Characterization of Intercellular Adhesion Molecule-1 Regulation by Epstein-Barr Virus-encoded Latent Membrane Protein-1 Identifies Pathways That Cooperate with Nuclear Factor k B to Activate Transcription*

The latent membrane protein-1 (LMP1) of Epstein-Barr virus induces gene transcription, phenotypic changes, and oncogenic transformation. One cellular gene induced by LMP1 is that for intercellular adhesion molecule-1 (ICAM-1), which participates in a wide range of inflammatory and immune responses. ICAM-1 may enhance the immune recognition of cells transformed by Epstein-Barr virus, and thus combat development of ma-lignancy. Despite growing understanding of the various signaling functions of LMP1, the molecular mechanisms by which LMP1 induces ICAM-1 are not understood. Here, we demonstrate that transcriptional activation by LMP1 is absolutely dependent upon a variant NF- k B motif within the tumor necrosis factor a (TNF a ) response element of the ICAM-1 promoter. Although the TNF a response element is sufficient for TNF a induction of the ICAM-1 promoter, LMP1 also required the cooperation of additional upstream sequences for optimal induction. Inhibitor studies of known LMP1-induced signaling pathways ruled out the involvement of c-Jun N-terminal kinase (JNK), p38 mitogen-activated protein kinase, and the Janus-activating tyrosine kinase 3 (JAK3), and confirmed NF- k B as a critical factor for induction of ICAM-1. However, although constitutive activation of NF- k B efficiently induced promoter activity, it was not sufficient to induce either ICAM-1 mRNA or ICAM-1 protein. Using signaling defective LMP1 mutants

the major oncogene of Epstein-Barr virus (EBV), a persistent herpesvirus that is associated with various malignant diseases (1). LMP1 transforms rodent fibroblasts (2), induces lymphomas in transgenic mice (3), and is essential for EBV-induced immortalization of human primary B-lymphocytes (4). The oncogenic properties of LMP1 are at least in part due to the up-regulation of anti-apoptotic proteins such as Bcl-2, A20, Mcl-1, and Bfl-1 (5)(6)(7)(8). In addition, LMP1 up-regulates components of the endogenous antigen processing pathway and some intercellular adhesion molecules, such as ICAM-1 and LFA-3 (9,10). These latter functions ensure that EBV-transformed lymphocytes can be recognized and regulated by cellular immune responses, which is an important feature of EBV persistence in healthy individuals that normally prevents the development of EBV-positive lymphomas (11).
Consistent with its diverse biological functions, LMP1 has been reported to trigger a number of different signaling pathways, including activation of NF-B (5,12), the mitogen-activated protein kinases, JNK and p38, leading to activation of AP-1 and ATF-2 transcription factors (13)(14)(15), and the JAK3/ STAT1 pathway (16). LMP1 mimics a ligand-independent, constitutive active receptor of the tumor necrosis factor receptor (TNFR) superfamily by binding TNFR-associated factor and TNFR-associated death domain (TRADD) to effect its signaling functions (17,18). The important regions of LMP1 are the so-called C-terminal activator regions (CTAR-1, -2, and -3), which initiate the signaling function of LMP1 by binding signaling molecules and which need to cooperate together for optimal function (19). CTAR1 induces NF-B and is probably involved in the p38 pathway; CTAR2 signals through NF-B, the p38 pathway, and the JNK pathway; and CTAR3 binds JAK3 (13,16,19).
The intercellular adhesion molecule, ICAM-1 (CD54), is an inducible cell surface glycoprotein and member of the immunoglobulin supergene family (20,21). ICAM-1 serves as a counterreceptor for a number of cell surface molecules such as LFA-1 (CD11a) and MAC-1 (CD11b), and it plays a central role in a wide range of inflammatory and immune responses (22). ICAM-1 is constitutively expressed at low levels on vascular endothelium and lymphocytes and at moderate levels on monocytes. Induction of high levels of ICAM-1 occurs in response to various inflammatory mediators, including bacterial lipopolysaccharide, phorbol esters, oxidant stress and pro-inflammatory cytokines, such as TNF␣, interleukin-1␤ (IL-1␤), and ␥-in-terferon (␥-IFN) (23,24). Furthermore, certain viruses (e.g. rhinovirus, respiratory syncytial virus, and EBV) are also known to up-regulate ICAM-1 surface expression (11,25,26). With regard to EBV, the LMP1 protein is the major effector of ICAM-1 up-regulation (27). The mechanism by which LMP1 effects up-regulation of ICAM-1 is poorly understood, but studies on the other stimuli of ICAM-1 suggest that all of the transcription factors activated by LMP1 are potential regulators of ICAM-1 transcription.
