INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Epstein-Barr virus (EBV) is a ubiquitous human herpesvirus that establishes a life-long latent infection within B lymphocytes. Latency is associated with the expression of as many as ten viral proteins, including a family of six nuclear proteins (EBNAs), three integral membrane proteins (LMPs) and the recently identified polypeptide RK-BARF0 (17, 24). Several of these, most notably the EBNA-3 proteins, elicit a strong EBV-specific cytotoxic T-lymphocyte immune response (23, 38, 43). Through differential expression of EBV proteins that are immunogenic, potentially oncogenic and those necessary for sustaining infection, an equilibrium is established between infected B cells and the host's immune surveillance. As a consequence, a relatively stable pool of latently infected cells is maintained by the host (31, 36, 56). Thus, EBV is particularly well adapted for persistence within B lymphocytes of an immune host, and though the virus encodes the oncogenic protein LMP-1, its association with human malignancy is rare, especially considering the high incidence of EBV infection worldwide.

As a result of the balanced relationship that exists between EBV and its host, pathogenesis associated with EBV latency is generally a consequence of either an immune deficiency, leading to EBV-induced immunoblastic lymphoma, or a known or suspected secondary genetic or environmental event that promotes full oncogenic transformation of latently infected cells (43). Nevertheless, expression of LMP-1 in some tumors, most notably in EBV-positive Hodgkin's lymphoma and nasopharyngeal carcinoma, strongly suggests that EBV indeed actively contributes to the oncogenic potential of these malignancies (5, 11, 14, 22, 41, 53, 61).

In contrast to tumors such as Hodgkin's lymphoma and nasopharyngeal carcinoma, LMP-1 is not expressed in EBV-positive Burkitt lymphoma (BL), in which the pattern of viral gene expression, referred to as type I latency, is restricted to the proteins EBNA-1 (required for EBV DNA maintenance) and RK-BARF0 and to two small noncoding but highly expressed RNA polymerase III transcripts, EBER-1 and EBER-2 (2, 6, 17, 18, 46). Neither RK-BARF0 or the EBERs are required for EBV-mediated immortalization of B cells in vitro (44, 54). Furthermore, although EBNA-1 has been linked to development of B-cell lymphomas in some transgenic mice (59, 60), there is as yet no evidence that EBNA-1 directly contributes to tumorigenicity in EBV-associated malignancies. Overall, these findings argue against a significant contribution by EBV to tumorigenic potential in established BL. Thus, the dominant factor responsible for maintenance of tumorigenicity in BL is believed to be the deregulated expression of the c-MYC proto-oncogene (in this paper, c-MYC refers to the gene and c-myc refers to the mRNA and cDNA), which occurs as the result of characteristic chromosomal translocations that juxtapose the c-MYC gene and immunoglobulin loci (reviewed in reference 29). Indeed, programmed expression of c-MYC in the B-cell compartment of transgenic mice predisposes these animals to protracted development of lymphoma (1).

