Latent infection membrane protein transmembrane FWLY is critical for intermolecular interaction, raft localization, and signaling

Abstract

Relatively little is known about the biochemical mechanisms through which the Epstein–Barr virus latent infection integral membrane protein 1 (LMP1) transmembrane domains cause constitutive LMP1 aggregation and continuous cytoplasmic C terminus-mediated signal transduction. We now evaluate the role of the three consecutive LMP1 hydrophobic transmembrane pairs, transmembrane domains (TM)1-2, TM3-4, and TM5-6, in intermolecular aggregation and NF-κB activation. LMP1TM1-2 enabled ≈40% of wild-type LMP1 cytoplasmic domain-mediated NF-κB activation, whereas TM3-4 or TM5-6 assayed in parallel had almost no effect independent of LMP1TM1-2. Alanine mutagenesis of conserved residues in LMP1TM1-2 identified FWLY_38–41 to be critical for LMP1TM1-2 intermolecular association with LMP1TM3-6. Further, in contrast to wild-type LMP1, LMP1 with FWLY_38–41 mutated to AALA_38–41 did not (i) significantly partition to lipid Rafts or Barges and effectively intermolecularly associate, (ii) enable cytoplasmic C terminus engagement of tumor necrosis factor receptor-associated factor 3, (iii) activate NF-κB, and thereby (iv) induce tumor necrosis factor receptor-associated factor 1 expression. Other LMP1 intermolecular associations were observed that involved LMP1TM1-2/LMP1TM1-2 or LMP1TM3-4/LMP1TM3-6 interactions; these probably also contribute to LMP1 aggregation. Because FWLY_38–41 was essential for LMP1-mediated signal transduction, and LMP1 activation of NF-κB is essential for proliferating B lymphocyte survival, inhibition of LMP1FWLY₄₁-mediated LMP1/LMP1 intermolecular interactions is an attractive therapeutic target.

Epstein–Barr virus (EBV) infection efficiently growth transforms resting primary B lymphocytes into proliferating lymphoblastoid cell lines (LCLs), causes infected B cell lymphoproliferative diseases in immune-suppressed people, and is a cause of Hodgkin's disease (for review, see refs. 1 and 2). EBV is also etiologically implicated in nasopharyngeal carcinoma. EBV latent infection membrane protein 1 (LMP1) is essential for EBV-induced lymphocyte proliferations and is also expressed in Hodgkin's disease and nasopharyngeal carcinoma.

LMP1 is an integral plasma membrane protein that consists of a 24-aa cytoplasmic N terminus, six hydrophobic transmembrane domains 1–6 (TM1–6), separated by five short reverse turns, and a 200-aa cytoplasmic C terminus [Fig. 1A (3, 4)]. The transmembrane domains mediate aggregation and association with plasma membrane microdomain lipid rafts (4–7). Aggregation is critical for LMP1 C-terminal cytoplasmic domain signaling (4, 8–13). N-terminally truncated LMP1 composed of TM5-6, and the cytoplasmic C terminus is expressed in lytic EBV infection, localizes diffusely in cytoplasmic membranes, and barely signals, whereas LMP1 deleted for TM3-4 is also diffuse in cytoplasmic membranes and only partially signals (14). LMP1 is palmitoylated, and mutation of C₇₈ to A blocks palmitoylation without affecting Raft association or signaling (7). Recently, an extensive mutation of a putative heptad repeat in LMP1 TM1 lessened protein stability, aggregation, and signaling, compatible with the notion that this sequence has a role in LMP1 aggregation (15).

Fig. 1.

The role of LMP1TM1-6 in NF-κB activation. (A) Schematic depiction of WT and deletion mutant Flag-tagged LMP1. Transmembrane domains are indicated by black, cytoplasmic domains by white, and extracellular domains by gray. LMP1 C-terminal cytoplasmic domains that engage TRAF and death domain proteins are indicated by 1 and 2, respectively. Numbers 71, 133, 129, 75, and 187 refer to codons in WT LMP1 B95 sequence. ΔN begins with codon 12. B and C show NF-κB activation by WT and/or mutant LMP1 in HEK293 cells that were cotransfected 24 h previously with an LMP1 expression plasmid, a 3X-NF-κB site-promoter and luciferase reporter plasmid, and a control GK-β-galactosidase expression plasmid. The averages and the error bars of fold induction in luciferase assay are based on three separate transfections. Similar results were observed in three independent experiments. WT and mutant LMP1 expression at similar levels was confirmed by Western blot.

