Control of Epstein–Barr virus reactivation by activated CD40 and viral latent membrane protein 1

Abstract

In humans, Epstein–Barr virus (EBV) establishes a persistent latent infection in peripheral resting B lymphocytes. Virus reactivation is highly restricted. Whereas in healthy humans the infection usually is benign, immunocompromised patients show an increased risk for EBV-associated malignancies, accompanied by an increase in virus replication and in the number of virus-infected cells. To search for viral and host factors regulating virus reactivation, we used conditionally EBV-immortalized B cells. We found that CD40–CD40 ligand interaction and the viral mimic of activated CD40, EBV latent membrane protein 1, suppress virus reactivation. Both inhibit anti-IgM or phorbolester-induced transcription of the viral immediate early protein BZLF1, which controls entry into the viral lytic cycle. The finding that latent membrane protein 1 and CD40 contribute to the regulation of latency may have important implications for the balance between EBV and its host in normal as well as in immunocompromised individuals.

Epstein–Barr virus (EBV) is a ubiquitous human lymphotropic herpesvirus that is associated with a number of malignancies, including Burkitt's lymphoma, nasopharyngeal carcinoma, lymphoproliferative diseases in immunocompromised individuals, and Hodgkin's disease (1). The oncogenic potential of the virus is reflected by its potent transforming capacity in vitro. Upon infection of primary human B lymphocytes in vitro, EBV induces continuous proliferation, leading to the establishment of so-called lymphoblastoid cell lines (LCLs). In these latently infected cells six EBV nuclear antigens (EBNA1, EBNA2, EBNA3A, EBNA3B, EBNA3C, and EBNA-LP) and three latent membrane proteins (LMP1, LMP2A, and LMP2B) are expressed. The same proteins are expressed during the early phase of in vivo infection (2, 3). EBNA2 is the central transcriptional regulator of viral and cellular genes that are involved in B cell immortalization and expressed in LCLs (4). LMP1 also is regulated by EBNA2, acts like a constitutively active receptor of the tumor necrosis factor receptor family (5), and transmits signals to the B cell analogous to those provided by binding of CD40 ligand to the CD40 receptor (6). LMP1 additionally shares functional characteristics with activated CD40. In EBV-immortalized B cells lacking EBNA2, CD40 activation and LMP1 expression result in the same phenotype of prolonged cell survival (7). Activated CD40 and LMP1 both protect B cells from apoptosis (8, 9). In transgenic mice, LMP1 mimics CD40 signals in some, but not all, aspects of B cell differentiation in vivo (10).

Although EBV shows a strong proliferation-inducing potential in vitro, viral infection in vivo is benign in the vast majority of cases. After primary infection the virus persists lifelong in the peripheral blood in resting memory B cells at a frequency of 1–50 viral genome-carrying cells per 10⁶ B cells (11–13). Viral RNA, for e.g. LMP2A, has been repeatedly detected in these cells, but the presence of viral proteins has not been documented (3, 12–15). In normal healthy individuals, virus replication is not detectable in the peripheral blood (16), whereas in patients with pertubations of the immune system the number of EBV-infected B cells in the peripheral blood is increased and viral replication is detectable (15, 17). These patients show an increased risk for EBV-associated malignancies (18–20). Because of the low number of EBV-infected cells in the periphery, it is very difficult to study virus reactivation in vivo. In vitro studies on virus reactivation were mostly done with Burkitt lymphoma cell lines, which, in contrast to LCLs, can be easily induced to enter the lytic cycle by treatment with various agents, including phorbol esters, calcium ionophores, and anti-Ig (21–23). Activation of the lytic cycle initially results in the expression of two viral genes, BZLF1 and BRLF1 (24, 25). The BZLF1 protein is an activator of early viral genes and ultimately of viral replication (4). Expression of BZLF1 alone suffices to initiate the entire lytic cascade (26, 27), and regulation of BZLF1 is thus of pivotal importance for regulation of viral reactivation.

For herpesviruses, viral replication and production of infectious progeny inevitably is associated with the death of the cell. EBV persistence in B cells and viral replication are thus two mutually exclusive phenomena. We reasoned that viral proteins expressed in in vitro EBV-immortalized cells and during primary infection in vivo, and/or cellular proteins induced by these viral proteins, should be involved in suppression of the lytic cycle. Even though some viral and cellular genes have been reported to participate in the maintenance of latency (28–30), studies systematically searching for a lytic cycle-suppressing activity among the EBV-encoded latent antigens have not been reported.

