Epstein-Barr Virus-Encoded Latent Membrane Protein 1 Activates the JNK Pathway through Its Extreme C Terminus via a Mechanism Involving TRADD and TRAF2

Abstract

The transforming Epstein-Barr virus-encoded latent membrane protein 1 (LMP1) activates signalling on the NF-κB axis through two distinct domains in its cytoplasmic C terminus, namely, CTAR1 (amino acids [aa] 187 to 231) and CTAR2 (aa 351 to 386). The ability of CTAR1 to activate NF-κB appears to be attributable to the direct interaction of tumor necrosis factor (TNF) receptor-associated factor 2 (TRAF2), while recent work indicates that CTAR2-induced NF-κB is mediated through its association with TNF receptor-associated death domain (TRADD). LMP1 expression also results in activation of the c-Jun N-terminal kinase (JNK) (also known as stress-activated protein kinase) cascade, an effect which is mediated exclusively through CTAR2 and can be dissociated from NF-κB induction. The organization and signalling components involved in LMP1-induced JNK activation are not known. In this study we have dissected the extreme C terminus of LMP1 and have identified the last 8 aa of the protein (aa 378 to 386) as being important for JNK signalling. Using a series of fine mutants in which single amino acids between codons 379 and 386 were changed to glycine, we have found that mutations of Pro³⁷⁹, Glu³⁸¹, Ser³⁸³, or Tyr³⁸⁴ diminish the ability of LMP1 CTAR2 to engage JNK signalling. Interestingly, this region was also found to be essential for CTAR2-mediated NF-κB induction and coincides with the LMP1 amino acid sequences shown to bind TRADD. Furthermore, we have found that LMP1-mediated JNK activation is synergistically augmented by low levels of TRADD expression, suggesting that this adapter protein is critical for LMP1 signalling. TRAF2 is known to associate with TRADD, and expression of a dominant-negative N-terminal deletion TRAF2 mutant was found to partially inhibit LMP1-induced JNK activation in 293 cells. In addition, the TRAF2-interacting protein A20 blocked both LMP1-induced JNK and NF-κB activation, further implicating TRAF2 in these phenomena. While expression of a kinase-inactive mutated NF-κB-inducing kinase (NIK), a mitogen-activated protein kinase kinase kinase which also associates with TRAF2, impaired LMP1 signalling on the NF-κB axis, it did not inhibit LMP1-induced JNK activation, suggesting that these two pathways may bifurcate at the level of TRAF2. These data further define a role for TRADD and TRAF2 in JNK activation and confirm that LMP1 utilizes signalling mechanisms used by the TNF receptor/CD40 family to elicit its pleiotropic activities.

Epstein-Barr virus (EBV) is a gammaherpesvirus associated with several human malignancies, including Burkitt’s lymphoma, lymphoproliferative disorders in immunocompromised individuals, Hodgkin’s disease, and nasopharyngeal carcinoma (30, 60). EBV infects normal resting B cells and induces their transformation into lymphoblastoid cell lines through the coordinate expression of a number of latent cycle proteins, including six nuclear proteins (EBNA 1, 2, 3A, 3B, and 3C and LP) and three membrane proteins (latent membrane protein 1 [LMP1], LMP2A, and LMP2B) (30).

Among these viral gene products, LMP1 has been a focus of interest due to its ability to growth transform certain rodent fibroblast cell lines (1, 56). Recombinant genetic analysis of the EBV genome has demonstrated that LMP1 is also essential for EBV-mediated immortalization of B cells (29, 30). Many of the phenotypic characteristics of lymphoblastoid cell lines have been attributed to the pleiotropic effects of LMP1 expression. Thus, transient overexpression of LMP1 in normal resting B cells induces DNA synthesis (44). In addition, LMP1 expression in B cells prevents cell death through the activation of a number of antiapoptotic proteins such as Bcl-2, Bcl-xL, Mcl-1, and A20 and induces up-regulation of B-cell activation markers such as CD23 and CD40, cell adhesion molecules such as ICAM1, LFA1, and LFA3, and cytokine production (19, 33, 50, 59). Expression of this viral protein in epithelial cells also results in phenotypic changes, induction of A20, and cytokine production and blocks differentiation, a property which may be important in the pathogenesis of nasopharyngeal carcinoma (5, 9, 11, 14, 39). Paradoxically, ectopic overexpression of LMP1 in certain established cell lines of epithelial or B-cell origin results in cytostatic or cytotoxic effects (7, 13, 17, 36).

Structurally, LMP1 is a 63-kDa phosphoprotein comprising a short 23-amino-acid (aa) N-terminal cytoplasmic domain, six putative membrane-spanning domains of 162 aa, and a long 200-aa C-terminal cytoplasmic tail. Mutational analysis has identified both transmembrane and C-terminal regions of the protein as being necessary for transformation and phenotypic changes (29, 41). Recent studies have demonstrated that LMP1 function requires oligomerization at the plasma membrane and emphasize the importance of the transmembrane segments in this process (12, 15). The 45 most-membrane-proximal residues of the LMP1 C terminus (aa 186 to 231) are critical for EBV-mediated transformation of primary B cells, but the long-term growth of these cells also requires the distal C-terminal sequences (aa 352 to 386) (25, 29).

Interestingly, these two functional domains of the LMP1 cytoplasmic tail can independently activate the transcription factor NF-κB (20, 24, 40). Transient assays with LMP1 deletion mutants have identified the extreme C-terminal activating region 2 (CTAR2) (aa 351 to 386) as the principal contributor to LMP1-induced NF-κB in the majority of cell lines, being responsible for 70 to 80% of the total NF-κB activation by LMP1. The proximal CTAR1 domain (aa 187 to 231) induces low levels of NF-κB, an effect which could be attributed to its ability to interact with tumor necrosis factor receptor (TNFR)-associated factors (TRAFs) through a P²⁰⁴xQ²⁰⁶xT²⁰⁸ motif important for association. Thus, TRAF1 and TRAF3 strongly bind to CTAR1, but TRAF2 interacts only weakly (6, 28, 42, 52). TRAF5 has also been reported to associate with CTAR1 (2).

The TRAF proteins have recently attracted much attention as important mediators of signal transduction induced upon activation of various members of the TNFR superfamily, including CD40, TNFRI, and TNFRII. TRAF2 and TRAF5 are of particular interest because their transient overexpression has been shown to activate NF-κB (47). A role for TRAF2 in LMP1-induced NF-κB activation through CTAR1 has been demonstrated by using a dominant-negative N-terminal-deletion TRAF2 mutant which blocks CTAR1-mediated NF-κB (6, 9, 28). This mutated TRAF2 is also able to partially inhibit CTAR2-induced NF-κB activation, and a similar although more potent effect has been noted following expression of TRAF2-interacting proteins such as TANK/I-TRAF (28) and A20 (9). Unlike CTAR1, the CTAR2 domain of LMP1 does not directly associate with TRAFs. However, a recent study has demonstrated that CTAR2 binds the TNFR-associated death domain (TRADD) protein and that this interaction may account for the ability of CTAR2 to mediate NF-κB activation (25).

