Defective Epstein-Barr Virus Genomes and Atypical Viral Gene Expression in B-Cell Lines Derived from Multiple Myeloma Patients

ABSTRACT

Epstein-Barr virus (EBV) is a human gammaherpesvirus that is causally associated with various lymphomas and carcinomas. Although EBV is not typically associated with multiple myeloma (MM), it can be found in some B-cell lines derived from MM patients. Here, we analyzed two EBV-positive MM-patient-derived cell lines, IM9 and ARH77, and found defective viral genomes and atypical viral gene expression patterns. We performed transcriptome sequencing to characterize the viral and cellular properties of the two EBV-positive cell lines, compared to the canonical MM cell line 8226. Principal-component analyses indicated that IM9 and ARH77 clustered together and distinct from 8226. Immunological Genome Project analysis designated these cells as stem cell and bone marrow derived. IM9 and ARH77 displayed atypical viral gene expression, including leaky lytic cycle gene expression with an absence of lytic DNA amplification. Genome sequencing revealed that the EBV genomes in ARH77 contain large deletions, while IM9 has copy number losses in multiple EBV loci. Both IM9 and ARH77 showed EBV genome heterogeneity, suggesting cells harboring multiple and variant viral genomes. We identified atypical high-level expression of lytic genes BLRF1 and BLRF2. We demonstrated that short hairpin RNA (shRNA) depletion of BLRF2 altered viral and host gene expression, including a reduction in lytic gene activation and DNA amplification. These findings demonstrate that aberrant viral genomes and lytic gene expression persist in rare B cells derived from MM tumors, and they suggest that EBV may contribute to the etiology of MM.

IMPORTANCE EBV is an oncogenic herpesvirus, but its mechanisms of oncogenesis are not fully understood. A role for EBV in MM has not yet been established. We analyzed EBV-positive B-cell lines derived from MM patients and found that the cells harbored defective viral genomes with aberrant viral gene expression patterns and cell gene signatures for bone marrow-derived lymphoid stem cells. These findings suggest that aberrant EBV latent infection may contribute to the etiology of MM.

INTRODUCTION

Epstein-Barr virus (EBV) is a human tumor virus responsible for a diverse spectrum of B-cell lymphomas and lymphoproliferative disease, NK/T-cell lymphomas, undifferentiated nasopharyngeal carcinomas, and 10% of gastric carcinomas (1, 2). EBV is also considered a suspect in the etiology of various B-cell inflammatory and autoimmune diseases, particularly multiple sclerosis (3). EBV infection of resting B cells induces a multistep developmental program resembling the germinal center reaction and memory B-cell differentiation program without normal antigenic and T-cell help (4, 5). Various EBV-associated B-cell malignancies involve additional host cell mutations, such as c-myc translocation in Burkitt’s lymphoma (BL), and loss of host immune function, as in posttransplant lymphoproliferative disease. Epithelial malignancies are thought to arise due to the ability of EBV to infect epithelial cells that normally support lytic cycle replication during persistent and latent infections (6, 7). Lytic gene expression is typically repressed in normal B-cell latent infection, but lytic cycle genes have been detected at variable frequencies in EBV-associated tumors (8), and EBV lytic genes have been shown to have oncogenic activities that could contribute to EBV carcinogenesis (9, 10).

Multiple myeloma (MM) is a B-cell malignancy arising from postgerminal plasmablasts accumulating in the bone marrow (reviewed in references 11 and 12). MM evolves through premalignant states of monoclonal gammopathy of uncertain significance (MGUS) to smoldering myeloma and myeloma (13, 14). A role for EBV in MM has been suspected but, to date, there is no compelling evidence to implicate EBV in the majority of MM cases (15–17). MM is a heterogeneous disease with diverse genetic aberrations, suggesting that there are genetic subtypes of MM (18). Several genetic and cell surface markers are used to diagnose MM, including the expression of cell surface CD138 (Syndecan 1) as a marker for plasma B-cell maturation (19, 20). EBV coinfection can be detected in a small percentage of CD138⁺ MM cells in some patients, while EBV may also infect CD138⁻ B cells in bone marrow in MM patients (21). EBV latent infection has been correlated with increased risk of the premalignant MGUS and MM (16). However, EBV genomes and gene products are rarely found in CD138⁺ MM-derived tumor cells and lesions, although exceptional cases have been reported (22).

Several MM-patient-derived cell lines have been characterized; a few of these were found to be EBV positive and subsequently were deemed to be EBV-immortalized lymphoblastoid cell lines (LCLs), rather than bona fide MM cell lines (23). Here, we examine two of these EBV-positive MM-patient-derived cell lines and find that they are not typical LCLs and more likely are B-cell lymphoma lines carrying defective EBV genomes. Defective EBV genomes have been identified in various EBV tumors and tumor-derived cell lines but not typically in EBV-positive cells from healthy donors or in vitro-transformed LCLs (8). Defective, rearranged EBV genomes have been identified in Hodgkin’s lymphomas (24), oral hairy leukoplakia (25), BL cell lines (26, 27), and, more recently, chronic active EBV (CAEBV) and extranodal NK/T (ENKT) cell lymphoma (8). Defective viral genomes have the capacity to express aberrant latency programs and lytic cycle genes, resulting in chronic reactivation (26, 28). Here, we show that EBV genomes in the MM-derived cell lines ARH77 and IM9 have heterogeneity, intergenic deletions, and aberrant gene expression, correlating with robust in vitro and in vivo growth properties, raising the possibility that these viral genetic variations contribute to EBV carcinogenesis.

RESULTS

Transcription profiling of MM-patient-derived EBV-positive cell lines.

