Downregulation of HLA-ABC expression through promoter hypermethylation and downmodulation of MIC-A/B surface expression in LMP2A-positive epithelial carcinoma cell lines

Epstein Barr Virus (EBV) is a human herpesvirus, and has been reported to be associated with nasopharyngeal carcinoma, gastric carcinoma, Burkitt’s lymphoma and Hodgkin’s lymphoma. In most of the associated tumors, the virus remains in a latently infected state. During latency, EBV expresses Latent Membrane Protein 2A (LMP2A) along with few other genes. We previously showed that LMP2A causes downregulation of HLA-ABC surface expression in EBV associated gastric carcinomas. However, the mechanism that leads to this downregulation remain unclear. We therefore analyzed methylation-mediated regulation of HLA-ABC expression by LMP2A. Interestingly, according to the ‘missing self’ hypothesis, when there is a decrease in HLA-ABC surface expression, expression of NKG2D ligands’ must be upregulated to facilitate killing by Natural Killer (NK) cells. Analysis of NKG2D ligands’ expression, revealed downregulation of MIC-A/B surface expression in response to LMP2A. Furthermore, the role of Unfolded Protein Response (UPR) in the regulation of MIC-A/B surface expression in cells expressing LMP2A was also investigated. Protein Disulfide Isomerase (PDI) mediated inhibition of MIC-A/B surface expression was observed in LMP2A expressing cells. Our current findings provide new insights in LMP2A arbitrated dysregulation of host immune response in epithelial cell carcinomas.


Results
Decreased expression of HLA-ABC by EBV LMP2A. EBV has been shown to stably alter gene expression through aberrant DNA methylation in EBV infected epithelial cells 36 . EBV latent protein, LMP2A cDNA was cloned within the EcoR1 site of the pcDNA 3.1 vector and stably transfected into EBV-negative gastric cancer cell, AGS. Clones which showed expression pattern of LMP2A similar to that of EBV-positive Korean gastric cancer cell, SNU-719 were selected for the study 12,19 . Protein level measurement of LMP2A expression was estimated in AGS and AGS-LMP2A cells ( Supplementary Fig. 1A). A schematic description of HLA-ABC promoter region is provided with enhancers sites (Fig. 1A) 37 . Transcription Factors (TFs) including Nuclear Factor kappa-lightchain-enhancer of activated B cells (NFkB), Class II major histocompatibility complex Transactivator (CIITA), and Regulatory Factor X (RFX5) are reported to bind the promoter region at the enhancer sites. We previously showed decreased protein expression of the above-mentioned transcription factors in LMP2A expressing gastric cancer cells 19 . A decrease in transcript-level expression of HLA-ABC in gastric cancer cell, AGS expressing LMP2A gene was evaluated through quantitative Real-Time PCR (qRT-PCR) (Fig. 1B). Since epigenetics play important role in regulation of gene expression, we investigated the role of epigenetic regulators in the regulation of HLA-ABC gene expression. First we identified the epigenetic regulators with increased expressions in, AGS expressing LMP2A (AGS-LMP2A) through quantitation of mRNA level compared to vector control cells (AGS) using qRT-PCR. An up-regulated expression of DNMT1, DNMT3B, and UHRF1 at transcript level was observed ( Supplementary Fig. 1B), further validated by immunblotting experiments (Fig. 1C). Next, we explored methylation status associated with repressive levels of HLA-ABC gene. MSP 