CXCR4 induces cell autophagy and maintains EBV latent infection in EBVaGC

Rationale: Epstein-Barr virus (EBV) is found in ~7% of gastric carcinoma cases worldwide, and all tumour cells harbour the clonal EBV genome. EBV can regulate pathways and protein expression to induce gastric carcinoma; however, the molecular mechanism underlying EBV-associated gastric carcinoma (EBVaGC) remains elusive. Methods: GEO microarray and molecular experiments were performed to compare CXCR4 expression between EBV-positive and EBV-negative gastric carcinoma (EBVnGC). Transfections with LMP2A plasmid or siRNA were carried out to assess the role of LMP2A in CXCR4 expression. The effects and mechanisms of CXCR4 on cell autophagy were analysed in vitro using molecular biological and cellular approaches. Additionally, we also determined the regulatory role of CXCR4 in latent EBV infection. Results: CXCR4 expression was significantly upregulated in EBVaGC tissues and cell lines. LMP2A could induce AKT phosphorylation to increase NRF1 expression, thereby binding to the CXCR4 promoter to increase its transcriptional level. Moreover, CXCR4 promoted ZEB1 expression to upregulate ATG7 synthesis, which could then activate autophagy. Moreover, CXCR4 increased the number of cells entering the G2/M phase and inhibited cell apoptosis via the autophagy pathway. Finally, CXCR4 knockdown was associated with elevated BZLF1 expression, but this effect was not influenced by autophagy. Conclusions: Our data suggested new roles for CXCR4 in autophagy and EBV replication in EBVaGC, which further promoted cell survival and persistent latent infection. These new findings can lead to further CXCR4-based anticancer therapy.


Introduction
EBV-associated gastric carcinoma (EBVaGC) accounts for ~7% of all gastric carcinoma cases worldwide [1]. Due to the high incidence of gastric cancer, EBVaGC is the most common among EBVrelated malignancies [2]. The Cancer Genome Atlas (TCGA) divides gastric cancer into four types, namely, EBVaGC, microsatellite unstable tumours, genomically stable tumours, and chromosomally unstable tumours, based on molecular biology [3]. Each EBV-carrying gastric carcinoma is of monoclonal origin, arising from a single EBV-infected cell, indicating that EBV may be associated with gastric cancer formation [4]. Similar to other herpesviruses, EBV exhibits two modes of infection: latent and lytic infections. The switch from latent to lytic infection starts with the immediate-early gene BZLF1, which activates the expression of EBV immediate-early and early lytic genes [5]. In infected tissues, proteins expressed during the lytic and latent viral life cycle lead to cellular alterations that contribute to EBV-Ivyspring International Publisher associated diseases [6]. EBVaGC belongs to the latency I type, with EBNA1 and LMP2A present in 98.1% or 53.8% of cases [7]. LMP2A is a transmembrane protein that functions in numerous signal transduction pathways and is involved in EBVaGC tumorigenesis [8]. It has been proven that LMP2A induces cell stemness in EBVaGC [9]. LMP2A promotes cell survival of B cells through the activation of PI3K/AKT signalling [10]. LMP2A was also associated with the induced expression of phosphorylated STAT3 and the upregulation of DNA methyltransferase DNMT1 [11] and DNMT3B [12] expression. In summary, LMP2A may play an important role in EBVaGC tumorigenesis.
Dysregulation of genes and pathways plays an essential role in tumorigenesis and tumour progression [13]. EBV can regulate host genes and their related pathways to promote cell proliferation and survival, which may directly promote EBVaGC development [14]. PD-L1 expression has been observed more frequently in EBVaGC than in EBV-negative gastric carcinoma (EBVnGC) lesions, suggesting that EBVaGC may evade immunological surveillance by T-lymphocytes via the PD-1/PD-L1 pathway [15]. Moreover, the canonical NF-κB signalling pathway is constitutively activated, and this plays an essential role in EBVaGC tumorigenesis [16]. ADORA1 has been reported to suppress BZLF1 induction, increase tumour cell growth with EBV infection and inhibit sensitivity to the cytotoxic effects of antiviral and anticancer agents [17]. Thus, therapy selection based on the mechanism of tumour development is required, and it is important to diagnose EBVaGC prior to treatment.
Recently, it has been reported that EBV latent genes can also modify chemokine receptor expression [18]. CXCR4, a seven transmembrane G-protein coupled-receptor, acts as the primary receptor for CXCL12. CXCR4 was first discovered during the HIV epidemic when it was noted that certain strains of the virus used CXCR4 as a co-receptor to infect CD4 + T cells [19]. It was proven that the peripheral trafficking of CXCR4 + /CD34 + /CD133 + VSELs was increased in gastric cancer [20] and pancreatic cancer [21], suggesting that CXCR4 may participate in the formation of gastrointestinal cancer. Inhibition of the CXCL12/CXCR4 axis significantly inhibited lymphoma proliferation and survival, suggesting that the CXCR4/CXCL12 axis may participate in EBV-associated lymphomagenesis [22]. CXCR4 could modulate the G-protein signalling pathway to influence the invasion and migration of nasopharyngeal carcinoma cells [23]. However, the association between CXCR4 and EBVaGC has not yet been reported. Our study aimed to explore the role and molecular mechanism of CXCR4 in EBVaGC tumorigenesis.

