In this study, we identified a high-occupancy EBNA1 binding site in the 5′ promoter control region of the divergently transcribed GKN1 and GKN2 genes. EBNA1 binding sites were observed in two independent ChIP-Seq data sets from EBV positive lymphoid BL cells Raji and EBV positive epithelial nasopharyngeal carcinoma cells C666-1 (Figure 1A). We confirmed these binding sites by conventional ChIP–qPCR in both cell lines (Figure 1B and C). EBNA1 was also shown to bind directly to these sites by EMSA with purified recombinant EBNA1 DBD protein (Figure 1D). We show that GKN1 and GKN2 mRNA levels are highly repressed in most cell lines relative to primary gastric tissue (Figure 2). To study the potential role of EBV and EBNA1 in the transcriptional control of GKN1 and GKN2, we generated an EBV positive AGS gastric carcinoma cell line. We show that EBV adopts a variant type I latency pattern in AGS cells (Figure 3), and that EBNA1 can bind to the GKN1/GKN2 promoter region in the cellular chromosome (Figure 4C). We also found that GKN1 and GKN2 mRNA were further suppressed in EBV positive AGS cells relative to control EBV negative AGS cells (Figure 4D). We then showed that Aza-treatment led to the increase expression of GKN1 and GKN2 (Figure 5A), and that EBV latent infection inhibits Aza activation of GKN2 (Figure 5B). We found that siRNA depletion of EBNA1 in EBV positive AGS cells leads to transcription activation of GKN2 (Figure 6). We also show that EBNA1 ectopic expression moderately increases basal, but inhibits the Aza-induced levels of GKN1 and GKN2 transcription (Figure 7). Taken together, these findings indicate that EBNA1 binds to the GKN1-GKN2 promoter control region in multiple cell types, and raise the possibility that EBNA1 contributes to the transcriptional and epigenetic repression of the GKN1 and GKN2 tumor suppressor genes in EBV positive GC.
EBV latent infection is known to increase the tumorigenic phenotype of gastric carcinoma cells [29–31]. GKN1 and GKN2 are reported to function as cell growth inhibitors and tumor suppressors in GC [20, 21, 23, 25–27]. Our mRNA expression data showing high-level mRNA expression only in primary normal gastric tissue are consistent with a role of GKN1 and GKN2 as a tumor suppressor. However, we were unable to show that over-expression of either or both GKN1 or GKN2 in AGS or AGS-EBV cause a cell cycle arrest or reduce viability (data not shown). This suggests that GKN1 and GKN2 function at earlier stages in tumor cell evolution, or in more complex tumor microenvironments. We speculate that EBNA1 may have a more pronounced effect on GKN1 and GKN2 expression in situations where EBV may infect primary gastric cells where basal expression of GKN1 and GKN2 are high and important for tumor suppression.
Previous published studies have shown that GKN1 and GKN2 transcription is subject to epigenetic suppression by DNA methylation in all forms of GC . Our studies are consistent with the role of DNA methylation in the epigenetic suppression of GKN1 and GKN2 in AGS cells. Treatment with Aza resulted in the 4-10 fold increase in GKN1 and GKN2 mRNA expression (Figure 5A), and MeDIP revealed enrichment of methylated DNA at the promoter regions (Figure 5C). AGS-EBV cells did show an increase in DNA methylation at several cellular sites, including regions surrounding the EBNA1 binding sites at the GKN1 promoter region (Figure 5C), and the HDAC3 and MAP3K7IP2 genes (Figure 5D). However, the presence of EBNA1 in AGS-EBV cells did not prevent Aza-induced demethylation at these sites. This suggests that EBNA1 may repress transcription from some promoters, like GKN2, through a mechanism distinct from DNA methylation. However, ectopic expression of EBNA1 alone produced a more complicated phenotype, causing a small increase in basal expression, but limiting the effects of Aza-induced demethylation (Figure 7). This may suggest that that EBNA1 may function differently when expressed ectopically, than when expressed in the context of the viral genome. Nevertheless, our findings suggest that EBNA1 perturbs the normal transcriptional regulation of the GKN1 and GKN2 genes.
The precise function of EBNA1 in transcription regulation remains unclear. EBNA1 has been implicated in the transcriptional activation and repression of both viral and cellular genes [32, 33]. EBNA1 can repress its own mRNA expression from the EBV Qp in type III latency, where repression has been linked to steric interference with RNA polymerase II binding to the transcription initiation site . On the other hand, EBNA1 can activate Cp and LMP1 promoters in type III latency where it may function as an enhancer-like factor [35–37]. EBNA1 has been implicated in transcription activation of some cellular genes, including the Nox2 gene involved in reactive oxygen species formation . EBNA1 may also affect host-cell transcription through a global remodeling of the host chromosome . Thus, EBNA1 may alter cellular transcription through multiple direct and indirect mechanisms.
Epigenetic modifications are known to play an important role in EBV-associated gastric carcinoma . Interestingly, AGS cells carrying EBV bacmid genomes had higher levels of methylated DNA at many tested sites (Figure 5D). This is consistent with the proposed role of EBV in the methylation of host tumor suppressor genes . This is also consistent with the findings that EBV positive GC has elevated DNA methylation at promoter regions of several key GC tumor suppressors, including gastrokine genes [39, 41–45]. While EBNA1 bound near DNA methylated regions of the GKN2, we were unable to show that EBNA1 modulates DNA methylation at the GKN1 and GKN2 sites (data not shown). However, it is possible that EBNA1 in association with another viral encoded or induced factor may stabilize GKN1 and GKN2 transcriptional repression through a chromatin-dependent and structural mechanism that reinforces DNA methylation. It is also possible that EBNA1 may regulate GKN1 or GNK2 only in tissue or tumor microenvironments that are not readily recapitulated in cell culture. While the function of EBNA1 binding to host cell chromosome sites remains an important area of investigation, more sophisticated infection models may be required to elucidate its potential role in altering host cell gene expression and carcinogenesis.