IRF7 activation by Epstein–Barr virus latent membrane protein 1 requires localization at activation sites and TRAF6, but not TRAF2 or TRAF3

Abstract

Epstein–Barr virus (EBV) latent infection membrane protein 1 (LMP1), a constitutively aggregated and activated pseudoreceptor, activates IFN regulatory factor 7 (IRF7) through RIP1. We now report that the LMP1 cytoplasmic carboxyl terminal amino acids 379–386 bound IRF7 and activated IRF7. IRF7 activation required TRAF6 and RIP1, but not TRAF2 or TRAF3. LMP1 Y₃₈₄YD₃₈₆, which are required for TRADD and RIP1 binding and for NF-κB activation, were not required for IRF7 binding, but were required for IRF7 activation, implicating signaling through TRADD and RIP1 in IRF7 activation. Association with active LMP1 signaling complexes was also critical for IRF7 activation because (i) a dominant-negative IRF7 bound to LMP1, blocked IRF7 association and activation, but did not inhibit LMP1 induced NF-κB or TBK1 or Sendai virus-mediated IFN stimulated response element activation; and (ii) two different LMP1 transmembrane domain mutants, which fail to aggregate, each bound IRF7 and prevented LMP1 from binding and activating IRF7 in the same cell, but did not prevent NF-κB activation. Thus, efficient IRF7 activation required association with LMP1 CTAR2 in proximity to LMP1 CTAR2 mediated kinase activation sites.

Epstein–Barr virus (EBV) latent infection membrane protein 1 (LMP1) is essential for B lymphocyte transformation into proliferating lymphoblastoid cell line (LCLs) (for review see ref. 1). LMP1 has a 24-aa cytoplasmic N-terminus, six hydrophobic transmembrane domains (amino acids 25 to 186) and a 200-aa cytoplasmic C-terminus (amino acids 187 to 386). The transmembrane domains induce LMP1 aggregation in plasma membrane lipid rafts and barges and constitutively activate two C-terminal cytoplasm signaling domains referred to as C-terminal activation regions (CTAR) or transformation effector sites (TES) 1 and 2, which were initially defined as amino acids 187–231 and 352–386, respectively (Fig. 1A) (2–9). CTAR1 amino acids 201 to 210 engage tumor necrosis factor receptor (TNFR)-associated factors (TRAFs) 1, 3, 2, and 5 and CTAR2 amino acids 376–386 engage TNFR-associated death domain proteins TRADD and RIP1 (4–12). Both signaling domains activate NF-κB, p38 mitogen-activated protein kinase (MAPK), and c-Jun N-terminal kinase, thereby altering cell gene expression, cell survival, and immune responses (13–17).

Fig. 1.

LMP1 activates IRF7 through CTAR2. (A) A schematic representation of LMP1WT, LMP1 1–231(CTAR1), and LMP1 Δ187–351 (CTAR2). TM, transmembrane domain. HEK293 (B) or DG75 (C) cells were cotransfected with pSG5-HA-IRF7 and pSG5 (lane 1), pSG5-FLAG-LMP1 WT (lane 2), pSG5-FLAG-LMP1 1–231 (lane 3) or pSG5-FLAG-LMP1 Δ187–351 (lane 4) along with an ISRE or NF-κB-dependent luciferase and a pGK-β-galactosidase reporter plasmid. To calculate relative luciferase activity, empty vector and LMP1 induced luciferase activities were set to 1 and 100%, respectively.

