EBNA1 binding and epigenetic regulation of gastrokine tumor suppressor genes in gastric carcinoma cells

Previously published ChIP-Seq data from Raji BL cells revealed a limited number of highly enriched EBNA1 binding sites based on peak scores and read numbers [17]. Further inspection revealed a strong EBNA1 binding site located upstream of the start sites for the divergently transcribed GKN1 and GKN2 genes (Figure 1A, upper track). A similar EBNA1 binding peak was observed in a separate ChIP-Seq experiment performed in EBV positive nasopharyngeal carcinoma cell line C666-1 (full ChIP-seq data to be published elsewhere), indicating that this binding occurs in both epithelial, as well as lymphoid cell types (Figure 1A, lower track). The center of the peak was located ~ 5 kb from GKN2 and ~15 kb from the GKN1 transcription start sites. ChIP-qPCR was used to validate the EBNA1 binding at the GKN1-GKN2 promoter region in both Raji and C666-1 cells (Figure 1B and C). qPCR indicated that EBNA1 bound to the GKN1-2 site with similar efficiency to a previously validated EBNA1-binding site at the PITPNB promoter. qPCR also indicated that EBNA1 binding to GKN1-GKN2 was specific since there was no detectable EBNA1 binding at either the cellular GAPDH locus or EBV OriLyt control regions, as expected (Figure 1B and C). The putative EBNA1 binding site at GKN1-GKN2 locus was identified by alignment with a consensus binding site in the EBV family of repeats (FR) region, and this sequence was then tested for direct binding to EBNA1 by EMSA (Figure 1D). Purified recombinant EBNA1 DBD protein was assayed for binding to probes containing the GKN1-GKN2 site (GKN1/2), FR, or a negative control sequence lacking a consensus EBNA1 binding site. We found that EBNA1 DBD bound efficiently to the GKN1-GKN2 site, as well as to the FR consensus, but did not bind to the control sequence. These findings suggest that EBNA1 interacts with GKN1-GKN2 promoter region through direct DNA-binding with the EBNA1 DBD.

We next measured GKN1 or GKN2 mRNA levels in various gastric carcinoma cells lines and in primary normal gastric tissue by qRT-PCR (Figure 2). We found that GKN1 and GKN2 are expressed at much higher levels (~3×10⁴ fold) in primary stomach tissue than in any of the cell lines tested (Figure 2A). Although at much lower levels than primary stomach tissue, GKN1 and GKN2 were both expressed at measurable levels in AGS gastric carcinoma cell lines, with GKN2 expressed at higher levels than all other GC cell lines, or EBV positive B-cell or NPC cell lines (Figure 2B). Very low levels of GKN1 or GKN2 could be detected in primary oral epithelial tissue, further indicating that these genes are specific for primary gastric-tissue.

To test the effect of EBV latent infection on GKN1 and GKN2 expression, we generated an AGS cell line containing EBV B95.8 bacmid. The AGS cell line was selected for hygromycin resistance and GFP positivity to ensure that it contained bacmid components. To characterize the AGS-EBV cell line, we first analyzed the EBV gene expression pattern (Figure 3). We found that AGS-EBV cells expressed detectable mRNA levels of EBNA1, BARF0, and LMP2A, but not LMP1, EBNA2, or EBNA3C (Figure 3). Although EBV B95.8 bacmid lacks the BART miRNAs, these findings are consistent with AGS-EBV cells adopting a variant type I latency gene expression program with some expression of LMP2A, similar to that observed in EBV positive gastric carcinoma tumor tissue [8, 10]. To further characterize AGS-EBV cells, we analyzed EBNA1 protein expression by Western blot (Figure 4A) and EBV DNA copy number by qPCR (Figure 4B). EBNA1 was detected as a single low abundance species at the expected molecular mass (Figure 4A). EBV copy number was measured by comparing EBV Ori-Lyt DNA to cellular GAPDH (Figure 4B). We found that AGS-EBV contained ~50% less copies of the viral genome compared to EBV positive LCLs. To determine if EBNA1 retained its DNA binding activity in AGS-EBV, we performed conventional ChIP assays (Figure 4C). We found that EBNA1 bound to the EBV Dyad Symmetry (DS) DNA, as well as to the GKN1-GKN2 binding site in AGS-EBV cells. We next asked whether GKN1 or GKN2 mRNA levels were affected by EBV latent infection in AGS cells by comparing RT-qPCR expression levels in AGS relative to AGS-EBV cells (Figure 4D). We found that GKN1 and GKN2 were repressed ~3-8 fold in AGS-EBV relative to AGS cells, suggesting that EBV latency promotes or stabilizes the transcriptional repression of GKN1 and GKN2.

