Regulation of EBV LMP1-triggered EphA4 downregulation in EBV-associated B lymphoma and its impact on patients’ survival

Epstein-Barr virus (EBV) is oncogenic and its infection is associated with multiple human diseases, especially lymphoproliferative disorders, such as infectious mononucleosis, Burkitt lymphoma (BL), Hodgkin disease, natural killer/T-cell lymphoma, posttransplant lymphoproliferative disorder (PTLD), and EBV⁺ diffuse large B-cell lymphoma (DLBCL).^1,2  In vitro, EBV can transform human primary B cells into lymphoblastoid cell lines (LCLs), which support its association with human malignancies. Like other human herpes viruses, EBV may be found in 2 states: latency and lytic replication. It is well documented that viral latent products contribute largely to EBV oncogenic activities. The main EBV latent products include 6 EBV nuclear antigens (EBNA1, 2, LP, 3A, 3B, 3C), 3 latent membrane proteins (LMP1, 2A, 2B), 2 EBV-encoded small RNAs (EBER1, 2), and microRNAs (Bam H1 rightwards transcripts).¹  According to previous studies, EBV uses various strategies to manipulate cell gene expression so that it can persist in infected B cells. These include altering the expression of protein tyrosine kinases (PTKs), cytokines, adhesion molecules, and antiapoptotic genes.^1,3-8  Among them, we are particularly interested in PTKs because they are important at many aspects of cell proliferation, differentiation, apoptosis, migration, and tissue development. In total, 90 PTKs have been defined following sequencing of the human genome.⁹  They can be divided into 2 major categories: 58 receptor tyrosine kinases (RTKs) and 32 non-RTKs.⁹  In the case of EBV-associated malignancies, several EBV gene products are involved in dysregulation of RTKs. EBV-encoded LMP1 elevates amounts of Recepteur d’Origine Nantais via NF-κB to promote B-cell proliferation and induce the migration and invasion of epithelial cells.^3,4  In addition, epidermal growth factor receptor is induced and activated by LMP1 through protein kinase Cδ and NF-κB.^10-12  Also, c-Met is upregulated by LMP1 via Ets1 in epithelial cells.¹³  Activation of discoidin domain receptor 1 by LMP1 contributes to the survival of Hodgkin lymphoma.¹⁴  Recently, LMP1 has been reported to increase the expression of fibroblast growth factor receptor 1, enhancing aerobic glycolysis and cell invasiveness.¹⁵  Furthermore, LMP1 activates the insulin-like growth factor 1 receptor via upregulation of its ligand, IGF1, to promote cell proliferation and colony formation.¹⁶  In addition, Zta, a lytic EBV transactivator, can upregulate the expression of Trk-related tyrosine kinase, which may enhance the metastasis of nasopharyngeal carcinoma.^5,6  Human epidermal growth factor receptor 2 (HER2) and HER3 are dysregulated by EBV-encoded BARF0, which not only increases the level of HER2/3 but also enhances cell-anchorage independence.¹⁷  In the cases mentioned above, many EBV products are responsible for increasing the expression or enhancing the kinase activity of RTKs to deliver signaling that favors cell proliferation, migration, or invasion. Herein, we wondered whether there are other mechanisms involved in EBV immortalization of B cells through other RTKs. In our previous study, a kinase display was used to investigate the differential expression of PTKs between primary B cells and LCL. Recepteur d’Origine Nantais and Trk-related tyrosine kinase were found to be upregulated.^3-6  In this manuscript, we reveal another RTK, namely erythropoietin-producing hepatocellular receptor A4 (EphA4), which is involved in EBV lymphoproliferation.