LMP1 also induces and activates IFN regulatory factor 7 (IRF7) (18–20), a critical transcription factor for type 1 IFN (IFN)-mediated innate and adaptive immune responses (21–24). IRF7 is comprised of DNA binding, constitutive activation, virus activated, inhibitory, and signal response domains and is activated by serine 477 and 479 phosphorylation (25). Although IRF7 is expressed at low levels in nonstimulated cells, IRF7 is critical for type I IFN expression, particularly in plasmacytoid dendritic cells (24, 26, 27). Toll-like receptor activation, LMP1 expression, viral infection, and other signals that activate IRF3 and induce IFN β (IFNβ), can up-regulate IRF7 expression (18, 21–24, 28, 29). IRF7 can bind and localize to MyD88 until ligand-activated Toll-like receptors 7, 8, or 9 engage MyD88 and activate IRF7, which enters the nucleus and up-regulates IFN stimulated response elements (ISREs) to activate IFNα (24, 27, 29, 30). IRF7 accumulation on MyD88 is likely to be important for rapid IFN α and β activation (22–24, 30). Because the same MyD88 residues affect IRF7 binding and activation, the importance of IRF7 binding for activation has been difficult to assess (30). LMP1 residues that activate IRF7 have not been identified (20). We therefore investigated the LMP1 residues required for IRF7 activation, assessed whether IRF7 binding to these residues is critical for activation, and evaluated the role of RIP1, TRAF2, TRAF3, and TRAF6 in IRF7 activation.

Results

LMP1 Primarily Activates IRF7 Through CTAR2.

To assess the roles of the two LMP1 C-terminal signaling domains in IRF7 activation, LMP1 mutants deleted for CTAR1 or CTAR2 (Fig. 1A) were tested for IRF7 activation by using an IFN stimulated response element (ISRE)-dependent luciferase reporter. An NF-κB dependent luciferase reporter was used as a control for a known CTAR1 or CTAR2 activity. Because HEK293 cells do not express detectable IRF7, LMP1 minimally activated an ISRE-dependent luciferase reporter (data not shown). IRF7 expression in HEK293 cells enhanced ISRE luciferase activity 12-fold. LMP1 expression further activated to 200-fold, without altering IRF7 expression levels (Fig. 1B and data not shown). LMP1 deleted for CTAR2 (LMP1 amino acids 1–231) further activated IRF7 mediated ISRE luciferase activity to 40-fold, approximately 3 times the level induced by IRF7 alone, but only 20% of the level induced by IRF7 and LMP1, whereas LMP1 deleted for CTAR1 (LMP1 Δ187–351) activated IRF7-mediated ISRE luciferase activity to >220-fold, more than 5 times the effect of CTAR1 and approximately 110% of LMP1 (Fig. 1B). Thus, almost all LMP1-induced, IRF7-mediated ISRE activation was CTAR2 mediated and CTAR1 independent. In contrast, IRF7 expression had no effect on NF-κB activation, and LMP1, LMP1 deleted for CTAR2, and LMP1 deleted for CTAR1 induced NF-κB activation to 50-, 20-, and 40-fold, respectively, irrespective of IRF7 coexpression (Fig. 1B and data not shown).

The relative roles of CTAR1 and CTAR2 in LMP1 induced IRF7-mediated ISRE effects were also evaluated in DG75 EBV-negative Burkitt's Lymphoma cells, which express TRAF1, an important mediator of CTAR1 effects on NF-κB (5). In DG75 cells, IRF7 expression enhanced ISRE luciferase activation 3-fold, LMP1 activated the IRF7 effects to 30-fold, LMP1 deleted for CTAR2 activated to 9-fold, and LMP1 deleted for CTAR1 activated to 30-fold (Fig. 1C). Thus, CTAR1 had a 3-fold effect and CTAR2 had a 10-fold effect similar to LMP1. In contrast, LMP1 activated NF-κB 15-fold in DG75 cells, whereas LMP1 deleted for CTAR2 activated 7-fold, and LMP1 deleted for CTAR1 activated 10-fold (Fig. 1C). Thus, in DG75 cells, CTAR2 mediated as much IRF7 activation as LMP1, and CTAR1 had much less effect on IRF7 activation, but contributed significantly to NF-κB activation.

IRF7 Binds to LMP1 CTAR2 Amino Acids 376–386 and Not to CTAR1.