GKN1 and GKN2 may be subject to epigenetic suppression in GC and tissue culture cell lines. To explore this possibility, we tested whether treatment with DNA demethylating agent 5′ azacytidine (Aza) or deacetylating agent (NaB) combined with phorbol ester (TPA) would activate GKN1 or GKN2 in AGS cells (Figure 5A). We found that Aza treatment led to a ~10 fold increase in GKN2, and ~ 4 fold increase in GKN1 transcription in AGS treated cells. In contrast, NaB/TPA treatment produced only ~2 fold activation of transcription. These findings suggest that both GKN1 and GKN1 are under active epigenetic repression through DNA methylation. To test if EBV had any effect on Aza induced activation of GKN1 or GKN2, we compared the effects of Aza treatment on AGS relative to AGS-EBV cells by qRT-PCR (Figure 5B). We found that GKN1 and GKN2 were efficiently activated by Aza treatment in AGS cells, but to a lesser extent in AGS-EBV. In these experiments, Aza-induced activation of GKN2 was completely eliminated, while activation of GKN1 was only partly attenuated. Aza-treatment also led to the ~8.4 fold increase in EBNA1 mRNA levels, suggesting that Aza treatment stimulates EBV lytic gene expression in AGS-EBV cell lines. These results suggest the EBV gene products prevent GKN2, and to a lesser extent GKN1 activation after Aza treatment.

To better understand the mechanism of GKN1 and GKN2 epigenetic repression, we examined the DNA methylation levels using methyl cytosine-specific antibody directed DNA immunoprecipitation (MeDIP) assay. We assayed the enrichment of methylated DNA in the GKN1-GKN2 control region in AGS and AGS-EBV cells with or without Aza treatment (Figure 5C). We observed a relative enrichment of methylated CpG at a region between the EBNA1 binding site and GKN2 transcription start site (GKN2_A). Aza treatment led to a decrease in MeDIP signal in most regions where a signal was detected, suggesting that Aza treatment led to a general reduction in DNA methylation. Similar loss of CpG methylation was observed at the alpha-globin gene (which does not bind EBNA1), and at several EBNA1 binding sites, including HDAC3 and MAP3K7IP (Figure 5D). We also noted that MeDIP signals were generally higher in EBV-AGS than in AGS cells, suggesting that EBV latent infection may promote or stabilize DNA methylation throughout the host genome.

To determine if EBNA1 contributed to the transcriptional repression of GKN1 and GKN2, we first attempted to deplete EBNA1 in AGS-EBV cells using siRNA (Figure 6). We generated a siRNA targeting the 3′ non-coding UTR of EBNA1 mRNA, which partially depleted EBNA1 protein in AGS-EBV cells (Figure 6B). Depletion of EBNA1 resulted in an ~5 fold activation of GKN2, with little detectable activation of GKN1 (Figure 6A). These findings suggest that EBNA1 may function as a transcriptional repressor of GKN2 in AGS-EBV cells.

