Mechanism of action of a novel latent membrane protein-1 dominant negative.

Latent membrane protein-1 (LMP1) is a signaling molecule expressed by Epstein-Barr virus during latency. LMP1 is essential for B-cell immortalization by Epstein-Barr virus and transforms rodent fibroblasts. It activates many distinct signaling pathways including the transcription factors NFkappaB and AP1. We have generated a mutant of LMP1 with four point mutations; amino acids 204, 206, and 208 were mutated to alanine, and amino acid 384 was mutated to glycine. This mutant, termed LMP1(AAAG), is not only unable to activate nuclear signaling pathways, but also inhibits signaling from wild type LMP1. We have demonstrated the effectiveness, selectivity, and mechanism of this inhibitory molecule. It inhibits LMP1-stimulated NFkappaB, STAT, and Jun transcriptional activity. It is selective, as it does not inhibit TNF or interleukin-2 signaling. We have demonstrated that it does not sequester the downstream signaling molecule, TRAF2, but instead binds LMP1 and interferes with its ability to bind TRAF2. This demonstrates the importance of the interplay between the signaling domains of LMP1 and the oligomeric structure of LMP1 for effective signaling. It identifies a tool that will be useful to probe LMP1 function in disease.

. LMP1 also plays a role in the immunogenicity of EBV by up-regulating a number of proteins involved in immune regulation such as the TAP-1 and TAP-2 peptide transporter components of the endogenous antigen processing pathway, major histocompatibility complex class I, and the intercellular adhesion molecules, ICAM-1 and LFA-3 (11,12).
LMP1 is a 63-kDa integral membrane protein in which three signaling domains have been characterized (Fig. 1A). The protein has six hydrophobic transmembrane segments and is thought to spontaneously oligomerize to stimulate intracellular pathways. The three signaling domains are entitled C-terminal activating region-1 (CTAR1), CTAR2, and CTAR3 (13)(14)(15). CTAR1 has been shown to bind the complex of cellular proteins belonging to the family of tumor necrosis factor receptor-binding proteins (TRAFs) (16 -18). CTAR2 binds TRADD and other signaling molecules (19,20). Both of these domains can activate NFB transcriptional activity independently but they function optimally together (21). Likewise, activation of the p38 kinase pathway is also mediated via both CTAR1 and CTAR2 (22). In contrast, LMP1 activates the JNK kinase pathways via CTAR2 (23,24). The activation of NFB and p38 have been shown to be required for a number of LMP1-regulated genes, demonstrating the importance of these pathways (22,25). Recently a third signaling domain of LMP1 was described and termed CTAR3 (14). This has been suggested to bind Jak 3 and activate the DNA binding of STAT1 (14). No function has been ascribed to the activation of the Jak-STAT pathway by LMP1, and the interaction of CTAR3 with the other signaling domains of LMP1 remains to be elucidated.
The functional domains of LMP1 were initially identified by deletional analysis. Recently, more detailed analysis has identified point mutations of critical residues that can abolish functions. By comparison of the LMP1 sequence with the functional domains of CD40, a PXQXT motif in CTAR1 was identified and substitution of the critical Pro 204 /Gln 206 /Thr 208 residues to alanines abolished TRAF binding and NFB activation from the CTAR1 of LMP1 (16,26,27). In the CTAR2 region, substitution of a critical Tyr 384 residue completely abolished TRADD binding, and the activation of NFB and AP1 (24,28,29). The critical residues in CTAR3 for Jak3 binding have not been identified.
This project was instigated to test the hypothesis that mutation of LMP1 could yield not only a nonsignaling form of the protein but also generate a dominant inhibitor. We have generated such a mutant, LMP1 AAAG , in which both CTAR1 and CTAR2 were inactivated by four point mutations: P204A, Q206A, T208A, and Y384G. This report describes the key features of the LMP1 AAAG mutant which enhance our understanding of the mechanism of LMP1 signaling. First, this mutant is unexpectedly defective in the ability to activate a STAT-regulated reporter, a function that was shown to be regulated by CTAR3. Second, the LMP1 AAAG mutant was found to function as a dominant inhibitory molecule when coexpressed with wild type LMP1. The efficiency, specificity and mechanism of the dominant negative molecule have been examined.

EXPERIMENTAL PROCEDURES
Cell Culture-Jurkat is a cell line derived from an EBV negative T cell lymphoma (30). 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 (31). DG75 is an EBV-negative Burkitt's lymphoma B cell line. Kit225 is a human leukemic cell line (32) that is dependent upon IL-2 for growth. The Kit225 cells were deprived of IL-2 for 24 h prior to transfection by washing the cells twice in phosphate-buffered saline. All the lymphoid cell lines were grown in suspension in RPMI 1640 medium supplemented with 10% fetal calf serum, 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 .