38 was performed at the promoter region of HLA-A, HLA-B and HLA-C. Promoter hypermethylation was detected in EBV-positive gastric cancer cell, SNU-719, and AGS-LMP2A cells (Fig. 1D). The involvement of epigenetic regulators was further validated LMP2A Mutants were screened for HLA-ABC expression. LMP2A acts as an important viral protein for the maintenance of latent infection. LMP2A consists of an ITAM region similar to B-Cell Receptor which includes eight tyrosine 14 . Two such mutants 1) PY1/PY2 and 2) Y74/85F were designed using 'Site-directed mutagenesis' in the ITAM region of LMP2A ( Fig. 2A). Mutated versions of LMP2A, pcDNA3.1/LMP2A-PY1/ PY2 [LMP2A(PY1/PY2)] and pcDNA3.1/LMP2A-Y74/85F [LMP2A(Y74/85F)] were stably introduced into AGS cells and were selected using G418. The transcript level expression of mutant LMP2A genes was also quantitated in AGS-LMP2A(PY1/PY2) and AGS-LMP2A(Y74/85F) compared to AGS cells expressing pcDNA3.1  Fig. 1C). Clones showing similar expression pattern as AGS-LMP2A were used in the study. HLA-ABC surface expression was also quantitated in AGS-LMP2A(PY1/ PY2) and AGS-LMP2A(Y74/85F) compared to AGS cells (Fig. 2C). HLA-ABC surface expression was further validated in HEK293 cells transiently expressing LMP2A and mutated versions of LMP2A ( Supplementary  Fig. 2C). Decreased surface expression of HLA-ABC was verified in AGS and HEK293 expressing LMP2A and mutated LMP2A(PY1/PY2) genes. However, AGS and HEK293 expressing mutated LMP2A(Y74/85F) showed upregulated surface expression of HLA-ABC compared to vector control cells. Mutants were also screened for the expression of epigenetic regulators, DNMT1, DNMT3B, and UHRF1. Role of Y74/85F mutation within the ITAM region of LMP2A in the expression of epigenetic regulators at the transcript level of DNMT1, DNMT3B and UHRF1 was validated in AGS-LMP2A(PY1/PY2) and AGS-LMP2A(Y74/85F) cells. Decreased transcript-level expression of DNMT1, DNMT3B, and UHRF1 was observed in case of AGS-LMP2A(Y74/85F). AGS-LMP2A(PY1/PY2) exhibited similar expression pattern as the AGS-LMP2A cells ( Supplementary Fig. 1D). Consistent with the previous results, immunoblotting analysis also showed that Y74 and Y85 in the ITAM region is responsible for the upregulated expression of epigenetic regulators (Fig. 2D). Increased methylated products of HLA-A, HLA-B, and HLA-C promoter regions were observed in AGS cells stably expressing LMP2A and LMP2A(PY1/PY2) genes through qRT-PCR (  Increased expression of UPR protein markers. NKG2D ligands' are recognized to be regulated by a number of cellular stress, thus regulating the activation of NK cells [40][41][42][43][44] . UPR has recently been reported to regulate the expression of certain NKG2D ligands' 45 . Gastric cancer cell, expressing LMP2A (AGS-LMP2A) was analyzed for expression of UPR proteins, iRE1-ɑ, BIP, PDI, and CHOP. Immunoblotting analysis revealed increased expression of UPR proteins in AGS-LMP2A (Fig. 5A). EBV-negative gastric cancer cell, SNU5 transiently expressing LMP2A and LMP2A(PY1/PY2) also showed increased expression of PDI ( Supplementary  Fig. 3A). Decreased expression of UPR proteins was detected in AGS-LMP2A (  www.nature.com/scientificreports www.nature.com/scientificreports/ similar expression pattern as AGS-LMP2A cells (Fig. 5D). UPR proteins were upregulated in response to LMP2A and LMP2A(PY1/PY2) by HEK293 cells (Fig. 5E).

Restoration of Mic-A/B surface expression.