Cell lines and culture conditions
AGS cells containing recombinant neomycinresistant EBV (AGS-EBV) and control AGS cells were purchased from American Tissue Culture Collection (USA). AGS-EBV cells were classified as latency I type according to the latent products [24]. All cell lines were maintained in DMEM (Gibco, Germany) containing 10% foetal bovine serum (Biological Industries, Israel) and 2% penicillin-streptomycin at 37 °C with 5% CO2.

RNA isolation and real-time quantitative RT-PCR
Total RNA was extracted from the cell lines using TRIzol (Invitrogen, USA) and was reverse transcribed using the First Strand cDNA synthesis kit (Takara, Japan). The products were used as templates for the PCR assays using a FastStart DNA Master SYBR Green Kit (Roche, Germany) in the Light Cycler 96 sequence detection system. PCR was carried out following the manufacturer's instructions. All reactions were performed in triplicate, and β-actin was used for normalization. The 2 -ΔΔCt method was used to assess the relative gene expression. The PCR primers are shown in Table 1.  TCCTGTGGCATCCACGAAACT  GAAGCATTTGCGGTGGACGAT  CXCR4  GCGTCTCAGTGCCCTTTTGT  TGAAGTAGTGGGCTAAGGGC  BZLF1  CCATACCAGGTGCCTTTTGT  GAGACTGGGAACAGCTGAGG  LMP2A  TGTCGCTGGCATACTCTTCA  GCGTGTTAGTCATCACCGTC  EBER1  AGGACCTACGCTGCCCTAGA  AAAACATGCGGACCACCAGC  EBNA1  GGTCGTGGACGTGGAGAAA  GGTGGAGACCCGGATGATG  ATG3  GTTGGAAACAGATGAGGCTACC  CTCAACTGTTAAAGGCTGCCG  ATG5  ATATCAGACAACGACTGAAAGACCT  TCCAACATTGGCTCAATTCCA  ATG7  GCCAAGATCTCCTACTCCAATC  CAGAAGTAGCAGCCAAGCTTGT  ATG12  TGGAACTCTCTATGAGTGTTTTGG  AGCTGTCTCTTCCGTGAAAATC  ATG16L1 TACGCAGCAAAGTCTGCATAAA CCAGGGCAGTAATCTTTCCCA Beclin AAAACCAACGTCTTTAATGCAAC GGAACAAGTCGGTATCTCTGAA

Western blotting analysis
Protein extraction was performed with a RIPA buffer mixture (RIPA:PMSF:phosphatase inhibitors, 100:1:1), and the proteins were then measured with a bicinchoninic acid assay kit (CWBIO, China). The quantified samples were run on 10% or 12% SDS-PAGE gels and were transferred onto PVDF membranes (Millipore, USA) using a transfer device. The membrane was blocked with 5% non-fat milk for 2 h at room temperature, incubated with the primary antibody at 4 °C overnight and then probed with secondary antibody for 2 h. Proteins of interest were visualized by an enhanced chemiluminescence detection system.

