H. pylori is a Gram-negative spiral-shaped bacterium, present in the gastric epithelium of over 50% of the world's population (Jones et al., 2010). H. pylori infection has been associated with a variety of diseases including chronic gastritis, peptic ulcers, and epithelial and lymphoid malignancies of the stomach (Moss, 2017). Chronic infection by H. pylori is the strongest known risk factor for intestinal and diffuse histomorphological type GC (Helicobacter and Cancer Collaborative Group, 2001; Moss, 2017). H. pylori genome diversity and the presence of bacterial virulence factors play an important role determining the outcome of H. pylori infection. The cytotoxin-associated gene A (cagA) encodes CagA, the major virulence factor of H. pylori, which is delivered into gastric epithelial cells via bacterial type IV secretion. CagA positive and vacuolating cytotoxin A (VacA) with allelic variant s1m1 strains are associated with increased disease severity (Wroblewski and Peek, 2013).

Infection of the gastric mucosa by this pathogen causes inflammation of the gastric tissue (non-atrophic gastritis), which initiates the “Correa cascade.” The actual pre-neoplastic cascade begins at multifocal atrophic gastritis, characterized by loss of the gastric glands due to chronic inflammation (Correa, 1988). It is known that H. pylori inhabits the glandular epithelium of the stomach, and the bacterium is commonly lost during the progression of precancerous lesions as a result of the replacement of these glands for intestinal-like epithelium (intestinal metaplasia) (Morson, 1955). Therefore, the oncogenic role of H. pylori resides in the initial steps of the gastric precancerous cascade. Hence, this review summarizes studies on host ncRNAs (lncRNAs/miRNAs) which are dysregulated in association with H. pylori infection.

Eradication of H. pylori reduces overall GC rates, but only in early pre-cancerous lesions (i.e., non-atrophic and multifocal atrophic gastritis) and not in advanced lesions (i.e., intestinal metaplasia and dysplasia; Chen et al., 2016). Thus, whether to implement programs aiming for H. pylori eradication remains an open question.

EBV is a linear, double-stranded DNA virus and a member of herpesviridae family. EBV was described more than 50 years ago in patients with Burkitt's lymphoma and was the first virus linked to cancer in humans (Young et al., 2016). EBV has a high prevalence worldwide and it is thought that in about 90% of adults, EBV establishes persistent infection (Cohen, 2000). EBV is best known as the cause of infectious mononucleosis in adolescence or young adulthood. EBV is linked to a variety of human tumors, including lymphoid (Burkitt's lymphoma, Hodgkin's disease, B cell lymphomas) and epithelial neoplasms [nasopharyngeal carcinoma (NPC) and GC; Young et al., 2016]. In the case of GC, EBV-associated gastric carcinoma (EBVaGC) is linked to the diffuse histomorphological type GC (Camargo et al., 2011; Carrasco-Avino et al., 2017). In addition, the effect of EBV infection on GC prognosis, evaluated in 4,599 patients by an international pooled analysis, shows that patients with EBV-positive tumors have increased overall survival rates than EBV-negative cases (Camargo et al., 2014).

Although the association between EBV and gastric carcinoma was first proposed in 1992 by Shibata and Weiss (1992), it took more than 20 years to be recognized (The Cancer Genome Atlas Research Network, 2014). The Cancer Genome Atlas (TCGA) consortium, in its novel molecular classification for GC, recognizes EBVaGC as one of the four proposed subtypes, representing about 9% of all gastric carcinomas (Figure 1). This subtype of GC harbors recurrent mutations in the PIK3CA gene, DNA hypermethylation, and amplifications of JAK2, CD274, PDCD1LG2, and ERBB2 (The Cancer Genome Atlas Research Network, 2014). Such characteristics imply altered proliferation, apoptosis and immune suppression and evasion (Sun et al., 2007). Interestingly, EBVaGC has been reported to possess the most extensive CpG island methylation on both human and viral genomes, which is more extensive than in any other tumor type in the TCGA database (Zouridis et al., 2012; The Cancer Genome Atlas Research Network, 2014; Gulley, 2015; Alarcon et al., 2017).

