The TCGA database (https://genome-cancer.ucsc.edu/) is an open and freely accessible online database of cancer genomes consisting of clinical and pathological data on 30+ different cancer types. Through the TCGA database, we were able to gather information on ccRCC patients, including RNA-seq expression, prognosis, and pertinent clinicopathological information. The correlation of WAS and immune infiltration in ccRCC was observed by ESTIMATE and ssGSEA analysis. Median RNA expression was used as a cutoff to stratify patients into elevated or low expression groups, while the area under the receptor operating characteristic (ROC) curve (AUC) was applied for qualitative and quantitative performance measure.

The National Biotechnology Information Center (NBIC) (https://www.ncbi.nlm.nih.gov/geo/) maintains the GEO database, which is an extensive gene expression database. Transcriptomic data was obtained from the GEO (GSE53757), an independent validation cohort of ccRCC patients.

UALCAN is a convenient and easy-to-use online application that can be used for analyzing publicly available information about cancer (24). Proteomics technologies enable CPTAC to characterize and quantify tumor samples by mass spectrometry, which is able to identify each tumor sample’s characteristic proteome and constituent proteins. UALCAN was used in this study to analyze the flux of WAS protein expression obtained from CPTAC.

The HPA database (https://www.proteinatlas.org/) provides information on 26,000 human proteins, as well as their distribution in tissues and cells (25). With the use of specific antibodies, it is possible to visualize the expression profiles of proteins in normal tissues, cancer cells and cell lines by immunohistochemistry. In this study, we investigated the expression of WAS in both normal and tumor tissues by utilizing the HPA database.

The KM plotter can draw survival curves using 530 ccRCC samples to assess survival prognosis of related genes (26). In ccRCC, this database was used to asses overall survival (OS) and progression-free survival (PFS) of WAS in addition to estimating hazard ratios (HR) and p-values.

STRING (https://cn.string-db.org/) is an online tool for searching and constructing PPI interaction networks using interacting genes and proteins. The following main parameters are used to create a WAS co-expression network: 1) Active interaction sources: Co-expression, Co-occurrence, Textmining, Experiments, Databases; 2) The meaning of the network edge: evidence; 3) Maximum number of interactions: 10; 4) Minimum Required Interaction Score: High Confidence (0.700). Gene Ontology (GO) enrichment and Kyoto Encyclopedia of KEGG pathway analysis of coexpression genes were performed with the ClusterProfiler package, the results of which were visualized using “ggplot2”.

Tumor immune estimation resource (TIMER2.0) (https://timer.cistrome.org/) is an online data platform, which can be used to systematically analyze the immune infiltration in various malignant tumors (27). In addition, tumor purity can be estimated from this database. “GENE” module was applied to examine the WAS expression and its associations with tumor-infiltrating lymphocyte (TIL). Expression scatter plots were created by the correlation module for the correlation and estimated statistical significance of WAS expression with immune cell marker genes.

The Gene Expression Profiling Interactive Analysis (GEPIA) online database (http://gepia.cancer-pku.cn/index.html) is a comprehensive platform consisting of 8587 normal and 9736 tumor samples from GTEx and TCGA data (28). As part of this study, we analyzed the relationship between WAS expression and various immune cell markers in ccRCC as well as OS and progression-free survival (PFS) in ccRCC.

In order to study the interaction between tumors and immune factor, TISIDB (http://cis.hku.hk/TISIDB/index.php) combines data sources from a variety of heterogeneous sources. This database can be used for predicting immunotherapy responses and identifying new immunotherapy targets, among others. TISIDB was employed in this study to examine the relationships between WAS and 28 TILs, 45 immunostimulators, 24 immunoinhibitors, 41 chemokines, and 18 receptors in ccRCC (29).

Patients with ccRCC in TCGA were divided into high expression and low expression group based on the median expression of WAS. DEGs between two groups were identified by R software “limma” with P < 0.05 and |Log2 (Fold Change)| > 1 as thresholds.

