Comparative Analysis of Signal Transduction by CD40 and the Epstein-Barr Virus Oncoprotein LMP1 In Vivo

Abstract

There is much evidence, based primarily on in vitro studies, indicating that the Epstein-Barr virus oncoprotein latent membrane protein 1 (LMP1) mimics an activated CD40 receptor. In order to investigate the extent of similarity between LMP1 and CD40 functions in vivo, we analyzed the cytoplasmic signaling properties of LMP1 and CD40 in B cells in a directly comparable manner. For this purpose, we generated transgenic mice expressing either LMP1 or a chimeric LMP1CD40 molecule, which constitutively activates the CD40 pathway, under the control of the CD19 promoter. LMP1 and LMP1CD40 were expressed at similar levels in a B-lymphocyte-specific manner. Similar to LMP1, LMP1CD40 suppressed germinal center (GC) formation and antibody production in response to thymus-dependent antigens, albeit to a greater extent than LMP1. Furthermore, the avidity of the antibodies produced against thymus-dependent antigens was lower for LMP1CD40 transgenic mice than for wild-type and LMP1 transgenic mice. GC suppression was linked to the ability of LMP1CD40 and LMP1 to downregulate mRNA and protein levels of BCL6 and to suppress the activity of the BCL6 promoter. In contrast to LMP1, LMP1CD40 caused an upregulation of CD69, CD80, and CD86 in B cells and a dramatic increase in serum immunoglobulin M. In addition, LMP1CD40 but not LMP1 transgenic mice had elevated numbers of marginal-zone B cells and increased populations of polymorphonuclear cells and/or neutrophils. Consistent with these findings, LMP1CD40 but not LMP1 transgenic mice showed signs of spontaneous inflammatory reactions and the potential for autoimmunity.

Latent membrane protein 1 (LMP1) is tightly linked to the development of most human malignancies associated with Epstein-Barr virus (EBV) infection (28). Multiple pieces of evidence support the oncogenic role of LMP1. LMP1 is detected in tumor cells and is essential for B-lymphocyte transformation by EBV in vitro (17). It has autonomous transforming activity since it can induce rodent fibroblast transformation in vitro and lymphomas in transgenic mice expressing LMP1 in a B-lymphocyte-specific manner (2, 20, 31). Finally, the expression of LMP1 in epithelial cells and B lymphocytes has pleiotropic effects which are characterized by broad alterations in cellular gene expression and result in the inhibition of apoptosis, the induction of proliferation, and the modulation of immune responses (18).

LMP1 is an integral membrane protein consisting of a cytoplasmic 24-amino-acid amino-terminal region, six transmembrane domains connected by short turns, and a cytoplasmic carboxyl-terminal region (CCT) of 200 amino acids (Fig. 1A). The transmembrane domains mediate constitutive aggregation of the protein at the plasma membrane, which is essential for LMP1 signaling. The CCT binds and activates intracellular signaling proteins that are also utilized by members of the tumor necrosis factor receptor (TNFR) superfamily. These proteins include members of the TNFR-associated protein family, the TNFR-associated death domain protein and the receptor interacting protein (13). The identification of common signaling factors that are engaged by LMP1 and TNFRs provided the molecular basis for the formulation of a model for LMP1 functioning as a constitutively activated TNFR. Indeed, in vitro experiments have demonstrated similarities in signal transduction by LMP1 and the B-cell growth factor receptor CD40 (21). Both molecules can activate the NF-κB, JNK, and p38 pathways and induce a highly overlapping spectrum of adhesion molecules and activation markers in B cells. LMP1 expression or CD40 activation can induce proliferation and the prevention of B-cell apoptosis, and CD40 activation can partially complement the loss of LMP1 expression in B-cell transformation by EBV in vitro.

FIG. 1.