ARH77 and IM9 were derived from MM patients but were subsequently recharacterized as EBV-positive LCLs (23). The RPMI 8226 cell line (referred to here as 8226) is considered to be a prototypical MM cell line. Consistent with this designation, flow cytometry analysis demonstrated that 8226 cells express high levels of cell surface CD138, while EBV-positive cell lines ARH77 and IM9, along with the EBV-positive BL line Mutu I and LCLs, express 10- to 100-fold lower levels of CD138 (Fig. 1A). To better assess the cellular origin of the ARH77, IM9, and 8226 cell lines, we performed high-throughput RNA sequencing (RNA-seq) analysis. Transcriptome profiling revealed that ARH77 and IM9 are more closely related to each other than they are to 8226 (Fig. 1B and C). Immunological Genome Project (ImmGen) analysis of RNA-seq data categorized ARH77 and IM9 as stem cell like and bone marrow derived (SC.LT34F.BM), while 8226 cells were categorized as myeloid derived (MF.ThioS) (Fig. 1D). Hierarchical clustering further revealed that ARH77 and IM9 share major differences from 8226 in expression patterns for B-cell transcription factors, immune function genes, and histones (Fig. 2). For example, NF-κB and BATF were upregulated in ARH77 and IM9, while RUNX1, E2F3, EBP, and PROX1 were exclusively upregulated in 8226. ARH77 and IM9 were upregulated in expression of CD22, HLA-DQ, and HLA-DR, while 8226 was upregulated for S100A4, CD9, TSPAN13, SPP1, and CXCL12, relative to ARH77 and IM9. Ingenuity Pathway Analysis (IPA) of ARH77 and IM9, compared to 8226, showed upregulation of functions in leukopoiesis and cell viability and downregulation of apoptosis and polyarthritis, further highlighting the clustering of gene expression and function in these EBV-positive B-lymphoma cell lines, compared to EBV-negative 8226 MM cells.

FIG 1.

Characterization of EBV-positive cell lines derived from MM patients. (A) FACS analysis of CD138 cell surface expression (fluorescein isothiocyanate [FITC]) on cell lines 8226 (blue), ARH7 (green), IM9 (orange), LCL (cyan), and Mutu I (magenta). (B) PCA of the three cell lines based on 3,087 genes different at an FDR of <1%, with at least a 10-fold difference between any two cell lines. This indicates that ARH77 and IM9 are more similar to each other than to 8226 (first principal component shows that 68% of the variation comes from this difference). (C) Overlap in top 10% ImmGen cell types most correlated with each cell line. (D) ImmGen cell types correlated with the three cell lines (top 10 in at least one cell line). Shades go from top rank (orange) to lower rank (purple, capped at 50).

FIG 2.

Differential gene expression analysis of 8226, ARH7, and IM9 cell lines. (A) Heatmap and hierarchical clustering of top 100 differentially expressed genes across the three cell lines as detected by RNA-seq. (B) Top 10 affected cellular functions detected by IPA among genes differentially expressed in ARH77 and IM9 cells, compared to the EBV-negative MM cell line 8226. (C to E) Heatmap of differentially regulated genes according to categories for immune function (C), histones (D), and transcription factors (E).

Atypical EBV gene expression in MM-patient-derived EBV-positive cell lines.

RNA-seq also captured a number of reads mapping to the EBV genome in both ARH77 and IM9, with 8226 showing no EBV reads (Fig. 3A). The major viral RNA species in both ARH77 and IM9 were EBV-encoded small RNAs (EBERs) and BHRF1. Surprisingly, IM9 cells expressed unusually high levels of BLRF2 and BLRF1, transcripts not typically observed in latent infections. AHR77 cells also expressed BLRF2 at high levels but had similarly high levels of other lytic genes, suggesting that these cells had background lytic transcription in addition to latency gene products. We validated the expression of BLRF2 in AHR77 and IM9 cells by real-time-quantitative PCR (RT-qPCR) (Fig. 3B and C) and Western blotting (Fig. 3D). BLRF2 protein was expressed at the highest levels in ARH77 cells, while RNA levels were higher in IM9 cells. BLRF2 protein and RNA could also be detected in Mutu I cells. Interestingly, BLRF1 was similarly upregulated in the same cells, suggesting that these cDNAs are coregulated or cotranscribed. All EBV-positive cells expressed EBV nuclear antigen 1 (EBNA1), as expected. EBNA2 was expressed at extremely low levels in ARH77 and Mutu I, while latent membrane protein 1 (LMP1) was expressed robustly in both ARH77 and IM9 but not Mutu I. BHRF1 RNA was detected by RT-qPCR in ARH77 but protein was detected mostly in IM9, suggesting that RNA levels do not always correlate with protein expression. Early antigen-diffuse (EA-D) was expressed at high levels in ARH77, while Zta was expressed at high levels in IM9. RT-qPCR indicated that neither IM9 nor ARH77 expressed high levels of EBNA3C or EBNA3B, compared to LCLs or Mutu I. Fluorescence-activated cell sorting (FACS) analysis of Zta and EA-D protein revealed that IM9 cells expressed Zta in 20.4% of cells and EA-D in 0.28% of cells, while ARH77 expressed Zta in 3.8% of cells and EA-D in 3.8% of cells (Fig. 3E and F), consistent with Western blot analysis of bulk cells. Both ARH77 and IM9 utilized the Cp typically associated with type III latency, while Mutu I utilized the Qp associated with type I latency (Fig. 3G). Taken together, these findings suggest that ARH77 and IM9 have atypical latency types, and both have leaky and partial lytic gene expression.

FIG 3.