Gene Expression Omnibus (GEO) microarray dataset analysis
To validate the expression profiles between EBVaGC and EBVnGC tissues, we obtained the GSE51575 microarray from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). We then analysed the mRNA levels of CXCR4 in EBVaGC and EBVnGC tissues.

Immunohistochemistry
Paraffin-embedded gastric carcinoma tissues were collected from the pathology departments of different hospitals in Shandong Province. EBVpositive cases were identified by in situ hybridization for EBV-encoded small RNA1 (EBER1). Sections were incubated for 1 h at room temperature with anti-CXCR4 antibody (1:200). Antigen-antibody complexes were visualized using a DAB chromogenic kit (Zsbio, China). The specimens were randomized, coded, and then analysed by two independent pathologists. The study was conducted in accordance with the ethical standards and the principles of the Declaration of Helsinki and was approved by the Ethical Committee of the Medical College of Qingdao University. Written informed consent was obtained from each participant before the start of the study.

Plasmid construction
The EBV latent gene LMP2A was cloned into pcDNA3.1 containing the GFP fluorescent protein coding sequence. NRF1 and ZEB1 plasmids were produced by GeneChem Co. (China). Plasmid transfection was performed using the Lipofectamine 2000 transfection reagent (Invitrogen, China) according to the manufacturer's protocols. A total of 2.5 μg of plasmid DNA per six-well plate was used for transfection. Transfection efficiency was measured by qRT-PCR. The cells were harvested after 48 h for downstream analyses.

RNA interference
The siRNA target sequences for LMP2A and CXCR4 mRNA were designed and synthesized by HANBIO (China). Then, 50 μM siRNA was transfected using Lipofectamine 2000 Reagent. The cells were harvested after 48 h to assay relative gene expression.

Transmission electron microscopy
Cells were fixed in 2.5% glutaraldehyde at 4 °C, rinsed with PBS, dehydrated in solutions of increasing ethanol concentrations, and embedded in water permeable London Resin White. Samples were sectioned at a thickness of 90 nm, stained with 1% uranyl acetate for 2 min and lead citrate for 5 min and then examined using an electron microscope.

Immunofluorescence
Cells were grown on cover slips for 12 h, washed with PBS at 37 °C, and fixed with 4% paraformaldehyde at room temperature for 10 min. Fixed cells were blocked in 1% BSA (with 0.15% Triton X-100 and 22.52 mg/ml glycine) for 1 h. The cells were stained with the primary antibodies overnight at 4 °C and incubated with the secondary antibody conjugated with Alexa Fluor 555 for 1 h at room temperature. The nuclei were counterstained with Hoechst 33258 stain (Beyotime Biotechnology, China). Immunofluorescence images were acquired using an Olympus BX63 fluorescence microscope.

Cell apoptosis detection
For apoptosis detection, 1 × 10 6 cells with different treatments were seeded into six-well plates, cultured for 48 h, washed twice with PBS, and then detected using the Annexin V-FITC Apoptosis detection kit (BD Biosciences, USA) according to the manufacturer's instructions. After incubating for 10-15 min in the dark at room temperature, specimens were analysed by fluorescence-activated cell sorting (FACS) using a FACSCalibur apparatus (BD Biosciences, USA).

Propidium iodide staining analysis
Cells were collected 48 h after the indicated treatments, washed with PBS, and fixed in 66% ethanol overnight at 4 °C. Using a propidium iodide (PI) flow cytometry kit (Abcam, USA), the cells were incubated with PI and RNase staining solutions for 30 min at 37 °C in the dark. Finally, based on the standard protocol, > 10,000 cells were detected by FACS using a FACSCalibur apparatus.

