Activation of the p38 Mitogen-activated Protein Kinase Pathway by Epstein-Barr Virus-encoded Latent Membrane Protein 1 Coregulates Interleukin-6 and Interleukin-8 Production*

The Epstein-Barr virus-encoded latent membrane protein 1 (LMP1) is a pleiotropic protein the activities of which include effects on gene expression and cell transformation, growth, and death. LMP1 has been shown to induce nuclear factor (NF)-κB and c-Jun NH2-terminal kinase/AP-1 activities in target cells, and in this study we demonstrate that LMP1 also engages the p38 mitogen-activated protein kinase cascade, leading to activation of the transcription factor ATF2. Mutational analysis of the LMP1 cytoplasmic COOH terminus revealed that p38 activation occurs from both the tumor necrosis factor receptor-associated factor (TRAF)-interacting, membrane-proximal COOH-terminal activating region (CTAR)1 domain (amino acids 186–231) and the extreme tumor necrosis factor receptor-associated death domain (TRADD) binding CTAR2 region (amino acids 351–386). Because LMP1 also engages signaling on the NF-κB axis through CTAR1 and CTAR2, we have examined whether these two pathways are overlapping or independent. We have found that inhibition of p38 by the highly specific inhibitor SB203580 did not affect NF-κB binding activity. Conversely, although the metabolic inhibitor D609 blocked NF-κB activation, it did not impair the ability of LMP1 to signal on the p38 axis, suggesting that these two LMP1-mediated pathways are primarily independent. Divergence of signals must, however, occur downstream of TRAF2 as a dominant negative TRAF2 mutant that blocks LMP1-induced NF-κB activation also inhibited p38 signaling. In addition, we have found that p38 inhibition significantly impaired LMP1-mediated interleukin-6 and -8 expression. Thus, p38 may play a significant cooperative role in regulating at least some of the pleiotropic activities of LMP1.

The Epstein-Barr virus-encoded latent membrane protein 1 (LMP1) is a pleiotropic protein the activities of which include effects on gene expression and cell transformation, growth, and death. LMP1 has been shown to induce nuclear factor (NF)-B and c-Jun NH 2 -terminal kinase/AP-1 activities in target cells, and in this study we demonstrate that LMP1 also engages the p38 mitogen-activated protein kinase cascade, leading to activation of the transcription factor ATF2. Mutational analysis of the LMP1 cytoplasmic COOH terminus revealed that p38 activation occurs from both the tumor necrosis factor receptor-associated factor (TRAF)-interacting, membrane-proximal COOH-terminal activating region (CTAR)1 domain (amino acids 186 -231) and the extreme tumor necrosis factor receptor-associated death domain (TRADD) binding CTAR2 region (amino acids 351-386). Because LMP1 also engages signaling on the NF-B axis through CTAR1 and CTAR2, we have examined whether these two pathways are overlapping or independent. We have found that inhibition of p38 by the highly specific inhibitor SB203580 did not affect NF-B binding activity. Conversely, although the metabolic inhibitor D609 blocked NF-B activation, it did not impair the ability of LMP1 to signal on the p38 axis, suggesting that these two LMP1-mediated pathways are primarily independent. Divergence of signals must, however, occur downstream of TRAF2 as a dominant negative TRAF2 mutant that blocks LMP1-induced NF-B activation also inhibited p38 signaling. In addition, we have found that p38 inhibition significantly impaired LMP1-mediated interleukin-6 and -8 expression. Thus, p38 may play a significant cooperative role in regulating at least some of the pleiotropic activities of LMP1.

Epstein-Barr virus (EBV) 1 is a human herpesvirus that is
associated with several types of malignancy. EBV infects resting B cells, stimulates their proliferation, and induces the outgrowth of virus-transformed lymphoblastoid cell lines expressing the nuclear antigens EBNA1, EBNA2, EBNA3A, EBNA3B, EBNA3C, and EBNA-LP and the latent membrane proteins LMP1, LMP2A, and LMP2B (1).
Among the nuclear and membrane proteins expressed as a consequence of EBV infection, the latent membrane protein 1 (LMP1) is of particular interest because it induces the oncogenic transformation of rodent fibroblast cell lines (2,3). LMP1 expression is also essential for EBV-mediated primary B cell transformation in vitro (1,4) and is associated with a number of human malignancies such as Hodgkin's disease, undifferentiated nasopharyngeal carcinoma, and EBV-related lymphoproliferative disease (1). Expression of this viral oncogene in B cells can induce a plethora of activities including up-regulation of cell surface markers such as CD23, CD40, and CD54 (intercellular adhesion molecule 1) and induction of anti-apoptotic proteins such as A20 and members of the Bcl-2 family (5-7; for review, see Ref. 8). In epithelial cells, LMP1 can also induce A20, CD40, and CD54 expression and block differentiation, a property that may be important in the pathogenesis of nasopharyngeal carcinoma (9 -11). Of particular interest is the reported ability of LMP1 to induce production of cytokines such as interleukin-6 (IL-6) (12) and IL-10 (13), suggesting that this viral protein may activate inflammatory and immune-regulatory responses following EBV infection.
Structurally, LMP1 is a 63-kDa phosphoprotein comprising a short 23-amino acid NH 2 -terminal cytoplasmic domain, six membrane-spanning domains of 162 amino acids, and a long 200-amino acid COOH-terminal cytoplasmic tail. Mutational analysis has identified COOH-terminal regions of the protein as being essential for transformation and phenotypic changes. The 45 most membrane-proximal residues of the LMP1 COOH terminus (amino acids 186 -231) are critical for EBV-mediated transformation of primary B cells, but recent studies suggest that the long term growth of these cells also requires the distal COOH-terminal sequences (amino acids 352-386) (4,14,15).