In addition to NF-κB, LMP1 expression signals for activation of the c-Jun N-terminal kinase (JNK) (also known as the stress-activated protein kinase) pathway, a phenomenon which is mediated through CTAR2 but not CTAR1 and translates to the induction of the transcription factor AP-1 (10, 18, 31). While the kinetics of LMP1-mediated JNK and NF-κB activation appear to be identical, these two pathways can be dissociated. Thus, inhibition of NF-κB by a constitutively active mutated IκBα does not impair the ability of LMP1 to signal on the JNK axis, and, conversely, expression of a dominant-negative SEK (JNKK) blocks LMP1-induced JNK but not NF-κB activation (10). Engagement of the JNK cascade also occurs upon stimulation of CD40, TNFRI, and TNFRII, an effect which is mediated via a TRAF2-dependent mechanism (43, 45, 51). Furthermore, TRAF2 has been shown to be the TNFRII-associated protein where bifurcation of the JNK and NF-κB pathways occurs (54).

The signalling components involved in LMP1-induced JNK activation are not known. In this study we have dissected the extreme C terminus of LMP1 and have identified the last 8 aa of the protein as being important for JNK signalling. This region was also found to be essential for association of LMP1 with TRADD and for CTAR2-mediated NF-κB activation. Furthermore, our data provide evidence for a role for TRADD and TRAF2 in this LMP1-activated pathway.

MATERIALS AND METHODS

DNA constructs.

pSG5-based LMP1 and LMP1 deletion mutants Δ(194-386), Δ(332-386), and Δ(187-351) have been previously described (24). pSG5LMP1^AxAxA was generated by site-directed mutagenesis with the QuickChange site-directed mutagenesis kit (Stratagene) and pSG5-LMP1 as the substrate. The mutated oligonucleotide primers used were 5′-CCTCCCGCACGCTCAAGCAGCTGCCGATGA-3′ and its complement. pSG5-LMP1Δ(187-351)/378STOP and pSG5LMP1^AxAxA/378STOP were generated by a similar approach with the mutated primers 5′-GATGACGACCCCCACTGACCAGTTCAGCTAAGC-3′ and its complement, with pSG5-LMP1Δ(187-351) and pSG5LMP1^AxAxA as substrates, respectively. These mutations generate a stop codon at position 378 of the amino acid sequence of LMP1 and were verified by sequencing. pSG5CD2.192-LMP1 and CTAR2 fine mutants have been previously described (12).

The GAL4 DNA binding domain fusions were constructed by PCR-mediated amplification of LMP1 cDNA fragments with primers with artificial EcoRI and BamHI sites, digestion with EcoRI and BamHI, and in-frame cloning into plasmid pGBT9 (Clontech). The primers used were 5′-AGTGATGAATTCCACCACGAT-3′ (LE/F) and 5′-GCTGCGGATCCTTAGTCATAGTA-3′ (LB/R) for the amplification of the LMP1 C terminus [aa 194 to 386, construct GALbd-LMP1(194-386)] and 5′-AGTCATGAATTCGGCCATGGC-3′ (C2E/F) and LB/R for the amplification of CTAR2 [aa 351 to 386, construct GALbd-LMP1(351-386)]. Construct GALbd-LMP1(194-345) was generated from GALbd-LMP1(194-386) by inserting a stop codon at position 345 of the amino acid sequence of LMP1 by site-directed mutagenesis. The primers used were 5′-GACAGACGGAGGCGGCTGACATAGTCATGATTCCG-3′ and its complement. Constructs GALbd-LMP1(194-386)/378STOP and GALbd-LMP1(351-386)/378STOP were generated by introduction of a stop codon at position 378 of the amino acid sequence of LMP1 by site-directed mutagenesis. All cloned LMP1 fragments were sequenced, and expression was verified by Western blot analysis. TRADD cDNA was cloned into the activation domain plasmid pGAD424 as an EcoRI/SalI fragment.

The hemagglutinin (HA)-p46SAPKγ-pcDNA3 vector was a gift from James Woodgett (The Ontario Cancer Institute, Ontario, Canada), and the wild-type and mutated TRAF2 expression vectors pcDNA3-TRAF2 and pcDNA3-TRAF2Δ(6-86) were kindly provided by Ken Kaye and Eliott Kieff (Harvard Medical School, Boston, Mass.), respectively. The A20 expression vector pSFFV-A20 and anti-A20 mouse monoclonal antibody (MAb) were a generous gift from Vishva Dixit (University of Michigan, Ann Arbor). Kinase-inactive NF-κB-inducing kinase (NIK) [pcDNA3-NIK(KK429-430AA)] and pRK-TRADD were kindly provided by David Wallach (Weizmann Institute of Science, Rehovot, Israel) and David Goeddel (Tularik, San Francisco, Calif.), respectively. The CD40 expression vector pcDNA3-CD40 has been previously described (10).

Cell lines, transfections, and reporter assays.

Human embryonic kidney (HEK) 293 and COS-1 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum and 2 mM glutamine. For transient transfections, 8 × 10⁵ HEK 293 cells were plated out on a 25-cm² flask and the following day were transfected by using a standard calcium phosphate technique. Transient transfections in COS cells were performed by using a modification of the DEAE-dextran method which gives high transfection efficiency.

Luciferase reporter and β-galactosidase assays were performed as previously described (9). Fifty nanograms each of a cytomegalovirus (CMV)-driven β-galactosidase expression plasmid and of the 3Enh.κBconA-Luc reporter, which contains three tandem repeats of the NF-κB sites from the IgGκ promoter, were routinely used to transfect 293 cells. Analysis of luciferase and β-galactosidase expression was always performed at 36 h posttransfection.

Oligomerization of CD2.

Oligomerization of CD2/LMP1 chimeric proteins was induced by the addition of 5% culture supernatant of the OX34 hybridoma (27), which produces mouse anti-CD2 MAbs, and of 1:100 polyclonal anti-mouse immunoglobulins (Dako Z0259) which had been dialyzed to remove azide. The same OX34 supernatant was used for all of the experiments performed. As a control, 293 cells were treated with supernatant from the G28.5 hybridoma, which produces anti-CD40 MAbs.

Yeast two-hybrid assays.

pGBT9 or pGBT9/LMP1 cytoplasmic domain fusion proteins and pGAD424 or pGAD424/TRADD hybrids were cotransformed into Saccharomyces cerevisiae Y190, and positive interactions were identified by β-galactosidase filter assays according to the instructions of the manufacturer (Clontech).

Immunoprecipitations, kinase assays, and immunoblotting.