Plasmids-Plasmid pSG5-LMP1 expresses a wild type LMP1 cDNA derived from the B95.8 strain of EBV, under the control of the SV-40 promoter of the SG5 vector. The pSG5-LMP1⌬(1-128) encodes an LMP1 protein deleted for the first 128 amino acids and corresponds to the truncated from of LMP1 (tr.LMP1) expressed during lytic cycle in B95.8 cells (13). The pSG5-LMP1 AAAG mutant was constructed in two steps. First, a pSG5-LMP1 AAA mutant was generated by site-directed mutagenesis, (23), to mutate the codons 204 -208 from PXQXT to AXAXA. The LMP1 gene was then polymerase chain reaction-amplified from the pSG5-LMP1 AAA mutant as a BglII fragment using a variant 3Јprimer (GAAGATCTTTAGTCATAGCCGCTTAGC) to mutate codon 384 from Tyr to Gly, and was cloned into the BglII site of pSG5 to create pSG5-LMP1 AAAG . The presence of all the correct mutations was confirmed by sequencing. The same polymerase chain reaction product was also cloned into pEGFP-C1 (CLONTECH) to generate the pEGFP/ LMP1 AAAG plasmid encoding a mutant LMP1 AAAG protein with an N terminus GFP tag. Generation of the plasmid pSG5-CD2/LMP1 has been reported previously (29). The TRAF2 expression vectors were kindly provided by Elliott Kieff (17), pcDNA3-TRAF2⌬(6 -86) encodes a dominant negative mutant form of TRAF2 that retains the ability to bind LMP1 but is unable to affect signaling functions such as activation of NFB.
The reporter plasmid, 3Enh-luc (33) was used to assay NFB activity. The STAT reporter, GRR luciferase is the luciferase analog of the GRRCAT reporter described previously (34). The lex-Jun fusion protein plasmid (from Dr. R. Treisman, ICRF, London) is a fusion of codons 3 to 199 of the bacterial protein, LexA and codons 1 to 194 of c-Jun, which contains the transactivation domain of Jun. The lex reporter has been described previously (35).
Transfections, Reporter, and ICAM-1 Assays-For transient expression, 0.5 to 1.5 ϫ 10 7 cells from a suspension culture were transfected by electroporation using a Bio-Rad Genepulser II electroporator at 270 or 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.
The activity of the luciferase reporter plasmids was measured 24 h post-transfection using a standard bioluminescence protocol (21). For chloramphenicol acetyltransferase assays, cells were lysed in buffer containing 10 mM Tris (pH 8.0), 1 mM EDTA, 150 mM NaCl, 0.65% Nonidet P-40. Samples were then assayed for CAT activity by radioisotope method (34).
The induction of ICAM-1 protein in transfected cells was routinely assayed by immunofluorescence staining of viable cells, followed by flow cytometry using a Becton Dickinson FACSCalibur analyzer as described previously (21). Briefly, at 48 h post-transfection the cells were washed and stained with a phycoerythrin-conjugated monoclonal antibody to human CD54 (MCA675PE; Serotec) at 4°C for 60 min. The transfected population was marked by the expression of co-transfected EGFP-C1 plasmid (CLONTECH), and the GFP-positive population was gated for analysis of ICAM-1 staining.
Immunoprecipitations and Western Blotting-For each immunoprecipitation, 15 ϫ 10 6 cells of the DG75 line in 0.5 ml of growth medium were electroporated at 270 V and 950 microfarads with up to 20 g of plasmid DNA as indicated in the Results section. At 24 h post-transfection, DG75 B-cells were washed twice with phosphate-buffered saline, then lysed for 45 min on ice in 800 l of Nonidet P-40-lysis buffer (0.5% Nonidet P-40, 50 mM HEPES buffer, 0.25 M NaCl, 2 mM EDTA) to which a 5% protease inhibitor mixture (Sigma; P8340) was added immediately prior to use. The lysates were clarified by centrifugation at 13,000 ϫ g for 5 min, and the soluble fraction was then precleared for 4 h with 20 l of Sepharose-Protein G or with Sepharose-Protein G to which 1 g of control antibody had been covalently cross-linked. The precleared lysate was then incubated for 4 h with 20 l of Sepharose-Protein G to which 1 g of OX34 or a mixture of 0. For Western blotting analysis performed in parallel with reporter assays, cells were washed in phosphate-buffered saline and lysed for 30 min on ice in luciferase lysis buffer and processed as described previously (36). Typically, the lysates from 2 ϫ 10 5 cells was applied to each track of the gel. LMP1 was detected with the mouse monoclonal antibodies CS.1-4 used at 1 g/ml, or with a 1/1000 dilution of a rabbit polyclonal serum raised to a ␤-galactosidase fusion protein containing 189 amino acids of the C terminus of LMP1 (37). TRAF2 was detected with rabbit polyclonal antibodies reactive with a C terminus epitope corresponding to amino acids 478 -497 (Santa Cruz; sc876) and was used at 1 g/ml. All primary antibody incubations were for 90 min at room temperature.