Of note, Eph receptor is the largest family of RTKs in humans, being composed of 9 EphAs and 5 EphBs. Usually, EphA receptors bind to glycosylphosphatidylinositol-linked Eph receptor-interacting protein (ephrin)-A ligands, and Eph B receptors bind to transmembrane ephrin-B ligands.^18,19  The interactions between Eph receptors and their ligands trigger both forward Eph kinase activity in Eph receptor-expressing cells and reverse Src family kinase activity in ephrin-expressing cells.¹⁹  These bidirectional signaling pathways make the Eph family distinct from other RTKs. Conventionally, their functions are recognized in developmental processes and tissue homeostasis.²⁰  Recently, dysregulation of Eph receptors and ephrins has been reported in many human cancers.^18,19,21,22  Of interest, various reports indicated that Eph can function in oncogenic promotion or tumor suppression in several human cancers at different stages or in different cancers.^18,19,21,22  Notably, the involvement of Eph in viral infection has been explored in some human oncogenic viruses. EphA2 is the entry receptor for Kaposi sarcoma-associated herpesvirus²³  and also acts with epidermal growth factor receptor as a coreceptor for hepatitis C virus entry.²⁴  Collectively, Ephs not only play roles in physiological activities but also are involved in pathogenesis.²⁵  According to our kinase display data, EphA4 is clearly downregulated in LCL, compared with uninfected B cells. Little is known about the role of EphA4 in B lymphocytes, so the expression and regulation of EphA4 in LCL and EBV-associated diseases will be explored in this article.

Peripheral blood mononuclear cells were kindly provided by anonymous donors at Taipei Blood Center of Taiwan Blood Service Foundation, and CD19⁺ B cells were purified as described previously.⁷  The methods of EBV virion (B95.8 strain) production and EBV infection of primary B cells have been described previously.⁷  Experiments using human samples were approved by institutional review boards of National Taiwan University Hospital, Taipei, Taiwan.

Human peripheral blood CD19⁺ B cells were infected by B95.8 EBV to establish LCLs in vitro. BJAB cells are EBV⁻ BL cells. All B-cell lines were cultured in complete RPMI medium containing 10% fetal calf serum, 1 mM glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. A nasopharyngeal carcinoma cell line, TW01, was maintained in complete Dulbecco modified Eagle medium containing 10% fetal calf serum, 1 mM glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin.

The tyrosine kinase display assays were performed as described previously.⁵  Briefly, total RNAs were extracted and complementary DNAs (cDNAs) of PTKs were generated and amplified by using [γ-³³P]-labeled degenerate primers designed for the conserved kinase domain. The resulting 170-bp polymerase chain reaction (PCR) products were gel-purified and digested separately using 16 restriction enzymes. The digested PCR products were resolved in a denaturing polyacrylamide gel. The patterns of restriction enzyme digestion were identified for each specific PTK in the data bank.

The EBNA1-, EBNA2-, Zta-, and LMP1-expressing plasmids were constructed as described previously.⁷  The luciferase reporter plasmid, driven from the EPHA4 promoter (nucleotides −1000 to +42), was inserted into the pGL3-basic vector (Promega). The LMP1-expressing lentivirus plasmid, pSIN-LMP1, and its mutant plasmids, including LMP1-deleted C-terminal activation region 1 (CTAR1; ΔCTAR1) with the deletion of LMP1 amino acids 194 to 232, deleted CTAR2 (ΔCTAR2) with deletion of LMP1 amino acids 351 to 386 and LMP1-deleted CTAR1/2 (ΔCTAR1/2) plasmids were reported previously.³  The shLMP1, shSp1 lentivirus plasmids and its vector control shLuc were constructed in the pLKO.1 plasmids as described previously.^3,26  The pLKO.1-shERK1/2 lentivirus plasmids were purchased from National RNAi Core Facility (Academia Sinica, Taipei, Taiwan) with sequences targeting human extracellular signal-regulated kinase 1#1 (ERK1#1) (5′-CTATACCAAGTCCATCGACAT-3′), ERK1#2 (5′-TCCCTGTCAAAGCTGTCACTT-3′), and ERK2 (5′-TATCCATTCAGCTAACGTTCT-3′). The EphA4 (pCDNA3.1-EphA4) full-length (wild-type [WT]) and mutant plasmids are kindly provided by Dr Tang-Long Shen (Department of Plant Pathology and Microbiology, National Taiwan University, Taipei, Taiwan). The Flag-tagged EphA4 WT and mutant lentivirus plasmids were constructed into pWPI lentiviral plasmid, which is a gift from Dr Min-Liang Kuo (Institute of Toxicology, National Taiwan University, Taipei, Taiwan).

