LMP1 Transmembrane Domain 1 and 2 (TM1-2) FWLY Mediates Intermolecular Interactions with TM3-6 To Activate NF-κB

Abstract

The Epstein-Barr virus oncoprotein LMP1 has six transmembrane domains (TMs) that enable intermolecular aggregation and constitutive signaling through two C-terminal cytosolic domains. Expression of both TMs 1 and 2 without the C terminus (TM1-2ΔC) and TMs 3 to 6 fused to the C terminus (TM3-6) results in partial association, which is substantially decreased by TM1 F₃₈WLY₄₁ mutation to A₃₈ALA₄₁. We now investigate whether TM1-2ΔC can functionally interact with TM3-6. TM1-2ΔC induced TM3-6 to mediate NF-κB activation at 59% of LMP1 levels, and the effect was dependent on TM1-2 F₃₈WLY₄₁. TM1-2ΔC even induced TM3-4 C terminus-mediated NF-κB activation to 44% of LMP1 levels. Surprisingly, this effect was TM1 F₃₈WLY₄₁ independent, indicative of a role for TMs 5 and 6 in TM1 F₃₈WLY₄₁ effects. TM3 W₉₈ was also important for TM1-2ΔC induction of TM3-6-mediated NF-κB activation, for association, and for TM1 F₃₈WLY₄₁ dependence on C-terminal NF-κB activation. These data support models in which the TM1 F₃₈WLY₄₁ effects are at least partially dependent on TM3 W₉₈ and a residue(s) in TMs 5 and 6.

The Epstein-Barr virus (EBV)-encoded latent infection membrane protein 1 (LMP1) is essential for EBV-infected lymphocyte outgrowth into lymphoblastoid cell lines (15, 31). LMP1 is expressed in EBV-associated lymphoproliferative disease in immune-deficient people, Hodgkin's disease, and nasopharyngeal carcinoma (for reviews see references 33 and 49). LMP1 can cause established rodent fibroblast cells to grow with less contact inhibition, serum, or anchorage dependence and with greater tumor potential in nude mice (54). In human lymphocytes, LMP1 induces activation markers, adhesion protein expression, cell adhesion, Bcl-2 expression, and antiapoptotic effects. In transgenic mice, immunoglobulin (Ig) enhancer-and promoter-regulated LMP1 expression results in clonal B-cell proliferations (34) and polyoma promoter-regulated LMP1 expression results in epidermal hypertrophy (11). Thus, LMP1 is important for EBV effects on B-lymphocyte and epithelial cell growth and survival.

Reverse genetic analyses identify three LMP1 components that are critical for EBV-mediated transformation of human lymphocytes to lymphoblastoid cell lines: (i) six transmembrane domains (TMs) that enable intermolecular aggregation, association with lipid rafts, and patching in the lymphoblast plasma membrane (3, 9, 22, 24, 37-40, 42, 55, 58); (ii) the first 44 amino acids (aa; aa 187 to 231) of the C-terminal cytosolic domain, which interact with TNF receptor-associated factors (TRAFs) (4, 13, 14, 19, 20, 25, 30, 31, 43-46, 50, 60); and (iii) the last 36 aa (aa 351 to 386) of the C terminus, which interact with death domain proteins, including TRADD and RIP (26-28). The critical LMP1 components mimic a constitutively activated TNFR, which signals through TRAFs and TRADD. The LMP1 C-terminal signaling domains activate NF-κB, p38, and c-Jun N-terminal kinase up-regulation; PI3K and Cdc42 are also activated (12, 16, 17, 21, 35, 48, 53, 55, 56). NF-κB activation is critical for EBV-transformed-lymphoblast survival (7).