EBV infects B cells by recognition of CD21 on the cell surface, from where it is transported to the cell nucleus. After infection, EBV can enter into a lytic cycle or switch to a latency stage. The coding and non-coding latency genes repress the lytic cycle and specific sets of these genes are associated with malignancies. The expression patterns of coding and non-coding latency genes that repress the lytic cycle in EBV-associated malignancies is well characterized, except for EBVaGC. The most commonly observed pattern in this disease is latency II-like (44.4%), defined by expression of EBNA1, EBERs, BARF1, and LMP2A genes (zur Hausen et al., 2000). The second most common pattern is latency I (42.9%), which is restricted to EBNA1, EBERs, and BARTs. Latencies II and III represent only few cases (Price and Luftig, 2015; Ribeiro et al., 2017). Specific sections within this review explain the role of each of the above-mentioned genes. Clearly, further research is warranted to clarify the specific expression patterns of coding and non-coding latency genes in EBVaGC.

Although H. pylori and EBV have been clearly identified as etiological agents of GC, only few studies have analyzed the role of co-infection in this process. The mechanism of interaction between H. pylori and EBV is not known. As both pathogens induce severe inflammatory response, chronic inflammation is a common theme. This observation is supported by a study in which co-infected patients presented enhanced inflammatory lesions compared to those only infected with H. pylori or EBV (Cardenas-Mondragon et al., 2013). Moreover, EBV positivity in H. pylori positive GC was associated to premalignant lesions and intestinal-type GC (Cárdenas-Mondragón et al., 2015). Thus, EBV probably collaborates with H. pylori through the induction of additive inflammatory response and enhanced inflammation.

Although only H.pylori is associated with the activation of E-cadherin/β-catenin/TCF-4 signaling pathway in normal gastric epithelial cells (Yu et al., 2011, 2014), both pathogens are linked to the activation of NF-κβ and MAP kinases oncogenic pathways in GC cell lines (Mohr et al., 2014; Liu and Cohen, 2015; Byun et al., 2016; Dela Pena-Ponce et al., 2017). Co-infection by EBV can also hinder the host response to H. pylori. Host protein SHP 1 interacts with H. pylori virulence factor CagA, causing dephosphorylation and inactivation of the bacterial protein. Thus, SHP 1 prevents oncogenic activity of CagA. However, EBV coinfection causes methylation of host SHP 1, silencing its expression. Consequently, the cooperative effect of EBV may increase the oncogenic potential of H. pylori CagA (Saju et al., 2016).

Epigenetic modifiers (DNMT1, DNMT3B, EZH2), tumor suppressor genes (RPRM, MGMT, TWIST1), miRNAs (let-7a/c, miR-155) and multiple signaling pathways are dysregulated in the process of gastric carcinogenesis (Schneider et al., 2010; Hayashi et al., 2013; Zhang et al., 2016). In intestinal-type GC, down-regulation of CDH1 by DNA methylation has been exclusively linked to the presence of CagA-positive H. pylori strains (Ferrasi et al., 2010). On the other hand, in EBV-positive GC, a distinct gene silencing pattern has been identified including p16, p14, APC, p73, RUNX3, MINT2, MINT31, FHIT, CRBP1, WWOX, and DLC-1, in spite of the extensive DNA methylation observed in these tumors (Geddert et al., 2011; Saito et al., 2013; He et al., 2015). CagA-positive H. pylori strains attenuate let-7a/c expression by histone modification and DNA methylation and lead to activation of the Ras pathway (Hayashi et al., 2013). Conversely, the EBV latent infection protein EBV nuclear antigen 1 (EBNA1) upregulates multiple let-7 family miRNAs, including let-7a in NPC, as well as GC cell lines. Cellular targets of let-7a, such as Dicer, are downregulated in both types of cell lines (Mansouri et al., 2014). H. pylori upregulates miR-155 (Xiao et al., 2009), which is a strong inhibitor of inflammatory response in EBV-associated tumors including lymphoid and epithelial cancers (Jiang et al., 2006; Du et al., 2011). Clearly, further studies are warranted to elucidate mechanisms driving commonalities in co-infection by both etiological agents of GC.