The assumption that EBV does not directly contribute to tumorigenicity in BL, however, has been challenged by the finding that a loss in tumorigenic potential is associated with spontaneous loss of the EBV genome in the Akata BL cell line, which in culture maintains the type I latency program characteristic of primary BL tumors (52). Here we report that EBV-associated tumorigenicity of Akata BL cells correlates with an increased resistance to apoptosis that is coincident with a down-regulation of c-MYC protein, but not mRNA, and increased expression of the antiapoptotic protein Bcl-2. Reinfection of EBV-negative Akata cells, but not stable expression of EBNA-1 alone, restored EBV-dependent regulation of c-MYC and tumorigenicity. These data therefore provide direct evidence that EBV can significantly contribute to tumorigenic potential in BL. The data also indicate that the type I latency program of EBV gene expression contributes to long-term survival of latently infected cells and suggest a novel model whereby EBV suppresses apoptosis by selectively down-regulating c-MYC protein expression under limiting growth conditions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Cell culture and EBV infection. Cell lines were maintained in RPMI 1640 medium supplemented with 2 mM L-glutamine and 10% defined fetal bovine serum (HyClone); the exceptions were HL60 and K562, which were maintained in media containing 20% serum. Akata, SavI, KemI, and MutuI are human group I BL cell lines that maintain a type I EBV latency. IB4 is an EBV-immortalized human B-lymphoblastoid cell line (LCL) that expresses all known EBV latency-associated genes (type III latency). HL60 and K562 are human promyelocytic and erythroleukemic cell lines, respectively. Clonal EBV-positive Akata cells were derived from colonies isolated in soft agar (see below) and designated A.1, A.3, A.5, etc. The clonal EBV-negative Akata cell lines 3F2, 2C1, and 2A8 (gift of J. Sixbey) were isolated by single-cell sorting (FACStar Plus; Becton Dickinson) following treatment of parental EBV-positive Akata cells with 50 µM hydroxyurea (9). Loss of the EBV genome in these cells was confirmed by the lack of EBNA-1 expression as determined by immunoblot analysis and by the inability to amplify EBV DNA by PCR with primers specific for the large internal (BamHI-W) repeat of the EBV genome and/or the inability to detect EBV DNA by fluorescence in situ hybridization (FISH) using a cosmid DNA probe containing the EBV SalI B fragment (9).

Cell death assays. Cultures were seeded at 2.5 × 10⁵ cells per ml in standard growth medium (containing 10% serum), and cell numbers were determined daily to assess growth rate. On days 2 through 6, an aliquot of cells was removed, washed three times in growth medium that contained 0.1% serum, and then resuspended in this reduced-serum medium at 5 × 10⁵ cells per ml. Viability was then determined daily, using trypan blue dye exclusion to monitor cell survival. For cytospin preparations, cells from 100 µl of culture were pelleted onto a glass slide in a Cyto-Tek chamber holder (Sakura), air dried, fixed in methanol for 5 min at room temperature, air dried, stained in 0.02% (wt/vol) modified Giemsa stain (Sigma) for 45 min at room temperature, rinsed several times in distilled water, air dried, and then mounted in Permount (Fisher).

Immunoblotting. For the detection of EBNA-1, LMP-1, c-MYC, and Bcl-2 by immunoblotting, 10⁶ cells were washed once in phosphate-buffered saline, lysed in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer (100 mM Tris-HCl [pH 6.8], 200 mM dithiothreitol [DTT], 4% SDS, protease inhibitor cocktail [Sigma], 20% glycerol, 0.2% bromophenol blue), immediately heated at 100°C for 5 min, and sonicated. Proteins were fractionated by SDS-PAGE in 10% acrylamide gels, transferred to an Immobilon P membrane (Millipore), and immunoblotted by using an enhanced chemiluminescence detection system (Amersham). Primary antibodies used for immunoblotting were mouse monoclonal antibodies to c-MYC (9E10 [13]), Bcl-2 (M0887; Dako), and LMP-1 (S12 [30]) and a polyclonal rabbit antiserum to EBNA-1 (gift of J. Hearing). Immunoreactive proteins were detected with secondary antibodies conjugated to horseradish peroxidase. All blots were subsequently stripped of antibody in 62.5 mM Tris-HCl (pH 6.8)-100 mM 2-mercaptoethanol-2% SDS (50°C for 30 min) and reprobed with a mouse monoclonal antibody to actin (N350; Amersham) as a control for protein loading. For analysis of protein expression in SCID mouse tumors, approximately equal amounts of tumor tissue were solubilized in SDS-PAGE sample buffer and processed as described above. For detection of EBNA-1 and LMP-1 in Fig. 7, 10⁷ cells were lysed in 200 µl of 50 mM Tris (pH 8.0)-150 mM NaCl-1 mM EDTA-1% Triton X-100-protease inhibitor cocktail-0.5 mM phenylmethylsulfonyl fluoride (PMSF) and then centrifuged at 12,000 × g for 5 min to remove insoluble material. The protein concentration of the supernatant was determined by the Bradford method (Bio-Rad), and 50 µg of protein was then subjected to SDS-PAGE and immunoblotting.

