to all relevant ethical and legal standards (Approval No 2021-27). The included NSCLC patients signed the informed consent for all the images, other personal and clinical data in this study.

The human lung cancer cell lines H460, A549, H1975, H1299, H661, SK-MES-1, H226 and PC9 were acquired from the cell bank of the Chinese Academy of Science (Beijing, China). They were cultured in the Dulbecco’s Modified Eagle Medium (DMEM)/RPMI 1640 medium containing 10% fetal bovine serum (Gibco, Waltham, MA, USA). Lentiviruses for CD146 overexpression (Lv-CD146) or CD146 silencing (Lv-shCD146) were acquired from Genechem Co. Ltd (Shanghai, China). NSCLC cells were infected with Lentivirus in the range of 5 to 20 multiplicity of infection (MOI) in polybrene (10 µg/mL). The pcDNA3-CD146 plasmid was utilized to overexpress CD146 in A549 cell lines. The H460 cell line expresses high levels of CD146 and was infected with Lv-shCD146 to produce H460-shRNA cells. After 72 hours of infection, cells were acquired for the subsequent experiments by growing in the presence of puromycin (5 µg/mL) for 1 week. All cell lines were authenticated using short-tandem repeat (STR) DNA fingerprinting (Baseclear, Leiden, The Netherlands) and routinely tested for mycoplasma by polymerase chain reaction (PCR). All cell lines were mycoplasma-free.

The CD146 expression level in H460, A549, H1975, H1299, H661, SK-MS-1, H226 and PC9 cells was analyzed using flow cytometry. Cells in the phase of logarithmic growth were collected, washed and resuspended in PBS. PE-IgG and PE-CD146 antibodies were added to the cell suspension (1 $\times$ 10${}^6$ cells/20 µL) and incubated on ice for 40 minutes in the dark. Then, the cells were collected by centrifuge and washed twice with phosphate buffered solution (PBS). They were then filtered with 200-mesh nylon and analyzed by FACS cytometry (BD Biosciences, New York, NY, USA). FlowJo software (v10.6.2, BD Biosciences) was used to determine the mean fluorescence intensity (MFI) of CD146 expression.

H460 and A549 cells in the phase of logarithmic growth were seeded into 24-well plates at a density of 1 $\times$ 10${}^5$ cells/well and grown overnight. Subsequently, the samples were gathered and rinsed using PBS, stabilized in 4% paraformaldehyde for a duration of 10 minutes, made permeable with 0.1% Triton X-100 for 10 minutes, obstructed using 1% Bovine Serum Albumin (BSA) for one hour, and thereafter subjected to incubation with rabbit primary antibody against CD146 (ab75769, Abcam, Cambridge, MA, USA, at a dilution of 1:250) for three hours at ambient temperature. The cells washed twice by PBS were incubated with red fluorescence second antibody (A-21245, Invitrogen, Carlsbad, CA, USA) and then stained with DAPI. Coverslips underwent preparation followed by examination with a laser scanning confocal microscope (Carl Zeiss Meditec AG, Oberkochen, Germany), which was outfitted with an Olympus IX81 digital camera (Olympus Corporation, Tokyo, Japan) and utilized a 20/0.75 numerical aperture objective for imaging.

Wound healing assay analysis was performed to evaluate the ability of tumor cell migration. The cells were grown in culture plates and a straight scratch was made using a 10 µL pipette tip. Photographs were captured at multiple intervals (0 hour, 12 hours, 24 hours, and 48 hours) after the wound creation. The width of the wound was measured using an ocular ruler to ensure uniformity across all wounds at the initial time point.

For evaluating cell invasion, Transwell cell culture inserts were utilized in a 24-well plate. A total of 2 $\times$ 10${}^4$ cells were incubated in the upper chamber with 200 µL of serum-free medium, while, in the lower chamber, 800 µL of medium containing 20% FBS was placed. Following a 24 h incubation period, cells on the underside of the insert were fixed using 100% methanol and subsequently stained with 0.5% crystal violet.

Cells grown on a 10 cm dish were thoroughly washed once with PBS and whole-cell lysates were then extracted using RIPA lysis buffer (Solabio, Beijing, China). Following centrifugation, the resulting supernatants containing protein lysates were collected and the protein concentration assessed using a bicinchoninic acid (BCA) Protein Measurement Kit (Beyotime, Shanghai, China). Equal quantities of the proteins from the whole-cell lysate were first subjected to the sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred onto polyvinylidene difluoride membranes. The membranes underwent a blocking process for 1 hour in milk (0.1 g/mL), then were incubated in the refrigerator with the primary antibody at 4 °C overnight. Subsequently, they were rinsed thrice with Tris-buffered saline (TBS)-Tween and further subjected to a 1-hour incubation with the secondary antibody. An Electrochemiluminescence (ECL) detection reagent (Solabio, Beijing, China) was used to visualize the signal. The primary antibodies used here are listed in Supplementary Table 1.

