Identification of differential proteomics in Epstein-Barr virus-associated gastric cancer and related functional analysis

Abstract

Background

Epstein-Barr virus-associated gastric cancer (EBVaGC) is the most common EBV-related malignancy. A comprehensive research for the protein expression patterns in EBVaGC established by high-throughput assay remains lacking. In the present study, the protein profile in EBVaGC tissue was explored and related functional analysis was performed.

Methods

Epstein-Barr virus-encoded RNA (EBER) in situ hybridization (ISH) was applied to EBV detection in GC cases. Data-independent acquisition (DIA) mass spectrometry (MS) was performed for proteomics assay of EBVaGC. Functional analysis of identified proteins was conducted with bioinformatics methods. Immunohistochemistry (IHC) staining was employed to detect protein expression in tissue.

Results

The proteomics study for EBVaGC was conducted with 7 pairs of GC cases. A total of 137 differentially expressed proteins in EBV-positive GC group were identified compared with EBV-negative GC group. A PPI network was constructed for all of them, and several proteins with relatively high interaction degrees could be the hub genes in EBVaGC. Gene enrichment analysis showed they might be involved in the biological pathways related to energy and biochemical metabolism. Combined with GEO datasets, a highly associated protein (GBP5) with EBVaGC was screened out and validated with IHC staining. Further analyses demonstrated that GBP5 protein might be associated with clinicopathological parameters and EBV infection in GC.

Conclusions

The newly identified proteins with significant differences and potential central roles could be applied as diagnostic markers of EBVaGC. Our study would provide research clues for EBVaGC pathogenesis as well as novel targets for the molecular-targeted therapy of EBVaGC.

Background

Epstein-Barr virus (EBV) is a ubiquitous human herpes virus originally discovered in Burkitt lymphoma [1]. It has been recognized as the primary virus to be directly involved in numerous malignant tumors. EBV-associated gastric cancer (EBVaGC) is the most common one among EBV-related malignancies. And it accounts for nearly 10% of gastric carcinoma worldwide with variable frequencies between geographic regions [2]. EBVaGC was also identified as one of the four molecular subtypes of GC according to a full-scale molecular genetic analysis published by the Cancer Genome Atlas (TCGA) [3]. The diverse properties of EBVaGC distinct from other GC types have been attracting extensive attention in the past thirty years, including unique epidemiological, pathological, clinical and molecular features.

The molecular patterns in EBVaGC are complicated comprised of various genetic and epigenetic abnormalities [4]. In any event, cellular gene expression plays a critical role in viral oncogenesis, thus it is quite necessary to clarify the differential proteins with their specific effects on EBVaGC. The proteomics research for infection of pathogenic microorganisms has been rapidly developing since proposed [5, 6]. It aims to figure out the key proteins that determine crucial biological activities encompassing pathogen infection and host defense, and also the mechanisms for these proteins to function. Great significance has been manifested in the proteomics of both pathogens in vitro or vivo and infected tissue or cells of host, especially for some common organisms such as Salmonella typhimurium, Shigella flexneri and Helicobacter pylori, etc. [7,8,9]. The identification of proteomic differences for important organisms may not only conduce to in-depth knowledge of their pathogenesis, but also provide novel targets for the treatment of related diseases [10]. As for EBV infection-induced GC, however, almost all current studies at protein level were focused on single element or large-scale datasets based on bioinformatics database [11]. A comprehensive research for the protein expression patterns in EBVaGC established by high-throughput assay remains lacking.

In the present study, the protein profile in EBVaGC tissue was explored and differentially expressed proteins between EBV-positive and negative GC was identified. Functional analysis was subsequently performed for the differential proteins. Furthermore, validation experiment and related analyses were conducted for highly associated protein. We intend to make a deeper illustration for the molecular patterns involved in EBVaGC pathogenesis, as well as provide new clues for the molecular-targeted therapy of EBVaGC.