In yeast two-hybrid assays (12), Gal-4 DNA binding domain fusions to the LMP1 C-terminal cytoplasmic domain amino acids 180–386 or to CTAR2 amino acids 355–386 interacted weakly with a Gal-4 activation domain fusion with IRF7 amino acids 194–411 or with TRADD amino acids 195–312, giving light blue colonies in β-galactosidase filter assays after a 30-min incubation. In contrast, a Gal-4 DNA binding domain fusion to LMP1 CTAR1 amino acids 180–351 did not interact with IRF7 amino acids 194–411, but interacted strongly with TRAF3, giving blue colonies after a 30-min incubation. Surprisingly, IRF7 interacted similarly with LMP1 CTAR2 amino acids 355–386 with Y₃₈₄YD₃₈₆ mutated to I₃₈₄D, whereas TRADD and RIP1 interacted less with CTAR2 amino acids 355–386 mutated to I₃₈₄D (12).

To determine whether LMP1 can associate with full-length IRF7 in human cells, HEK293 cells were cotransfected with expression vectors for hemagglutinin (HA) epitope-tagged IRF7 and FLAG-tagged LMP1 wild type (WT), LMP1 with the CTAR1 null mutant, AA_204/206 (AA), LMP1 with the CTAR2 null mutant, I₃₈₄D (ID), LMP1 with both CTAR1 and CTAR2 null mutants (DM), or LMP1 amino acids 1–231, which is deleted for CTAR2 (Figs. 1A and 2A and data not shown). Cell lysates were immune precipitated with anti-HA agarose beads and LMP1 binding was assessed by Western blot (Fig. 2A). HA-IRF7 immune precipitated at similar and surprisingly high levels with LMP1 WT, AA, ID, and DM, but not with LMP1 amino acids 1–2-31, which is deleted for CTAR2 (Fig. 2A and data not shown). Although the yeast two hybrid data indicated that IRF7 binds similarly to CTAR2 and CTAR2 with the I₃₈₄D mutation, which is null for all other CTAR2 activities (10, 12), the high level association of IRF7 with LMP1 CTAR2 in HEK293 cells was not anticipated.

Fig. 2.

IRF7 binds to LMP1 amino acids 376–386. (A) HEK293 cells were cotransfected with pSG5-HA-IRF7 and pSG5 (lane 1), pSG5-FLAG-LMP1 WT (lane 2), pSG5-FLAG-LMP1 AA (lane 3), pSG5-FLAG-LMP1 ID (lane 4), or FLAG-LMP1 DM (lane 5). (B) HEK293 cells were cotransfected with pSG5-HA-IRF7 and pSG5-FLAG-LMP1 Δ187–351 WT (lane 1) or deletion mutants, Δ374–386 (lane 2), Δ372–375 (lane 3), Δ366–371 (lane 4), Δ356–365 (lane 5), or Δ351–375 (lane 6). HA-IRF7 immunoprecipitates (IP) and whole cell extracts (WCE) were analyzed by LMP1, HA, and FLAG Western blots (WB).

Because LMP1 requires only C-terminal amino acids 376–386 to mediate TRADD binding and NF-κB activation (10), the residues required for IRF7 binding were further defined. Deletion of CTAR2 amino acids 352–375 or amino acids 372–375 from LMP1 Δ187–351 did not inhibit IRF7 binding, although deletion of amino acids 366–371 reduced IRF7 binding by 50% and deletion of amino acids 374–386 abolished IRF7 binding (Fig. 2B). Notably, deletion of amino acids 352–375 from LMP1 Δ187–351 leaves only amino acids 376–386, which were sufficient for full IRF7 binding (Fig. 2B, lane 6).