Measurement of c-MYC stability and synthesis. For cycloheximide-chase experiments, protein synthesis was inhibited in cells that were in either mid-log- or stationary-phase growth by addition of cycloheximide (Calbiochem-Novabiochem) to the cell culture at a final concentration of 25 µg/ml. An aliquot of 3 × 10⁶ cells was harvested at 0, 15, 30, 60, 120, and 240 min after addition of cycloheximide, washed once in phosphate-buffered saline, resuspended in lysis buffer (50 mM Tris-HCl [pH 8.0], 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40 [NP-40], 1 mM PMSF), and incubated for 20 min at 4°C. Lysate containing protein from the equivalent of 10⁶ cells from each chase point was then analyzed for c-MYC expression by immunoblotting with anti-c-MYC antibody (9E10) as described above. For pulse-chase analysis, EBV-negative and -positive Akata cells (10⁸) in log- or stationary-phase growth were washed twice in methionine- and cysteine-free RPMI 1640 medium (BioWhittaker) containing L-glutamine and 10% serum. Cells were then incubated at 37°C in this medium for 30 min, followed by a pulse-labeling period of 15 min in the same medium containing 1 mCi of Tran³⁵S-label (ICN). Cells were then washed twice in medium containing a fivefold excess of methionine and cysteine and incubated in this medium for up to 240 min. At chase times of 0, 15, 30, 60, 120, and 240 min, 1.25 × 10⁷ cells were removed, pelleted, resuspended in 1 ml of lysis buffer (50 mM Tris-HCl [pH 8.0], 5 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1 mM PMSF), and incubated for 20 min at 4°C. All lysates were then precleared with Protein G Plus-Agarose (Oncogene Research Products) to remove nonspecific adsorbents. Lysate containing protein from the equivalent of 10⁷ cells was processed by immunoprecipitation with 2 µg of antibody to c-MYC (Ab-3; Oncogene Research Products) and 20 µl of Protein G Plus-Agarose. Immunoprecipitates were washed three times in SNNTE buffer (5% sucrose, 1% NP-40, 500 mM NaCl, 50 mM Tris-HCl [pH 7.5], 5 mM EDTA) and once in radioimmunoprecipitation assay buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate), resuspended in 30 µl of loading buffer (100 mM Tris-HCl, 200 mM DTT, 4% SDS, 20% glycerol, 0.2% bromophenol blue), boiled for 4 min, and then subjected to SDS-PAGE in a 10% acrylamide gel. Gels were fixed in 10% acetic acid-30% methanol for 30 min, washed three times in distilled water for 10 min, soaked in fluor solution (1 M sodium salicylate, 5% glycerol) for 20 min, dried, and exposed to Kodak BioMax MR film. To calculate protein half-life, immunoprecipitated c-MYC from each chase point was quantified by phosphorimage analysis (Molecular Dynamics) and normalized for total protein, and the resulting data were subjected to linear regression analysis.

Analysis of RNA. Total cellular RNA was isolated from 10⁷ cells with RNAzol B as recommended by the manufacturer (Tel-Test), followed by extraction with an equal volume of phenol-chloroform and then chloroform prior to ethanol precipitation. For Northern blot analysis, 10 µg of RNA per sample was fractionated by electrophoresis in a 1.2% agarose-2.2 M formaldehyde gel (48) and blotted onto a GeneScreen Plus membrane (DuPont). RNA blots were prehybridized for 4 h at 45°C followed by hybridization at the same temperature to a ³²P-labeled human c-myc cDNA probe under conditions described previously (39). Following hybridization, blots were washed at 62°C (49) and processed by autoradiography. Blots were then stripped and rehybridized to a probe specific for 28S rRNA to control for differences in RNA loading of the gel.