To explore the immune cell abundance in ccRCC tissues, CIBERSORT (30) was employed to evaluate the proportions of 22 immune cell types using a deconvolution algorithm by the R package with default parameters. In addition, the ESTIMATE scores (ES), tumor purity (TP), stromal scores (SS), and immune scores (IS) for each ccRCC sample were evaluated using the ESTIMATE algorithm (31) of the “estimate” package.

The immunophenoscore (IPS) is a machine learning-based scoring system applied for the prediction of patients’ responses to ICI treatment based on the weight average Z scores representing immune-related genes expression in cell types (32). High IPS scores reflect increased immunogenicity. As targeted therapy is widely used to treat ccRCC, risk scores were used to predict the drug sensitivity based on half-maximal inhibitory concentrations (IC50) for each ccRCC patient from the Genomics of Drug Sensitivity in Cancer (GDSC) website (33) using the R package “pRRophetic” (34).

Renal tubular epithelial cell line (HK-2) and ccRCC cell line (ACHN, Caki-2, 786-O, 769-P, A498) were purchased from Cell bank of Chinese Academy of Sciences. The HK-2 cells were cultured with DMEM (BI, Israel) and the other cells with RPMI1640 (BI, Israel). There was 10% fetal bovine serum (FBS) added to all media, as well as 1% penicillin and streptomycin. A humidified incubator with 5% carbon dioxide and 37°C was used for culture of each cell type.

Approved by the ethics committee of the affiliated Yantai Yuhuangding hospital of Qingdao university, a total of 20 eligible ccRCC patients were enrolled. ccRCC samples were acquired from patients who underwent radical nephrectomy, or partial nephrectom. The pathological specimens were confirmed by two independent pathologists.

Total RNA was extracted with Trizol reagent (Pufei, Shanghai) according to the manufacturer’s instructions. RNA was reverse transcribed to cDNA using the Promega M-MLV kit. The primer sequence of WAS for qPCR is forward 5′-CACGAGAACCAGCGACTCTTTG-3′ and reverse 5′-CCACAATGCTCCTTGGTCCAGT-3′.

We carried out all statistical analyzes with R (version 4.2.1) and visualizing the results with ggplot2 (version 3.3.3). In order to determine whether there are any differences between ccRCC tissues and normal surrounding tissues, the Mann-Whitney U test and paired t-test were performed. Execute the KM graph to build the survival curve. For KM plot, GEPIA, and TISIDB, HR, and P values are described using the log-rank test. In order to determine the relationship of WAS expression with immune infiltration levels, immunomodulators and chemokines, spearman’s correlation coefficients were calculated. A very weak correlation was determined by <0.2, a weak correlation by < 0.4, a moderate correlation by < 0.6, a strong correlation by < 0.8, and a very strong correlation by < 1.0. Statistical significance was determined by P < 0.05.

The pan-cancer analysis demonstrated that WAS expression was elevated in most cancer types, including breast invasive carcinoma, head and neck squamous cell carcinoma, renal clear cell carcinoma, and renal papillary cell carcinoma compared with adjacent tissues ( Figure 1A ).

To determine the mRNA and protein expression of WAS in ccRCC, the expression profile from TCGA, GEO, and UALCAD was analyzed. Our analysis of Figure 1B shows that the WAS mRNA in ccRCC samples (539 cases) of TCGA was higher than that in the normal samples (72 cases) (P < 0.001), which was similar to the data from TIMER database. From the TCGA database, Figure 1C represents that WAS expression in ccRCC was higher than that in matched normal tissues (n = 72). The same information can also be verified in the GEO database (GSE53757) ( Figure 1D ). By using the CPTAC analysis of UALCAN, we were able to conduct the protein expression analysis of WAS. The results showed that protein expression of WAS (n = 110) in ccRCC was significantly higher than that in normal tissues (n = 84) ( Figure 1G ). Furthermore, qPCR was performed to examine the expression of WAS in 20 pairs of ccRCC cases and adjacent normal tissues. At the same time, we verified that WAS gene expression in most ccRCC cell lines was higher than that in the control group through qPCR experiment ( Figure 1F ). It was observed that WAS mRNA ( Figures 1E, F ) and protein levels ( Figure 1G ) were higher in most ccRCC tissues than in paired adjacent normal tissues. Validation of protein expression was conducted using Human Protein Atlas (HPA) database ( Figures 1H, I ). And we found elevated expression levels of WAS protein in cancer tissues, particularly within renal tumor cells. Based on the above data, WAS gene was increased significantly in ccRCC, which might indicate that it can be used as a potential diagnostic biomarker.