Interestingly, these two functional domains of the LMP1 cytoplasmic tail can also activate the transcription factor NF-B, with the extreme COOH-terminal activating region 2 (CTAR2, amino acids 351-386) being the principal contributor to this effect in the majority of cell lines (6,16). This phenomenon can be attributed to the ability of CTAR2 to associate with tumor necrosis factor (TNF) receptor-associated death domain (TRADD) (15), a protein that mediates NF-B signaling from aggregated TNF receptor I (TNFRI) (17). The proximal CTAR1 domain of LMP1 (amino acids 187-231) induces low levels of NF-B through its direct interaction with TNFR-associated factor 2 (TRAF2) (18,19). TRAF3 has also been shown to bind CTAR1 (20). Importantly, these CTAR1-TRAF interactions occur through a P 204 xQ 206 xT 208 TRAF binding motif, common to the cytoplasmic tails of some TNFR family members such as CD40 and CD30. In addition to NF-B, LMP1 expression signals for activation of a Ras/MAPK/ERK pathway (21) and of the JNK (c-Jun NH 2 -terminal kinase, also known as the stressactivated protein kinase, SAPK) cascade (22)(23)(24), a phenomenon that is mediated through CTAR2 and results in the induction of the transcription factor c-Jun/AP-1 (22,24).
In this study we demonstrate that EBV-encoded LMP1 can also activate the p38 MAPK pathway. This phenomenon occurs through both CTAR1 and CTAR2 domains of the protein and appears to be mediated by the adaptor protein TRAF2. p38 activation has been observed in response to a variety of stimuli, including hyperosmotic shock, UV radiation, lipopolysaccharide treatment, and stimulation by certain cytokines such as IL-1 and TNF-␣ and requires phosphorylation of a closely spaced tyrosine and threonine residue in the activation domain of the protein (for review, see Refs. 25 and 26). Among the downstream targets of p38 are the heat-shock protein 27 (hsp27) and the transcription factors ATF2 (27), Elk-1 (28), CHOP/GADD153 (29), and Max (30). In addition, inhibition of p38 activity has been shown to interfere with TNF-mediated NF-B transactivation and to influence TNF-induced apoptosis (31) and IL-6 production (32)(33)(34). In agreement with these data we demonstrate that inhibition of p38 signaling impairs the ability of LMP1 to induce IL-6 and IL-8 expression.

MATERIALS AND METHODS
DNA Constructs-pSG5-based LMP1 and LMP1 deletion mutants ⌬(332-386) and ⌬(187-351) have been described previously (6). pSG5-LMP1 AxAxA was generated by site-directed mutagenesis using the QuickChange TM site-directed mutagenesis kit of Stratagene and pSG5-LMP1 as substrate. The mutated oligonucleotide primers used were: 5Ј-CCTCCCGCACGCTCAAGCAGCTGCCGATGA-3Ј and its complementary. pSG5-LMP1 AxAxA /378STOP was generated by a similar approach using the mutated primers 5Ј-GATGACGACCCCCACTGAC-CAGTTCAGCTAAGC-3Ј and its complementary and pSG5LMP1 AxAxA as substrate. These mutations generate a STOP codon at position 378 of the amino acid sequence of LMP1 and were verified by sequencing. pSG5CD2.192-LMP1 has been described previously (35). pIND-LMP1 was generated by excision of LMP1 coding sequences from pSG5-LMP1 using EcoRI digestion and subsequent insertion into the EcoRI site of the pIND vector (Invitrogen).
The 181-bp IL-8 promoter sequences (Ϫ135 to ϩ46) were PCR amplified from human genomic DNA using the following primers: 5Ј-GTGAGATCTGAAGTGTGATGACTCAGG-3Ј (IL8P-Forward), which contains an artificial BglII site, and 5Ј-GTGAAGCTTGAAGCTTGTGT-GCTCTGC-3Ј (IL8P-Reverse), which contains an artificial HindIII site. The PCR product was then digested with BglII/HindIII and inserted into the corresponding restriction sites of the luciferase reporter plasmid pGL2-Basic (Promega) to generate IL8-Luc. To generate the mu-tAP-1/IL8-Luc vector that contains the same IL-8 promoter sequences but with a double mutation that distorts the AP-1 consensus, the 5Ј-GTGAGATCTGAAGTGTGATATCTCAGG-3Ј (mut-IL8P-Forward) was used together with IL8P-Reverse. The PCR product was again digested with BglII/HindIII and ligated into pGL2-Basic.
Cell Culture, Transfections, and Reporter Assays-HEK 293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 2 mM glutamine. For transient transfections, 8 ϫ 10 5 HEK 293 cells were plated out on a 25-cm 2 flask and the following day were transfected using a standard calcium phosphate technique. HeLa and Rat-1 cells, cultured in RPMI and 10% fetal calf serum supplemented with 2 mM glutamine were transfected using the PrimeFector lipofection kit (EquiBio Ltd, Kent, U. K.). The xanthogenate compound D609 (tricyclodecan-9-yl-xanthogenate potassium) was purchased from Calbiochem and dissolved in growth medium immediately before use. The p38 inhibitor SB203580 was purchased from Calbiochem, dissolved in dimethyl sulfoxide, and stored at Ϫ20°C.
Luciferase reporter and ␤-galactosidase assays were performed as described previously (12). 50 ng each of a CMV-driven ␤-galactosidaseexpressing plasmid and of 3Enh.B-ConALuc reporter, which contains three tandem repeats of the NF-B sites from the Ig promoter, were used routinely to transfect 293 cells. Alternatively, 293 cells were transfected with 50 ng of IL8-Luc or mutAP-1/IL8-Luc and 50 ng of ␤-galactosidase, whereas HeLa cells were transfected with 1 g each of these reporter constructs. Analysis of luciferase and ␤-galactosidase expression was performed at 36 h post-transfection.