Aberrant EBV gene expression in ARH7 and IM9. (A) EBV genes sorted by maximum expression across two cell lines quantified as log₂-scaled 1 + fragments per kilobase of transcript per million reads mapped (FPKM) values. (B) RT-qPCR for EBV genes BLRF2 (primer pairs 1 and 2), BLRF1, BHRF1, BSRF1, BLLF1, BLLF2, BLLF3, EBNA1, LMP1, EBNA2, EBNA3B, EBNA3C, BHLF1, BNRF1, Zta (BZLF1), Rta (BRLF1), and EA-D (BMRF1), relative to the cellular β-glucuronidase (GUSB) gene, in cell lines 8226, ARH77, IM9, Mutu I, LCL, and Raji. (C) Schematic showing positions of primers for BLRF1, BLRF2_1, and BLRF2_2 (green bars). (D) Western blot analysis of BLRF2, EBNA1, EBNA2, LMP1, EA-D, Zta, BHRF1, GAPDH, and actin from cell lines. 8226, ARH77, IM9, Mutu I, LCL, and Raji. (E) Flow cytometry for EA-D and Zta expression in ARH77, IM9, Mutu I, and LCL. (F) Quantification of percentage of cells with EA-D and Zta from the FACS analysis shown in panel E. (G) RT-qPCR analysis of promoter utilization for Qp or Cp/Wp in Mutu I, LCL, ARH77, and IM9.

Since partial lytic cycle gene expression was observed for both ARH77 and IM9, we assayed viral DNA copy numbers by RT-qPCR (Fig. 4). We found that untreated IM9 and ARH77 had EBV genome copy numbers comparable to those of Raji but lower than those of Mutu I and greater than those of LCLs (Fig. 4A). We next assayed the ability of sodium butyrate (NaB)/phorbol-12-tetradecanoate-13-acetate (TPA) to induce lytic DNA amplification. While Mutu I cells could undergo >10-fold increases in DNA copy numbers, ARH77 and IM9 levels increased <2-fold (Fig. 4B). This same trend was observed for extracellular viral DNA (Fig. 4C). RT-qPCR analysis of viral RNA revealed that NaB/TPA treatment could moderately induce some viral latent and lytic genes, such as LMP1 in ARH77 and EBNA1, LMP1, EBNA3A, Zta, BLRF1, and BLRF2 in IM9 (Fig. 4D and E). In contrast, Mutu I cells showed a broader distribution of viral lytic genes, including Zta, EA-D, BLRF1, and BLRF2, and significantly greater induction after treatment with NaB/TPA (Fig. 4F), while LCLs could be induced by the tetrahydro-β-carboline C60 but not NaB/TPA (Fig. 4G).

FIG 4.

EBV copy numbers and restricted lytic replication in ARH77 and IM9. (A) EBV copy numbers were determined by RT-qPCR comparing viral Ori-Lyt to cellular GAPDH in 8226, ARH77, IM9, Mutu I, LCL, and Raji. (B) EBV copy numbers were determined after treatment with NaB/TPA or C60 for 48 h, relative to untreated control, for ARH77, IM9, LCL, and Mutu I. (C) Extracellular EBV copy numbers were assayed after treatment with NaB/TPA for 96 h, relative to untreated control, for ARH77, IM9, Mutu I, and LCL. (D to G) RNA gene expression was assayed by RT-qPCR for EBV genes after treatment with NaB/TPA or C60 for 48 h (or untreated control), relative to cellular GUSB, for ARH77 (D), IM9 (E), Mutu I (F), and LCL (G). Student’s t test was used to calculate P values. * indicates P < 0.05.

EBV intragenic deletions and genome heterogeneity in MM-patient-derived EBV-positive cell lines.

To better understand the molecular basis for the unusual viral gene expression patterns, we sequenced the genomes of ARH77 and IM9 using Illumina next-generation sequencing and mapped the reads to EBV reference genomes (Fig. 5). We found that both ARH77 and IM9 had various internal deletions, copy number variations, and genome heterogeneity. ARH77 had an ∼10-kb deletion at the BamHI A rightward transcripts (BART) locus (EBV nucleotide positions ∼138 to 148 kb), which has been reported as a frequent event in CAEBV and NK/T cell lymphomas (8). ARH77 also had deletions at EBNA2 and BHLF1 loci. IM9 had loss of copy number at several EBV genomic loci, including the BDLF1 to LF3 loci (EBV nucleotide positions ∼120 to 143 kb), BARF1 to BLLF3 loci (EBV nucleotide positions ∼67 to 76 kb), and LMP1/2 loci. We validated by RT-qPCR the copy number change in the BSLF1 to BLLF3 region that is lost in IM9 (Fig. 5B and C). In addition, a number of sites were found to have heterogeneous sequences (44 heterogeneous sites were called in ARH77 and 46 sites in IM9, with a total of 3 common sites), suggesting multiple genomes in the population.

FIG 5.

DNA sequences of ARH77 and IM9. (A) Next-generation sequencing of ARH77 and IM9 aligned to the wild-type EBV genome (NC_007605.1). Relative read count coverage is shown in orange, polymorphisms are indicated in the inner circle, and an EBV standard gene map shown in the outer circle. (B) UCSC genome browser showing ARH77 and IM9 sequence reads at BSLF1 to BLRF2 gene loci, with the RT-qPCR validation primers for positions A, B, C, and D shown below. (C) DNA RT-qPCR for primers A, B, C, and D, relative to Cp, for genomes from ARH77, IM9, Mutu I, and LCL.

MM-patient-derived EBV-positive cell lines are resistant to serum starvation in vitro and highly aggressive in SCID mouse models.