To study the role of latent EBV gene products involved in the control of viral reactivation we used LCLs expressing a conditional EBNA2-estrogen receptor fusion protein (EREB2 cells) (31). In these cells virus reactivation can be studied in the presence and absence of active EBNA2 and EBNA2-dependent genes. We show here that LMP1 as well as activated CD40 can suppress EBV lytic cycle induction in cells expressing no EBNA2 and that both proteins act through inhibition of BZLF1 transcription. We propose that LMP1 or CD40 signaling contribute to the maintenance of in vivo latency of EBV.

Materials and Methods

Cell Lines.

All cell lines were grown in RPMI 1640 medium supplemented with 10% FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 units/ml penicillin, and 100 μg/ml streptomycin. Estrogen was added to the medium of EREB2 cells at a final concentration of 1 μM. Akata, Elijah, Mutu I, and P3HR1 are Burkitt lymphoma cell lines (32–35). EREB2–1, EREB2–5, and EREB2–8 are estrogen-dependent conditional LCLs (31, 36). The SVLMP1–11C and SVLMP1–13C cell lines are EREB2–5 cells stably transfected with LMP1 under the control of the simian virus 40 promoter (7). The pHEBo-1A cell line is transfected with the control plasmid pHEBo. CD40 ligand-expressing mouse LTK⁻ cells were a kind gift from J. Banchereau, Baylor Institute For Immunology Research, Dallas (37).

Cell Stimulation and Transfection.

For culture without estrogen, cells were washed three times in estrogen-free medium. For induction of EBV's lytic cycle, cells were treated with either 20 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma), plate-bound goat anti-human IgM (Sigma), or goat anti-human IgG (ICN) (5 μg/well on 24-well plates). BZLF1 expression was screened after 24 h, and viral capsid antigen (VCA) expression after 72 h by intracellular fluorescence-activated cell sorter (FACS) analysis, indirect immunofluorescence, or immunoblot analysis. CD40 ligand stimulation was achieved by cocultivation of EBV-positive cells with irradiated (7,500 rad) CD40 ligand-expressing LTK⁻ cells or LTK⁻ cells as a control. EREB2 cells were cotransfected with 5 μg LMP1 or LMP2A and green fluorescent protein (GFP) by electroporation. LMP1 and LMP2A were expressed from the simian virus 40 promoter in pHEBo (7) and GFP from pEGFP-C1 (CLONTECH).

FACS Analysis, Immunofluorescence, and Antibodies.

For intracellular staining cells were fixed in 4% paraformaldehyde, washed in staining buffer (PBS with 1% BSA and 0.03% saponin), and incubated with mouse mAb to VCA (gp125, BALF4) (Chemicon) or mouse mAb to BZLF1 (B21; kindly provided by M. Rowe, University of Wales, Cardiff). FITC goat F(ab)₂ anti-mouse IgG (Sigma) was used as secondary antibody for staining of positive cells. Isotype controls were either mouse IgG1 (Dianova, Hamburg, Germany) or mouse IgG2a (PharMingen). For surface IgM staining, cells were stained with a mouse mAb to human IgM (Dianova). Cells were analyzed by using a Becton Dickinson FACScan with cellquest analysis software.

For indirect immunofluorescence assays, cells were washed in PBS, fixed in methanol/acetone at 4°C, stained with anti-VCA or anti-BZLF1 mAbs and Cy3-conjugated anti-mouse IgG as secondary antibody (Dianova), and analyzed by fluorescence microscopy.

Immunoblots.

Cellular extracts were prepared by sonification in sample buffer (0.13 M Tris⋅HCl, pH 6.8/6% SDS/10% α-thioglycerol), separated on 10% polyacrylamide gels, and transferred to nitrocellulose (Amersham Pharmacia Hybond ECL). Filters were blocked in 5% low-fat milk in PBS and then incubated with mouse mAb to BZLF1 or a human antiserum recognizing EBV early antigens. Immunoreactive proteins were detected by peroxidase-conjugated goat anti-mouse IgG antibody or rabbit anti-human immunoglobulins (Sigma) and enhanced chemiluminescence (ECL system, Amersham Pharmacia).

Northern Blots.