Total RNAs were extracted by using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. Reverse transcription and quantitative PCR (RT-Q-PCR) has been described previously.²⁶  EphA4 messenger RNA (mRNA) was determined using forward primer 5′-GATAGCAAGCCCTCTGGAAG-3′, reverse primer 5′-CCAATCAGTTCGTAGCCAGTT-3′, and Roche no. 20 universal probe. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA were determined by using forward primer 5′-TCCACTGGCGTCTTCACC-3′, reverse primer 5′-GGCAGAGATGATGACCCTTTT-3′, and Roche no. 45 universal probe. MAGOH mRNA were determined by using forward primer 5′-AAAGAGGATGATGCATTGTGGt-3′, reverse primer 5′-TCTTCTCCAATGACGATTTCA T-3′, and Roche no. 51 universal probe.

Cells were lysed using radioimmunoprecipitation assay buffer and western blotting was performed as per previous study.⁷  Antibodies (Abs) were used as follows: EphA4 (Santa Cruz Biotechnology and ECM Biosciences), β-actin (Sigma-Aldrich), phospho-Akt S473 (Cell Signaling), Akt (Santa Cruz Biotechnology), phospho-IκB-α-S32/36 (Cell Signaling), p65 (Santa Cruz Biotechnology), phospho-JNK T183/Y185 (Cell Signaling), JNK (Millipore), phospho-ERK T202/Y204 (Cell Signaling), ERK (Santa Cruz Biotechnology), Sp1 (Santa Cruz Biotechnology). Abs against EBV viral products including EBNA1, EBNA2, LMP1, Zta were as reported previously.^7,27 

The method of production and infection with lentivirus was reported previously.⁷  For EphA4 WT and mutant-expressing lentivirus infection, 5 × 10⁵ LCLs were infected with lentivirus at a multiplicity of infection (MOI) of 4. For LMP1, Sp1, and ERK1/2 knockdown, LCLs were seeded at a density of 1 × 10⁶ cells/mL and infected with shLuc control, shLMP1, shSp1, or shERK1/2 lentivirus at MOI of 1. BJAB cells were seeded in 1 × 10⁶ cells/mL and infected with pSIN-LMP1 and its deletion mutant expressing lentivirus at a range of 0.5 to 4 MOI.

LCLs were seeded at 1 × 10⁶ cells/mL and treated with 20 μM phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002, 20 μM JNK inhibitor SP600125, 20 μM MEK inhibitor PD98059, 2.5 μM NF-κB inhibitor BAY 11-7082 for 48 hours, respectively. TW01 cells were treated with 500 nM Sp1 inhibitor mithramycin for 24 hours. The inhibitors mentioned above were purchased from Merck Millipore. Dimethylsulfoxide served as a solvent control.

TW01 cells were seeded at a density of 1 × 10⁵ cells per well in a 12-well plate and then cotransfected with 0.5 μg of EPHA4 promoter luciferase reporter plasmids, 0.25 μg of pSG5-LMP1, and 0.05 μg of green fluorescent protein (GFP)-expressing plasmids (pEGFP-C1; Promega) using T-Pro Non-liposome transfection Reagent II (T-Pro NTR II; T-Pro Biotechnology) according to the manufacturer’s instructions. Cells were harvested and the luciferase activities and GFP fluorescent intensity were detected using the Bright-Glo Luciferase Assay System kit (Promega) 2 days posttransfection. The relative fold induction of luciferase activity from each transfectant was first standardized with enhanced GFP, followed by normalization to the control vector pGL3.

Lentivirus-infected LCLs were seeded at 1 × 10⁴ cells per well in 96-well plates for 5 days. Before the indicated time point, cells were treated with AlamarBlue (Thermo Fisher Scientific) for 4 hours and the absorbance measured according to the manufacturer’s instructions.