To investigate the phenotypic differences in EBV-positive versus EBV-negative MM-derived cell lines, we assayed the responses of ARH77, IM9, and 8226 cells for cell cycle profiles under cell culture conditions with high (12%) or low (2%) serum concentrations (Fig. 6A). We found that ARH77 and IM9 were resistant to low serum levels, while 8226 cells underwent a G₁/S arrest (Fig. 6A). We next tested the growth and tumorigenic properties of these cell lines in SCID mouse xenograft models. Identical numbers of cells were injected intravenously and followed by mCherry-luciferase tagging and IVIS imaging (Fig. 6B to D). We found that IM9 grew aggressively and led to lethal proliferative disease within 3 weeks. Similarly, ARH77 resulted in rapid proliferative disease and mouse lethality within 6 weeks. In contrast, 8226 cells migrated primarily to bones of the hip and spine and did not cause disseminating B-cell proliferative disease. Another MM cell line (MM1.S) had very similar growth kinetics, compared to 8226, in SCID mice (Fig. 6C). These findings indicate that IM9 and ARH77 are more aggressive than two established MM cell lines in mouse models of B-cell proliferative disease.

FIG 6.

Serum starvation and mouse tumor growth properties of 8226, ARH77, and IM9. (A) The 8226, ARH77, and IM9 cell lines were grown in RPMI 1640 medium containing 12% or 2% FBS and were analyzed by FACS for cell cycle analysis by propidium iodide staining. (B) Kaplan-Meir survival curves of mice (n = 10) injected intravenously with 8226 (black), IM9 (blue), ARH77 (green), Mutu I (purple), B95.8-LCL (orange), and Mutu-LCL (red) cells. Survival was increased for animals engrafted with 8226, compared with EBV-positive B-cell lines (IM9, ARH77, Mutu I, B95.8-LCL, and Mutu-LCL). ***, P < 0.0001, log-rank (Mantel-Cox) test. The survival of animals engrafted with IM9 was significantly lower than that of animals engrafted with other EBV-positive B-cell lines (ARH77, Mutu I, B95.8-LCL, and Mutu-LCL). ***, P < 0.0001, log-rank (Mantel-Cox) test. (C) Survival curves for EBV-negative MM lines MM1.S (n = 5) and 8226 (n = 10), There was no significant difference in the survival curves for mice engrafted with MM1 or 8226 (log-rank [Mantel-Cox] test, P = 0.5372). (D) In vivo cell growth as measured by whole-body bioluminescence flux (photons/s). ***, P = 0.0003, one-way analysis of variance, followed by Pearson’s correlation.

BLRF2 transcript regulates expression of cellular genes involved in oxidative stress and control of EBV lytic reactivation.

We designed short hairpin RNA (shRNA) targeting BLRF2 (shBLRF2), which is typically expressed during the lytic cycle and has been implicated in several functions, including inhibition of cyclic GMP-AMP synthase (cGAS) (29, 30) and interaction with the tegument protein BNRF1 implicated in chromatin regulation (31–33). Knockdown of BLRF2 led to decreased levels of expression of a number of variable EBV latent and lytic genes in ARH77, IM9, and Mutu I cells (Fig. 7A). The shBLRF2 also reduced the transcriptional activation of several viral lytic genes, including immediate early genes Zta and Rta, early lytic gene EA-D, and late lytic gene BNRF1, after NaB/TPA induction in all three cell lines (Fig. 7B). shBLRF2 also inhibited the amplification of viral DNA copy numbers after NaB/TPA treatment (Fig. 7C). These findings suggest that BLRF2 positively regulates lytic cycle gene expression and DNA replication.

FIG 7.

BLRF2 depletion limits lytic gene expression in IM9 and ARH77. (A) Western blot analysis of IM9, ARH77, or 8226 after shControl (shCtrl), shBLRF2-3, or BLRF2-4 transduction. Cells were assayed for BLRF2, Zta, EA-D, EBNA1, LMP1, EBNA2, PARP1, Rad21, and actin expression at 7 days postransduction. (B) RNA expression for EBV genes after shBLRF2 transduction in untreated or NaB/TPA-treated (24 h) ARH77, IM9, and Mutu I cells (ΔΔC_T method). (C) DNA copy number for EBV genomes (Ori-Lyt) relative to cellular GAPDH in untreated or NaB/TPA-induced (48 h) cells after shCtrl or shBLRF2. Student’s t test was used to calculate P values. * indicates P < 0.05.

BLRF2 knockdown alters cellular gene expression.

To investigate a potential role of BLRF2 in controlling cellular gene expression, we performed RNA-seq for ARH77 and IM9 cells before and after BLRF2 knockdown (Fig. 8). Numerous cellular genes were affected by BLRF2 shRNA, but only a few were common to both cell types; among these were RGS1, CHAC1, and P4HA2, which were downregulated after shBLRF2 transduction (Fig. 8A). The pathways most affected were cell death and survival and transactivation of RNA, based on IPA (Fig. 8B). RGS1 has been implicated in promoting B-cell and other malignancies (34, 35), P4HA2 has been implicated in B-cell lymphoma progression (36), and CHAC1 has been implicated in glutathione metabolism and the endoplasmic reticulum stress response (37), suggesting that BLRF2 may contribute to the control of cellular genes important for EBV tumorigenesis. We validated several of these changes by RT-qPCR (Fig. 8C).

FIG 8.

Cellular gene responses to BLRF2 depletion in ARH77 and IM9. (A) Differentially regulated genes in response to BLRF2 shRNA depletion in ARH77 or IM9 cells. (B) IPA of gene function and categories affected by BLRF2 shRNA depletion in ARH77 and IM9 cells. (C) RT-qPCR validation of genes (RGS1, CHAC1, P4HA2, AICDA, CXCL10, and BAG1) deregulated by BLRF2 shRNA depletion in ARH77 or IM9 cells. Student’s t test was used to calculate P values. *, P < 0.05.