Microscopy-For LMP1 staining, DG75 cells were stained with a mouse monoclonal antibody to LMP1, LMPO25 (10 g/ml), using standard protocols with Texas Red-conjugated anti-mouse IgG (EY Laboratories) as a secondary antibody. For two color analysis of GFP-tagged LMP1 AAAG and CD2-tagged wild type LMP1, CD2 staining was performed using OX34 (10 g/ml) and detected with Texas Red conjugate as described previously (38). The cells were visualized on a Leica TCS4D confocal microscope. A Z series was performed and equatorial region was illustrated in all cases.

RESULTS
Generation of a Nonfunctional LMP1 Mutant, LMP1 AAAG -To create a nonsignaling LMP1, a form of LMP1 was generated in which four of critical residues in CTAR1 and CTAR2 are mutated. A schematic of this mutant, called LMP1 AAAG , is shown in Fig. 1A. Fig. 1A also shows a schematic of a rare but naturally occurring truncated LMP1 (tr.LMP1) that can be expressed in a limited number of strains of EBV, and is postulated to inhibit LMP1 func- tion (39). These two mutants and a wild type form of LMP1 were expressed in DG75 B-cells and analyzed by Western blotting (Fig.  1B). The molecular weight of the LMP1 AAAG is the same as wild type LMP1 (compare second and third lanes of Fig. 1B), whereas the truncated form (Fig. 1B, 4th lane) runs at a reduced molecular weight.
The ability of LMP1 AAAG to activate LMP1 signaling pathways was tested. Jurkat cells were co-transfected with a range of amounts of an expression vector for wild type LMP1 or LMP1 AAAG together with an NFB reporter construct (3enh luciferase) that contains three copies of an NFB-binding site upstream of the luciferase gene. After 18 h the cells were lysed and a luciferase assay was performed. Fig. 2A shows that while wild type LMP1 signals strongly to NFB, LMP1 AAAG does not activate NFB reporter activity. This experiment was also performed in two B-cell lines, Eli-BL cells and DG75, and an epithelial cell line, 293. LMP1 AAAG was not able to stimulate NFB transcriptional activity in any of these lines (data not shown).
It was possible that mutation of CTAR1 and CTAR2 would only affect signaling to pathways activated by CTAR1 and CTAR2. For this reason we tested the ability of LMP1 to trigger STAT transcriptional activity as LMP1 effects on STAT signaling have recently been mapped to a new domain entitled CTAR3 (Fig. 1A). A GRR reporter assay was chosen to test for STAT transcriptional activity. This is a well characterized reporter construct based on the STAT-binding site from the Fc␥R promoter. It has been shown to bind various STATs (34,40) and has been used as a reporter in many different studies (34,38). Eli-BL cells were co-transfected with the GRR reporter (10 g) together with various amounts of both wild type LMP1 and LMP1 AAAG expression vectors. Fig. 2B shows that while wild type LMP1 was able to stimulate GRR luciferase transcriptional activity, LMP1 AAAG showed no stimulatory capacity for this reporter. This was somewhat surprising and suggests that either CTAR1 or CTAR2 are required for the activation of STAT transcriptional activity.
Protein expression of LMP1 AAAG was examined to demonstrate that the mutant expressed at a similar level to wild type. Fig. 2C shows the results from the expression of wild type and LMP1 AAAG in Eli-BL cells. Cells were transfected with a range of amounts of an LMP1 expression vector, cells were lysed, and cellular proteins were resolved by SDS-PAGE. Western blot analysis was performed with an anti-LMP1 antibody (CS.1-4) and similar levels of both wt LMP1 and LMP1 AAAG were seen (Fig. 2C). Similar data were also obtained with DG75 and 293 cells (data not shown).