Following stimulation or transfection, cells were lysed in 300 to 500 μl of kinase lysis buffer (20 mM Tris [pH 7.6], 0.5% Triton X-100, 250 mM NaCl, 3 mM EGTA, 3 mM EDTA, 2 mM sodium vanadate, 10 μg of aprotinin per ml, 10 μg of leupeptin per ml, and 1 mM dithiothreitol) for 15 min on ice. Cell debris was removed by centrifugation, and the protein concentration was determined with a commercially available (Bio-Rad) protein assay. JNKs were immunoprecipitated from 250 μg of total protein extracts by using 1 μg of anti-HA antibody (Boehringer) and 25 μl of protein G-Sepharose (Pharmacia) for 2 to 3 h. Following immunoprecipitation, beads were washed once with kinase lysis buffer and twice with assay buffer (20 mM HEPES [pH 7.5], 20 mM β-glycerophosphate, 10 mM MgCl₂, 1 mM dithiothreitol, 50 μM sodium vanadate, and 1 μg of leupeptin per ml). After the last wash, the beads were drained with a fine-gauge Hamilton syringe and resuspended in 40 μl of assay buffer containing 1 μg of glutathione S-transferase (GST)–c-Jun (aa 1 to 79) substrate (Stratagene), 20 μM cold ATP, and 3 μCi of [γ-³²P]ATP. Kinase reactions were carried out at 30°C for 30 min and stopped by addition of 40 μl of 6× Laemmli buffer and boiling for 5 min. Samples were then analyzed on a sodium dodecyl sulfate–10% polyacrylamide gel. The gel was then stained with Coomassie blue to verify that equal amounts of substrate were used, and autoradiography and phosphorimaging were performed after the gels were dried. Immunoblot analysis of anti-HA immunoprecipitates with a JNK-specific antibody was also performed to demonstrate that comparable amounts of HA-p46SAPKγ-pcDNA3 were analyzed in cotransfection experiments.

For LMP1 and TRAF2 immunoblotting, 20 μg of total cell lysates isolated as described above was analyzed on a 10% gel, and LMP1 or TRAF2 expression was detected with the anti-LMP1 MAbs CS.1 to -4 (49) or the TRAF2(C-20) polyclonal antibody (Santa Cruz) and enhanced chemiluminescence (Amersham). For A20 immunoblotting, 100 μg of lysates was analyzed on a 7.5% gel; immune complexes were then detected by using enhanced chemiluminescence.

RESULTS

The last 8 aa of LMP1 are critical for JNK activation.

We have recently demonstrated that LMP1-mediated JNK activation is mediated exclusively through the CTAR2 domain (aa 351 to 386) of the protein (10). As CTAR2 also activates NF-κB, we have dissected the extreme C terminus of LMP1 to identify whether these two activities colocalize to the same region of the molecule. For this purpose, LMP1 deletion mutants lacking the terminal 8 aa were constructed. Thus, by using site-directed mutagenesis, a stop codon was introduced at aa 378 of the simian virus 40-driven pSG5-LMP1Δ(187-351) plasmid, which expresses a CTAR1 deletion version of LMP1, and of pSG5-LMP1^AxAxA, which contains a triple P²⁰⁴xQ²⁰⁶xT²⁰⁸→AxAxA mutation (Fig. 1A). This triple mutation has been previously shown to block CTAR1-mediated NF-κB by abrogating TRAF binding to this LMP1 domain (6, 9).

FIG. 1.

The last 8 aa of LMP1 are critical for JNK and NF-κB signalling. (A) Schematic representation of the LMP1 protein and the deleted LMP1 gene sequences used in this study. Solid black lines represent wild-type (wt) LMP1 sequences, and dotted lines denote deleted LMP1 sequences. CTAR1 is located at residues 194 to 232, and CTAR2 is located at residues 351 to 386. The asterisks represent a triple P²⁰⁴xQ²⁰⁶xT²⁰⁸→AxAxA mutation. (B) Induction of NF-κB-dependent transcriptional activity by LMP1 and LMP1 deletion mutants. HEK 293 cells were transfected with 1 μg of pSG5-based constructs in the presence of 50 ng of NF-κB-regulated luciferase reporter plasmid 3Enh.κBconA-Luc and 50 ng of β-galactosidase expression vector. Relative luciferase values (RLV), which represent the luciferase values normalized on the basis of β-galactosidase expression, were determined at 36 h posttransfection. The data shown represent fold increases in RLV relative to the vector control (vec), which was given the arbitrary value of 1, and are representative of at least five independent experiments. (C) Effects of wild-type and mutated LMP1 expression on JNK activity. HEK 293 cells were transfected with 1 μg of pSG5 or pSG5-based LMP1 expression vectors in the presence of 0.5 μg of the HA-tagged JNK1 expression vector HA-p46SAPKγ-pCDNA3. At 36 h posttransfection, HA-JNK was immunoprecipitated from 250 μg of cell lysates by using anti-HA antibody, and kinase assays were performed as described in Materials and Methods. JNK activity was assessed by the ability of the immunoprecipitate to phosphorylate GST–c-Jun substrate. Results of a representative assay are shown (second panel). The same lysates were analyzed for wild-type or mutated LMP1 (upper panel) and JNK (third panel) expression. Numbers on the left are molecular weights in thousands. Relative levels of JNK activation were quantitated on a phosphorimager and are presented in histogram form (lower panel). At least four independent experiments were performed and gave similar results.

The effects of these mutants on NF-κB and JNK activation were analyzed in transiently transfected HEK 293 cells. The data were compared to those obtained following expression of wild-type LMP1 or deletion mutants lacking the entire cytoplasmic tail [LMP1Δ(194-386)], the CTAR2 region [LMP1Δ(332-386)], or CTAR1 [LMP1Δ(187-351)] (Fig. 1A); these mutated LMP1 constructs have been previously shown to differentially affect NF-κB and JNK signalling (10, 24, 28). By using luciferase reporter assays, it was found that expression of LMP1Δ(187-351) or LMP1^AxAxA induced comparable NF-κB levels, which were approximately 75% of the wild-type LMP1 activity (Fig. 1B). Transfection of the CTAR1-containing pSG5-LMP1Δ(332-386) plasmid had only a small effect on NF-κB, while as previously documented, deletion of the entire cytoplasmic tail completely abrogated the effect (8, 24). Interestingly, removal of the last 8 aa of LMP1 abolished the ability of both LMP1Δ(187-351) and LMP1^AxAxA to signal on the NF-κB axis (Fig. 1B).

In order to determine whether the same 8-aa sequence contributes to JNK activation, HEK 293 cells were transiently transfected with 1 μg of pSG5 or pSG5-based LMP1-expressing constructs in the presence of 0.5 μg of HA-p46SAPKγ-pcDNA3, a CMV-driven HA-tagged JNK expression vector. Lysates from transfected cells were isolated at 36 h posttransfection, immunoprecipitated with an HA-specific antibody, and assayed for kinase activity in an immune complex kinase assay with a GST–c-Jun (aa 1 to 79) fusion protein as the substrate. As shown in Fig. 1C, expression of LMP1Δ(187-351) or LMP1^AxAxA induced JNK activity to wild-type levels. However, removal of the last 8 aa from these mutated LMP1 sequences impaired their ability to signal for JNK activation. Transfection of CTAR1 [LMP1Δ(332-386)] or C terminus deletion LMP1 [LMP1Δ(194-386)] had no significant effect, in agreement with previous reports (8, 31). Immunoblot analysis with MAbs CS.1- to 4 (49) was also performed to confirm LMP1 expression from these plasmids (Fig. 1C). As has been documented previously (24), the CTAR1 deletion [Δ(187-351)] constructs were not detectable by immunoblotting, but expression was confirmed by immunofluorescence staining with the CS.1 MAb (data not shown).