Generation of EBV/EBNA-1⁺ cell lines. To generate EBV-negative Akata cells that stably expressed EBNA-1, the EBNA-1 open reading frame was inserted into the BamHI restriction site of the retroviral expression vector pLXSN (33) to yield pLE₁SN, in which expression of EBNA-1 is under control of the retroviral 5' long terminal repeat. Ten micrograms of pLE₁SN or pLXSN (to generate control cell lines) was used to transfect 8 × 10⁶ EBV-negative Akata cells by electroporation as described previously (50). At 48 h posttransfection, cells were placed in 24-well tissue culture plates (10⁴ cells per well) in medium containing 200 µg of G418 per ml. To generate clonal EBV/EBNA-1⁺ cell lines, single cells were isolated from pools of G418-resistant EBNA-1-positive cells by cell sorting (FACStar Plus; Becton Dickinson) into 96-well plates containing conditioned medium. Following expansion of these cells under selection with G418, clones that expressed EBNA-1 were identified by immunoblotting.

Growth in soft agar and tumorigenicity assays. For the analysis of growth potential in soft agar, 10⁴ cells were suspended in 3 ml of 0.33% agarose (SeaPlaque; FMC Corporation) in standard growth medium (without phenol red) and plated onto a 3-ml layer of solidified 0.66% agarose in the same medium in a 6-cm-diameter tissue culture dish. Once the top layer of agarose had solidified, the cells were incubated at 37°C in a humidified atmosphere of 5% CO₂ for 4 weeks and fed weekly with 1 ml of liquid medium containing 0.33% agarose. Each cell line tested was plated in triplicate.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Enhanced growth potential of BL is associated with type I EBV latency. EBV-negative Akata BL cells are nontumorigenic relative to their EBV-positive counterparts when assayed for growth in soft agar and tumor induction in athymic nude mice (52). Akata BL cells may therefore provide an ideal system to address the contribution of EBV to tumorigenicity in BL. However, the low plating efficiency of EBV-positive Akata cells in soft agar (1 to 3% [52]) raised the possibility that enhanced growth potential associated with EBV infection in Akata BL cells was not a stable phenotype. We therefore initially addressed whether EBV-positive Akata cells and several clones of EBV-negative Akata cells had consistent and measurably different growth properties. The EBV-negative clones analyzed either had lost the EBV genome spontaneously (two clones) or were generated by treatment with hydroxyurea (three clones), which hastens the loss of EBV episomes in these cells (9). Under the low dose of hydroxyurea used (50 µM), which is approximately 40-fold lower than that used therapeutically, no toxic effects were observed, nor did we observe any increase in genomic instability as a result of treatment with hydroxyurea (9). Loss of the EBV episome was confirmed by failure to detect EBV DNA by PCR and/or fluorescence in situ hybridization (FISH) (data not shown). As illustrated by representative data presented in Fig. 1A, EBV-positive Akata cells consistently exhibited a greater growth potential in soft agar relative to EBV-negative cells. Although an EBV-negative cell line would occasionally exhibit a plating efficiency similar to those of EBV-positive cells (1 to 3%), the resulting colonies from these cells were always markedly smaller than those derived from EBV-positive cells (see below). Thus, our findings were consistent with previous observations suggesting that EBV contributes to the growth potential of Akata BL cells (52).

View larger version (56K): [in this window] [in a new window]   FIG. 1.   Enhanced growth potential of EBV-positive Akata cells is independent of LMP-1. (A) Clonal EBV-negative and parental EBV-positive Akata cells were plated in 0.33% agarose at a density of 104 cells per 6-cm-diameter dish. Colonies were photographed 4 weeks after plating. (B) Individual colonies derived from the EBV-positive parental Akata cells were transferred from agarose to liquid culture, briefly expanded, and analyzed for expression of EBNA-1 and LMP-1 by immunoblotting. Each lane contained protein from 106 cells. IB4 is an EBV-immortalized LCL used as a control for detection of LMP-1 expression.