lncRNAs make up the largest portion of the mammalian non-coding transcriptome (Derrien et al., 2012). lncRNAs regulate a plethora of cellular processes and aberrant expression of lncRNAs could play an important role in gastric carcinogenesis (reviewed in Li et al., 2016; Yoshida and Kimura, 2017). A few studies have explored the role of lncRNAs in GES-1, a normal epithelial gastric cell line, infected with H. pylori. Twenty-four hours post-infection, 23 and 21 lncRNAs were upregulated and downregulated, respectively. Specifically, lncRNAs XLOC_004562, XLOC_005912, and XLOC_000620 were highly expressed, whereas XLOC_004122 and XLOC_014388 were downregulated (Table 1). These latter two lncRNAs were validated in clinical samples positive for H. pylori infection and further re-expressed post H. pylori eradication (Yang et al., 2015). Profiling of multiple lncRNAs and coding transcripts in H. pylori-infected GES-1 cells has identified a network of multiple dysregulated lncRNA-mRNAs involved in the development of H. pylori-related pathologies, including GC (Zhu et al., 2015). Later studies have identified specific examples of this, such as decreased expression of lncRNA AF147447 in H. pylori infection. This non-coding transcript inhibits cell proliferation and invasion through downregulation of oncogene MUC2 (Zhou et al., 2016). More recent studies have begun to shed light on the transcriptional crosstalk between lncRNAs and mRNAs in GC, proving that this is a promising field in biomedicine (Zhang et al., 2018) which will broaden our understanding of the pathogenesis of this disease. Furthermore, tissue-specificity and ability to predict clinical subtypes of cancer by their future behavior suggest great potential for these ncRNAs as biomarkers for diagnosis and prognosis (Du et al., 2013; Schmitt and Chang, 2016).

In this section, we focus on the role of miRNAs at the early stages of the pre-neoplastic process of GC since, as mentioned above, it is at these stages where H. pylori is directly involved (Chen et al., 2016). We do not make an in-depth discussion of miRNAs that are differentially expressed in GC as, at this stage, H. pylori is most commonly not present at the lesions.

miRNAs have an important role in the control of inflammatory response associated with H. pylori infection. Downregulation of let-7 miRNAs, the first known human miRNA family, was observed in H. pylori-positive clinical samples, particularly those infected with CagA-positive strains. Downregulation of let-7b results in overexpression of TLR4, which activates NF-κB and increases the expression of COX-2 (Teng et al., 2013). Similarly, members of the miR-200 family, including miR-141, miR-200a/b/c, and miR429, as well as miR-375 and miR-103, were downregulated in H. pylori-infected gastritis samples. These findings correlate with increased neutrophil and/or mononuclear cell tissue infiltration, as well as overexpression of pro-inflammatory cytokines including IL-1β, IL-6, IL-8, and TNF-α (Isomoto et al., 2012). miR-124 has tumor suppressive functions and was reported as a negative regulator of polyamine catabolic enzyme spermine oxidase (SMOX) (Murray-Stewart et al., 2016). SMOX is induced in bacterial infection and chronic inflammation, including H. pylori-associated gastritis. When induced in the nucleus, SMOX-dependent production of hydrogen peroxide contributes to oxidative DNA damage and subsequent tumorigenesis. Consequently, chronic gastritis patients with lower levels of miR-124 expression displayed higher levels of SMOX and were more likely to develop high-grade gastric lesions.

On the other hand, several miRNAs are upregulated in H. pylori infection. miR-21 was the first reported miRNA to be influenced by H. pylori, found overexpressed in H. pylori-infected cell lines, gastritis, and GC tissues (Zhang et al., 2008). Importantly, miR-21 acts as an oncogenic miRNA by targeting the tumor suppressor gene RECK, involved in tumor metastasis and angiogenesis (Gutiérrez et al., 2016). miR-155 expression is induced in H. pylori-infected GES-1, as well as in gastritis tissue samples (Xiao et al., 2009). Interestingly, expression of this miRNA is induced in H. pylori-infected T cells and primary macrophages in a Foxp3-dependent manner, suggesting a functional interplay between miR-155 and the host immune response (Fassi Fehri et al., 2010). miR-155 also targets IkB kinase-ε (IKK-ε), Sma- and Mad-related protein 2 (SMAD2), and Fas-associated death domain protein (FADD), reducing the release of pro-inflammatory cytokines such as IL-8 and GROα (CXCL1) and attenuating the inflammatory response associated with H. pylori infection (Xiao et al., 2009). miR-223 is also induced upon H. pylori infection (Matsushima et al., 2011; Wang et al., 2016). This miRNA downregulates pro-inflammatory cytokines (IL-6, IL-8, IL-12, and TNF-α) and inhibits the activation of macrophages through downregulation of CD40, CD68, CD80, and CD163. Therefore, miR-223 also limits the host inflammatory response to H. pylori (Wang et al., 2016). Additionally, CagA-positive H. pylori strains induce expression of NF-κB, which binds to the promoter of miR-223. miR-223 targets ARID1A, promoting cell proliferation and migration. In clinical samples miR-223 is upregulated in the tumor, while ARID1A is downregulated significantly in comparison to the non-tumor adjacent mucosa from the same patients (Yang et al., 2018).

