Oncogene (1998) 17, 2383 ± 2392 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Epstein ± Barr virus latent membrane -1 (LMP1) signalling is distinct from CD40 and involves physical cooperation of its two C-terminus functional regions

J Eike Floettmann1, Aristides G Eliopoulos2, Matthew Jones1, Lawrence S Young2 and Martin Rowe1

1Department of Medicine, University of Wales College of Medicine, Heath Park, Cardi€ CF4 4XX; 2Institute of Cancer Studies, The Medical School, The University of Birmingham, Birmingham B15 2TJ, UK

The Epstein ± Barr virus (EBV) encoded Latent Mem- expressed both in LCL and in many EBV-associated brane Protein-1 (LMP1) mimics a constitutively active malignancies. LMP1 is considered to be a classical receptor molecule, and has been shown to activate NF- oncogene because of its ability to fully transform kB and the MAPK and JNK pathways. Two regions immortal rodent ®broblast cell lines (Baichwal and within the cytosolic domain of LMP1 have been found to Sugden, 1988; Wang et al., 1985). In EBV-infected e€ect cell signalling. One of these, the carboxy-terminal human cells, LMP1 is essential for immortalisation of activation region-1 (CTAR1), binds members of the resting B (Kaye et al., 1993) and for TRAF family of , and the other (CTAR2) binds continued proliferation of the derived LCLs (Kilger et TRADD, suggesting that LMP1 transduces signals al., 1998). Furthermore, expression of LMP1 in cell similarly to the Tumour Necrosis Factor Receptor family lines causes a plethora of e€ects, including changes in of receptors. The ability to bind TRAFs, to activate NF- cell surface marker expression reminiscent of the kB and the JNK pathway, to upregulate cellular genes activation process initiated by EBV infection of such as CD54 (ICAM-1 adhesion molecule), and to normal resting B cells; e.g. upregulation of CD23, a€ect cell growth and apoptosis has led to the suggestion CD40 and CD54 (Peng and Lundgren, 1992; Wang et that LMP1 signalling is similar to, or even identical to al., 1988, 1990). LMP1 also e€ects cell cycle CD40. However, we now show that while ligand-induced progression by initiating DNA synthesis in resting B CD40 signalling is impaired in the Jurkat line, cells (Peng and Lundgren, 1992) and by partially

LMP1 was fully functional; therefore demonstrating that blocking lines in the G2/M phase (Floettmann et LMP1 and CD40 signalling di€er. Mutated LMP1 al., 1996). In addition, LMP1 can also protect cells genes, in which one or other of the CTAR1 and CTAR2 from apoptosis. In B cells this protection is at least in domains was non-functional, behaved more like CD40 in part due to upregulation of anti-apoptotic proteins, being unable to upregulate the CD54 cell surface marker Bcl-2 (Henderson et al., 1991; Rowe et al., 1994) and in Jurkat cells. However, the CTAR1 domain of LMP1, probably A20 (Laherty et al., 1992). In non-B cells, which shared a TRAF-binding sequence motif with LMP1 does not cause upregulation of Bcl-2 but does CD40, di€ered from CD40 in being unable to activate upregulate A20 (Fries et al., 1996; Laherty et al., 1992). NF-kB in Jurkat. Cotransfection experiments with In epithelial cells LMP1 also inhibits cellular differ- LMP1 mutants demonstrated that CTAR1 can coopera- entiation (Dawson et al., 1990), which could be an tive with CTAR2 on separate LMP1 molecules, provided important feature of LMP1's role in EBV-associated that they exist within the same oligomeric complex. epithelial malignancies. The signalling events involved in mediating LMP1 Keywords: Epstein ± Barr virus; LMP-1; NF-kB; CD54; e€ects have begun to emerge. Thus, the NF-kB CD40; TNF-R transcription factor was shown to be activated in response to LMP1 expression (HammarskjoÈ ld and Simurda, 1992) and, more recently, LMP1 was also shown to activate the JNK and MAPK kinase Introduction pathways (Eliopoulos and Young, 1998; Kieser et al., 1997; Roberts and Cooper, 1998). At least four TNF- Infection of human resting B lymphocytes in vitro with receptor associated factors (TRAF1, TRAF2, TRAF3 Epstein ± Barr virus (EBV) results in the outgrowth of and TRAF5) and the tumor necrosis factor receptor- immortal lymphoblastoid cell lines (LCLs) (reviewed associated death domain protein (TRADD) have been by Kie€, 1996; Rickinson and Kie€, 1996). Infection reported to directly associate with LMP1 (Devergne et with EBV in vivo is associated with a number of al., 1996; Izumi and Kie€, 1997; Mosialos et al., 1995; malignant diseases of lymphoid and epithelial origin Sandberg et al., 1997) and thus potentially act as signal (Rickinson and Kie€, 1996) and the Latent membrane transducers. protein-1 (LMP1) is one of the few viral genes to be The LMP1 molecule itself comprises a short N- terminal cytoplasmic domain, six membrane spanning helixes that anchor the protein in the membrane, and a long cytoplasmic carboxy-terminus. Two functional domains, the C-terminal activation region-1 (CTAR1) Correspondence: M Rowe Received 27 March 1998; revised 21 May 1998; accepted 21 May and -2 (CTAR2) have been identi®ed that are 1998 independently able to activate NF-kB in certain cells Cooperative signalling functions in LMP1 JE Floettmann et al 2384 (Huen et al., 1995; Mitchell and Sugden, 1995; Paine et tion of an expression vector for Green Fluorescent al., 1995). Mutational analysis of CTAR1 has led to Protein (GFP); this assay is a modi®cation of that the identi®cation of a PxQxT motif, at residues 204 ± developed by Pilon et al. (1991), which used CD2 as a 208 in the 386 amino acid LMP1 protein of the B95.8 marker gene for transfection eciency, and which is EBV isolate, that is essential for binding of TRAFs now