Analysis of EPHB2 expression in pan-cancer
Based on the TIMER2.0 database, the EPHB2 expression level in 38 cancer types was first analyzed and the results suggested that EPHB2 was distinctly differentially expressed between the normal group and the tumor group in 19 malignant tumors. As shown in Fig. 1A, in these 19 cancer types, EPHB2 was significantly high expression in 13 malignant tumors, including BRCA, BLCA, COAD, CHOL, CESC, ESCA, HNSC, LUSC, LUAD, LIHC, READ, STAD and UCEC, whereas significantly low expression in 6 malignant tumors, involving KICH, KIRC, KIRP, PRAD, SKCM and THCA (Fig. 1A). Then, we further verified the EPHB2 expression in these 19 malignant tumors using the GEPIA2 database. As Shown in Fig. 1B-1K, when compared with corresponding normal controls, the EPHB2 expression level in CESC, COAD, CHOL, ESCA, HNSC, LUSC, LUAD, READ, STAD and UCEC was markedly increased, which are accord with the results of Fig. 1A. And there is no statistical difference in the expression of EPHB2 in BRCA, BLCA, KIPR, KIRC, KICH, LIHC, PRAD, SKCM and THCA. Therefore, we can further confirm that EPHB2 expression was significantly upregulated in CESC, COAD, CHOL, ESCA, HNSC, LUSC, LUAD, READ, STAD and UCEC, which suggests that EPHB2 might play an important part in the carcinogenesis of these10 cancer types.
Prognostic value analysis of EPHB2
Then, the disease-free survival (RFS) and overall survival (OS) analysis for EPHB2 in CESC, COAD, CHOL, ESCA, HNSC, LUSC, LUAD, READ, STAD and UCEC were conducted respectively using the GEPIA2 database. As exhibition in Figure 2, for OS, the patients with high expression of EPHB2 in CESC and COAD was significantly worse compared with the corresponding normal control group. For RFS, high expression of EPHB2 suggested poor prognosis only in CESC (Figure 3). Taken together, high expression of EPHB2 may have a significant adverse effect on the prognosis of multiple cancers, especially CESC.
Prediction and analysis of mRNA-miRNA co-expression network
Next, we predict the upstream ncRNAs that may regulate EPHB2, consisting of miRNAs and lncRNA. First, based on the starBase database, we explored the upstream miRNAs that might potentially combined with EPHB2 with a program Num ≥2 as the screening criteria, and 22 miRNAs were finally found. As presented in Figure 4A, we built a miRNA-EPHB2 regulatory network using the Cytoscape software. EPHB2 should negatively correlate with miRNA based on the mechanism of miRNA action on target gene expression. Thus, we performed the correlation analysis of EPHB2 and miRNAs. There were 19 miRNAs related to EPHB2, and the other three miRNAs were not. As listed in Figure 4B, hsa-miR-27b-3p, hsa-miR-150-5p and hsa-miR-30e-5p were significantly negatively related to EPHB2 while hsa-miR-30a-5p and hsa-miR-10a-5p were positively related toEPHB2 in CESC. We further determined the analysis of hsa-miR-27b-3p, hsa-miR-150-5p and hsa-miR-30e-5p expression in CESC (Figure 4C-4E) and found only hsa-miR-150-5p shows differential expression in CESC. The analysis of survival suggested that high expression of hsa-miR-150-5p was positively related to the prognosis of patients with CESC (Figure 4F-4G). Besides, we also identified the site of binding of EPHB2 and hsa-miR-150-5p from the starBase database (Figure 4H). All these findings indicated that the most potential regulatory miRNA of EPHB2 in CESC might be hsa-miR-150-5p.
Prediction and analysis of mRNA-miRNA-lncRNA co-expression network
Based on the starBase database, the has-miR-150-5p’s upstream lncRNAs were also explored. And 187 potential lncRNAs were finally predicted totally. Based on the Cytoscape softwarea, a lncRNA-has-miR-150-5p regulatory network was established for visualization (Figure 5A). Then we analyzed these lncRNAs expression and prognostic value based on the UALCAN database and identified that only the analysis results of expression and survival of AC008771.1 and AC073957.3 were statistically significant in CESC (Figure 5B-E). The correlation between the two kinds of lncRNAs and EPHB2 or has-miR-150-5p was further explored. As presented in Figure 5F-5I, AC073957.3 correlates positively with EPHB2, and negatively with has-miR-150-5p. However, the correlation between AC008771.1 and EPHB2 and has-miR-150-5p has no statistical significance. Take the above analysis results together, we concluded that AC073957.3 is one of the most likely lncRNAs upstream of the hsa-miR-150-5p/EPHB2 axis in CESC.
EPHB2 with various immune cells correlation analysis in CESC
In this part, we tested whether EPHB2 showed the gene expression characteristics of high immune cell infiltration in CESC. As presented in Figure 6A, as EPHB2 copy numbers fluctuate in CESC, there is an obvious difference in immune cell infiltration levels according to the TIMER1.0 database, including B cell, dendritic cell, neutrophil and CD8+T cell. Numerous tumors occur and develop with the involvement of immune infiltration, so we further analyzed the immune cell infiltration and EPHB2 expression level correlation using the R language package. EPHB2 was remarkably positively relevant to 7 immune cells types (Th2 cell, NK cell, Eosinophils, Tgd, Mast cell, NK CD56bright cells and Tem) and negatively correlated with 2 immune cells types (Tcm and B cell) in CESC (Figure 6B-6K).