The Role of TRAF2 Binding to the Type I Interferon Receptor in Alternative NFκB Activation and Antiviral Response*

Type I interferons (IFNs) play critical roles in the host defense by modulating gene expression through the IFN-dependent activation of STAT and NFκB transcription factors. Previous studies established that IFN activates NFκB through a classical NFκB pathway that results in IκBα degradation and formation of p50-containing NFκB complexes, as well as an alternative pathway that involves NFκB-inducing kinase and TRAF2, which results in the formation of p52-containing NFκB complexes. In this study, we examined the interaction of TRAF proteins with the type I IFN receptor. We found that TRAF2 was directly coupled to the signal-transducing IFNAR1 subunit of the IFN receptor. By immunoprecipitation, overexpression of epitope-tagged IFNAR1 constructs, and glutathione S-transferase pulldown experiments, we demonstrate that TRAF2 rapidly binds to the IFNAR1 subunit of the IFN receptor upon IFN binding. The membrane proximal half of the IFNAR1 subunit was found to directly bind TRAF2. Moreover, analysis of mouse embryo fibroblasts derived from TRAF2 knock-out mice demonstrated that TRAF2 plays a critical role in the activation of the alternative NFκB pathway by IFN, but not the classical NFκB pathway, as well as in the antiviral action of IFN. Our results place TRAF2 directly in the signaling pathway transduced through the IFNAR1 subunit of the IFN receptor. These findings provide an important insight into the molecular mechanisms by which IFN generates signals to induce its biological effects.

The family of tumor necrosis factor (TNF) 2 receptor-associated factors (TRAFs) functions as adaptor molecules for TNF superfamily members by associating with the intracellular domain of these receptors to mediate downstream signaling events such as NFB and AP-1 activation (1). TRAF proteins, originally discovered because of their ability to bind to the p75 TNF receptor, represent a family of six proteins containing a conserved C-terminal domain, which enables their interaction with various members of the TNF receptor superfamily. Moreover, TRAF proteins can also integrate NFB activation to signal transduction by other receptors, including the family of Toll-like receptors, the LMP1 protein of Epstein-Barr virus, and lymphotoxin-␤ (2). TRAF proteins play critical roles in innate immune responses, inflammatory processes, and programmed cell death, which are also processes regulated by the type I IFN family of proteins (IFN␣, IFN␤, and IFN). However, the role of TRAF proteins in the cellular response to IFNs has been explored only recently (3,4). Type I IFNs bind to a ubiquitously expressed cell-surface receptor comprising the IFNAR1 and IFNAR2 chains. The IFN signal transduction pathway involves the recruitment of STAT proteins to the IFN receptor subunits and STAT activation by receptor-associated Janus kinase tyrosine kinases (5)(6)(7).
Besides the classical JAK/STAT signaling pathway, type I IFNs also activate the NFB transcription factor that regulates the expression of genes involved in cell survival and in immune and inflammatory responses (8 -11). In mammals, the NFB protein family includes NFB1 (p105 processed to p50), NFB2 (p100 processed to p52), RelA, RelB, and c-Rel. Although p50/ RelA and p52/RelB heterodimers are the NFB complexes most often observed in cells, other combinations of Rel homodimers and heterodimers also form. The NFB1 and NFB2 precursor proteins undergo proteolytic processing into the p50 and p52 proteins, respectively. Although the processing of p105 is constitutive and largely cotranslational, the proteolytic processing of p100 protein results from its ligand-inducible phosphorylation and subsequent ubiquitinylation.
In common with a variety of stimuli, IFN␣/␤ promotes the dissociation of the cytosolic inactive NFB-IB complexes via the serine phosphorylation and degradation of IB, leading to NFB translocation to the nucleus and DNA binding (12), which is denoted as the classical NFB activation pathway. IFN␣/␤ also induces NFB activation by an alternative pathway dependent on NFB-inducing kinase and TRAF proteins that results in the processing of the p100/NFB2 precursor into p52. We showed previously that expression of a dominant-negative TRAF2 construct inhibited IFN-promoted NFB activation (8), suggesting a role for TRAF2 in coupling of the IFN receptor signaling to NFB activation.