BJAB vector control and LMP1-expressing cells were harvested. DNA-protein complexes were immunoprecipitated using anti-Sp1 antibody as described previously.²⁸  The DNA was extracted and analyzed by PCR with the EPHA4 promoter spanning the Sp1-binding sites. The amplification of the GAPDH promoter region was as described previously.⁷ 

Tonsil, PTLD, and DLBCL biopsies were obtained from the National Taiwan University Hospital. Immunohistochemistry (IHC) assays of EphA4 and LMP1 were performed using the Starr Trek Universal Detection system (Biocare) according to the manufacturer’s protocol, and in situ hybridization of EBER assays was described in our previous study.³ 

The microarray data sets of DLBCL were downloaded from the National Center for Biotechnology Information (NCBI) and Gene Expression Omnibus (GEO) database (accession number GSE4475). Overall survival was analyzed by generating Kaplan-Meier plots. Patients were divided into high and low EphA4 groups, according to the median expression of EphA4.

Statistical analysis was conducted using GraphPad Prism software program. Quantitative data were reported as mean ± standard error of the mean and compared using the unpaired Student t test. Clinical correlation was determined by Fisher exact test. For Kaplan-Meier survival analysis, a log-rank test was used to compare the difference between 2 groups. Data were considered statistically significant at P < .05.

Because RTKs play a critical role in EBV-infected cells, the RTK expression profile was analyzed by kinase display assay using degenerate primers.⁵  According to the BstNI, HhaI, and MnlI restriction enzyme digestion profile, downregulation of EphA4 was found in 3 LCLs compared with control primary B cells (Figure 1A). To confirm this result, mRNA and protein expression levels of EphA4 were detected in cells infected with EBV for 3, 7, 14, 21, or 28 days (defined as LCL), compared with uninfected primary B cells. Clearly, mRNA and protein expression levels of EphA4 were reduced to ∼50% to 10% post-EBV infection, compared with primary B cells (Figure 1B-C). Expression of EBNA1 and LMP1 indicated that the primary B cell was successfully transformed by EBV (Figure 1C). In addition to EBV infection, B cells can be activated by various treatments including anti-CD40 antibody plus interleukin-4 (IL-4), lipopolysaccharide (LPS), and poly I:C to mimic T-cell–dependent or –independent activation.²⁹  To clarify whether B-cell activation can decrease EphA4 expression, EphA4 transcripts were measured after 3 days of incubation. As shown in Figure 1D, EphA4 was downregulated not only by EBV infection but also by anti-CD40 antibody plus IL-4 treatment. This fit previous findings that EBV LMP1 functionally mimics the constitutively active CD40 receptor associated with tumor necrosis factor receptor-associated factors.^30,31  Meanwhile, treatments of LPS or poly I:C which triggered the Toll-like receptor 4 or 3 signal pathway did not alter EphA4 expression.

To determine whether this effect is common in B cells infected by EBV, expression of EphA4 mRNA was detected by RT-Q-PCR in another 8 LCL cell lines established in our laboratory.⁷  Compared with primary B cells, EphA4 transcripts were also detected at 10% to 50% normal levels in LCLs (Figure 1E). To verify these observations, total RNA and protein were collected from 2 paired primary B cells and their corresponding LCLs. Consistent with Figure 1B-C, EphA4 mRNA and protein expression were reduced in LCLs, suggesting that EBV infection influenced the expression of EphA4 at the transcriptional and translational levels (Figure 1F-G).

Many Epstein-Barr viral products are able to regulate gene expression. To determine which Epstein-Barr viral gene was involved in EphA4 suppression, selected viral genes were expressed ectopically in the EBV⁻ nasopharyngeal carcinoma cell TW01 through transfection. Based on the results of Figure 2A-H, we demonstrated that LMP1 was the candidate protein that inhibits EphA4 expression. Other EBV gene products, including Zta, EBNA2, EBER1, EBER2, and BARF0, had no effect on the EphA4 expression level (Figure 2C-H). This LMP1-mediated EphA4 downregulation was also observed in a dose-dependent manner in EBV⁻ BL cell BJAB transduced with LMP1 lentivirus, at mRNA and protein levels (Figure 2I-J). Of note, upregulation of EphA4 protein and mRNA expression was seen in LCLs using an shLMP1 knockdown approach (Figure 2K-L). Taken together, all of the evidence indicated that LMP1 was the viral protein responsible for EphA4 downregulation.