LMP1 and LMP1CD40 expression in transgenic mice. (A) Schematic representation of LMP1 and LMP1CD40 molecules. The chimeric molecule LMP1CD40 was generated by replacing the cytoplasmic carboxyl-terminal (C) region of LMP1 (rectangle) with the cytoplasmic region of CD40 (circle). (B) RT-PCR analysis of LMP1, LMP1CD40, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression in B cells (B) and B-cell-depleted populations (NB) of spleen cells isolated by FACS from LMP1 (LMP1 Tg) and LMP1CD40 (LMP1CD40 Tg) transgenic mice or their wild-type littermates. (C) Representative Western blot of whole-cell extracts from equal numbers of cells showing the expression of LMP1 and LMP1CD40 in B cells (B cell) and B-cell-depleted populations of spleen cells (non B cell) isolated by FACS from LMP1 (LMP1 Tg) and LMP1CD40 (LMP1CD40 Tg) transgenic mice or their wild-type littermates. Actin levels were determined and used as a cell extract loading control. (D) Western blot of whole-cell extracts from equal numbers of cells showing the expression of LMP1 in FACS-isolated splenic B cells from LMP1 transgenic mice (LMP1Tg) or EBV-transformed human lymphoblastoid cell lines (LCL1 and LCL2). Actin levels were determined and used as a cell extract loading control.

The oncogenic activity of LMP1 is not limited to the activation of growth stimulatory or antiapoptotic pathways. LMP1 is also capable of modulating immune responses to prevent the recognition and destruction of EBV-infected B cells by T lymphocytes (26). This is possible because of the induction of immunosuppressive molecules, such as interleukin-10, which antagonize T-cell responses. In addition, although it enhances certain humoral responses, LMP1 expression in a B-cell-specific manner in transgenic mice prevents germinal center (GC) formation, which may shield LMP1-expressing EBV-infected cells from mounting effective T-cell-mediated responses (30).

Members of our laboratory previously demonstrated that replacement of the LMP1 CCT with the cytoplasmic region of CD40 results in a chimeric molecule (LMP1CD40) which can mediate constitutive CD40 signaling towards the NF-κB and JNK pathways in vitro, similar to LMP1 (12). In order to investigate the effects of chronic CD40 signaling in vivo, specifically in B cells, and to directly compare the immunomodulatory properties of LMP1 and CD40, we generated and studied transgenic mice expressing LMP1 or LMP1CD40 in B cells. Our experiments highlight similar effects of and important differences in LMP1 and CD40 signaling in B cells in vivo.

MATERIALS AND METHODS

Generation of plasmids and transgenic mice.

The coding sequences of LMP1 and LMP1CD40 (12) were blunted and cloned into the SmaI site of the pUC18V.5-hCD19 vector (a kind gift of N. Uyttersprot and K. Rajewsky, University of Cologne). pUC18V.5-hCD19 contains the EcoRI-KpnI fragment of the human CD19 gene, which includes the first and second exon, and it was mutagenized in order to eliminate both AUG codons of the first exon that precede the coding sequences of LMP1 and LMP1CD40. In a subsequent cloning step, the recombinant EcoRI-KpnI fragment of pUC18V.5-hCD19 was used to replace the corresponding fragment in the pSP65V7.J5 vector (a kind gift of N. Uyttersprot and K. Rajewsky), which contains additional sequences of the human CD19 gene (32). Fertilized CBA × C57BL/6 hybrid (F2) zygotes were microinjected as previously described (5). Independent lines from two founders of each transgenic mouse type backcrossed to the C57BL/6 strain six times were analyzed. All of the transgenic mice analyzed in this study were over 4 months old and were heterozygotes, and their littermates that did not carry the transgene are referred to as wild-type mice. All of the analyzed mice carrying the same transgene had the same phenotype.

Immunizations and immunohistochemical analysis of splenic structure.

Mice were injected intraperitoneally with 5 × 10⁸ sheep red blood cells (SRBC) in phosphate-buffered saline (PBS). Serum samples and spleens were collected before or 12 days after immunization. The spleens were harvested, embedded in O.C.T. compound (BDH Laboratory Supplies, Dorset, United Kingdom), frozen in liquid nitrogen, and stored at −70°C. Cryostat sections (5 μm thick) mounted on gelatinized slides were fixed in cold acetone for 10 min. The slides were rehydrated in PBS, endogenous peroxidase was quenched with 0.5% H₂O₂ in PBS for 20 min, and the sections were preincubated for 30 min with 5% fetal bovine serum (FBS) in PBS. For double immunostaining of GCs and B cells, the sections were incubated with horseradish peroxidase (HRP)-conjugated peanut agglutinin (PNA; 25 μg/ml; Sigma Chemical Co., St. Louis, Mo.) and biotinylated rat anti-mouse B220 (clone RA3-6B2; BD Pharmingen). Visualization of staining with biotinylated antibodies was performed by the use of avidin-alkaline phosphatase (Vectastain ABC kit; Vector Laboratories, Inc.) according to the manufacturer's instructions. Peroxidase activity was visualized with diaminobenzidine (Sigma Chemical Co.), and alkaline phosphatase activity was developed with naphthol AS-MX phosphate (Sigma Chemical Co.) and Fast Blue BB salt (Sigma Chemical Co.). All antibodies were diluted in PBS containing 5% FBS, and incubations were performed under humidified conditions at room temperature.