Methods

Sample preparation

The ethics committee of the First Hospital of China Medical University has approved the project. Signed informed consents were obtained from every participant. The subjects enrolled in this study were GC patients receiving surgical treatment in our hospital from September 2012 to October 2019. Screening criteria were having no other primary tumors and not undergoing any preoperative radiochemotherapy. Gastric tissue specimens were gained from each patient after surgical operation including cancer with adjacent cancer-free tissue. Two senior gastrointestinal pathologists made the histopathological diagnosis independently. Fresh GC tissue and adjacent normal tissue were randomly taken out from each case and divided into several parts with the size to fit for single use. Samples for EBV detection, hematoxylin–eosin (HE) staining and immunohistochemical staining were fixed with 10% formalin and embedded in paraffin. And samples for proteomics research were frozen in liquid nitrogen immediately and stored at − 80 °C.

Determination of EBV infection in GC

Epstein-Barr virus-encoded RNA (EBER) in situ hybridization (ISH) was applied to EBV detection for 140 GC cases using an EBER test kit (Beijing Zhongshan Jinqiao). In brief, tissue paraffin sections were cut into 4–6 μm-thick pretreated with dimethylbenzene and 100% ethanol. Each slice was incubated with 300–400 μl gastric enzyme for 30 min at 37℃. After dehydration by gradient ethanol, we added 10–20 μl EBER probe solution on each slice for hybridization and incubated them in moist chamber for 1 h at 37 °C. Then the sections were washed with PBS and incubated with peroxidase-labeled anti-digoxin antibody for 30 min at 37 °C. Finally, all tissue sections were stained with DAB (5-15 min) and restained with hematoxylin (5–10 s).

Quantitative proteomics of EBVaGC

Data-independent acquisition (DIA) mass spectrometry (MS) was performed by Genechem Co., Ltd. (Shanghai, China) to assay the proteomics of EBVaGC [12]. Briefly, total protein was extracted from tissue specimens and measured with BCA kit. We took 20 μg protein from each extract and mixed them with 6X sample loading buffer. The solutions were tested by SDS-PAGE (250 V, 40 min) and the gels were stained with Coomassie Blue. Filter-aided sample preparation (FASP) was adopted to extract and quantify peptides from 200 μg protein solution. All the peptides mix were graded by 1260 infinity II high performance liquid chromatography (HPLC) system (Agilent Technologies Inc.). We collected 48 components and 12 fractions after merging. 6 μl sample was taken from each fraction, mixed with 1 μl 10 × iRT peptides and separated by nano-LC. Finally, DIA-based MS analysis was conducted with LC–MS including Easy nLC system (Thermo Fisher Scientific) and Oribitrap Fusion Lumos system (Thermo Fisher Scientific). In addition, the MS based on data-dependent acquisition (DDA) was also performed and a spectrogram database was established for quality control.

Determination of protein expression in tissue

Immunohistochemistry (IHC) staining was employed to detect protein expression in tissue [13]. In short, paraffin-embedded tissue specimens were cut into 4 μm-thick sections. Tissue sections were dewaxed, rehydrated with gradient ethanol, incubated in 10 mmol/l citrate buffer (pH 6.0) and heated for 90 s. Endogenous peroxidase was blocked with 3% hydrogen peroxide (10 min). Tissue collagen was spoilt with 10% normal goat serum (10 min) for reducing non-specific binding. Rabbit polyclonal antibody for target protein (Abcam, UK) was used as primary antibody to incubate the samples for 1 h at room temperature. After washing with PBS, the samples were incubated with biotin-labeled secondary antibody (Fuzhou Maixin Biotech) and followed by streptavidin–horseradish peroxidase (HRP), both for 10 min at room temperature. Then the samples were stained with DAB (DAB-0031, Fuzhou Maixin Biotech), dehydrated and fixed with resin. Finally, the stained tissue sections were observed by experienced pathologists under inverted microscope. IHC staining was scored for each tissue section with positive staining based on the area (25%, 50%, 75%, 100%) and intensity (+ , +  + , +  + +). The final score was set to range from 1 to 4 after conversion.