Type I EBV latency in BL confers resistance to apoptosis under growth-limiting conditions. Although EBV-positive Akata cells demonstrated a greater growth potential than their EBV-negative counterparts in soft agar, we did not observe a significant difference in growth properties during standard culture in liquid medium, as EBV-positive and -negative Akata cells exhibited identical growth rates and reached similar saturation densities (Fig. 2A). However, EBV-negative cells rapidly lost viability after 5 to 6 days in culture without feeding, whereas EBV-positive cells often remained viable for an additional 2 to 3 days (data not shown). This finding suggested that the mechanism(s) responsible for the increased growth potential associated with EBV in Akata BL cells was manifest only under growth-limiting conditions, such as would occur in soft agar or in vivo.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   The type I EBV latency program promotes BL cell survival following serum deprivation. (A) Clonal EBV-positive (shaded symbols) and EBV-negative (open symbols) Akata cells were seeded at 2.5 × 105 cells per ml in standard growth medium containing 10% serum, and cell concentration was monitored daily for 6 days to assess growth rate. (B) Cells from the log (day 2) and stationary (day 5) phases of growth were washed and transferred to growth medium containing 0.1% serum, in triplicate, at a density of 5 × 105 cells per ml. Cell viability was determined daily by trypan blue dye exclusion. Results are representative of seven independent experiments.

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Serum deprivation induces morphological changes in BL cells that are characteristic of apoptosis. EBV-positive (A.3) and -negative (2C1) Akata cells were maintained in medium containing 0.1% serum for 3 days after reaching the stationary phase of the growth cycle. Cytospin preparations shown here indicate that the EBV-negative cells displayed changes typical of apoptotic cells, including condensation of chromatin into micronuclei and vacuolation. By contrast, EBV-positive cells appeared relatively healthy.

Suppression of apoptosis by EBV is associated with down-regulation of c-MYC expression. To determine if altered expression of c-MYC might account for EBV-mediated suppression of apoptosis, we examined c-MYC protein levels in EBV-positive and EBV-negative Akata cells at specific intervals of the cell growth cycle. Additionally, we examined the expression of several anti- and proapoptotic proteins of the Bcl-2 family. During log-phase growth (Fig. 4, days 1 to 3), there were no detectable differences in the levels of c-MYC protein between EBV-positive and EBV-negative cells. However, consistent with their fate when deprived of survival factors, c-MYC was dramatically down-regulated in EBV-positive cells as they entered the stationary phase of growth. By contrast, EBV-negative cells expressed c-MYC at levels comparable to those observed during log-phase growth. Both EBV-positive and EBV-negative Akata cells showed increased expression of the antiapoptotic protein Bcl-2 as they approached the stationary phase of growth, and EBV-positive Akata BL cells generally had higher steady-state levels of Bcl-2 protein than the EBV-negative Akata cells (Fig. 4). We did not observe any EBV- or growth phase-associated differences in the expression of the antiapoptotic proteins Mcl-1 and Bcl-X_L or the proapoptotic protein Bax (data not shown).

View larger version (37K): [in this window] [in a new window]   FIG. 4.   EBV- and growth phase-associated expression of c-MYC and Bcl-2. Clonal EBV-positive (A.3 and A.15) and EBV-negative (3F2, 2C1, and 2A8) Akata BL cells were seeded (2.5 × 105 per ml) in complete growth medium, and 106 cells were harvested daily from the experiment presented in Fig. 2A for the first 5 days in culture to monitor expression of c-MYC and Bcl-2 by immunoblot analysis. Detection of actin served as a loading control. Results are representative of three independent experiments.

View larger version (57K): [in this window] [in a new window]   FIG. 5.   c-MYC is a common target for down-regulation in group I BL cell lines. The group I BL cell lines KemI, SavI, and MutuI were seeded in complete growth medium (2.5 × 105 per ml), and cell samples were harvested on days 1 through 5 to monitor expression of c-MYC as for Fig. 4.