Overall, LMP1 induces B lymphocyte activation and adhesion molecule expression and causes cellular adhesion. In immortalized rodent fibroblasts, LMP1 lowers serum dependence, lessens contact inhibition, enables anchorage-independent growth, and increases tumorigenicity in nude mice (8–11, 16–19). Reverse genetic and biochemical analyses of LMP1 effects or of LMP1's essential role in EBV conversion of B lymphocytes to LCLs indicate that the LMP1 cytoplasmic C terminus has two domains that are essential for NF-κB activation and B lymphocyte growth transformation (13, 20–22). The LMP1 C terminus between these two domains is not important for Janus kinase 3 activation or EBV effects on lymphocyte growth or survival (7, 8, 20–26). The two critical domains recruit tumor necrosis factor receptor-associated factors (TRAF)1, -2, -3, and -5 (27–37) or tumor necrosis factor receptor-associated death domain proteins, such as TRADD and RIP (13, 24, 38). Through these domains, LMP1 activates NF-κB (18, 19), c-Jun N-terminal kinase (16, 39, 40), and p38 mitogen-activated protein kinase (41). LMP1-induced NF-κB activation is critical for LCL survival (42). The experiments described here further investigate the role of LMP1 transmembrane domains in constitutive aggregation and NF-κB activation.

Materials and Methods

Cell Lines and Antibodies. Human embryonic kidney (HEK)293 and EBV-negative human Burkitt lymphoma (BJAB) cells were cultured in DMEM and RPMI medium 1640, respectively, supplemented with 10% FCS, 2 mM l-glutamine, penicillin, and streptomycin at 37°C with 5% CO₂. S12 mouse monoclonal antibody against LMP1 was purified from hybridoma (43) supernatant. Antibodies to TRAF3 (H-122), TRAF1 (H-3), CD40 (C-20), Lyn (H-6), hemagglutinin (HA) (F-7), and GST (B-14) were from Santa Cruz Biotechnology, and FLAG (M2) antibody was from Sigma.

Plasmids. Flag-tagged EBV B95-8 strain WT LMP1 (FLMP1) cDNA deletion mutants were made by PCR. Substitution mutants were made by QuikChange site-directed mutagenesis (Stratagene). Mutants were cloned into Flag-modified pSG5 (Stratagene) and sequence-confirmed. A plasmid with a three-MHC class I NF-κB site-enhanced minimal promoter-luciferase reporter and a pGK-β-gal reporter control plasmid were used to evaluate NF-κB activation in transient transfection assays (7). BJAB cell transfectants that stably express Flag-tagged WT or mutant LMP1, FLMP1 were made by subcloning FLMP1 cDNA into pTER10, which has the murine elongation factor 1α promoter for FLMP1 expression and a neomycin-resistant gene under the control of the MC1 promoter (Stratagene).

Transfections and Reporter Gene Assays. HEK293 cells were transfected in six-well plates by using Effectene (Qiagen, Chatsworth, CA). Twenty-four hours after transfection, cells were lysed in reporter lysis buffer (Promega). Luciferase (Promega) and β-galactosidase (Galacton-Plus, Tropix, Bedford, MA) activities were assayed, and luciferase values were normalized for β-galactosidase activity. FLMP1 expressing BJAB cell lines were established by electroporation of BJAB cells with pTER10 FLMP1 expression plasmids followed by selection using 1 mg/ml G418 sulfate (Gemini Biological Products, Calabasas, CA).

Immune Precipitations, GST Pull-Downs, and Immunoblotting. HEK293 cells were transfected in 60-mm plates. Twenty-four hours after transfection, cells were lysed in buffer containing 0.5% Nonidet P-40, 50 mM Tris (pH 7.4), 150 mM NaCl, 5 mM EDTA, 20% glycerol, 20 mM NaF, 1 mM Na₃VO₄, 10 mM Na₃P₂O₇, 25 mM β-glycerophosphate, 1 mM PMSF, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Cell lysates were incubated with anti-FLAG M2-agarose beads (Sigma) for immunoprecipitation or glutathione-sepharose beads (Amersham Pharmacia) for GST pull-down at 4°C for 3 h, respectively. Precipitates were washed four times with lysis buffer and immunoblotted (7).