Data analysis

The raw data of DIA-MS was processed with Spectronaut Pulsar X (v12, Biognosys AG). After normalization, differentially expressed proteins between EBV-positive and negative GC were identified. The threshold were set as absolute fold change (FC) > 1.5 and P < 0.05 corrected with 1% false discovery rate (FDR). Protein–protein interaction (PPI) information was downloaded from the STRING online tool (v11.0, https://string-db.org) and PPI network was constructed with Cytoscape software (v3.6.1). Funrich database (v3.1.3) was applied to gene enrichment analysis including expression site, Gene Ontology (GO) and biological pathways. The online datasets of gene expression profiling by microarray about EBVaGC were searched in Gene Expression Omnibus (GEO) database and analyzed with GEO2R package. Data processing and mapping was performed using R-project (v4.0.3) and Rstudio software (v1.3.1093). SPSS (v22.0) software was employed to analyze the data of validation experiments, including χ² test, independent t test or Mann–Whitney U test, Kaplan–Meier test, log rank test and Cox regression, etc.. All the tests were judged as statistically significant when |FC| > 2.0 and P < 0.05 after correction with Benjamini-Hochberg (BH) method (FDR).

Results

Identification of EBVaGC subjects

Based on the proven method of EBER-ISH, the nucleus of EBV-infected cells could be strongly stained after disposal following kit instructions [14]. A total of 7 tissue specimens with positive EBER signals out of the 140 GC cases were identified as EBV-positive GC group (A1-A7, Additional file 2: Fig. S1). Meanwhile, another 7 GC samples without positive staining were picked as EBV-negative GC group (B1–B7) matched by gender and age (± 5 years). The basic information and pathological characteristics of all subjects in the two groups were shown in Additional file 1: Table S1.

Characteristics of the protein profile in EBVaGC

The proteomics study for EBVaGC was conducted with the above 7 pairs of GC cases. A total of 137 differentially expressed proteins in EBV-positive GC group were identified compared with EBV-negative GC group (Table 1). Among them, GBP5, C5AR1, THRAP3, P3H3 and MDK were the top 5 differential proteins in the 47 up-regulated records. For the 90 down-regulated proteins, TMEM168, AKR7A3, MFAP4, EPHB2 and BCAM had the top 5 FC values. The clustered expression profile of all differential proteins in assayed tissue was shown in Fig. 1. And their detailed expression levels in each sample were listed in Additional file 1: Table S2.

Table 1 The differentially expressed proteins between EBV-positive and negative GC

Fig. 1

The clustered heat map of the differentially expressed proteins in EBVaGC. Several representative proteins are labeled

PPI network of the differentially expressed proteins in EBVaGC

To investigate the potential gene–gene interactions in EBVaGC, a PPI network was constructed for all above differentially expressed proteins. First, PPI information was collected from the String online database and 96 proteins showed interactions with at least one or more proteins. Based on their interactions and combined scores, the interaction degree for each protein was calculated with the cytoHubba plug-in in Cytoscape software. All the proteins were divided into 5 levels according to their interaction degrees: (1) > 20: 1; (2) 15–20: 4; (3) 10–15: 9; (4) 5–10: 29; and (5) < 5: 53. It was shown that several proteins had relatively high interaction degrees and might be the hub genes in EBVaGC, including ITGB2, CDC5L, CYBB, HLA-DRB1 and ATP6V1D (Fig. 2).

Fig. 2

The PPI network of the differentially expressed proteins in EBVaGC. The gradient color of circles from yellow to red represents the interaction degree of proteins from low to high

Gene enrichment analysis of the differentially expressed proteins in EBVaGC

Next, gene enrichment analysis was performed for these differentially expressed proteins to explore their potential biological function involved in EBVaGC. The expression sites of genes were predicted at first, which comprised of diverse cancer tissue, normal tissue and cell lines. The differential proteins between EBV-positive and negative GC were found to be significantly expressed in numerous cell lines and tissue such as H293 cell (P = 1.23E-14), CaOV3 cell (P = 9.47E-14), CD8 cell (P = 8.19E-13), ascites cancer cell (P = 1.98E-12) and colorectal cancer (CRC) tissue (P = 6.02E-12). Their fold enrichment were 1.99, 2.73, 2.67, 2.57 and 2.32, respectively (Fig. 3).