The 5Ј-flanking region of the ICAM-1 gene contains numerous potential regulatory elements that could be involved in the activation of the promoter, some of which are tissue-specific and cytokine-dependent (23). For example, ␥-IFN and TNF␣ have been shown to mediate ICAM-1 induction at the level of transcription, using different signal transduction pathways and specific activator sites (28 -30). The interferon response element maps to Ϫ76 bp to Ϫ66 bp, whereas the TNF␣-responsive element (TRE) ranges from Ϫ227 bp to Ϫ177 bp. The TRE was shown to be both necessary and sufficient to induce ICAM-1 promoter activation by TNF␣, and it critically required the binding of the NF-B family of transcription factors, specifically p65 homodimers, to a variant B-site (30). However, flanking sequences surrounding this B binding site are also required for transcription factor binding and transactivation in TNF␣-mediated induction of ICAM-1 (31). The ICAM-1 TRE also contains Sp-1 and C/EBP binding sites located upstream of the modified NF-B site, and binding of C/EBP homo-or heterodimers to the C/EBP site was reported to be necessary for maximal induction of ICAM-1 by TNF␣ (32).
The studies described here were designed to define the molecular mechanisms involved in LMP1-induced ICAM-1 expression in lymphocytes. We carried out promoter deletion analysis and luciferase reporter assays to investigate the role of LMP1 at the transcriptional level and to determine the activator site within the ICAM-1 promoter. Inhibition of specific signaling pathways induced by LMP1 was achieved using chemical inhibitors for the p38 and JAK3-pathways, and dominant inhibitory molecules for the SEK and NF-B pathways. The importance of these individual pathways for LMP1-induced ICAM-1 promoter and protein expression at the cell surface was thus determined and revealed a hitherto unrecognized function of LMP1 to activate ICAM-1 surface expression. Finally, we will show that this novel function cooperates with one of the Cterminal activator regions of LMP1 and is essential for optimal ICAM-1 induction.

MATERIALS AND METHODS
Cell Lines-Jurkat is a cell line derived from an EBV-negative T cell lymphoma (33). Eli-BL is an EBV-positive B cell line established from a Burkitt's lymphoma, and it displays a latency I form of infection in which Epstein-Barr virus nuclear antigen 1 is the only viral protein detected (34). DG75 is an EBV-negative Burkitt's lymphoma B cell line, and the derived DG75-tTA-LMP1 line contains a stable transfected tetracycline-regulated LMP1 expression plasmid; this transfectant, together with the control DG75-tTA transfectant, has been described previously (35). All the lymphoid cell lines were grown in suspension in RPMI, 10% fetalm calf serum supplemented with 2 mM glutamine and antibiotics (200 units/ml penicillin and 200 g/ml streptomycin), and were maintained at 37°C in a humidified atmosphere with 5% CO 2 . The DG75 transfectants were maintained in 1 g/ml tetracycline and were drug-selected with 0.8 mg/ml hygromycin B plus 2 mg/ml G418 (DG75-tTA-LMP1) or with 0.8 mg/ml hygromycin B only (DG75-tTA).
Plasmids and Inhibitors-The ICAM-1 reporter constructs containing 5Ј regions upstream of the ICAM-1 gene in front of a luciferase gene were obtained from Harry C. Ledebur and have been described elsewhere (30). The 3Enh-luc reporter plasmid, with three B elements upstream of a minimal conalbumin promoter driving the expression of the firefly luciferase gene (36), and the p(IL-6B) 3 plasmid, containing three copies of the IL-6 promoter B site in front of a luciferase gene (37), were used to assay NF-B activity. Plasmid pSG5-LMP1 expresses wild-type LMP1 cloned from the B95.8 strain of EBV and has been described previously (38). The LMP1 mutant CTAR1 Ϫ plasmids replace amino acids Pro 208 , Gln 210 , and Thr 212 with alanines, and the CTAR2 Ϫ mutant replaces Tyr 384 with glycine; these mutants were described elsewhere (14,39). The green fluorescent protein expression plasmid, pEGFP-C1, was purchased from CLONTECH, and the pMEKK1-expressing plasmid was from BioLabs.
A constitutively active IB␣/GFP fusion protein vector was generated by amplifying the IB␣ gene from the pCMV IB␣⌬N plasmid (kindly provided from Dean W. Ballard, Howard Hughes Medical Institute, Nashville, TN) using a forward 5Ј-primer, which binds to the base pairs after the codon 36, and a 3Ј-primer lacking the stop codon of the protein. The purified PCR fragment was then cloned into the BglII site of the pEGFP-C1 plasmid to generate plasmid EGFP-IB␣DN. The hemagglutinin-tagged kinase vectors, HA-p46SAPKy-pcDNA3 and HA-p38, and the dominant inhibitor SEKDN vector (40) were provided by Aristides Eliopoulos (CRC Institute, Birmingham, United Kingdom). The pyridinyl imidazole SB20380 (Calbiochem), a specific inhibitor of the p38 MAPK pathway, was prepared as a 20 mM stock solution in dimethyl sulfoxide and a JAK3 inhibitor (Calbiochem) was also prepared in dimethyl sulfoxide as a 25 g/ml stock solution. Both inhibitors were added to the cultures at a dilution of 1/1000. The deacetylation inhibitor sodium butyrate (Sigma) was used in a final concentration of 1 mM.