Raft and Barge Flotation. Lysates for Raft (7) or Raft and Barge (44) flotation assays were made with Triton X-100 extraction buffers and extensive Dounce homogenization. For Raft and Barge flotation, 2 × 10⁷ cells were resuspended for 30 min at 4°C in 0.5 ml of TENT [50 mM Tris (pH 7.4)/150 mM NaCl/5 mM EDTA/0.5% Triton X-100], supplemented with phosphatase and protease inhibitors, and extracted on ice for 30 min. For Raft and Barge flotation, cell lysates were adjusted to 50% with Optiprep (Nycomed, Oslo). One milliliter of the 50% cell extract at 4°C was placed at the bottom of an SW50 centrifuge tube (Beckman Coulter) and overlaid with 1 ml each of 40%, 30%, and 20% Optiprep and 0.4 ml of 10% Optiprep in TENT. After centrifugation at 170,000 × g at 4°C for 4 h, 0.5-ml fractions were aspirated from the top and analyzed by SDS/PAGE and immunoblot.

Results

LMP1TM1-2 Is Required for Signaling. The role of the LMP1 transmembrane domains in signaling was evaluated by transient cotransfection of HEK293 cells with a WT or transmembrane domain mutant LMP1 expression plasmid, a luciferase reporter plasmid that has three MHC class I NF-κB sites upstream of a minimal promoter, and an NF-κB-independent β-galactosidase reporter plasmid to control for transfection efficiency. LMP1 mutants included LMP1TM1-2, LMP1TM3-4, LMP1TM5-6, LMP1TM1-4, and LMP1TM3-6; all were N-terminally Flag-tagged and had at least one R for LMP1 N-terminal anchoring and the full LMP1 cytoplamsic C terminus (Fig. 1 A). As expected, WT LMP1 and LMP1ΔN strongly activated NF-κB in HEK293 cells and LMP1TM5-6, also known as D1LMP1 (9), minimally activated NF-κB (Fig. 1B and data not shown). Surprisingly, LMP1TM3-4 or TM3-6 did not activate NF-κB, whereas LMP1TM1-2 had ≈40% WT LMP1 effects and LMP1TM1-4 had ≈70% WT LMP1 activity (Fig. 1B and data not shown). These data indicate that TM1-2 is required for NF-κB activation and is unique among the transmembrane pairs in providing substantial constitutive activation.

LMP1FWLY_38–41 Is Critical for Signaling. The unique capacity of TM1-2 to constitutively signal led us to attempt to identify specific critical residues within TM1-2. Mutations were introduced into the LMP1 cDNA to target sequences conserved in human and rhesus LMP1 (Fig. 2A). The ring residues in TM1, FWLY₄₁, were also mutated even though they are not fully conserved in rhesus (VWFF₃₈), in part motivated by the putative role of a conserved HIV gp41 W in cholesterol interaction and an observed role in membrane fusion (45).

Fig. 2.

Identification of LMP1TM1-2 residues required for NF-κB activation. (A) Shown is alignment of human and rhesus LMP1TM1-2 amino acids 25–60 and 23–58, respectively. The underlying asterisk highlights identical residues. The sequence of LMP1 A mutants M1-M7 is shown below the WT sequences. The first and the second transmembrane domains are overlined. (B) NF-κB activation in HEK293 cells by Flag-tagged WT or M1-M7 LMP1 mutants as in Fig. 1. Similar results were obtained in three independent experiments.

Surprisingly, LMP1TM1AAAA_30–33, with four consecutive L to A mutations targeting the conserved putative TM1 heptad repeat (15), was similar to WT LMP1 in NF-κB activation (M1 in Fig. 2B). Further, TM1-2 AAA_45–47 mutated for a conserved SDW in the first outer reverse turn (4), TM2A₅₇ mutated for a conserved S, and TM2ASA_56–58, mutated for a partially conserved YSF, were mildly impaired relative to WT LMP1 (M3, M4, and M6 in Fig. 2B). Most significantly, mutations that altered TM1FWLY₄₁ to TM1AAAA₄₁ or TM1AALA₄₁ or included the TM1AALA₄₁ mutation were markedly impaired (M2, M5, and M7 in Fig. 2B). As expected, LMP1 with mutations in the two critical C-terminal cytoplasmic NF-κB activation domains did not activate NF-κB (FDM in Fig. 2B). All WT and mutant LMP1s were expressed at similar levels, and small differences in expression among experiments did not correlate with altered activity (data not shown). These results indicate that TM1FWLY_38–41 is critical for LMP1 constitutive signaling in HEK293 cells, that other conserved residues may contribute to signaling, and that the TM1 putative heptad repeat is not critical.