LMP1 AAAG Acts as a Dominant Negative Molecule-Since the LMP1 AAAG could not activate LMP1 signaling pathways, it was postulated that it may be able to inhibit wild type LMP1 activity by preventing the formation of a functional signaling complex. NFB transcriptional activity was tested by co-transfection of the reporter with various amounts of an expression vector for wild type LMP1 (from 0.1 to 4 g) in combination with a constant 5 g of LMP1 AAAG . Fig. 3A shows that expression of LMP1 AAAG dramatically inhibits the activation of NFB by wild type LMP1. A ratio of 1 g of wild type LMP1 plasmid to 5 g of LMP1 AAAG plasmid inhibited NFB transcriptional activity by greater than 90%. Over a number of experiments, the expression of equal levels of LMP1 AAAG to wild type LMP1 inhibited signaling by 50 -75%, demonstrating the efficiency of the mutant. A dose response of LMP1 AAAG (0 -12 g) was tested with a fixed amount of wild type LMP1 expression vector (2 g) (Fig. 3B). This shows that the inhibition correlates with amount of LMP1 AAAG .
A naturally occurring truncated LMP1, tr.LMP1, which lacks amino acids 1-127, has been postulated to inhibit LMP1 signaling (39). We compared the LMP1 AAAG mutant with the putative inhibitory effects of the truncated from of LMP1 (Fig.  3C). A direct comparison of the two molecules suggests that a much lower level of expression of LMP1 AAAG is efficient relative to tr.LMP1 (compare 3rd bar with 5th bar and 4th bar with 6th bar). In this experiment, the co-transfection of 3 g of LMP1 AAAG expression vector inhibited wild type LMP1 activation of NFB by greater than 90%, whereas 3 g of the expression vector for tr.LMP1 had very little effect.
LMP1 AAAG Inhibits STAT Transcriptional Activity and Jun Transactivation-Two other nuclear signals stimulated by LMP1 were also tested. The STAT reporter (GRR luciferase) was tested in Eli-BL cells (Fig. 4A). This showed that FIG. 2. LMP1 AAAG cannot activate nuclear signals. Jurkat cells were transfected with a range of concentrations of wild type LMP1 (solid squares) or LMP1 AAAG (empty circles) expression plasmids together with the NFB reporter, 3enh luciferase (A), or a STAT reporter, GRR luciferase (B). Samples were harvested and the luciferase activity was detected as described under "Experimental Procedures." C, expression of wt LMP1 and LMP1 AAAG was determined by Western blotting with the CS.1-4 mixture of monoclonal anti-LMP1 antibodies. LMP1 AAAG effectively inhibited STAT transcriptional activity stimulated by LMP1. A Jun-transactivation assay was also performed (Fig. 4B). Jurkat cells were co-transfected with a mammalian expression vector for a chimeric protein with the DNA-binding domain from the bacterial lex protein and the transactivation domain of c-Jun (amino acids 1-194) (41) together with a reporter that contains two binding sites for lexprotein upstream of the CAT gene (35). This reporter therefore gives a measurement of Jun phosphorylation in vivo, which requires the activity of Jun N-terminal kinase (JNK) or p38 stress activated protein kinase. When co-transfected with this reporter, LMP1 could induce Jun transactivation, and this activation was inhibited by coexpression of the LMP1 AAAG (Fig. 4B).
LMP1 AAAG Inhibits Induction of ICAM-1 by LMP1-It is important to demonstrate that the LMP1 AAAG can also block downstream biological functions of LMP1, i.e. endogenous protein changes, and not just affect reporter constructs. Fig. 5 demonstrates that LMP1 AAAG efficiently inhibits the LMP1mediated induction of the ICAM-1 (CD54) adhesion molecule. Jurkat cells were co-transfected with a GFP vector and a wild type LMP1 plasmid. The GFP vector allows the identification of transfected cells by flow cytometry. The cells were left for 48 h and expression of ICAM-1 was determined using phycoerythrin-conjugated CD54 antibodies and flow cytometric analysis. Fig. 5A (top panel) shows the expression of ICAM-1 in Jurkat cells transfected with EGFP and control vector DNA. When Jurkat cells were transfected with the GFP-LMP1 AAAG expression vector no increase in ICAM-1 levels was detected demonstrating that LMP1 AAAG could not signal effectively (Fig.  5A, 2nd panel). In contrast, cells transfected with wild type LMP1 and EGFP expression vectors showed a dramatic increase in ICAM-1 expression (Fig. 5A, 3rd panel). This can be represented in two ways. The first consists of determining the percentage of cells whose fluorescence intensity crosses an arbitrary threshold gate. In the example shown in Fig. 5A, expression of wt LMP1 caused an increase from 2.7 to 34.2% of ICAM-1 positive cells. The effects upon ICAM-1 can also be determined by measuring the increase in mean fluorescence intensity which, in Fig. 5A, shows a 3-fold increase in the level of expression of ICAM-1 caused by expression of wt LMP1 (control transfectant, mean fluorescence intensity of 68; wt LMP1 transfectant, mean fluorescence intensity 195). When wild type LMP1 and GFP-LMP1 AAAG expression vectors were co-transfected, no increase in either the percentage of positive cells or the mean fluorescence intensity was detected (Fig. 5A,  compare top and bottom panels). A dose response of LMP1 plasmid was performed in the presence and absence of a constant amount of LMP1 AAAG . Whether the data are expressed as mean fluorescence intensity (Fig. 5B) or as a percentage of ICAM-1 positive cells (not shown), LMP1 AAAG dramatically inhibits LMP1 signaling. These results demonstrate that LMP1 AAAG can inhibit endogenous cellular gene expression in addition to synthetic reporter genes.