Taken together, host miRNAs play a preponderant role in the inflammatory response against H. pylori, representing a potential bridge between H. pylori, chronic inflammation and progression of precancerous lesions. Table 1 summarizes the main studies of host miRNAs which are dysregulated upon infection by H. pylori.

In latent EBV infections, the virus hinders the host functional immunity. During latent infection, EBV expresses limited levels of protein but high levels of ncRNAs, such as Epstein-Barr Virus-Encoded RNAs (EBERs) and Bam HI A rightward transcripts (BARTs). This suggests that the ncRNAs are related to evasion of the viral immune response (Albanese and Tagawa, 2017). Pfeffer, were the first group to report that miRNAs are also encoded by viruses (Pfeffer et al., 2004). The herpesvirus family encodes the vast majority of viral miRNAs (Pfeffer et al., 2005). These DNA viruses, including EBV, encode their own miRNAs and/or modulate the expression of host miRNAs to facilitate infection cycles. It has also been evidenced that EBV miRNAs may contribute to gastric carcinogenesis (Yau et al., 2014). EBV miRNAs are encoded in two main clusters: the Bam HI fragment H rightward open reading frame 1 (BHRF1) cluster and BART cluster (Pfeffer et al., 2004).

It is long known that the most abundantly expressed viral transcripts during EBV latent infection are small non-coding EBV-encoded RNAs termed EBER1 and EBER2 (Rymo, 1979). These ncRNAs are 167 and 172 nucleotides long, respectively (Lerner et al., 1981), and form double-stranded RNA-like structures. The genes encoding EBER1 and EBER2 are separated by 161 nucleotides and transcribed individually in the same direction using RNA polymerase III (Rosa et al., 1981). These transcripts are expressed in all EBV-positive cells and, thus, in all EBV-associated malignancies. EBERs can be detected in EBV-positive cells using in situ hybridization technique as a routine test to determine EBV infection in clinical samples (Weiss and Chen, 2013).

In the host, EBERs are involved in the promotion of cellular growth. Reports show that EBERs may confer an apoptosis-resistant phenotype to epithelial cells. EBV infection induces expression of insulin-like growth factor (IGF-I) and secreted IGF-I acts as an autocrine growth factor. Moreover, EBV-positive GC and NPC biopsies consistently express IGF-I, whereas EBV-negative biopsies do not (Iwakiri et al., 2003, 2005).

BARTs are a family of transcripts from the Bam HI A region of the EBV genome. BARTs show extraordinarily high expression levels in EBV-infected epithelial cancers, but not in EBV-transformed lymphocytes (Qiu et al., 2011). It has been proposed that BARTs are key players in epithelial malignancies such as NPC and EBVaGC. Reports indicate that the expression of BARTs is regulated by interferon regulatory factors and NF-κB signaling pathway (Chen et al., 2005; Samanta et al., 2006). Interest in the functional role of BARTs has led to the discovery of viral miRNAs (miR-BARTs) in this region (Pfeffer et al., 2004).

EBV was the first described virus to encode miRNAs (Pfeffer et al., 2004). EBV uses the host cell machinery to produce its miRNAs, analogous to the biogenesis of host cell miRNAs (Kim and Lee, 2012). Briefly, the miRNA sequences are mainly transcribed by host RNA polymerase II to form stem-loop structures, which are subsequently processed by Drosha and DiGeorge syndrome chromosomal region 8 (DGCR8) and Dicer. The host DGCR8 endonuclease complex cleaves the primary miRNA, resulting in a precursor miRNA which is transported from the nucleus to the cytoplasm. Subsequently, this precursor miRNA is processed in the cytoplasm by host protein Dicer to generate the mature miRNA.

As stated above, EBV miRNAs are encoded in two clusters. The first EBV miRNA cluster is located within a transcript that encodes the BHRF1 protein and includes three miRNA precursors (miR-BHRF1−1,−2, and−3), which produce four mature EBV miRNAs. miR-BHRF1-1 is located in the 5′ UTR (untranslated region) while miR-BHRF1−2 and−3 are located in the 3′ UTR of the BHRF1 mRNA (Pfeffer et al., 2004). The second cluster is located in intronic segments within the BART transcripts and includes 22 miRNA precursors (miR-BART1-22), which generate 40 mature miRNAs (Cai et al., 2006; Grundhoff et al., 2006; Zhu et al., 2009).