a well established method for analysing LMP1 and CTAR1 mediated NF-kB activation (Devergne et function (Huen et al., 1995; Peng and Lundgren, 1992; al., 1996; Sandberg et al., 1997). Similarly, a six amino Rowe et al., 1994). acid sequence within the far carboxy terminus of the As shown in Figure 1a, LMP1 and ligand-activated molecule, at 379 ± 384, was found to be essential for CD40 both activated NF-kB to similar levels in the CTAR2 mediated NF-kB activation (Floettmann and Jurkat line. Jurkat is CD40-negative and, therefore, Rowe, 1997), and this region of CTAR2 binds CD40L only caused activation of NF-kB in cells TRADD (Izumi and Kie€, 1997). transfected with CD40. In contrast, Eli-BL cells In many respects, LMP1 appears to mimic a expressed endogenous CD40 which was re¯ected in constitutively active CD40-like receptor. While extra- the observation that CD40L activated NF-kB in the cellular binding of CD40-ligand mediates oligomerisa- vector control transfected Eli-BL cells. The CD40- tion of CD40 molecules and subsequent association of mediated activation of NF-kB in Eli-BL was consis- TRAFs with the cytosolic tails of the receptor complex, tently slightly higher than the activation mediated by LMP1 is constitutively active in a ligand-independent LMP1. However, it should be noted that in neither cell fashion through the spontaneous formation of line was there a synergistic e€ect of LMP1 and ligated oligomers via the multiple hydrophobic membrane- CD40 upon NF-kB activation, which is consistent with spanning regions. Similarities between CD40 and the hypothesis that the two receptors trigger similar LMP1 functions include their ability to bind TRAFs signalling pathways activating NF-kB. through a PxQxT motif, and to activate both NF-kB In addition to the activation of NF-kB in Eli-BL and the JNK pathway. Furthermore, LMP1 and CD40 cells, ligation of exogenous and/or endogenous CD40 show remarkable similarities in their downstream resulted in upregulation of cell surface CD54 that was e€ects upon cellular phenotype, including upregula- more ecient that that induced by LMP1 (Figure 1b). tion of a similar set of cell surface proteins (e.g. CD54), Importantly, while expression of LMP1 in Jurkat cells and in their e€ects upon cell di€eration, proliferation caused upregulation of CD54, ligation of CD40 and apoptosis (Eliopoulos et al., 1997; Kilger et al., completely failed to do so (Figure 1b) despite causing 1998; van Kooten and Banchereau, 1997; Zimber- ecient activation of NF-kB. The CD54 results in Strobl et al., 1996). Jurkat demonstrate a clear di€erence in the signalling In the present work, we have compared LMP1 and events following CD40 ligation and signalling from the CD40 signalling in di€erent cell types, and show that constitutive active LMP1 complex. there are di€erences. We also addressed the question of the signi®cance of LMP1 having two separate The CD40-homologous TRAF-binding CTAR1 domain functional domains. The results show that the of LMP1 has di€erent signalling properties to CD40 CTAR1 and CTAR2 regions can physically cooperate with each other in oligomeric complexes of LMP1 To identify possible reasons for the observed differ- molecules to generate a signalling event that is ences in LMP1 and CD40 mediated signalling, the qualitatively distinct from the e€ects mediated by e€ects of the individual CTAR1 and CTAR2 func- CTAR1 or CTAR2 alone. tional domains of LMP1 were examined by transfect- ing various LMP1 mutants into either Jurkat or Eli- Results BL. The premise for these experiments was that CTAR1 contains a TRAF binding domain that is Cell-dependent di€erences between LMP1 and CD40 very similar to the TRAF binding domain in CD40, signalling whereas CTAR2 binds TRADD via a sequence that has no structural or functional homology with CD40. It is well-established that LMP1 functions can be However, mutants that carried a deletion (LMP.349D) in¯uenced by the host cell type, and we have previously or a point mutation (LMP.Y384G) in CTAR2, and noted di€erences in the ability of LMP1 to upregulate were thus functional in CTAR1 only, unexpectedly NF-kB and CD54 in B cell lines and T cell lines (Huen failed to activate NF-kB in Jurkat (Figure 2a), which et al., 1995). Furthermore, even though CD54 contrasts with the ability of CD40 to activate NF-kBin expression is regulated by NF-kB, there was not these cells (Figure 1a). Even when greatly over- always a good correlation between NF-kB activation expressed relative to the amount of wild-type LMP1 and CD54 induction by LMP1, suggesting that CD54 required to give maximal NF-kB activation, these induction can be used as an indicator of LMP1 CTAR2-defective LMP1 mutants failed to activate NF- functions in addition to NF-kB activation. Therefore, kB (data not shown). Mutants carrying a deletion in the ®rst set of experiments we compared the abilities (LMP.186D352) or point mutations (LMP.AAA) in of LMP1 and CD40 to activate NF-kB and to induce CTAR1 and which were thus functional in CTAR2 CD54 in the Eli-BL B cell line and in the Jurkat T cell only, were shown to activate NF-kB to about 30% of line. The cells were transiently transfected with LMP1 wild-type LMP1 levels (Figure 2a) with maximal or CD40 and activation of NF-kB was measured by activation of about 40% achieved by greatly over- cotransfection of a NF-kB-regulated luciferase reporter expressing either of the mutant LMP1 proteins. Thus, plasmid. Induction of CD54 was measured by ¯ow- in Jurkat cells, neither CTAR1 nor CTAR2 alone was cytometric analysis of CD54 expression in the able to activate NF-kB as eciently as did wild-type transfected cell subpopulation identi®ed by cotransfec- LMP1, and CTAR1 function was most severely Cooperative signalling functions in LMP1 JE Floettmann et al 2385