Flow cytometric analysis
For each reaction, 1×10 6 cells were suspended in 100 μl FACS buffer (1% BSA/PBS). Subsequently, the cells were incubated for 30 min with an anti-CXCR4 antibody or a matching isotype control at 4 °C. After washing three times with FACS buffer, the cells were incubated for 30 min at 4 °C with a secondary goat anti-rat fluorescein isothiocyanate antibody (Alexa Fluor 647) and washed three times with FACS buffer. Cells were detected by FACS using a FACSCalibur apparatus.

Luciferase reporter assay
To test the effect of transcription factors on the expression of putative target genes, wild-type and mutant 3′-UTR oligonucleotides were subcloned downstream of the firefly luciferase gene in the PGL3-basic vector (GeneChem Co.). HEK293T and AGS cells were seeded into 24-well plates (7 × 10 4 cells/well), and after 12 h of inoculation, cells were transfected with 1.5 µg of recombinant plasmid (wild-type or mutant 3′UTR), 1.5 µg of NRF1 or ZEB1 plasmid, and 300 ng of pRL-TK plasmid per well. Cell lysates were collected after 48 h, and luciferase activity was detected using the dual luciferase reporter system (Promega, USA).

Detection of EBV DNA copy number
EBV nucleic acid amplification fluorescence quantification kits were purchased from DaAn (Guangzhou, China). According to the manufacturer's instructions, extracted DNA was added to the corresponding reagent in the kit. Afterwards, the copy number of EBV DNA was determined using the following procedure: 93 °C for 120 s, followed by 10 cycles of 93 °C for 45 s and 55 °C for 60 s and 30 cycles of 93 °C for 30 s and 55 °C for 45 s. The results were calculated according to a standard curve.

Statistical analysis
Data were analysed using Student's t-test and two-way repeated-measure analysis of variance (ANOVA). Analyses were performed using GraphPad Prism software (GraphPad Software, USA). All data are expressed as the mean ± standard error of the mean (SEM).

The expression of CXCR4 in EBVaGC tissues and cells
We downloaded GSE51575 (EBVaGC expression profiling) from the GEO, including 12 EBV + and 14 EBVgastric carcinoma tissues. The results revealed that CXCR4 was upregulated in EBVaGC compared to EBVnGC samples ( Figure 1A). Next, we analysed CXCR4 expression between 38 EBV-positive (proven by in situ hybridization with EBER1) and 48 EBV-negative gastric carcinoma patients via immunohistochemistry (IHC), and the results were consistent with the expression found in the GSE51575 dataset ( Figure 1B). The expression level of CXCR4 protein is depicted in Table 2. The relationships between CXCR4 expression and patient clinicopathological features, including sex, age, lymph node metastasis, tumour differentiation and invasion depth, were also analysed (Table 3). Finally, CXCR4 expression was significantly higher in AGS-EBV cells than in AGS cells ( Figure 1C-D). The surface expression of CXCR4 protein was also detected to verify the above results ( Figure 1E).

EBV increases CXCR4 expression via the PI3K-AKT pathway
The PI3K/AKT pathway can control a variety of cellular activities, such as proliferation, survival, motility, and morphology. We compared AKT and p-AKT expression between AGS-EBV and AGS cells. As expected, EBV infection led to an increase in AKT phosphorylation (Figure 2A-B). Additionally, we treated AGS-EBV cells with LY294002, which suppressed CXCR4 expression in a dose-dependent manner ( Figure 2C), indicating the involvement of PI3K-AKT signalling in EBV-induced CXCR4 upregulation.  To explore the exact role of the PI3K/AKT pathway in CXCR4 expression, we first searched for transcription factor-binding sites (TFBSs) in the CXCR4 promoter using Jaspar (http://jaspar.genereg. net/). PI3K/AKT signalling was confirmed to be necessary for the nuclear translocation of NRF1. Three TFBSs for NRF1 were identified in the CXCR4 promoter ( Figure 2F), and the dual luciferase reporter assay showed that upregulated NRF1 expression significantly activated CXCR4 promoter activities in 293T and AGS cells ( Figure 2G-H). Additionally, NRF1 expression was suppressed in AGS-EBV cells treated with LY294002 ( Figure 2C-D). Additionally, the overexpression of NRF1 could induce CXCR4 expression in AGS cells ( Figure 2E). These data suggest that the PI3K/AKT signal is involved in CXCR4 expression via the NRF1 factor.