Total cellular RNA was extracted by using an RNAeasy kit according to the manufacturer's instructions (Qiagen, Chatsworth, CA). Ten micrograms of RNA was separated on 1% agarose/formaldehyde gels (20 V, overnight), transferred to Hybond N⁺ membrane (Amersham Pharmacia), and hybridized with a ³²P-labeled BZLF1 probe. The BZLF1 probe consisted of a 0.9-kb ApoI fragment of the BZLF1 cDNA.

Results

Induction of the Lytic Cycle of EBV Is Inhibited by Viral LMP1.

Previous reports studying in vitro reactivation of EBV from a latent to a lytic state were either done in EBV-immortalized LCLs expressing all viral latency antigens or in EBV-positive group I Burkitt lymphoma cell lines expressing only EBNA1. These in vitro systems represent proliferating cells and therefore do not mimic B cells latently infected by EBV in vivo. To address the question whether genes involved in immortalization are involved in the regulation of virus latency, we used B cells immortalized by P3HR1 virus whose EBNA2 defect was complemented by an EBNA2–estrogen receptor fusion protein. In these cells, proliferation depends on estrogen (31). Upon hormone withdrawal, EBNA2 is inactivated and EBNA2-dependent viral and cellular genes are down-regulated (31, 38), resulting in growth arrest. In EREB2 cells, induction of the lytic cycle can be studied in the presence and absence of EBNA2 and EBNA2 target gene products.

To determine the percentage of cells undergoing virus production quantitatively, we established intracellular FACS analysis as an alternative to indirect immunofluorescence staining (39) using mAbs recognizing gp125, a component of the VCA, or the immediate early protein BZLF1. A comparison of indirect immunofluorescence staining and intracellular FACS analysis gave comparable results (data not shown). For induction of the lytic cycle, EREB2–1 cells were treated either with PMA or anti-IgM in the presence or absence of estrogen and stained for BZLF1 or VCA expression after 24 and 72 h, respectively. In the presence of estrogen, virus reactivation was nearly completely blocked, whereas in the absence of functional EBNA2, EREB2–1 cells were highly permissive for induction of the lytic cycle (Fig. 1). Inhibition of lytic cycle antigen expression was not caused by estrogen as induction of the lytic cycle in P3HR1 Burkitt lymphoma cells was not affected by estrogen (data not shown). EREB2–1 cells cultured in the presence or absence of estrogen were tested for comparable levels of surface IgM, to exclude that the observed difference in induction of the lytic cycle does not simply reflect different surface IgM levels (data not shown). The differences in the inducibility of the lytic cycle in the presence and absence of estrogen were not restricted to EREB2–1 cells. They also could be observed in two additional, independently generated EREB2 cell lines, EREB2–5 and EREB2–8 (data not shown). It has recently been shown that LMP2A, whose expression is regulated by EBNA2, inhibits virus reactivation after anti-IgM stimulation of EBV-positive LCLs, but not after PMA treatment (28). Because in the conditional LCLs not only anti-IgM-dependent but also PMA-dependent reactivation was inhibited in an EBNA2-dependent manner, we concluded that there must be an additional inhibitor of virus reactivation besides LMP2A. One likely candidate was the second EBNA2-dependent LMP of EBV, LMP1. LMP1 engages signaling pathways that result in the activation of NF-κB, AP-1, and STAT transcription factors, which are key modulators of cellular responses (40–43). To address the question, whether LMP1 interferes with lytic cycle induction, we tested EREB2–5 cells constitutively expressing LMP1 under the control of the simian virus 40 promoter (SVLMP1–11C or SVLMP1–13C) in the absence of functional EBNA2 (7). These cells constitutively express LMP1 to a level slightly higher than EREB2 cells in the presence of estrogen (data not shown). In LMP1-expressing cells, but not in cells transfected with a control plasmid (pHEBo-1A), PMA-induced or anti-IgM-induced virus reactivation was inhibited in the absence of EBNA2 to the same extent as in the parental cell line expressing functional EBNA2 (Fig. 2). At the time of PMA or anti-IgM stimulation (24 h after estrogen withdrawal) LMP1 was down-regulated in untransfected and control cells, whereas the LMP1 level in SVLMP1 cell lines remained constant (data not shown). Additionally, LMP2A and B messages were not detectable by Northern blot analysis (data not shown). This finding indicates that LMP1 also is able to inhibit virus reactivation. All EREB2 transfectants were checked for comparable surface IgM expression to ensure comparable anti-IgM stimulation (data not shown). To compare the effects of LMP1 and LMP2A expression on anti-IgM-dependent induction of the lytic cycle we performed transient transfection experiments in EREB2–1 cells. Freshly estrogen-depleted EREB2–1 cells were cotransfected with plasmids expressing LMP1 or LMP2A and a plasmid expressing GFP. Two days after transfection, cells were stimulated with anti-IgM, and GFP-positive cells were analyzed 3 days later by FACS for VCA expression. Transfection of LMP1 reduced the number of VCA-expressing cells to 22.2 ± 7.8% and transfection of LMP2A to 67.1 ± 3.8% as compared with the vector control. This finding shows that both LMP1 and LMP2A can inhibit EBV reactivation by anti-IgM in EREB2 cells.