The expression of these microRNAs depends on the latency state and is cell-type specific. miR-BHRF1s are expressed during lytic infection and in latency III (Cai et al., 2006). Meanwhile, miR-BARTs are expressed in all EBV latency types. miR-BHRF1s are expressed in lymphoblastoid cell lines (latency III), but not in NPC (Cosmopoulos et al., 2009) and EBVaGC (Kim et al., 2007; Qiu et al., 2011; Shinozaki-Ushiku et al., 2015). Only sporadic expression of BHRF1s at very low levels was found in some Burkitt's lymphoma and Hodgkin's disease biopsies (Qiu et al., 2011). Conversely, miR-BARTs are expressed in EBV-infected lymphoblastoid cell lines, Burkitt's lymphoma, NPC and EBVaGC, showing significantly higher overall expression in epithelial cancers NPC and EBVaGC in comparison to EBV-positive B lymphoma. As expression of BHRF1 miRNAs is not consistently deregulated in the tumors, it is expected that the oncogenic role of EBV miRNAs can be attributed to the BARTs (Qiu et al., 2011). Specific miR-BARTs can induce transition between latency states and lytic replication. miR-BART6-5p maintains type I/II latency in cultured cells, as its inhibition results in increased expression of viral oncoproteins latent membrane protein 1 (LMP1) and EBNA2, important in type III latency (Iizasa et al., 2010).

EBV miRNAs favor evasion of the host immune system and maintain EBV latency by suppressing apoptosis of infected cells (Seto et al., 2010; Albanese and Tagawa, 2017). Viral miRNAs can regulate inflammation by targeting several chemokines and cytokines that regulate antiviral inflammatory responses. For example, several viral miRNAs can act collectively to suppress the release of pro-inflammatory cytokine IL-12 caused by and subsequently modulate the inflammatory response of CD4+ and CD8+ T cells (Albanese et al., 2016; Tagawa et al., 2016). EBV miRNAs can also target several viral genes to downregulate expression of some EBV proteins such as EBNA1 (Albanese et al., 2016), latent membrane proteins LMP1 (Lo et al., 2007) or LMP2A (Lung et al., 2009). Very low expression of viral antigens allows the virus to escape the immune system and maintain persistent infection.

Similarly, Marquitz and colleagues (Marquitz et al., 2012) demonstrated that EBV infected GC cell line shows limited viral protein expression whereas viral BART miRNAs are abundantly expressed. Moreover, BARTs affected growth properties of these EBV-infected cells, suggesting that they can contribute to the development of epithelial malignancies. In an another study they showed that both BART clusters can independently inhibit apoptosis in GC cell line and that the pro-apoptotic protein Bcl-2 interacting mediator of cell death is a target of multiple BART miRNAs (Marquitz et al., 2011). These observations were further extended by identification of several mRNA targets for EBV BART miRNAs that encode pro-apoptotic proteins (CASZ1, DICE1, OCT1, CREBBP, SH2B3, PAK2, and TP53INP1; Kang et al., 2015).

Specifically, the EBV miR-BART5 is highly expressed in epithelial cells. This miR-BART targets the proapoptotic protein p53 up-regulated modulator of apoptosis (PUMA) mRNA, resulting in translational repression of the protein in NPC and EBV-infected GC cells and promoting host cell survival (Choy et al., 2008). miR-BART20-5p reduces apoptosis and enhances growth of GC cells by targeting apoptosis-inducing factor Bcl-2-associated agonist of cell death (BAD) mRNA (Kim et al., 2015). Recently, high BART20-5p expression levels were associated with worse recurrence-free survival of EBVaGC patients (Kang et al., 2017). Likewise, miR-BART4-5p targets BH3 interacting domain death agonist (BID), a Bcl-2 family gene, and suppresses apoptosis in GC cells (Shinozaki-Ushiku et al., 2015).

