The Tumor Immune Estimation Resource (TIMER) (https://cistrome.shinyapps.io/timer/) web server is a resource that systematically analyzes immune infiltrates across different cancer types (31). To evaluate the expression difference of CD146 between tumor and adjacent normal tissues, we used the TIMER database to study the RNA sequence data of different cancer types in TCGA (The Cancer Genome Atlas). The immune cell abundance was estimated by the TIMER algorithm. Correlation modules were used to determine the relationship between the RNA-seq expression profile data of CD146 in ccRCC and immune cells, including CD4+ T cells, CD8+ T cells, regulatory T (Treg) cells, T follicular helper (Tfh) cells, Type 1 T helper (Th1) cells, Type 2 T helper (Th2) cells, natural killer (NK) cells, myeloid dendritic cells, monocyte, neutrophils, M1 macrophages, and M2 macrophages. The gene markers of immune cells were also correlated with CD146 expression using gene modules. These gene markers referenced are cited in previous publications (32–34).

DNA methylation during carcinogenesis has an impact on not only gene expression, but also the prognosis of cancer patients (35). MethSurv (https://biit.cs.ut.ee/methsurv/) is a web portal that provides survival analysis based on DNA methylation biomarkers using TCGA data. DNA methylation of CD146 at CpG sites and the prognostic value of these CpG sites in ccRCC were analyzed by MethSurv.

Tumor specimens were collected from 140 ccRCC patients diagnosed with ccRCC treated with radical or partial nephrectomy at the Department of Urology of the Chinese People’s Liberation Army (PLA) General Hospital (Beijing, China) from January 2013 to December 2019. The medical records of clinic-pathologic data from our institutional database, including age, gender, T stage, N stage, M stage, and Fuhrman grade, were retrospectively reviewed. All patients were staged according to the eighth edition of the AJCC-UICC TNM classification (36). Fuhrman classification was used to attribute nuclear grade (37). The clinic-pathologic data are reported in Table 1 . All samples of cancer tissue had been pathologically confirmed as ccRCC by two pathologists. All patients were informed and signed a consent for the use of clinical data for scientific purposes. The present study was approved by the ethics committee of the Chinese PLA General Hospital.

To determine CD146 protein expression in ccRCC, IHC staining of CD146 was conducted on cancer and paracancer tissues for 140 ccRCC cases from our cohort. For IHC staining, a tissue microarray (TMA) was obtained from the tissue bank at the Department of Urology of the Chinese PLA General Hospital. IHC staining of TMA tissues was performed with antibodies against CD146 (Abcam, ab75769). The standard protocols were followed as previously described (38). Slides were scanned using an Axio Image Z2 Microscope (Zeiss) and the TissueFAXS imaging system (TissueGnostics GmbH, Austria). All images were analyzed by TissueQuest and StrataQuest software (TissueGnostics GmbH, Austria). Staining intensity was scored 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). Staining range was scored on a 4-point scale (0 = 0%, 1 = 1%∼24%, 2 = 25%∼49%, and 3 = 50%∼100%). The final IHC score was obtained by multiplying the intensity scores with the staining range. ccRCC patients with a final IHC score ≥4 were included in the high CD146 group, whereas those with a final IHC score<4 were included in the low CD146 group (39).

Western blot assays were used according to standard techniques as previously reported (40). Antibodies against CD146 (ab75769; Abcam) and β-actin (#3700; CST) were used.

UALCAN (http://ualcan.path.uab.edu/) is a web resource for online analysis of gene transcriptional data and clinical data of cancers from TCGA (41). We obtained differential expression of CD146 mRNA and protein of CD146 in ccRCC tumor tissues and adjacent normal tissues in the UALCAN database. DNA hypermethylation at promoters can lead to gene silencing (42). To further identify the mechanisms underlying the upregulation of CD146 in ccRCC, the methylation levels of CD146 in the ccRCC dataset were also analyzed by the UALCAN database. The differential expression of CD146 and its promoter methylation status in patients with various tumor grade (grades 1, 2, 3, and 4), tumor stage (stages 1, 2, 3, and 4) and ccRCC subtype (ccA and ccB) were also compared.