Fig. 3

The top 10 significant items in the enrichment analysis of expression sites for the differentially expressed proteins in EBVaGC. FE, fold enrichment

Then we focused on the GO-term enrichment analysis including cellular component (CC), molecular function (MF) and biological process (BP). Top 10 records sequenced by P values were picked for each term. Regarding CC, three items were suggested to significantly enrich the differentially expressed proteins, which were exosomes (P < 0.001), lysosome (P = 0.001) and mitochondrion (P = 0.032). And their fold enrichment respectively were 2.43, 2.34 and 2.26 (Fig. 4A). One term in MF, catalytic activity, showed significant enrichment effect for those proteins (P = 0.006, fold enrichment = 3.70, Fig. 4B). As for BP, the differential proteins were observed to be significantly enriched in two items, energy pathways (P < 0.001) and metabolism (P < 0.001). Both their fold enrichment were 3.01 (Fig. 4C).

Fig. 4

The top 10 significant items in the enrichment analysis of GO-term for the differentially expressed proteins in EBVaGC. A cellular component; B molecular function; C biological process

Moreover, a pathway analysis was performed to seek the possible biological pathways in which the differentially expressed proteins in EBVaGC might function. The records with top 10 P values were also selected. Only one item, ethanol degradation II (cytosol), demonstrated significant enrichment effect for those proteins (P = 0.047, fold enrichment = 43.75). And its percentage of enriched genes was 4.2% (Fig. 5).

Fig. 5

The top 10 significant items in the pathway analysis for the differentially expressed proteins in EBVaGC

Verification for the differentially expressed proteins in EBVaGC with GEO datasets

To elucidate the features of protein profiles in EBVaGC comprehensively, GEO database was also utilized to search high-throughput experimental data related to EBVaGC. A dataset of microarray gene expression profiling (GSE51575) was retrieved, containing 12 EBV-positive and 14 negative GC cases. We screened all the overlapping genes from differential records between GEO dataset and our array, including 15 up-regulated and 10 down-regulated genes. Interestingly, GBP5 was the only top gene with the highest fold change in both datasets. It was also suggested to be significantly up-regulated in EBV-positive GC compared with EBV-negative GC (P = 1.19E-03, log₂FC = 3.21, Additional file 1: Table S3), indicating that GBP5 might be a highly associated protein with EBVaGC. The expression levels of GBP5 in all tissue samples were presented in Fig. 6.

Fig. 6

The expression levels of GBP5 gene (mRNA) in EBVaGC from the microarray gene expression profiling (GSE51575) in GEO datasets

Validation for GBP5 expression in EBVaGC

Finally, a validation experiment was conducted to confirm the close association of GBP5 protein with EBVaGC. IHC staining was performed to detect GBP5 expression in a total of 255 tissue specimens including 7 EBV-positive and 248 EBV-negative GC cases with their corresponding adjacent normal tissue. The basic characteristics of GC subjects were presented in Additional file 1: Table S4. Representative photomicrographs of tissue cell staining were shown in Fig. 7. In EBV-positive GC, the staining signals of GBP5 protein were brown in color and located in epithelial cell membrane and cytoplasm, while no marked staining was found in adjacent normal tissue (Fig. 7A vs. B). Furthermore, GBP5 protein was also brown-stained in the membrane of lymphocytes among EBV-positive GC tissue (Fig. 7C, D). As for EBV-negative GC, neither epithelium nor mesenchyme has obviously positive staining in tissue specimens (Fig. 7E, F).

Fig. 7

The expression levels of GBP5 protein in EBVaGC by IHC staining. A, a EBV-positive GC tissue (× 100), positive staining in epithelial cell membrane and cytoplasm (score = 4); B, b adjacent normal tissue of A, a (× 40), negative staining in epithelial cell membrane and cytoplasm; C, c EBV-positive GC tissue (× 100), positive staining in the membrane of lymphocytes (score = 4); D&d, amplified visual field of C, c (× 200); E, e EBV-negative GC tissue (× 40), negative staining in epithelial cell membrane and cytoplasm; F, f EBV-negative GC tissue (× 40), negative staining in the membrane of lymphocytes