In this study, we examined the interaction of TRAF proteins with the type I IFN receptor. We found that TRAF2 binds to the signal-transducing IFNAR1 subunit of the type I IFN receptor.
By immunoprecipitation, overexpression of epitope-tagged IFNAR1 constructs, and GST pulldown experiments, we demonstrate that TRAF2 rapidly bound to the IFNAR1 subunit upon IFN addition to cells. The interaction of TRAF2 with the IFNAR1 subunit was localized to the membrane proximal portion of the receptor. Furthermore, analysis of mouse embryo fibroblasts (MEFs) derived from TRAF2 knock-out (KO) MEFs demonstrated that TRAF2 plays an important role in coupling alternative NFB activation to the type I IFN receptor and in the induction of antiviral activity by IFN.

EXPERIMENTAL PROCEDURES
Biological Reagents and Cell Culture-Recombinant human IFN␣ (IFN␣Con1) and rat IFN␤ were provided by InterMune (Brisbane, CA) and Biogen Idec, Inc. (Cambridge, MA), respectively. The biological activity of the IFN preparations was expressed in terms of international reference units/ml using the appropriate National Institutes of Health reference standard as described previously (13). Human Daudi cells were cultured in RPMI 1640 medium with 10% bovine calf serum. HT1080 fibrosarcoma cells and mouse L cells expressing various IFNAR1 constructs were cultured in Dulbecco's modified Eagle's medium with 10% bovine calf serum (14). Wild-type (WT) and TRAF2-KO MEFs were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (15). Cells were maintained in the presence of penicillin (100 IU/ml) and streptomycin (100 g/ml) at 37°C with 5% CO 2 .
Constructs and Transfection Conditions-The cDNAs for TRAF2 and IFNAR1 were amplified by PCR and cloned into the pcEF expression vector that provides a c-Myc epitope tag at the C terminus of the protein (7). Transfection of cells (10 7 ) was accomplished by electroporation (capacitance of 300 microfarads, 250 V) with 20 g of plasmid DNA and 500 g of salmon sperm DNA. Stable transfectants were selected for neomycin resistance (0.4 mg/ml G418).
GST Fusion Constructs and GST Pulldown Assays-cDNAs encoding amino acids 462-557 (AR1), 462-507 (AR1 N ), and 507-557 (AR1 C ) of the IFNAR1 intracellular domain were amplified from IFNAR1 cDNA and cloned into the EcoRI and XhoI sites of pGEX-KG, which provides the proteins with an amino-terminal GST tag (17). GST fusion proteins were expressed in Escherichia coli strain BL21(DE3) by transformation with the plasmid construct and affinity-purified on glutathione-Sepharose (GE Healthcare) as described previously (18). For GST pulldown assays, lysates from control or IFN-treated (ϳ2 ϫ 10 7 ) cells were incubated with GST fusion proteins bound to glutathione-agarose beads at 4°C overnight. The bound proteins were washed extensively and eluted with Laemmli buffer, resolved by SDS-PAGE (7.5%), blotted onto PVDF membranes, and probed as indicated.
Antiviral and Antiproliferative Assays-To determine antiviral activity, MEF cultures were incubated overnight with rat IFN␤, followed by infection with vesicular stomatitis virus (VSV) or encephalomyocarditis virus (EMCV) for 2 h at 0.1 plaque-forming units/cell. At 24 h post-infection, the VSV yield in the medium was assayed by plaque formation on Vero cells as described previously (11). At 24 h post-infection, the cytopathic effect of EMCV on STAT1-KO MEF cultures was determined by uptake of the tetrazolium dye 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide as a measure of cell viability.
To determine antiproliferative activity, MEF cultures were plated at 1 ϫ 10 5 cells in 25-cm 2 flasks and treated with rat IFN␤. 3 days after IFN addition, cells were harvested by trypsinization and enumerated in a Coulter Counter (19).

