The RNA sequencing (RNA-seq) data (HTSeq-FPKM), along with corresponding clinical data and immune subtypes, were acquired from UCSC Xena (https://xena.ucsc.edu/, originated from TCGA database) (16).

The expression patterns of EphA genes in TCGA tumors were depicted through a boxplot graph. Subsequently, heatmaps were generated for 18 distinct tumor types, employing log2 (fold change) values to highlight discrepancies in EphA gene expression between primary tumors and adjacent normal tissues. Additionally, Spearman’s correlation test was employed to compute gene expression correlations among EphA members across 33 cancer types. To scrutinize the differential expression of EphA family genes across diverse cancer types, the “Wil-cox. test” was applied. For graphical representation, we utilized the “ggpubr” and “pheatmap” R packages to craft a box plot and a heatmap, respectively. The exploration of correlations within the EphA family genes involved the utilization of the “corrplot” R package.

We conducted univariate Cox regression and Kaplan-Meier (KM) analyses, facilitated by the “survminer” and “survival” R packages, to evaluate the influence of EphA on the survival outcomes associated with various cancers. Furthermore, the Cox proportional hazard model was applied to assess the connection between EphA gene expression and the prognosis of diverse cancer types. The “survival” and “forest plot” packages enabled the creation of a forest plot to visually present the results.

We gauged the extent of immune and stromal cell infiltration across diverse cancers using immune scores and stromal scores metrics from ESTIMATE (17). The relationship between these scores and EphA expression was assessed using the Spearman Correlation Coefficient. Additionally, six distinct immune subtypes were delineated to quantify immune infiltration in the tumor microenvironment (TME) (18). Employing Analysis of Variance (ANOVA), we examined the link between immune infiltration types and SEMA3 expression within the TME based on immune subtypes acquired from TCGA pan-cancer data.

The Tumor Immune Estimation Resource(TIMER, https://cistrome.shinyapps.io/timer/) was developed for quantifying immune cell infiltration across 10,897 cancer samples from TCGA (19, 20). Leveraging TIMER gene modules, we analyzed EphA expression across diverse cancer types and explored its relationship with immune cell infiltration levels. Furthermore, correlation modules were utilized to examine the connections between EphA expression and gene biomarkers linked to tumor-infiltrating immune cells, utilizing established gene biomarkers (21, 22).

KIRP tumor and adjacent normal kidney specimens were analyzed from a total of 157 patients with KIRP as part of a study approved by Tianjin Medical University Cancer Institute and Hospital. All patients were treated with radical or partial nephrectomy and rendered disease-free.

Immunohistochemical staining was performed on the sections from surgical specimens fixed in 10% formalin and embedded in paraffin according to a standard method (23). Briefly, tissue sections were incubated with anti-EphA3 antibody (ab126261, Abcam; 1: 50 dilution) overnight at 4°C, and then incubated with secondary antibody (ab207995, Abcam; 1:100 dilution) followed by avidin-biotin peroxidase complex (DAKO) at room temperature for 30min. Finally, color development was performed with 3, 3′-diaminobenzidine. The immunostained slides were evaluated separately by two pathologists. The intensity of antibody staining was used to semiquantitatively quantify the expression of EphA3 in cancer cells. Staining intensity was categorized as follows: absent staining as 0, weak as 1, moderate as 2, and strong as 3.

Statistical analyses were conducted using R 4.0.2 (https://www.r-project.org/). A linear mixed-effect model was employed to compare gene expression patterns between tumor and normal samples. Boxplots were utilized to illustrate gene expression variation across different cancer types. Univariate and multivariate Cox regression or Log-rank test were used to examine the relationship between gene expression and overall survival (OS) of patients. The correlation between gene expression and stemness scores, stromal scores, immune scores, and estimate scores was assessed using Spearman or Pearson correlation methods.