Structurally, LMP1 resembles a CD40 receptor but its activity is ligand-independent. LMP1 delivers signaling through its long CTARs.^30,31  To determine which CTAR domains were involved in EphA4 downregulation, lentiviruses containing LMP1 full-length (WT), CTAR1-deleted mutant (ΔCTAR1), CTAR2-deleted mutant (ΔCTAR2), and both CTAR1- and CTAR2-deleted mutant (ΔCTAR1/2) were transduced into BJAB cells. It seems that both CTAR1 and 2 of LMP1 are involved in this repression (Figure 3A). Usually, these CTARs associate with tumor necrosis factor receptor-associated factors and activate downstream PI3K/Akt, NF-κB, and MAPK, including the JNK, ERK, and p38 pathways.³²  To dissect which signaling pathway was necessary for LMP1-mediated EphA4 downregulation, LCLs were treated with MEK inhibitor PD98059, JNK inhibitor SP600125, PI3K inhibitor LY294002, or NF-κB inhibitor Bay11-7082. Expression of EphA4 was increased in LCLs treated with PD98059 but not other inhibitors (Figure 3B). Furthermore, LMP1-triggered EphA4 downregulation through the ERK pathway was confirmed by silencing both ERK1/2 by lentiviral transduction of LCLs (Figure 3C).

LMP1 downregulated the expression of EphA4 at the transcriptional and translational levels (Figure 2) through the ERK pathway (Figure 3B-C). ERK has been reported to phosphorylate Sp1 at T453 and T739 to regulate the promoter activity of targeted genes³³  and we speculated that LMP1 may inhibit EphA4 promoter activity through Sp1. To test this hypothesis, EphA4 protein was detected in Sp1-knockdown LCLs and its amount was augmented (Figure 4A). Meanwhile, EphA4 protein amounts were restored in LMP1-overexpressing BJAB cells, when cells were simultaneously knocked down of Sp1 (Figure 4B). These data suggested that Sp1 was involved in LMP1-mediated EphA4 downregulation. The effect of Sp1 on EphA4 promoter activity was investigated further. The EPHA4 promoter (−1000 to +42) sequence was analyzed and 4 Sp1-binding sites were predicted. Clearly, EphA4 promoter activity was upregulated in Sp1-silenced LCLs and in Sp1-knockdown, LMP1-expressing TW01 cells (Figure 4C-D). Also, addition of the Sp1 inhibitor mithramycin, which interferes with Sp1 binding to the guanine-cytosine sequence, to the LMP1 transfectants increased the EphA4 promoter activity (Figure 4E). Furthermore, to determine whether LMP1 promotes Sp1 binding to the EPHA4 promoter, a chromatin immunoprecipitation assay was performed in BJAB cells expressing LMP1. We showed that Sp1 bound to the EPHA4 promoter at the region of −51 to −60 nt in BJAB cells, following LMP1 expression (Figure 4F).

Like other EphA family members, EphA4 contains an extracellular domain, transmembrane (TM) and juxtamembrane (JM) domains, and an intracellular kinase domain.¹⁹  To understand the biological function of EphA4 in B cells, EphA4 full-length (WT) and mutants (including EphA4 with a kinase-dead mutant with V653M [KD] and EphA4 with mutations of the tyrosine autophosphorylation site Y569F and Y602F of the JM region (2M), as illustrated in Figure 5A), were delivered to LCLs by lentivirus infection. The transduced LCLs expressed WT or mutated EphA4 at the physiologic level, compared with primary B lymphocytes (Figure 5B-C right panel). Cell proliferation was inhibited significantly in the LCLs expressing EphA4 WT and KD but not in the LCLs expressing EphA4 2M (Figure 5B-C). These data showed that EphA4 repressed LCL proliferation through its JM region, suggesting that the kinase domain is not important for EphA4 repression of LCL proliferation.