Figure 1.

EBV reactivation in the presence and absence of active EBNA2. EREB2–1 cells incubated in medium containing estrogen (plus estrogen; EBNA2 is active) or deprived of estrogen for 24 h (minus estrogen; EBNA2 is inactive) were plated on 24-well plates at a density of 2.5 × 10⁵ cells/ml. Cells were stimulated by transfer into wells precoated with anti-human IgM (5 μg/well) or addition of 20 ng/ml PMA or were left untreated. Cells were harvested 24 (A) or 72 (B) h after stimulation, intracellularly stained for (A) BZLF1 or (B) VCA expression, and analyzed by FACS analysis. Data shown are means of triplicates ± SD of three independent experiments.

Figure 2.

LMP1 can inhibit EBV reactivation in the absence of EBNA2. Untransfected cells, cells transfected with the vector control (pHEBo-1A), and stably transfected EREB2–5 cells constitutively expressing LMP1 (SVLMP1–13C and SVLMP1–11C) were tested for virus reactivation in the absence of estrogen. Cells were stimulated with PMA or plate-bound anti-human IgM for 72 h. Virus reactivation was examined by intracellular staining for VCA expression and FACS analysis. Data shown are means ± SD of three independent experiments.

Activation of CD40 Also Inhibits EBV's Lytic Cycle.

LMP1 exhibits similar effects on B cell activation as activation of CD40 by its ligand (7, 10, 44). We therefore asked whether activation of CD40 also could inhibit induction of the lytic cycle. In the absence of estrogen, EREB2–1 cells were cocultivated with LTK⁻ cells expressing CD40 ligand or LTK⁻ control cells and stimulated for EBV reactivation. Both PMA-mediated and anti-IgM-mediated induction of the lytic cycle were significantly inhibited by CD40 ligand-expressing cells but not LTK⁻ control cells (Fig. 3A). The degree of inhibition depended on the number of CD40 ligand-expressing LTK⁻ cells. At a ratio of EREB2–1 cells to CD40 ligand-expressing LTK⁻ cells of 1:1, induction of the lytic cycle by PMA was inhibited by 75% and induction by anti-IgM was down to background level. In LTK⁻ controls, PMA rather induced reactivation slightly (Fig. 3A), presumably because of a LTK⁻-dependent feeder effect rescuing dying EREB2 cells. In samples cocultivated with CD40 ligand-expressing LTK⁻ cells this increase in reactivation appears to be hidden by the CD40 ligand-dependent inhibition. CD40 ligand-dependent inhibition of EBV reactivation also was observed in the EBV-positive, EBNA2-defective Burkitt lymphoma cell line P3HR1 after PMA treatment (data not shown) as well as in the EBV-positive, LMP1-negative group I Burkitt lymphoma cell lines Akata, Elijah, and Mutu I upon anti-Ig stimulation (Fig. 3 B and C). This finding demonstrates that inhibition of EBV reactivation by CD40 ligand is a general phenomenon and not restricted to the mutant EBV strain P3HR1.

Figure 3.