Based on the IHC staining results, related analyses for the association of GBP5 protein with GC clinicopathological parameters and prognosis were further performed. Foremost, we found that GBP5 expression had significant or borderline association with multiple GC clinicopathological parameters (Table 2). The positive rates were significantly higher in the following GC subgroups compared with control subgroups, including deeper invasion of gastric wall (muscularis + serosa, P = 0.042), positive vascular cancer embolus (P = 0.021) and positive extranodal tumor implantation (P = 0.011). However, no significant association between GBP5 expression and GC prognosis was found in either univariate or multivariate analysis after adjustment by the impacted factors of overall survival (Additional file 1: Table S5 and Additional file 1: Table S6). Moreover, an additional correlation was observed between GBP5 expression and EBV infection. GBP5 protein tended to be expressed in EBV-positive GC (P = 0.054), and its IHC staining score in the 7 EBV-positive GC cases was markedly higher than EBV-negative GC (3.2 ± 1.6 vs. 1.2 ± 1.5, P = 0.002, Table 3).

Table 2 The association between GBP5 protein expression and clinicopathological parameters of GC

Table 3 The association between GBP5 protein expression and EBV infection in GC

Discussion

Undoubtedly, thorough study for the molecular features of EBVaGC is of great pathological and clinical values. Here, a comprehensive analysis was presented for the protein profile in EBVaGC tissue based on DIA-MS. A total of 137 differentially expressed proteins were identified between EBV-positive and negative GC. PPI network and gene enrichment analysis were successively performed for all differential proteins. Combined with the gene expression profiling in GEO datasets, a highly associated protein (GBP5) with EBVaGC was screened out and validated with IHC staining. As far as we concerned, for the first time our study integrally revealed the protein expression patterns in EBVaGC along with the potential biological function of differentially expressed proteins. In addition, we also firstly reported the highly associated protein with EBVaGC followed by preliminary validation.

Virus-host interactions within infected cells are the core parts during EBV-induced carcinogenesis. Compared with the relatively simple proteomics in virus, the number of genes and complexity of proteomics in host are much more than the former. Besides, the expression levels of various oncogenes and tumor suppressor genes in the infected host cells could vary with the stimulation of viral gene products [15, 16]. Therefore, the proteomic analysis in EBVaGC is quite valuable, and the proteins with remarkable differences and central roles maybe potential diagnostic markers of EBVaGC. Lots of differentially expressed proteins in EBVaGC were newly identified in our study. Although the evidence about their direct relations with EBVaGC is limited, some hints have been manifested in their respective association with EBV infection and GC initiation such as several top proteins like GBP5, C5AR1 and THRAP3 [17,18,19]. Furthermore, a few crucial genes in EBVaGC were excavated from the differential proteins by means of network analysis. The PPI network showed several proteins with relatively strong interactions such as ITGB2, CDC5L, CYBB and HLA-DRB1. Consistently, previous reports have also suggested that they may serve as hub genes in many diseases especially carcinoma [20,21,22,23]. Considering both the differential profile and PPI network, a highly studied hub gene (HLA-DRB1) is noteworthy, which was concurrently one of the top 10 up-regulated records in the assay. Its expression and polymorphisms were shown to be associated with both EBV infection and GC [24, 25]. In general, the establishment of protein profiles in EBVaGC greatly improved the access to its molecular research. The key proteins with significantly differential expression and hub roles could be selected as potential biomarkers for EBVaGC detection. However, substantial discovery studies are needed to confirm that.