Figure 1 Analysis of LMP1 and activated CD40 function in Jurkat and Eli-BL cells. Cells were transfected either with vector DNA or with DNA vectors expressing LMP1 or CD40, and were cultured without (open columns) or with (®lled columns) CD40-ligand. (a) Cells cotransfected with an NF-kB regulated luciferase reporter were assayed for luciferase at 24 h post-transfection. The basal luciferase activity obtained with vector control transfectants was subtracted from the activity of the other transfectants, and the data were nor- malized to the activity of wild-type LMP1 (100%). The results are the mean (+s.d.) of at least three separate experiments. The basal NF- kB activity in Jurkat was barely detectable, and expression of LMP1 caused 50 ± 100-fold induction, whereas the basal NF-kB in Eli-BL was higher and expression of LMP1 caused only a 3 ± 5-fold activation of NF-kB. (b) cells cotransfected with a Green Fluorescence Protein marker gene were harvested at 48 h post-transfection for ¯ow-cytometric immuno¯uorescence analysis of CD54 induction. CD54 was detected with a phycoerythrin-conjugated anti-CD54 monoclonal antibody and two-colour analysis was performed to quantitate the CD54 expression in the GFP-marked transfected cell population. The basal CD54 expression obtained with vector control transfectants was subtracted from the expression of the other transfectants, and the data were normalized to the activity of wild- type LMP1 (100%). The data for each construct are the mean (+s.d.) of at least three separate experiments Cooperative signalling functions in LMP1 JE Floettmann et al 2386 impaired. In Eli-BL cells, CTAR1 demonstrated CTAR1 and CTAR2 cooperate in LMP1-mediated signi®cant activation of NF-kB that reached up to upregulation of CD54 35% of wild-type levels, but again CTAR2 was more ecient and showed about 75% of wild-type function The above results con®rm and extend previous (Figure 2a). ®ndings (Huen et al., 1995) reporting cell type specific The di€erent eciencies with which CTAR1 and activation of NF-kB and upregulation of CD54 by CTAR2 domains of LMP1 activated NF-kB did not LMP1 and LMP1 derived deletion-mutants. They also correlate with the ability to upregulate CD54 (Figure suggest that, at least Jurkat cells, both CTAR1 and 2b). Thus, in Jurkat, CTAR1 or CTAR2 alone caused CTAR2 are required to be functional in the same cell little or no upregulation of CD54 expression, whilst in to activate CD54. We therefore performed a series of Eli-BL cells, CTAR1 and CTAR2 each activated CD54 cotransfection experiments to test whether wild-type to about 50% of the levels achieved by wild-type LMP1 function could be reconstituted by coexpression LMP1. Therefore, in terms of NF-kB activation and of mutant CTAR1 and CTAR2 LMP genes. This set CD54 upregulation, the e€ects of CD40 ligation are of experiments was performed only in the Jurkat line more closely mimicked by the non-homologous since the results were expected to be more easily CTAR2 domain of LMP1 than by the PxQxT interpreted. As shown in Figure 3, the LMP.Y384G containing CTAR1. mutant was again unable to activate NF-kB, while the

Figure 2 Function of CTAR1 and CTAR2 domains of LMP1 in Jurkat and Eli-BL cells. Two LMP1 mutants with a non- functional or deleted CTAR2 (LMP.Y384G and LMP.349D) were compared with two LMP1 mutants with a non-functional or deleted CTAR1 (LMP.AAA and LMP.186D352) for their ability to activate NF-kB(a) and to induce CD54 (b) as described in Figure 1 Cooperative signalling functions in LMP1 JE Floettmann et al 2387

Figure 3 Synergistic e€ect of CTAR1 and CTAR2 functional domains of LMP1. The LMP.Y384G and LMP.AAA mutant LMP1 genes were cotransfected into Jurkat cells, and compared with cells transfected with either one of the mutant genes or with wild-type LMP1. Activation of NF-kB(a) and induction of CD54 (b) were assayed as described in Figure 1

LMP.AAA mutant activated NF-kB to about 25% of wild-type LMP1 levels; but coexpression of the two mutants caused a synergistic twofold increase in NF- kB activation to about 50% of wild-type LMP1 levels. More remarkably, Figure 3 shows that although neither mutant alone was able to signi®cantly a€ect CD54 expression, cotransfection resulted in upregula- tion of CD54 to about 50% of levels achieved with wild-type LMP1. Similar results were obtained when the LMP.Y384G CTAR2-inactive mutant was sub- stituted with the LMP.349D CTAR2 deletion mutant (data not shown). This demonstrates that ecient upregulation of CD54 in Jurkat cells requires the presence of functional CTAR1 as well as functional Figure 4 Western blot demonstrating expression of the LMP1 CTAR2 in the transfected cell. mutants used in this study. Jurkat cells were transfected with the A Western blot of LMP1 and mutants used in this indicated LMP1 genes and were harvested after 24 h and processed for SDS ± PAGE and Western blotting. The blot was study is shown in Figure 4. The results show that the probed with the CS.1 ± 4 pool of monoclonal antibodies reactive di€erent LMP1 mutants were expressed at comparable with LMP1 levels and that cotransfection of LMP1 mutants did not result in increased expression of the individual mutant proteins which by itself could theoretically result in increased upregulation of CD54. As has been could be achieved if CTAR1 and CTAR2 were active documented previously (Huen et al., 1995), the on LMP1 mutants that formed separate homo- LMP1(186D352) was not detectable by Western oligomers; whilst the latter scenario would only arise blotting but expression was