Alanine mutations were used to further evaluate LMP1 residues within 376–386 that are required for IRF7 binding, ISRE activation, and NF-κB activation (Fig. 3 and data not shown). Alanine mutations of LMP1 amino acids 376–378 had almost no effect on IRF7 binding, ISRE activation, or NF-κB activation, whereas mutations of amino acids 379–381 or 376–383 almost completely or completely abolished IRF7 binding, ISRE activation, and NF-κB activation (Fig. 3 A and B, and data not shown). Furthermore, alanine mutations of amino acids 382–383 or amino acids 384–386, or mutation of Y₃₈₄YD₃₈₆ to I₃₈₄D did not abrogate IRF7 association, but substantially reduced (I₃₈₄D) or abrogated ISRE and NF-κB activation (Fig. 3 A and B and data not shown). These data indicate that CTAR2 amino acids 379–381 is particularly critical for IRF7 binding and amino acids 379–386 for ISRE and NF-κB activation.

Fig. 3.

LMP1 amino acids 379–381 are critical for both IRF7 binding and activation. (A) HEK293 cells were cotransfected with pSG5-HA-IRF7 and pSG5 (lane 1), pSG5-FLAG-LMP1 Δ187–351 WT (lane 2) or alanine mutants for amino acids 376 to 383 (376–383A, lane 3), 376 to 378 (376–378A, lane 4), 379 to 381 (379–381A, lane 5), or 382 to 383 (382–383A, lane 6). HA-IRF7 IPs and WCEs were analyzed by FLAG and HA WBs. (B) ISRE activation by LMP1 amino acids 376–383 alanine mutants was analyzed by using luciferase reporter assays. HEK293 cells were cotransfected with pSG5, pSG5-FLAG-LMP1 Δ187–351 WT or alanine mutants plus HA-IRF7, ISRE dependent-luciferase and pGK-β-galactosidase reporter plasmids. LMP1 Δ187–351 induced luciferase activity was set to 100% to calculate relative luciferase activities in each experiment.

IRF7 Amino Acids 194–411 Bind Strongly to CTAR2 and Block IRF7 Activation.

Because LMP1 amino acids 379–381 are critical for IRF7 binding and activation, but are also essential for NF-κB and JNK activations (10, 12, 31, 32), assessment of the specific role of IRF7 binding in IRF7 activation required another approach. IRF7 amino acids 194–411 associated with LMP1 CTAR2 at least as strongly as IRF7, but lacks domains that are critical for transactivation, dimerization, and nuclear translocation (Fig. 4A) (25, 33). We therefore tested whether IRF7 amino acids 194–411 has a dominant-negative effect on IRF7 binding to LMP1 in HEK293 cells. IRF7 amino acids 194–411 bound to LMP1 CTAR2. Increasing amounts of IRF7 amino acids 194–411 blocked IRF7 binding to LMP1 CTAR2 (Fig. 4B, compare lanes 2 and 3–5). IRF7 amino acids 194–411 also blocked LMP1 CTAR2-mediated IRF7 activation (Fig. 4C, compare lanes 3 and 5–7), while having no effect on LMP1 CTAR2-mediated NF-κB activation (Fig. 4D, compare lanes 2 and 5–7). The IRF7 amino acids 194–411 dominant-negative effect on IRF7 binding and activation without inhibition of NF-κB activation supports an important role for IRF7 binding to CTAR2 in LMP1-mediated IRF7 activation.

Fig. 4.