Consequently, miR-BARTs play an important role in carcinogenesis. It was reported that several EBV-encoded miRNAs, including miR-BART22, contribute to EBVaGC by targeting N-myc downstream regulated gene 1 (NDRG1) (Kanda et al., 2015). NDRG1 is an essential gene for maintaining epithelial cell differentiation and a suppressor of metastasis. miR-BART11 was shown to promote monocyte differentiation to macrophages by attenuating expression of forkhead box P1 (FOXP1), a key molecule involved in this transformation (Song et al., 2016). Biopsies from both epithelial cancers, NPC (Liao et al., 2014) and GC (Liu et al., 2017), are often infiltrated with inflammatory cells, including tumor-associated macrophages that serve as a barrier against the infiltration of CD8+ T cells into GC. In the case of NPC, tumor infiltration by macrophages is tightly associated with poor prognosis (Liao et al., 2014). Targeting of FOXP1 by miR-BART11 induces proliferation of GC cells and activates NF-κB signaling (Song et al., 2016). In a study which analyzed 52 EBVaGC tissues samples for viral miRNAs, miR-BART4-5p was the most abundantly expressed miR-BART, followed by miR-BART11-3p, miR-BART2-5p, miR-BART6-3p, miR-BART9-3p, and miR-BART18-5p. Additionally, an interaction among miR-BART9-3p, miR-200a and E-cadherin was demonstrated suggesting that miR-BARTs contribute to malignant transformation via regulation of host miRNAs and epithelial-to-mesenchymal transition (EMT) (Tsai et al., 2017). Other EBV BART-miRNAs, including miR-BART1-3p, 5-5p, 7-3p, 15-3p, 19-3p, and 22-3p are also expressed in EBVaGC tissue samples and GC cell lines (Qiu et al., 2011; Kim et al., 2013), and have distinct expression levels in EBV-related epithelial cancers compared to lymphoid malignancies. The functions of these and additional miR-BARTs are included in Table 2.

Contrary to most published studies, a few publications bring controversial view on the strictly pro-tumorigenic role of miR-BARTs. For example, it has also been proposed that miR-BART6-3p acts as a tumor suppressor miRNA. miR-BART6-3p directly targets lncRNA LOC553103 and induces downregulation of CDH2, β-catenin, SNAIL, MMP2, and MMP9, reversing EMT (He et al., 2016). Choi et al. (2013) have shown that miR-BART15-3p, expressed in cultured GC cells infected with EBV, targets baculovirus inhibitor of apoptosis repeat-containing ubiquitin-conjugating enzyme (BRUCE), a member of the inhibitor of apoptosis proteins. These findings show that some miR-BARTs can possess anti-tumor activities that have, to date, been poorly described.

EBV modulates the host immune response and participates in cell transformation not only via its own miRNAs but also through regulation of host miRNAs. In this sense, several studies have investigated the expression of viral and host miRNAs in EBVaGC. For example, Iizasa et al. (2010) reported that viral miR-BART6-5p targets host cell Dicer by direct binding to the 3′-UTR of the mRNA, and proposed it is likely a major viral miRNA suppressor. In consequence, host cell miRNA production is impaired, thus enabling EBV to remain in a specific latency state and evade the host immune system.

Through a sequential series of studies, the group of Fukayama showed that expression the miR-200 family was decreased in EBVaGC, as well as in several GC cell lines infected with EBV (Shinozaki et al., 2010). They also reported that Bam HI A fragment rightward reading frame (BARF)0, EBNA1, EBERs, and LMP2A contributed to the downregulation of the mature miR-200 family, indicating a synergetic effect of latency type I genes in EBVaGC. Importantly, in vitro, downregulation of miR-200 transcripts lead to upregulation of E-cadherin transcription repressors ZEB1 and ZEB2. In consequence, decreased expression of miR-200 resulted in inhibition of E-cadherin expression and induction of EMT, which was initially observed by their group to be more frequent in EBVaGC than in EBV-negative GC (Sudo et al., 2004). Confirming these findings, Marquitz et al. (2014) performed miRNA profiling of an EBV-infected GC cell line and reported the downregulation of several miRNAs from the miR-200 and let-7 family. Moreover, they identified several other downregulated miRNAs, among them miR-143-3p, miR-146b-5p, miR-148-3p. Another downregulated host miRNA in EBVaGC is miR-34a. Expression of this miRNA is inhibited by viral protein EBNA1 (Kim et al., 2017), which is expressed by all EBV-infected cells, and is an essential gene for the establishment and maintenance of latency (Ribeiro et al., 2017). Downregulation of miR-34a in EBV-infected GC cells causes upregulation of NOX2 and results in augmented ROS production and increased cell viability (Kim et al., 2017).