Immunoprecipitations, Kinase Assays, and Immunoblotting-JNK in vitro kinase assays were performed as described previously (24). For phospho-p38 immunoblots or p38 kinase assays, cells were lysed in 300 -500 l 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 g/ml aprotinin, 10 g/ml leupeptin, and 1 mM DTT) for 20 min on ice. Cell debris were removed by centrifugation, and the protein concentration was determined using a commercially available Bio-Rad protein assay. p38 MAPK was immunoprecipitated from 200 -250 g of total protein extracts using 1 g of anti-HA antibody (Boehringer Mannheim) for 2 h and 25 l of protein G-Sepharose (Amersham Pharmacia Biotech) for an additional 1 h. After immunoprecipitation, beads were washed twice with kinase lysis buffer and twice with assay buffer (25 mM Tris, pH 7.5, 5 mM ␤-glycerophosphate, 10 mM MgCl 2 , 2 mM DTT, 100 M sodium vanadate, and 1 g/ml leupeptin). After the last wash, the beads were drained using a fine gauge Hamilton syringe and resuspended in 50 l of assay buffer containing 2 g of GST-ATF2(19 -96) substrate (New England Biolabs) and 200 M ATP. Kinase reactions were carried out at 30°C for 30 min and stopped by the addition of 25 l of 3 ϫ Laemmli buffer and boiling for 5 min. Samples were then analyzed on a 12.5% SDS-polyacrylamide gel, and ATF2 phosphorylation was detected by immunoblot using a phospho-specific ATF2 antibody (New England Biolabs) which reacts with Thr 69 /Thr 71 doubly phosphorylated ATF2 followed by densitometric analysis using a Bio-Rad GS-690 imaging densitometer. Immunoblot analysis of anti-HA immunoprecipitates using a control p38 antibody (New England Biolabs) was also performed to demonstrate that comparable amounts of HA-p38 were analyzed in cotransfection experiments. Phosphorylation of p38 was determined by immunoblot analysis of 50 g of cell extracts using a phospho-specific p38 MAPK (Thr 180 /Tyr 182 ) antibody (New England Biolabs).
For LMP1 and TRAF2 immunoblotting, 25 g 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. 1-4 (36) or the TRAF2(C-20) polyclonal antibody (Santa-Cruz) and ECL (Amersham Pharmacia Biotech).
Electrophoretic Mobility Shift Assays (EMSAs)-Cell nuclei isolated by resuspending cells in a solution containing 10 mM HEPES, pH 7.9, 1.5 mM magnesium chloride, 10 mM potassium chloride, 0.5 mM DTT, and protease inhibitors were subjected to lysis in a buffer constituting 20 mM HEPES, pH 7.9, 25% glycerol, 1.5 mM magnesium chloride, 0.42 M sodium chloride, 0.2 mM EDTA, 0.5 mM DTT, and protease inhibitors. The protein concentration of isolated nuclear extracts was determined by the Bio-Rad protein assay, according to the manufacturer's instructions. For EMSAs, a 29-bp HIV-B probe (5Ј-GATCAGGGACTTTC-CGCTGGGGACTTTCC-3Ј), a 22-bp collagenase TRE probe (5Ј-AGCT-TGATGAGTCAGCCGGATC-3Ј), a 24-bp Jun2 TRE probe (5Ј-AGCTAGCATTACCTCATCCCGATC-3Ј), and a 24-bp IL-8 probe containing the AP-1 consensus (5Ј-GAAGTGTGATGACTCAGGTTT-GCC-3Ј) were made by annealing complementary synthetic oligonucleotides and end labeling using [ 32 P]ATP (Amersham Pharmacia Biotech) and T4 polynucleotide kinase (Boehringer Mannheim) followed by spincolumn purification. Binding reactions containing 10 g of nuclear protein extract, 1 g of poly(dI-dC) (Amersham Pharmacia Biotech), 0.1 ng of probe labeled to a specific activity of 2 ϫ 10 8 cpm/g, and binding buffer (4% glycerol, 1 mM EDTA, 5 mM DTT, 0.01 M Tris⅐HCl, pH 7.5, and 5 mM KCl) in a volume of 20 l were incubated for 30 min at room temperature and then resolved by electrophoresis through a 5% polyacrylamide gel in a 0.55 ϫ TBE buffer. Gels were then dried and exposed to x-ray film for autoradiography (Kodak). In supershift experiments, nuclear extracts were preincubated for 45 min with antibodies directed against c-Fos (Santa Cruz, sc-052X) or c-Jun (Santa Cruz, sc-045X) and then for a further 30 min with radiolabeled probe before being subjected to EMSA as described above. For ATF2 depletion experiments, 10-g nuclear extracts were incubated with 1 g of anti-ATF2 or control c-Rel antibody (Santa Cruz, sc-242X or sc-272X, respectively) for 2 h on ice. Depletion was achieved by the addition of antirabbit IgG-coated agarose beads during the last 60 min of the incubation. After a brief centrifugation at 12,000 ϫ g, the extracts were subjected to EMSA as above. For EMSAs using recombinant ATF2, 0.5 g of full-length ATF2 protein (Santa Cruz, sc-4007) was incubated for 30 min with collagenase TRE, Jun2 TRE, or IL-8/AP-1 probe in the presence of poly(dI-dC) and binding buffer as described above and was then analyzed on a 5% polyacrylamide gel electrophoresis.
Reverse Transcription PCR and ELISA-cDNA synthesis from total RNA and semiquantitative reverse transcription PCR were performed as described previously (12). A 208-bp IL-8 cDNA fragment was generated using the primers 5Ј-CAGTTTTGCCAAGGAGTGCTA-3Ј (IL8-Forward) and 5Ј-AACTTCTCCACAACCCTCTGC-3Ј (IL8-Reverse) under the following conditions: denaturation at 94°C for 45 s, annealing at 52°C for 45 s, and extension at 72°C for 50 s for 24 cycles followed by a final extension step at 72°C for 50 min. PCR products were analyzed on a 1.5% agarose gel, transferred on Hybond N ϩ , and hybridized with the oligonucleotide probe 5Ј-TGATTGAGAGTGGACCACA-3Ј (IL8-Probe). Primers and amplification for GAPDH have been described previously. IL-8 versus GAPDH hybridization signals were quantified on a Molecular Dynamics PhosphorImager.
The presence of IL-6 or IL-8 protein in the supernatants of cultured cells was determined by ELISAs (Pelikine human IL-6 or IL-8 ELISA, CLB, Netherlands) as described (12). 12 h after transfection, cells were washed with phosphate-buffered saline and plated on a 24-well plate at a density of 40,000 cells/ml in 1 ml of complete medium in the presence or absence of inhibitors. Supernatants were analyzed 24 h later. Alternatively, 40,000 293EcR-LMP1 cells were pretreated for 1 h with 20 M SB203580 and then induced to express LMP1 by treatment with 10 M ponasterone A; supernatants were analyzed 24 h later.