IRF7 amino acids 194–411 binds strongly to CTAR2 and inhibits IRF7 activation without affecting NF-κB activation. (A) A schematic representation of IRF7. DBD, DNA binding domain; CAD, constitutive activation domain; VAD, virus activated domain; ID, inhibitory domain; SRD, signal response domain. (B) HEK293 cells were transfected with pSG5-FLAG-LMP1 Δ187–351 (lane 1), pSG5-FLAG Δ187–351 plus pSG5-HA-IRF7 (lane 2), pSG5-FLAG-LMP1 Δ187–351 plus pSG5-HA-IRF7 with increasing amounts of pSG5-Myc-IRF7 194–411 (lanes 3 to 5), or pSG5-FLAG-LMP1 Δ187–351 plus pSG5-Myc-IRF7 194–411 (lane 6, Myc-IRF7 194–411 amounts same as lane 3). FLAG IPs and WCEs were analyzed by FLAG, Myc and HA WBs. The effect of IRF7 194–411 on LMP1 Δ187–351-mediated ISRE (C) or NF-κB (D) activation was analyzed by luciferase reporter assays. LMP1 Δ187–351-induced luciferase activity was set to 100% to calculate relative luciferase activity of each experiment. (E) HEK293 cells were transfected with IRF7 alone (lane 2), TBK1 plus IRF7 (lane 3), TBK1 plus IRF7 194–411 (lane 4), or TBK1 plus IRF7 with increasing amounts of Myc-IRF7 194–411 (lanes 5 to 7). TBK1 plus IRF7-induced luciferase activity was set to 100% to calculate relative luciferase activity of each experiment. (F) HEK 293 cells were transfected with IRF7 (lanes 2 and 3), IRF7 194–411 (lane 4), or IRF7 plus increasing amounts of IRF7 194–411 (lanes 5 to 7). After 8 h, cells were mock-infected (lane 1 and 2) or infected with 200 HA units of Sendai virus (lanes 3 to 7). ISRE luciferase reporter activity was measured at 18 h postinfection. Sendai virus plus IRF7 induced luciferase activity was set to 100% to calculate relative luciferase activity of each experiment.

To exclude an effect of IRF7 amino acids 194–411 on inhibition of an aspect of IRF7 signaling other than LMP1 binding, we investigated the potential effect of IRF7 amino acids 194–411 on TRAF-family member-associated NF-κB activator (TANK)-binding kinase 1 (TBK1)-mediated IRF7 activation in HEK293 cells (Fig. 4E, compare lanes 3 to 5–7). IRF7 expression was necessary for TBK1 mediated ISRE activation and IRF7 194–411 did not replace IRF7. Notably, increasing amounts of IRF7 194–411 did not inhibit TBK1-mediated IRF7 activation of an ISRE (Fig. 4E). Similarly, IRF7 194–411 did not replace IRF7 in mediating Sendai virus-induced IRF7 activation and increasing doses of IRF7 194–411 had only small inhibitory effects on Sendai virus infection-mediated IRF7 activation (Fig. 4F, compare lanes 3 and 7). These data further support the model that IRF7 194–411 specifically inhibits LMP1-mediated IRF7 activation by blocking IRF7 binding to LMP1.

Nonaggregating and Nonsignaling LMP1 Mutants Bind IRF7 and Prevent LMP1-Mediated IRF7 Activation, but Not NF-κB Activation.

LMP1 with F₃₈WLY₄₁ mutated to A₃₈ALA₄₁ is deficient in spontaneous aggregation, localization to lipid rafts, engagement of TRAFs, and NF-κB activation (34, 35). To further evaluate the role of IRF7 binding in LMP1-mediated IRF7 activation, LMP1 A₃₈ALA₄₁ was investigated as a potential nonactivating competitive binder of IRF7. LMP1 A₃₈ALA₄₁ bound approximately 2-fold more to IRF7 than LMP1, indicating that IRF7 binding does not require LMP1 aggregation or signaling (Fig. 5A). The enhanced binding to LMP1 A₃₈ALA₄₁ may be due to deficient IRF7 activation, because activation would result in nuclear translocation. Indeed, LMP1 A₃₈ALA₄₁ was 80% deficient relative to LMP1 in ISRE activation (Fig. 5B, compare lanes 3 and 4) and 70% deficient relative to LMP1 in NF-κB activation (Fig. 5C, compare lanes 2 and 3). Importantly, increasing expression of LMP1 A₃₈ALA₄₁ dominantly repressed WT LMP1 ISRE activation 60%, 80%, and 90% (Fig. 5B, compare lanes 3 and 5–7) and had almost no effect on LMP1 mediated NF-κB activation (Fig. 5C, compare lanes 2 and 4–6). These data indicate that LMP1 A₃₈ALA₄₁ blocks IRF7 activation by sequestering IRF7 from binding to WT LMP1. Thus, binding to aggregated and signaling competent LMP1 is required for IRF7 activation.