Table 1 describes the baseline characteristics of ccRCC patients. As is shown in Table 1 and Figure 2 , elevated expression of WAS was significantly correlated with distant metastasis, tumor stage, pathological stage, histological grade, laterality, and overall survival (OS) (P < 0.05). The association between WAS expression and other clinicopathological features, such as gender, serum calcium level, hemoglobin, and age, was not statistically significant (P > 0.05). In general, these results indicated that WAS expression level was associated with TNM stages, pathological/histological grade, and prognosis of ccRCC patients.

The distribution of WAS expression, survival status of ccRCC patients, and expression profiles of WAS were shown in Figure 3A . With the increase of risk score in ccRCC patients, the number of dead ccRCC patients increased gradually. Additionally, the Kaplan-Meier survival curves indicated that high-WAS patients had poorer survival and a higher risk of death than those of low-WAS patients (P = 0.0176) ( Figure 3B ). The ROC analysis demonstrated that the elevated WAS expression allowed for highly accurate ccRCC diagnosis (AUC = 0.935, 95% CI: 0.908–0.961) ( Figure 3C ). Additionally, we verified that WAS expression was correlated with OS in ccRCC using GEPIA (P < 0.01) ( Figure 3D ). A significant correlation between WAS expression and DFS (Disease Free Survival) was not detected ( Figure 3E ).

DEG analysis between the high- and low-WAS groups in the TCGA cohort showed 475 up-regulated and 1020 down-regulated DEGs ( Figure 4A ). The heatmap suggested the top twenty significant up- and down-regulated genes ( Figure 4B ).

An interactive map was created between WAS and its 10 co-expressed genes based on the STRING database, and constructed a PPI network based on this map ( Figure 5A ). The functional annotations in Figure 5B show that WAS is related to Fc receptor signaling pathway, immune response-activating cell surface receptor signaling pathway, immune response-regulating cell surface receptor signaling pathway involved in phagocytosis, Fc-gamma receptor signaling pathway involved in phagocytosis, Fc-gamma receptor signaling pathway, phagocytosis, and so on. Figures 5C, D illustrate the components of cells as well as their molecular functions. KEGG pathway analysis results showed that the top 10 related genes shown in the PPI network are involved in T cell receptor signaling pathway, Fc gamma R-mediated phagocytosis, tight junction, adherens junction, endocytosis, natural killer cell mediated cytotoxicity, renal cell carcinoma and B cell receptor signaling pathway, etc ( Figure 5E ). As is presented in Figures 5F–J , the WAS expression level was positively correlated with ARPC3 (Actin Related Protein 2/3 Complex Subunit 3) (r = 0.413, P < 0.001), BTK (Bruton Tyrosine Kinase) (r = 0.765, P < 0.001), GRB2 (Growth Factor Receptor Bound Protein 2) (r = 0.486, P < 0.001), LCP2 (Lymphocyte cytosolic protein 2) (r = 0.766, P < 0.001), and WIPF1 (WAS/WASL Interacting Protein Family Member 1) (r = 0.639, P < 0.001).