LMP1 Expression Induces Phosphorylation of p38 and
ATF2-To determine the effects of LMP1 on the p38 MAPK pathway, an ecdysone-inducible system was used to provide regulatable expression of LMP1. This system is based on the binding of the steroid hormone ecdysone analog ponasterone A to a heterodimeric receptor comprising a modified ecdysone receptor and the retinoid X receptor (RXR) which allows the subsequent activation of an ecdysone-responsive promoter to express the target gene (37). HEK 293 cells carrying the pVgRXR plasmid (293EcR), which encodes the receptor subunits, were transfected with pIND-LMP1, which contains the Cell lysates were analyzed for p38 phosphorylation by immunoblot using an antibody that specifically recognizes the phosphorylated form of the protein. Data are representative of at least five independent experiments. C, LMP1 induces p38 kinase activity. HEK 293EcR-LMP1/cl.4 or control cells were transfected with HA-p38 and 24 h later were stimulated with 10 M ponasterone A for various time intervals (0, 3, 6, 12, or 24 h). HA immunoprecipitates of cell lysates were then analyzed for kinase activity using GST-ATF2(19 -96) fusion protein as substrate. ATF2 phosphorylation was determined by immunoblot analysis using an antibody specific for the phosphorylated form of ATF2. A maximum of an 8.1-fold increase in ATF2 phosphorylation levels of 293EcR-LMP1 cells was observed at 12 h of treatment compared to control cultures. As positive control, cells were treated with sodium salicylate and analyzed for kinase activity as above (first lane from left). Immunoblot analysis of anti-HA immunoprecipitates using a control p38-specific antibody was also performed to demonstrate that comparable amounts of HA-p38 were analyzed in cotransfection experiments (lower panel). Three independent experiments were performed and gave similar results. LMP1 coding sequences under the control of ecdysone-response elements. Selected stable clones (293EcR-LMP1) were then examined for ponasterone A-inducible gene expression. Results from a representative clone (293EcR-LMP1/cl.4) are shown in Fig. 1A. Induction of LMP1 after the addition of 10 M ponasterone A was observed as early as 6 h and increased further at 12 h to remain stable for up to 24 h of treatment. The levels of LMP1 achieved after treatment with ponasterone A were comparable to those expressed in X50-7, an EBV-transformed B cell line (Fig. 1A, X50-7 lane). Control cultures, stably transfected with empty pIND vector, showed no LMP1 expression (Fig. 1A). Cell lysates were also analyzed for p38 phosphorylation by immunoblot using an antibody that specifically recognizes the phosphorylated form of the protein (34,38). These experiments demonstrated a transient increase in endogenous p38 phosphorylation levels in parallel with the induction of LMP1 expression (Fig. 1B), and similar results were obtained with an additional 293EcR-LMP1 clone (293EcR-LMP1/cl.5, data not shown). The levels of p38 phosphorylation in vector control-transfected cultures remained unaffected (Fig. 1B).
To demonstrate that the observed LMP1-mediated p38 phosphorylation is functional, the ability of immunoprecipitated p38 to activate ATF2 was examined. The transcription factor ATF2 is a known substrate for p38 MAPK, being phosphorylated primarily on threonine residues 69 and 71, and this event increases its transactivating properties. HEK 293EcR-LMP1/ cl.4 or control cells were transiently transfected with HAtagged p38 expression vector and 24 h later were stimulated with 10 M ponasterone A for various time intervals (0, 3, 6, 12, or 24 h). Lysates from these cultures were isolated, immunoprecipitated with an HA-specific antibody, and assayed for kinase activity using GST-ATF2(19 -96) fusion protein as sub-strate. ATF2 phosphorylation was then determined by immunoblot analysis using an antibody specific for the phosphorylated form of ATF2. As shown in Fig. 1C, no increase in p38 kinase activity was observed after a 3-h exposure to ponasterone A, in agreement with the lack of induction of LMP1 expression and p38 phosphorylation at this time point. A significant increase in ATF2 phosphorylation levels was, however, noted at 6 h and peaked at 12 h of treatment, in parallel with p38 phosphorylation. A small decrease in p38 kinase activity was then observed at 24 h of stimulation (Fig. 1C, upper panel). As a control for this experiment, HA-p38-transfected 293EcR-LMP1/cl.4 cells treated for 10 min with 20 mM sodium salicylate, which has been shown to induce p38 activation (38), were used (Fig. 1C, first lane). Immunoblot analysis of HA immunoprecipitates using an anti-ATF2 control antibody verified that comparable amounts of HA-p38 were analyzed (Fig. 1C, lower  panel). In similar immune complex kinase assays, the anti-HA immunoprecipitate failed to phosphorylate GST substrate (data not shown). Thus, inducible LMP1 expression triggers the phosphorylation of p38 and ATF2.
Activation of the p38 MAPK pathway was also observed after transient expression of LMP1. HeLa cervical carcinoma cells were transfected with increasing doses (1 or 2.5 g) of pSG5-LMP1, an SV40 early promoter-driven LMP1 expression vector, in the presence of 0.5 g of HA-p38, and kinase activity was determined 36 h later by immune complex kinase assays using GST-ATF2(19 -96) as substrate. These experiments demonstrated that increasing levels of LMP1 expression correlated with increased p38 kinase activity (Fig. 2, A and B), and similar results were obtained in HEK 293 cells (see Fig. 4C).
LMP1-mediated p38 Activation Requires Oligomerization at the Plasma Membrane and Can Be Primarily Dissociated from NF-B Activation-Recent studies have demonstrated that LMP1-mediated NF-B activation requires oligomerization of the protein at the plasma membrane (35,39). Thus, chimeric molecules in which the extracellular and transmembrane regions of CD2, CD4, or nerve growth factor receptor have been linked to the cytoplasmic tail of LMP1 can activate NF-B signaling only after antibody-or ligand-induced aggregation of the chimera.