con®rmed by immuno- if CTAR1 and CTAR2 were present within the same ¯uorescence staining with the CS.1 antibody, and it hetero-oligomeric complex of LMP1 mutants. We appeared to be expressed comparably to the other therefore devised the following experiment to test LMP1 mutants. Furthermore, as shown in Figure 2, these two possibilities. As described schematically in the LMP1(186D352) mutant was clearly functional in Figure 5, two complementary CTAR1/CTAR2 mu- the Eli-BL cell line. tants were designed with the intention that one mutant with an inactivated CTAR2 (LMP.Y384G) would spontaneously form homo-oligomers, whereas the Cooperation of CTAR1 and CTAR2 occurs within the second mutant (CD2-229), comprising a fusion protein same oligomeric LMP1 complex of the transmembrane and extracellular domains of The observed cooperation between CTAR1 and CD2 fused to residues 229 ± 386 of LMP1, would form CTAR2 (Figure 3) could arise in one of two possible separate oligomers upon cross-linking with extracellu- ways: either two separate signalling events combine lar anti-CD2 antibodies (Figure 5b and c). additively to e€ect full wild-type LMP1 function or, a First, in order to demonstrate that CD2-LMP1 physical association (direct or indirect) between fusion proteins do indeed form oligomeric complexes CTAR1 and CTAR2 initiates a qualitatively distinct separately from molecules containing the multiple signalling event. In the former scenario, cooperation membrane-spanning sequences of LMP1, immunopre- Cooperative signalling functions in LMP1 JE Floettmann et al 2388 cipitation experiments were performed (Figure 6). any LMP.Y384G in the anti-CD2 immunoprecipitates, Because the extracellular domain of rat CD2 is heavily although the CD2-192 was quantitatively recovered. As glycosylated, the CD2-LMP1 fusion proteins migrate a positive control for these experiments, LMP.Y384G as broad bands on SDS-polyacrylamide gel electro- was coexpressed with the CD2-LMP1 chimera which phoresis (SDS ± PAGE) and they characteristically comprised a full-length LMP1 (1 ± 386) fused at its show weaker bands of a lower molecular weight that amino terminus to the transmembrane and extracel- may be incompletely glycosylated polypeptides or lular domains of CD2; this chimera is known to be breakdown products. With the CD2-229 chimera, one spontaneously functional (Floettmann and Rowe, of the lower molecular-weight bands migrated similarly 1997) and is therefore assumed to spontaneously form to wild-type LMP1 which complicated interpretation of oligomers. In this control immunoprecipitation the coimmunoprecipitation experiments. In these experi- LMP.Y384G was shown to associate with the CD2- ments, therefore, we used the CD2-192 chimera LMP1 protein. Thus, the lack of multiple tansmem- containing LMP1 residues 192 ± 386 (i.e. including brane regions in the CD2-192 and CD2-229 chimerae CTAR1) which was preferable because it produced a abrogates their ability to form hetero-oligomers with larger protein that was clearly distinguished from wild- LMP1, but they do form homo-oligomers with type LMP1 on SDS ± PAGE. Jurkat cells were themselves following antibody-mediated cross-linking cotransfected with di€erent combinations of LMP1 via the CD2 domain. mutants and fusion proteins and, after 18 h culture, the As shown in Figure 7a, the CD2-229 chimera was membrane proteins were solubilized with NP40 unable to activate NF-kB in its monomeric state but detergent and immunoprecipitated with anti-CD2 when cross-linked with anti-CD2 monoclonal anti- antibody. Western blots of the immunoprecipitates bodies plus secondary rabbit anti-mouse IgG anti- were then probed with LMP1-speci®c monoclonal bodies, activation of NF-kB was observed at levels antibodies. As seen in Figure 6, cells expressing slightly higher than were achieved with the LMP.AAA LMP.Y384G alone did not yield LMP1 sequences CTAR1-inactivated mutant. Cross-linking the CD2-229 with anti-CD2 immunoprecipitates. Cells coexpressing protein caused a small increase in CD54 expression to both LMP.Y384G and CD2-192 also failed to yield the low levels obtained spontaneously with the

Figure 5 A schematic representation of the experiment in which the LMP.Y384G mutant (functional CTAR1 domain only) and the CD2.229 fusion protein (functional CTAR2 domain only) were expressed as separate homo-oligomers. Schematic A shows the components of the CD2.229 chimeric protein which contains the extracellular and transmembrane domains of rat CD2 (non- functional) fused to the C-terminus of LMP1 which is deleted for CTAR1. Schematic B illustrates the plasma membrane localisation of coexpressed LMP.Y384G and CD2.229 proteins. The LMP.Y384G protein spontaneously forms functional homo-oligomers, while CD2.229 remains as scattered inactive monomers. Schematic C illustrates the e€ect of adding cross-linking antibodies directed to determinants on the extracellular CD2 domain of CD2.229 proteins. Now, the CD2.229 proteins are induced to oligomerise, with concommitent activation of CTAR2 function, but these oligomers remain separate from the LMP.Y384 oligomers Cooperative signalling functions in LMP1 JE Floettmann et al 2389 reference LMP.AAA mutant (Figure 7b). Crucially, the cross-linked CD2-229 protein showed only a slight when compared to the synergistic e€ect of coexpressing additive e€ect upon CD54 when coexpressed with LMP.AAA and LMP.Y384G upon CD54 induction, LMP.Y384G despite showing similar levels of NF-kB activation to contransfected LMP.Y284G/LMP.AAA (Figure 7a). These results demonstrate that CTAR1 and CTAR2 have to locate to the same oligomeric complex in order to synergistically a€ect upregulation of CD54.