Fig. 5.

LMP1 transmembrane domain mutant LMP1 A₃₈ALA₄₁ binds to IRF7 and inhibits LMP1-mediated IRF7 activation without affecting NF-κB activation. (A) HEK293 cells were cotransfected with HA-IRF7 and either FLAG-LMP1 WT (lane 2) or FLAG-LMP1 A₃₈ALA₄₁ (lane 3). HA IPs and WCEs were analyzed by FLAG and HA WBs. (B and C) HEK293 cells were cotransfected with HA-IRF7 and FLAG-LMP1 WT (lane 3), FLAG-LMP1 A₃₈ALA₄₁ (lane 4), or FLAG-LMP1 WT and increasing amounts of FLAG-LMP1 A₃₈ALA₄₁ (lanes 5 to 7). ISRE (B) or NF-κB (C) luciferase reporter activity was measured. LMP1 WT induced luciferase activity was set to 100% to calculate relative luciferase activity of each experiment.

Another signaling-deficient LMP1 mutant, D1LMP1, which is comprised of the last two LMP1 transmembrane domains and the full C-terminus, was also tested for its ability to sequester IRF7 from LMP1 and prevent IRF7 activation, without affecting LMP1 signaling. IRF7 also associated with D1LMP1 at a higher level than with LMP1 [see supporting information (SI) Fig. S1A]. D1LMP1 strongly and dose dependently blocked LMP1-mediated IRF7 activation (Fig. S1B, compare lanes 3 and 5–7) and had almost no effect on LMP1-mediated NF-κB activation (Fig. S1C, compare lanes 2 and 4–6). These data further support the model that IRF7 binding to LMP1 signaling complexes is required for IRF7 activation.

LMP1 Does Not Require RIP1 for IRF7 Association, but Requires RIP1 for IRF7 Activation.

Because LMP1 activation of IRF7 is RIP1 dependent and is associated with both RIP1 and IRF7 K63-linked polyubiquitination (20), we investigated the potential role of RIP1 in LMP1 association with IRF7. LMP1 associated with IRF7 in WT mouse embryonic fibroblasts (MEFs) and at least as well with IRF7 in RIP1 knock out (KO) MEFs, indicating that RIP1 does not mediate IRF7 association with LMP1 (Fig. 6). However, RIP1 was critical for LMP1-induced IRF7 activation, as assayed by ISRE-dependent promoter responses (data not shown) and previously reported (20).

Fig. 6.

RIP1 is not required for LMP1 interaction with IRF7. Control WT (lanes 1 and 2) or RIP1 KO (lanes 3 and 4) mouse embryonic fibroblasts were cotransfected with pSG5-HA-IRF7 and either pSG5 or pSG5-FLAG-LMP1. HA IPs and WCEs were analyzed by LMP1, HA and RIP1 WBs.

TRAF6, but Not TRAF2 or TRAF3, Is Critical for LMP1 CTAR2-Mediated IRF7 Activation.

Because LMP1 induces IRF7 and RIP1 K63-linked polyubiquitination (20), the role of TRAF2, TRAF3, or TRAF6 K63 E3-ubiqutin ligases in IRF7 activation was investigated. LMP1 activated IRF7 equally well in WT and TRAF2 or TRAF3 KO MEFs (Fig. 7). In contrast, LMP1 mediated IRF7 activation was >80% reduced in TRAF6 KO MEFs (Fig. 7). TRAF6 reconstitution in TRAF6 KO MEFs resulted in approximately 30% of WT MEF IRF7 activation and LMP1 expression in the TRAF6 reconstituted MEFs resulted in 180% of WT MEF IRF7 activation (Fig. 7). These data indicate that TRAF6, but not TRAF2 or TRAF3, is required for LMP1 CTAR2 mediated IRF7 activation.

Fig. 7.