To determine whether oligomerization is also important for LMP1-mediated p38 activation, a CD2/LMP1 chimera comprising the extracellular and transmembrane domains of CD2 (amino acids 1-212) linked to the cytoplasmic COOH terminus of LMP1 (amino acids 192-386) (construct pSG5CD2.192-LMP1) (35) was used to transfect Rat-1 fibroblasts. CD2 and LMP1 expression levels in stable transfectants were verified using flow cytometry and immunoblot analysis, respectively. Expression in a representative clone (Rat-1/CD2.192LMP1 cl.4) is shown in Fig. 3A. In these assays LMP1 appeared as a broad band presumably caused by CD2 glycosylation (Fig. 3A,  right panel). These cultures were subsequently transfected with 2 g of HA-p38 and 36 h later were either left untreated or stimulated for 3 h with OX34 anti-CD2 mAb and crosslinking anti-mouse IgG, as described (35,39). Cell lysates were immunoprecipitated with a HA-specific antibody and assayed for kinase activity using GST-ATF2(19 -96) fusion protein as substrate. As shown in Fig. 3B, untreated Rat-1/CD2.192LMP1 cells demonstrated only basal levels of ATF2 phosphorylation (fifth lane); however, a significant increase in p38 kinase activity was observed after aggregation of the chimera (sixth lane), suggesting that oligomerization is essential for activation of p38 by LMP1. ATF2 phosphorylation levels in control cultures remained unaffected (Fig. 3B, first two lanes).
Because LMP1 also engages NF-B signaling, experiments were carried out to establish whether the p38 and NF-B pathways are overlapping or can be dissociated. For this purpose, we first examined whether inhibition of LMP1-mediated p38 activation could influence NF-B signaling. Rat-1/ CD2.192LMP1 cl.4 cells were stimulated for 3 h with OX34/IgG in the presence of 20 M SB203580, a pyridinyl imidazole compound that has been shown previously to inhibit p38 activity in response to a variety of stimuli (32,40,41). In agreement with these reports, treatment with SB203580 significantly inhibited ATF2 phosphorylation induced by CD2 cross-linking (Fig. 3B, seventh lane). The requirement for high concentrations of this compound to block p38 kinase activity compared with amounts required in vitro could be attributed to the ability of SB203580 to bind unactivated as well as activated p38 and to compete with ATP in vivo (42); high p38 levels in transfected cells would require increased amounts of inhibitor. Parallel cultures were analyzed for NF-B activation by EMSAs using a 32 P-labeled HIV-B probe. It was found that treatment of Rat-1/CD2.192LMP1 cells with SB203580 did not impair the ability of LMP1 to activate NF-B after CD2 engagement (Fig. 3C,  sixth and seventh lanes). In addition, no inhibition in antibody- FIG. 4. Both CTAR1 and CTAR2 domains of LMP1 contribute to p38 MAPK activation. A, schematic representation of the LMP1 protein and the deleted LMP1 gene sequences used in this study. Solid black lines represent wild type (wt), and dotted lines denote deleted LMP1 sequences. CTAR1 is located at residues 194 -232 and CTAR2 at residues 351-386. The asterisks represent a triple P 204 xQ 206 xT 208 3 AxAxA mutation. B, induction of NF-B-dependent transcriptional activity by LMP1 and LMP1 deletion mutants. HEK 293 cells were transfected with 1 g of pSG5-based constructs in the presence of 50 ng of NF-B-regulated luciferase reporter plasmid BConA-Luc and 50 ng of ␤-galactosidase expression vector. Relative luciferase values (RLV), which represent the luciferase values normalized on the basis of ␤-galactosidase expression, were determined at 36 h post-transfection. Data shown represent fold increase in RLV relative to vector control, which was given the arbitrary value of 1 and are the mean (ϮS.D.) of at least three independent experiments. C, induction of p38 kinase activity by LMP1 and LMP1 mutants. HEK 293 cells were transfected with 0.5 g of HA-p38 and 1 g of pSG5 or pSG5-based LMP1 expression vectors; p38 kinase activity was assessed using immune complex kinase assays and GST-ATF2(19 -96) as substrate. Densitometric analysis using a Bio-Rad GS-690 imaging densitometer showed the following increases in ATF2 phosphorylation levels compared with control vector, which was given the arbitrary value of 1: LMP1, 7.5; LMP1⌬(332-386), 1.9; LMP1⌬(187-351), 5.5; LMP1 AxAxA , 4.7; LMP1 AxAxA /378STOP, 0.9. Three independent experiments were performed and gave similar results.
induced JNK activation was observed at this SB203580 concentration (data not shown). We then examined the effects of the metabolic inhibitor D609 on p38 and NF-B activation. This compound has been shown previously to prevent nuclear translocation of NF-B and subsequent NF-B transactivation in response to a variety of stimuli, including LMP1 expression (24,43). In agreement with these reports, exposure of OX34/ IgG-stimulated Rat-1/CD2.192LMP1 cl.4 cells to 25 g/ml D609 significantly inhibited LMP1-mediated NF-B binding activity (Fig. 3C, far right lane). Lysates from parallel cultures, cotransfected with HA-p38 expression vector, were analyzed for p38 kinase activity. It was found that far from reducing kinase activity, D609 increased ATF2 phosphorylation in response to LMP1 signaling. This is in agreement with previous reports showing that antioxidants and potent NF-B inhibitors increase p38 kinase activity in response to various stimuli, such as TNF-␣ and phorbol 12-myristate 13-acetate (41). Our data demonstrate that compounds that efficiently block LMP1mediated NF-B activation in Rat-1 cells do not impair its ability to signal on the p38 axis, and conversely, inhibition of p38 activity does not influence NF-B binding, indicating divergence of signals.
Both CTAR1 and CTAR2 Domains of LMP1 Mediate p38 Activation-Previous studies have shown that LMP1 activates signaling on the NF-B axis through two distinct regions in its cytoplasmic COOH terminus, namely CTAR1 (amino acids 187-231) and CTAR2 (amino acids 351-386). To determine whether these LMP1 domains are also implicated in p38 signaling, LMP1 deletion and point mutants with inactivated CTAR1 or CTAR2 were used (Fig. 4A).