DISCUSSION

MM is a heterogeneous disease of mature antibody-producing B cells colonizing the bone marrow and disrupting osteocyte homeostasis. Although EBV is not found in the majority of MM lesions and pathogenic B cells, EBV has been suspected to contribute to the etiology of MM. Here, we examined two EBV-positive cell lines derived from MM patients to better understand the potential contribution of EBV to various B lymphomas. We found that EBV-positive ARH77 and IM9 cells shared common features distinct from those of the canonical EBV-negative MM cell line 8226. ImmGen analysis of RNA-seq data indicated that ARH77 and IM9 fell into the categories of SC.LT34F.BM and germinal center B cells (38, 39). The stem-cell-like feature derived from bone marrow is intriguingly MM-like. In contrast, 8226, which expresses high levels of cell surface CD138, fell into the category of macrophages derived in the peritoneal cavity. Transcription factor and immune gene profiling showed significant differences among all three cell lines, although EBV-positive cells shared increased expression of histone genes and NF-KB, BATF, and CD22. Elevated CD22 levels have been found on MM precursor cells (40). These findings raise the possibility that EBV-positive B cells may contribute to some stages or cellular subtypes found in the complex heterogeneity of MM disease.

IM9 and ARH77 cells were found to have aberrant EBV gene expression. IM9 and ARH77 also showed some discordance between RT-qPCR analysis of RNA and Western blot analysis of protein expression. ARH77 expressed very little or none of EBNA2 or EBNA3s by RNA or protein analysis. IM9 did express EBNA2 protein, but RNA levels for EBNA2 and EBNA3s were significantly lower than those of LCLs. Both ARH77 and IM9 expressed LMP1, although both had lower RNA levels than LCLs. ARH77 and IM9 utilized the Wp/Cp promoter and not the Qp for EBNA1 expression. Both ARH77 and IM9 had leaky lytic gene expression, with unusually high levels of BLRF2 and BLRF1 RNA, and elevated levels of Zta. ARH77 had the highest protein levels of BLRF2 and EA-D. Neither IM9 nor ARH77 could be induced by NaB/TPA to amplify viral genomes to levels comparable to those of Mutu I. Taken together, these findings indicate that ARH77 and IM9 have a type III/II latency with aberrant lytic gene expression and abortive lytic replication.

IM9 and ARH77 cells were found to have defective viral genomes. Illumina sequencing and PCR validation revealed anomalies in the EBV genomes of IM9 and ARH77. In addition to intragenic deletions in the BART locus, we found significant copy number loss for EBNA2, EBNA3A, EBNA3B, and EBNA3C in ARH77. Interestingly, the BART deletion in ARH77 was located at EBV nucleotide positions 138319 to 148525, not exactly overlapping the B95.8 strain deletion (EBV nucleotide positions 139724 to 151554). We found numerous loci with copy number loss in IM9, including the BHLF1 locus, BARF1 to BLLF3 loci (EBV nucleotide positions ∼67 to 76 kb), BDLF1 to LF3 loci (EBV nucleotide positions ∼120 to 143 kb), and LMP1 and LMP2 loci. Some of these deletions may account for discrepancies in measurements of RNA and protein expression levels. For example, the deletion in the EBNA2 region in ARH77 is located at EBV nucleotide positions 36551 to 37074, while our RT-qPCR primers for EBNA2 amplify EBV nucleotide positions 36095 to 36181. This deletion may account for the loss of EBNA2 protein but not its RNA in ARH77. Similarly, a deletion in ARH77 at EBV nucleotide positions 39134 to 39498 did not affect RT-qPCR primers for BHLF1, which amplify EBV nucleotide positions 38180 to 38250. Both cell types had sequence heterogeneity, suggesting that more than one viral genotype is present in the cell population. Intragenic loss of viral DNA is found in various viral tumors, including loss of EBNA2 in BL cells (41) and loss of BART genes in ENKT and CAEBV cells (8). Defective viral genomes are frequently observed in tumor cells with integrated EBV and are known to be oncogenic drivers in human papillomavirus- or Merkel cell polyomavirus-associated cancers (42, 43).

IM9 and ARH77 cells were found to be resistant to serum starvation and highly aggressive in mouse models of B-cell lymphoproliferation. Although EBV is retained in these tumor cell lines and is likely to be under positive selection, it is not yet known how EBV may contribute functionally to these robust tumor cells. Both IM9 and ARH77 expressed LMP1, which likely contributes to their oncogenic phenotype. The unusually high-level expression of BLRF2 and BLRF1 is not yet known to contribute to viral oncogenicity. We tested the effect of BLRF2 by shRNA depletion, and we found little phenotypic effect under most growth conditions, including low serum levels. BLRF2 depletion led to codepletion of the BLRF1 transcript, suggesting that these are coregulated. BLRF2 depletion altered host cell gene expression, including expression of several genes common to both IM9 and ARH77 that have functions related to carcinogenesis. BLRF2 was previously shown to be a component of the EBV virion and to interact with the BNRF1 tegument protein (44, 45). BLRF2 was found to interact with several cellular proteins, including serine/threonine-protein kinase 2 and DVL2, important for Notch signaling (44). BLRF2 is also the orthologue of the Kaposi sarcoma-associated herpesvirus inhibitor of cGAS (KicGAS) and was found to bind to cGAS in vitro (30). Although the precise role of BLRF2 upregulation in IM9 and ARH77 is not yet known, our data suggest that BLRF2 can contribute to changes in cellular gene expression important for B-cell lymphomagenesis or viability. BLRF2 has been shown to have partial nuclear localization (44), and its association with BNRF1, which can bind to a complex of DAXX with histone H3.3-H4, suggests that it may have a broad function in chromatin regulation. BLRF2 depletion also inhibited viral lytic gene activation in ARH77, IM9, and Mutu I cells and downregulated lytic DNA amplification in Mutu I cells. Taken together, our findings suggest that BLRF2 is upregulated in IM9 and ARH77 and its depletion affects cellular and viral gene expression, potentially through chromatin regulation.