The levels of EphA mRNA were assessed across 33 cancer types using data sourced from UCSC Xena, aiming to uncover the diversity inherent within the EphA family. The results revealed prominent high expression levels for most EphA genes in multiple cancer types. However, this trend was not mirrored by three genes: EphA5, EphA6, and EphA8, which exhibited relatively diminished expression levels ( Figure 1A ). When examining individual EphA genes like EphA2 and EphA10, significant up-regulation was observed in cases of esophageal carcinoma (ESCA) and cholangiocarcinoma (CHOL). Conversely, EphA2 and EphA10 displayed a down-regulation in kidney chromophobe (KICH) and glioblastoma multiforme (GBM) ( Figure 1B ). Particularly noteworthy was the correlation analysis that pinpointed the strongest pairwise correlation between EphA1 and EphA2 among the nine genes (Correlation coefficient =0.47). This finding implied potential shared characteristics or functions between these two genes. In contrast, EphA1 and EphA5 demonstrated a distinct negative correlation (Correlation coefficient =−0.36, Figure 1C ), suggesting intricate co-expression interactions involving numerous EphA genes across diverse cancer types.

On closer inspection, it became evident that nearly all EphA genes exhibited discernible disparities in expression between cancerous tissue samples and their normal counterparts ( Figure 2 ). Moreover, significant differences in expression emerged across the spectrum of different cancer types.

To deeply delve into the prognostic implications of EphA family genes, we examined the impact of the expression level of each gene on the prognosis of patients with specific cancers ( Supplementary Figure 1 ). Through the application of the Cox proportional hazard model, the prognostic value of the nine EphA genes was evaluated across pan-cancer scenarios ( Figure 3 , Supplementary Table 1 ). The findings underscored a connection between EphA gene expression levels and the OS of patients, albeit with nuanced ramifications contingent on the specific cancer types. For example, heightened EphA5 expression correlated with an unfavorable prognosis of papillary renal cell carcinoma (KIRP) and uveal melanoma (UVM), whereas predicting improved survival in pancreatic adenocarcinoma (PAAD). Similarly, elevated EphA2 expression indicated a poor prognosis for patients with colon adenocarcinoma (COAD), diffuse large B-cell lymphoma (DLBC), low-grade glioma (LGG), and pancreatic adenocarcinoma (PAAD), yet correlated with higher survival rates in those with kidney chromophobe (KICH) and pheochromocytoma and paraganglioma (PCPG). Importantly, EphA genes emerged as independent prognostic markers for several distinct cancer types.

EphA genes occupy a critical position within the immune system due to their intricate involvement in the development, mobilization, and activation of both innate and adaptive immune cells (15, 24). Within the context of human malignancies, six distinct types of immune infiltration which were defined as the relative abundance of a set of immune cell populations, ranging from tumor promoting to tumor inhibiting, namely C1 (wound healing), C2 (INF-r dominant), C3 (inflammatory), C4 (lymphocyte depleted), C5 (immunologically quiet), and C6 (TGFβ dominant) (18). In the scope of our investigation, we conducted a thorough examination of immune infiltration patterns across the TCGA pan-cancer dataset, aligning them with the expression profiles of EphA genes ( Figure 4A ). Remarkably, the data illuminated connections between the expression levels of EphA genes and diverse categories of immune infiltration, except for EphA5, EphA6, and EphA8. Notably, EphA1 and EphA2 exhibited heightened expression in the C1 and C2 subtypes, indicating a plausible involvement in tumor promotion. This assumption found support in the observation of poorer survival rates among patients exhibiting these types of immune infiltration (C1 and C2). In contrast, EphA6 and EphA10 displayed elevated expression in the C3 and C5 subtypes, suggesting a potential tumor-suppressive effect. Nevertheless, it’s important to highlight that these correlations contradicted the roles of certain EphA members as promoters of cancer, which in some cases translated to diminished survival rates. A case in point was the poor prognosis linked to heightened EphA6 expression in UCEC ( Supplementary Figure 1 ). These conflicting outcomes can potentially be attributed to the intricate biological diversity and multifaceted molecular interactions inherent to the progression of tumors. The implications of EphAs on the immune dynamics of the TME hold the promise of guiding novel pathways for the development of treatment strategies.

Furthermore, we delved into the stromal and immune scores indicative of tumor growth and metastatic potential. Leveraging the ESTIMATE algorithm, EphA genes were subjected to a comprehensive analysis involving these scores ( Figures 4B, C ). EphA genes with high stromal scores suggested heightened complexity within the TME, possibly intensifying tumor malignancy. An illustration of this was found in the close association between EphA3 expression and elevated stromal score across diverse cancers, whereas EphA10 expression demonstrated an inverse pattern. The associations between distinct gene expression levels within the EphA family and the estimated scores across various tumors underscored the diverse impacts that these genes might exert on the TME.