These mutated LMP1 constructs were first analyzed for their effects on NF-B activation using luciferase reporter assays in transiently transfected HEK 293 cells. It was found that deletion of CTAR2 (construct pSG5-LMP1⌬(332-386)) significantly reduced NF-B activity to approximately 20% of wild type LMP1 (Fig. 4B). Deletion of CTAR1 (construct pSG5-LMP1((187-351)) induced only a small reduction in NF-B levels to approximately 75% of the wild type molecule. Comparable levels of NF-B activation were observed after expression of pSG5-LMP1 AxAxA , which contains a P 204 xQ 206 xT 208 3 Ax-AxA mutation and therefore acts as a CTAR2 effector. Indeed, this triple mutation has been shown previously to block CTAR1-mediated NF-B by abrogating TRAF binding to this LMP1 domain (12,19). Because the extreme COOH terminus of LMP1 is important for association with TRADD and CTAR2 signaling (15,58), we have introduced a stop codon at amino acid position 378 of pSG5-LMP1 AxAxA (construct pSG5-LMP1 AxAxA /378STOP, Fig. 4A) and examined the ability of this mutant to engage NF-B signaling. Removal of the last 8 amino acids from LMP1 AxAxA completely abolished the ability of this EBV protein to signal on the NF-B axis (Fig. 4B). These data verify that CTAR2 is the major NF-B-activating domain and demonstrate that the LMP1 constructs used are functional.
To determine whether CTAR1 and CTAR2 also contribute to p38 activation, HEK 293 cells were transfected with 1 g of pSG5 or pSG5-based LMP1 constructs together with 0.5 g of HA-tagged p38 expression vector. Lysates from transfected cells were subjected to immune complex kinase assays using GST-ATF2(19 -96) fusion protein as substrate. As shown in Fig. 4C, expression of LMP1⌬(187-351) or LMP1 AxAxA induced significant p38 kinase activity, albeit lower than wild type LMP1, whereas transfection of the CTAR1-expressing pSG5-LMP1⌬(332-386) construct had only a small effect on ATF2 phosphorylation. Removal of the last 8 amino acids from LMP1 AxAxA abolished LMP1-mediated p38 signaling, in agreement with the loss of its functional CTAR1 and CTAR2 domains (Fig. 4C, middle panel). Immunoblot analysis using the CS.1-4 mAbs (36) was also performed to confirm LMP1 expression from these plasmids (Fig. 4C, top panel). As has been documented previously (6), the CTAR1-deleted LMP1⌬(187-351) construct was not detectable by immunoblotting, but expression was confirmed by immunofluorescence staining with the CS.1 mAb (data not shown). Thus, both the TRAF-interacting CTAR1 and TRADD-binding CTAR2 domains of LMP1 can activate p38 signaling with CTAR2 being the major contributor.
TRAF2 Is a Mediator of LMP1-activated p38 Signaling-The inability of LMP1 AxAxA /378STOP mutated construct, which lacks both the TRAF-and TRADD-interacting domains of LMP1 to induce p38 kinase activity, suggests that these adap- tor proteins are important for p38 signaling. TRADD and TRAF2 in particular are known to regulate TNFR, CD40, and LMP1-mediated JNK and NF-B signaling (44 -47, 58). We therefore examined the ability of TRAF2 to regulate p38 activation by LMP1. For this purpose, HEK 293 cells were cotransfected with 1 g of pSG5LMP1 and HA-p38 in the presence or absence of 1 g of a CMV-driven TRAF2⌬(6 -86) (18). This NH 2 -terminal deleted TRAF2-expressing construct has been shown to act as a dominant negative mutant of TRAF2 activities because it blocks JNK and NF-B signals from CD40, TNFRII, and LMP1 (12, 18, 19, 44 -47, 58). TRAF2 associates with TRADD, and dominant negative TRAF2 can also inhibit TNFRI-and TRADD-mediated signals (44,48).
Expression of LMP1 and TRAF2 in transiently transfected 293 cells was verified by immunoblot analysis (Fig. 5, A and B). Immune complex kinase assays demonstrated that although TRAF2⌬(6 -86) alone had no effect on ATF2 phosphorylation, expression of this mutated TRAF2 protein reduced LMP1-mediated p38 kinase activity (Fig. 5C, first four lanes). This observation, coupled with the ability of TRAF2 overexpression to activate p38 MAPK (Fig. 5C, far right lane) suggests that TRAF2 is a mediator of LMP1-activated p38 signaling. Overexpression of TRADD also induced ATF2 phosphorylation in the presence of the anti-apoptotic cowpox virus crmA protein; but TRAF3, another LMP1-associated protein, had no significant effect (data not shown).
The p38 Pathway Regulates IL-6 and IL-8 Secretion in Response to LMP1 Expression-To identify the physiological consequences of p38 activation by LMP1, we have examined the effects of p38 inhibition on interleukin production. Previous data have demonstrated the importance of p38 signaling in TNF-mediated IL-6 synthesis (32-34), and we have shown recently that LMP1 expression induces IL-6 production via a pathway involving TRAFs (12).
HeLa cells were transiently transfected with pSG5-LMP1 or control vector, and condition supernatants were analyzed 36 h later using an IL-6-specific ELISA. These assays demonstrated that LMP1 induced a significant, 130-fold increase in IL-6 production compared with vector-transfected cells. However, in the presence of 20 M SB203580, a dramatic decrease in LMP1induced IL-6 secretion was observed (Fig. 6A). Immunofluores-cence staining for the detection of LMP1 expression in transfected cultures showed that more than 35% of the cells were positive for LMP1 in every experiment performed (data not shown).
The same supernatants were also analyzed for IL-8 production. Interestingly, we have found that LMP1 potently enhanced IL-8 protein synthesis, inducing an approximately 100fold increase in IL-8 secretion, and this effect was inhibited in the presence of nontoxic concentrations of SB203580 (Fig. 6B).
To examine the generality of the observed phenomenon, 293EcR-LMP1/cl.4 cells were induced to express LMP1 in the presence or absence of 20 M SB203580. Whereas untreated cells produced very little IL-8, treatment with ponasterone A dramatically increased IL-8 synthesis in 293EcR-LMP1 but not in control cultures (Fig. 6C). However, in the presence of p38 inhibitor, LMP1-mediated IL-8 secretion was reduced. The observed differences in the levels of IL-8 secreted from HEK 293 and HeLa cells is a cell line-dependent phenomenon rather than the result of differential LMP1 expression, as transient transfection of pSG5-LMP1 in 293 cells induces IL-8 production at levels comparable to ponasterone-treated cells (data not shown). Overall, these data suggest that the p38 pathway may influence the ability of LMP1 to induce both IL-6 and IL-8 production.
