Expression of EPB41L1 in human KIRC
Firstly, we analyzed the transcription levels of EPB41L1 in KIRC tumors from a series of studies linked to the Oncomine database and found that the mRNA expression of EPB41L1 in KIRC tissues was obviously lower compared to normal tissues (P ˂0.05). As shown in Figure 1a-d, the mRNA expression of EPB41L1 was among the top 30%, although the differences were not more than 2-fold between KIRC patients and normal tissues.
We used the HPA database to ﬁnd normal and KIRC sections from several patients with staining for protein 4.1N. Antibodies used in the HPA database were HPA054104. Immunohistochemistry for the 4.1N in the HPA database showed that protein 4.1N highly expressed in normal cell cytoplasm and plasma membrane but was almost undetectable in KIRC tissue (Figure 1e).
EPB41L1 expression in subtype of human KIRC
To further prove the specificity of EPB41L1 in KIRC, we integrated various clinic factors of KIRC samples in the TCGA database, for example, cancer stages, tumor grade, KIRC subtype, nodal metastasis status, patients’ gender, and age, and to compare the transcription levels of EPB41L1 in each group. The results showed that KIRC patients, compared with normal subjects, still maintained a low transcription level of EPB41L1 (Figure 2a-f). Hence, EPB41L1 had the potential to be kidney biopsy-based markers for screening KIRC high-risk patients.
Down-expression of EPB41L1 Predicts Poor Prognosis of KIRC
Then we investigated the prognostic value of EPB41L1 in KIRC. As shown by the KM curve, there was a close relationship between the expression of EPB41L1 and the survival of KIRC patients that the low expression of EPB41L1 caused poor overall survival and disease-free survival (Figure 3a-b).
The mutation of EPB41L1 in KIRC and its prognosis
The occurrence of most tumors and the prognosis are related to gene alterations. Here, we evaluated the frequency of EPB41L1 mutations in 537 sequencing data of KIRC patients in the TCGA database through cBioPortal. Result indicated that there were only 3 cases of EPB41L1 mutations (0.6%), one of the mutation sites occurred in FERM_C (M299I), the other one occurred between FA and SAB (Q464AFS*6) (Figure 3c). But the KM curves analysis of the prognostic value between the EPB41L1 altered group and the unaltered group showed no significant (p=0.44) (Figure 3d). This result suggested that the poor prognosis caused by EPB41L1 low expression is not due to its mutation.
Co-expression genes correlated with EPB41L1 in KIRC
We speculated that the role of EPB41L1 in KIRC might be closely related to the function of its neighbor genes in KIRC. We used the LinkedOmics database to analyze the co-expressed genes of EPB41L1 in 533 KIRC cases (Additional file 1). As shown in Figure 4a, there were 3,218 genes represented by dark red dots, having an obviously positive connection with EPB41L1. Conversely, there were 2,881 genes, represented by dark green dots, having a notably negative correlation with EPB41L1 (false discovery rate [FDR] <0.001). 50 significant gene sets were shown by the heat map whether they were positively and negatively correlated with EPB41L1 (Figure 4b-c).
GO and KEGG Analysis of EPB41L1-Related Co-expressed Genes in KIRC
The outcomes of GO analysis, carried out by GSEA in LinkedOmics, indicated that differentially expressed genes correlated with EPB41L1 were mainly located in cell projection membrane, ciliary part, and basolateral plasma membrane, where they primarily participated in cell tissue migration and cell-cell adhesion. They acted as transmembrane receptor protein kinase activity and extracellular matrix structural constituent (Figure 4d–f). The functions of these differentially expressed genes were principally enriched in adhesion junction and Rap1 signaling pathway which is a typical cell adhesion-related pathway, through the KEGG pathway analysis (Figure 4g).
To confirm the adhesion of EPB41L1 in KIRC, we constructed the EPB41L1 plasmid and transferred it to 786-O cells to overexpress EPB41L1 (Figure 5a). The results of the wound healing assay showed that in pEGFP-C3-EPB41L1 groups the scratch width was significantly wider than that in the control group (Figure 5b-c). The results indicated that increasing the expression of EPB41L1 could weaken cell migration.
Construction of Co-expression Gene PPI Network
The top 500 significantly co-expressed genes were built into a protein-protein network by using the STRING database, and Cytoscape (MCODE plug-in) was used to establish the most important module, highlighted in yellow (Figure 6a-b). Based on the degree score, the module with the highest score consisting of GNG7, PIK3R3, TBXA2R, GPR4, FPR2, TACR1, PMCH, EDNRB, APP, AGTR1, GNA15, and CHRM3 was identified as potential hub genes (Figure 6c).
Prognostic analysis of Hub gene in KIRC
The overall survival of hub genes in KIRC was analyzed by GEPIA database, which indicated that the 8 genes (including APP, TBXA2R, PIK3R3, AGTR1, GNG7, CHRM3, TACR1, and EDNRB) displayed a severe decline in the overall survival rate in low expression groups (Figure 5c). Next, we used the UCSC Cancer Genomics Browser to hierarchically cluster these 8 hub genes with EPB41L1 and found that the expression pattern between EPB41L1 and APP gene was consistent (Figure 7a). And there was a high correlation coefficient between EPB41L1 and APP through GEPIA analysis (Spearman's correlation = 0.76) (Figure 7b). Therefore, APP might be the most attractive target in cell migration and adhesion among hub genes.
APP Expression in KIRC Patients
Subsequently, we screened a series of datasets from the Oncomine database, such as Gumz Renal, Jones Renal, Beroukhim Renal, and Higgins Renal, to identify the expression of APP in KIRC (Figure 7c-f). The expression of APP at cancer stages, tumor grades, KIRC subtype, nodal metastasis status, patients’ gender, and age, and in the TCGA database showed that the down-regulation of APP expression nearly existed in all subtypes of KIRC (Figure 8a-f).