To determine whether WAS expression was correlated with immune cell infiltration in ccRCC, we employed the TIMER2.0 online platform and the ssGSEA algorithm. Genetic techniques used to study immune infiltration must take into account the purity of tumor cells in clinical cancer samples. In our research, WAS expression was adversely correlated with the purity of ccRCC (r = -0.363, P < 0.001). Additionally, we observed a strong correlation between WAS and TIL abundance ( Figure 6A ). We used the ssGSEA algorithm to identify the levels of immune cell infiltration in TCGA ccRCC samples, and the vast majority of which were positively correlated with WAS expression ( Figure 6B ). The ESTIMATE algorithm revealed a profound association of WAS with stromal score (r = 0.50, P < 0.001), immune score (r = 0.93, P < 0.001), ESTIMATE score (r = 0.83, P < 0.001), and tumor purity (r = -0.82, P < 0.001) in ccRCC ( Figure 6C ). As predicted by the ESTIMATE algorithm, high WAS expression patients were significantly higher in immune score (P < 0.001), stromal score (P < 0.001), and ESTIMATE score (P < 0.001) compared with low WAS expression patients ( Figure 6D ). In contrast, the ESTIMATE algorithm was able to determine that elevated WAS expression groups had a lower tumor purity than low WAS expression group ( Figure 6D ). For example, the correlation between WAS and T cells, cytotoxic cells, Th1 cells, ADC, Tregs, B cells, TFH, T helper cells, and other immune cells was greater than 0.5. Based on the TIMER database, we found that the WAS expression was correlated with the infiltration of CD8 T cells (r = 0.688), B cells (r = 0.496), monocytes (r = 0.627), neutrophils (r = 0.652), macrophages (r = 0.726), T cell regulation (r = 0.422), NK cells (r = 0.364) and bone marrow dendritic cells (r = 0.673) ( Figure 6E ). All p-values are much less than 0.001. Additionally, Box and violin plot showed the specific fraction of 22 immune cells in each ccRCC sample by the CIBERSORT algorithm ( Figures 6F–H ). These results suggested that WAS was critically correlated with ccRCC immune infiltration.

To further confirm the relationship between WAS expression and immune cell infiltration levels in ccRCC, the association between WAS and different biomarkers of TILs (CD8+/CD4+ T cells, NK cells, B cells, monocytes, DCs, TAMs, M1 macrophages, M2 macrophages, neutrophils and T cell exhaustion-related subtypes) in ccRCC was investigated using TIMER, TCGA and GEPIA databases. It was found that most TIL markers in ccRCC were associated with WAS. There were also several functional T cells under analysis, including Tfh/Th1/Th2/Th17 cells, Tregs, and exhausted T cells. Particularly, WAS showed a strong association with a majority of immune markers sets of TILs in ccRCC ( Table 2 ).

Obviously, WAS was significantly associated with most marker sets, including monocytes, TAMs, M2 macrophages, and T cell exhaustion ( Table 2 ). Our study demonstrated that WAS are markedly correlated with the chemokine ligands of TAMs (IL10, CCL2 and CD68) and T-cell exhaustion (CTLA4, PD-1, PD-L1, GZMB, LAG3, and HAVCR2) in ccRCC, with the same M1 macrophages (IRF5) and M2 macrophages (MS4A4A, CD163, VSIG4) (P < 0.001; Figures 7A–E ). Additionally, based on the GEPIA database, we further assessed the association between WAS expression and the aforementioned markers of M1 macrophages, M2 macrophages, TAMs, monocytes, and T-cell exhaustion. Moreover, the results were similar to those found in TIMER ( Table 3 ). Thus, our research suggested that WAS may regulate the exhaustion of T cells and macrophage polarization in ccRCC.

Additionally, correlation analysis represented that WAS expression was positively correlated with T-cell exhaustion-related markers (CTLA4, PD-1, PD-L1, LAG3, TIM-3 and GZMB) in ccRCC (r = 0.198~0.771, P < 0.001) ( Figure 7E ).