LMP1 AAAG Is Selective as It Does Not Inhibit TNF or Interleukin-2 Signaling-The usefulness of a dominant-negative molecule is enhanced if it is also specific for its intended target. We therefore examined whether LMP1 AAAG might interfere with the signaling pathways activated by other receptor molecules. The selectivity of the LMP1 AAAG inhibitor was tested in two different systems. Jurkat cells were transfected with the 3enh (NFB) luciferase reporter and increasing amounts of the LMP1 AAAG vector. The cells were left overnight to ensure expression, and were then stimulated with TNF. The results in Fig. 6A show that doses of LMP1 AAAG which inhibit wt LMP1 signaling by more than 75% had no effect on TNF activation of NFB. The LMP1 AAAG molecule retains the putative Jak3binding site (CTAR3) and while it appears unable to signal to STAT transcriptional activity, it could theoretically act to sequester Jak3 and thus inhibit Jak 3-dependent signaling from other receptors. This possibility was tested in Kit225 cells, a leukemic cell line that is stimulated by the cytokine, IL-2. In these cells, IL-2 utilizes Jak3 and activates STAT5 and STAT3, which have been previously shown to activate the GRR reporter (34). The Kit225 cells were transfected with the LMP1 AAAG expression plasmid, and the ability of IL-2 to activate the GRR reporter was tested. Fig. 6B shows that IL-2 was fully functional in the presence of similar levels of LMP1 AAAG that were shown to inhibit wt LMP1 signaling. This provides evidence for the specificity of LMP1 AAAG and suggests that it acts proximal to the LMP1 molecule itself rather than binding or sequestering signaling machinery that may be utilized by other signaling pathways.
LMP1 AAAG Cellular Localization Is Indistinguishable from that of Wild Type LMP1-All the mutants of LMP1 utilized in this study were expressed in DG75 B-cells and their cellular localization was investigated by staining with the LMP1 antibody, LMPO25. Cells were transfected with 5 g of each expression vector and left to express overnight. The cells were fixed with 2% paraformaldehyde and acetone and then stained as described previously. The staining was visualized with a Leica confocal microscope and a Z series was acquired. Fig. 7A shows the staining of wild type LMP1, LMP1 AAAG , and tr.LMP1. The staining of wild type LMP1 produced a characteristic pattern, with an accumulation of LMP1 into a "cap" at one end of the cell. The pattern of staining produced by LMP1 AAAG was indistinguishable from wild type LMP1. In The similarity of the wild type LMP1 and LMP1 AAAG was striking and was confirmed by expression of GFP-tagged forms of both proteins. This prevents any artifacts due to fixation and allowed the visualization of LMP1 in live cells. The results obtained (data not shown) were very similar to that seen by staining with LMPO25 (Fig. 7). To see if wild type LMP1 and LMP1 AAAG could colocalize, constructs that could be distinguished had to be utilized. We used a fusion protein comprising the extracellular and transmembrane domains of rat CD2 tagged to the N terminus of wild type LMP1, and a GFP-tagged LMP1 AAAG . Fig. 7B shows the image of an equatorial region of a DG75 cell which had been co-transfected with expression vectors for the CD2-WTLMP1 and LMP1 AAAG -GFP. The cells were fixed and stained for rat CD2 using the OX34 antibody and a Texas Red secondary antibody. GFP was visualized in the fluorescein isothiocyanate channel and a specific channel for Texas Red was used to visualize CD2-tagged LMP1. A comparison of the two pictures shows an overlapping distribu-tion of the two proteins. Taken together, the results suggest that both wild type LMP1 and LMP1 AAAG normally localize to the same subcellular region and that, if coexpressed in the same cell, may physically interact.