The experiments described here investigate intermolecular interactions among TMs to identify interactions that can constitutively enable C terminus-mediated activation of NF-κB. TMs 1 and 2 are critical and even partially sufficient for signaling, since expression of TMs 1 and 2 fused in frame to the C terminus (TM1-2) induces 40% of full-length LMP1 (TM1-6)-mediated NF-κB activation, whereas fused to the C terminus, TMs 3 and 4, TMs 5 and 6, or TMs 3 to 6 (TM3-4, TM5-6, and TM3-6, respectively) induce almost no signaling (10, 58). However, TMs 3, 4, 5, and 6 are also important for TM1-6 signaling, since TM1-2 induce only 40% of TM1-6-mediated NF-κB activation and TMs 1 to 4 fused to the C terminus (TM1-4) induce only 75% of TM1-6-mediated NF-κB activation (58). Within TMs 1 and 2, mutation of 7 of 11 TM1 leucines to alanines does not affect LMP1 intermolecular association but abrogates NF-κB activation (32). However, mutation of four of five TM1 leucines that are conserved in EBV and rhesus lymphocryptovirus LMP1s has no affect on NF-κB activation, indicating that the leucines have a less specific role than anticipated from the more extensive leucine mutagenesis (58). Further alanine mutational analyses of TM1 in the context of full-length LMP1 identifies TM1 residues F₃₈WLY₄₁, which precede the external first reverse turn (Fig. 1A), as critical for raft association, for TM1-2ΔC intermolecular association with TM3-6, and for TM1-6 NF-κB activation (58). Surprisingly, TM1-2ΔC with a mutation of F₃₈WLY₄₁ (TM1-2ΔC A₃₈ALA₄₁) can still associate with LMP1 TM1-2 (58). The potential functional role of TM1-2ΔC and TM1-2ΔC A₃₈ALA₄₁ in mediating NF-κB activation through intermolecular interaction with TM3-6 has not been directly investigated and is the objective of these experiments.

FIG. 1.

(A) Schematic depiction of LMP1 in a membrane with the critical transformation effector sites 1 and 2 (TES1 and TES2). The six TMs are shown. The locations of aromatic amino acids TM1 F₃₈WLY₄₁, TM3 W₉₈, TM4 Y₁₀₆, TM5 Y₁₅₉, TM6 WW_164-5, and TM6 W₁₇₃, which were mutated, are shown. (B) Schematic diagram of LMP1 deletion mutants used in this study. TMs are shown as black, cytoplasmic domains as white, and extracellular domains as gray.

MATERIALS AND METHODS

Cell lines and antibodies.

Human embryonic kidney 293 (HEK293) cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum, 2 mM l-glutamine, penicillin, and streptomycin at 37°C with 5% CO₂. S12 mouse monoclonal antibody against LMP1 was purified from hybridoma supernatant (41). Antibodies to hemagglutinin (HA) (F-7) and glutathione S-transferase (GST) (B14) were purchased from Santa Cruz Biotechnology. Antibody to FLAG (M2) was purchased from Sigma.

Plasmids.

EBV B95-8 strain FLAG-tagged full-length LMP1, FLMP1 (FTM1-6), FTM5-6, and FTM3-6, GST-tagged TM1-2ΔC (GTM1-2ΔC F₃₈WLY₄₁) and GTM1-2ΔC F₃₈WLY₄₁ mutant (GTM1-2ΔC A₃₈ALA₄₁), and HA-tagged full-length wild-type LMP1 (HWT) and its F₃₈WLY₄₁ mutant (HM5) have been described (58). HA-tagged TM1-2ΔC (HTM1-2ΔC) and its mutant HTM1-2ΔC A₃₈ALA₄₁ were created by inserting a stop codon after the amino acid 75 codon with a QuikChange (Stratagene) kit using HWT and HM5 as templates, respectively. HA-tagged TM1-4ΔC (HTM1-4ΔC) and its mutant HTM1-4ΔC A₃₈ALA₄₁ were created by inserting a stop codon after the amino acid 133 codon with the QuikChange (Stratagene) kit using HWT and HM5 as templates, respectively. FTM3-6 W₉₈A, Y₁₀₆A, and W₉₈Y₁₀₆AA double mutants were created by using a QuikChange (Stratagene) kit with FTM3-6 as a template. FTM5-6 Y₁₅₉A, FTM5-6 WW_164-5AA, and FTM5-6 W₁₇₃A mutants were created by using a QuikChange (Stratagene) kit with FTM5-6 as a template. Mutations were confirmed by nucleotide sequencing. A plasmid with five Igκ NF-κB sites preceding a minimal promoter and luciferase reporter and a pGK-β-gal reporter control plasmid were used to evaluate NF-κB activation in transient-transfection assays with HEK293 cells (58).