The effect of LMP2A on CXCR4 expression
To gain mechanistic insights into CXCR4 overexpression caused by EBV infection, we detected CXCR4 and p-AKT expression levels by transfection with LMP2A or siLMP2A. qRT-PCR with primers for LMP2A was used to determine the transfection efficiency. The results showed significant increases in CXCR4 and p-AKT expression in AGS cells following LMP2A plasmid transfection ( Figure 3A-B). Moreover, we showed that knockdown of LMP2A (using siLMP2A in AGS-EBV cells) had strong inhibitory effects on CXCR4 and p-AKT expression ( Figure 3C-D). Moreover, exogenous expression of LMP2A significantly induced NRF1 expression ( Figure 3E-F). Collectively, these results indicated that CXCR4 expression in EBVaGC partly depends on the LMP2A-PI3K/AKT-NRF1 pathway.

The induction of autophagy by CXCR4
To investigate whether autophagy was induced by CXCR4 in EBVaGC cells, LC3B was detected by Western blotting in AGS-EBV cells transfected with siCXCR4. The assay revealed that siCXCR4 transfection could suppress the autophagy-related protein LC3B ( Figure 4A). Furthermore, confocal fluorescence microscopy was used to visualize autophagosome formation in cells treated with control or siCXCR4 ( Figure 4B). To confirm these observations, we performed transmission electron microscopy to analyse the samples, as this method is the gold standard for evaluating autophagy ( Figure  4C). Compared with those in control cells, the numbers of autophagosomes decreased in the presence of siCXCR4. To further determine the effect of CXCR4 on cell autophagy, autophagy-related genes were analysed at the mRNA and protein levels after knocking down CXCR4 ( Figure 4D-E). The results showed that CXCR4 could significantly induce ATG7 expression. We found ZEB1 binding sites in the ATG7 promoter through Jaspar (http://jaspar.genereg.net/) and constructed the corresponding ATG7 wild-type and mutant reporter plasmids ( Figure 4G). The results showed that exogenous expression of ZEB1 significantly promoted the luciferase reporter activity of the ATG7 wild-type promoter construct but not the mutant promoter ( Figure 4H-I). We found, for the first time, that the ZEB1 transcription factor activates ATG7 transcription and is involved in autophagy regulation in EBVaGC. Jung et al. previously showed that UHRF1 greatly induced the expression of ZEB1 through CXCR4 [25], and our data confirmed these findings. These results indicate that CXCR4 could stimulate ATG7 expression to increase cell autophagy via ZEB1.

The role of LMP2A in cell autophagy
Studies have shown that LMP2A inhibits anoikis and luminal clearing in acini to prevent cell death by inducing autophagy [26]. However, the effect of LMP2A on cell autophagy in EBVaGC is still unclear. Our results showed that LMP2A could induce cell autophagy in AGS cells ( Figure 5A-C). In addition, LMP2A promoted many autophagy-related genes, especially ATG7 ( Figure 5D-E). These data suggest that LMP2A activates autophagy by increasing ATG7 expression.