Our characterization of two EBV-positive B-lymphoma cell lines suggests that defective viral genomes may be more prevalent in human cancers than previously appreciated. This is consistent with recent findings from Okuno et al. showing a high frequency of intergenic deletions in CAEBV and ENKT cell tumors (8). Defective genomes may also account for the aberrant viral gene expression, either through loss of coding DNA or rearrangement of regulatory elements controlling genes, such as BLRF1 and BLRF2. The potential of lytic cycle genes in EBV oncogenesis has been demonstrated in various contexts, including a requirement for the lytic activator BZLF1 (Zta) to promote lymphomagenesis in humanized mouse models (46). These finding underscore the diversity of strategies that EBV may utilize to increase the fitness and tumorigenicity of infected cells.

MATERIALS AND METHODS

Cell lines.

The EBV-negative MM cell line 8226 and the EBV-positive B-cell lines ARH77 and IM9 were obtained from Yulia Nefedova, Wistar Institute. The EBV-positive BL cell line Mutu I was obtained from Jeffrey Sample, Penn State University Hershey Medical School. LCLs were generated from human primary B cells transformed with Mutu I or B95.8 virus. The EBV-positive Raji cell line was purchased from ATCC. The 8226, ARH77, IM9, Mutu I, LCL, and Raji cells were maintained in RPMI 1640 medium containing 12% fetal bovine serum (FBS) and antibiotics (penicillin and streptomycin).

Antibodies used in Western blotting.

Rabbit polyclonal anti-BLRF2 was kindly provided by Ayman El-Guindy, Yale School of Medicine. Anti-EBNA1 and anti-Zta were generated in-house, and anti-poly(ADP-ribose) polymerase 1 (PARP-1) (210–302-R100; Alexis), anti-Rad21 (ab992; Abcam), anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (2118; Cell Signaling), rat monoclonal anti-EBNA2 (MABE8; Millipore), mouse monoclonal anti-LMP1 (M0897; Dako), anti-EA-D (ab30541; Abcam), anti-BHRF1 (MAB8188; Millipore), and peroxidase-conjugated anti-actin antibody (A3854; Sigma) were purchased.

RNA extraction and RT-qPCR.

RNA was isolated from 2 × 10⁶ cells using the RNeasy kit (Qiagen) and then was treated with DNase I by using the DNase treatment and removal kit (Ambion). RT-qPCR was performed with the SYBR green probe in an ABI Prism 7900 system, using the ΔC_T method for relative quantitation. Primer sequences are listed in Table 1.

TABLE 1.

RT-qPCR primers

Sequence	Primer name
CGCCCTGCCTATCTGTATTC	GUSB_Fwd
TCCCCACAGGGAGTGTGTAG	GUSB_Rev
CCAACCGCGAGAAGATGA	Actin_Fwd
CCAGAGGCGTACAGGGATAG	Actin_Rev
TCCAGAATTGACGGAAGAGGTT	LMP1_Fwd
GCCACCGTCTGTCATCGAA	LMP1_Rev
GGTCGTGGACGTGGAGAAAA	EBNA1_Fwd
GGTGGAGACCCGGATGATG	EBNA1_Rev
CCAGCGCCAATCTGTCTACA	EBNA2_Fwd
TGATGGCGGCAGATAAAGC	EBNA2_Rev
AGCCATTCTCCGCAGGTTT	EBNA3A_Fwd
CTGTATGCCTGGTAACCCATAGG	EBNA3A_Rev
CGATCCCTTGGATGTCCATACT	EBNA3B_Fwd
CAACGCCTCTGCTTAACAATTG	EBNA3B_Rev
GCCGGGCTGTCAAGCA	EBNA3C_Fwd
CCCACTATCGAGTATCAGGTTTGAT	EBNA3C_Rev
ATTGCACCTTGCCGCCACCTTTG	Zta_Fwd
CGGCATTTTCTGGAAGCCACCCGA	Zta_Rev
GAGCGATGAGAGACCCATATTC	Rta_Fwd
GAACATACCTTCCCGGCTATC	Rta_Rev
TTGGGCAGGTGCTGTTGAT	EA-D_Fwd
TGCCCACTTCTGCAACGA	EA-D_Rev
AGGCTCTGAAGCAACAGGTC	BLRF2_1_Fwd
TGGAGCAGAAGGAACAGAGG	BLRF2_1_Rev
TGTCAGCTCCACGCAAAGTCAGAT	BLRF2_2_Fwd
AGGACCTGTTGCTTCAGAGCCTTA	BLRF2_2_Rev
CAGGACCAGTTCTACTCCTACA	BLRF1_Fwd
ACTAAGACAAGCGTCAGAAGTG	BLRF1_Rev
GATCTCGGCCGTTGTCTCAT	BSRF1_Fwd
AGGCGGAGTTGATGAAAGGG	BSRF1_Rev
CATCTACAGATTCCAGGCTTACTTG	BLLF1_Fwd
AGCTTCCAATTAACGTCACCA	BLLF1_Rev
CACCTGTTGAGGTGACCGTG	BLLF2_Fwd
TCCACATCCACAAAACCGGG	BLLF2_rev
TTGACCCGAGGATCCAAACC	BLLF3_Fwd
CCATGGGTGATAGTGGGCTG	BLLF3_Rev
CGGGTTGGTGGCACTGTT	BHLF1_Fwd
GCATGGCGAAGTAGACAGGTTA	BHLF1_Rev
GCGGCCTTCACGAATGC	BNRF1_Fwd
GGAACGTGTTGTCCCTAACCTC	BNRF1_Rev
TGCCTGAACCTGTGGTTGG	Wp/Cp_Fwd
GTGCGCTACCGGATGGC	Qp_Fwd
CATGATTCACACTTAAAGGAGACGG	Wp/Cp/Qp_Rev
CCAGCATCACATTAAACAGGAG	BAG1_FWD
AGAATGGCAAGAAAGGTGGA	BAG1_REV
AACTTCTTGCCAACCAAACTG	RGS1_FWD
GCTTTACAGGGCAAAAGATCA	RGS1_REV
GCATGGTCACCTTCAAGCTACT	AICDA_FWD
CAACCCATTATTAGGGTGTCTTG	AICDA_REV
AAGCAGTTAGCAAGGAAAGGTC	CXCL10_FWD
GACATATACTCCATGTAGGGAAGTGA	CXCL10_REV
CCTGAAGTACCTGAATGTGCGAGA	CHAC1_FWD
GCAGCAAGTATTCAAGGTTGTGGC	CHAC1_REV
CTTCCTGCCTCAGGTATTGC	P4HA2_FWD
CCCAGAGTTTCATGGTCACA	P4HA2_REV

RNA-seq.