Gene Transfection-For transient expression, 0.5 to 1 ϫ 10 7 cells from a suspension culture were transfected by electroporation using a Bio-Rad GenePulser II electroporator at 280 V and 950 microfarads at room temperature in 500 l of growth medium. The cells were reseeded in 5 ml of fresh growth medium and were then incubated under normal conditions. Transfection efficiency ranged from 10% to 20% for Eli-BL and from 40% to 50% for Jurkat, as assessed by cotransfection with the EGFP-C1 expression vector and flow cytometry analysis.
Assay for Reporter Activity-The activity of the different reporter plasmids was measured at 18 -24 h after transfection. Cells were washed twice in phosphate-buffered saline and lysed in 150 l of lysis buffer containing 100 mM HEPES, pH 8.0, 2 mM magnesium chloride, 5 mM dithiothreitol, and 2% Triton X-100. Luciferase activity in 50 l of clarified lysate was analyzed in a Berthold LB9501 luminometer following injection of 100 l of 0.5 mM luciferin (Amersham Pharmacia Biotech) dissolved in luciferin assay reagent (30 mM glycylglycine, pH 7.9, 1 mM MgCl 2 , 0.1 mM EDTA, 30 mM dithiothreitol, 0.3 mM coenzyme A, 0.5 mM ATP). Light release was integrated over 10 s.
Assay for Cell Surface ICAM-1 Protein by Flow Cytometry-The induction of ICAM-1 protein in transfected cells was assayed by immunofluorescence staining of viable cells, followed by flow cytometry using a Becton Dickinson FACSCalibur analyzer as described previously (19). Briefly, at 48 h after transfection, the cells were washed and stained with a phycoerytherin-conjugated monoclonal antibody to human CD54 (MCA675PE; Serotec) at 4°C for 60 min. The transfected population was marked by the expression of cotransfected EGFP-C1 plasmid, and this population was gated for analysis of ICAM-1 staining.
Detection of Proteins by Immunoblotting -Cells were washed in phosphate-buffered saline and lysed for 30 min on ice in luciferase lysis buffer. The lysates were centrifuged for 5 min at 13,000 ϫ g. An aliquot of the clarified lysate was added to an equal volume of 2ϫ gel sample buffer (0.1 M Tris buffer, pH 6.8, 0.2 M dithiothreitol, 4% sodium dodecyl sulfate, 20% glycerol, 0.1% bromphenol blue) and boiled for 2 min. The solubilized proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane for immunoblotting using an alkaline phosphatase chemiluminescent detection protocol (41). LMP1 was detected by first incubating the membranes for 1 h with 1 g/ml CS. 1-4 (42) in I-Block (Tropix Inc.) followed by incubating for 1 h with a 1/10,000 dilution of alkaline phosphatase-conjugated goat anti-mouse IgG (Bio-Rad 170-6461). The IB and EGFP-IBDN proteins were detected with 1 g/ml rabbit polyclonal antibodies to IB␣ (Santa Cruz sc371) followed by an alkaline phosphatase-conjugated goat anti-rabbit IgG (Bio-Rad 170-6518). Specific antibody-protein complexes were detected using CDP-Star (Tropix Inc.) development reagent.
Kinase Assays-JNK and p38 in vitro kinase assays were performed as described elsewhere (40). Briefly, at 48 h after transfection with either HA-p46SAPKy-pcDNA3 or HA-p38, cells were lysed in 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 dithiothreitol). An aliquot of 250 g of protein extract was used for immunoprecipitating HA-p38 or HA-JNK using 1 g of anti-HA antibody (Roche Molecular Biochemicals; 12CA5) on Sepharose beads. The washed immunoprecipitates were then subjected to a kinase reaction, and phosphorylation was determined by Western blot. In the JNK assay, phosphorylation of Jun was determined by immunoblot analysis using phospho-c-Jun (Ser 63 ) antibody (New England Biolabs 9261), and in the p38 assay phosphorylation of ATF2 was determined using phospho-ATF2 (Thr 71 ) antibody (New England Biolabs 9221).
RNase Protection Assay-Total RNA was isolated from DG75 cells using Ultraspec solution (Biotecx) according to manufacturer's recommendations. Quantification and purity of RNA was assessed by A 260 / A 280 absorption, and an aliquot of 15 g of RNA from each sample was used in the RNase protection assay.
The assay was performed using the Riboquant Multiprobe RNase protection assay system as described by the manufacturer (PharMingen). The customized Multiprobe ICAM1 template set and T7 RNA polymerase were used to synthesize 32 P-labeled antisense riboprobes, which were hybridized to the RNA sample in solution at 54°C for 16 h. After digestion of free probe and other single-stranded RNA with RNases A and T1, the labeled "RNase-protected" fragments were phenol:chloroform:isoamyl alcohol-extracted and ethanol-precipitated. The fragments were than resolved on 6% denaturing polyacrylamide gels and detected by autoradiography or using a phosphorimager (Bio-Rad). Approximately 5000 cpm of "unprotected" labeled probe served as a reference to identify the fragments.