View larger version (37K): [in this window] [in a new window]   FIG. 6.   EBV down-regulates c-MYC expression through a posttranscriptional mechanism. (A) Northern blot analysis of c-myc mRNA and 28S rRNA (loading control) levels in EBV-negative (3F2 and 2A8) and EBV-positive (A.3 and KemI) BL cells at 2, 3, 5, and 6 days after seeding in complete growth medium at 2.5 × 105 cells per ml. Each lane contained 10 µg of total cellular RNA. (B) Cycloheximide (CHX)- and pulse-chase analysis of c-MYC t1/2 in EBV-negative (3F2) and EBV-positive (A.3) Akata BL cells during the log and stationary phases of the cell growth cycle. At chase times of 0, 15, 30, 60, 120, and 240 min, c-MYC levels were analyzed by immunoblotting (CHX-chases) or by immunoprecipitation and SDS-PAGE of pulse-labeled protein. Following quantification by phosphorimage analysis of immunoprecipitated c-MYC at each chase point, the data were subjected to linear regression analysis to determine the t1/2 of c-MYC (see text). (C) Half-lives of c-MYC during log-phase growth in the human promyelocytic and erythroleukemia cell lines HL60 and K562, respectively, and two additional group I BL cell lines (KemI and SavI), were analyzed by CHX-chase experiments as in panel B.

EBV infection, but not EBNA-1, restores resistance to apoptosis and tumorigenicity to EBV-negative BL. It was possible that the parallel loss of the EBV genome and tumorigenic potential in some Akata cells was merely coincidental or that a propensity to lose the EBV genome was an effect, rather than the cause, of the spontaneous loss of tumorigenicity. To directly test the contribution of EBV to tumorigenicity, we reinfected EBV-negative Akata cells and assessed the consequences of a restored EBV latency on tumorigenic potential. Moreover, since it had been suggested that EBNA-1 alone might directly contribute to the oncogenic potential of EBV (59, 60), we also established several Akata cell lines that constitutively expressed EBNA-1 in an EBV-negative background.

View larger version (46K): [in this window] [in a new window]   FIG. 7.   Restoration of restricted EBV latency and establishment of EBV-independent EBNA-1 expression in EBV-negative Akata cells. (A) 2A8 Akata cells reinfected with EBV (lines 2A8.1, 2A8.2, and 2A8.3) express normal levels of EBNA-1 but, like their EBV-positive counterparts A.15 and A.17, do not express detectable LMP-1. The IB4 LCL provides a positive control and reference for detection of LMP-1. (B) Stable expression of EBNA-1 in EBV-negative Akata cells. A.3 and A.11 are EBV-positive Akata cells; LXSN.1, LXSN.2, and LXSN.3 are G418-resistant EBV-negative Akata cell lines that contain the expression vector pLXSN without DNA encoding EBNA-1; E1.3,12 and E1.2,9 are clonal EBV/EBNA-1+ Akata cells derived from the nonclonal E1.3 and E1.2 lines, respectively. All lanes contained 50 µg of total cellular protein.

View larger version (34K): [in this window] [in a new window]   FIG. 8.   EBV infection, but not EBNA-1 alone, regulates c-MYC levels and sensitivity to apoptosis. (A) Immunoblot analysis of c-MYC throughout the cell growth cycle in reinfected and EBV/EBNA-1+ Akata cells. Cells were seeded as described in Fig. 2 and harvested at the indicated day postseeding. (B) EBV infection, but not EBNA-1 alone, protects Akata cells from apoptosis during stationary-phase growth. The indicated cells were deprived of serum, and percent viability was determined at the indicated intervals by trypan blue dye exclusion. Results are representative of two independent experiments.