Total RNA was isolated from 1.5 × 10⁶ cells using the Direct-zol microprep kit (Zymo Research) following the manufacturer’s protocol. RNA samples were submitted to the Wistar Institute genomics core facility for initial analysis of RNA quality, with each sample having an RNA integrity number (RIN) value greater than 8.5 (TapeStation; Agilent Technologies). Sequencing library preparation for ARH77, IM9, and 8226 samples was completed using the ScriptSeq RNA-seq library preparation kit. Sequencing was done with an Illumina NextSeq 500 system in high-output mode to generate ∼140 × 10⁶ reads (2 × 75 bp) across three multiplexed and pooled samples. Sequencing library preparation for the shBLRF2 experiment was then completed using the QuantSeq 3′-mRNA kit (Lexogen) to generate Illumina-compatible sequencing libraries according to the manufacturer’s instructions. Sequencing was done with an Illumina NextSeq 500 system in high-output mode to generate ∼4 × 10⁸ reads (1 × 75 bp) across 12 multiplexed and pooled samples.

RNA-seq analysis.

RNA-seq data were aligned using the Bowtie 2 (47) algorithm against the hg19 human genome version, followed by the use of RSEM (v1.2.12) software (48) to estimate read counts and reads per kilobase of transcript per million reads mapped (RPKM) values, using UCSC human transcriptome information and EBV genome version NC_007605.1 for the ARH77, IM9, and 8226 experiment and Ensemble transcriptome version GRCh37.p13 for the shBLRF2 experiment. DESeq2 (49) was used on raw counts to estimate the significance of expression differences between any two experimental groups and to generate normalized counts that were Z-score scaled for clustering and principal-component analysis (PCA). EdgeR was used to estimate significance between individual cell types (50). Overall gene expression changes were considered significant with a false discovery rate (FDR) of <5%. Gene set enrichment analysis was done using Qiagen IPA software using the “diseases and functions” option, and selected functions with a significant predicted activation Z-score of at least 2 were reported. Expression values of genes most different among the three cell lines (FDR of <1%, at least 10-fold difference) for each sample were correlated with data from the ImmGen database using Spearman correlation, and the top 10 correlated cell types were reported.

Whole-genome sequencing.

Genomic DNA was purified with the Wizard genomic DNA purification kit (A1120; Promega). Sequencing library preparation was then completed using the Illumina DNA preparation kit according to the manufacturer’s instructions. Sequencing was done with an Illumina NextSeq 500 system in high-output mode to generate a total of ∼4 × 10⁸ paired-end 75-bp reads for pooled ARH77 and IM9 samples.

Whole-genome sequence analysis.

Whole-genome sequencing raw reads were aligned against EBV genome version NC_007605.1 using the BWA algorithm (51). Analysis of heterogeneous sites was performed following the procedure described by Čejková et al. (52). Specifically, deduplicated uniquely mapped reads with alignment quality scores of >Q30 were considered and trimmed 4 bp from the 3′ terminus. Nucleotide positions supported by at least 6 reads with at least 3 reads from both strands that showed a frequency of ≥20% of the less-frequent allele were further examined. EBV heterogeneous sites located within homopolymeric tracts (defined as a stretch of 6 or more identical nucleotides) or within a 2-nucleotide distance of those tracts were omitted from further analysis. Heterogeneous positions separated from each other by less than 7 bp were also omitted. After the filtering, a total of 44 heterogeneous sites were called in ARH77 and 46 sites in IM9. There were 3 overlapping heterogeneous sites between the two cell lines.

Lentiviral transduction.

pLKO.1 vector-based shRNA constructs for BLRF2 were generated with the target sequences 5′- TCGCCTGGAGTCTGAGAATAA-3′ (shBLRF2-3) and 5′- AGCAAAGAAAGTGCAAATTTC-3′ (shBLRF2-4). shControl was generated in pLKO.1 vector with the target sequence 5′- TTATCGCGCATATCACGCG-3′. Lentiviruses were produced by cotransfection with envelope and packaging vectors pMD2.G and pSPAX2 in 293T cells. ARH77, IM9, or 8226 cells were infected with lentiviruses carrying pLKO.1-puro vectors by spin infection at 450 × g for 90 min at room temperature. The cell pellets were resuspended and incubated in fresh RPMI 1640 medium and then were treated with 2.5 μg/ml puromycin at 48 h after the infection. The RPMI 1640 medium with 2.5 μg/ml puromycin was replaced every 2 to 3 days. The cells were collected after 7 days of puromycin selection and then subjected to subsequent assays.

Relative DNA copy number assay.

Cells (1 × 10⁶ cells per sample) were collected and resuspended in 100 μl of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris [pH 8.0]). After brief sonication, immunoprecipitation dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris [pH 8.0], 167 mM NaCl) was added to 1 ml, followed by incubation with proteinase K for 2 to 3 h at 50°C. Three hundred microliters of lysate was then removed and subjected to phenol-chloroform extraction and ethanol precipitation. Precipitated DNA was assayed by RT-qPCR. Relative EBV copy numbers were determined using primers for the Ori-Lyt region of EBV and normalized by the cellular DNA signal for the GAPDH or actin gene locus. Relative copy numbers for the BLRF region were determined using primers across the region (A, B, C, and D) and normalized by the EBV Cp DNA signal. Primers for DNA RT-qPCR are provided in Table 2.