Transfections and reporter gene assays.

HEK293 cells, in six-well plates, were transfected with 300 ng, or as indicated in Fig. 2, of each expression construct using Effectene (QIAGEN, Chatsworth, CA). The amount of total DNA transfected was kept same by adding pSG5. 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. Luciferase values were normalized for β-galactosidase activity (58).

FIG. 2.

LMP1 TM1-2ΔC induction of TM3-6-mediated NF-κB activation is TM1F₃₈WLY₄₁ dependent. Full-length FLAG-tagged LMP1 (FTM1-6) or FTM3-6 and increasing amounts of HTM1-2ΔC or its mutant HTM1-2ΔC A₃₈ALA₄₁ were expressed in HEK293 cells. The HEK293 cells were also transfected with a plasmid with five Igκ NF-κB sites preceding a minimal promoter and luciferase reporter and a pGK-β-gal reporter control plasmid. Cells were lysed after 24 h and assayed for luciferase, β-galactosidase, and LMP1 C terminus expression. Luciferase values were normalized for β-galactosidase activity. FTM1-6-induced luciferase activity was set to 100% as an internal marker for each experiment. The presented data are representative of three independent transfections done in duplicate. The LMP1 C terminus (CT) was identified with S12 monoclonal antibody (41).

Immunoblotting.

Twenty-four hours after transfection, HEK293 cells were lysed in reporter lysis buffer (Promega). Lysates were resolved by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The proteins were transferred to a nitrocellulose membrane and immunoblotted with S12 antibody to the LMP1 C terminus (41) and F-7 antibody to HA (Santa Cruz Biotechnology).

GST pull-downs (GST PD) and immunoblotting.

HEK293 cells, in 60-mm plates, were cotransfected with 500 ng each of the expression plasmids using Effectene (QIAGEN, Chatsworth, CA). 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 phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Cell lysates were incubated with glutathione-Sepharose beads (Amersham Pharmacia) at 4°C for 3 h. Precipitates were washed four times with lysis buffer and immunoblotted with S12 antibody to the LMP1 C terminus (41) and B14 antibody to GST (Santa Cruz Biotechnology) (58).

Calculation of intermolecular interaction efficiency.

Immunoblots after the GST PD were developed and analyzed on Image Station 4000R (Kodak) using molecular imaging software (Kodak). GST PD efficiency was calculated by comparing band intensities between whole-cell extract (WCE) blotted for GST and GST PD blotted for GST. Similarly efficiency of intermolecular interaction was calculated by comparing band intensities between WCE blotted for the LMP1 C terminus and GST PD blotted for the LMP1 C terminus and correcting it for GST PD efficiency. Intermolecular interaction efficiency between GTM1-2ΔC and FTM3-6 was normalized as 100%, and relative intermolecular interaction percentage was calculated accordingly.

RESULTS

F₃₈WLY₄₁ is critical for TM1-2ΔC induction of NF-κB activation through TM3-6.

To investigate whether N-terminally HA epitope-tagged HTM1-2ΔC, which is TM1-2 without the LMP1 C-terminal cytoplasmic signaling domain, can induce NF-κB activation through intermolecular interaction with FTM3-6, which is FLAG-tagged TM3-6 with the C-terminal cytoplasmic signaling domain, TM1-2ΔC was expressed with FTM3-6 in HEK293 cells (Fig. 1 and 2). A luciferase reporter plasmid with five Igκ NF-κB sites upstream of a minimal promoter and an NF-κB-independent pGK-β-gal reporter plasmid were cotransfected to monitor NF-κB activation and transfection efficiency, respectively. In the experiments reported here β-galactosidase level varied ±10% with no evidence for systematic bias. N-terminally FLAG epitope-tagged LMP1 (FTM1-6) activated NF-κB 30- to 150-fold and was an internal control in each experiment. HTM1-2ΔC or FTM3-6 activated NF-κB at ∼10% of FTM1-6 levels (Fig. 2). Cotransfection of FTM3-6 with increasing amounts of HTM1-2ΔC activated NF-κB at 19%, 60%, and 67% of FTM1-6 levels, respectively (Fig. 2). Thus, TM1-2ΔC can induce TM3-6 to mediate more than 60% of LMP1 NF-κB activation.