To determine whether specific amino acids are responsible for JNK and NF-κB activation by the extreme C terminus of CTAR2, we next used a series of fine mutants in which single amino acids between codons 379 and 386 of LMP1 were changed to glycine (Fig. 2A). These pSG5-LMP1Δ(187-351)-based constructs were first transfected into HEK 293 cells together with an NF-κB-dependent luciferase reporter plasmid (3Enh.κBconA-Luc) and a β-galactosidase expression vector, and relative NF-κB activity was determined. As shown in Fig. 2B, mutations of Pro³⁷⁹, Glu³⁸¹, Ser³⁸³, or Tyr³⁸⁴ to glycine diminished CTAR2-mediated NF-κB activation to almost-background levels, in agreement with a previous report for B-cell lines (12). Expression of a Val³⁸⁰- or Asp³⁸⁶-to-glycine mutant had only a minimal effect on CTAR2-mediated NF-κB activity.

FIG. 2.

Single point mutations within aa 379 to 385 severely impair CTAR2-mediated NF-κB and JNK activation. (A) Schematic representation of LMP1 CTAR2 mutants with single amino acid substitutions. (B) Activation of NF-κB in HEK 293 cells by the CTAR2 fine mutants described in panel A. Relative luciferase values (RLV) (fold increase) of full-length LMP1 (bar 2), LMP1Δ(187-351) (bar 3), and LMP1Δ(187-351)/378 STOP (bar 4) were also determined and are shown for comparison. The results shown are representative of those from three independent experiments. Bars correspond to those in panel C. (C) Activation of JNK signalling by fine CTAR2 mutants. HEK 293 cells were transfected with the pSG5-based constructs described in panel A in the presence of 0.5 μg of p46SAPKγ-pcDNA3, and JNK activity was determined by immune complex kinase assays with GST–c-Jun as the substrate (upper panel). Immunoblot analysis of anti-HA immunoprecipitates with a JNK-specific antibody was also performed to demonstrate that comparable amounts of HA-p46SAPKγ-pcDNA3 were analyzed in cotransfection experiments (middle panel). Levels of GST–c-Jun phosphorylation were quantitated on a phosphorimager. The data shown represent fold increases in JNK activation relative to the vector control (vec), which was given the arbitrary value of 1. Three independent experiments were performed and gave similar results. Consistent levels of LMP1 expression were verified by immunofluorescence staining (data not shown). wt, wild type.

These mutated constructs were subsequently examined for their ability to activate the JNK pathway. HEK 293 cells were transfected with 1 μg of LMP1Δ(187-351)-based mutants together with 0.5 μg of HA-p46SAPKγ-pcDNA3, and lysates were isolated, immunoprecipitated with an HA-specific antibody, and assayed for kinase activity with GST–c-Jun (aa 1 to 79) as the substrate. These fine-mapping experiments demonstrated that single amino acid mutations which abrogate CTAR2-mediated NF-κB also abolish its ability to signal on the JNK axis (Fig. 2C). Thus, both NF-κB and JNK signals from CTAR2 require overlapping extreme C-terminal LMP1 sequences.

TRADD is a mediator of JNK activity from the CTAR2 domain of LMP1.

Recent evidence demonstrates that TRADD specifically interacts with CTAR2 and mediates NF-κB activation from this region (25). Using yeast two-hybrid assays, we have confirmed this association and found that deletion of the last 8 aa of LMP1 abrogates its ability to interact with TRADD (data not shown). Thus, this extreme C-terminal sequence of LMP1 appears to be critical for TRADD binding as well as activation of the NF-κB and JNK pathways, suggesting that TRADD is a mediator of these signals from the CTAR2 region of LMP1.

In order to demonstrate that this death domain protein is involved in JNK activation by LMP1 and in the absence of dominant-negative TRADD mutants (22, 23), we have examined the ability of TRADD to augment LMP1 signals on the JNK axis. For this purpose, we have taken advantage of the ability of LMP1 to act as a constitutive receptor (12, 15). Thus, a chimeric molecule comprising the extracellular and transmembrane domains of CD2 (aa 1 to 212) linked to the cytoplasmic C terminus of LMP1 (aa 192 to 386) (construct pSG5 CD2.192-LMP1 [Fig. 3A]) can activate signalling on the JNK axis only following CD2 engagement (Fig. 3B and C). As shown in Fig. 3D, treatment of CD2.192-LMP1- but not vector control-transfected 293 cells with OX34 anti-CD2 MAb and cross-linking anti-mouse immunoglobulin G (IgG) induces JNK activation in a time-dependent manner. A small increase in JNK activity observed in extracts from untreated CD2.192-LMP1-expressing cultures could be attributed to spontaneous aggregation following ectopic expression of this chimeric protein. Immunoblot analysis of transfected cells with the anti-LMP1 MAbs CS.1-4 revealed at least two molecular weight species which probably represent differential glycosylation of the CD2 domain (Fig. 3D) (12).

FIG. 3.

LMP1-mediated JNK activation requires oligomerization of its cytoplasmic C terminus and is synergistically augmented by TRADD expression. (A) Schematic representation of LMP1 (left panel) and a chimera (CD2.192-LMP1) comprising the extracellular and transmembrane domains of CD2 fused to the cytoplasmic terminus of LMP1 (right panel). (B) Schematic representation of the plasma membrane localization of LMP1 and the CD2.192-LMP1 chimera. LMP1 spontaneously forms functional homo-oligomers, while CD2.192-LMP1 chimeric molecules are distributed on the cell membrane essentially as inactive monomers. (C) Following CD2 cross-linking with OX34 anti-CD2 MAb and IgG, the CD2.192-LMP1 chimera aggregates on the cell membrane and forms oligomers, thereby mimicking the constitutive aggregation of LMP1. (D) The CD2.192-LMP1 chimera activates the JNK pathway following receptor aggregation. HEK 293 cells were transiently transfected with 1 μg of pSG5CD2.192-LMP1 or control vector in the presence of 0.5 μg of p46SAPKγ-pcDNA3 and 36 h later were treated with OX34 anti-CD2 MAb and cross-linking mouse IgG for 0, 0.5, 1, 2, or 6 h. Cell lysates were then analyzed for LMP1 expression by immunoblotting (upper panel) and for JNK activity with immune complex kinase assays and GST–c-Jun as the substrate (middle and lower panels). Numbers on the left are molecular weights in thousands. vec, vector. (E) LMP1-mediated JNK activation is synergistically augmented by TRADD expression. HEK 293 cells were transfected with CD2.192-LMP1 or control vector and p46SAPKγ-pcDNA3 as described above, in the presence of 1 μg of crmA expression vector and increasing concentrations of pRK-TRADD (0, 0.1, 0.25, or 0.5 μg). Twenty-four hours later, cells were treated with OX34 and IgG for 2 h before being analyzed for LMP1 and TRADD expression (upper two panels) and JNK activity (third panel). JNK protein levels from HA immunoprecipitates were also determined (fourth panel). JNK activities were quantitated on a phosphorimager, and results are depicted in histogram form (lower panel). The data shown represent fold increases in JNK activation relative to the untreated control (bar 1), which was given the arbitrary value of 1. At least two independent experiments were performed and gave similar results.