CD40 ligand can inhibit EBV reactivation. (A) A total of 2.5 × 10⁵ EREB 2–1 cells deprived of estrogen were stimulated with PMA or anti-IgM in the absence or presence of increasing numbers of irradiated CD40 ligand-expressing LTK⁻ cells (50 ×, 150 ×, and 250 × 10³ cells) or LTK⁻ cells as a control. Seventy two hours after stimulation cells were stained for intracellular VCA expression and analyzed by FACS. Data shown are means of three independent experiments ± SD. Values marked by * are significantly different from those of cells incubated without LTK⁻ cells (P < 0.05, unpaired Student's t test). For cocultivation with 150 × and 250 × 10³ CD40 ligand-expressing cells and anti-IgM stimulation the SD values fall within the symbol. (B) A total of 2.5 × 10⁵ Akata cells were mixed with 2.5 × 10⁵ CD40 ligand-expressing LTK⁻ cells or LTK⁻ control cells, stimulated with plate-bound anti-IgG for 72 h and stained for VCA expression by intracellular FACS analysis. Data shown are means of three independent experiments ± SD. The data marked by * are significantly different from those of cells incubated without LTK⁻ cells (P < 0.05, unpaired Student's t test). (C) Immunoblot of early antigen (EA) expression in Mutu I and Elijah cells after anti-IgM stimulation. A total of 2.5 × 10⁵ Mutu I or Elijah cells were stimulated with plate-bound anti-IgM and cocultivated with 2.5 × 10⁵ irradiated CD40 ligand-expressing LTK⁻ cells or LTK⁻ cells as a control. Forty eight hours after stimulation cells were harvested for preparation of cell extracts and analyzed by immunoblot using a human antiserum reactive with EBV early antigens. The positions of the molecular mass standards (kDa) are indicated on the left.

LMP1 and Activated CD40 Inhibit Activation of BZLF1 Transcription.

As shown above, induction of the lytic cycle in hormone-treated proliferating EREB2–1 cells was blocked at the level of BZLF1 expression (Fig. 1A). We therefore asked whether LMP1 and activated CD40 recapitulate inhibition of viral reactivation at the level of BZLF1, which is in fact the case. Constitutive LMP1 expression in SVLMP1–11C and SVLMP1–13C cells resulted in suppression of BZLF1 expression (Table 1), whereas BZLF1 expression was not inhibited in untransfected EREB2–5 cells or EREB2–5 cells transfected with the vector control. Likewise, in PMA-treated or anti-IgM-treated EREB2–1 cells BZLF1 induction was inhibited by CD40 ligand-expressing LTK⁻ cells in a dose-dependent manner but not by LTK⁻ control cells as revealed by immunoblotting for BZLF1 (Fig. 4). To determine whether inhibition of BZLF1 induction occurs at the transcriptional level, EREB2–5 cells constitutively expressing LMP1 (SVLMP1–11C) and cells transfected with the control vector (pHEBo-1A) were treated with PMA for 4 and 16 h. Total cellular RNA was analyzed by Northern blotting using a BZLF1-specific probe (Fig. 5). The BZLF1 probe detected a 1.0-kb monocistronic BZLF1 and a less abundant 2.8-kb bicistronic BRLF1/BZLF1 transcript. Both transcripts could be detected in unstimulated pHEBo-1A cells and were clearly up-regulated after 16 h of PMA stimulation. In SVLMP1–11C cells, no BZLF1 message could be detected before and after stimulation, indicating that the observed block of BZLF1 synthesis is at the transcriptional level. The RNA data are in accordance with the level of BZLF1-positive cells detectable by immunofluorescence (Table 1). To analyze the effect of CD40 ligand on BZLF1 RNA levels, EREB2–1 cells were treated with PMA for 16 h and cocultivated with CD40 ligand-expressing LTK⁻ cells or LTK⁻ control cells at a ratio of 5:1 or 1:1. Cocultivation with CD40 ligand-expressing cells resulted in a dose-dependent inhibition of BZLF1 transcription after PMA stimulation (Fig. 5).

Table 1.

LMP1 inhibits BZLF1 induction

Stimulus	Cell line
	EREB2-5	pHEBo-1A	SVLMP1-13C	SVLMP1-11C
No stimulus	<0.1%	0.6%	0.6%	<0.1%
PMA	6.9%	16.0%	3.7%	0.8%
Anti-IgM	53.1%	19.2%	3.7%	1.1%

EREB2-5 transfectants were incubated in medium without estrogen for 24 h and then plated on 24-well plates at a density of 2.5 × 10⁵ cells/ml. Stimulation was achieved by addition of 20 ng/ml PMA or transfer into wells precoated with anti-human IgM (5 μg/well). Twenty-four hours after stimulation cells were stained for BZLF-1 expression by indirect immunofluorescence. The percentages of BZLF-1-positive cells of one representative experiment are shown. 

Figure 4.