Disulfide isomerases are known to regulate cancer migration through modulation of invasive properties 35 . PDI is a well-known UPR protein recognized to catalyze proteolytic shedding of cell surface proteins 33 . We sought to determine the role of PDI and viral oncoprotein, LMP2A in regulation of MIC-A/B surface expression. An increased surface level expression of MIC-A/B was observed in AGS-LMP2A and SNU-719 cells upon siRNA-mediated knockdown of LMP2A expression (Fig. 6A). Role of PDI in surface expression of MIC-A/B in LMP2A-expressing epithelial cell carcinoma was investigated through knockdown study of PDI, mediated by siRNA. MIC-A/B protein and surface-level expressions were restored in AGS-LMP2A cells (Fig. 6B,C) and SNU-719 cells (Fig. 6D,E). Knockdown analysis reveals the role of PDI in regulation of MIC-A/B surface expression through proteolytic shedding. AGS cells stably expressing the Y74/85F mutant version of LMP2A failed to show similar results as that of AGS-LMP2A cells. LMP2A ITAM region mutation (Y74/85F) was unable to exhibit similar phenotype as normal LMP2A gene. Previous reports show activation of PI3k-Akt pathway 15 and Sonic Hedgehog (Shh) pathway 19 by the viral latent protein, LMP2A. In order to investigate the role of these signaling pathways in regulation of MIC-A/B expression, we used Forskolin (PI3k/ Akt inhibitor) and LY294002 (Shh inhibitor) in a dose-dependent manner. The surface expression of MIC-A/B was significantly enhanced in AGS-LMP2A cell (Supplementary Fig. 4D) and SNU-719 cell ( Supplementary  Fig. 4E) upon Forskolin treatment. These results suggest that decreased MIC-A/B surface expression in EBV infected epithelial cell carcinomas is due to activation of Shh pathway. Next, we further analyzed the role of UPR in MIC-A/B surface expression through thapsigargin (an UPR inducer drug) treatment. AGS (Fig. 6F,G) and

Discussion
We previously reported EBV has developed strategies to avoid detection of infected cells by CTLs and compromising the host's ability to overcome EBV infection in EBVaGC 17,18 . The current study addresses the role of EBV latent protein, LMP2A in the regulation of host immune response through epigenetic regulators and UPR in LMP2A-positive epithelial cell carcinomas. We first explored the role of epigenetics in the downregulation of HLA-ABC surface expression in response to viral oncogene, LMP2A. Furthermore, since alterations in epigenetic marks have recently been linked to the regulation of HLA class I gene expression 46 , we, therefore, proceeded to investigate the role of LMP2A in regulating the gene expression through epigenetic regulators. The promising aspect of epigenetic regulators in tumorigenesis has recently been investigated in several carcinomas 47 . Our results provide evidence for LMP2A mediated increased methylation of the HLA-ABC promoter region along with elevated levels of epigenetic regulators; DNMT1, DNMT3B, and UHRF-1. Hence it is significant that methylation mediated modification of HLA-ABC promoter region is utilized by the virus to target the availibility of HLA-ABC on the surface of infected cells. Methylation mediated downregulation of HLA-ABC expression was justified by treatment with 5′-azacytidine, which acts as a demethylating agent. An increased HLA-ABC surface expression was observed upon demethylation treatment in LMP2A-expressing gastric cancer cell, AGS and SNU-719. We also showed that LMP2A(Y74/85F) mutant failed to exhibit smiliar phenotype as LMP2A validating the role played by Tyr74 and Tyr85 in the downregulation of HLA-ABC expression.
Apart from CTL response, NK cell response provides protection against viral infection. NKG2D ligands' expression on the infected cells play important role in recognition of virus-infected cells by the counteracting NK cell. Expression of NKG2D ligands' is transcriptionally regulated by various types of cellular stresses associated with transformation, viral infection, or other stress signals to the host cell. The stress signals that alerts the immune cells includes DNA damage response 42 , oxidative stress 40 , and UPR 45 during cancer progression. In an www.nature.com/scientificreports www.nature.com/scientificreports/ earlier study, EBV was shown to downregulate MIC-B through BARTs 27 . LMP2A-deficient EBV infected lymphoblastoid cell lines (LCLs) showed increased susceptibility to CTLs 48 . HLA-E expression also plays important role in desensitization of NK cells 49 . UPR proteins acts as an important regulator for viral persistence and function. Our current findings were in good agreement with EBV latent protein, LMP2A mediated loss of MIC-A/B surface expression. Meanwhile expression of UPR proteins was also altered in LMP2A expressing epithelial cell carcinomas. It might be suggested that these expression alterations of UPR proteins might result in regulation of MIC-A/B expression. PDI is localized on the plasma membrane as well as within the cytoplasm, where it is reported to facilitate proteolytic shedding of cell surface proteins 35 . We showed that PDI resulted in regulation of MIC-A/B surface expression in LMP2A-expressing epithelial cell carcinomas. Thus, PDI could be used as drug target to prevent tumour immune evasion. We verified the role of ITAM (Tyr74 and Tyr85) in the downregulated expression of MIC-A/B mediated by PDI. Replacement of LMP2A with LMP2A ITAM mutant (Y74/85F) was successful in releasing the effect viral protein on the surface availibility of MIC-A/B in epithelial cell carcinomas. ITAM region is reported to activate several cellular signaling cascades including PI3k-Akt and Shh pathways. Thus the two pathways were targeted using LY294002 and Forskolin, respectively. Restoration of MIC-A/B surface expression was observed in response to Forskolin treatment, suggesting the role of the Shh pathway in the regulation of MIC-A/B surface expression. The current study highlights, regulation of CTL response through decreased HLA-ABC expression along with modulation of NK cell response through downregulation of MIC-A/B surface expression in LMP2A expressing epithelial cell carcinomas (Fig. 7).