To determine whether TRADD and LMP1 coactivate JNK, HEK 293 cells were transfected with 1 μg of CD2.192-LMP1 or control vector and increasing amounts (0, 0.1, 0.25, or 0.5 μg) of pRK-TRADD, a CMV-driven Myc-tagged TRADD expression vector, together with 0.5 μg of HA-p46SAPKγ-pcDNA3. These amounts of pRK-TRADD were chosen on the basis of optimal activation of NF-κB and low frequency of apoptosis induction (references 23 and 25 and our unpublished data). In addition, these experiments were performed in the presence of the cowpox virus gene crmA, which has been shown to protect against the preapoptotic effects of TRADD (23).

Following OX34 and IgG treatment, a 6.5-fold increase in JNK activity was observed in CD2.192-LMP1-transfected versus control vector-transfected 293 cells. This effect was significantly potentiated in the presence of increasing amounts of TRADD, resulting in a maximal 14.8-fold increase in JNK activity in the presence of 0.5 μg of pRK-TRADD (Fig. 3E). TRADD expression alone had no significant effect on JNK levels, in agreement with a previous report (35). Immunoblot analysis was performed to verify LMP1 and TRADD expression in transfected cells (Fig. 3E). The observed synergy between TRADD and LMP1 in JNK activation, in the absence of an effect by TRADD alone, provides direct evidence for a functional role for TRADD in this phenomenon.

Involvement of TRAF2 in LMP1-induced JNK signalling.

In view of the ability of TRADD to recruit TRAF2 upon TNFRI oligomerization and the central role of TRAF2 in TNF-induced NF-κB and JNK signalling, we next examined the contribution of this molecule to LMP1-mediated JNK activation. For this purpose, a dominant-negative N-terminal deletion form of TRAF2 [TRAF2Δ(6-86)] was used (28); such TRAF2 mutants have been previously shown to inhibit LMP1-induced NF-κB as well as NF-κB and JNK activation mediated by various members of the TNFRI superfamily (9, 23, 28, 43, 45, 47).

Thus, to determine the effects of dominant-negative TRAF2 on LMP1-induced JNK, CMV-driven TRAF2Δ(6-86) was overexpressed in HEK 293 cells, together with pSG5-CD2.192-LMP1 and HA-p46SAPKγ-pcDNA3. Following OX34 and IgG treatment, lysates were subjected to immune complex kinase assays with GST–c-Jun (aa 1 to 79) as the substrate. These experiments showed that expression of the dominant-negative TRAF2 mutant partially reduced LMP1-mediated JNK activity, inducing a maximum 40% decrease in c-Jun phosphorylation in the presence of 2.5 μg of TRAF2Δ(6-86) (Fig. 4A). Transfection of TRAF2Δ(6-86) alone had no effect on JNK levels. CD2.192-LMP1 and dominant-negative TRAF2 expression was verified by immunoblotting (Fig. 4A).

FIG. 4.

Involvement of TRAF2 in LMP1-mediated JNK activation. (A) The effects of dominant-negative N-terminally deleted TRAF2 [TRAF2Δ(6-86)] on LMP1-mediated JNK activation were determined by using the CD2.192-LMP1 chimera. HEK 293 cells were transiently transfected with 1 μg of CD2.192-LMP1 or control vector (vec) in the presence of increasing amounts of TRAF2Δ(6-86) (0, 1, or 2.5 μg) and 0.5 μg of p46SAPKγ-pcDNA3 and 36 h later were treated for 2 h with OX34 and IgG before being analyzed for LMP1 and mutant TRAF2 expression (upper two panels), JNK activity with GST–c-Jun as the substrate (third panel) and HA-JNK levels (fourth panel). Numbers on the left are molecular weights in thousands. GST–c-Jun phosphorylation levels were quantitated on a phosphorimager, and results are depicted in histogram form (lower panel). Data shown represent fold increases in JNK activation relative to the untreated control (bar 1), which was given the arbitrary value of 1. Three independent experiments were performed and gave similar results. TRAF2Δ(6-86) conferred only a partial inhibition of LMP1-induced JNK activation. (B) Expression of dominant-negative TRAF2 mutant abrogates TNF-mediated JNK activation. HEK 293 cells transiently transfected with p46SAPKγ-pcDNA3 and TRAF2Δ(6-86) as described above were treated for 30 min with 15 ng of TNF-α per ml before being analyzed for mutant TRAF2 expression (upper panel), JNK activity (second panel), and JNK levels (third panel). JNK activities were quantitated on a phosphorimager, and results are depicted in histogram form (lower panel), with the untreated control (bar 1) given the arbitrary value of 1. Three independent experiments were performed and gave similar results. (C) Effect of TRAF2Δ(6-86) on JNK activation mediated by expression of full-length LMP1. HEK 293 cells were transiently transfected with 1 μg of pSG5-LMP1 or control vector in the presence of increasing amounts of TRAF2Δ(6-86) (0, 1, 2.5, or 5 μg) and 0.5 μg of p46SAPKγ-pcDNA3 and 36 h later were analyzed for LMP1 and mutant TRAF2 expression (upper two panels), JNK activity with GST–c-Jun (aa 1 to 79) as the substrate (third panel), and HA-JNK levels (fourth panel). GST–c-Jun phosphorylation levels were quantitated on a phosphorimager, and results are depicted in histogram form (lower panel). Mutant TRAF2 expression was detected at 1 μg following a longer exposure of the film. At least five independent experiments were performed and gave similar results. TRAF2Δ(6-86) conferred only a partial (40 to 60%) inhibition of full-length LMP1-induced JNK activation. (D) Unlike LMP1, TRAF2Δ(6-86) induces a dramatic decrease in CD40-mediated JNK activation. HEK 293 cells were transiently transfected with 1 μg of pcDNA3-CD40 or control vector in the presence of increasing amounts of TRAF2Δ(6-86) (0, 1, or 2.5 μg) and 0.5 μg of p46SAPKγ-pcDNA3, and 36 h later cell lysates were analyzed for CD40 and mutant TRAF2 expression (upper two panels), JNK activity (third panel), and JNK expression in HA immunoprecipitates (fourth panel). Kinase activities were quantitated on a phosphorimager, and results are depicted in histogram form (lower panel). Data are representative of those from two independent experiments.

As a control for this experiment, the effect of TRAF2Δ(6-86) expression on TNF-α-mediated JNK activity was determined. For this purpose, HEK 293 cells, which express only TNFRI (47), were transiently transfected with various amounts (0, 1, or 2.5 μg) of TRAF2Δ(6-86) in the presence of 0.5 μg of HA-p46SAPKγ-pcDNA3. At 36 h posttransfection, cells were left untreated or stimulated with 15 ng of TNF-α per ml for 30 min before being analyzed for JNK activity. It was found that TRAF2Δ(6-86) expression conferred a potent, concentration-dependent inhibitory effect on TNF-α-mediated JNK activity. Thus, unlike the case for CD2.192-LMP1, transfection with 2.5 μg of TRAF2Δ(6-86) resulted in a complete block of c-Jun phosphorylation (Fig. 4B).