Discussion

We have shown that CD40L treatment of Jurkat T cells transfected with CD40 causes activation of NF- kB, but not upregulation of CD54. This is in contrast to LMP1 mediated signalling which results in activation of NF-kB and also upregulation of cell surface CD54. This observation highlights the fact that NF-kB activation alone is not sucient to activate CD54 in Jurkat cells, and it indicates the requirement for additional signal factor(s). Since the ability of activated CD40 to upregulate CD54 was demonstrated Figure 6 Immunoprecipitation experiments demonstrating that LMP1 and CD2/LMP1 fusion proteins can form separate homo- here in the Eli-BL B cell line, the results with the oligomers. Shown is a Western blot, probed with CS.1 ± 4 anti- Jurkat line demonstrate that CD40 mediated signalling LMP1 antibodies, of proteins immunoprecipitated with the OX34 may somehow be impaired in certain host cells while anti-CD2 antibody from Jurkat cells cotransfected with the LMP1 signalling is una€ected. A subsequent investiga- indicated combinations of LMP1 and CD2/LMP1 expression tion of a wider panel of BL lines (Henriquez et al., vectors. The tCD2 DNA encodes only the extracellular and transmembrane domains of rat CD2; the CD2-192 DNA encodes 1998) showed that while most BL lines tested were the extracellular and transmembrane domains of rat CD2 fused to responsive to CD40 signalling, as was the Eli-BL line the cytosolic 192 ± 386 aa of LMP1; and the CD2-LMP1 DNA studied in the present work, some BL lines were clearly encodes the extracellular and transmembrane domains of rat CD2 defective for CD40 signalling while retaining the ability fused to the full-length 1 ± 386 aa of LMP1 to produce a seven to eciently transduce LMP1 signals. Therefore, the transmembrane domain protein that retains the full function of wild-type LMP1 but which contains an extracellular CD2 Jurkat T cell line is not unique in dislaying di€erential domain. Immunoprecipitates from cells transfected with ability to support CD40 and LMP1 signals. These LMP.Y384G (lane 1) or with the tCD2 gene (lane 2), did not observations indicate the use of di€erent signalling contain any LMP1 sequences. Immunoprecipitates from cells pathways by CD40 and LMP1 which usually converge transfected with CD2-192 (lanes 2 and 4) or with CD2-LMP1 gene (lanes 5 and 6), contained LMP1 sequences from the fusion to e€ect similar downstream functions. Therefore, if proteins, as expected. Coexpression of LMP.Y384G with the one subscribes to the view that LMP1 mimics CD40, CD2-LMP1 resulted in coprecipitation of the LMP.Y384G this could give LMP1 the advantage to remain fully protein (lane 6) as demonstrated by the appearance of a 63 kD functional in cellular backgrounds in which CD40 band of the same size as the LMP.Y384G detected in whole cell signalling is impaired. extracts (lane 8). In contrast, coexpression of LMP.Y384G with the CD2-192 did not result in coprecipitation of the LMP.Y384G Although CD40 and LMP1 each have two domains protein (lane 4) that are able to activate NF-kB, only one of each pair

Figure 7 The synergistic e€ect of CTAR1 and CTAR2 functional domains of LMP1 is not observed if CTAR1 and CTAR2 are expressed in separate LMP1 oligomers. Jurkat cells were transfected with the indicated LMP1 expression vectors, and were incubated without (open columns) or with (®lled columns) anti-CD2 and anti-mouse IgG cross-linking antibodies. Activation of NF-kB(a) and induction of CD54 (b) were assayed as described in Figure 1 Cooperative signalling functions in LMP1 JE Floettmann et al 2390 shows signi®cant . Thus, a TRAF contrast, activation of NF-kB by CTAR2, was only binding PxQxT motif is present in CD40 at aa 250 ± 254 partially impaired in Jurkat; nevertheless, LMP1 and also in the CTAR1 region of LMP1 at aa 204 ± 208. mutants active for CTAR2 alone were unable to Interestingly, while LMP1 CTAR1 was reported to bind upregulate CD54. Cotransfection experiments with TRAF1 as well as TRAF2, TRAF3 and TRAF5 mutated LMP1 genes in which either CTAR1 or through its PxQxT motif (Brodeur et al., 1997; CTAR2 was inactivated, showed that the CTAR1 and Devergne et al., 1996; Mosialos et al., 1995; Sandberg CTAR2 domains of LMP1 can cooperate to et al., 1997), CD40 seems to associate only with TRAF2 upregulate CD54. Using a chimeric CD2/CTAR2 and TRAF3 through this motif (Ishida et al., 1996; construct it was possible to exclude the possibility of Rothe et al., 1995). It has been noted that di€erences in the observed cooperation between CTAR1 and CTAR2 the sequences adjacent to the PxQxT motif may being the additive e€ect of two independent signals in¯uence TRAF binding and signalling from CD40 from separate CTAR1 and CTAR2 homo-oligomers. and LMP1 (Eliopoulos et al., 1997; Franken et al., Thus, coexpression of spontaneous oligomers of 1996). It is possible, therefore, that the inability of LMP.Y384G (functional CTAR1 only) and separate TRAF1 to bind CD40 could explain di€erences between antibody-mediated oligomers of the CD2-229 (CD2/ CD40 and LMP1 mediated signals. Similarly it has been LMP1 chimera with CTAR2 only) did not upregulate observed that while overexpression of TRAF1 