TRAF6, but not TRAF2 or TRAF3, is critical for LMP1-mediated IRF7 activation. TRAF2 (A), TRAF3 (B), or TRAF6 (C) WT or KO mouse embryonic fibroblasts were cotransfected with pSG5-HA-IRF7 and pSG5 (lanes 1 and 3) or pSG5-FLAG-LMP1 (lanes 2 and 4) along with ISRE-dependent luciferase and pGK-β-galactosidase reporter plasmids. To reconstitute TRAF6 in TRAF6 KO MEFs, pCDNA3-FLAG-TRAF6 was transfected (lanes 5 and 6). After 50 h, luciferase assays were performed and normalized for transfection efficiency by β-galactosidase activity. To calculate relative luciferase activity, empty vector and LMP1 induced luciferase activities in WT MEFs were set to 1 and 100%, respectively.

Discussion

The data presented here indicate that IRF7 associates at a high level with LMP1 CTAR2 and is induced by LMP1 CTAR2 to efficiently activate of an ISRE responsive promoter. In contrast, IRF7 neither binds nor associates with LMP1 CTAR1 and is not induced by LMP1 CTAR1 to significantly activate an ISRE-responsive promoter. Furthermore, the LMP1 transmembrane domains only require CTAR2 for WT IRF7 binding and activation, indicating that CTAR2 is the principal IRF7 activator.

The high level of IRF7 association with LMP1 CTAR2 in human cells is surprising because IRF7, TRADD, and RIP1 bind CTAR2 at similar low levels in yeast two-hybrid assays and TRADD and RIP1 associate at low levels with LMP1 in human cells (10). Further, mutation of CTAR2 Y₃₈₄YD₃₈₆ did not affect IRF7 binding or association, but abrogated TRADD and RIP1 binding and significantly reduced or abrogated IRF7 and NF-κB (10, 12) activation. These data implicate Y₃₈₄YD₃₈₆-dependent signaling through TRADD, RIP1, or another death domain protein in IRF7 activation. TRADD is implicated in TNFRI and LMP1 CTAR2 activation of NF-κB, JNK, and p38 (10, 12, 36) and our data confirm that RIP1 is essential for LMP1-mediated IRF7 activation (20). Downstream of RIG-I, TRADD assembles FADD, RIP1, TRAF3, and TANK in complexes that induce TBK1 or IKKε-mediated IRF3 and IRF7 phosphorylation (37). TRADD may have a similar role downstream of LMP1.

Our data indicate that IRF7 association with constitutively aggregated LMP1 CTAR2 complexes is critical for efficient IRF7 activation. Abrogating IRF7 association with signaling competent LMP1 CTAR2 complexes by expressing a dominant-negative IRF7 or either of two signaling-incompetent, transmembrane domain mutant LMP1 decoys, blocked IRF7 activation, without affecting NF-κB activation. Thus, LMP1 must stably associate with CTAR2 in the context of aggregated LMP1 molecules, where a nearby CTAR2 can activate an IRF7 kinase.

In binding and activating IRF7, NF-κB, JNK, and p38, LMP1 CTAR2 mimics an activated TLR 7, 8, or 9 (23, 24, 30, 38). Downstream of TLR 7, 8, or 9, IRF7 association with MyD88 may be critical for efficient receptor mediated IRF7 activation (22–24, 30). IRF7 and MyD88 form a death domain complex with IRAK1, which likely recruits TRAF6, TRAF3, or TRAF2 K63-linked E3 ubiquitin ligases, to activate IKKε or TBK1 to phosphorylate IRF7 (22, 30). Similarly, downstream of RIG-I, Cardif assembles a complex with IRF7, TRADD, TRAF3, TANK, FADD, and RIP1, and recruits TBK1 or IKKε to phosphorylate IRF7 (37). The importance of IRF7 assembly on adapters that are part of kinase-activating complexes that mediate IRF7 phosphorylation can likely be assessed by using the decoy LMP1s used here.