In order to demonstrate that the observed partial effect of dominant-negative TRAF2 on LMP1-induced JNK is genuine and not particular to the chimeric CD2 molecule used, we transfected HEK 293 cells with wild-type pSG5-LMP1 in the presence of increasing amounts (0, 1, 2.5, or 5 μg) of TRAF2Δ(6-86) and 0.5 μg of HA-p46SAPKγ-pcDNA3. At 36 h posttransfection, lysates were subjected to immune complex kinase assays with GST–c-Jun (aa 1 to 79) as the substrate. Again, it was found that expression of dominant-negative mutated TRAF2 confers only partial (40 to 60%) inhibition of LMP1-mediated JNK activity (Fig. 4C), and similar results were obtained with the P²⁰⁴xQ²⁰⁶xT²⁰⁸→AxAxA mutated LMP1-expressing pSG5-LMP1^AxAxA construct (data not shown). LMP1 and mutated TRAF2 expression was verified by immunoblotting (Fig. 4C). Expression of TRAF2Δ(6-86) in COS-1 cells gave similar levels of inhibition of LMP1-mediated JNK (data not shown).

Interestingly, in contrast to the case for LMP1, dominant-negative TRAF2Δ(6-86) had a profound effect on CD40-mediated JNK. Ectopic expression of CD40 readily induces signalling cascades including JNK activation, presumably as a result of spontaneous aggregation of CD40 monomers on the cell membrane, a phenomenon which mimics the effects of CD40 ligation (10, 47, 51). In agreement with these data, transient transfection of HEK 293 cells with 1 μg of pcDNA3-CD40, a CMV-driven CD40-expression vector, induced an 8.4-fold increase in JNK activity (Fig. 4D). In the presence of 1 or 2.5 μg of TRAF2Δ(6-86) this effect was significantly reduced, by 65 and 85%, respectively (Fig. 4D). Immunoblot analysis of the same lysates verified mutated TRAF2 and CD40 expression in the transfected cells (Fig. 4D). In these assays, CD40 appeared as a broad band, which is probably due to CD40 glycosylation.

Overall, these experiments demonstrate the involvement of TRAF2 in LMP1-mediated JNK activation. Interestingly, however, dominant-negative TRAF2 has only a partial effect on LMP1-induced JNK, compared to an almost complete inhibition of this signalling pathway induced by TNF-α treatment or CD40 expression. Similarly, while TRAF2Δ(6-86) overexpression abolishes CTAR1-mediated NF-κB in HEK 293 cells, it confers only partial inhibition of CTAR2-activated NF-κB signalling (28).

A20 blocks both LMP1-induced NF-κB and JNK activation.

Recent studies using yeast two-hybrid and functional analyses have demonstrated that the antiapoptotic protein A20 interacts with TRAF2 and blocks TNF-induced signalling (26, 55). Our previous work has shown that A20 overexpression in simian virus 40-transformed keratinocytes inhibits NF-κB activation from both the CTAR1 and CTAR2 domains of LMP1 (9). To examine whether a similar phenomenon can be observed in HEK 293 cells, 1 μg of pSG5-LMP1, pSG5-LMP1Δ(332-386), or pSG5-LMP1^AxAxA was cotransfected with various amounts of A20 (0, 0.1, 0.25, 0.5, or 1 μg) in the presence of NF-κB-driven luciferase reporter and β-galactosidase expression plasmids. As shown in Fig. 5A, A20 significantly inhibits wild-type, CTAR1-, and CTAR2-mediated NF-κB in 293 cells. Thus, transfection of 0.5 μg of A20 decreased wild-type LMP1-induced NF-κB activation by 56%, and an even-more-pronounced 85% decrease was observed in the presence of 1 μg of A20 expression vector. The effects of A20 on CTAR2-mediated NF-κB activation were of a similar degree (Fig. 5A).

FIG. 5.

The zinc finger, TRAF2-interacting protein A20 potently inhibits both LMP1-induced NF-κB and JNK activation. (A) Effects of A20 expression on NF-κB activity induced by wild-type LMP1, CTAR1, [LMP1Δ(332-386)], and CTAR2 effector (LMP1^AxAxA). HEK 293 cells were transiently transfected with NF-κB-driven luciferase reporter and β-galactosidase expression plasmids and with 1 μg of pSG5 or pSG5-based LMP1-expressing constructs in the presence of increasing amounts of A20 (0, 0.1, 0.25, 0.5, or 1 μg). Relative luciferase values (RLV) are depicted in histogram form; the RLV of vector control-transfected cells was given the arbitrary value of 1. Data are representative of those from at least three independent experiments. (B) Effects of A20 expression on LMP1-mediated JNK activation. HEK 293 cells were transiently transfected with 1 μg of pSG5 (vec) (lanes 1, 3, and 5) or pSG5-LMP1 (lanes 2, 4, and 6) in the presence of increasing concentrations of A20 (0, 0.5, or 1 μg) and 0.5 μg of p46SAPKγ-pcDNA3. Thirty-six hours later cell lysates were analyzed for LMP1 (upper panel) or A20 (second panel) expression. The lower band in the A20 immunoblot represents nonspecific protein. The same lysates (250 μg) were subjected to immune complex kinase assays with GST–c-Jun (aa 1 to 79) as the substrate (third panel). Numbers on the left are molecular weights in thousands. Data were analyzed on a phosphorimager and are depicted in histogram form as fold increases compared to the vector control (bar 1), which was given the arbitrary value of 1 (lower panel). Immunoprecipitates were also analyzed for JNK levels (fourth panel).

In order to determine whether expression of this TRAF2-interacting protein also inhibits LMP1-induced JNK activation, HEK 293 cells were transiently transfected with 1 μg of pSG5-LMP1 in the absence or presence of various amounts of A20 (0.5 or 1 μg) and 0.5 μg of HA-p46SAPKγ-pcDNA3. These amounts of A20 were chosen on the basis of a significant (more than 50%) inhibition of LMP1-induced NF-κB activation. Lysates from these transfectants were subjected to immune complex kinase assays with GST–c-Jun (aa 1 to 79) as the substrate. It was found that A20 confers a potent inhibitory effect on LMP1-mediated JNK activation, inducing a maximum 80% reduction in the presence of 1 μg of A20 expression vector (Fig. 5B). Immunoblot analysis was used to verify LMP1 and A20 expression in transfected cells (Fig. 5B). Similar results were obtained when A20 was coexpressed with CD2.192-LMP1 in HEK 293 cells; following OX34 and IgG stimulation, an 85% decrease in c-Jun substrate phosphorylation levels was observed in the presence of 1 μg of A20 expression vector (data not shown).

Overall, these data demonstrate that the TRAF2-interacting protein A20 significantly inhibits both the LMP1-induced NF-κB and JNK signalling pathways.

Involvement of NIK in LMP1-mediated NF-κB but not JNK signalling.

NIK is a mitogen-activated protein kinase kinase kinase identified in a yeast two-hybrid screen for TRAF2-interacting proteins (37). Overexpression of NIK in target cells activates NF-κB, while a kinase-inactive NIK mutant protein has been shown to be a potent inhibitor of NF-κB induced by both TNF treatment and TRAF2 expression (37, 54).