In summary, EBV-LMP2A acts as a potent regulator of host immune response. Evasion of immune response acts as a marker for poor prognosis in cancer patients, particularly in epithelial carcinomas. These regulation of www.nature.com/scientificreports www.nature.com/scientificreports/ immune response through promoter hypermethylation and UPR proteins provides better understanding about the strategies utilised by EBV to overcome recognition by immune cells.

Materials and methods
Materials. Thapsigargin, 5′-azacytidine, Forskolin, along with LY294002 were purchased from Sigma Aldrich and dissolved in DMSO for experimental treatment, while G418 was purchased from Gibco and solubilized in PBS for the selection of clones stably expressing pcDNA-3.1 along with the gene of interest. The HLA-ABC and MIC-A/B antibody was procured from BD Biosciences. Antibodies against DNMT1, DNMT3B, UHRF1, MIC-A, MIC-B, LMP2A, and β-ACTIN were purchased from Abcam and BIP, CHOP, IRE1 ɑ and PDI primary antibodies, along with anti-rabbit, anti-mouse, anti-rat horseradish peroxidase (HRP)-conjugated secondary antibodies were procured from Cell Signaling Technology. Antibodies' information is provided in Supplementary Table 3.
Cell Culture. Human gastric cancer cell lines, AGS (EBV-negative) and SNU-719 (EBV-positive) 12,18,19 were cultured in RPMI 1640 (Gibco) supplemented with 10% (v/v) heat-inactivated fetal bovine serum, FBS (Gibco). Human Embryonic Kidney 293 (HEK293, EBV-negative) 19  RT-PCR and quantitative real-time PCR. RNA was extracted through Tripure isolation reagent (TRIZOL, Roche) and quality assayed using spectroscopy (Eppendorf BioPhotometer). 2ug of the total RNA was subjected to reverse transcription for 90 min at 42 °C using oligo dT primers and M-MuLV reverse transcriptase (Fermentas). Quantitative RT-PCR was performed using SYBR Green core PCR reagents (Applied Biosystems) and HPRT was used as the endogenous control. The qRT-PCR reactions and analysis were carried out in the 7500 Sequence Detection System (Applied Biosystems) 12,19 . Primer sequences used for qRT-PCR analysis in this study are provided in Supplementary Table 1. www.nature.com/scientificreports www.nature.com/scientificreports/ Bisulfite Modification and Methylation Specific PCR (MSP). 1 × 10 5 cells were harvested and dissolved in 100 mL 5% PBS was used for further experiment. EpiTect Fast LyseAll Bisulfite Kit (Qiagen) was used to carry out bisulfite modification as per the manufacturer's instructions. The modified DNA samples were finally stored at −20 °C until further use. The methylation status of the cell lines was determined using MSP. Primer sequences used for MSP in this study are shown in Supplementary Table 2. Polymerase chain reaction (PCR) was carried out in a volume of 50 μl high GC buffer with 100 ng or less template DNA with TaKaRa EpiTaq HS (TaKaRa). There were 35 cycles of denaturation at 98 °C for 10 s, annealing at 60 °C for 30 s, and extension at 72 °C. The amplified product was assayed using DNA gel electrophoresis and photographed using a Gel Doc chemiluminescence detector (BioRad). The qRT-PCR reaction was performed using SYBR Green core PCR reagents and bisulfite modified genomic DNA as a template and methylated GAPDH was used as the endogenous control. Primer sequences used for qRT-PCR for methylation study are provided in Supplementary Table 2.