LMP1FWLY_38–41 Is Required for Constitutive Signaling in Human B Lymphoblasts. Whereas tumor necrosis factor (TNF) receptors associate with TRAFs and activate NF-κB only in response to cognate TNFs, LMP1 constitutively associates with a substantial fraction of the TRAF1 and TRAF3 in LCLs, B lymphoblasts, or epithelial cells and constitutively activates NF-κB (27–32, 46, 47). To evaluate the importance of LMP1TM1FWLY_38–41 in TRAF association in lymphoblasts, Flag-tagged WT LMP1, LMP1AALA₄₁, and LMP1DM, which is mutated for both cytoplasmic signaling domains, were stably expressed at similar levels in BJAB B lymphoblasts. The LMP1 expression level was similar to that of EBV-transformed LCLs. FLMP1, FLMP1AALA₄₁, and FLMP1DM were immunoprecipitated, and the level of stable TRAF3 association was assayed (LMP1AALA₄₁ is M5 in Fig. 3A). As expected, FLAG antibody immunoprecipitation of Flag-tagged WT LMP1 resulted in coimmunoprecipitation of ≈20% of TRAF3, whereas TRAF3 was not detected in coimmunoprecipitates with LMP1DM or LMP1AALA₄₁ (Fig. 3A). Thus, LMP1TM1FWLY_38–41 is critical for constitutive TRAF3 recruitment in lymphoblasts.

Fig. 3.

Role of LMP1FWLY_38–41 in TRAF3 binding and TRAF1 induction. (A) FLMP1 was immunoprecipitated with FLAG-specific M2 antibody-conjugated beads from BJAB cells stably expressing similar amounts of Flag-tagged LMP1 M5 AALA₄₁ mutant, of Flag-tagged WT LMP1, as a positive control, or of Flag-tagged LMP1DM, TRAF and death domain double mutant as a negative control. Cells were lysed and immunoprecipitates were immunoblotted with TRAF3 or LMP1-specific antibody. Whole-cell extract (WCE) is equivalent to 10% of the cell extract. (B) Whole cell lysates were immunoblotted for TRAF1, LMP1, or Lyn. Representative data are shown. Similar data were obtained with two other sets of BJAB-derived stable expressing cell lines.

To more directly assay the role of LMP1FWLY_38–41 in an LMP1 signaling effect on BJAB B lymphoblast cell gene expression, we assayed the level of TRAF1 induction (27, 28) in cells that stably express WT LMP1, LMP1AALA₄₁, or LMP1DM. TRAF1 was up-regulated in WT LMP1 expressing BJAB cells but not in the BJAB cells that stably express LMP1TM1AALA₄₁ or LMP1DM, despite equivalent loading of other cell proteins and equivalent levels of WT and mutant LMP1 expression (LMP1AALA₄₁ is M5 in Fig. 3B). Thus, TM1FWLY_38–41 is critical for constitutive TRAF1 up-regulation.

LMP1FWLY_38–41 Is Critical for Raft and Barge Association in BJAB Lymphoblasts. In lymphoblasts, WT LMP1 is partially associated with detergent-insoluble membrane Rafts and the cytoskeleton, whereas LMP1 with null mutations (LMP1DM) or deletions of both C-terminal cytoplasmic signaling domains is increasingly Raft-associated and not associated with the cytoskeleton, indicating that Raft association is an intrinsic property of the transmembrane domains, whereas cytoskeleton association is mediated by TRAF engagement (5–7, 48–50).