TABLE 2.

Genome primers

Sequence	Primer name
TCGCCTTCTTTTATCCTCTTTTTG	Ori-Lyt_5′
CCCAACGGGCTAAAATGACA	Ori-Lyt_3′
CGGTGCGTGCCCAGTT	GAPDH_5′
CTACTTTCTCCCCGCTTTTTTTT	GAPDH_3′
GCCATGGTTGTGCCATTACA	Actin_5′
GGCCAGGTTCTCTTTTTATTTCTG	Actin_3′
GGCGGGAGAAGGAATAACG	Cp_5′
CTTGAGCTCTCTTATTGGCTATAATCC	Cp_3′
GGTTCCCTCGTTGAAGCACT	A_5′
GCTGTGGCAAGGACGATAGA	A_3′
ACGGGGGCGGACATTATTTA	B_5′
GGCCATCGTTCTGTTTAGCC	B_3′
TCTGGAAGGCGTAGCGTATG	C_5′
GGACCTTCCCCATTTCGCA	C_3′
GAACACGCTGCTGACGGATT	D_5′
AATCTGACTTTGCGTGGAGC	D_3′

Extracellular EBV copy number assay.

Cells were treated with or without NaB (1 mM) and TPA (20 ng/ml) for 96 h. Two hundred microliters of culture supernatant from NaB/TPA-treated or untreated cells (1 × 10⁶ cells/ml) were collected, digested with 10 μl of DNase I (2,000 U/ml) for 1 h at 37°C, and then heat inactivated at 65°C for 10 min. Two hundred microliters of TE (10 mM Tris [pH 8.0], 1 mM EDTA) buffer and 4 μl of proteinase K (25 mg/ml) were added to each sample. The mixture was incubated at 50°C for 2 h and then subjected to phenol-chloroform extraction and ethanol precipitation. Precipitated DNA was dissolved in 100 μl of double-distilled water and then assayed by RT-qPCR using primers for the EBV Cp region, and EBV copy numbers were determined by the standard curve method. A standard curve was generated using serial 10-fold dilutions of Namalwa cell genomic DNA (at 2 copies of EBV genome per cell), varying from 40,000 copies to 4 copies of EBV DNA per reaction.

Cell cycle analysis.

ARH77, IM9, or 8226 cells were washed twice in 1× phosphate-buffered saline (PBS), resuspended in 75% ethanol, and incubated at 4°C for 30 min. Cells were collected by centrifugation, resuspended in 1× PBS supplemented with 20 μg/ml propidium iodide (Sigma) and 6 μg/ml RNase A (Sigma), and then subjected to FACS analysis using FloJo software (Ashland, OR).

FACS for Zta and EA-D.

A total of 1 × 10⁶ cells were harvested and washed in cell staining buffer (PBS containing 5% FBS and 0.09% sodium azide). After the cell staining buffer was decanted, 100 μl of cell suspension was mixed with antibodies to Zta (in-house-generated rabbit anti-Zta) and EA-D (mouse monoclonal antibody; VWR, Radnor, PA) and incubated at room temperature for 30 min. Cells were then washed three times with cell staining buffer, and secondary antibodies conjugated to Alexa 647 (anti-rabbit IgG) and Alexa 555 (anti-mouse IgG; BioLegend, San Diego, CA) were added prior to incubation of the cells at room temperature for 30 min. Cells were again washed three times with cell staining buffer, and flow cytometry was performed on a LSR II flow cytometer (BD Biosciences, Bedford, MA). Forward scatter and side scatter gating was used to exclude dead cells and debris, and pulse geometry gating was used to ensure that single cells were selected for analysis. Data were analyzed using FloJo software.

Animal study.

NSG mice (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) were bred in-house at the Wistar Institute under protocol 112092. Mice were enrolled at 8 weeks of age and housed in microisolator cages in a designated, specific pathogen-free facility, where they were fed sterile food and water ad libitum. Mice were euthanized via CO₂ administration according to AAALAC euthanasia guidelines.

Mice (five males and five females per treatment group) were engrafted intravenously via tail vein injection with a cell suspension (>98% viability) of 2.5 × 10⁶ 8226, ARH77, IM9, Mutu I, B95.8-LCL (transformed with the EBV B95.8 strain), or Mutu-LCL (transformed with the EBV Mutu I strain) cells resuspended in 1× PBS (pH 7.4) and maintained on ice. All cells were transduced to express mCherry-eLuciferase prior to engraftment, to enable bioluminescent imaging and monitoring of cell growth in vivo. For imaging studies, mice were injected with d-luciferin (Gold Biotechnology) intraperitoneally at a dose of 7.5 mg/kg, in a dose volume of 10 ml/kg body weight, 15 min prior to imaging; this was the optimal interval between luciferin injection and bioluminescent imaging, as determined by an initial kinetic curve for these cell lines in mice. Mice were anesthetized using isoflurane prior to imaging. Total body flux (photons per second) was quantitated throughout the study using IVIS imaging software. In experiments for Fig. 6C, EBV-negative myeloma cell lines MM1.S and 8226 were injected intravenously as described above. Animals were weighed three times per week and monitored daily; mice were sacrificed when they lost 20 percent of their body weight or experienced lethargy, labored breathing, impaired locomotion, or failure to respond to stimuli.

Data availability.

RNA-seq and EBV genome DNA-seq data are available in the NCBI GEO database under accession number GSE165196 and a subseries under accession numbers GSE165194 (RNA-seq data for the three cell lines at baseline), GSE165195 (shBLRF2), and GSE171494 (genome).