To investigate ICAM1 mRNA levels, control plasmid and LMP1-or MEKK1-expressing plasmids were transiently transfected together with rCD2GFP into DG75 cells. After 48 h, transfections were pooled and living positive transfected cells were stained with the rat CD2specific antibody, Ox34 (American Tissue Culture Collection), and sorted using MACS immunobeads (Miltenyi Biotec) according to manufacturer's recommendations. Total RNA was extracted from cells and analyzed for ICAM1 mRNA expression by RNase protection assay.

LMP1 Induces ICAM-1 Surface Expression and Promoter
Activation-LMP1 up-regulates ICAM-1 protein expression in various cell lines. This is illustrated in Fig. 1A, showing the results of flow cytometry analysis of ICAM-1 expression on Jurkat cells transfected with a control vector or with the SG5-LMP1 expression vector. The cells were cotransfected with a GFP expression vector so that the transfected population could be identified and gated to allow analysis of ICAM-1 expression following staining of the cells with phycoerytherin-conjugated CD54 antibodies. In this representative experiment, the vector control-transfected cells showed basal ICAM-1 surface expression with a mean fluorescence intensity of 29.4 arbitrary units (  Fig. 1B, induction of ICAM-1 in Jurkat cells LMP1 caused a maximal 3-fold increase in ICAM1protein expression. In separate independent experiments, the maximal induction of ICAM-1 by LMP1 in Jurkat cells typically ranged between 2-and 6-fold at 48 h after transfection. In Eli-BL cells (Fig. 1C), a similar magnitude of ICAM-1 induction by LMP1 was observed, but the dose-response curve differed in that Eli-BL cells were responsive to as little as 0.1 g of SG5-LMP1 plasmid, whereas Jurkat cells required 5 to 10 times more plasmid to elicit a similar response. Nevertheless, the optimal induction of ICAM-1 surface expression in both cell lines was achieved at a plasmid concentration of between 1 and 3 g of SG5-LMP1, which was used in subsequent experiments.
Since LMP1 is known to activate transcription factors, we investigated the effects of LMP1 on the ICAM-1 promoter activation. The 5Ј region of the ICAM-1 gene is well described, and the transactivator sites and responsive regions for different members of the TNF␣ receptor family have been identified (30,43). Only 1381 base pairs of the 5Ј-ICAM-1 gene region are required for the ICAM-1 induction by the proinflammatory cytokines TNF␣, Il-1␤, and ␥-IFN. Fig. 2A shows a schematic structure of the promoter region with the potential transcription factor binding sites, and the characterized TNF␣ response region (TRE). The nuclear transcription factors involved in the regulation of ICAM-1 promoter include: Ap1, NF-B, C/EBP, Ets, STAT, and Sp1. The locations of their binding sites in the ICAM-1 5Ј-regulatory region are shown, together with the mapped AP2 and AP3 sites, and the translational start sites (CAT/TATA boxes).
As a member of the TNFR superfamily, LMP1 might be predicted to activate ICAM-1 using the TRE region within the ICAM-1 promoter. To test this possibility, we used a series of luciferase reporters regulated by different regions of the ICAM-1 promoter (30). The results of a representative experiment are shown (Fig. 2B) in which the luciferase plasmids were cotransfected with SG5-LMP1 into Eli-BL (black bars) and Jurkat (white bars) cells, and the reporter activities measured at 24 h after transfection. LMP1 induced the ICAM-1 reporters with a similar profile in both cell lines. The maximal luciferase induction of 5-6-fold was only obtained with the full-length 1.3ICAM1-luc construct. Mutants deleted for 779 or 1126 5Ј bp in the regions upstream of the TRE site (0.5-ICAM1 and TRE-ICAM1, respectively) showed reduced activation levels but were still induced 3-4-fold by LMP1. In contrast, neither the reporter mutant deleted for the TRE (delTRE-ICAM1) nor the mutant deleted for 1248 5Ј-bases (0.1-ICAM1) was induced by LMP1. It should be noted that the nonresponsive 0.1-ICAM1reporter contained the intact interferon response element. These results suggest that LMP1 uses the same essential region as TNF␣ to activate the ICAM-1 promoter.