The specific programs of viral gene expression found in EBVaGC can target cell signaling pathways leading to increased proliferation, cell survival, immune invasion, augmented epithelial-to-mesenchymal transition (EMT) and acquisition of stemness features [15]. For instance, Zhao et al. reported 13 pathways deregulated in EBVaGC, including mitogen-activated protein kinase (MAPK), Wnt and focal adhesion etc., which could facilitate rapid tumor growth [26, 27]. Coincidently, some differential proteins mentioned above were indicated to participate in the genesis of gastric adenocarcinoma or stromal tumors via these classical pathways such as GBP5, C5AR1 and THRAP3 [28,29,30]. Beyond that, EBVaGC-specific cellular pathways have also been increasingly explored [11]. For example, Want et al. found alterations in macromolecular biosynthetic processes, and deregulation of cholesterol transport and lipoprotein clearance pathways was also evident in EBVaGC [26, 31]. Novel findings were observed in our prediction for the biological function of differentially expressed proteins in EBVaGC. They were shown to be enriched in the metabolic pathways of energy including mitochondrion or biochemical substances like ethanol degradation, along with catalytic activity. The metabolic landscape of EBVaGC was investigated before and aberrant metabolism in EBVaGC was well accepted. Significant down-regulation of genes involved in metabolic pathways has been proved especially biochemical metabolism such as amino acids, lipids and carbohydrates [32, 33]. So far, however, rare study has referred to the change of energy pathways in EBVaGC. Only one gene set enrichment analysis by Sohn et al. revealed that EBVaGC had significant genetic alterations in pathways involving energy production [34]. Some clues could be extracted from the association between EBVaGC and mitochondrion-related pathways. An original research showed that EBV-encoded BARF1 was down-regulated in EBV-positive malignant cells and induced caspase-dependent apoptosis via mitochondrial pathway [35]. Another report suggested that the expression of CCL21 by EBVaGC cells protected CD8( +) CCR7( +) T lymphocytes from apoptosis via mitochondria-mediated pathway [36]. Therefore, it is reasonable to infer that the differential proteins in EBVaGC might function in the dysregulation of energy metabolism by mediating mitochondrial pathways, and even affect the survival of EBV-infected GC cells. Nevertheless, all the hypotheses about concrete mechanisms need further verification.

Combined our high-throughput assay with public database, a highly associated protein of EBVaGC, GBP5, was found out with the highest fold change of differential expression both in the present study and GEO dataset. IHC staining also confirmed its overexpression in EBVaGC tissue. GBP5 (Guanylate binding protein 5) is a member of IFN-inducible subfamily of guanosine triphosphatases (GTPases) and exert critical roles in cell-intrinsic immunity against diverse pathogens including EBV [37]. The expression level of GBP5 was increased in the peripheral blood mononuclear cells of patients with chronic active EBV infection [18]. The involvement of GBP5 in the immune microenvironment of GC has also been preliminarily explored. A previous IHC experiment demonstrated that GBP5 had cytoplasmic and membranous expression in GC cells while no signals in non-neoplastic stomach [30]. Meanwhile, EBV could invade into B-lymphocytes, epithelial cells and fibroblasts through different mechanisms, thus the up-regulation of GBP5 might appear in both epithelia and mesenchyme. All these phenomena were consistent with our assay. Moreover, further analyses revealed that GBP5 protein was correlated with some malignant GC clinicopathological features. Considering GBP5 also took parts in innate immune activation and the regulation of inflammasomes related to cancer [38], its overexpression might be defensively activated in lesion when poor differentiation arose in GC cells. Importantly, GBP5 protein was validated to have a higher expression trend in GC tissue with EBV infection compared with EBV-negative GC, which laid a more convinced association with EBVaGC. Hence, GBP5 protein could be a promising EBVaGC-related marker with the function as an anti-EBV factor and effector of immune defense against GC progression simultaneously, in spite of the need to further investigation.

To be acknowledged, however, only the most representative protein GBP5 was validated with IHC and further analyzed. More proteins with the potential to be EBVaGC-related markers except for GBP5 might be hidden in other differential records from DIA-MS or GEO database. And it is quite necessary to validate them in future studies.

Conclusions

In summary, we conducted a comprehensive analysis of the protein profile in EBVaGC mainly by the aid of DIA-MS. A few differentially expressed proteins were newly identified between EBV-positive and negative GC, and several hub genes were subsequently revealed. The proteins with significant differences and potential central roles could be applied as diagnostic markers of EBVaGC. They were also predicted to be involved in the biological pathways related to energy and biochemical metabolism. Additionally, a highly associated protein (GBP5) was screened out by a joint analysis with GEO database and validated with IHC staining, which might be a key protein in EBVaGC. Our study could provide research clues for EBVaGC pathogenesis as well as novel targets for the molecular-targeted therapy of EBVaGC.