Background
Gastric cancer (GC) is the fourth most common cancer in the world, ranking second in the causes of cancer death [
1]. It is a complex disease with great heterogeneity that can be divided into four molecular groups based on genomic characteristics and clinical features, including chromosomal instability (CIN), genomically stable (GS), microsatellite instability (MSI) and EBV-associated GC (EBVaGC) [
2]. EBV is detected in GC cells rather than in noncancerous gastric mucosa, and shows a clonal nature in neoplastic cells. It is therefore considered to have a causal role in GC [
3,
4]. Molecular characterization of EBVaGC has been described recently [
2]. However, the pathogenic mechanism of EBVaGC remains elusive.
Gene misregulation plays a critical role in tumorigenesis and progression [
5]. Regulation of gene expression includes a great variety of mechanisms that increase or decrease the specific gene products. Gene regulatory network is a collection of molecular regulators that interact with each other to govern the gene expression and function, which has been getting increasing attention for facilitation of gaining insight into the transcriptional and epigenetic regulation patterns in cancers [
6,
7]. At the transcriptional level, transcription factors (TFs) are the main regulators. They can bind to the DNA regions of enhancer or promoter adjacent to the target genes that they regulate [
8,
9]. Noncoding RNAs (ncRNAs) have been shown to regulate gene expression serving as an important type of epigenetic regulation mechanism [
10,
11]. Two of the main types of ncRNAs, which are microRNAs (miRNAs) and long ncRNAs (lncRNAs), can suppress each other as competing endogenous RNAs (ceRNAs) and form a regulatory ceRNA network (lncRNAs–miRNAs–mRNAs) to regulate target mRNAs [
12]. In addition, not only mammals but also viruses encode miRNAs. EBV was the first virus in which viral miRNAs were found. Recently, it has been commonly accepted that EBV also encodes for plenty of miRNAs, such as BART cluster and BHRF cluster [
13,
14]. These miRNAs were observed to promote viral latency or cancer development by targeting both viral and cellular genes [
15‐
17].
Given the importance of TFs and ncRNAs, it is of great interest to construct gene regulatory networks based on TFs and ncRNAs for exploring the biological processes of EBVaGC. With the increasing availability of multi-level expression data from cancer and normal tissues, new opportunities for the extraction and integration of large data sets such as gene expression omnibus (GEO) may help to provide a more comprehensive understanding of cancer [
18,
19]. In this study, we integrated expression data to identify differentially expressed mRNAs and the corresponding TFs, miRNAs, and lncRNAs involved in EBVaGC. Regulatory networks including TF–mRNA, lncRNA–miRNA–mRNA, EBV encoded miRNA–mRNA and their overlap were analyzed, which possibly provide a new avenue for investigating the regulation mechanisms of EBVaGC.
Discussion
The genetic and epigenetic regulation mechanisms can be clarified by examining mRNAs, TFs, miRNAs, lncRNAs and their networks. Our study conducted integrated analysis of gene regulatory networks based on TFs, miRNAs and lncRNAs targeting differentially expressed genes, and revealed key elements and their interactions associated with molecular mechanisms of EBVaGC.
Firstly, a total of 104 differentially expressed genes between EBvaGC and normal controls were identified from GEO databases using the GEO2R program in the present research. The functional analysis showed that these genes were mainly associated with digestion, G-protein coupled receptor binding, gastric acid secretion, etc. KEGG enrichment analysis also illustrated that the differential genes were mainly involved in the gastric acid secretion and protein digestion and absorption. Acid secretion exerts the greatest impact of all gastric functions on the occurrence of stomach disorders [
28]. Our findings highlighted the probable importance of the regulation of these key genes and vital biological behaviors in EBVaGC, which warranted further investigations.
Furthermore, a set of gene regulatory networks were constructed by targeting these differentially expressed genes. At transcriptional level, studies have revealed that gene misregulation is often due to the aberrant expression of TFs. Based on the TF network, we identified some hub TFs associated with EBVaGC, including IRX3, NKX6-2, PTGER3 and SMAD5. Iroquois homeobox 3 (IRX3) plays vital roles in embryonic development, it has recently been reported to participate in tumor progression. Choi et al. [
29] found that NKX6 participated in differentiation of gastrin-producing G cells in the stomach antrum. Prostaglandin E-receptor was observed to induce growth inhibition in gastric cancer cells [
30]. Nagasako et al. [
31] reported that up-regulated SMAD5 mediated apoptosis of gastric epithelial cells induced by Helicobacter pylori infection. These TFs may individually or comprehensively participate in EBVaGC pathogenesis by regulating their target genes, such as SST (Somatostatin) and GDF5 (growth differentiation factor 5). SST is important for regulating motor activity and the secretion of gastrin-stimulated gastric acid in the gastrointestinal tract [
32], and GDF5 serves as a regulator of cell growth and differentiation in both embryonic and adult tissues. Their aberrant expressions were reported to be associated with varieties of cancers [
33‐
36].