Comparison of the Effects of LMP1 and TNF␣ on the ICAM-1 Promoter-Since the TRE has been shown to be essential and sufficient for TNF␣-induced up-regulation of ICAM-1 (30), we tested mutant reporters with or without this region to investigate the similarities between LMP1 and TNF␣. We analyzed the luciferase activity of 1.3ICAM1, TRE-ICAM1, and delTRE-ICAM1 (see Fig. 2A) together with a full-length reporter in which the variant B-site in the TRE had been inactivated by point mutation (NF-Bneg-ICAM1). In one representative experiment shown in Fig. 3, cells transfected with each of these reporters were cotransfected either with SG5-LMP1 or with SG5 vector; the SG5 vector transfectants were then treated with TNF␣ at 12 h after transfection and for 6 h prior to harvesting. The results in Fig. 3 show that LMP1 induced the TRE-ICAM1-luc reporter only to 58% of the full-length promoter activity, whereas TNF␣ stimulated the TRE-reporter to 140% of the full-length promoter activity. The reduced inducibility of the TRE-ICAM-1 reporter by LMP1 suggests that, in contrast to TNF␣ stimulation, the TRE region is not sufficient for optimal induction of ICAM-1 by LMP1.
The results with the delTRE-luc reporter confirmed that the TRE is essential for induction of ICAM-1 promoter activity both by LMP1 and by TNF␣. Within this region, the variant NF-B motif has been reported to be critical for transcriptional activation of ICAM-1 by TNF␣ (30), and we therefore tested the importance of this site for LMP1-induced ICAM-1 promoter activation. Using the B-negative mutant ICAM1 reporter, we always observed an increased background promoter activity of about 2-fold above the basic luciferase reporter activity, which suggests that this functional B site regulates basal ICAM-1 levels as well as the inducibility. However, the B-negative reporter was not induced further either by LMP1 or by TNF␣ (Fig. 3). Taken together, the results in Fig. 3 indicate that LMP1 differs from TNF␣ in utilizing regions upstream of the TRE to maximize the inducibility achieved from the TRE, and that within the TRE the variant B site is critical for induction both by LMP1 and by TNF␣.
Activation of NF-B Is the Major Event in LMP1-induced Up-regulation of ICAM-1-We wanted to further investigate the importance of the LMP1-induced NF-B pathway in regulating ICAM-1. Therefore, we examined the effect of the NF-B To enable endogenous and transfected IB␣ to be distinguished, we designed a constitutive active IB␣ that was deleted for the first 36 amino acids (thus removing the two regulatory phosphorylation sites) and fused to GFP (EGFP-IB␣DN; see Fig. 4A). Expression of this construct following transfection into Jurkat cells was determined by immunoblotting with a rabbit anti-IB␣ antibody (Fig. 4B). In Jurkat cell extracts, this antibody always detected endogenous IB␣ as low molecular mass bands, between 39 and 44 kDa, and the transfected EGFP-IB␣DN as a higher molecular mass band, between 68 and 70 kDa, of similar intensity to the endogenous IB␣. The inhibitory function of EGFP-IB␣DN on LMP1-induced NF-B signaling was analyzed in reporter assays using the B-dependent luciferase reporters 3Enh-luc and IL-6(B) 3luc, which contain triple repeats of B response elements from the Ig promoter and the IL-6 promoter, respectively. EGFP-IB␣DN was fully functional, being able to inhibit LMP1-induced activation of the 3Enh-luc reporter by 93%, and the IL-6(B) 3 -luc reporter by 94% (Fig. 4C).
Having confirmed the effectiveness of EGFP-IB␣DN as an inhibitor of the LMP1-induced NF-B pathway, we investigated its effect on LMP1-induced ICAM-1 up-regulation. Increasing amounts of LMP1 with or without EGFP-IB␣DN were transfected into Jurkat lymphocytes, and the effects upon luciferase activity of the full-length 1.3ICAM-1 promoter reporter (Fig. 5A) as well as ICAM-1 protein expression (Fig. 5B) were measured. The results show that LMP1-induced ICAM-1 promoter activity was completely inhibited by EGFP-IB␣DN at low doses of SG5-LMP1 (Յ1 g), and was inhibited by about 80% at the highest input doses (2 and 4 g) of SG5-LMP1. It should be noted that the constitutive SV40 promoter of the SG5-LMP1 plasmid itself is not completely unaffected by EGFP-IB␣DN. However, although EGFP-IB␣DN reduced LMP1 expression by up to 50% at the lowest doses of SG5-LMP1, at the higher doses, there was no significant effect on LMP1 expression (data not shown). The flow cytometry analysis of the cell surface ICAM-1 protein expression in the same transfected cell population revealed that LMP1 induction of ICAM-1 protein is completely abolished in the presence of the NF-B inhibitor. These experiments were also performed with the Eli-BL B cell line and showed similar results (data not shown).