EBV⁺ PTLD exhibits similar biological features to LCLs. Thus, to address whether EphA4 was downregulated in PTLD biopsies, we examined EphA4 expression levels in 14 EBER⁺ PTLD biopsies and 5 EBER⁻ tonsil biopsies by IHC assay. The results of the IHC assay showed that all cases of tonsil biopsies were positive for EphA4 staining. In addition, EphA4 was predominantly expressed in the cytoplasm and cell membranes in tonsil biopsies. On the other hand, 13 of 14 cases of PTLD (92.9%) were negative for EphA4 expression (Figure 6; Table 1). In addition, LMP1 was responsible for downregulation of EphA4 in our studies (Figures 2-4). Therefore, the expression levels of LMP1 were measured in EBER⁺ PTLD biopsies by IHC assay and 12 of 14 cases of PTLD (85.7%) were positive for LMP1 staining (Figure 6; Table 1).

According to 2016 World Health Organization (WHO) classification, EBV⁺ DLBCL of the elderly, which is equivalent to EBV⁺ DLBCL, not otherwise specified, displays a similar EBV viral protein expression as LCL.^2,34  To confirm that EphA4 expression is correlated with EBV infection and LMP1 expression, 16 EBER⁺ DLBCL and 11 EBER⁻ DLBCL biopsies from patients without HIV infection or any iatrogenic immunosuppression were examined. The IHC assay indicated that all cases of EBV⁻ DLBCL biopsies were positive for EphA4 staining (Figure 7A; Tables 1 and 2). In contrast, 10 of 16 cases of EBV⁺ DLBCL (62.5%) were negative for EphA4 expression, 3 showed very weak, positive signals, and 3 were stained positively (Figure 7A; Tables 1 and 2). That the presence of EBV was associated with reduced EphA4 expression was statistically significant (Table 2, P = .0011, Fisher exact test). On the other hand, 12 of 15 cases of EBV⁺ DLBCL (80%) were positive for LMP1 staining (Figure 7A; Table 1). LMP1 expression was also inversely correlated with EphA4 expression (Table 2, P = .0375, Fisher exact test). Next, we asked whether there was correlation between EphA4 expression and patients’ survival. So, we analyzed the survival data of patients with DLBCL from GEO data sets (accession number GSE4475). The profile of overall survival in patients was used to estimate whether patients with low EphA4 expression have worse prognosis. Based on the median expression value of EphA4, 123 patients were grouped into 2 clusters; 60 patients had high EphA4 and 63 patients had low EphA4. Patients with low EphA4 had worse overall survival (Figure 7B, P = .0414). Our results indicated that low EphA4 expression may potentially contribute to a poor prognosis for these DLBCL patients.

Physiologically, Eph receptors and their ligands are critical for many developmental processes. Pathologically, the Eph family is dysregulated in a variety of human diseases, especially cancers. During cancer formation, Eph can function in tumor suppression or tumor promotion, depending on the type of cancer, the interacting ligands and cross-talk with other RTKs. Among them, EphA2 is a good example. Limited expression of EphA2 is required for mammary gland development but its overexpression may enhance the malignancy of breast cancer.³⁵  In brain tumors, the presence of ephrin-A1 determines the direction of a reciprocal loop between EphA2 and Akt, which results in inhibition or promotion of the tumor grade.³⁶  In addition to neuronal development, EphA4 has been found to be involved in tumor progression.^{18,19,21,22,37,38}  For example, EphA4 is upregulated in colon cancer with liver metastasis.²²  In invasive cervical carcinoma, EphA4 expression is decreased due to chromosomal deletions, with loss of heterozygosity.²¹  EphA4 mRNA is downregulated or lost in metastatic melanoma.³⁷  The regulatory mechanisms of EphA4 mRNA expression have been explored in terms of transcription. In the cervical cancer cell line HeLa and U373 glioma cells, EphA4 is reduced due to mRNA instability through regulation of its 3′ untranslated regions by HuR.³⁹  In hepatocellular carcinoma, microRNA-10a targets EphA4 mRNA and overexpression of EphA4 inhibits cell migration and invasion.⁴⁰  Moreover, a Pax3/FKHR oncogenic fusion protein has been shown to bind directly to the promoter region of EphA4, thus increasing EphA4 transcription in SaOS-2 cells.⁴¹  Furthermore, DNA methylation of EphA4 is observed in acute lymphoblastic leukemia.⁴²  Our results showed that EBV LMP1 inhibits EphA4 promoter activity through Sp1 (Figure 4).