LMP1 AAAG Binds Wild Type LMP1-The images of wild type LMP1 and LMP1 AAAG together with the specificity of LMP1 AAAG suggested an hypothesis that LMP1 AAAG may be forming a complex with wild type LMP1. This complex may inhibit the signaling of wild type LMP1. This hypothesis was tested by investigating whether wild type LMP1 could bind LMP1 AAAG by performing co-precipitation experiments. The CD2-tagged wtLMP1 expression vector was transfected alone or together with either LMP1 AAAG or tr.LMP1 vectors into DG75 B cells, and cell lysates were immunoprecipitated with a specific CD2 antibody (OX34). The presence of CD2-wtLMP1, LMP1 AAAG , and tr.LMP1 in the immunoprecipitates was assayed by Western blotting with a rabbit polyclonal antibody to LMP1. The results are illustrated in Fig. 8. The anti-CD2 antibody specifically precipitated CD2-wtLMP1 (lane 1) but did not pull-down LMP1 AAAG (lane 4) or tr.LMP1 (lane 5) when these LMP1 molecules were expressed alone. However, inspection of the immunoprecipitates in the second lane shows that when LMP1 AAAG is coexpressed with CD2-wtLMP1, LMP1 AAAG does indeed form a complex with the wild type protein. In contrast, the truncated LMP1 (third lane) does not complex with the wild type protein. The lower panel shows the input lysates used for the immunoprecipitations and shows that similar levels of LMP1 AAAG and tr.LMP1 were expressed. These results demonstrate that LMP1 AAAG will form a complex with wild type LMP1.
LMP1 AAAG Prevents the Recruitment of TRAF2 to Wild Type LMP1-A complex between wild type LMP1 and LMP1 AAAG may be unable to bind signaling molecules. To investigate this possibility, co-immunoprecipitation experiments were performed to investigate the ability of the TNF adaptor molecule, TRAF2, to bind LMP1. TRAF2 is postulated to be a major component of the LMP1 signaling machinery. DG75 cells were transfected with vectors for wild type LMP1, LMP1 AAAG , and the truncated LMP1 (tr.LMP1), in the presence of TRAF2. Cells were left overnight to allow expression. The cells were then lysed and the LMP1 was immunoprecipitated with antibodies FIG. 6. LMP1 AAAG does not affect TNF or interleukin-2 signaling. A, Jurkat cells were transfected with an NFB reporter with and without wt LMP1 and LMP1 AAAG constructs as indicated. The cells were incubated overnight to allow expression, and one set of cultures which had not been transfected with wt LMP1 vector were instead stimulated with TNF (10 ng/ml) for 6 h. Samples were then assayed for luciferase. B, Kit225 cells were cultured in the absence of IL-2 for 24 h and then co-transfected with 10 g of STAT reporter (GRR luciferase) and either SG5 empty vector or SG5-LMP1 AAAG as indicated. Replicate cultures were either stimulated overnight with IL-2 (20 ng/ml) or were left untreated before assaying for luciferase.  -hand panel). Localization of the constructs was investigated after overnight expression by staining using LMPO25 antibody with a Texas Red anti-immunglobulin antibody conjugate. Cells were visualized by confocal microscopy. B, DG75 were co-transfected with an expression vector for a rat CD2 wild type LMP1 fusion protein and a GFP-tagged LMP1 AAAG expression vector (right-hand panel). Cells were left to express the constructs overnight. The CD2 tagged was visualized using OX34 antibody and a Texas Red anti-immunglobulin antibody conjugate. Cells were visualize by confocal microscopy.
CS-3 and CS-4. These immunoprecipitates were resolved by SDS-PAGE and the presence of TRAF2 in the immunoprecipitates was detected by Western blotting. The results from a representative experiment are illustrated in Fig. 9A. These show that TRAF2 binds to wild type LMP1 but binds at a substantially reduced level to both LMP1 AAAG and tr.LMP1. Experiments were performed to determine whether LMP1 AAAG could inhibit the recruitment of TRAF2 to wild type LMP1. This was done by co-transfecting a range of amounts of a vector for GFP-tagged LMP1 AAAG (EGFP-LMP1 AAAG ) with a constant amount of wild type LMP1 (3 g) and TRAF2 (6 g) vectors. Cells were lysed and LMP1 was immunoprecipitated as before. Fig. 9B shows that LMP1 AAAG prevents the interaction of TRAF2 with wild type LMP1 (top panel). The bottom panel (Fig.  9B) shows the expression of both wild type LMP1 and the GFP-tagged LMP1 AAAG .