Immunofluorescence analysis of mouse IgG deposits in kidneys.

Mouse kidneys were harvested, embedded in O.C.T. compound (BDH Laboratory Supplies), frozen in liquid nitrogen, and stored at −70°C. Cryostat sections (5 μm thick) mounted on gelatinized slides were fixed in cold acetone for 10 min. The slides were rehydrated in PBS and incubated for 30 min with 5% FBS in PBS. Antibody deposits were visualized by incubation with fluorescein isothiocyanate-conjugated horse anti-mouse immunoglobulin G (IgG) (Vector Laboratories, Inc.) at room temperature for 1 h.

ELISAs for total and SRBC-specific antibody levels in sera and for antibody avidities.

Maxisorp microtiter plates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with 100 μl of solubilized membrane extract from SRBC (50 μg/ml in PBS). Serum samples were added as serial dilutions in PBS containing 0.05% Tween 20 and 1% bovine serum albumin (BSA) and were incubated at 37°C for 1 h. Subsequently, the plates were incubated with 100 μl of HRP-labeled goat anti-mouse IgG (1:5,000 in PBS containing Tween 20 and 1% BSA; Southern Biotechnology Associates, Inc.) at 37°C for 1 h. Enzyme-linked immunosorbent assays (ELISAs) were developed with 0.5 mg of o-phenyldiamine dihydrochloride (Sigma Chemical Co.)/ml in 50 mM phosphate-citrate buffer, pH 5.0, containing 0.03% H₂O₂ and then stopped with 2 M H₂SO₄, and optical densities at 490 nm from duplicate wells were measured by use of a microplate reader (MRX; Dynatech Laboratories, Inc., Chantilly, Va.). Optical density values were converted to arbitrary units by comparison to a standard titration curve according to standard approaches (7, 14, 15). A standard curve for determinations of serum antibodies in nonimmunized mice was obtained from data for one wild-type animal. A standard curve for determinations of anti-SRBC-specific antibodies in immunized mice was derived from data for a pool of sera from 10 wild-type mice that were immunized with SRBC. Preimmune sera were analyzed and had no detectable levels of anti-SRBC antibodies.

For determinations of the avidities of SRBC-specific IgG antibodies, a constant amount of each test serum, chosen to give an optical density of approximately 1.5, was added to Maxisorp microtiter plates that were precoated with a solubilized membrane extract from SRBC. After incubation at 37°C for 1 h, the plates were incubated at room temperature for 15 min with serial dilutions of chaotropic (NH₄)₂SCN followed by extensive washing with PBS containing 0.05% Tween 20. Subsequently, an ELISA was performed as described in detail above. The avidity index was calculated as the concentration of (NH₄)₂SCN that produced an optical density at 492 nm identical to that in the wells containing a twofold dilution of the constant amount of test serum (27).

Total IgM and IgG antibodies in sera were measured by ELISAs as described in detail above. Polyclonal rabbit anti-mouse IgM plus IgG (DACO) was used as a capture antibody, and HRP-labeled goat anti-mouse IgM or IgG (Southern Biotechnology Associates, Inc.) was used as a detection antibody.

ELISA for anti-dsDNA serum antibodies.