To test the role of LMP1FWLY_38–41 in Raft vs. cytoskeletal association, we compared in parallel the extent of WT LMP1, LMP1AALA₄₁, and LMP1DM association with Rafts and the cytoskeleton in extracts of BJAB cells that stably express WT or mutant LMP1. A significant amount of WT LMP1 was constitutively associated with Raft fractions, more remained in the detergent-soluble cell fractions, and a very substantial amount was associated with the cytoskeleton pellet (Fig. 4A). As expected, a larger fraction of LMP1DM was raft-associated, most of the rest was in the detergent-soluble cell fraction, and almost none was associated with the cytoskeleton pellet (Fig. 4A). In contrast, LMP1AALA₄₁ was almost entirely in the detergent-soluble cell fraction and was neither Raftnor cytoskeleton-associated (Fig. 4 A). These results indicate that LMP1FWLY_38–41 is critical for LMP1 Raft association as well as for signaling-mediated cytoskeleton association.

Fig. 4.

Role of LMP1FWLY_38–41 in LMP1 association with Raft or Barge membrane fractions. (A) Stable Flag-tagged WT LMP1-, LMP1 M5 AALA₄₁-, or LMP1DM-expressing BJAB cells were lysed in 1% Triton X-100 at 4°C, and lysates were fractionated by flotation sucrose step gradient. Fractions were analyzed by immunoblot for LMP1. Ten percent of the WCE is a positive control for the immunoblot. (B) Cells were lysed in 0.5% Triton X-100 at 4°C, and lysates were fractionated by flotation Optiprep gradient density centrifugation (44). Fractions were analyzed by immunoblot for LMP1. Numbers indicate fractions from top of the gradient. WCE is 10% of the starting cell extract used as a positive control. Data shown are representative of three independent experiments.

Because HIV-1 gp41 is partially associated with larger detergent-resistant membrane lipid aggregates, which have been termed the “Barge” fraction (44), we investigated the extent to which LMP1 also associates with larger or higher-density detergent-resistant aggregates. WT LMP1 partially associated with both the Raft and Barge fraction (fractions 2–3 and 5, respectively, in Fig. 4B), and LMP1DM was significantly more proportionately Raft- and Barge-associated (Fig. 4B). In contrast, LMP1AALA₄₁ was absent from Raft and Barge fractions and was completely in the detergent-soluble cell fraction (Fig. 4B). These data indicate that LMP1 associates with both the Raft and Barge fractions, and that LMP1FWLY_38–41 is critical for Raft and Barge association.

LMP1AALA₄₁ Can Associate with WT LMP1, LMP1TM3–6, or LMP1AALA₄₁. To test the relative ability of WT LMP1 and LMP1AALA₄₁ to associate with WT and mutant LMP1, HEK293 were transfected with plasmids expressing Flag-tagged WT or mutant LMP1 and HA-tagged WT or mutant LMP1. Cells were lysed in nonionic detergent, dounce-homogenized, Flag-tagged protein was immunoprecipitated, and the amount of HA-tagged protein that coimmunoprecipitated with the Flag-tagged protein was determined by SDS/PAGE electrophoresis and immunoblot for Flag and HA tags (Fig. 5). Less than 10% of HA-tagged LMP1AALA₄₁ (HM5) coprecipitated with FLMP1 (FWT), ≈10% coprecipitated with FLMP1TM3-6 (FTM3-6), and ≈10% coprecipitated with FLMP1AALA₄₁ (FM5) in an efficient immunoprecipitation (Fig. 5A). At least 20% of WT LMP1 (HWT) coprecipitated with FWT or FLMP1ΔN, and WT LMP1 did not coimmunoprecipitate with FLMP1TM5-6 in multiple experiments (Fig. 5B). Further, ≈10% of HA-tagged WT LMP1 coprecipitated with Flag-tagged LMP1ΔC, or LMP1ΔN/C (Fig. 5 B and C). These results indicate that LMP1AALA₄₁ can still associate with WT LMP1, LMP1AALA₄₁, or LMP1TM3-6, that the LMP1 cytoplasmic N and C termini are not necessary for LMP1 inter-molecular association, and that LMP1TM5-6 does not associate with WT LMP1. This latter result is consistent with the ineffective signaling from LMP1TM5-6 and the absence of a dominant-negative effect of LMP1TM5-6 on WT LMP1 signaling (9, 10).

Fig. 5.

Association of WT or LMP1 M5 AALA₄₁ with WT LMP1 or mutant LMP1 in HEK293 cells. HA-tagged WT or M5 AALA₄₁ mutant LMP1 (HWT or HM5) expression vector was transfected with Flag-tagged WT or mutant LMP1 expression vector into HEK293 cells. After 24 h, cells were lysed, Flag-tagged LMP1 was precipitated with M2 antibody-conjugated beads, and precipitated proteins were analyzed by SDS/PAGE and immunoblot for HA and Flag tags. Data shown are one of three representative experiments. WCE is equivalent to 10% of the WCE.