Since GST N-terminally tagged TM1-2ΔC (GTM1-2ΔC) with TM1 F₃₈WLY₄₁ mutated to A₃₈ALA₄₁ (GTM1-2ΔC A₃₈ALA₄₁) is deficient in association with FTM3-6 (58), HTM1-2ΔC A₃₈ALA₄₁ was tested for effect on FTM3-6-mediated NF-κB activation. HTM1-2ΔC and HTM1-2ΔC A₃₈ALA₄₁ localize similarly to the plasma membranes of B lymphoblasts in which they are transiently expressed (data not shown). In contrast with HTM1-2ΔC, which induced NF-κB activation through FTM3-6 at 67% of FTM1-6 levels, HTM1-2ΔC A₃₈ALA₄₁ induced NF-κB activation at only 19% of FTM1-6 levels (Fig. 2). HTM1-2ΔC A₃₈ALA₄₁ was expressed similarly to HTM1-2ΔC (data not shown), and FTM3-6 expression was not significantly affected by HTM1-2ΔC or HTM1-2ΔC A₃₈ALA₄₁ expression (Fig. 2). These data indicate that TM1-2ΔC intermolecular interaction with TM3-6 reconstitutes 67% of TM1-6-mediated NF-κB; this reconstitution is mostly dependent on TM1-2 F₃₈WLY₄₁.

LMP1 TM1 F₃₈WLY₄₁ is not critical for NF-κB activation through TM3-4.

Since LMP1 TM1-4 activates NF-κB at nearly 80% of TM1-6 levels (58), we proceeded to determine whether HTM1-2ΔC can also induce TM3-4 to mediate NF-κB activation. HTM1-2ΔC or HTM1-2ΔC A₃₈ALA₄₁ was expressed in cells with FLAG-tagged TM3-4 (FTM3-4) or FTM3-6 (Fig. 3). Individually, HTM1-2ΔC, FTM3-6, and FTM3-4 activated NF-κB at up to ∼12% of FTM1-6 levels (Fig. 3 and Table 1). As expected from the preceding experiments, HTM1-2ΔC activated NF-κB through FTM3-6 at 62% of FTM1-6 levels and HTM1-2ΔC A₃₈ALA₄₁ was deficient and activated FTM3-6 to only 30% of FTM1-6 levels (Fig. 3 and Table 1). HTM1-2ΔC activated NF-κB through FTM3-4 to 41% of the FTM1-6 levels (Fig. 3 and Table 1). Surprisingly, HTM1-2ΔC A₃₈ALA₄₁ also activated NF-κB through FTM3-4 to 40% of FTM1-6 levels (Fig. 3 and Table 1). These data indicate that TM1-2ΔC can activate NF-κB through TM3-4 at a lower level than through TM3-6 and that TM1 F₃₈WLY₄₁ is unimportant for TM1-2ΔC activation of NF-κB through TM3-4. Thus, TMs 5 and 6 are important in TM1 F₃₈WLY₄₁-specific interaction with TM3-6.

FIG. 3.

LMP1 TM1-2ΔC can induce TM3-4-mediated NF-κB activation, and TM1 F₃₈WLY₄₁ is not critical for this activation. NF-κB activation was measured after HTM1-2ΔC or its mutant HTM1-2ΔC A₃₈ALA₄₁ was expressed alone or with FTM3-6 or FTM3-4 in HEK293 cells. NF-κB activation was measured by cotransfection with a plasmid with five Igκ NF-κB sites preceding a minimal promoter and luciferase reporter and a pGK-β-gal reporter control plasmid and analyzed as described in the legend for Fig. 2.

TABLE 1.