Maxisorp microtiter plates (Nunc) were coated overnight at 4°C with 100 μl of bovine thymus double-stranded DNA (dsDNA) (10 μg/ml in bicarbonate buffer; dsDNA type I; Sigma Chemical Co.). Serum samples were added as serial dilutions in PBS containing 0.05% Tween 20 and 0.5% gelatin and were incubated at 37°C for 4 h. Subsequently, the plates were incubated with 100 μl of HRP-labeled goat anti-mouse IgG (1:5,000 in PBS containing Tween 20 and 0.5% gelatin; Southern Biotechnology Associates, Inc.) at 37°C for 1 h. ELISAs were developed with 0.5 mg of o-phenyldiamine dihydrochloride (Sigma Chemical Co.)/ml as described above.

Flow cytometric analysis.

Single-cell suspensions were prepared from spleens, red blood cells were excluded by the use of Gey's lysis solution, and debris was removed with a cell strainer (40-μm pore size; BD Falcon). The cells were stained for a panel of cell markers by incubation in PBS-BSA for 15 min on ice with titrated concentrations of reagents (BD Pharmingen). The cells were analyzed on a Coulter EPICS XL-MCL flow cytometer. Cell sorting was performed with a FACS Vantage SE system (Becton Dickinson). B cells were sorted twice from splenocytes by gating at B220^high, and the purity of the sorted population exceeded 99%.

Luciferase reporter assays.

EBV-negative Burkitt's lymphoma cells (DG75) were electroporated with equal amounts of a human bcl6 promoter-driven luciferase reporter construct (pBCL657/1261Luc) (19) and a β-galactosidase expression construct to normalize transfection efficiencies in the presence or absence of the LMP1 or LMP1CD40 expression vector. Relative luciferase activities were determined at 24 h posttransfection by dividing the luciferase activity of each sample by the corresponding β-galactosidase activity as described previously (29). The results are given as means ± standard deviations (SD) of relative luciferase activities from three experiments.

Western blot analysis.

Whole-cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting as described previously. The antibodies used were rabbit anti-BCL6 (clone N-3; Santa Cruz), mouse anti-actin (ICN), mouse monoclonal anti-LMP1 amino terminus (OT22CN; a kind gift from J. Middeldorp), anti-mouse Ig-HRP (Amersham Biosciences), and anti-rabbit IgG-HRP (Southern Biotechnology Associates, Inc.).

RESULTS

Constitutive CD40 signaling increases the population of MZ B cells, macrophages, and polymorphonuclear cells in the spleen.

The expression of LMP1 and LMP1CD40 was analyzed in corresponding transgenic mice. Reverse transcription-coupled PCR (RT-PCR) analysis demonstrated the B-cell-restricted expression of LMP1 and LMP1CD40 mRNAs in the spleens of transgenic mice (Fig. 1B). LMP1 and LMP1CD40 mRNAs were expressed at similar levels, as determined by semiquantitative RT-PCR (data not shown). LMP1 or LMP1CD40 expression was absent from all other tissues examined, including the thymus, the small intestine, and the brain. To determine the protein levels of LMP1 and LMP1CD40, we subjected whole-cell extracts from sorted spleen cells of transgenic mice to Western blot analysis with a monoclonal antibody (OT22CN) that recognizes the amino terminus of LMP1 (Fig. 1C). LMP1 and LMP1CD40 were expressed at similar levels in B cells and were absent from B-cell-depleted splenocytes. Furthermore, the expression level of LMP1 in B cells of transgenic mice was comparable to that in EBV-transformed human lymphoblastoid cell lines (Fig. 1D). The effect of LMP1 and LMP1CD40 on splenic cell populations was examined by fluorescence-activated cell sorting (FACS) analysis. LMP1CD40, but not LMP1, caused an increase in the numbers of macrophages and polymorphonuclear cells (Fig. 2A). The ratio of CD4 to CD8 singly positive T cells was unaffected by LMP1 or LMP1CD40, although an increase in the percentages of these cells among splenocytes was evident for LMP1CD40-expressing transgenic mice (data not shown). Notably, LMP1CD40, but not LMP1, caused a significant increase in the population of IgM^high IgD^low cells, which consist of immature and marginal zone (MZ) B cells (Fig. 2B). An increase in the MZ B-cell population was verified by the increase in IgM^highIgD^high CD21^high and B220⁺ CD21^high CD1d^high cells in LMP1CD40 transgenic mice relative to wild-type or LMP1 transgenic mice (data not shown). In addition, mature follicular B cell levels (IgM^low IgD^high) were decreased in LMP1CD40 transgenic mice but not in LMP1 transgenic mice (Fig. 2B). An examination of cell surface markers on B lymphocytes identified an LMP1CD40-dependent and specific increase in the activation markers CD69, CD80, and CD86 (Fig. 2C). Taken together, our data indicate that constitutive CD40 signaling, unlike LMP1 signaling, increases the populations of macrophages, polymorphonuclear cells, and MZ B cells and induces a distinct set of activation markers in B cells.