potenti- CD54. This demonstrates that cooperation between ates LMP1-mediated NF-kB activation (Devergne et al., CTAR1 and CTAR2 has to occur within a hetero- 1996), it has no e€ect on CD40-mediated NF-kB oligomeric LMP1 complex. Furthermore, it suggests activation (Rothe et al., 1995). There is no sequence that this cooperation between CTAR1 and CTAR2 homology between the second NF-kB activation results in a signalling event that is qualitatively distinct domain within CD40, which maps to the TRAF6 from the signals generated by CTAR1 and CTAR2 binding at aa 230 ± 245 (Ishida et al., 1996), and either separately. CTAR1 or CTAR2 of LMP1. Furthermore, TRAF6 The existence of a cooperative function between appears not to be directly involved in LMP1-mediated CTAR1 and CTAR2 of LMP1 gives further insight signalling (Brodeur et al., 1997). Therefore, despite the into how LMP1 signals. It also cautions against over- presence of a functional PxQxT TRAF binding motif in interpretation of data obtained with mutants non- CD40 and LMP1, the only evidence that CD40 and functional for CTAR1 or CTAR2, and particularly LMP1 utilize the same mechanisms to activate NF-kB with mutants containing large deletions that may comes from experiments where dominant-negative adversely a€ect physical cooperation between CTAR1 TRAF2 over-expression was shown to block NF-kB and CTAR2. In agreement with this, there is evidence activation from both molecules (Devergne et al., 1996; that di€erences in signalling from individual CTAR1/2 Rothe et al., 1995). However, rather than directly domains and whole length LMP1 might result from interacting with other TRAFs at the receptor, the changes in the components of the TRAF complexes e€ects of dominant-negative TRAF2 could result from bound to LMP1. Thus, aa 187 ± 386 of LMP1 bind less an interference with other signal complexes further TRAF1 and 2 than deletion mutants encoding amino down the pathway. This was demonstrated by the acids 187 ± 231 only (Devergne et al., 1996; Sandberg et ability of TRAF2 to interfere with NF-kB activation al., 1997). from LMP1 CTAR2 which itself does not bind TRAF2 These data with LMP1 raise the question of whether directly (Devergne et al., 1996). similar cooperation may occur between the two It is conceivable that the di€erent mechanisms by signalling domains of CD40. Intriguingly, it has been which CD40 and LMP1 activate NF-kB may result in suggested that CD30 may undergo a conformational the translocation of di€erent NF-kB species to the change when TRAF2 simultaneously engages both of nucleus. Whilst both CD40 and LMP1 are reported to its signalling domains (Boucher et al., 1997). It is activate p50/RelA heterodimers (Herrero et al., 1995; possible, therefore, that physical cooperation between Rothe et al., 1995) it is not clear that this is sucient two cytoplasmic signalling domains to produce a to activate CD54 ; it has been reported qualitatively new signalling event is a general feature that RelA/cRel heterodimers may play a crucial role in of the CD40/TNFR superfamily of receptors. CD54 gene expression (Ledebur and Parks, 1995). Alternatively, or in addition, it is possible that other signalling pathways such as the SEK/JNK/AP1 path- Materials and methods way (Reinhard et al., 1997) normally activated by the association of TRAF2 with members of the TNF- Cell lines receptor superfamily, or even the MAPK pathway Jurkat is a cell line derived from a EBV negative T cell (Roberts and Cooper, 1998), are involved in regulation (Brattsand et al., 1990). Eli-BL is an EBV- of CD54 gene expression and may be selectively positive B cell line established from a Burkitt's lymphoma impaired in Jurkat T cells but not in Eli-BL cells. and it displays a latency I form of infection in which Whatever the reasons for the di€erences between EBNA1 is the only viral protein detected (Rowe et al., CD40 and LMP1 function in Jurkat cells, this line 1992). All cells were grown in RPMI/10% FCS supplemented with 200 U of penicillin and 200 mgof proved useful for investigating the structural require- streptomycin per ml at 378Cina5%CO atmosphere. ments for LMP1 function. Consistent with previous 2 reports that the extent of CTAR1 and CTAR2- mediated signalling in LMP1 mutants can vary with Plasmids the cellular background (Huen et al., 1995), we found Most of the plasmids used in this work have been described that the Jurkat is defective for CTAR1 mediated elsewhere (Eliopoulos and Young, 1988; Floettmann and activation of NF-kB and upregulation of CD54. In Rowe, 1997; Huen et al., 1995). Plasmid pSG5-LMP1 Cooperative signalling functions in LMP1 JE Floettmann et al 2391 expresses wildtype LMP1 cloned from the B95.8 virus, and Flow-cytometric CD54 expression assay was used to derive various mutants: plasmid pSG5- LMP.349D expresses LMP1 amino acids 1 ± 349; plasmid Induction of CD54 expression by LMP1 was analysed by pSG5-LMP.186D351 expresses LMP1 deleted for amino ¯ow-cytometry using Green Fluorescence Protein as a marker acids 187 to 350; plasmid pSG5-LMP.Y384G expresses for transfected cells. The monoclonal antibody to human LMP1withaminoacid384changedfromtyrosineto CD54 was phycoerytherin-conjugated (MCA675PE; Serotec) glycine; and plasmid pSG5-LMP.AAA expresses LMP1 and staining was carried out with a single incubation at 08C with the amino acids proline 204, glutamine 206 and for 60 min. Cells transfected with CD2/LMP1 chimeras that threonine 208 mutated to alanine. The pSG5-CD2 expresses were incubated with cross-linking antibodies were pretreated a truncated rat CD2 protein (amino acids 1 ± 212) that lacks with 20% normal mouse serum for 30 min at 08C prior to the cytosolic tail, and was used to generate chimeric CD2/ staining for CD54. LMP1 proteins: pSG5-CD2-LMP1 expresses truncated CD2 fused to the amino terminus of the whole length LMP1 Detection of proteins by Western blotting (amino acids 1 ± 386); pSG5-CD2-192 expresses truncated CD2 fused to amino acids 192 ± 386 of LMP1; and pSG5- Cells were harvested and analysed for the expression of CD2-229 expresses truncated CD2 fused to amino acids proteins by SDS ± polyacrylamide gel electrophoresis and 229 ± 386 of LMP1. Plasmid pcDNA3-CD40 is a human Western blotting as previously described (Floettmann and CD40 expression vector (Eliopoulos and Young, 1998). The Rowe, 1997). Nitrocellulose transfers were blocked in Blotto Green Fluorescent Protein expression plasmid, pEGF-N1, (5% w/v dried skimmed milk in TBS) for 2 h before binding was obtained from Clontech. primary antibody. LMP1 and LMP1 encoding chimeras were detected using 1 mg/ml CS. 1 ± 4 (Rowe et al., 1987) in Blotto. All blots were incubated with 1 : 2000 dilution of rabbit anti- Gene transfection mouse IgG antibodies (Z0259: Dako) rabbit anti mouse All cell lines were transfected by electroporation using a diluted 1 : 2000 in Blotto followed by an incubation with goat Biorad Genepulser II electroporator. Cells were harvested anti-rabbit-alkaline phosphatase (Sigma; A-3812) diluted by centrifugation and resuspended at 1.56107 cells/ml in 1 : 10 000 in TBS-tween. Speci®c antibody-protein complexes growth medium. An aliquot of 0.5 ml was then added to were detected using a BCIP/NBT chromogenic substrate the electroporation cuvette (Biorad, 0.4 mm width) and (Biorad). 2 mg) of each plasmid was added. After electroporation at 270 V and 950 mF, the cells were reseeded in 5 ml growth Immunoprecipitations medium and incubated under normal conditions. Jurkat cells were electroporated with di€erent combina- tions of expression plasmids, and 24 h later aliquots of Oligomerisation of CD2 and ligation of CD40 26107 cells were harvested, rinsed with cold phosphate Unless otherwise stated, oligomerisation of CD2/LMP1 balanced salt solution (PBS), and kept in a hypotonic chimeric proteins was induced by the addition of 5% bu€er (10 mM HEPES pH 7.9, 0.5 mM KCl, 0.5 mM culture supernatant of the OX34 hybridoma (Je€eries et MgCl2,0.1mM EGTA, 0.5 mM DTT) on ice for 30 min al., 1985) which produces murine anti-CD2 monoclonal as described elsewhere (Gires et al., 1997). The cells were antibodies and addition of 1 : 100 polyclonal anti-mouse then dounced using a 27 gauge needle and lysates were immunoglobulins Z0259 (Dako) which had been dialysed centrifuged. The particulate fraction was resuspended in to remove the azide from the antibody preparation. 400 mlcoldTBS/0.5%NP40andincubatedonice.After Transfected and endogenous CD40 was ligated with 30 min the lysate was centrifuged again and the super- 10 mg/ml CD40 ligand (Immunex). natant was added to 100 ml of a Sepharose-Protein A beads (P-3391; Sigma) preincubated with rabbit anti-mouse IgG antibodies (Z0259; Dako) and OX34 murine antibodies to Assay for NF-kB activity rat CD2. Following incubation on a rotating mixer at 48C The activity of NF-kB was assessed using a luciferase overnight, the beads were washed three times in RIPA reporter plasmid with three kBelementsupstreamofa bu€er (50 mM Tris pH 7.5, 150 mM NaCl, 1% NP40, 0.5% minimal conalbumin promoter driving the expression of the Na deoxycholate, 0.1% SDS) and once in PBS. Beads were ®re¯y luciferase gene (Arenzana-Seisdedos et al., 1993). At resuspended in 100 mlGelSampleBu€er(50mM Tris pH 24 h post-transfection, cells were washed twice in phosphate 6.8, 4% SDS, 10% glycerol, 5% b-mercaptoethanol, 0.01% bu€ered saline and lysed in 150 ml lysis bu€er containing bromophenol blue), and boiled for 2 min prior to analysis 100 mM HEPES pH 8.0, 2 mM magnesium chloride, 5 mM by Western blotting. dithiothreitol, and 2% Triton X-100. Luciferase activity was measured by adding 50 ml of clari®ed lysate to a scintilation tube and then analysing release in a Berthold LB9501 luminometer following injection of 100 mlof0.5mM luciferin (USB) dissolved in Luciferin Assay Reagent Acknowledgements This work was funded by grants from the Welsh Scheme (30 mM glycylglycin pH 7.9, 1 mM MgCl2,0.1mM EDTA, 30 mM DTT, 0.3 mM coenzyme A, 0.5 mM ATP). Light for Health and Social Science, from the release was integrated for 10 s. The luciferase activity in each Research Fund and grant ERB CHR XCT 940651 from sample was normalised for variations in transfection the European Community. AE and LSY were supported by eciency as determined by the percentage of GFP positive the Cancer Research Campaign and the Medical Research cells. Council, UK.