The LinkedOmics database (http://www.linkedomics.org/login.php) is acknowledged as a web portal that analyses multi-omics data from TCGA datasets (43). We searched for the differentially expressed genes related to CD146 in ccRCC using the LinkFinder module. The correlation results were analyzed by the Pearson correlation coefficient and were visualized by volcano plot and heat maps. To obtain description information, the differentially expressed genes related to CD146 were annotated using Gene Ontology (GO) analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, and gene set enrichment analysis (GSEA) via the LinkInterpreter module.

The TISIDB database (http://cis.hku.hk/TISIDB) is a web server for the interplay between the tumor and immune system, which can assist in the prediction of immunotherapy responses (44). In our study, the associations between CD146 expression and lymphocytes, immunomodulators, and chemokines were analyzed by the TISIDB database. A ‘rho’ value greater than 0.2 and less than −0.2 was considered as a significant correlation at p < 0.05 (45).

To explore whether CD146 could be used as a therapeutic target in cancer patients, we investigated the correlation between CD146 expression and drug response. CD146 gene expression levels in 30 cancer cell lines were obtained from the Cancer Cell Line Encyclopedia (CCLE; https://portals.broadinstitute.org/ccle/). Drug response data in cancer cell lines were downloaded from the Cancer Therapy Response Portal (CTRP, https://portals.broadinstitute.org/ctrp.v2.2/) (46). The correlation between CD146 expression and drug response area under the curve (AUC) was analyzed by Pearson correlation coefficient analysis in each cancer cell type. The percentage of drugs significantly correlated with CD146 was obtained. The ratio of drugs related to CD146 in 10 different cancer types with 30 cell lines was represented by histogram. The correlation between CD146 and 545 drug responses under the curve in 21 renal cancer cell lines were analyzed by SangerBox and shown by volcanic plots. We considered a p-value of less than 0.05 as statistically significant.

Statistical analysis was performed using the GraphPad Prism version 8.0 (GraphPad software, USA) and Statistical Package for Social Sciences (SPSS 17.0 for Windows, SPSS, Chicago, IL). The measurement data are presented as mean ± SD. An independent samples t test was used to analyze the differential expression levels of CD146 mRNA between the ccRCC tissues and the adjacent normal tissues from TCGA databases. Correlations between CD146 expression and clinicopathological characteristics were analyzed by the Pearson’s Chi squared test. Overall survival (OS) analysis and progression-free survival (PFS) analysis were performed by Kaplan–Meier plots and the differences were compared using the log-rank test. Univariate and multivariable analyses were performed using the Cox proportional hazards regression models. A two-tailed P value of 0.05 was considered statistically significant.

Compared with normal tissues, CD146 was upregulated in cholangiocarcinoma, head and neck squamous cell carcinoma, kidney renal clear cell carcinoma, liver hepatocellular carcinoma, pheochromocytoma and paraganglioma, prostate adenocarcinoma, and thyroid carcinoma cancers, while CD146 was downregulated in bladder urothelial carcinoma, breast invasive carcinoma, cervical squamous cell carcinoma and endocervical adenocarcinoma, kidney chromophobe, lung adenocarcinoma, lung squamous cell carcinoma, and uterine corpus endometrial carcinoma cancers ( Figures 1A, B ). Consistent with the mRNA expression data, we found that CD146 protein was highly expressed in the ccRCC cancer tissues compared with paracancer tissues ( Figure 1C ). The positive staining of CD146 was mainly located in the plasma and membrane ( Figure 1C ). The results of UALCAN database analysis further confirmed that CD146 protein expression was higher in primary ccRCC cancer tissues than that in paracancer tissues ( Figure 1D ). The western blot assay also confirmed that CD146 is higher in ccRCC tumor tissues than that in paracancer tissues ( Figure 1E ). This evidence strongly indicated that the mRNA and protein expressions of CD146 were significantly upregulated in ccRCC.