View larger version (82K): [in this window] [in a new window]   FIG. 9.   EBV infection restores tumorigenic potential to EBV-negative Akata cells. (A) Growth of EBV-positive (A.15), EBV-negative (2A8), and reinfected 2A8 (2A8.2) Akata cells in soft agar. Shown are low- and high-magnification micrographs of cell colonies resulting from growth in 0.33% agarose. (B) Tumorigenicity of Akata BL cells assayed by tumor induction in SCID mice (representative results). Each mouse received an injection of EBV-negative cells (3F2 or 2A8) in the left hind flank and EBV-positive (A.15) or reinfected 2A8 (2A8.2) cells in the right flank (2 × 107 cells per injection). The mouse that received the 3F2 and A.15 cells was photographed 5 weeks postinjection; the mouse that received the 2A8 and 2A8.2 cells was photographed 7 weeks postinjection. The results from all SCID mouse assays are presented in Table 1.

View larger version (42K): [in this window] [in a new window]   FIG. 10.   Tumors in SCID mice express EBNA-1 but not LMP-1. Representative results from immunoblot analysis of EBNA-1 and LMP-1 expression in tumors that resulted from injection of EBV-positive A.15 and reinfected 2A8 Akata cells are shown. Protein from IB4 cells was used as a positive control; tissue harvested from a site of injection of EBV-negative Akata cells served as a negative control (lane 1). Results of the analysis of all 13 tumors obtained are presented in Table 1.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

We have demonstrated that EBV can directly contribute to the tumorigenic potential of BL in the context of a type I latency program. With the possible exception of EBNA-1 (see below), EBV genes expressed during type I latency have not been directly linked to either B-cell immortalization or oncogenic transformation associated with EBV infection (24). In fact, the EBER and RK-BARF0 genes are dispensable for EBV-induced immortalization of B lymphocytes in vitro (44, 54). B-cell immortalization by EBV, however, occurs in the context of a type III latency (expression of all 12 EBV latency-associated genes), and thus one cannot exclude a possible role for EBNA-1, RK-BARF0, or the EBERs in growth-related functions of the type I latency program. The reservoir of EBV in healthy infected individuals consists of B lymphocytes that, like BL cells, preclude the expression of known growth-promoting EBV proteins (7, 37, 42, 57). Thus, one or more of the EBV genes expressed during the type I latency program may perform a critical role in modulating B-cell growth or survival in vivo.

The concept that the type I latency program of EBV promotes BL cell growth and survival is strongly supported by data presented here demonstrating that (i) EBV-positive Akata BL cells are more resistant to apoptosis than are EBV-negative Akata cells under growth-limiting conditions; (ii) under conditions that favor increased sensitivity to apoptotic stimuli, there is an EBV-dependent reduction in c-MYC, an oncoprotein known to augment the apoptotic program in BL (35) and other cell lineages (3, 12, 40); (iii) EBV-positive Akata BL cells generally, but not always, express higher levels of the antiapoptotic protein Bcl-2 than do their EBV-negative counterparts throughout all stages of the cell growth cycle; and (iv) tumorigenicity in Akata BL cells is strictly dependent on a latent EBV infection characteristic of BL tumors. Based on these observations, we propose that EBV contributes to tumorigenicity in BL by inhibiting c-MYC-induced apoptosis through at least two mechanisms: a modest up-regulation of Bcl-2 expression and, most importantly, a concomitant decrease in c-MYC expression, apparently at the level of translation, under growth-limiting conditions.

Although earlier studies failed to detect Bcl-2 expression in BL cells that maintain type I latency (21, 34), our results clearly indicate that Bcl-2 is expressed in Akata and other group I BL cell lines (Fig. 4 and data not shown). However, the level of Bcl-2 expression needed to effectively inhibit apoptosis in group I BL cells equals or exceeds the much higher levels of Bcl-2 observed in LCLs and BL cells that maintain type III latency (34). Because EBV-negative Akata cells express only slightly lower levels of Bcl-2 than their EBV-positive counterparts, Bcl-2 levels characteristic of EBV-positive BL cells that maintain type I latency seem unlikely to be sufficient alone to fully restore tumorigenicity to EBV-negative Akata BL cells.