CD40 ligand inhibits BZLF1 expression in a dose-dependent fashion. Immunoblot of BZLF1 expression in EREB2–1 cells after anti-IgM and CD40 ligand activation. A total of 2.5 × 10⁵ cells cultured in estrogen-free medium were stimulated with plate-bound anti-IgM or PMA in the absence or presence of increasing numbers of irradiated CD40 ligand-expressing LTK⁻ cells (10 ×, 50 ×, 150 ×, and 250 × 10³ cells) or LTK⁻ cells as a control. Twenty-four hours after stimulation, cells were harvested for preparation of cell extracts. To adjust for comparable protein concentrations of cell extracts, LTK⁻ cells were added to a final concentration of 2.5 × 10⁵ LTK⁻ cells directly before preparation of the extracts.

Figure 5.

LMP1 and activated CD40 inhibit BZLF1 transcription. PHEBo-1A (stably pHEBo-transfected), SVLMP1–11C (stably SVLMP1-transfected), and untransfected EREB2–1 cells were incubated without estrogen for 24 h and then stimulated with PMA. PHEBo-1A and SVLMP1–11C cells were stimulated for 0, 4, and 16 h and EREB2–1 cells for 16 h in the absence or presence of irradiated CD40 ligand-expressing LTK⁻ cells or LTK⁻ cells as controls. BZLF1 expression was monitored by Northern blot analysis using total cellular RNA and a labeled 0.9-kb BZLF1 probe. The positions of the 1.0-kb BZLF1 and the 2.8-kb BRLF1/BZLF1 transcripts are indicated by arrows. The 28S and 18S bands of the ethidium bromide-stained RNA gels served as loading controls.

Discussion

EBV establishes a lifelong persistence in its host. The viral reservoirs are resting memory B lymphocytes. It is not known whether these persistently infected memory cells express any viral gene product. LMP2A transcripts have been detected in latently EBV-infected memory cells (3, 12–14), but not consistently in all instances tested (15). To ensure viral persistence in a long-lived nonproliferating cell pool, lytic virus replication either has to be suppressed or an equilibrium has to be guaranteed between virus production and continuous regeneration of the latently infected cellular reservoir by means of new infection. Virus reactivation in normal healthy individuals is not detectable in the peripheral blood (16). It is probably restricted to oropharyngeal lymphoid tissues (13, 45, 46).

Following the hypothesis that B cell immortalization and lytic virus replication are mutually exclusive phenomena we assumed that one or several of the EBV latent proteins must contribute to suppression of the lytic cycle. To test this assumption, we used the EREB2 system in which the function of EBNA2 can be switched on and off reversibly by addition or deprivation of estrogen. In the absence of estrogen, EBNA2-dependent genes are down-regulated and cell proliferation arrests. In the absence of functional EBNA2, this cell culture system turned out to be highly permissive for virus reactivation by stimuli like PMA and antigen receptor crosslinking. Switching on EBNA2 and EBNA2-regulated genes suppressed viral reactivation by these stimuli. We also could show that the EBNA2 target LMP1 can independently of other EBNA2-regulated genes inhibit the viral lytic cycle induction by anti-IgM or PMA to the same degree as functional EBNA2. LMP1 does so by blocking the transcription of the EBV immediate early protein BZLF1, which acts as a master switch for induction of the lytic cycle. This is a completely novel function of LMP1 that partly overlaps with the well-documented function of LMP2A to specifically inhibit EBV reactivation by anti-IgM treatment (28). In contrast to LMP2A, LMP1 blocked BZLF1 induction independently of the way of stimulation. A comparison of LMP1 and LMP2A in EREB2 cells indicated that both exhibit inhibitory activity on anti-IgM-induced EBV reactivation and may thus act as a double safeguard. It is noteworthy that LMP1 is up-regulated in group I Burkitt lymphoma cell lines by agents like anti-IgM and PMA that induce the lytic cycle (47, 48). The rate of lytic cycle induction is high if BZLF1 induction precedes LMP1 synthesis (as described for Akata cells), and induction of the lytic cycle is substantially less efficient in cells in which LMP1 is induced before or concomitantly with BZLF1 (48). Induction of LMP1 in group I Burkitt lyphoma cell lines may explain why most Burkitt lymphoma cell lines are poor virus producers upon anti-IgM or PMA treatment with the notable exception of Akata cells.