As is shown in Table 1 , CD146 was more expressed in higher grade cancers compared to lower grades cancers ( Table 1 ). Kaplan-Meier analysis demonstrated that high expression of CD146 was significantly associated with poor OS [hazard ratio (HR) = 3.677, p = 0.0028] and PFS (HR = 3.493, p = 0.0009) in ccRCC patients ( Figures 2A, B ). Univariate Cox regression analyses showed an association between OS with age, T stage, N stage, M stage, Fuhrman grade, and CD146 expression. Moreover, multivariate Cox regression analysis showed that CD146 expression (HR = 4.655, p = 0.007), Fuhrman grade (HR = 3.472, p = 0.005), and M stage (HR = 3.625, p = 0.004) were independent prognostic factors for ccRCC patients ( Table 2 ). Univariate Cox regression analysis indicated that T stage, N stage, M stage, Fuhrman grade, and CD146 expression were correlated with PFS. Furthermore, multivariate Cox regression revealed that CD146 expression (HR = 5.829, p = 0.001), Fuhrman grade (HR = 2.927, p = 0.007), and M stage (HR = 3.028, p = 0.005) were independent prognostic indicators for ccRCC patients ( Table 3 ). These results suggest that CD146 was upregulated in ccRCC and associated with worse prognosis.

DNA methylation levels of CD146 were significantly lower in ccRCC cancer tissues compared with normal samples ( Figure 3A ). The methylation status of CD146 was high in late-stage and high-grade tumors ( Figures 3B, C ). Furthermore, correlation analysis indicated that expression of CD146 mRNA was significantly negatively correlated with its methylation status ( Figure 3D ). Among 18 predicted CpG sites of CD146, 15 CpG sites, including cg08187057, cg09042577, cg25484790, cg081861493, cg21096399, cg18890215, cg24827784, cg18165196, cg14976391, cg17466841, and cg11287851, were significantly correlated with the prognosis of ccRCC ( Table 4 ). Consistently, CpG sites of CD146, including cg08187057, cg09042577, cg25484790, cg18890215, cg24827784, cg14976391, and cg17466841, showed higher methylation levels in ccRCC, indicating that the CD146 methylation in these CpG sites was correlated with poor prognosis in ccRCC patients ( Figure 3E ). These results revealed that the methylation levels of CD146 act as an effective prognostic biomarker for ccRCC, demonstrating that CD146 may have a pivotal role in tumor progression.

The results of the co-expression pattern of CD146 showed that 1904 genes were positively correlated with CD146, while 1696 genes were negatively correlated with CD146 ( Figure 4A ). Heat maps displayed the top 50 genes positively and negatively associated with CD146 ( Figures 4B, C ). GO term annotation showed that co-expressed genes of CD146 join mainly in endothelium development, response to virus, T cell activation, adaptive immune response, extracellular structure organization, angiogenesis, regulation of immune effector process, negative regulation of defense response, regulation of innate immune response, leukocyte differentiation, etc. ( Figure 4D ). KEGG pathway analysis indicated enrichment in the Notch signaling pathway, Th1 and Th2 cell differentiation, microRNAs in cancer, Th17 cell differentiation, leukocyte transendothelial migration, and the Rap1 signaling pathway ( Figure 4E ). These results indicate that the CD146 expression network influences the immune microenvironment greatly in ccRCC.