The recent molecular classifications of GC proposed by the TCGA, also identified a differential miRNA expression profile for EBVaGC. Upregulation of miR−196a and−196b was observed in all GC subtypes compared to normal tissue samples, except this subtype (The Cancer Genome Atlas Research Network, 2014). A study by Treece and collaborators (Treece et al., 2016) that analyzed expression of a custom miRNA panel in EBV-positive versus EBV-negative tumors showed significant downregulation of miR-196b in the infected tumors, confirming what was identified by the TCGA. The same study demonstrated significant upregulation of miR-155, miR-185, and miR-378 in EBV-positive tumors, also consistent with previously reported upregulation of miR-155 and miR-185 in the TCGA database. miR-155 is upregulated in both, in EBV-positive tumors (Jiang et al., 2006; Du et al., 2011) and in H. pylori-infected gastric tissues, and suppresses the host inflammatory response, as described above in the section about host miRNAs associated with H. pylori infection.

In summary, EBV-infected cells express viral miRNAs and proteins which regulate host miRNA expression. This helps infected cells to evade the host immune response, favoring chronic infection and enhancing cell viability.

Compared with viruses, exosomes possess an increased ability to enter a wider range of cell types. In vitro experiments showed that when exposed sufficient time, most cell types internalized at least a part of stained exosomes independently of their cell origin (van Dongen et al., 2016). The exact mechanisms of exosome entry, specific cell targeting, and cargo delivery remains unclear. Generally, exosomes can deliver their content by fusing to the recipient cell's cytoplasmic membrane. Nevertheless, receptor-specific entry has also been described. Incorporation of viral glycoprotein gp350 has been reported in exosomes produced by EBV-infected cells. These exosomes selectively target B cells through the EBV entry receptor CD21. This gp350–CD21 interaction is so strong that exosomes derived from EBV-transformed cells can cause competitive inhibition of EBV entry in B cells, preventing their infection (Vallhov et al., 2011). In contrast, viral entry has been broadly studied and is very specific for each virus and cell type, requiring specific receptors and co-receptors that bind to envelope glycoproteins. How extracellular vesicles and exosomes exploit viral entry routes for cargo delivery and how viruses utilize exosomes for virus, viral genome or protein delivery has been comprehensively reviewed elsewhere (Nolte-'t Hoen et al., 2016; van Dongen et al., 2016; Raab-Traub and Dittmer, 2017).

A study by Pegtel et al. (2010) evidenced that EBV-infected B cells secrete functional viral miRNA miR-BHRF1-3 via exosomes. This miRNA is transferred to and acts by silencing expression of target genes in uninfected recipient cells. Furthermore, they also detected EBV-encoded BART Cluster 1 miRNAs in circulating non-B cell lymphocytes, where EBV DNA is not present. They suggest that the presence of miR-BARTs in these cells can be attributed to exosomal transfer. It is known that miRNAs from the miR-BART cluster 1 cause translational repression of EBV LMP1 (Lo et al., 2007). They propose exosomal miRNA transfer from EBV-infected cells to uninfected recipient cells as a gene silencing mechanism in adjacent (in a paracrine fashion) or distant cells. Accordingly, Choi et al. have described that miR-BART15-3p is secreted via exosomes by EBV-infected GC cells, targeting the inhibitor of apoptosis BRUCE (Choi et al., 2013). They also proposed that this miR-BART is preferentially loaded into exosomes and may be delivered into neighboring immune cells, favoring their apoptosis. Though interesting, no experiments were carried out to support this conclusion.

Aside from non-coding RNAs, exosomes derived from EBV-positive cells contain functional viral proteins. LMP1 is the major EBV oncoprotein, as it is required for B lymphocyte transformation, and is expressed in neoplastic cells of EBV-positive lymphoid malignancies (Raab-Traub and Dittmer, 2017). Dukers et al. (2000) have for the first time described direct immunosuppression in a DNA virus, previously thought to be restricted to RNA viruses. This group demonstrated that EBV is capable of inducing T cell anergy via a novel direct route, possibly mediated by secretion of EBV-encoded LMP1. EBV-infected cells secrete LMP1-containing exosomes, which inhibit T cell proliferation and natural killer cell cytotoxicity (Dukers et al., 2000; Flanagan et al., 2003); suggesting that these exosomes contribute to the immune suppression that allows the virus to survive. LMP1 also increases the upload and release of fibroblast growth factor 2 (FGF2), a potent angiogenic factor, into exosomes (Ceccarelli et al., 2007). Another study that investigated exosomes released from NPC cells harboring latent EBV (Meckes et al., 2010) showed that exosomes contained EBV-LMP1. Furthermore, uptake of these exosomes resulted in the activation of phosphoinositide 3-kinase (PI3K)/AKT and MAPK/ERK signaling pathways in the recipient cells. Exosomes loaded with LMP1 are associated with increased oncogenicity. How LMP1 enters or manipulates the host exosome pathway is still not clear. The group of Meckes first showed that the host tetraspanin protein CD63 regulates packing of LMP1 into exosomes, as well as LMP1-mediated enhancement of vesicle production, exosomal trafficking, noncanonical NF-κB signaling (Hurwitz et al., 2018), and mTOR signaling (Hurwitz et al., 2017). Next, they investigated the interactome of LMP1 where, among the probable interacting partners of LMP1, they identified components of exosomes or other extracellular vesicles, including molecules involved in vesicular trafficking, such as CD63, syntenin-1, ALIX, TSG101, and HRS. Based on these results, they conclude that LMP1 likely modifies pathways involved in exosome trafficking and biogenesis (Rider et al., 2018).