Clearly, LMP1 AAAG interferes with the ability of wild type LMP1 to bind TRAF2 in the cellular context. It is possible that this is a result of LMP1 AAAG sequestering wild type LMP1 to a different subcellular location, away from TRAF2, While this appears unlikely from the results in Fig. 7, we conducted additional experiments to examine whether LMP1 AAAG interfered with the ability of wild type LMP1 to bind TRAF2 in vitro. In this set of experiments, LMP1 molecules were first precipitated from lysates of transfected DG75 B-cells and were then incubated with TRAF2-containing lysates from separately transfected DG75 cells. Fig. 9C shows that the while immune precipitates of wild type LMP1 can effectively bind exogenously added TRAF2, the complex of wild type LMP1 with LMP1 AAAG could not bind TRAF2. Together these immunoprecipitation experiments suggest a mechanism whereby LMP1 AAAG binds to wild type LMP1 and prevents the recruitment of TRAF2. This prevents LMP1 signal transduction to the nucleus. DISCUSSION This study describes a novel inhibitor of the oncogenic protein LMP1. It characterizes the efficacy, selectivity, and mechanism of the inhibitor. The efficacy was tested by inhibition of reporter constructs and ICAM-1 protein expression changes induced by LMP1. It was shown to be both nonfunctional and to be dominant inhibitory when expressed at similar levels to wild type LMP1. It was shown to be selective as it does not inhibit TNF or IL-2 signaling. Finally an investigation into the mechanism demonstrates that it does not sequester downstream signaling molecules but binds wild type LMP1 and inhibits the binding of at least one signaling molecule, TRAF2.
All described previously, mutants of LMP1 that are unable to activate NFB have had deletions of either CTAR1 or CTAR2. LMP1 AAAG is the first mutant that is unable to activate NFB and contains only point mutations. This reduces the likelihood of dramatic structural effects that can result from amino acid deletions. It strengthens the hypothesis that all the signals required for NFB activation lie completely within CTAR1 and CTAR2.
While in terms of NFB, this is not surprising, it is still an important confirmation of our expectations. However, the fact that proper functioning of CTAR1 and CTAR2 is required for GRR luciferase activity, and thus STAT activation, was unexpected since Gires et al. (14) recently described CTAR3 as a Jak-binding domain. Our result does not necessarily contradict the results from Gires et al. (14) but casts them in a new light. It suggests that while CTAR3 may be a distinct signaling motif of LMP1, it does not function in isolation to activate STATs. The mechanism of this activation is intriguing and is currently the subject of investigation. Gires et al. (14) measured STAT1 DNA binding whereas this present study has measured STAT transcriptional activity. STAT proteins have been shown to be modulated by both tyrosine and serine phosphorylation (34,40,42) and the Jak kinases can only phosphorylate tyrosine residues. The serine phosphorylation of STATs has been shown to be important for transcriptional activity but not DNA binding. Distinct signaling pathways have been shown to be utilized by both IL-2 (34) and interferon signaling (42) for the regulation of STAT transcriptional activity. Thus, while CTAR3 may control STAT tyrosine phosphorylation, CTAR1 or CTAR2 may be required for other signaling pathways necessary for STAT transcriptional activity.
The interplay between the domains of LMP1 has been postulated previously. It has been shown that for optimal functions, CTAR1 and CTAR2 must be in the same signaling complex (21). The inhibitory effects of LMP1 AAAG further elucidates the importance of cooperation between LMP1 molecules. It is apparent that this cooperation is not just of a structural nature but also requires the cooperative binding of signaling molecules. We have shown that LMP1 AAAG does not prevent normal cellular localization of LMP1 (Fig. 7). Furthermore, a complex of wild type LMP1 and LMP1 AAAG is unable to bind TRAF2, a key signaling molecule, both in vitro and in whole cells (Fig. 9). This suggests that the cooperation between the domains of LMP1 is required between molecules for the regulation of gene expression events. Furthermore, this cooperation acts at the level of the recruitment of receptor proximal signaling molecules and not at the level of cooperating nuclear signals, demonstrating the importance of the formation of an intact signaling complex. TRAF2 has been shown to bind the TNF receptor, to which LMP1 has been compared, as a trimeric FIG. 8. LMP1 AAAG binds wild type LMP1. Western blots, probed with a rabbit antiserum to LMP1, indicating the LMP1 molecules co-immunoprecipitating with wild type LMP1. For each immunoprecipitation, 15 ϫ 10 6 DG75 cells were co-transfected with 10 g each of the SG5 expression plasmids as indicated, and lysates were prepared 24 h post-transfection. The wild type LMP1 was expressed as a CD2/ LMP1 fusion protein which was precipitated with OX34 monoclonal antibody to rat CD2. complex. If it binds LMP1 using a similar mechanism, it is likely that the presence, in a signaling complex of a protein that cannot bind TRAF2 will inhibit TRAF binding to other members of the signaling complex. There may also be interplay between CTAR3 and one or both of the other signaling domains of LMP1. Since STAT regulation is proposed to be controlled by CTAR3, but LMP1 AAAG , in which CTAR3 is intact, cannot signal to the STAT reporter, some cooperation between CTAR1 or CTAR2 and CTAR3 must exist.