Binding of TRAF2 to the IFNAR1 Chain of the Type I IFN
Receptor-We reported previously that expression of a dominant-negative TRAF2 construct in the highly IFN-sensitive human Daudi lymphoblastoid cell line blocked IFN-promoted NFB DNA-binding activity and p100 processing (8). Because TRAF proteins by definition are receptor-associated factors, we investigated whether TRAF2 is associated with the IFN receptor. Lysates were prepared from control and IFN-treated (1,000 units/ml, 15 min) Daudi cells, immunoprecipitated with antibodies directed against the extracellular domains of the IFNAR1 or IFNAR2 subunit (20), and then immunoblotted with Abs directed against TRAF2. As shown in Fig. 1A, although TRAF2 was found not to be basally associated with either the IFNAR1 or IFNAR2 chain of the IFN receptor, IFN treatment induced the association of TRAF2 with the IFNAR1 subunit, but not with the IFNAR2 subunit. Next we examined the time course of TRAF2 association with IFNAR1. Daudi cells were treated with IFN for varying times, lysed, and immunoprecipitated with anti-IFNAR1 antibody. As shown in Fig. 1B, IFN induced the association of TRAF2 with the IFNAR1 subunit within 5 min, and TRAF2 remained bound to IFNAR1 up to 1 h after IFN addition. In contrast, although TRAF6 was expressed in Daudi cells, TRAF6 was not bound to IFNAR1 either before or after IFN treatment (data not shown). The results demonstrate that IFN induces the specific association of TRAF2 with the IFNAR1 subunit of the IFN receptor. IFN signaling involves both JAK-mediated tyrosine phosphorylation and PI3K/Akt-mediated serine phosphorylation. To determine whether phosphorylation plays a role in the IFN-induced interaction of IFNAR1 with TRAF2, Daudi cells were treated with genistein (a tyrosine kinase inhibitor) or LY294002 (a PI3K inhibitor) as we have used previously (10,13,21). As shown in Fig. 1C, inhibitors of either IFN-activated signaling pathway reduced the IFN-induced interaction of IFNAR1 with TRAF2. However, this approach does not discern whether phosphorylation of TRAF2 and/or IFNAR1 is involved in this interaction, which is addressed by GST pulldown assays.
Role of the Intracellular Domain of IFNAR1 in IFN-dependent TRAF2 Binding-TRAF proteins mediate signal transduction by interacting with the intracellular domain of various members of the TNF receptor superfamily. To further characterize the interaction of TRAF2 with the IFNAR1 subunit, Daudi cells were transiently transfected with various Myc epitope-tagged IFNAR1 constructs. 48 h after transfection, cell lysates prepared from control and IFN-treated (1,000 units/ml, 15 min) Daudi cells were immunoprecipitated with anti-c-Myc antibodies and immunoblotted for IFNAR1 or TRAF2. As shown in Fig. 2A, IFN induced the association of TRAF2 with full-length IFNAR1, but not with IFNAR1 without its intracellular domain. Full-length and intracellular domain-deleted IFNAR1 were expressed at similar levels in transiently transfected Daudi cells. These results demonstrate that the interaction of TRAF2 with IFNAR1 is IFN-dependent and localize the site of TRAF2 interaction to the intracellular domain of IFNAR1.
To define the region of the intracellular domain of IFNAR1 required for TRAF2 binding, various portions of the IFNAR1 chain were expressed as GST fusion proteins, and GST pulldown assays were performed. Daudi and HT1080 fibrosarcoma cells were treated with IFN (1,000 units/ml, 15 min), cell extracts were prepared and incubated with GST fusion proteins, and GSTbound material was immunoblotted for TRAF2 or GST. As shown in Fig. 2B, none of the IFNAR1-GST constructs bound TRAF2 in extracts prepared from control cells. However, the full-length IFNAR1 intracellular domain as well as the membrane proximal, amino-terminal half of the intracellular domain of IFNAR1 (AR1 N ) bound TRAF2 from extracts prepared from IFN-treated Daudi cells or HT1080 cells (Fig. 2B). TRAF2 was not bound to the carboxyl-terminal half of the IFNAR1 intracellular domain (AR1 C ) in extracts prepared from control or IFN-treated cells, although the various IFNAR1-GST constructs were present at equivalent amounts. These results demonstrate that the membrane proximal half of the intracellular domain of the IFNAR1 chain directly binds TRAF2 upon IFN treatment. Pretreatment of Daudi cells with genistein or LY294002 blocked the IFN-induced binding of TRAF2 to the full-length IFNAR1 intracellular domain-GST construct (Fig.  2C). Moreover, because the GST constructs do not undergo IFN-induced phosphorylation, the results presented in Figs. 1 and 2 indicate that upon IFN treatment, phosphorylated TRAF2 binds to the membrane proximal half of IFNAR1.