The LMP1 CTAR2 downstream kinase(s) remains to be identified and may differ in various EBV-infected cell types. IRF7 can bind to TRAF3 and to RIP1, which may recruit TBK1 or IKKε to phosphorylate and activate IRF7 (20, 23, 24, 38–40). TRADD and to a lesser extent RIP1 directly interact with LMP1 CTAR2. TRADD may further stabilize CTAR2 interaction with RIP1, recruit TRAF6 to K63 polyubiquitinate RIP1 and attract IKKγ, TBK1, or IKKε. IRAK1 is critical for LMP1 mediated NF-κB activation (41, 42) and may also have a role in LMP1-mediated IRF7 activation.

LMP1 activation of IRF7 results in down-regulation of the EBNA1 Qp promoter in latency III (28) and in up-regulation of the LMP1 promoter in latency II and III (33). Activated IRF7 may also alter cell gene transcription and modify IRF4 effects on cell growth (43).

In primary EBV infection, LMP1 induced IFN is likely critical for innate and acquired immune containment of EBV latency III infected B lymphocyte proliferation. LMP1 expression with EBV replication in epithelial cells (44, 45) and with latency III infection in lymphoblasts, should activate IRF7, NF-κB, JNK, and p38, and synergistically induce IFN expression (46–48). IFN can prevent EBV from converting other lymphocytes to proliferating lymphoblasts (49, 50). IRF7 activation is also uniquely critical for plasmacytoid dendritic cell-based adaptive immune T cell responses, which eliminate most latency III infected lymphoblasts (for review, see refs. 24 and 51). Indeed, the potentially oncogenic EBNA2, EBNALP, EBNA3, EBNA3C, and LMP1 latency III transformation program may have not evolved without LMP1 induction of IFN-mediated innate and adoptive immune responses (24). This latter hypothesis can be tested by infection of Rhesus macaques with Rhesus EBV, which has been reverse-genetically modified so as to express a dominant-negative inhibitor of LMP1 mediated IRF7 activation (52, 53).

Materials and Methods

Cells, Plasmids, Antibodies, Transfections, and Reporter Gene Assays.

RIP1 KO MEF was kindly provided by Philip Leder (Harvard Medical School), and pSG5-HA-IRF7 was a gift from Jeff Sample (St. Jude Children's Research Hospital). TRAF3 KO MEF was kindly provided by Genhong Cheng (UCLA), and TRAF2 and TRAF6 KO MEFs were previously described (41). pISRE-Luc plasmid was purchased from Stratagene. pSG5-FLAG-LMP1 Δ187–351 alanine mutated vectors were developed by using the Quikchange mutagenesis kit (Stratagene). Antibodies to HA and Myc were purchased from Santa Cruz Biotechnology. An anti-FLAG antibody was purchased from Stratagene. An anti-RIP1 antibody was purchased from BD Biosciences. An anti-tubulin antibody was purchased from Sigma Aldrich. For transient transfection of HEK293 cells, Effectene Transfection Reagents were used according to the manufacturer's directions (Qiagen. For electroporation of DG75 cells, Bio-Rad Gene Pulser was used as described (54). Luciferase assays were performed as previously described (41, 54). Luciferase data shown here represent three independent experiments.

Yeast Two-Hybrid Analysis.

The screening method and analyses were previously described (10).

Immunoprecipitation.

Ten million cells were lysed in buffer containing 20 mM Tris·HCl (pH 7.5), 100 mM NaCl, 0.5 mM EDTA, 1% nonidet P-40, phosphatase inhibitor mixture (Calbiochem) and protease inhibitor mixture (Roche). Lysates were precleared with protein A/G agarose beads (Santa Cruz) and incubated at 4°C for overnight with either anti-HA antibody (HA-probe F-7, Santa Cruz Biotechnology) or anti-FLAG (Sigma) antibody-conjugated agarose beads. After washing three times with lysis buffer, protein complexes were eluted with either HA (Covance) or FLAG (Sigma) peptides for 30 min at 20°C and analyzed by western blot. The band intensity of western blots was measured by using a Molecular Imaging Station from Kodak. Western blot data shown here represent experiments that were performed three times.