LMP1TM1-2 and LMP1TM3-4 Can Mediate Intermolecular LMP1 Association. The surprising result that FLMP1AALA₄₁ (M5) inter-molecularly associated with HALMP1AALA₄₁ and with HAWTLMP1 indicates that LMP1 has sequence(s) other than FWLY that can mediate LMP1 intermolecular association. To further investigate the interactions of LMP1TM1-2 and TM3-4 with LMP1, we compared the ability of WT LMP1, LMP1TM1-2ΔC, LMP1TM3-4ΔC, LMP1TM1-4ΔC, and LMP1TM1-6ΔC expressed as GST fusion proteins in HEK293 cells to bind coexpressed FLMP1 or FLMP1TM3-6. GWT, GTM1-6ΔC, GTM1-4ΔC, and GTM3-4ΔC were similar in association with FLMP1 (Fig. 6A and B). In contrast, GTM1-2ΔC associated with WT LMP1 substantially less well (Fig. 6B). These data indicate that LMP1TM3-4 has a substantial role in intermolecular association with LMP1.

Fig. 6.

The role of LMP1TM1-2 and -TM3-4 in LMP1 intermolecular association is shown. (A) Schematic diagram of the GST WT or mutant LMP1 fusion proteins that were expressed with Flag-tagged WT LMP1 or Flag-tagged LMP1TM3-6 in HEK293 cells. The coordinates for these constructs are the same as in Fig. 1A.(B and C) GST-LMP1 was pulled out of lysates that were made 24 h after HEK293 cell transfection. Proteins bound to glutathione beads were analyzed by SDS/PAGE and immunoblot with Flag or GST antibodies. Representative data from one of more than three experiments are shown. WCE is equivalent to 10% of the WCE.

The relative roles of LMP1TM1-2, TM3-4, and TM3-6 in intermolecular interaction were further compared by analyzing the ability of FLMP1TM3-6 to associate with GWT, GTM1-2ΔC, GTM1-4ΔC, GTM1-6ΔC, or GTM3-4ΔC (Fig. 6C). FLMP1TM3-6 associated at a high level with GWT and GTM1-2ΔC and at a moderate level with GTM1-6ΔC, corrected for GTM1-6ΔC expression level (Fig. 6C). FLMP1TM3-6 associated to a lesser extent with GTM3-4ΔC and GTM1-4ΔC (Fig. 6C). These data confirm that LMP1TM1-2 has an important role in intermolecular association with LMP1TM3-6, and that LMP1TM3-4 also has a role in intermolecular association with LMP1TM3-6.

LMP1FWLY_38–41 Is Critical for LMP1TM3-6 Association but Not for LMP1TM1-2 Association. The role of LMP1FWLY_38–41 in inter-molecular binding of LMP1TM1-2 or LMP1TM3-6 was further evaluated by assessing the association of FLMP1TM3-6 or FLMP1TM1-2 with GTM1-2ΔC or GTM1-2ΔC M5 AALA in HEK293 cells. At least 10% of FLMP1TM3-6 associated with GTM1-2ΔC, but LMP1TM3-6 did not associate with GTM1-2ΔC AALA₄₁ (Fig. 7A). In contrast, LMP1TM1-2 associated at ≈2% level with GTM1-2ΔC or GTM1-2ΔC AALA₄₁ (Fig. 7B). These data indicate that LMP1TM3-4 intermolecular association with LMP1TM1-2 is robust and depends on TM1-2 FWLY_38–41, whereas LMP1TM1-2 intermolecular association with TM1-2 is at a lower level and independent of FWLY_38–41.

Fig. 7.

Analysis of the role of LMP1TM1-2 FWLY_38–41 in intermolecular association with LMP1TM1-2 or LMP1TM3-6. HEK293 cells were transfected with expression vectors for GST, GST-LMP1TM1-2ΔC, or GST-LMP1TM1-2ΔC M5 AALA₄₁. After 24 h, cells were lysed, GST-LMP1 was pulled down with glutathionesepharose beads, and the precipitates were analyzed by SDS/PAGE and immunoblot with Flag and GST antibodies. Data shown are representative of three independent experiments. WCE is equivalent to 10% of the WCE.