The effect of CXCR4 on cell apoptosis and cell cycle progression
Due to the potential role of CXCR4 in tumorigenesis, we explored whether CXCR4 affects cell proliferation and apoptosis through autophagy. As depicted in Figure 6A, siCXCR4 treatment resulted in an increase in cell apoptosis, but cell apoptosis was conversely suppressed in siCXCR4+rapamycin (autophagy inducer)-treated cells. Our results suggested that CXCR4 may affect cell apoptosis via autophagy. As shown in the cell cycle assay, transfection with siCXCR4 decreased the accumulation of G2/M cells, and this effect was inhibited by further incubation with rapamycin ( Figure 6B). These findings indicated that CXCR4 could also induce G2/M phase arrest by autophagy. The effect of rapamycin on autophagy activation is shown in Figure S1.

The role of CXCR4 and autophagy in EBV latency
As it has been hypothesized that autophagy is critical to EBV lytic reactivation and production of viral particles [27]; therefore, we wondered whether CXCR4 is involved in EBV latency in EBVaGC via autophagy. AGS-EBV cells were treated with siCXCR4, rapamycin, siCXCR4+rapamycin, or the corresponding controls. A significantly higher BZLF1 expression was observed in cells treated with siCXCR4 than in cells treated with the control. A similar effect was also observed in the siCXCR4+rapamycin group. However, no difference in BZLF1 expression was observed between the rapamycin and DMSO groups ( Figure 7A-B). It was further verified through immunofluorescence analysis that BZLF1 expression was upregulated in AGS-EBV cells treated with siCXCR4 and siCXCR4+rapamycin ( Figure  7C). The aforementioned data support the notion that rather than autophagy, CXCR4 is required for EBV latency.
Additionally, we further examined the effect of CXCR4 expression on EBV latency. Our results showed that CXCR4 could decrease the EBV DNA copy number with or without the presence of TPA ( Figure 7D-E). We also detected latency-related genes and found that CXCR4 promoted the EBV-encoded latent genes LMP2A and EBNA1 ( Figure 7F). Therefore, CXCR4 may play an important role in maintaining EBV latent infection in EBVaGC development.