To determine whether NIK is involved in LMP1-induced NF-κB activation, HEK 293 cells were transiently transfected with 1 μg of pSG5-LMP1, pSG5-LMP1Δ(332-386), or pSG5-LMP1^AxAxA and various amounts (0, 0.1, 0.25, 0.5, or 1μg) of a CMV-driven kinase-inactive NIK mutant [NIK(KK429-430AA)] in the presence of NF-κB-driven luciferase reporter and β-galactosidase expression plasmids. Relative NF-κB activity was assessed at 36 h posttransfection. It was found that NIK(KK429-430AA) significantly inhibited wild-type, CTAR1-, and CTAR2-mediated NF-κB in 293 cells in a concentration-dependent manner (Fig. 6A). Thus, transfection of 0.5 μg of NIK(KK429-430AA) inhibited LMP1-induced NF-κB activation by 50%, and an even-more-pronounced 75% decrease was observed in the presence of 1 μg of kinase-inactive NIK. Overexpression of NIK(KK429-430AA) also had a very potent inhibitory effect on CTAR1-mediated NF-κB activation. These effects were of approximately the same magnitude as those observed upon transfection of 0.5 and 1 μg of A20, respectively (Fig. 5A).

FIG. 6.

NIK is a component of LMP1-mediated NF-κB but not JNK signalling. (A) Transfection of kinase-inactive NIK [NIK(KK429-430AA)] blocks NF-κB activation induced by 1 μg of wild-type LMP1, CTAR1 [LMP1Δ(332-386)], and CTAR2 effector (LMP1^AxAxA). Relative luciferase values are depicted in histogram form; the RLV of vector control-transfected cells was given the arbitrary value of 1. Results are representative of those from three independent experiments. (B) Kinase-inactive NIK does not inhibit LMP1-induced JNK activation. HEK 293 cells were transiently transfected with 1 μg of pSG5 (vec) (lanes 1, 3, and 5) or pSG5-LMP1 (lanes 2, 4, and 6) in the presence of increasing concentrations of NIK(KK429-430AA) (0, 0.5, or 1 μg) and 0.5 μg of p46SAPKγ-pcDNA3. Thirty-six hours later cell lysates were analyzed for LMP1 expression (upper panel) by immunoblotting or subjected to immune complex kinase assays with GST–c-Jun (aa 1 to 79) as the substrate (second panel). Numbers on the left are molecular weights in thousands. Data were quantitated on a phosphorimager and are depicted in histogram form as fold increases compared to the vector control (bar 1), which was given the arbitrary value of 1 (lower panel). Immunoprecipitates were also analyzed for JNK levels (fourth panel).

We then examined whether expression of the kinase-inactive NIK mutant could also block LMP1-induced JNK activation. For this purpose, amounts of NIK(KK429-430AA) which confer a significant inhibition of LMP1-induced NF-κB levels (0.5 or 1 μg) were cotransfected with 1 μg of pSG5-LMP1 or empty vector and 0.5 μg of HA-p46SAPKγ-pcDNA3. Lysates from transfected cells were isolated at 36 h posttransfection and examined for LMP1 expression (Fig. 6B) or immunoprecipitated with an HA-specific antibody and assayed for kinase activity in vitro with GST–c-Jun as the substrate. As shown in Fig. 6B, expression of NIK(KK429-430AA) does not influence LMP1-mediated JNK activation, and similar results were obtained following transfection of pSG5-LMP1^AxAxA (data not shown). This is in contrast to A20, which dramatically inhibited LMP1-mediated JNK signalling (Fig. 5B). Overall, these data demonstrate that NIK is a component of the LMP1-mediated NF-κB pathway but not the JNK signalling pathway downstream of TRAF2.

DISCUSSION

Expression of the EBV-encoded LMP1 induces a plethora of activities in target cells. These include the oncogenic transformation of rodent fibroblast cell lines, up-regulation of various cell surface markers and antiapoptotic proteins, cytokine production, and differentiation blockade in epithelial cells. Furthermore, LMP1 expression is essential for EBV-induced B-cell immortalization in vitro. The signalling pathways which may mediate these phenomena have recently attracted much attention, but the precise organization of LMP1 signal transduction remains unknown. LMP1 expression leads to the rapid activation of the transcription factor NF-κB, an effect mediated independently by two domains in the cytoplasmic C terminus of the protein: CTAR1 (aa 187 to 231) and CTAR2 (aa 351 to 386). More recent studies indicate that LMP1 also mediates activation of a Ras/mitogen-activated protein kinase (MAPK)-dependent pathway (46) as well as of the JNK cascade. LMP1-induced JNK/AP-1 activation maps entirely to the CTAR2 domain and occurs with kinetics that mirror those of NF-κB activation (10, 31).

The effects of CTAR1 on NF-κB could be attributed to its ability to bind molecules of the TRAF family. Indeed, the membrane-proximal LMP1 domain strongly associates with TRAF1 and TRAF3 but is also capable of binding TRAF2 and TRAF5 (2, 6, 28, 42, 52). The latter proteins are of particular interest, as they mediate NF-κB activation by CD40 and certain other TNFR family members. This could also possibly account for the ability of LMP1 to mimic many of the effects of CD40 ligation on cell growth, cytokine production, and induction of cell surface markers (7, 9, 10, 16, 18, 32, 39). Unlike CTAR1, CTAR2 does not directly bind TRAFs; however, a dominant-negative TRAF2 mutant has been shown to partially inhibit CTAR2-induced NF-κB (9, 28). This phenomenon can be explained by the ability of CTAR2 to interact with TRADD (25). TRADD was first identified by virtue of its association with the intracellular death domain of TNFRI in response to TNF-α cross-linking, where it acts as a platform for the recruitment of other proteins, one of which is TRAF2, and this interaction leads to NF-κB activation (23).

The organization and molecular components of LMP1-mediated JNK signalling are, however, unknown. In this study we have dissected the cytoplasmic C tail of LMP1 and found that sequences critical for JNK activation are localized in the extreme C terminus of CTAR2. Thus, deletion of the last 8 aa (aa 378 to 386) abrogates the ability of LMP1 to signal on the JNK axis. Interestingly, the same sequences appear to be important for CTAR2-mediated NF-κB induction. The significance of this extreme C-terminal LMP1 domain for signalling is further evidence by the observation that single point mutations within aa 379 to 385 severely impair CTAR2-mediated NF-κB and JNK activation in HEK 293 cells. Importantly, deletion of the last 8 aa also abrogates the interaction of the LMP1 C terminus with TRADD, suggesting that this adapter protein may be critical for LMP1 signalling. To investigate the contribution of TRADD to LMP1-induced JNK activation, we have used a chimeric molecule consisting of the extracellular and transmembrane domains of CD2 fused to the cytoplasmic C terminus of LMP1 (CD2.192-LMP1). We have found that induction of JNK activity following receptor cross-linking in CD2.192-LMP1-transfected cells is synergistically augmented by low levels of TRADD expression. TRADD can also potentiate CTAR2-mediated NF-κB (reference 25 and our unpublished observations). Taken together, these data provide functional evidence for the contribution of this death domain protein in LMP1 signalling and confirm a role for TRADD in JNK activation.