To further explore the contribution of EBNA1 to GKN1 and GKN2 transcription regulation, we tested the effect of ectopic expression of EBNA1 alone on Aza-induced levels of GKN1 or GKN2 transcription in GC cells. AGS cells were transduced with an EBNA1 expressing lentivirus. Stable AGS-EBNA1 (pLU-EBNA1) or AGS- control vector (pLU-Vec) cell lines were selected and assayed for GKN1 and GKN2 mRNA levels. We observed that AGS-EBNA1 cells had a ~2-3 fold higher basal level of GKN1 and GKN2 mRNA relative to parental AGS cells (Figure 7A). However, Aza-induced mRNA levels of GKN1 and GKN2 were attenuated in AGS-EBNA1 compared to AGS cells (Figure 7B). Aza treatment also led to a large increase in EBNA1 mRNA levels (Figure 7C). To determine if the effects of EBNA1 on GKN1 and GKN2 were specific to AGS cells, we transduced another EBV negative GC cell line MKN74 with EBNA1 lentivirus (Figure 7D-F). Similar to AGS cells, we found that EBNA1 increased basal levels of GKN1 and GKN2, but inhibited the ability of Aza to further induce GKN1 and GKN2 mRNA levels (Figure 7D and E). We confirmed that Aza-treatment was effective by measuring EBNA1 mRNA levels, which increased ~7 fold (Figure 7F). These findings suggest that ectopic expression of EBNA1 can increase basal, but inhibit Aza-induced levels of GKN1 and GKN2 transcription in EBV-negative GC cell lines.

EBV-positive Burkitt’s lymphoma Raji cells, EBV positive nasopharyngeal carcinoma C666-1 cells, and gastric carcinoma cell lines (a gift from Dr. Antonia R. Sepulveda, Columbia University) including GTL, MKN28, MKN74, SNU638 were maintained in RPMI containing 10% FBS and supplemented with antibiotics (penicillin and streptomycin). Gastric carcinoma AGS cells (ATCC No. CRL-1739) were maintained in F-12K containing 10% FBS. Primary mouth epithelial cells were provided by Dr. Manjunatha Benakanakere, University of Pennsylvania and cultured in Keratinocyte-SFM medium. EBV-LCL was established by primary infection of peripheral blood mononuclear cells (PBMC) with EBV BAC virions generated from stimulated 293-EBV cells [46, 47]. EBV-LCL contains a hygromycin B resistant EBV bacmid were maintained in RPMI containing 10% FBS, hygromycin B (100 μg/ml), glutamax (Invitrogen), and antibiotics. AGS-EBV cells were generated from AGS cells co-cultivated with EBV-LCL by adapting a previously published co-cultivation method described AGS cells infection with rEBV through cell-to-cell contact [48] with some modification. Briefly, EBV-LCL was induced by 20 ng/mL 12-O-tetradecanoylphorbol-13-acetate (TPA) and 1 mM sodium butyrate (NaB) 24 hours prior to co-cultivation. The induced EBV-LCL cells were washed with PBS twice to completely remove the inducing agents, resuspended with complete RPMI medium at 10⁶ cells/ml before the co-incubation. AGS cells were plated in 6-well Plates 24 hours before co-cultivation, then 60 - 70% confluent AGS cells in 1 ml complete F-12K medium were incubated with 10⁶ induced and washed EBV-LCL cells in 1 ml complete RPMI. 24 hours later, 2 ml of Serum free F-12K medium was added to each well to reduce the FBS concentration to 5% to prevent cell overgrowth. 3 days after co-incubation, EBV-LCL cells were removed from the co-cultures and the AGS cells were thoroughly washed with PBS at least 5 times to remove any of the donor EBV-LCL cells. The infected AGS cells was then incubated with fresh F-12K medium with 10% FBS and 100 μg/ml hygromycin B and the selection medium was changed every 2 to 3 days until the infected AGS cells formed Hyg B selection colonies with GFP expression (usually 3 to 4 weeks after co-cultivation). The selection colonies were then pooled and tested for EBNA1 expression and EBV genome copy number before subject to experiments. AGS-EBV cells were maintained in F-12K medium with 10% FBS and 100 μg/ml hygromycin B.