Although LMP1 plays a crucial role in EBV-mediated malignancies, LMP1 has seldom been detected in EBV-positive gastric tumors. A study by Sato et al. (2017) provides a hint toward the mechanism explaining this controversial observation. They demonstrated that co-culture of LMP1-positive and -negative gastric cancer cells leads to elimination of LMP1-positive cells, through the release of LMP1-loaded exosomes that mediate epidermal growth factor receptor (EGFR) activation in LMP1-negative cells. Assuming that the microenvironment and neighboring cells can also influence viral latency eliminating cells expressing the viral latent gene LMP1 during the early phase of EBV infection of gastric cells, LMP1-negative cells gradually overgrow and replace LMP1-positive cells.

EGFR can be secreted from cells into exosomes and other microvesicles. Subsequent uptake of these EGFR-enriched exosomes by endothelial cells induces activation of MAPK and Akt pathways, and triggers endogenous expression of vascular endothelial growth factor (VEGF), followed by the activation of VEGF receptor-2 (VEGFR-2) (Al-Nedawi et al., 2009). After exosome internalization in an EBV-negative epithelial cell line, exosome-derived LMP1 induced expression of EGFR. Importantly, these non-infected cells produced exosomes that also contained high levels of EGFR (Meckes et al., 2010). In a similar fashion, LMP1 activates expression of hypoxia-inducible factor-1α (HIF1α), a transcriptional regulator under hypoxic conditions that promotes a more aggressive tumor phenotype. LMP1 increases exosome delivery of transcriptionally active HIF1α (Aga et al., 2014). These findings show that, via exosome-mediated transfer, HIF1α promotes pro-metastatic effects in recipient cells.

In summary, exosomes from EBV-infected cells have been shown to incorporate a wide range of molecules such as LMP1, EGFR, FGF2, PI3K, and HIF1α, as well as multiple kinases (Figure 2). Taken together, these observations suggest that the functional properties of exosomes derived from EBV-infected cells mimic the properties of LMP1-expressing cells and thus, possibly regulate signaling pathways in both an autocrine and paracrine manner (Meckes et al., 2013; Meckes, 2015; Raab-Traub and Dittmer, 2017).

Secretion of CagA oncoprotein can induce malignant neoplasms in mammals and is strongly associated with severe gastric lesions, particularly GC. In host cells, CagA affects multiple signaling pathways by acting as an extrinsic scaffold protein (Hatakeyama, 2014; Yong et al., 2015). CagA-positive H. pylori infection may also be associated with diseases outside of the stomach (Franceschi and Gasbarrini, 2007).

To date, only one study (Shimoda et al., 2016) has investigated whether host exosomes may be modified by H. pylori in a manner similar to that in which viruses hijack host exosomes. This study reported that serum-derived exosomes in patients infected with CagA-positive H. pylori were also loaded with CagA. When CagA expression was induced in gastric epithelial cells these cells also secreted exosomes containing CagA. Additionally, it was demonstrated that exosomes deliver functional CagA into gastric epithelial cells, as they were able to induce elongated cell shape in these cells. These findings indicate that CagA can be delivered to distant organs or tissues via secretion of CagA-enriched exosomes into circulation. Thus, these exosomes may be involved in the development of extra-gastric diseases associated with CagA-positive H. pylori infection. Furthermore, exosome-mediated CagA delivery may contribute to the increased risk of colorectal (Strofilas et al., 2012) and pancreatic (Risch et al., 2014) cancers in individuals infected with CagA-positive H. pylori strains.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.