Diaminobenzidine detection kit (Maixin-Bio Co. Ltd., Fuzhou, China) was utilized to conduct immunohistochemical staining based on the manufacturers’ instructions. In brief, NSCLC tissues were subjected to standard histological processing of dewaxing and rehydration. To restore antigenicity, the sections were heated for two minutes in a pressure cooker. Next, endogenous peroxidase activity was blocked by treating the sections with 3% H${}_2$O${}_2$ for ten minutes, followed by incubation with the goat serum for ten minutes to prevent non-specific antibody binding. The section was then incubated in the primary antibody for one hour at 37 °C following by incubation with biotin-labeled secondary antibody for 10 minutes and streptavidin-peroxidase conjugate for an additional 10 minutes. To visualize peroxidase activity, a 0.02% solution of diaminobenzidine was used as the chromogen. Subsequently, the section was lightly counterstained with the hematoxylin, mounted using Permount, and examined under light microscope.

All the statistical analyses in our study were carried out using SPSS 18.0 software (IBM Corp., Chicago, IL, USA). The results were shown as the mean $\pm$ standard error of mean. Quantitative data and categorical data were analyzed by two-tailed Student’s t-test as well as the $\chi$${}^2$ test, respectively. Kaplan-Meier with log-rank analysis were used to determine the survival between the two subgroups. A p-value $\le$ 0.05 was considered to represent statistical significance.

To explore the potential molecular mechanism involving CD146, the RNA-seq profiles and clinical data of LUAD and lung squamous cell carcinoma (LUSC) samples were obtained from TCGA database. GSVA and GSEA analyses were then performed on the DEGs found between the high- and low-CD146 expression groups using R studio in R software (version 4.1.3, RStudio, Inc., Boston, MA, USA). In LUAD, 28 signaling pathways were found to be positively associated with CD146 expression, and 14 pathways were negatively associated with CD146 expression. In LUSC, 29 positively related-pathways and 18 negatively related-pathways were found to be associated with CD146 expression (Fig. 1A). In both LUAD and LUSC, CD146 expression was involved in signaling regulation of the EMT pathway. GSEA analysis showed a positive association between the CD146 expression and EMT in NSCLC (Fig. 1B). This result was consistent with those obtained from KEGG analysis showing that CD146 expression was related to TGF-$\beta$, EMT, cell connection and other pathways, all of which are closely associated with tumor metastasis (Fig. 1C).

Fig. 1.

Potential CD146-regulated signaling pathways. (A) Gene set variation analysis (GSVA), and (B) Gene Set Enrichment Analysis (GSEA) of correlations between CD146 expression and epithelial-mesenchymal transition (EMT)-related gene signatures in The Cancer Genome Atlas (TCGA) non-small cell lung cancer (NSCLC) dataset. (C) Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of CD146-related pathways in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) datasets from TCGA database.

A total of 150 NSCLC tissue samples confirmed by pathologic diagnosis were collected. Immunohistochemical staining was conducted to assess the expression level of CD146. Samples were assigned to high- (n = 95) or low-CD146 (n = 55) expression groups separated by the median expression value. The associations between CD146 expression and the clinical features of NSCLC and patient demographics were shown in Table 1. High CD146 expression was notably related to advanced TNM stages, including lymph node metastasis, and distant metastasis (p 0.05), but no significant correlations were observed with age, gender, histopathological type, or tumor size.

We next analyzed whether CD146 expression was the risk factor for OS. Multivariate analysis revealed that high CD146 expression as well as TNM stage was the risk factor for worse survival of NSCLC patients (Table 2). Kaplan–Meier survival analysis indicated that patients with NSCLC and high CD146 expression had shorter OS (Fig. 2).

Fig. 2.

The prognostic value of CD146 expression in NSCLC patients. Kaplan-Meier survival analysis showing that patients with NSCLC and high CD146 expression had a significantly shorter survival time.

To assess its biological function in NSCLC, the expression level of CD146 in the eight NSCLC cell lines was first evaluated by Western blot and flow cytometry (Fig. 2). H460 cells were used in subsequent studies due to their high expression of CD146 compared to the other cell lines, while A549 cells were also used due to their low expression (Fig. 3A,B). Next, the siCD146 sequence (5${}^{\prime }$-GGAACTACTGGTGAACTATGT-3${}^{\prime }$) was cloned into puromycin-resistant lentivirus vector (GenePharma, Shanghai, China) and cultured in medium with puromycin (10 µg/mL) for 48 h. The CD146 overexpression lentivirus (GenePharma) was also constructed and transfected. Transfection efficiency was confirmed by the knockdown and overexpression effects, as observed by multiplex immunohistochemistry and Western blot (Fig. 3C,D).

Fig. 3.

Construction of NSCLC cell lines with stable CD146 overexpression or silencing. (A) Western Blot, and (B) flow cytometry analysis of CD146 expression in NSCLC cell lines. (C) Immunofluorescence analysis of CD146 expression levels in H460 and A549 cells, and cellular localization. (D) Construction of cell lines with high or low CD146 expression. Gray scale analysis of CD146/GAPD. ns, not significant, *** p 0.001.