Correlation analyses indicated that WAS expression in ccRCC was significantly correlated with immunostimulants (P < 2.2e-16), such as C10orf54 (VSIR) (r = 0.522), CD27 (r = 0.802), CD28 (r = 0.66), CD40LG (r = 0.59), CD48 (r = 0.808), CD70 (r = 0.39), CD80 (r = 0.625), CD86 (r = 0.776), CXCR4 (r = 0.473), ICOS (r = 0.7), KLRC1 (r = 0.448), KLRK1 (r = 0.699), LTA (r = 0.77), MICB (r = 0.659), TMIGD2 (r = 0.496), TNFRSF8 (r = 0.732), TNFRSF9 (r = 0.651), TNFRSF17 (r = 0.554), TNFRSF18 (r = 0.644), TNFRSF13B (r = 0.707), TNFRSF14 (r = 0.572) and IL2RA (r = 0.439) ( Figure 8A ). The expression of WAS in ccRCC was also significantly associated with immunoinhibitors (P < 2.2e-16), such as BTLA (r = 0.649), CD244 (r = 0.744), CD96 (r = 0.777), CSF1R (r = 0.72), CTLA4 (r = 0.702), IL10 (r = 0.516), IL10RB (r = 0.419), LAG3 (r = 0.761), LGALS9 (r = 0.768), PDCD1 (r = 0.788), PDCD1LG2 (r = 0.53), and TIGIT (r = 0.768) ( Figure 8B ). According to these results, WAS may contribute to the regulation of immune interactions and be involved in the regulation of tumor immune escape.

Chemokines are responsible for regulating immune cell infiltration. This study investigated the association between WAS expression and chemokines. The results presented that WAS expression was positively correlated with CCL3 (r = 0.507), CCL4 (r = 0.65), CCL5 (r = 0.798), CCL8 (r = 0.349), CCL17 (r = 0.398), CCL19 (r = 0.469), CCL22 (r = 0.541), CXCL9 (r = 0.627), CXCL10 (r = 0.534), CXCL11 (r = 0.52), CXCL13 (r = 0.626), CXCL16 (r = 0.618), XCL1 (r = 0.647) and XCL2 (r = 0.673) ( Figure 9A ). All P-values were < 0.001. Meanwhile, we also demonstrated that WAS expression was considerably positively correlated with chemokine receptors (P < 0.001), including CCR1 (r = 0.571), CCR2 (r = 0.684), CCR4 (r = 0.508), CCR5 (r = 0.781), CCR6 (r = 0.316), CCR7 (r = 0.604), CCR8 (r = 0.567), CXCR3 (r = 0.837), CXCR4 (r = 0.473), CXCR5 (r = 0.617), CXCR6 (r = 0.774) and XCR1 (r = 0.431) ( Figure 9B ). These results further demonstrated that WAS may operate as an immunoregulatory factor in ccRCC.

This study was further expanded by determining the machine learning-based score (IPS) that predicted patients’ response to ICI treatment. Four subtypes of IPS values (CTLA4_pos_PD1_neg, CTLA4_neg_PD1_pos, CTLA4_neg_PD1_neg and CTLA4_pos_PD1_pos) were carried out to predict the responses to anti-PD1 and anti-CTLA4 treatment among ccRCC patients. We found that the response rate of anti-PD1 and anti-CTLA4 were elevated in high-risk score patients (P < 0.001). Similar finding was mirrored for the combination treatment of anti-PD1 and anti-CTLA4 (P < 0.001) ( Figure 10A ). A heatmap ( Figures 10B, C ) also showed a positive correlation between WAS and most immune checkpoint genes.

An IC50 analysis of eight drugs was performed to determine the predictive effect of WAS for treatment responses to chemotherapy and target therapy. The estimated IC50 values of Bexarotene, Bortezomib, Dasatinib, Doxorubicin, Mitomycin C, Paclitaxel, Ruxolitinib, and Sunitinib in high-WAS patients were significantly elevated compared to low-WAS patients, which indicating the high-WAS patients showed a stronger drug resistance (P < 0.05) ( Figures 11A, B ). Similarity, patients in low-WAS group were associated with increased sensitivity to other drugs relative to high-WAS patients (P < 0.05) ( Figure S1 ).