References

Arenzana-Seisdedos F, Fernandez B, Dominguez I, Jacque Boucher LM, Marengere LEM, Lu Y, Thukral S and Mak JM, Thomas D, Diaz-Meco MT, Moscat J and Virelizier TW. (1997). Biochem. Biophys. Res. Commun., 233, 592 ± JL. (1993). J. Virol., 67, 6596 ± 6604. 600. Baichwal VR and Sugden B. (1988). Oncogene, 2, 461 ± 467. Cooperative signalling functions in LMP1 JE Floettmann et al 2392 Brattsand G, Cantrell DA, Ward S, Ivars F and Gullberg M. Kieser A, Kilger E, Gires O, Ueng M, Kolch W and (1990). J. Immunol., 144, 3651 ± 3658. Hammerschmidt W. (1997). EMBO J., 16, 6478 ± 6485. Brodeur SR, Cheng G, Baltimore D and Thorley-Lawson KilgerE,KieserA,BaumannMandHammerschmidtW. DA, T-L. (1997). J. Biol. Chem., 272, 19777 ± 19784. (1998). EMBO J., 17, 1700 ± 1709. Dawson C, Rickinson AB and Young LS. (1990). Nature, Laherty CD, Hu HM, Opipari AW, Wang F and Dixit VM. 344, 777 ± 780. (1992). J. Biol. Chem., 267, 24157 ± 24160. Devergne O, Hatzivassiliou E, Izumu KM, Kaye KM, Ledebur HC and Parks TP. (1995). J. Biol. Chem., 270, 933 ± Kleijnen MF, Kie€ E and Mosialos G. (1996). Mol. Cell. 943. Biol., 16, 7098 ± 7108. Mitchell T and Sugden B. (1995). J. Virol., 69, 2968 ± 2976. EliopoulosA,StackM,DawsonC,KayeK,RoweMand Mosialos G, Birkenbach M, Yalamanchili R, VanArsdale T, Young L. (1997). Oncogene, 14, 2899 ± 2916. Ware C and Kie€ E. (1995). Cell, 80, 389 ± 399. Eliopoulos A and Young LS. (1998). Oncogene, 16, 1725 ± Paine E, Scheinman AS, Baldwin Jr, AS and Raab-Traub N. 1736. (1995). J. Virol., 69, 4572 ± 4576. Floettmann JE and Rowe M. (1997). Oncogene, 15, 1851 ± Peng M and Lundgren E. (1992). Oncogene, 7, 1775 ± 1782. 1858. Pilon M, Gullberg M and Lundgren E. (1991). J. Immunol., Floettmann JE, Ward K, Rickinson AB and Rowe M. 146, 1047 ± 1051. (1996). Virology, 223, 29 ± 40. Reinhard C, Shamoon B, Shyamala V and Williams LT. Franken M, Devergne O, Rosenzweig M, Annis B, Kie€ E (1997). EMBO J., 16, 1080 ± 1092. and Wang F. (1996). J. Virol., 70, 7819 ± 7826. Rickinson AB and Kie€ E. (1996). Fields Virology, Vol. 2. Fries KL, Miller WE and Raab-Traub N. (1996). J. Virol., Fields BN, Knipe DM and Howley PM. (eds). Lippincott- 70, 8653 ± 8659. Raven Publishers: Philadelphia, pp. 2397 ± 2446. Gires O, Zimber-Strobl U, Gonella R, Ueng M, Marschall Roberts ML and Cooper NR. (1998). Virology, 240, 93 ± 99. G, Zeidler R, Pich D and Hammerschmidt W. (1997). Rothe M, Sarma V, Dixit VM and Goeddel DV. (1995). EMBO J., 16, 6131 ± 6140. Science, 269, 1424 ± 1427. HammarskjoÈ ld M-L and Simurda MC. (1992). J. Virol., 66, Rowe M, Evans HS, Young LS, Hennessy K, Kie€ E and 6496 ± 6501. Rickinson A. (1987). J. Gen. Virol., 68, 1575 ± 1586. Henderson S, Rowe M, Gregory C, Croom-Carter D, Wang Rowe M, Lear A, Croom-Carter D, Davies AH and F, Longnecker R, Kie€ E and Rickinson A. (1991). Cell, Rickinson AB. (1992). J. Virol., 66, 122 ± 131. 65, 1107 ± 1115. Rowe M, Peng-Pilon M, Huen DS, Hardy R, Croom-Carter Henriquez N, Floettmann E, Salmon M, Rowe M and D, Lundgren E and Rickinson AB. (1994). J. Virol., 68, Rickinson AB. (1998). (Submitted for publication). 5602 ± 5612. Herrero JA, Mathew P and Paya CV. (1995). J. Virol., 69, Sandberg M, Hammerschmidt W and Sugden B. (1997). J. 2168 ± 2174. Virol., 71, 4649 ± 4656. Huen DS, Henderson SA, Croom-Carter D and Rowe M. van Kooten C and Banchereau J. (1997). Curr. Opin. (1995). Oncogene, 10, 549 ± 560. Immunology, 9, 330 ± 337. Ishida T, Tojo T, Aoki T, Kobayashi N, Ohishi T, Watanabe Wang D, Liebowitz D and Kie€ D. (1985). J. Virol., 43, T, Yamamoto T and Inoue J. (1996). Proc. Natl. Acad. Sci. 831 ± 840. USA, 93, 9437 ± 9442. Wang D, Liebowitz D, Wang F, Gregory C, Rickinson A, Izumi KM and Kie€ ED. (1997). Proc. Natl. Acad. Sci. USA, Larson R, Springer T and Kie€ E. (1988). J. Virol., 62, 94, 12592 ± 12597. 4173 ± 4184. Je€eries WA, Green JR and Williams AF. (1985). J. Exp. Wang F, Gregory C, Sample C, Rowe M, Liebowitz D, Med., 162, 117 ± 127. Murray R, Rickinson A and Kie€ E. (1990). J. Virol., 64, Kaye KM, Izumi KM and Kie€ E. (1993). Proc. Natl. Acad. 2309 ± 2318. Sci. USA, 90, 9150 ± 9154. Zimber-Strobl U, Kempkes B, Marschall G, Zeidler R, van Kie€ E. (1996). Fields Virology, Vol. 2. Fields BN, Knipe Kooten C, Banchereau J, Bornkamm GW and Ham- DM and Howley PM (eds). Lippincott-Raven Publishers: merschmidt, W. (1996). EMBO J., 15, 7070 ± 7078. Philadelphia, pp. 2343 ± 2396.