pLU-EBNA1 Lentivirus expression vector was constructed by PCR amplification of EBNA1 with primers (GCGGGATCCTCTGACGAGGGGCCAGGTACAGGACCT and ATCGTCGACTCACTCCTGCCCTTCCTCACCCTCATC) introducing a 5′ BamH I and 3′ Sal I site cloned in frame to pLU-TCMV-FMCS-pPURO. AGS or MKN74 cells were infected with Lentivirus expressing pLU-EBNA1 or pLU control vector generated freshly from 293T cells. Infected AGS or MKN74 cells were then selected with 2.5 μg/ml Puromycin for 10 to 14 days. The selected cells were pooled and treated with or without 5'-azacytidine for 48 hours then subjected to RT-PCR.

siRNA against EBNA1 and siRNA Control (Cat. No. D-001810-01-20) were all purchased from Dharmacon. siRNA directed against EBNA1 3′UTR were synthesized using the target sequence CGGAGAUGACGGAGAUGAAUU. Transfection of siRNA duplexes was conducted by using Oligofectamine (Dharmacon), following manufacturers specifications.

ChIP assays were performed as described previously [49]. Quantification of precipitated DNA was determined using real-time PCR and the delta Ct method for relative quantification (ABI 7900HT Fast Real-Time PCR System; Applied Biosystems). Primers for ChIP assays are available upon request. The following antibodies were used for ChIP assays: rabbit anti-EBNA1 (305/10 wk), anti- rabbit IgG (Santa Cruz Biotechnology, sc-2027). Anti-Actin HRP (Sigma, A3854) and rabbit anti-EBNA1 (305/10 wk) were used for Western Blotting. EBNA1 specific antibody was raised against a bacterial EBNA1 protein lacking the GA-repeat region using New Zealand white rabbits and then affinity purified using the same bacterial protein coupled to sepharose beads.

DNA fragments were labeled using T4 polynucleotide kinase (NEB) in the presence of [γ-³²p]ATP and purified using G-25 spin columns (GE Healthcare). Purified EBNA1 DNA binding domain [17] was incubated with probes at room temperature for 30 min in a total volume of 20 μl buffer containing 10mM TrisCl PH 8.0, 100 mM KCl, 1 mM EDTA, 10 mM MgCl₂, 0.05 μg/μl poly(dI-dC), 0.5 μg/μl bovine serum albumin, 0.05% NP40, 35 mM β-mercaptoethanol, and 10% glycerol. The samples were then separated by electrophoresis on a native 5% polyacrymide gel. Gels were dried and analyzed using a Typhoon phosphorImager system. The forward probe sequences are gatccggatacagattaggatagcatatactaccca (FR), tttcttattttccaggcagcatatactatctcaagctcaccatgtc (GKN1/2) and cgccgccggggcctgcggcgcctcccgcccgggcatggggccgcgc (negative control, LANA binding sites at KSHV TR).

For cell lines, RNA was isolated from 2 × 10⁶ cells using RNeasy Kit (Qiagen) and then further treated with DNase I. Reverse transcriptase PCR (RT-PCR) was performed as previously described [50]. Real-time PCR was performed with SYBR green probe in an ABI Prism 7900 according to the manufacturer’s specified parameters. Primer sequences for RT-PCR are available upon request. Total human stomach RNA (Cat. No. 540037) was purchased from Agilent Technologies.

Genome copy number assay was performed as described previously [51]. The total cellular DNA was assayed by real-time PCR using primers for the EBV Ori-Lyt region and normalized by the cellular GAPDH DNA region. The primers for Ori-Lyt and GAPDH are available upon request.

Methyl-DNA immunoprecipitation (MeDIP) methods have been described previously [51]. MeDIP DNA were analyzed by real-time PCR with delta Ct method for quantification. The primers for MeDIP are available upon request.