Discussion

The data reported here begin to explain the constitutive localization of a substantial fraction of LMP1 to large cap(s) in EBV-transformed lymphoblast cell plasma membranes and the ability of those LMP1 aggregates to constitutively signal through their cytoplasmic C termini (4, 31, 49, 51, 52). Importantly, in these experiments we have consistently found high-level intermolecular association among LMP1 molecules, which is mediated by the transmembrane domains. By alanine mutational analysis, transient transfections into HEK293 cells, and NF-κB assays, we identified a unique role for LMP1TM1-2 in inducing NF-κB activation from the LMP1 cytoplasmic C terminus and identified conserved residues in LMP1TM1-2 that contribute to its activity. Of these, the most significant was FWLY_38–41, which was essential for LMP1 association with Raft and Barge fractions, for stable association with TRAF3, and for TRAF1 expression in human B lymphoblasts. The LMP1AALA_38–41 mutation also abrogated LMP1TM1-2 high-level intermolecular association with LMP1TM3-6, without affecting LMP1TM1-2 intermolecular association with LMP1TM1-2. This mutation therefore provides a genetic and biochemical link among LMP1TM1-2 intermolecular association with LMP1TM3-6, WT LMP1 localization to Raft and Barge fractions, and constitutive signaling.

Tryptophan residues have been implicated in hydrophobic ring and Raft interactions; FWLY_38–41 may mediate interactions with other ring residues in LMP1 or with cholesterol-rich microdomains (53–56). The failure of LMP1AALA_38–41 to associate with Raft and Barge membrane microdomains and to signal, supports the hypothesis that LMP1 transmembrane domain-mediated intermolecular aggregation in Raft or Barge microdomains is an important nucleation event for signaling. If so, inhibitors of cholesterol synthesis or Raft formation could affect LMP1 signaling. Irrespective of the precise molecular role of FWLY_38–41 in Raft interaction, FWLY_38–41 is essential for LMP1 signaling, and this sets the stage for the development of assays to identify peptide mimetics or small molecules that can inhibit FWLY_38–41 interactions and LMP1 signaling.

The data presented here identify other LMP1/LMP1 inter-molecular associations that may also be important for constitutive localization in large plasma membrane patches and for signaling. LMP1TM1-2 intermolecularly associated at a high level with LMP1TM3-6 through FWLY_38–41 but also bound at a lower level to LMP1TM1-2; LMP1TM1-2/LMP1TM1-2 association was FWLY_38–41-independent. Further, LMP1TM3-4 inter-molecularly associated with LMP1TM3-6 at a low level and contributed significantly to LMP1TM1-2 signaling in that LMP1TM1-4 had 70% of WT LMP1 signaling, substantially more than LMP1TM1-2 alone. Moreover, within TM1-2, TM2 S₅₇, TM2 YSF₅₈, and TM1-2 SDW₄₇ may also participate in LMP1 C-terminal cytoplasmic domain signaling based on the affect of A point mutations on otherwise WT LMP1-mediated NF-κB activation. Cumulatively, these residues may have significant effects. Thus, these data are consistent with a model in which LMP1 intermolecular aggregation and signaling depend primarily on FWLY_38–41-mediated intermolecular association and secondarily on intermolecular associations of LMP1TM1-2 with LMP1TM1-2 and of LMP1TM3-4 with LMP1TM3-6.

Surprisingly, LMP1TM3-6 did not activate NF-κB. This observation is not likely to be an artifact of protein misfolding or mislocalization. LMP1 constructs used in these transmembrane pair studies had very hydrophilic, DYKDDDDK, Flag epitope N termini followed by at least one R to firmly anchor the N terminus in the cytoplasm. Furthermore, the Flag epitope constructs were expressed at similar levels in transfected HEK293 cells. Moreover, FTM3-6 associated well in HEK293 cells with GST-W T LMP1, GST-TM1-2ΔC, or with FLMP1AALA₄₁ and was specific in associating with GST-TM1-2ΔC and not with GST-TM1-2ΔCAALA₄₁. Thus, the failure of FTM3-6 to activate NF-κB is in striking contrast to the 40% activity of FTM1-2. These data are most consistent with a uniquely important role for TM1-2 in physiologically significant LMP1/LMP1 intermolecular interaction.