The striking finding that EBV targets the down-regulation of c-MYC protein under growth-limiting conditions in BL suggests that this is the principal mechanism by which EBV promotes cell survival, particularly given that c-MYC is the primary mediator of apoptosis in BL (35) and that Akata BL cells are null for p53 (15). To determine if down-regulation of c-MYC is indeed responsible for the increased survival of BL cells, we have attempted to repress expression of c-MYC in EBV-negative Akata cells during stationary phase, using antisense c-myc oligonucleotides that have been successfully used in BL (35). Unfortunately, we were unable to demonstrate a repression of c-MYC by such means and therefore have been unable to directly test whether repression of c-MYC is sufficient to promote group I BL cell survival. Interestingly, our observations regarding BL cells contrast those for EBV-transformed lymphoblastoid cells, in which expression of c-MYC protein is sustained under growth-restrictive conditions (8), most likely through an EBV-dependent stabilization of c-myc mRNA that occurs during type III latency (26, 27). Although these opposing effects of EBV on c-MYC expression likely result from the dominant influence of the respective EBV latency program maintained by a cell, we cannot exclude the possibility that the EBV-dependent decrease in c-MYC protein observed here is unique to BL. Presumably, the documented ability of cells that maintain type III latency to substantially up-regulate the antiapoptotic proteins Bcl-2 and A20 would protect such cells having sustained c-MYC expression (16, 19, 21, 28, 47).

Although expression of the EBV EBNA-1 protein may be responsible for the protracted development of B-cell lymphoma in some lines of transgenic mice (59, 60), our observations indicate that neither EBV-dependent tumorigenicity nor regulation of c-MYC and Bcl-2 expression in BL can be attributed to EBNA-1 alone. Other EBV gene products that are consistently expressed in BL are the small RNAs EBER-1 and EBER-2 and the more recently identified RK-BARF0 protein (2, 17). Whereas the function of RK-BARF0 is unknown, at least two functions have been ascribed to the EBER RNAs: first, EBERs bind to the double-stranded RNA-dependent protein kinase PKR and disrupt the ability of PKR, a protein with potential tumor suppressor activity, to inhibit translation (10, 25, 32, 51); second, EBER RNAs bind to the ribosomal protein L22 and sequester L22 to the nucleus (58). Given that the EBERs are capable of targeting two proteins associated with the translational process and our data suggesting that down-regulation of c-MYC expression is most likely mediated at the level of translation, the possibility that the EBERs contribute to cell survival as defined here is particularly attractive.

Although our observations provide important insights into the long-standing debate over the role of EBV in BL, equally important are the implications of these findings with respect to the maintenance of EBV latency in the healthy EBV-immune host. Because the EBV-dependent down-regulation of c-MYC in BL appears to occur through a posttranscriptional mechanism, this type of regulation may also be operational within normal latently infected B cells in vivo. Thus, EBV may contribute to its persistence by promoting the survival of proliferating latently infected B cells. Although a major reservoir of EBV in the peripheral blood appears to be resting B cells that lack EBNA-1 mRNA but which express LMP-2A transcripts (37), cells that express EBNA-1 mRNA in the absence of transcripts that encode the other EBNA proteins and LMP-1 are detectable as well (7, 57). This finding suggests that there is a subpopulation of infected B cells in vivo that maintains a pattern of EBV gene expression similar or identical to that observed in BL. Such B cells may be actively dividing in response to physiological signals and essential to maintain a critical pool of latently infected B cells that is necessary to sustain long-term infection (reviewed in reference 56). The potential of EBV to usurp physiological pathways of cell proliferation, coupled with its ability to limit cell death under growth-restrictive conditions as shown here, would enable EBV to circumvent the need for virus-induced proliferation and the associated expression of the LMP-1 oncoprotein and other EBV proteins, such as the EBNA-3 family, that are capable of evoking a strong cellular immune response.