Discussion
We identified new classes of differentially expressed genes (DEGs) through a high-throughput screening of GSE51575, among which CXCR4 was selected for further study. Interestingly, the dataset showed that EBV could induce CXCR4 expression in EBVaGC tissues and cells. The mechanism of EBV and CXCR4 expression has not yet been clearly articulated. Studies have shown that CXCR4 is regulated by many factors. CD164 induces activation of the CXCR4 promoter in the nucleus [28]. Extracellular signalling (such as TGF-β) could increase CXCR4 expression in epithelial cells [29]. Moreover, EBV-encoded latent products could regulate CXCR4 expression. Nakayama et al. found that stable expression of EBNA2 and LMP1 downregulated CXCR4 expression [30], and it has been proven that transfection of EBNA2 or LMP1 decreased CXCR4 expression in B cell lymphomaderived cells [31]. However, different EBV-related carcinomas show different types of latent infection and express different latent genes. Our study found that LMP2A induced activation of the PI3K/AKT pathway to upregulate NRF1 expression. NRF1 can bind the promoter region of CXCR4 to enhance its transcriptional level. LY294002, an inhibitor of PI3K, effectively reduced NRF1 and CXCR4 expression. Collectively, these findings indicate that LMP2A increased CXCR4 expression via the PI3K/AKT-NRF1 pathway.  Macroautophagy, also called autophagy, is a highly evolutionarily conserved dynamic recycling system for cell homeostasis [32]. Autophagy selectively degrades long-lived proteins, removes dysfunctional organelles under basal conditions, and promotes cell survival. Autophagy has been implicated in various human diseases, including neurodegenerative diseases [33], cancers [34], heart failure [35], and inflammatory bowel disease [36]. It has been shown that autophagy is dysregulated by EBV to facilitate viral replication and contribute to the onset and maintenance of EBV-associated malignancies [37]. We determined that CXCR4 could promote ZEB1 binding to the ATG7 promoter, thereby activating ATG7 transcriptional expression and inducing the formation of autophagosomes in EBVaGC. Fotheringham et al. found that LMP2A also increased autophagosome formation and the expression of proteins in the autophagosome pathway [38]. LY294002, an autophagy inhibitor, decreased the numbers of double-membraned autophagosomes and autophagic vacuoles [39]. Overall, LMP2A regulation of cell autophagy may partly depend on CXCR4-ZEB1-ATG7 signalling. LMP2A could potentially be required for EBV-induced lymphomas in humans because it is largely dispensable for EBV-induced B cell transformation in vitro [40]. Due to the survival effect of autophagy, LMP2A takes advantage of the autophagy process to enhance EBV-infected cell immortalization. Previous studies have shown that if autophagy inhibition persists, cells can activate pro-apoptotic genes such as BBC3 via FOXO3 and thus become sensitized to cell death [41]. Ravegnini et al. showed that autophagy plays a key role in maintaining gastrointestinal stromal tumour cell survival by providing energy supporting the high energy requirement for metabolism and growth [42]. We demonstrated that transfection with siCXCR4 promoted cell apoptosis in AGS-EBV cells, and rapamycin partially inhibited this effect, which revealed that CXCR4 inhibited cell apoptosis by autophagy. Moreover, Zheng et al. reported that there is a clear crosstalk between autophagy and cell cycle regulation, determining cellular life and death decisions [43]. Beclin increased several G2/M checkpoint-related proteins, PLK1 and CDC25C, to induce the G2/M cell cycle transition [44]. Autophagy induces G2/M cell cycle arrest, accompanied by an upregulation of p21 protein [45]. Similarly, CXCR4 induced the number of G2/M cells, and rapamycin enhanced the proportion of G2/M phase cells. These findings indicated that CXCR4 decreased cell apoptosis and increased G2/M arrest in EBVaGC by promoting autophagy.
EBV alternates between latent and lytic modes of infection. During EBV latency, the viral genome is replicated and maintained in circular form at a constant copy number in the absence of virion production [46]. EBV establishes latent infection in EBV-associated tumours in the presence of latency proteins [47]. BZLF1 expression controls the switch from latent to lytic infection. BZLF1 expression is regulated at the transcriptional level by several activator and repressor proteins that bind various elements of the promoter (Zp). We silenced CXCR4 with siRNA and detected an upregulation of BZLF1 expression, suggesting that CXCR4 acts as a negative regulator of BZLF1 expression. Stephanie et al. showed that C7 reactivated the EBV lytic cycle through subsequent ERK1/2-autophagy activation in EBV-positive epithelial cancers [48]. However, De et al. found that inhibition of autophagy enhanced BZLF1 expression and cell replication in EBV-positive Burkitt's lymphoma [49]. Based on the mechanism of action in EBV lytic reactivation, lytic inducers can be categorized into autophagy-dependent (C7 and iron chelators) and autophagy-independent (HDACi) subclasses [50]. Based on our results, we found that rapamycin has no function in regulating BZLF1 expression, which suggests that siCXCR4 could induce BZLF1 expression by elements other than autophagy. Moreover, inhibition of BZLF1 expression in EBVaGC was beneficial to maintain EBV latency. CXCR4 could decrease the EBV DNA copy number and induce the latent EBV products LMP2A and EBNA1, which confirms the hypothesis that CXCR4 partly contributes to EBV latency.
In conclusion, the key finding in our study is that CXCR4 expression is induced in EBVaGC tissues and cell lines. Moreover, LMP2A promotes CXCR4 transcription via the PI3K/AKT-NRF1 pathway, and CXCR4 activates autophagosome formation by regulating ZEB1-ATG7 transcription. Eventually, CXCR4 inhibits BZLF1 expression and enhances latent gene expression. CXCR4 expression induced by LMP2A could not only contribute to cell growth but also persistently induce EBV latent infection in EBVaGC. Thus, CXCR4 is of importance in EBVaGC development, providing new implications that can be used for novel therapeutic strategies targeting autophagy and EBV reactivation. However, there are still some issues requiring further study. CXCR4 inhibitors or modulators are likely to be toxic, and whether CXCR4 contributes to the prognosis of EBVaGC patients' remains to be elucidated.