Summary of NF-κB activation induced by LMP1 constructs^a

LMP1 (n)	Signaling (% of control ± SD)
	Self	HTM1-2ΔC	HTM1-2ΔC A38ALA41	HTM1-4ΔC	HTM1-4ΔC A38ALA41
FTM1-6 (4)	100
FTM1-4 (4)	78 ± 9
FTM3-6 (4)	12 ± 3	59 ± 3	22 ± 5
FTM3-6 W98A (3)	16 ± 1	45 ± 8	47 ± 2
FTM3-6 Y106A (3)	16 ± 4	75 ± 10	30 ± 5
FTM3-6 W98Y106AA (3)	18 ± 4	52 ± 8	48 ± 3
FTM3-4 (4)	9 ± 2	44 ± 3	46 ± 9
FTM5-6 (4)	6 ± 1	18 ± 3	13 ± 4	20 ± 1*	13 ± 0.2*
FTM5-6 Y159A (3)	9 ± 1			20 ± 3
FTM5-6 WW164-5AA (3)	7 ± 2			19 ± 3
FTM5-6 W173A (3)	9 ± 2			22 ± 3

^aNF-κB activation by full-length FLAG-tagged LMP1 (FTM1-6) was normalized as 100%. NF-κB activation by individual transfection of a LMP1 construct is termed “self.” The values shown are averages from four or three independent experiments (n) done in duplicate, except for values shown with an asterisk, which are from three independent experiments done in duplicate.

TM1-2ΔC and TM1-4ΔC can weakly activate NF-κB through TM5-6.

To determine whether TM1-2ΔC can activate NF-κB through TM5-6, FLAG-tagged TM5-6 (FTM5-6) was expressed in cells with HTM1-2ΔC or HTM1-2ΔC A₃₈ALA₄₁ (Fig. 4 and Table 1). Whereas HTM1-2ΔC induced FTM3-6-mediated NF-κB activation to 62% of FTM1-6 levels, HTM1-2ΔC-induced FTM5-6 mediated NF-κB activation to only 20% of FTM1-6 levels (Fig. 4A and Table 1). In the same experiment, HTM1-2ΔC A₃₈ALA₄₁ increased FTM5-6-mediated NF-κB activation to 17% of FTM1-6 levels (Fig. 4A and Table 1). Although FTM5-6 expression is lower than in comparison to FTM3-6 levels (Fig. 4A), FTM5-6 expression at a higher level than FTM3-6 with HTM1-2ΔC or HTM1-2ΔC A₃₈ALA₄₁ also induced NF-κB activation to only 13% and 11% of FTM1-6 levels, respectively (Fig. 4B and Table 1). These data indicate that TM1-2ΔC can induce low levels of TM5-6-mediated NF-κB activation and that TM1 F₃₈WLY₄₁ may have a small role in TM5-6-mediated NF-κB activation.

FIG. 4.

LMP1 TM1-2ΔC can weakly induce TM5-6-mediated NF-κB activation, and TM1 A₃₈ALA₄₁ has little effect. NF-κB activation was measured after HTM1-2ΔC or its mutant HTM1-2ΔC A₃₈ALA₄₁ was expressed alone or with FTM3-6 or FTM5-6 in HEK293 cells. NF-κB activation was measured by cotransfection with a plasmid with five Igκ NF-κB sites preceding a minimal promoter and luciferase reporter and a pGK-β-gal reporter control plasmid and analyzed as described in the legend for Fig. 2. Two independent experiments, one where expression of FTM5-6 is low in comparison to FTM3-6 (A) and one where expression of FTM5-6 is high (B), are shown.

To evaluate whether TM1-4ΔC can better induce TM5-6-mediated NF-κB activation, HTM1-4ΔC or HTM1-4ΔC A₃₈ALA₄₁ were assayed for induction of FTM5-6-mediated NF-κB activation (Table 1). HTM1-4ΔC was similar to HTM1-2ΔC in inducing FTM5-6-mediated NF-κB activation at 20% of FTM1-6 levels (Table 1). However, HTM1-4ΔC A₃₈ALA₄₁ was more deficient in this context and induced FTM5-6-mediated NF-κB activation to only 13% of the FTM1-6 levels (Table 1). These data indicate that TMs 3 to 6 are an important functional unit in these assays and that TM1 F₃₈WLY₄₁ may have a role in NF-κB activation through TM5-6.