Azacytidine treatment. The cell lines were plated at low density, incubated with 2uM and 4uM 5′-Aza-2′-deoxycytidine for 6 days, independent sets were replenished with fresh 5′-Aza-2′-deoxycytidine every 24 hr. The cells were harvested after treatment for surface expression of HLA-ABC.
Western blot analysis. Cells were harvested and lysed utilizing Nonidet P-40-lysis buffer [20 mM Tris-HCl (pH 8.0), 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 2 mM EDTA, 200 mM Na 3 VO 4, 100 mM phenylmethylsulfonyl fluoride, protease inhibitor cocktail (Roche) and phosphatase inhibitor (Roche, Mannheim, Germany)] for 30 min on ice, further centrifuged at 16,000 g for 5 min at 4 °C and the supernatant was collected for further analysis. Equal amounts of protein (25-50 mg) were subjected to 10-12% SDS-polyacrylamide gel electrophoresis and electrotransferred onto a nitrocellulose membrane (GE Healthcare). The membrane was subjected to blocking with 5% BSA in TBS containing 0.1% Tween 20 (Sigma Chemical Co.) for an hour at room temperature. Primary antibodies mentioned previously were incubated with membranes at 4 °C overnight. Membranes were then washed in TBS-T, probed with secondary antibodies conjugated to horseradish peroxidase (Cell Signaling Technology) for an hour (1:3000 dilution in TBS-T), and then washed with TBS-T. The antibody-bound protein bands were detected through enhanced chemiluminescence reagent (Amersham Biosciences) and photographed using a VersaDoc chemiluminescence detector and Quantity ONE software (BioRad) 12,19 .
RNA interference. The siRNA sequence for inhibiting LMP2A mRNA expression was 5′-AACUCCCAAUAUCCAUCUGCU-3′, identical to that previously reported 12,19,50 . The siRNA sequence for inhibiting PDI was 5′-GGACCAUGAGAACAUCGUC-3′ similar to that previously reported 51 . A control scrambled sequence was also used as Mock (Sigma Aldrich, St. Louis, MO). The siRNAs were delivered utilizing Lipofectamine 2000 (Invitrogen) and the cells were further assayed at different time intervals.
Cell surface analysis of HLA-ABC. 1 × 10 5 cells were dissolved in 100 uL 5% PBS and incubated with allophycocyanin (APC) conjugated anti-HLA-ABC monoclonal antibody for a time period of 30 min in dark at 4 °C. The amount of antibody used was as per the manufacturer's instructions. Stained samples were then subjected to PBS wash, further resuspended in 1 ml 5% PBS and endured to Fluorescence-Activated Cell Sorting (FACS) Analysis through FACS Calibur (Becton Dickinson). The results were analyzed using Cell Quest Pro software (B.D. Biosciences) 19 .
Cell surface analysis of MIC-A/B. 1 × 10 5 cells were dissolved in 100 uL 5% PBS and incubated with phycoerythrin (PE) conjugated anti-MIC-A/B monoclonal antibody for a time period of 30 min in dark at 4 °C. The amount of antibody used was as per the manufacturer's instructions. Stained samples were then subjected to PBS wash, further resuspended in 1 ml 5% PBS and endured to Fluorescence-Activated Cell Sorting (FACS) Analysis through FACS Calibur (Becton Dickinson). The results were analyzed using the Cell Quest Pro software (B.D. Biosciences).
Thapsigargin (Tg) treatment. The cell lines were plated at low density, incubated with 1uM and 2uM thapsigargin for a time period of 6 hr. The cells were then harvested for protein level expression of PDI and surface expression of MIC-A/B. Statistical analysis. Results obtained from all the experiments were reported as the mean ± s.e.m. The significance of the analyzed data was calculated by a standard student's t-test. The gene expression levels were compared by unpaired two-tailed t-test. Only P < 0.05 was considered statistically significant.