FIG. 2.

FACS analysis of splenocytes derived from LMP1 and LMP1CD40 transgenic mice. Splenocytes from LMP1 and LMP1CD40 transgenic mice and their wild-type littermates were stained with anti-CD11b and anti-Ly-6G (A) or anti-IgM and anti-IgD (B) and then analyzed by FACS. (C) Splenic B lymphocytes from transgenic mice and their wild-type littermates were analyzed by FACS for expression of the activation markers CD69, CD25, CD80, and CD86. Five animals from each group were analyzed.

LMP1CD40 and LMP1 differentially affect antibody production.

A previous study demonstrated defective T-cell-dependent humoral responses in transgenic mice expressing LMP1 in B lymphocytes (30). Unlike LMP1, in nonimmunized mice LMP1CD40 caused a significant increase in total IgM levels in sera (Fig. 3A). LMP1 and LMP1CD40 did not affect the levels of total serum IgG. Upon the immunization of mice with a T-cell-dependent antigen, both LMP1 and LMP1CD40 caused a decrease in the levels of antigen-specific IgG production, indicating a defect in T-cell-dependent humoral immune responses (Fig. 3B). Furthermore, the avidities of the antigen-specific antibodies were significantly reduced in LMP1CD40-expressing mice but not in LMP1-expressing mice (Fig. 3C). These findings indicate a stronger suppression of humoral immune responses by constitutive CD40 signaling than by LMP1 signaling.

FIG. 3.

Humoral responses in LMP1 and LMP1CD40 transgenic mice. (A) Total IgM and IgG levels were measured in sera of 5-month-old mice by ELISA. Average values ± SD for sera from five animals in each group are shown. (B) Mice were immunized intraperitoneally with SRBC, and sera from five animals in each group were examined by ELISA for SRBC-specific IgG antibodies 12 days after injection. (C) Avidity indexes of serum SRBC-specific IgG antibodies were measured by ELISA 12 days after injection for mice that were immunized with SRBC. In all cases, average values ± SD for sera from five animals in each group are shown. A statistical analysis for significant differences was performed by a two-tailed unpaired Student t test (1*, P < 0.05; 2*, P < 0.001; 3*, P < 0.005 compared to the wild-type group). (D) Spleen sections from mice at 12 days postimmunization were immunostained for GCs (PNA⁺, brown [arrows]) and B lymphocytes (B220⁺, blue). A representative image of a spleen section from one of five animals per group is shown.

GC suppression by LMP1CD40 and LMP1.

The defects in T-cell-dependent humoral immune responses in LMP1CD40-expressing mice prompted an examination of the splenic structures in immunized mice. The expression of LMP1 in B cells in transgenic mice has been shown to suppress GC formation. LMP1CD40 suppressed GC formation to a larger extent than did LMP1 (Fig. 3D). The numbers and sizes of GCs in LMP1CD40-expressing mice were smaller than those of GCs in LMP1-expressing mice. No GCs were observed in unimmunized transgenic mice or wild-type littermates. B-cell areas and follicular dendritic cell networks showed a physiological organization in LMP1CD40 and LMP1 transgenic mice (data not shown). These findings are consistent with the suppression of humoral responses in the transgenic mice.

LMP1CD40 and LMP1 downregulate BCL6 expression.