This observation inevitably raises the question of a possible role for TRAF2 in LMP1-induced JNK activation downstream of TRADD. TRAF2 recruitment to the TNFRI-TRADD complex has been shown to mediate JNK as well as NF-κB activation following TNFRI cross-linking (35). In addition, transient overexpression of TRAF2 induces JNK activity in the absence of TNFR aggregation (43, 45, 54). Further evidence to support a role for TRAF2 in JNK signalling emerges from recent findings suggesting that CD40, which directly binds TRAF2, is a potent activator of JNK (51) and that lymphocytes from TRAF2 dominant-negative transgenic mice are impaired in CD40L-induced JNK activation (34, 58).

To determine the contribution of this molecule to LMP1-induced JNK signalling, we have used a N-terminally deleted TRAF2 mutant [TRAF2Δ(6-86)] which exerts a potent, dominant-negative effect on NF-κB and JNK activity mediated by transient CD40 expression or TNF-α treatment (47) (Fig. 4B and D). This dominant-negative TRAF2 was transfected in HEK 293 cells in the presence of CD2.192-LMP1; following CD2 receptor cross-linking, only a partial inhibition of JNK activation was observed, and similar results were obtained when TRAF2Δ(6-86) was coexpressed with wild-type LMP1 or LMP1^AxAxA (Fig. 4A and C and data not shown). Interestingly, expression of TRAF2Δ(6-86) in 293 cells also confers only a partial blockade of CTAR2-mediated NF-κB activation (28). While these data demonstrate that TRAF2 is a component of CTAR2 signalling, the inability of dominant-negative TRAF2 to completely abolish these signals may indicate an additional contribution(s) from another TRADD-associated protein(s). The preapoptotic protein RIP, for example, interacts with TNFRI-bound TRADD without disrupting the TRADD-TRAF2 complex, and its overexpression induces both JNK and NF-κB activation (21, 35). The role of RIP in LMP1 CTAR2-mediated signalling is presently unknown. Alternatively, it is possible that TRAF2 is bound in a stable complex with other proteins and that large amounts or prolonged incubations following transfection are required for TRAF2Δ(6-86) to displace endogenous wild-type TRAF2. Indeed, a number of TRAF2-interacting proteins have been identified, such as TRAF1, TANK/I-TRAF, and cellular inhibitors of apoptosis (c-IAPs), among others (4, 47, 48, 53), which may influence TRAF2 heterocomplex stability and signalling. Consistent with this possibility is the observation that an increase in the amount of TRAF2Δ(6-86) from 2.5 to 5 μg significantly decreased JNK activation induced by expression of 1 μg of LMP1, from 45 to 60% (Fig. 4C). A similar requirement for large amounts of dominant-negative TRAF2 to elicit a significant inhibitory effect on JNK activation induced by TNF-α has been previously described (22). Thus, differences in the affinity and/or stoichiometry of TRADD/TRAF2-associated factors may be responsible for the ability of dominant-negative mutated TRAF2 to completely block TNFRI but not LMP1 CTAR2 signals.

In this context it is also of interest that CD40 but not LMP1 CTAR1 can activate the JNK/AP-1 pathway in 293 cells (10), a phenomenon which may reflect differences in the interaction of TRAFs with the cytoplasmic tails of CD40 and LMP1. Thus, despite a common PxQxT TRAF-binding motif, TRAF1 interacts with LMP1 CTAR1 directly but with the CD40 cytoplasmic tail only indirectly (6, 42, 47). In addition, while TRAF2 strongly binds CD40, it interacts only weakly with CTAR1 (6, 47, 52), and there is some evidence that CTAR1 and CD40 signalling may be quantitatively and qualitatively different (11a). Alternatively, these data may indicate a disruption in the wiring of signals leading to JNK activation downstream of CTAR1/TRAF2.

The contribution of TRAF2 in CTAR2-mediated signalling is further emphasized by the ability of the TRAF2-interacting proteins A20 and NIK to influence NF-κB and/or JNK activation from this LMP1 C-terminal domain. The NF-κB-inducible zinc finger A20 protein inhibits TNF-α-mediated NF-κB and AP-1 transactivation, presumably by interfering with TRAF2 signal transduction. Indeed, A20 has been shown to block TRAF2-induced NF-κB activation (55). The ability of A20 to also inhibit interleukin-1-induced NF-κB activation (26, 55), which is mediated by TRAF6 (3), suggests that A20 may function as a promiscuous inhibitor of TRAF activities. In this context, our data demonstrating that A20 expression suppresses both LMP1-induced NF-κB and JNK activation while dominant-negative TRAF2 has only a partial effect may indicate an additional role for other TRAF family members in CTAR2-mediated signalling. Interestingly, overexpression of TANK has also been shown to confer a more potent inhibitory effect on CTAR2-mediated NF-κB than the dominant-negative TRAF2 mutant (28). Unlike A20, the TRAF2-interacting protein kinase NIK appears to regulate LMP1-induced NF-κB but not JNK activation. Thus, expression of the kinase-inactive NIK mutant [NIK(KK429-430AA)] significantly impaired wild-type LMP1-, CTAR1-, and CTAR2-mediated NF-κB but had no effect on JNK signalling. This observation coupled with the reported ability of a dominant-negative SEK to block LMP1-induced JNK but not NF-κB (10) suggests that these two signalling pathways bifurcate at the level of TRAF2.

Thus, the organization of LMP1 signalling so far appears to be similar but not identical to that of CD40 or TNFRI (reviewed in reference 8). The CTAR1 domain, which binds TRAF1, TRAF2, and TRAF3, mediates low NF-κB activity via a CTAR1-TRAF2-NIK connection but fails to induce JNK in 293 cells. NIK may in turn activate the recently identified IκB kinase (IKK), which induces phosphorylation and degradation of IκBα and release of functional NF-κB (38, 57). Indeed, LMP1 appears to activate NF-κB through phosphorylation of IκBα (10, 20). CTAR2 mimics TNFRI by exploiting TRADD as its signalling adapter. Recruitment of TRAF2 to the LMP1-TRADD complex may modulate JNK/AP-1 and NF-κB signalling but not to the same extent as in TNFRI. Induction of NF-κB may occur via a NIK-dependent cascade similar to that of CTAR1, while the signalling component leading to SEK–JNK–AP-1 activation downstream of TRAF2 is presently unknown. Expression of A20 disrupts both JNK and NF-κB signals. Additional TRADD-interacting molecules which regulate JNK activation from CTAR2 may exist.

The identification of the signalling mechanisms used by the EBV-encoded LMP1 so far reveals important similarities with the pathways activated by TNFR or CD40 cross-linking and may explain its ability to recapitulate many of the functions of this receptor superfamily. However, the present study also highlights interesting differences in the nature of TRADD-dependent effects mediated via CTAR2, which may have important implications for the transforming ability of LMP1.

ADDENDUM

After submission of this paper, Sylla et al. (55a) reported that LMP1-mediated NF-κB activation occurs via a NIK-dependent pathway.