TM5 Y₁₅₉, TM6 WW₁₆₅, and TM6 W₁₇₃ are not critical for TM1-4ΔC induction of TM5-6-mediated NF-κB activation.

The possibility that F₃₈WLY₄₁ in TM1-4ΔC interacts with similarly positioned tyrosines or tryptophans at the external side of TM5 or TM6 was evaluated by comparing NF-κB activation induced by HTM1-4ΔC through FTM5-6 or FTM5-6 mutants (Fig. 1 and Table 1). HTM1-4ΔC activated FTM5-6, FTM5-6 Y₁₅₉A, FTM5-6 WW_164-5AA, and FTM5-6 W₁₇₃A to similar levels of ∼20% of FTM1-6 levels (Table 1). Thus, TM5-6 Y₁₅₉, WW_164-5, and W₁₇₃ individually are not critical for TM1-4ΔC-induced TM5-6-mediated NF-κB activation.

TM3 W₉₈ is important for TM1-2ΔC induction of TM3-6-mediated NF-κB activation.

The possibility that TM1 F₃₈WLY₄₁ may interact with similarly positioned aromatic amino acids at the external side of TM3 or TM4 was also evaluated by comparing levels of HTM1-2ΔC and HTM1-2ΔC A₃₈ALA₄₁ activation of NF-κB through FTM3-6 or FTM3-6 mutated for W₉₈A, Y₁₀₆A, or W₉₈Y₁₀₆AA (Fig. 1 and 5 and Table 1). In the experiment shown in Fig. 5, HTM1-2ΔC-induced FTM3-6 mediated NF-κB activation to 55% of FTM1-6 levels and FTM3-6 Y₁₀₆A mediated NF-κB activation to 82% of FTM1-6 levels (Fig. 5), although the average in repeated experiments was 59% ± 3% and 75% ± 10%, respectively (Table 1). As expected, HTM1-2ΔC A₃₈ALA₄₁ was markedly deficient in FTM3-6- and FTM3-6 Y₁₀₆A-mediated NF-κB activation, with overall inductions of 22% and 30%, respectively, much less than the 59% and 75% with HTM1-2ΔC and marginally above the 12% induction by FTM3-6 alone (Table 1).

FIG. 5.

LMP1 TM1-2ΔC induction of NF-κB activation through TM3-6 is partially dependent on TM3 W₉₈. HTM1-2ΔC or its mutant HTM1-2ΔC A₃₈ALA₄₁ were expressed in HEK293 cells alone or with FTM3-6 or its mutant FTM3-6 W₉₈A, FTM3-6 Y₁₀₆A, or FTM3-6 W₉₈Y₁₀₆AA. NF-κB activation was measured by cotransfection with a plasmid with five Igκ NF-κB sites preceding a minimal promoter and luciferase reporter and a pGK-β-gal reporter control plasmid and analyzed as described in the legend for Fig. 2.

Most interestingly, HTM1-2ΔC induced lower NF-κB levels through FTM3-6 mutants W₉₈A and W₉₈Y₁₀₆AA than through FTM3-6 or FTM3-6 mutant Y₁₀₆A (Fig. 5 and Table 1). HTM1-2ΔC-induced FTM3-6 W₉₈A and FTM3-6 W₉₈Y₁₀₆AA mediated NF-κB activation to only 45% and 52%, respectively, versus 59% for FTM3-6 levels (Table 1 and Fig. 5). These data indicate a role for TM3 W₉₈ in TM1-2ΔC induction of TM3-6-mediated NF-κB activation.