BCL6 is a transcription factor which is essential for GC formation. To determine whether GC suppression by LMPCD40 and LMP1 was associated with a decrease in BCL6 expression, we determined the levels of BCL6 in B cells from wild-type littermates or transgenic mice by Western blotting. BCL6 was detected in splenic B cells from unimmunized and immunized wild-type mice. Consistent with previous reports, BCL6 levels increased upon immunization. However, the expression of LMP1CD40 or LMP1 led to a dramatic repression of the BCL6 protein in both unimmunized and immunized animals (Fig. 4A). An examination of the steady-state levels of BCL6 mRNA by RT-PCR revealed a dramatic repression of BCL6 expression by both LMP1 and LMP1CD40 in unimmunized and immunized animals (Fig. 4B). To determine whether LMP1 and LMP1CD40 can affect the transcription of BCL6 directly, we investigated the effects of LMP1 or LMP1CD40 expression on the activity of the human bcl6 promoter, which is highly similar to the mouse bcl6 promoter. LMP1 and LMP1CD40 were expressed at similar levels (data not shown) and dramatically repressed the expression of a luciferase reporter gene driven by the human bcl6 promoter (Fig. 4C) (19). Our results indicate that LMP1CD40 and LMP1 suppress GC formation at least in part by repressing bcl6 transcription directly.

FIG. 4.

Downregulation of BCL6 expression in LMP1 and LMP1CD40 transgenic mice. Splenocytes harvested prior to or after immunization with SRBC were examined for BCL6 expression by Western blot analysis with an anti-BCL6 antibody (A) or RT-PCR (B). Actin protein (A) or GAPDH mRNA (B) levels were used as controls for the protein or mRNA sample content, respectively. (C) Luciferase reporter assay. The effect of LMP1 or LMP1CD40 on the activity of a BCL6 promoter-driven luciferase reporter is shown. Average values ± SD from three transfections are shown.

LMP1CD40 expression causes spontaneous inflammatory responses and creates the potential for autoimmunity.

LMP1CD40 expression disrupted the relative abundance of B-cell populations in the spleen. Similar alterations have been associated with spontaneous autoimmune and inflammatory responses (9, 10). For this reason, LMP1CD40 and LMP1 transgenic mice were examined for signs of autoimmunity by determining the levels of anti-thymus dsDNA antibodies in sera. LMP1CD40 transgenic mice had significantly higher levels of autoantibodies against dsDNA than did the wild type, whereas LMP1 transgenic mice did not differ from their wild-type littermates (Fig. 5A). In addition, LMP1CD40 transgenic mice had IgG deposits in the kidneys, whereas LMP1 transgenic mice and their wild-type littermates did not (Fig. 5B). A histological examination of the livers of the mice revealed perivascular inflammation for LMP1CD40 transgenic mice, which was not observed for LMP1 transgenic mice or their wild-type littermates (Fig. 5C).

FIG. 5.

Autoimmune and inflammatory responses in LMP1CD40 transgenic mice. (A) Average values of anti-dsDNA IgG antibody levels in sera from five animals for each group are shown. A statistical analysis for significant differences was performed by a two-tailed unpaired Student t test (#, P < 0.005). (B) Immunofluorescent detection of IgG deposits in kidney sections from LMP1 and LMP1CD40 transgenic mice and their wild-type littermates. A representative image of a kidney section for one of five animals per group is shown. (C) Hematoxylin-eosin staining of liver sections shows perivascular inflammation for LMP1CD40 transgenic mice which is absent from LMP1 transgenic mice or their wild-type littermates. A representative image of a liver section from one of five animals per group is shown.

DISCUSSION

The functional relationship of LMP1 to the B-cell growth and differentiation factor CD40 is largely based on evidence from studies with cell culture systems, which cannot reveal the full spectrum of immunomodulatory functions of these proteins. In order to investigate and directly compare the signaling capabilities of the two molecules in vivo, we generated transgenic mice that express LMP1 or a constitutively activated CD40 molecule (LMP1CD40) in B cells. LMP1CD40 activates the CD40 pathway analogously to the LMP1 pathway. LMP1 and LMP1CD40 were expressed at similar levels, which allowed for a valid comparison of the long-term phenotypic effects of LMP1 and CD40 signal transduction in B cells. LMP1 and LMP1CD40 induced similar and distinct immunomodulatory effects. No signs of malignancy were detected in the transgenic mice up to 5 months of age.