Sp1, which is expressed ubiquitously, can serve as a transactivator or repressor to regulate the promoter activities of target genes via phosphorylation.⁴³  For example, FGF-2 stimulation enhances Sp1 binding to repress the promoter of platelet-derived growth factor receptor-α in smooth muscle cells through ERK-mediated phosphorylation of Sp1 at T453 and T739.⁴⁴  HER2 also has been reported to suppress the RECK promoter by ERK-mediated Sp1phosphorylation.⁴⁵  In this study, LMP1 activated the ERK pathway to inhibit the EphA4 promoter through Sp1 binding (Figure 4). We provide this new insight into how a viral oncoprotein can regulate Eph expression.

The function of EphA4 associated with cancer progression is controversial. As a tumor suppressor, EphA4 downregulates ERK phosphorylation to inhibit migration and invasion in non-small cell lung cancer cells.³⁸  However, EphA4 can act as tumor promoter; for example, EphA4 interacts directly with FGFR1 through its cytoplasmic domains, phosphorylates the docking protein, FRS2α, and enhances cell migration and proliferation.^46,47  In our case, EphA4 prevents LCL from proliferation (Figure 5), suggesting that it probably plays an inhibitory role in EBV⁺ B-lymphoproliferative disorders. This is confirmed by the IHC results of PTLD and DLBCL (Figures 6A-7A). In this study, expression of EphA4 was undetectable in the cells of EBER⁺ PTLD biopsies (Figure 6; Table 1). Because the incidence of EBV in PTLD cases is over 90%,²  these data imply that EphA4 is involved in the pathogenesis of EBV⁺ PTLD.

In addition to PTLD, our results indicated that cells in EBV⁺ DLBCL biopsies exhibited lower expression levels of EphA4 than those in EBV⁻ biopsies (Figure 7A; Table 2), verifying the inverse correlation of EphA4 and EBV infection in DLBCL patients. It has been reported that patients with EBV⁺ DLBCL have worse overall survival and progression-free survival than their EBV⁻ counterparts,^34,48  yet the underlying mechanism has not been fully understood. Here, we provided observation to show that low EphA4 correlated with poor survival outcome for DLBCL (Figure 7B). Also, some EphA4 ligands, such as ephrin-A1, A2, A3, A4, and B1 were downregulated post-EBV infection (data not shown). The results of overexpression of EphA4 in LCL indicated that EphA4 prevention of B-cell proliferation may be ligand-independent.

Taken together, LMP1-mediated EphA4 repression accelerates B-cell proliferation post-EBV infection. Our clinical results suggest that downregulation of EphA4 in patients with PTLD and DLBCL may provide new insights into pathogenesis and a poor prognostic marker for EBV-associated B-cell malignancies. EphA4 may be considered as a potential therapy target for PTLD and DLBCL.

The authors thank Tim J. Harrison of University College London Medical School (London, United Kingdom) for reviewing the manuscript critically. The authors thank Taipei Blood Center of Taiwan Blood Service Foundation for providing whole blood.

This work was supported by the Ministry of Science and Technology (grant: 103-2320-B-002-038-MY3), the National Health Research Institute (grant: NHRI-EX105-10306BI), Excellent Translational Medicine Research Projects of National Taiwan University College of Medicine and National Taiwan University Hospital (grant: 105R39012) (C.-H.T.), and by the Ministry of Science and Technology (grant: 103-2320-B-182-028-MY3) (S.-J.L.).

Contribution: Y.-C.H. designed experiments, performed experiments, analyzed the data, and cowrote the manuscript; S.-J.L. and C.-H.T. designed experiments and cowrote the manuscript; K.-M.L. and J.L. performed experiments; Y.-C.C. performed experiments and analyzed the data; C.-W.L., T.-L.S., and M.-R.C. provided materials; and S.-C.Y., C.-L.C., and C.-K.C. provided materials and analyzed the data.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Ching-Hwa Tsai, Graduate Institute of Microbiology, College of Medicine, National Taiwan University, Room 719, No. 1, 1st Section, Jen-Ai Rd, Taipei 10051, Taiwan; e-mail: chtsai@ntu.edu.tw.