Renal carcinoma ranks as the 13^th most prevalent cancer worldwide, with an escalating incidence rate (25). Papillary renal cell carcinoma (KIRP), constituting 10% to 15% of kidney cancer cases, stands as the second most common subtype (26). In the metastatic context, the prognosis for KIRP patients remains bleak due to the absence of effective therapeutic options (27). The optimal treatment strategy for advanced KIRP continues to be a subject of debate. Nonetheless, recent clinical studies have unveiled promising outcomes for both molecularly targeted therapies and immunotherapy within this subtype (28). Given the elevated risk association of most EphA genes with poor survival in KIRP patients, our research expanded to explore the interplay between EphA genes, diverse immune subtypes, stem cells, and the TME in KIRP. The connection between EphA gene expression and distinct immune subtypes in KIRP echoed the patterns observed across all 33 TCGA cancer types. Of particular note, EphA1, EphA 2, EphA 3, and EphA 7 exhibited significant associations with immune infiltration profiles in KIRP ( Figure 5A ). Subsequently, we delved into the correlation between EphA expression and stromal score, revealing positive associations for EphA3 and EphA6 (P < 0.05) in KIRP, in which EphA3 demonstrated the most robust correlation (r =0.49) ( Figure 5B ). On the other hand, EphA2, EphA4, EphA5, EphA7 EphA 8, and EphA 10 did not display significant correlations with stromal scores, implying their potential origin from the tissue stroma in KIRP. Moreover, EphA3 exhibited correlations with the immune score, a metric assessing the presence of infiltrating immune cells (P = 0.0011) and tumor purity (Estimate score) (P < 0.0001).

Upon a deeper exploration of EphA3’s role, pronounced connections surfaced between EphA3 expression and various immune cell types, such as CD4⁺ T cells (R =0.171, P =5.79E-03), CD8⁺ T cells (R =0.236, P =1.32E-04), B cells (R =0.221, P =3.60E-04), dendritic cells (R =0.295, P =1.62E-06), and neutrophils (R =0.215, P =4.94E-04). However, these correlations were notably absent for tumor purity (R=-0.014, P=8.21E-01) and macrophages (R =-0.049, P =4.42E-01) ( Figure 5C ). Leveraging the TIMER database, the Kaplan-Meier curves highlighted a substantial linkage between KIRP patients survival and EphA3 expression (P =0.001), as well as the infiltration of CD8⁺ T cells (P =0.024) and B cells (P =0.035) ( Figure 5D ). Overall, our findings underscore the potential role of EphA3 in governing immune cell infiltration in KIRP. Alongside B cell and CD8⁺ T cell infiltration, EphA3 emerges as a pivotal modulator influencing clinical outcomes for KIRP patients.

Further delving into the connection between EphA3 expression and immune cell infiltration, we uncovered EphA3’s association with markers of various immune cell types, including B cells (CD19 and CD79A), monocytes (CD86 and CD115), M1 macrophages (INOS, IRF5, and COX2), Th2 cells (GATA3 and IL13), and Treg cells (FOXP3, CCR8, STAT5B, and TGFb) ( Table 1 ). This revelation suggested that EphA3 may play a regulatory role as an immunomodulator in the renal cancer microenvironment.

EphA3 was found to be up-regulated in advanced-stage KIRP samples analyzed by TCGA datasets (HR =0.228, P <0.001, Figure 5E ). To further elucidate the association between EphA3 expression and KIRP, differences in the level of EphA3 protein expression between cancer cells and adjacent normal cells were compared in KIRP tissue specimens by immunohistochemistry. The high and low groups were defined based on the median EphA3 expression ( Figure 5F ). The result showed that the expression of EphA3 protein was significantly associated with metastases (TNM) stage (P =0.0017), tumor diameter (P <0.0001), and age (P =0.0030). No significant association between the expression of EphA3 and sex (P =0.8635) was found ( Table 2 ).