We also could show that LMP1 shares its capability to inhibit lytic cycle induction with ligand-activated CD40. Inhibition of anti-IgM- or PMA-induced lytic cycle induction by CD40 ligand was observed in EREB2 cells and additionally in three different Burkitt lymphoma cell lines. Our finding that both LMP1 and activated CD40 inhibit EBV reactivation in vitro is another example of CD40 mimicry by LMP1. For a discussion of the in vivo significance of our findings it is noteworthy that CD40 activation not only prevents induction of virus replication in hormone-deprived EREB2 cells or Burkitt lymphoma cell lines, it also inhibits spontaneous EBV reactivation in resting memory B cells that have been naturally infected by the virus in vivo (L. D. Klaman and D. A. Thorley-Lawson, personal communication). We propose the following scenario for a role of LMP1 and CD40 activation during establishment and maintenance of EBV latency in vivo. After primary infection of B cells, LMP1 and, independently from LMP1, LMP2A expression may suppress the lytic cycle in EBV-infected cells entering the B cell memory compartment in which EBV persists (11–13, 15). During this B cell differentiation process LMP1, and in some instances LMP2A, transcription (15) is finally shut down. If recirculating resting B cells encounter antigen, receptor crosslinking will lead to B cell apoptosis (49) and in the case of an EBV-positive B cells also to induction of the lytic cycle and release of infectious viral progeny. In the presence of T cells help by means of CD40–CD40 ligand interaction, both B cell apoptosis (50) and viral reactivation will be inhibited. Stimulation of CD40 with an anti-CD40 antibody has been reported to induce EBV reactivation in LCLs in the absence of other stimuli (51). This result seems contradictory to our findings but may reflect the fact that soluble anti-CD40 mAb has a different biological activity than the same antibody immobilized on the surface of L cells or CD40 ligand presented by L cells (37). Our in vitro data are supported by a recent report on in vivo experiments showing that CD40 stimulation inhibits reactivation of MHV68, another gammaherpesvirus, which is closely related to EBV (52). CD40-dependent control of virus reactivation has important implications for the development of EBV-associated malignancies. Burkitt's lymphoma and posttransplant lymphoproliferative disease preferentially arise under conditions of severe perturbation of the immune system (18–20). In acute malaria and under immunosuppression after organ transplantation, the number of EBV-infected B cells in the peripheral blood is strongly increased (15, 17). In contrast to healthy carriers, immunocompromised patients additionally show viral replication in the peripheral blood, which reflects a more permissive environment for virus reactivation. Two different, but not mutually exclusive, mechanisms may account for increased viral replication in immunocompromised individuals: (i) stimulation of EBV-positive memory B cells by antigen in the absence of T cell help by means of CD40–CD40 ligand interaction, and (ii) an impaired cytotoxic T lymphocyte (CTL) response against antigens of EBV's lytic cycle, which usually control spontaneous or antigen-induced reactivation. Immediate early proteins are targets of anti-EBV CTL responses and cells expressing these proteins are likely to be eliminated by a CTL response (53–55). The detection of immediate early and early transcripts but no viral replication in EBV-positive B cells in the peripheral blood of healthy virus carriers supports the idea that cells entering the lytic cycle are eliminated by CTLs before they produce infectious virus (15, 54, 56). A cytotoxic T cell response against immediate early and early antigens and CD40 ligand-dependent suppression of spontaneous and antigen-induced virus reactivation may thus account for the tight control of EBV lytic cycle in the peripheral blood. A similar double safeguard mechanism has recently been demonstrated for the control of MHV-68 reactivation (52). A loss of this control mechanism may give rise to increased viral reactivation and an accumulation of EBV-positive B cells in the peripheral blood of immunosuppressed patients, which in turn might increase the risk of these patients to develop an EBV-associated malignancy.

In summary, we have described a mechanism in which EBV exploits the immune system to maintain viral latency. Stimulation of EBV-infected B cells by CD40 ligand inhibits lytic cycle induction and most likely also contributes to the maintenance of viral latency in long-lived, nonproliferating memory B cells in vivo. EBV's own mimic of activated CD40, LMP1, also blocks lytic cycle induction and thus may aid to establish viral latency in B cells.

Abbreviations

LMP

latent membrane protein

EBV

Epstein–Barr virus

EBNA

EBV nuclear antigen

LCL

lymphoblastoid cell line

PMA

phorbol 12-myristate 13-acetate

VCA

viral capsid antigen

FACS

fluorescence-activated cell sorter

GFP

green fluorescent protein