Activation of NF-B Is Essential but Not Sufficient for ICAM-1 Induction-For confirmation that NF-B is the major activator of ICAM-1 expression, we tested the effects of a constitutive activated form of MEKK1, a potent inducer of the NF-B pathway. The active kinase expression plasmid, when transfected into Jurkat cells, was shown to induce the 3Enh-luc NF-B dependent luciferase reporter by 110-fold (data not shown). Therefore, we transfected MEKK1 with and without SG5-LMP1 into Jurkat cells and measured its effect on the ICAM-1 reporter activation as well as on the expression of cell surface ICAM-1 protein. In the representative experiment shown in Fig. 6, MEKK1 induced the ICAM-1 full-length promoter (1.3ICAM-1) 465-fold over background level, which is 63 times higher than the LMP1-induced reporter luciferase activity (Fig. 6A). In contrast, flow cytometry analysis of cell surface ICAM1 protein expression (Fig. 6B) revealed that, although the cells transfected with LMP1 showed a substantial 10-fold increase in ICAM-1 mean fluorescence intensity, the cells transfected with MEKK1 alone showed no induction of ICAM-1 expression. Cotransfection of LMP1 and MEKK1 showed that the MEKK1did not interfere with the ability of LMP1 to upregulate ICAM-1 protein (Fig. 6B). These results show that NF-B activation alone is sufficient for activation of the ICAM-1 reporter, but is not sufficient to effect up-regulation of ICAM-1 protein. This suggests that other signaling pathways of LMP1, in addition to NF-B, are required to up-regulate ICAM-1 protein expression.
Other Signaling Pathways Known to Be Induced by LMP1 Are Not Required to Induce ICAM-1 Protein-In addition to NF-B, the known LMP1-induced signaling pathways include the JNK pathway, which leads to activation of the c-Jun transcription activator; the p38 pathway, resulting in ATF2 translocation; and a JAK3/STAT pathway. The importance of these individual pathways in LMP1-mediated up-regulation of ICAM-1 was analyzed using specific inhibitors. To inhibit the JNK pathway, we used SEKDN, a dominant negative form of JNK kinase that is deleted for the phosphorylation site and is therefore unable to phosphorylate the Jun-activating kinase, JNK (13). To inhibit the p38 pathway, we used a chemical inhibitor SB203580, which specifically blocks p38 phosphorylation and translocation without affecting other signaling pathways (44). A chemical inhibitor was also used to specifically inhibit JAK3 (45). These inhibitors were all shown to affect their respective targets without affecting the expression of LMP1 from the SG5-LMP1 vector (Fig. 7A and data not shown). The inhibition of LMP1-mediated JNK activation with SEKDN was less efficient than was the inhibition of other pathways (Fig. 7A), but the 60 -70% inhibition was similar to that previously reported by other workers (40). Having confirmed the functionality of the inhibitors, we analyzed their effects on LMP1-induced ICAM-1 protein (Fig. 7B). As a positive control for inhibition of ICAM-1 induction, we cotransfected EGFP-IB␣DN together with SG5-LMP1 in Jurkat cells. The results shown demonstrate that, although LMP1-induced up-regulation of ICAM-1 protein was completely inhibited by EGFP-IB␣DN, the other inhibitors tested did not affect the upregulation of ICAM-1 protein.
MEKK1 Cannot Induce ICAM-1-mRNA-Since LMP1 and MEKK1 induce a similar subset of signaling pathways and MEKK1 very efficiently induces ICAM-1 promoter but was not able to induce ICAM-1 protein, we wanted to investigate those differences in more detail. The inability of MEKK1 to induce ICAM-1 surface expression suggests that LMP1 regulates additional pathways. To further analyze the level at which LMP1 and MEKK1 signal differed, we investigated the effects of MEKK1 on ICAM-1 mRNA levels. DG75 cells were transfected with MEKK1 or LMP1 together with rat CD2 to identify the transfected cells. Cells were incubated at 37°C for 24 h to allow expression. The transfected cells were stained with OX34 monoclonal antibody to rat CD2, and were separated by immunomagnetic beads. This resulted in greater than 90% purity of rat CD2-positive cells as assayed by fluorescence-activated cell sorting (data not shown). Total RNA from the purified cells was isolated and mRNA for ICAM-1 was assayed by RNase protection using specific ICAM-1 probes. The results of these experiments are shown in Fig. 8. The upper panel shows a graphic representation of a single experiment and clearly shows that LMP1 can up-regulate ICAM-1 mRNA (compare second lane with first lane). However MEKK1 (third lane) could not induce ICAM-1 mRNA. The lower panel shows the average mRNA levels from the phosphorimager analysis of three different experiments. Although LMP1 induced ICAM-1 mRNA 2-3-fold, MEKK1 had no effect on the ICAM-1 mRNA levels. These data suggest that MEKK1 lacks a key event required for the effective transactivation of the ICAM-1 gene.