Equally as surprising, HTM1-2ΔC A₃₈ALA₄₁ did not differ from HTM1-2ΔC and induced FTM3-6 W₉₈A- and FTM3-6 W₉₈Y₁₀₆AA-mediated NF-κB activations to 47% and 48%, respectively, of the FTM1-6 levels (Table 1 and Fig. 5). Thus, TM3 W₉₈ is important for TM1-2ΔC induction of TM3-6-mediated NF-κB activation and specifically for TM1 F₃₈WLY₄₁ effective induction of TM3-6-mediated NF-κB activation.

TM3 W₉₈A is important for intermolecular association with TM1-2ΔC.

The potential effect of TM3 W₉₈A on TM1-2ΔC association with TM3-6 was investigated by transfecting GST-tagged TM1-2ΔC (GTM1-2ΔC) or GTM1-2ΔC A₃₈ALA₄₁ and FTM3-6, FTM3-6 W₉₈A, FTM3-6 Y₁₀₆A, or FTM3-6 W₉₈Y₁₀₆AA into HEK293 cells. The transfected cells were lysed, and GST-tagged proteins were pulled down with glutathione-conjugated Sepharose beads. The lysates and the immunoprecipitates were immunoblotted for GST and TM3-6 and analyzed with Kodak Imager (Fig. 6). In general, GTM1-2ΔC associated with ∼24% (normalized to 100%) of FTM3-6, whereas GST did not associate with FTM3-6 (Fig. 6). GTM1-2ΔC associated with 120% of FTM3-6 Y₁₀₆A (Fig. 6), consistent with the slightly higher NF-κB activation than with FTM3-6 (Fig. 5 and Table 1). Interestingly, GTM1-2ΔC associated with 68% and 54% of FTM3-6 W₉₈A and W₉₈Y₁₀₆AA, respectively, indicative of slight deficiency relative that of FTM3-6 levels (Fig. 6). Furthermore, GTM1-2ΔC A₃₈ALA₄₁ was substantially deficient in intermolecular association with FTM3-6 at 15% in comparison to 100% with GTM1-2ΔC (Fig. 6). GTM1-2ΔC A₃₈ALA₄₁ was similarly deficient in association with FTM3-6 Y₁₀₆A and was slightly less deficient in association with FTM3-6W₉₈A and W₉₈Y₁₀₆AA, with relative intermolecular associations of 21% and 24%, respectively (Fig. 6). These data are consistent with a model in which TM1 F₃₈WLY₄₁ and TM3 W₉₈ are important for TM1-2 association with and signaling from TM3-6 and TM3 W₉₈ is important for TM1 F₃₈WLY₄₁.

FIG. 6.

LMP1 TM1-2ΔC interaction with TM3-6 is partially dependent on TM3W₉₈. GST alone, GST-tagged TM1-2ΔC (GTM1-2ΔC), or its mutant GTM1-2ΔC A₃₈ALA₄₁ was expressed in HEK293 cells with FTM3-6, FTM3-6 W₉₈A, FTM3-6 Y₁₀₆A, or FTM3-6 W₉₈Y₁₀₆AA. After 24 h, GST, GTM1-2ΔC, or GTM1-2ΔC A₃₈ALA₄₁ was adsorbed to glutathione-conjugated Sepharose beads. The proteins pulled down by the beads were eluted with sodium dodecyl sulfate gel loading buffer (GST PD) and were identified by polyacrylamide gel electrophoresis and immune blotted for the LMP1 C terminus (CT) (top) or GST (bottom). Three percent of the WCE was also immune blotted for the LMP1 C terminus (middle). The data were analyzed by using Molecular Imaging software (Kodak). The band intensity for wild-type intermolecular association with GTM1-2ΔC was normalized as 100%, and relative intermolecular association percentages were computed accordingly. The data shown are representative of three independent experiments.

DISCUSSION

These experiments further investigate the role of LMP1 TMs in intermolecular interactions that induce C-terminal signaling as measured by NF-κB activation. The data confirm a special role for TMs 1 and 2, for TM1-2ΔC interaction with TM3-6, and for TM1 F₃₈WLY₄₁ in LMP1 intermolecular interactions and provide a surprisingly robust functional dimension to these interactions. TM1-2ΔC induced TM3-6 C terminus-mediated NF-κB activation to 59% of the LMP1 levels, and the effect was critically dependent on TM1 F₃₈WLY₄₁. TM1-2ΔC even induced TM3-4-mediated NF-κB activation at a reduced level of 44%, and this effect was TM1 F₃₈WLY₄₁ independent, indicative of an important role for TMs 5 and 6 in TM1 F₃₈WLY₄₁ effects. Furthermore, TM3 W₉₈ was deficient in TM1-2ΔC induction of NF-κB activation and association and was not afffected by TM1 F₃₈WLY₄₁ mutation.