CD146 was positively correlated with infiltrating levels of neutrophils (rho = 0.312, p = 7.44e-12), myeloid dendritic cells (rho = 0.533, p = 3.36e-35), active CD4+ T cells (rho = 0.377, p = 5.63e-17), CD8+ T cells (rho = 0.203, p = 1.14e-05), Treg cells (rho = 0.316, p = 3.78e-12), Tfh cells (rho = -0.251, p = 4.58e-08), NK cells (rho = 0.308, p = 1.38e-11), activated mast cells (rho = 0.307, p = 1.49e-11), monocytes (rho = 0.256, p = 2.59e-08), macrophages (rho = 0.229, p = 6.56e-07), M1 macrophages (rho = 0.189, p = 4.46e-05), and M2 macrophages (rho = -0.333, p = 2.01e-13) ( Figure 5A ). In addition, CD146 was significantly correlated with the gene markers of monocytes, macrophages, M1 macrophages, M2 macrophages, neutrophils, NK cells, dendritic cells, Th1 cells, Th2 cells, Tfh cells, and Treg cells ( Table 5 ). CD146 was significantly correlated with immune stimulators, such as TNFRSF4 (rho = 0.618, p < 2.2e-16), ENTPD1 (rho = 0.616, p < 2.2e-16), TMEM173 (rho = 0.505, p < 2.2e-16), and RAET1E (rho = 0.389, p < 2.2e-16) ( Figure 5B ). The expression of CD146 was also associated with immune inhibitors, including KDR (rho = 0.523, p < 2.2e-16), TGFB1 (rho = 0.424, p = 2.2e-16), ADORA2A (rho = 0.352, p < 4.97e-17), and IDO1 (rho = 0.303, p < 1.03e-12) ( Figure 5C ). CD146 expression was significantly correlated with CCL14 (rho = 0.566, p < 2.2e-16), CCL26 (rho = 0.300, p < 1.82e-12), CCL28 (rho = 0.214, p < 6.64e-7), and CX3CL1 (rho = 0.232, p < 6.21e-8) ( Figure 5D ). Meanwhile, CD146 expression was significantly associated with chemokine receptors, including CCR10 (rho = 0.443, p < 2.2e-16), CXCR4 (rho = 0.388, p < 2.2e-16), CCR6 (rho = 0.183, p < 2.14e-5), and CCR7 (rho = 0.180, p < 2.88e-5) ( Figure 5E ). These results support the findings that CD146 may function as an immunoregulatory factor in ccRCC.

As presented in the previous results, CD146 methylation in ccRCC correlates with prognosis in ccRCC. To elucidate the effect of CD146 methylation on the progression of ccRCC, we assessed the correlation of CD146 methylation with immune infiltration using TISIDB platforms. The result revealed that the methylation status of CD146 was negatively correlated with NK cells (rho = -0.153, p = 0.006), Th1 cells (rho = -0.119, p = 0.034), Th2 cells (rho = -0.242, p = 1.27e-5), and γδT cells (rho = -0.134, p = 0.0166) ( Figure 6A ). Similarly, the methylation status of CD146 was negatively associated with immune stimulators, such as C10orf54 (rho = -0.248, p = 8.13e-06), CD276 (rho = -0.296, p = 8.58e-08), CXCR4 (rho = -0.149, p = 0.0078), and ENTPD1 (rho = -0.287, p =2.04e-07) ( Figure 6B ), while being positively associated with immune inhibitors, such as CD274 (rho = 0.220, p = 7.84e-05), CTLA4 (rho = 0.134, p = 7.84e-05), HAVCR2 (rho = 0.122, p = 0.0289), and KDR (rho = 0.287, p = 2.11e-07) ( Figure 6C ). The methylation status of CD146 was also negatively associated with chemokines and receptors, such as CCL14 (rho = 0.341, p = 5.46e-10), CCR6 (rho = -0.122, p = 0.03), CCR10 (rho = -0.311, p = 1.73e-08), and CXCR4 (rho = -0.149, p = 0.0078) ( Figure 6D ). These results indicated that the methylation status of CD146 is positively associated with immunosuppressive status in ccRCC.

As is shown in Figure 7A , the ratio of drugs significantly correlated with CD146 expression in 10 different cancer cell line types is presented. The percentage of drugs significantly related with CD146 in renal cancer cell lines accounted for 5.5%. As is shown in Figure 7B , the relationship between CD146 and 545 drug response AUCs in renal cancer cell lines is presented. The drugs associated with CD146 in renal cancer cell lines are shown in Table 6 . High CD146 expression was significantly correlated with a better response of inhibitors of topoisomerase I+II, including topotecan, SN-38, and etoposide. In addition, brivanib, inhibitor of VEGFR1/2, was also confirmed to be related to CD146 expression. Besides, inhibitors of DNA replication, including gemcitabine and clofarabine, were correlated with CD146 expression. Moreover, high CD146 expression was related to drug resistance of quizartinib, GSK1059615, BRD-K92856060, and AC55649. Overall, CD146 has the potential to become a therapeutic target for clinical treatment of ccRCC.