How well does LMP1CD40 model CD40 signaling? It was previously shown that LMP1CD40 can activate the NF-κB and JNK pathways in vitro, and these pathways can be activated by CD40 as well (12). The expression of LMP1CD40 in the B lymphocytes of transgenic mice led to the upregulation of the cell surface activation markers CD69, CD80, and CD86, which are also upregulated in response to CD40 activation. LMP1CD40 suppressed BCL6 mRNA and protein levels, similar to CD40 activation (1, 8, 24). Finally, the suppression of GC formation which was caused by LMP1CD40 expression has also been observed in mice by the administration of CD40 agonists over a period of 1 week (6). Taken together, these findings indicate that LMP1CD40 constitutively activates the CD40 pathway. Furthermore, our experiments established the fusion of the cytoplasmic regions of TNFRs to the amino terminus and transmembrane domains of LMP1 as a valid approach for generating constitutively active receptor molecules and animal models of diseases associated with the hyperactivity of these receptors. The ability to express these activated receptor molecules in a tissue-specific manner permits the study of the phenotypic effects of constitutive receptor signaling in specific cell types or tissues.

Some common phenotypic features of chronic constitutive LMP1 and CD40 signaling are the suppression of GC formation and the impaired production of antibodies to T-cell-dependent antigens. Although a previous study (6) demonstrated that the long-term activation of CD40 by CD40 agonists in mice suppresses GC formation, our study clearly identifies B cells as the target of CD40 activity that alone can mediate GC suppression. It has been proposed that GC suppression by constitutive CD40 activation may serve as a regulatory mechanism for the downregulation of humoral immune responses (6). LMP1 may have usurped this aspect of CD40 signaling to avoid the environment of GCs and possibly to limit the exposure of EBV-infected B cells to T-cell recognition and negative selection procedures. Our data suggest that GC suppression is mediated by the downregulation of BCL6, which is essential for the development of GCs. Previous studies have demonstrated a negative effect of CD40 activation on BCL6 expression and the downregulation of BCL6 in EBV-infected B-lymphoma cells (1, 4, 8, 24). Our data are consistent with these findings and demonstrate for the first time that LMP1 or CD40 signaling is sufficient to suppress BCL6 expression in B cells in vivo. The suppression of BCL6 by LMP1 may also prevent an antagonistic action of BCL6 towards the transcription factor NF-κB (25), which is activated by LMP1 and plays a critical role in B-lymphocyte transformation by EBV (3). The suppression of BCL6 expression by LMP1 or CD40 signaling is most likely caused by a direct effect on the promoter of BCL6, as determined by the effect of LMP1 and LMP1CD40 on human BCL6 promoter-driven reporter gene expression. The phenotypic characteristics of the transgenic mice reported here may also have been due to a spatial and/or temporal interference of constitutive LMP1 or CD40 signaling with endogenous CD40 signaling. This may have happened due to competition between endogenous CD40 and LMP1 or LMP1CD40 for common signaling molecules such as TNFR-associated protein family members.

Unlike LMP1, a comparable expression of LMP1CD40 in B cells led to autoimmune reactions and perivascular inflammation in the liver. These phenotypes are consistent with elevated numbers of MZ B cells and polymorphonuclear cells, respectively, which are induced by LMP1CD40 but not LMP1. MZ B cells represent the fastest responding B-cell population in response to blood infections, and their population is elevated in cases of autoimmune disorders (22). Qualitative and quantitative aspects of B-cell receptor signaling primarily control the development of MZ B cells. Our data indicate that CD40 signaling can also drive B-cell development towards the MZ B-cell lineage at the expense of follicular B cells, which may explain the potential for autoimmunity that is observed in LMP1CD40 transgenic mice. The activation of CD40 by CD40 ligand overexpression has also been linked to inflammation and autoimmunity in other transgenic mouse models (16, 23). However, in these other models the effect of CD40 activation in B cells cannot be distinguished from its effect on other CD40-expressing cell populations such as macrophages, dendritic cells, or epithelial cells. The LMP1CD40 transgenic mouse generated for this study allowed for the analysis of phenotypic effects of CD40 signaling specifically in B cells. LMP1 expression was not sufficient to disturb the relative abundance of B-cell populations in the spleen and to induce autoimmune or inflammatory reactions. The elevated levels of serum IgM induced by LMP1CD40 but not LMP1 are also consistent with a potential for autoimmunity (11). These phenotypic differences between LMP1 and CD40 signaling may reflect the evolution of a viral strategy to modulate the signaling mechanism of LMP1 in a way that limits the exposure of LMP1-expressing B cells to immune surveillance.