The LMP1 TMs are remarkably hydrophobic. TM1, -2, -3, -4, -5, and -6 have 15, 17, 14, 17, 17, and 13 leucines, isoleucines, phenylalanines, and valines, respectively (18), whereas 6 hydrophobic residues can be adequate for membrane insertion (23, 29). These very hydrophobic domains are likely to engage in intramolecular interactions between antiparallel adjacent α-helices, perhaps similar to the interactions described for rhodopsin alpha-helical transmembrane domains (36, 47). Interactions between antiparallel alpha helices may have contributed to the evolution of TM3 W₉₈ interaction with TM1 F₃₈WLY₄₁ in intermolecular aggregation and signaling, since TM1 and TM3 are both linked to TM2 and are antiparallel to TM2.

Accumulated data are consistent with models in which constitutive LMP1 signaling is due to intermolecular interactions among LMP1 TM segments. TM1-2 can intermolecularly aggregate and constitutively signal (10, 58). Furthermore, TM1-2ΔC can intermolecularly associate with and induce TM3-6 to activate NF-κB; these effects are TM1 F₃₈WLY₄₁ dependent (58). Moreover, the studies presented here define a role for TM3 W₉₈ in effecting the intermolecular interaction of TM1 F₃₈WLY₄₁ with TM3-6 and constitutive signaling through TM3-6. Tryptophan and tyrosines are frequently at the outer margin of TMs and together with phenyalanines are implicated in hydrophobic stacking interactions, including interactions among aromatic residues and cholesterol-rich membrane lipid raft microdomains (1, 2, 8, 51, 52, 57-59); TM1 F₃₈WLY₄₁ is critical for stable LMP1 interaction with lipid rafts (58). The epistatic effect between mutations in TM1 F₃₈WLY₄₁ and TM3 W₉₈ is consistent with direct or indirect physical interaction of TM1 F₃₈WLY₄₁ and TM3 W₉₈. This stacking interaction could stabilize LMP1 intermolecular associations and contribute to signaling complex stability, consistent with the importance of TM3 W₉₈ in TM1 F₃₈WLY₄₁ association with TM3-6 and in signaling through TM3-6. However, in the context of transient LMP1 expression in 293 cells, as described here, LMP1 with the W₉₈A mutation is not significantly different from full-length wild-type LMP1 (TM1-6) in signaling and LMP1 with the W₉₈Y₁₀₆AA mutation is only marginally decreased (data not shown). The data are therefore most consistent with a model in which other similarly positioned aromatic residues in LMP1 can substitute for TM3W₉₈ in supporting intermolecular interactions of TMs 3 to 6 with TMs 1 and 2.

Intermolecular associations among LMP1 TMs 1 and 2 have been previously described (58), and LMP1 TMs 1 to 6 without the C-terminal cytoplasmic signaling domains have also been reported to activate a signaling-deficient LMP1 with seven TM1 leucines mutated to alanines (32). The data presented here are the first demonstration that TM1-2ΔC can induce robust NF-κB activation through intermolecular interactions with TMs 3 to 6 and the induction is TM1 F₃₈WLY₄₁ and TM3 W₉₈ dependent.

These LMP1 TM intermolecular interactions and constitutive signaling effects begin to explain the role of the TMs in enabling constitutive LMP1 signaling. Lymphoblastoid cell lines are dependent on LMP1-mediated NF-κB activation (5-7), and inhibition of TM1-2 intermolecular interactions with FTM3-6 is a potential therapeutic target for EBV-associated lymphoproliferative disease, Hodgkin's disease, and nasopharyngeal carcinoma.