Involvement of p38 in the Transcriptional Control of IL-8 Synthesis-Because the ability of LMP1 to induce IL-8 production is novel, we have examined this phenomenon in more detail. We first determined whether LMP1 activates IL-8 at the transcriptional level. For this purpose, RNA isolated from HeLa cells transiently transfected with LMP1 or control vector was subjected to semiquantitative reverse transcription PCR using primers specific for IL-8 and using GAPDH as internal control. These experiments showed that IL-8 mRNA levels increased approximately 12-fold in response to LMP1 expression (Fig. 7A). Interestingly, treatment with 20 M SB203580 dramatically reduced IL-8 RNA levels in LMP1-transfected HeLa cells, suggesting that the p38 MAPK pathway modulates the ability of LMP1 to induce IL-8 synthesis at the transcriptional level.
To analyze further the effects of LMP1 on IL-8 transcription, direct studies on the IL-8 promoter were performed. A reporter plasmid (IL8-Luc) containing 181 bp of the IL-8 promoter sequences linked to the luciferase gene was constructed and transfected in HeLa and HEK 293 cells together with wild type or mutated LMP1 expression vectors. LMP1 expression was found to induce IL-8 promoter activity significantly in both HeLa (6.0 Ϯ1.4-fold increase) and 293 cells (12.6 Ϯ3.2-fold increase, Fig. 7B). Transfection of the CTAR1-expressing LMP1⌬(332-386) had only a marginal effect whereas expression of CTAR2 (constructs LMP1⌬(187-351) and LMP1 AxAxA ) significantly induced IL-8 promoter activity in both cell lines, in agreement with the function of CTAR2 as the predominant contributor of LMP1 signals. Transfection of HeLa or HEK 293 cells with LMP1 AxAxA /378STOP, which is inactive for both NF-B and p38, failed to induce IL-8 promoter activity above background levels (Fig. 7B).
Nucleotide sequence analysis of the 181-bp IL-8 promoter has revealed the existence of three cis-acting elements that have been shown to be functional in cell lines: a TGACTCA AP-1 binding motif 100 bp upstream the TATA box, an AGTT-GCAAAT C/EBP (NF-IL6) binding site 64 bp upstream of the TATA box, and a GGGAATTTCC NF-B binding site 52 bp upstream of TATA box (49). To examine the contribution of the AP-1 site to LMP1-mediated IL-8 promoter activity, a 2-nucleotide mutation within the AP-1 motif (TATCTCA), which abolishes transcription factor binding (49), was introduced in IL8-Luc. This mutated construct (mut AP-1/IL8-Luc) was cotransfected with pSG5-LMP1 or control vector into HeLa or HEK 293 cells and examined for luciferase activity. It was found that mutation of the AP-1 site from the IL-8 promoter significantly inhibited its activity in response to LMP1 expression by 50 -60% (Fig. 7C). These results suggest that AP-1 is a positive regulator of IL-8 promoter activation by LMP1. Unlike AP-1, mutations that abrogate C/EBP binding on IL-8 promoter had no effect on its activation after LMP1 expression (data not shown). A role for NF-B in IL-8 promoter regulation was demonstrated by the ability of a constitutively active IB␣ mutant (IB␣[S32A/S36A]), which has been shown previously to interfere with LMP1-mediated NF-B (24), to abolish LMP1induced IL8/Luc and mutAP-1/IL8-Luc reporter activity (Fig.  8). Thus, NF-B confers a potent positive regulatory role in IL-8 promoter activity.
To gain more insight into the composition of the protein complex bound to the IL-8 AP-1 consensus, we have performed EMSAs using nuclear extracts from vector or LMP1-transfected HEK 293 cells. As shown in Fig. 9A, LMP1 expression dramatically increased binding activity on the IL-8 AP-1 motif. The addition of anti-Fos antibody in nuclear extracts did not influence the mobility or intensity of binding, whereas the addition of anti-Jun/AP-1 antibody supershifted the complex. Exposure of nuclear extracts from LMP1-transfected HEK 293 cells to anti-ATF2 antibody and subsequent precipitation of the formed immune complex reduced the intensity of protein-DNA interaction by approximately 30 -40%, suggesting that ATF2 is a component of the complex bound to the IL-8 AP-1 site (Fig.  9A). Parallel depletion experiments using a control anti-c-rel antibody failed to induce any significant change in AP-1 binding (data not shown). To verify that ATF2 binds the IL-8 AP-1 consensus, EMSAs were performed using recombinant ATF2 protein. Previous studies have shown that ATF2 binds the Jun2 TRE sequence from the c-Jun promoter but not the collagenase promoter TRE (50,51). In agreement with these reports, recombinant ATF2 strongly interacted with a 32 P-labeled Jun2 TRE oligonucleotide (Fig. 9B, right lane) but did not bind the collagenase AP-1 sequence (Fig. 9B, left lane). Interestingly, recombinant ATF2 was found to interact with the IL-8 promoter AP-1 motif, although to less extent than with Jun2 TRE (Fig. 9B, center lane).
To determine whether the p38 MAPK pathway is involved in IL-8 promoter regulation, the p38 inhibitor SB203580 was used to treat HeLa or HEK 293 cells transfected with pSG5-LMP1 or control vector and IL-8 promoter constructs. It was found that p38 inhibition significantly reduced LMP1-mediated IL-8 promoter activity by 45-60% in both cell lines (Fig. 7C).

DISCUSSION
The EBV-encoded LMP1 has recently attracted much attention as it appears to function as a constitutively activated TNF family receptor. Indeed, LMP1 has been shown to interact with TRAFs and TRADD and to mimic many of the phenotypic consequences of CD40 or TNFR activation.