LMP1 AAAG was substantially more efficient than the naturally occurring truncated from of LMP1 (tr.LMP1) (Fig. 3). Furthermore, localization and immunoprecipitation experiments, which characterized the mechanism of LMP1 AAAG , show tr.LMP1 to function very differently. Truncated LMP1 does not oligomerize with wild type LMP1 (Fig. 8 and Ref. 39) and it has substantially reduced TRAF2 binding (Fig. 9) suggesting that it cannot sequester this signaling molecule. Our data suggests that whatever mechanism tr.LMP1 utilizes, it is less efficient than that of LMP1 AAAG .
The regulation of the Jak-STAT signaling pathway by LMP1 is of interest since constitutive STAT activity has been shown to be a feature of leukemic cell lines (43). Lymphoblastoid cell lines, B-cells that are immortalized by Epstein-Barr virus, contain STAT complexes that can bind DNA (44). 2 The importance of these DNA binding complexes is unknown but the ability of LMP1 to regulate not only the DNA binding (14) but also the transcriptional activity of STATs, as shown in the present study, suggests that the STATs that are binding DNA in lymphoblastoid cell lines are also likely to be transcriptionally active. STATs may offer novel therapeutic targets for LMP1-associated disease and the ability of LMP1 AAAG to inhibit STAT transcriptional ability increases its usefulness. LMP1 AAAG will be a useful tool to characterize the role of LMP1 in survival and disease pathogenesis. It has advantages over currently available technology, such as antisense technology or even LMP1-deleted recombinant viruses, as it is specific and allows the study of LMP1 in cells that have already been infected with EBV. LMP1 AAAG is being cloned into an inducible vector to make stable cell lines with this mutant to test its effects on the growth of lymphoblastoid cell lines, as a model of post-transplant lymphoproliferative disorder. Furthermore, inducible LMP1 AAAG may be useful in characterizing the role of LMP1 in the growth of nasopharyngeal carcinomas tumors transplanted into nude mice (45). With regard to Hodgkin's disease, published data suggests that LMP1 down-regulation of the cell surface protein, CD99, plays a critical role in the progression to Hodgkin's disease (46). LMP1 AAAG inhibits the LMP1 mediated down-regulation of the CD99 promoter 3 suggesting that it may be inhibitory for disease progression. Other disease models are currently under investigation.
In summary, this study enhances our understanding of LMP1 signaling specifically and as a model of the TNF receptor superfamily. We have characterized the effectiveness and mechanism of a receptor-based dominant inhibitor, LMP1 AAAG . This mutant suggests cooperation within the LMP1 oligomer in binding of the signaling molecule, TRAF2. The application of this type of receptor based inhibitory molecule may prove useful in establishing the function and mechanism of action of both LMP1 and other receptors in vitro and in a cell specific fashion in vivo. It has also been suggested that LMP1 AAAG may have useful therapeutic applications.  ). B, the dose responsive dominant-negative effect of LMP1 AAAG on wt LMP1 binding to TRAF2 was assayed. LMP1 AAAG was expressed as an EGFP-LMP1 AAAG -tagged protein so that the expression levels obtained with the indicated doses of pEGFP-LMP1 AAAG DNA (1-8 g) could be compared with the expression of the lower molecular weight wt LMP1 coexpressed from 3 g of SG5-LMP1 plasmid. A Western blot of the LMP1 immunoprecipitates was probed with rabbit anti-TRAF2 antibodies (upper blot), and the input lysates were probed with CS.1-4 anti-LMP1 antibodies (lower blot). C, the in vitro binding of TRAF2 to wild type LMP1 or a complex of wild type LMP1 and LMP1 AAAG was investigated by adding lysates that contained TRAF2 to immunoprecipitates of LMP1 from DG75 cells that had been transfected with wild type LMP1 or both wild type LMP1 and LMP1 AAAG . The immunoprecipitates were then resolved by SDS-PAGE and the presence of newly bound TRAF2 was investigated by Western blotting.