The role of CD146 in cell invasion and migration were tested by wound healing and transwell assays. These showed that CD146 knockdown significantly inhibited the migration and invasion of H460 cells compared with the negative control. In addition, CD146 overexpression significantly enhanced the cell migration and invasion of A549 cells compared with negative control (Fig. 4).

Fig. 4.

In vitro migration and invasion assays. (A) Wound healing assays showed that inhibition of CD146 expression significantly reduced H460 cell migration, whereas CD146 overexpression increased the migration of A549 cells (n = 3, ***p 0.001, Scale bars: 500 µm). (B) Transwell assay showed that CD146 silencing decreased the invasion of H460 cells, and CD146 overexpression increased invasion of A549 cells. *** p 0.001. Scale bars: 500 µm. (C) Transwell assay showed that CD146 silencing decreased the migration of H460 cells, and CD146 overexpression increased migration of A549 cells. *** p 0.001. Scale bars: 500 µm.

Tumor metastasis is a complex process regulated by a variety of factors, including changes in the microenvironment, interaction between cells and the cell matrix, loss of cell adhesion, the migration as well as invasion of tumor cells. A prerequisite and necessary condition for tumor cell metastasis is EMT. The results presented above suggest that CD146 can promote the migration and invasion of NSCLC. Subsequently, we further explored the molecular pathways mediated by CD146 in promoting the migration as well as invasion of NSCLC. Moreover, western blot demonstrated that CD146 silencing reduced the expression of mesenchymal-associated marker proteins and transcription factors, such as N-cadherin, vimentin, snail and twist. Conversely, CD146 overexpression resulted in the upregulation of these proteins. Meanwhile, the expression of E-cadherin was decreased (Fig. 5).

Fig. 5.

CD146 induces epithelial-mesenchymal transition (EMT) in NSCLC. Western blot analysis of shCD146 H460 cells showed significant downregulation of N-cadherin, vimentin, snail and twist compared with vector control cells. Conversely, pcDNA3.1-CD146 A549 cells showed significant upregulation of N-cadherin, vimentin, snail and twist compared with vector control cells. * p 0.05, ** p 0.01, *** p 0.001.

To study the molecular mechanism by which CD146 promotes NSCLC invasion and metastasis, transcriptome profiling via RNA sequencing was performed on the NSCLC cell line H460 and its CD146-knockdown cell line H460-shRNA. Differential expression analysis identified 324 DEGs (174 upregulated and 150 downregulated) at false discovery rate (FDR) 0.05, log${}_2$ $|$fold change$|$ $>$1 after CD146 knockdown (Fig. 6A,B).

Fig. 6.

Identification of differentially expressed genes (DEGs) between H460 cells and H460 cells following transfection with shCD146 vector, and functional enrichment analysis of the DEGs. Volcano plot (A) and heatmap (B) of DEGs showing the upregulated and downregulated genes. Bubble plots of the most enriched GO terms (C) and KEGG pathways (D) for all DEGs. GSEA analysis suggested that CD146 expression was significantly associated with positive regulation of cell migration (E), focal adhesion pathways (F), the PI3K/Akt pathway (G), and the WNT pathway (H).

GO and KEGG analyses were performed to explore CD146-related downstream signaling pathways based on the identified DEGs. Cell adhesion and cell junction, both of which are closely related to tumor metastasis, were found to be associated with CD146 (Fig. 6C,D). GSEA confirmed the significant positive regulation of cell migration, focal adhesion pathways, the PI3K/Akt pathway, and the WNT pathway in CD146-high expression cells (Fig. 6E–H).

We next evaluated whether CD146 can activate the PI3K/Akt pathway. The phosphorylation levels of Akt and PI3K in H460 cells decreased following knockdown of CD146 expression, whereas CD146 overexpression considerably increased activation of the PI3K/Akt pathway based on phosphoprotein levels (Fig. 7). Moreover, inhibition of PI3K activity using PI3K-IN-1 markedly decreased N-cadherin, vimentin, snail and twist expression, and increased E-cadherin expression. In contrast, the activation of PI3K using 740YP markedly increased N-cadherin, vimentin, snail and twist expression. The above results suggest that CD146 mediates EMT in NSCLC by activating the PI3K/Akt pathway, thus facilitating the migration and invasion of NSCLC.

Fig. 7.

CD146 induces the epithelial-mesenchymal transition (EMT) in NSCLC by activating the PI3K/Akt/Snail pathway. Western blot analysis showing that CD146 silencing in H460 cells elevated E-cadherin expression and reduced N-cadherin, vimentin, snail, and twist expression. Treatment of sh-CD146 H460 cells with 740YP decreased E-cadherin expression, but elevate N-cadherin, vimentin, snail, twist expression, as well as PI3K and AKT phosphorylation compared to vector control cells. CD146 overexpression in H460 cells resulted in inhibition of E-cadherin and elevation of N-cadherin, vimentin, snail, and twist expression. Treatment of pcDNA3.1-CD146 A549 cells with PI3K-IN-1 increased the expression of E-cadherin, but reduced N-cadherin, vimentin, snail and twist expression, as well as PI3K and AKT phosphorylation compared with vector control cells. ** p 0.01.