CTAR1 and CTAR2 Provide Qualitative Different Signals to ICAM-1-LMP1 regulation of ICAM-1 clearly requires as yet uncharacterized signals in addition to NF-B. The full nature of these signal remains elusive. In the case of LMP1 signaling, the cooperation of the C-terminal activator regions CTAR1 and CTAR2 has been shown to be essential for LMP1 function (19). Furthermore, histone acetylation of genes and their promoters was identified to be involved in coordinate regulation of transcription (46). We wanted to test if these mechanisms could also be important for LMP1-induced ICAM-1 up-regulation. Therefore, we investigated two point mutants of LMP1, defec-tive for either one (CTAR1 Ϫ ) or the other (CTAR2 Ϫ ) C-terminal activator region and MEKK1 in the presence of a chemical compound that has been shown to inhibit histone deacetylation and thus prevent gene silencing. Eli-BL cells were transfected with control or LMP1-, CTAR1 Ϫ -, CTAR2 Ϫ -, and MEKK1-expressing plasmid together with EGFP-C1 to control for transfection. After 24 h the cells were stimulated with sodium butyrate, and, after an additional 12-h incubation, the ICAM-1 surface expression was analyzed using flow cytometry. The ICAM-1 surface expression of these transfected cells is shown in Fig. 9 as mean values of three independent experiments. LMP1 induces ICAM-1 expression as seen before (column 2 compared with column 1). The CTAR1 Ϫ mutant (column 3), which has intact CTAR2 and CTAR3 domains, induced only half of the ICAM-1 protein compared with LMP1, and the CTAR2 Ϫ mutant with intact CTAR1 and CTAR3 induced even less (column 5). However, after incubation with the deacetylation inhibitor sodium butyrate, only the CTAR1 Ϫ -transfected cells showed ICAM-1 expression comparable to wild type LMP1 levels (column 4 compared with column 2), whereas the ICAM-1 levels of the CTAR2 Ϫ transfected cells remained low. In the case of the MEKK1-transfected cells, the ICAM-1 expression levels showed no significant difference between cells treated with or without the deacetylation inhibitor. The ICAM-1 surface expression in MEKK1-transfected cells remained at basal levels (columns 7 and 8).
These data suggest that the deacetylation inhibitor, sodium butyrate, can cooperate with CTAR2 and -3 to generate LMP1 wild type ICAM-1 expression. Therefore, we conclude that LMP1 induces a novel additional signal or pathway to induce ICAM-1 protein, which can be mapped to the CTAR1 region of LMP1. DISCUSSION Although it has been recognized for some time that the EBV-encoded LMP1 is responsible for the up-regulation of ICAM-1 (10), the mechanism of activation was unclear. The biological and signaling properties of LMP1 share many features with the pro-inflammatory cytokines TNF␣ and ␥-IFN (9, 16 -18), both of which transcriptionally regulate ICAM-1 expression through distinct regulatory elements in the ICAM-1 promoter. Our results show that LMP1 acts primarily through the TRE located at Ϫ178 bp to Ϫ227 bp, but not the interferon response element located at Ϫ76 bp to Ϫ66 bp. However, LMP1 and TNF␣ are subtly different in their regulation of ICAM-1 (Fig. 3), which highlights the complex range of signaling path-ways and transcription factors involved in ICAM-1 activation. Consistent with this observation, although a similar range of signal transduction factors are known to associate with LMP1 and TNFR1, and an overlapping spectrum of signaling pathways are activated, there are important differences. For example, the death domain at the C terminus of TRADD is involved in TNFR1/TRADD interaction, whereas LMP1 recruits TRADD via its N-terminal domain, and the different topology of TRADD binding affects the mechanisms by which LMP1 and TNFR1 activate NF-B and JNK (47,48).
A direct comparison of LMP1 and TNF␣ signaling (Fig. 3) showed that the TRE is both essential and sufficient for TNF␣induced ICAM-1, as reported previously (30); in contrast, this site was essential but not sufficient for optimal induction by LMP1 (Figs. 2 and 3). Taken together, these data suggest that LMP1 might differ from TNF␣ by targeting additional transcription activation site(s) located upstream of the TRE between Ϫ574 bp and Ϫ1381 bp. In this respect, potential sites include an AP1 binding site at position Ϫ1284 and a second NF-B site at position Ϫ531. Other upstream regulatory enhancer elements have also been described, which are thought to be more important for constitutive ICAM-1 expression rather than for inducible ICAM-1 (29), although it is possible that they have a role in LMP1 induction of ICAM-1.
Whatever the role of upstream elements, the TRE appears to be critically involved in inducible regulation of ICAM-1 by various stimuli. Within the TRE, a critical feature is the variant NF-B binding site, which is essential for transcriptional up-regulation of ICAM-1 mediated by LMP1 (Fig. 3), TNF␣ (30), and other cytokines (24). The critical role for NF-B in LMP1-mediated up-regulation of ICAM-1, suggested by our analysis with mutant reporters, was supported by complementary experiments showing that a constitutively active IB␣ efficiently abolished LMP1-mediated up-regulation both of the full-length 1.3ICAM-1 reporter and expression of ICAM-1 protein (Fig. 5). Our results shed new light on an area of confusion since two previous studies have reported the use of IB␣ to inhibit the ability of LMP1 to up-regulate ICAM-1 protein expression, but there was disagreement about the efficiency of this effect (49,50). Our data, in line with those of Devergne and colleagues (50), show that efficient blocking of NF-B activation can completely abrogate the ability of LMP1 to up-regulate ICAM-1 protein.