Expression of this viral protein induces a plethora of activities in target cells. These include the oncogenic transformation of rodent fibroblast cell lines, up-regulation of anti-apoptotic proteins and cell surface markers, cytokine production, and differentiation blockade in epithelial cells. Furthermore, LMP1 expression is essential for EBV-induced B cell immortalization in vitro. The signaling pathways that mediate these phenomena are a subject of intense investigation (for review, see Ref. 52). Previous studies have demonstrated that LMP1 expression leads to the rapid activation of the transcription factor NF-B, an effect mediated independently by two domains in the cytoplasmic COOH terminus of the protein: CTAR1 (amino acids 187-231) and CTAR2 (amino acids 351-386) (6,16). More recent reports indicate that LMP1 also mediates activation of a Ras/MAPK-dependent pathway (21) as well as the JNK/AP-1 cascade (22)(23)(24).
In this study we provide evidence of another signaling pathway activated by LMP1. Thus, we have shown that inducible LMP1 expression leads to activation of the p38 MAPK, as evidenced by the ability of LMP1 to induce p38 phosphorylation. Furthermore, p38 immunoprecipitated from LMP1-transfected cells induced phosphorylation of the transcription factor ATF2, one of the downstream targets of p38 MAPK. This phe-nomenon depends on oligomerization of LMP1 on the cell membrane because a CD2/LMP1 chimera, comprising the extracellular and transmembrane domain of CD2 and the cytoplasmic tail of LMP1, can induce p38 kinase activity only after antibody-induced aggregation of the chimera. Mutational analysis of the LMP1 cytoplasmic COOH terminus identified both TRAF-interacting CTAR1 and TRADD-binding CTAR2 domains contributing to p38 signaling. To exclude the possibility that induction of an autocrine loop is responsible for p38 activation, conditioned supernatants from vector or LMP1-transfected HEK 293 cells were used to treat untransfected HEK 293 cultures. These experiments showed that none of the conditioned media was able to induce ATF2 phosphorylation above background levels (data not shown), suggesting that LMP1 activates the p38 MAPK cascade directly.
Because LMP1 also engages NF-B signaling, experiments were performed to determine whether these two pathways run on the same axis or in parallel. We have found that inhibition of p38 MAPK by treatment with the highly specific inhibitor SB203580 did not affect LMP1-induced NF-B binding activity. Conversely, although the metabolic inhibitor D609 impaired LMP1-mediated NF-B activation, it did not inhibit signaling on the p38 axis, suggesting that these two LMP1-activated pathways can be primarily dissociated. This is in agreement with our previous work demonstrating NF-B and JNK signal divergence in response to LMP1 expression (24). However, we cannot exclude the possibility of LMP1-mediated NF-B transactivation being a target for p38. Indeed, a role for p38 MAPK in TNF-or CD40-induced NF-B transactivation in the absence of a direct effect on NF-B binding activity has been reported recently (33,41,53), and our preliminary results using a luciferase reporter plasmid containing B elements from the Ig enhancer (12) suggest that SB203580 induces a small inhibition in LMP1-mediated NF-B transcriptional activity.
Our data demonstrate the involvement of TRAF2 in LMP1mediated p38 activation. The ability of TRAF2 also to engage signaling on the NF-B axis suggests that bifurcation of NF-B and p38 signals occurs downstream of this adaptor protein.
Indeed, in this study we have shown that transient overexpression of TRAF2 significantly induced p38 kinase activity and that a dominant negative TRAF2⌬(6 -86) molecule that has been shown previously to interfere with LMP1-mediated NF-B partially inhibited p38 activation.
The functional implications of LMP1-mediated p38 activation have been demonstrated by the ability of SB203580 to down-regulate LMP1-activated IL-6 and IL-8 production. These cytokines play an important role in initiation and maintenance of acute inflammatory responses. IL-8 is a potent chemotactic agent and may contribute to the accumulation of a T cell infiltrate in EBV-positive malignancies such as nasopharyngeal carcinoma and Hodgkin's disease, where LMP1 is expressed. In addition, IL-8 has been shown to induce angiogenesis and haptotactic migration and to enhance metastatic potential in melanoma cells (54,55). These data coupled with the recently reported ability of LMP1 to induce up-regulation of matrix metalloproteinases (56) suggest that in addition to its transforming potential, LMP1 may contribute to metastasis of EBV-associated tumors. The effects of p38 MAPK on LMP1mediated IL-8 production were found to occur at the transcriptional level as treatment of HeLa cells with SB203580 significantly inhibited up-regulation of IL-8 RNA. This is in agreement with the reported ability of SB203580 to block TNFmediated IL-6 RNA and protein synthesis (32,40).
To gain more insight into this novel LMP1-mediated phenomenon, we have examined the effects of LMP1 expression on IL-8 promoter and found that transfection of HeLa or HEK 293 cells with this viral protein significantly enhanced promoter activity. Elements important for IL-8 promoter regulation are localized within the first 135 bp of the 5Ј-flank and include an AP-1 and a NF-B binding site (49). LMP1 is known to engage signaling on the NF-B axis, and inhibition of this pathway by coexpression of a constitutively active IB␣[S32A/S36A] mutant abolished wild type or AP-1-mutated IL-8 promoter activity. Thus, LMP1-mediated NF-B activation can positively regulate IL-8 activity.
Furthermore, we have found that LMP1 expression induced nuclear protein binding on the AP-1 element of IL-8 promoter, whereas mutations within this element significantly impaired LMP1-mediated IL-8 transcriptional induction. Taken together, these data support a positive regulatory role for LMP1activated AP-1-bound proteins in IL-8 synthesis. Using EM-SAs, we have identified c-Jun and ATF2 proteins as components of the bound complex. Binding sites for c-Jun/ ATF2 have been found in several promoters including those of c-Jun, ␤-interferon, E-selectin, and the human urokinase enhancer (50,51,57). A role for p38 in IL-8 promoter regulation is demonstrated by the inhibitory effect of SB203580 on LMP1induced IL-8 transcriptional activity. However, the ability of LMP1 to induce c-Jun phosphorylation through activation of JNK suggests that this kinase pathway may also contribute to modulation of IL-8 expression. A similar complex regulation of the E-selectin promoter by JNK, p38, and NF-B has been reported recently (57).
Overall, in this study we demonstrate the ability of a viral protein, namely EBV-encoded LMP1, to activate the p38 MAPK pathway. This phenomenon appears to be important for IL-6 and IL-8 production and may play a significant cooperative role in regulating additional LMP1 activities.