Noncoding RNAs (ncRNAs) are also important part of the regulatory network involved in post-transcriptional regulation of genes. By building ceRNA network, our results also revealed several novel miRNAs and lncRNAs that were possibly involved in gene regulation associated with EBVaGC. The top five miRNAs were hsa-miR-4446-3p, hsa-miR-5787, hsa-miR-1915-3p, hsa-miR-335-3p and hsa-miR-6877-3p. Kim et al. [
37] observed that miR-4446-3p was upregulated by compression in breast cancer cells. Aberrantly expression of miR-5787 was supposed significantly down-regulated in serum and might be involved in the process of glucose metabolism in colorectal cancer [
38]. miR-1915 inhibits Bcl-2 to modulate multidrug resistance by increasing drug-sensitivity of human colorectal cancer cells [
39]. Overexpression of miR-335 significantly inhibited cell proliferation, migration and invasion in GC cells [
40]. Little is known about miR-6877-3p, the only research reported that its expression was associated with ovary development in cyprinus carpio [
41]. In addition, two unreported lncRNAs, RP5-1039K5.19 and TP73-AS1 were identified in the ceRNA regulation network, which may become the candidate targets for in-depth study of EBVaGC.
Additionally, miRNAs are not solely produced by metazoans, but also by viruses, which opened a new window for the research. Up to date, 44 mature EBV coding miRNAs have been identified, many of which have been proven to promote carcinogenesis by targeting host genes [
13]. In our study, we built an EBV related miRNA regulation network and found that CXCL10 and SMAD5 were regulated by EBV-miR-BART1-3p and EBV-mir-BART22. EBV-miR-BART1 was observed to be involved in regulating metabolism-associated genes [
42] and induced tumor metastasis [
43] in nasopharyngeal carcinoma. Zhou et al. [
44] found that CXCL10/CXCR3 axis can promote the invasion of GC via PI3 K/AKT pathway-dependent MMPs production. As for EBV-mir-BART22, it is a brand new miRNA without prior study. Interestingly, its target gene SMAD5 was also identified as a hub TF associated with EBVaGC in our study.
Intriguingly, when taking an overview on the various regulation networks in the current study, some overlapping genes and regulators were observed in the cross network. Firstly, CXCL10 was the common target gene in the three diverse regulation networks. It could be regulated by the transcription factor PTGER3, miR-6877-3p and EBV-miR-BART1-3p at the same time. Secondly, GDF5 was the target gene of transcription factor SMAD5 and miR-6877-3p. Moreover, SMAD5 was simultaneously regulated by EBV-mir-BART22. In addition, both CXCL10 and GDF5 were in the same ceRNA network that they can be regulated by miR-6877-3p and the two unreported lncRNAs, RP5-1039K5.19 and TP73-AS1. Furthermore, the expression levels of GDF5, CXCL10, SMAD5 and PTGER3 were also different between EBVaGC and EBVnGC. There were also differences between EBVaGC and other molecular subtypes of GC for these genes. In addition, in the histological verification experiment, differential expressions of the two main target genes GDF5 and CXCL10 were observed between EBVaGC and non-tumor tissues as well as EBVnGC. These results indicate that GDF5 and CXCL10 and their misregulation may play important roles specifically in EBVaGC related mechanisms. CXCL10 is a strong angiostatic factors, and it may be involved in the recruitment of tumour-infiltrating T cells [
45]. It has been reported that TGF-β produced by breast cancer cells induces the GDF5 expression in the endothelial cells, which in its turn stimulates the angiogenesis both in vivo and in vitro [
46]. Dysregulation of these two genes may lead to the activation of pathways related to cancer hallmarks like angiogenesis and tumour-promoting inflammation to promote EBVaGC, which needs further investigated. These identified key elements and their network regulation may offer new perspectives on mechanisms of EBVaGC.
Conclusion
In summary, in current study, we provided a framework for revealing the key elements and their regulatory network involved in EBVaGC. Some hub TFs associated with EBVaGC, including IRX3, NKX6-2, PTGER3 and SMAD5 were found to regulate their target genes. We also identified five miRNAs hsa-miR-4446-3p, hsa-miR-5787, hsa-miR-1915-3p, hsa-miR-335-3p, hsa-miR-6877-3p and two unreported lncRNAs, RP5-1039K5.19 and TP73-AS1 in the ceRNA regulation network. EBV related miRNAs EBV-miR-BART1-3p and EBV-mir-BART22 were observed to regulate CXCL10 and SMAD5. Further, some overlapping genes and regulators were observed in the three diverse regulation networks, such as CXCL10, GDF5, PTGER3, SMAD5, miR-6877-3p, RP5-1039K5.19, TP73-AS1, EBV-miR-BART1-3p and EBV-mir-BART22. Moreover, CXCL10, GDF5, PTGER3 and SMAD5 were also differentially expressed among the four molecular subtypes of GC. The histological verification experiment showed differential expressions of the two main target genes GDF5 and CXCL10 between EBVaGC and non-tumor tissues as well as EBVnGC. Therefore, the misregulation of target genes GDF5 and CXCL10 may be specifically involved in EBVaGC mechanisms. This study provides a new insight into understanding the mechanism based on gene regulation of EBVaGC, and further molecular experiments are needed to confirm the findings.