Despite the essential role for NF-B, it is clear that activation of this transcription factor alone is unable to induce ICAM-1 protein expression. Thus, we have observed previously that transfection and overexpression of p50 and p65 NF-B species has no effect upon ICAM-1 protein expression in lymphoid cells. 2  whose effects include activation of NF-B besides AP-1 and p38 activation (51,52), is able to activate ICAM-1 transcription more efficiently than does LMP1 but does not affect ICAM-1 protein expression (Fig. 6). These results implicate other LMP1-activated signal pathways in up-regulating ICAM-1 protein. However, our experiments with specific inhibitors mitigate against a role for the Jun/AP-1, p38/ATF-2, and JAK3/ STAT pathways (Fig. 7). It could be argued that, because the inhibition of the JNK/AP-1 pathway by the dominant-negative SEKDN molecule was not as efficient as the other inhibitors used in this study (Fig. 7), it is possible that the residual LMP1-induced AP-1 activation is sufficient to cooperate with NF-B to up-regulate ICAM-1 protein. However, this is contra-dicted by the observations (i) that the SEKDN inhibitor did not even partially inhibit up-regulation of ICAM-1 (Fig. 7), and (ii) that MEKK1, which activates both AP-1 and NF-B, was unable to up-regulate ICAM-1 protein. With regard to the lack of affect of SB203580 inhibition of p38 on LMP1-induced ICAM-1, it is noteworthy that this inhibitor has been shown to abrogate two other biological functions of LMP1, the induction of IL-6 and IL-8 (40). During the preparation of this report, the activation of Cdc42 (a small GTPase) in fibroblasts was identified as yet another signaling function for LMP1 (53). However, this function was reported to be mediated by the transmembrane domains of LMP1 (53), whereas activation of ICAM-1 requires only domains on the C-terminal cytosolic regions of LMP1 (19). Therefore, we are drawn to conclude that another, as yet uncharacterized, signaling pathway(s) cooperates with NF-B to control ICAM-1 regulation by LMP1. The ability of NF-B to activate the ICAM-1 reporter but not ICAM-1 protein expression (Fig. 6) is intriguing. There are a number of possible factors that could be involved, including mRNA stabilization, promoter accessibility regulated by nucleosomes, and various post-transcriptional mechanisms. With regard to mRNA stability, both phorbol esters and ␥-IFN have been shown to stabilize the otherwise labile ICAM-1 mRNA in murine fibroblast and monocytic cell lines (54,55). Furthermore, stabilization of mRNA has been shown to be a feature of LMP1-mediated up-regulation of Bfl-1 in lymphocytes by as yet unknown mechanisms (8). However, using the same human lymphoid cell model (BJAB cells transfected with a tetracycline-regulated LMP1 expression vector) that was used to demonstrate Bfl-1 mRNA stabilization, we found that LMP1 does not affect the stability of ICAM-1 mRNA (data not shown).
The major difference regarding MEKK1 and LMP1 lies in the ability to induce mRNA, since LMP1 induces ICAM-1 mRNA 2-3-fold but MEKK1-transfected cells have no increased ICAM-1 mRNA levels (Fig. 8). Thus, the LMP1-specific effect seems to act at the level of transcription of the endogenous ICAM-1 gene, which is not completely reflected in the luciferase reporter assays. Signal-regulated acetylation events and their effects on gene transcription are recognized as another mechanism of gene activation (46,56,57). We have shown ( Fig.  9) that the deacetylation inhibitor, sodium butyrate, can substitute for a function of LMP1, which locates to the CTAR-1 region. This identifies a new mechanism for activation of ICAM-1 by LMP1. The precise nature of this activation process is not clear, because sodium butyrate has other effects on cells, such as regulation of proliferation (58), and induction of apoptosis (59), which may be due to mechanisms in addition to its effects on histone deacetylation. Furthermore, this new function of CTAR1, which can be substituted by sodium butyrate, cannot alone explain our data. Thus, although activation of NF-B was the only known signaling function of LMP1 that was shown to be critical for induction of ICAM-1 protein by LMP1 (Fig. 7), sodium butyrate was unable to cooperate with the NF-B activated by the LMP1-CTAR2 Ϫ mutant or by MEKK1 (Fig. 9). There are several possible explanations for this; one being that there is yet another unknown signaling function of LMP1. The precise analysis of these new features of LMP1 signaling is ongoing.
In summary, the present study demonstrates that NF-B is a key signaling pathway stimulated by LMP1 in the transcriptional regulation of ICAM-1, and that the other known signaling pathways activated by LMP1 appear not to be critical for up-regulation of ICAM-1. However, NF-B activation is not sufficient by itself to up-regulate ICAM-1 protein. We described a new cooperating signaling mechanism induced by LMP1, which may act at the level of promoter accessibility and maps to the CTAR1 domain of LMP1. Further characterization of this pathway and the detailed analysis of the signal-regulated acetylation event and gene transcription are the next critical steps toward our understanding of LMP1 function.