By analyzing IHC and TCGA data, our experiment elucidated that abnormally expressed ERCC6 and ERCC8 were associated with clinicopathological behaviors and survival of GC. Furthermore, by performing bioinformatics analysis of GO, KEGG, GSEA and gene-gene interaction analysis, our research extended the existing knowledge of ERCC6/ERCC8 in GC.
We initially detected protein expressed levels of ERCC6/ERCC8 in GC and para-cancerous tissues. The results indicated that their expressions were significantly decreased in GC, in comparison to adjacent tissues despite individual or joint expression. Then we investigated associations between protein expression of ERCC6 and ERCC8 and clinicopathological parameters, and the analysis showed that overexpressed ERCC6, ERCC8 and ERCC6-ERCC8 were significantly related to favorable clinicopathological features, which are key factors that have great impact on disease progression. RNA-seq data revealed identical results with ERCC6 that higher ERCC6 expression was associated with favorable T stage, while overexpressed ERCC8 was associated with unfavorable clinicopathological parameters. We suspected that the discrepancy may be due to some potential mechanisms that resulted in the instability of ERCC8 protein in GC progression. And cancer cells lacking ERCC6 or ERCC8 protein, which are responsible for DNA repair, may exhibit a more malignant and poorly differentiated phenotype. A recent research reported that ERCC6 deficiency could result in heterochromatin loss and exacerbates cellular aging (27). Defects in ERCC6 and ERCC8 will influence the coupling of transcription and repair to a certain extent, thus leading to declining DNA repair capacity(28). Physiologically, DNA repair capacity could be related to expression levels of proteins involved in DNA repair activities(29). Previous studies have also reported that a downregulation of DNA repair genes is related to late stage cancers and malignant transformation(30). Therefore, it is conceivable that the expressed status of DNA repair genes could reflect the capacity of a cell to meet repair demands after being stimulated by a carcinogen. We suggested that ERCC6 and ERCC8 downregulation could induce persistent existence of unrepaired DNA lesions, leading to decreased DNA repair capacity and increased cancer susceptibility, and finally resulting in cancer progression.
Further to explore the prognostic value of ERCC6 and ERCC8, we studied the correlation between ERCC6 and ERCC8 expression and survival in GC patients with both IHC and RNA-seq data. According to univariate survival analysis based on IHC, higher ERCC6 protein expression was associated with better prognosis while double negative ERCC6 and ERCC8 expression indicated worse overall survival of GC patients. And RNA-seq data also showed that overexpressed ERCC8 was related to a better OS of GC patients. When adjusting for certain parameters in the Cox multivariate analysis, analyses results of ERCC6 and ERCC8 expression with IHC and RNA-seq data no longer maintained independent predictive power, which may be due to the complexity of tumor progression. A previous lab study showed that knockdown of ERCC6 could sensitize HCT116 cells to 5-Fluorouracil in xenograft mouse models and colorectal cancer patients with high ERCC6 expression exhibited shorter overall survival(31). As for other DNA repair family genes, high ERCC5 expression was shown to correlate with shorter survival times compared with low ERCC5 expression(32), whereas decreased ERCC1 expression was reported to predict a favorable prognosis in gastric cancer(33). Overall, our data suggested that expressed levels of ERCC6 and double negative ERCC6-ERCC8 protein, and ERCC8 mRNA, to some extent, may possess potential prognostic value in GC, and some certain factors should also be taken into account to estimate GC prognosis more comprehensively in the further analysis.
Next, bioinformatic analyses were conducted to better investigate biological functions and regulation networks of ERCC6 and ERCC8 in GC progression. First we queried the 10 most relevant genes of ERCC6 and ERCC8 through String and then performed GO and KEGG analyses with the obtained results. Enrichment analysis of ERCC6 and ERCC8 and their relevant genes showed similar results. Both the two genes were mainly involved in the composition of transcriptional initiation complexes and exerted influences on diverse nucleotide excision repair pathways. Similarly in other experiments, researchers have identified ERCC6 and ERCC8 as core NER genes(34–36). KEGG pathway analysis results further revealed that ERCC6 also functioned in Huntington’s disease and ERCC8 showed significant impacts in ubiquitin mediated proteolysis. Consistent with our analyses, one study have reported that ERCC8 are involved in the formation a complex which exhibits ubiquitin ligase activity(37). Furthermore, we conducted GSEA analysis to explore ERCC6 and ERCC8 associated regulation networks in GC. Usually a set of genes which exhibit certain patterns of up or downregulation when an already known pathway related to tumorigenesis is activated, are defined as an oncogenic pathway signature. Here in our study, oncogenic signatures analysis suggested that ERCC6 was significantly associated with the oncogenic signatures of EIF4E, TBK1, BCAT, mTOR, JAK2 and CSR related regulation networks. As for ERCC8, the results indicated a significant relationship with TBK1, PIGF, BCAT, RB, ERBB2, GCNP, SRC and CYCLIN_D1 associated oncogenic regulation networks. These days emerging evidence has illustrated that PI3K/AKT/mTOR pathway deregulation plays an important part in GC progression(38). Currently, one study conducted by Riquelme mentioned that two mTOR pathway genes, EIF4E and mTOR, were overexpressed in GC cells(39). It has been found that ERBB2 could mediate the activation of PI3K(40). Moreover, some studies have reported the environment-dependent inhibition or activation role of TBK1 in mTOR signaling(41–43). Therefore, given all the above results, we suspected that ERCC6 and ERCC8 could regulate GC progression through the regulation of PI3K/AKT/mTOR pathway. Because of the similar and identical functions and pathways found in our analysis, we then did gene-gene interaction analysis to figure out the potential associations between ERCC6 and ERCC8. The results demonstrated that there did exist direct physical interactions and pathways between ERCC6 and ERCC8, which was supported by one previous study(44). And indirect interactions including prediction, co-expression, colocalization and shared protein domains were also revealed in the results. From these we suggested the existence of alliance mechanisms between ERCC6 and ERCC8, which needs further in-depth study.