Activation of the Alternative NFB Pathway Is Directed through the IFNAR1 Chain-IFN activates a TRAF-dependent alternative NFB pathway, which results in the formation of p52-p65 NFB complexes (8). We established previously that when expressed in mouse L cells, the human IFNAR1 chain functions as a species-specific transducer for human IFN (14,22). Utilizing this system, we examined whether signaling mediated through the intracellular domain of IFNAR1 is required for IFN-induced p100 processing. Mouse L cells expressing either full-length human IFNAR1 or an IFNAR1 chain lacking the intracellular domain were treated with human IFN␣ (1,000 units/ml, 15 min), assayed by pulldown assays with p65-GST, and immunoblotted with p52. As shown in Fig. 3A, IFN induced the formation of p52-p65 complexes in mouse cells expressing full-length human IFNAR1, but not in cells transfected with the human IFNAR1 chain lacking the intracellular domain or an empty vector construct. Moreover, p52-p65 complexes were not detected in either transfectant  MAY 23, 2008 • VOLUME 283 • NUMBER 21

Role of TRAF2 in Interferon Action
in the absence of IFN treatment. These results demonstrate 1) the species specificity of the interaction between the human IFNAR1 chain and human IFN and 2) that the intracellular domain of the human IFNAR1 is required for induc-tion of the alternative NFB signaling pathway by human IFN.
Role of TRAF2 in IFN-induced NFB Signaling Pathways-To further define the role of TRAF2 in the IFN-induced activation of the NFB signaling pathway, we employed fibroblasts derived from TRAF2-KO MEFs. IFN induces IB␣ degradation (the classical NFB pathway), as well as an alternative NFB pathway that is independent of IB degradation (10). To examine the contribution of TRAF2 to the classical NFB pathway, IFN-promoted IB␣ degradation was assessed in extracts prepared from WT and TRAF2-KO MEFs treated with IFN for varying times. As shown in Fig. 3B, IFN induced IB␣ degradation with a similar time course in both WT and TRAF2-KO MEFs, indicating that TRAF2 is not required for IFN-induced IB degradation. To determine whether TRAF2 is required for induction of the NFB-dependent DNA binding, TRAF2-KO MEFs were stimulated with IFN␤, and NFB activation was examined by EMSA with a consensus B oligonucleotide probe. As shown in Fig. 3C, IFN induced NFB DNA-binding activity within 30 min, and activity returned to near base-line levels by 2 h after IFN addition. The kinetics of NFB activation are similar to what we have observed previously in IFN-treated MEFs (11). Because TNF also induces NFB activity via the classical pathway (23), TRAF2-KO MEFs were treated with TNF for comparison. As shown in Fig. 3C, TNF also induced NFB activation in TRAF2-KO MEFs. However, the kinetics of NFB induction by TNF, as well as the NFB complexes , which was bound to glutathione-agarose beads (pulldown assays). The proteins were resolved by SDS-PAGE, blotted onto PVDF membranes, and probed with anti-TRAF2 or anti-GST antibody. Blots were visualized by enhanced chemiluminescence. C, Daudi cells were pretreated with Me 2 SO as a vehicle (veh), genistein (gen; 100 M), or LY294002 (Ly; 10 M) for 1 h prior to IFN addition (1,000 IU/ml), and after IFN treatment for 30 min, cell lysates were incubated with AR1-GST fusion protein bound to glutathione-agarose beads. The proteins were resolved by SDS-PAGE, blotted onto PVDF membranes, probed with anti-TRAF2 antibody, and visualized by enhanced chemiluminescence.  (14) were treated with human IFN␣ (1,000 units/ml, 15 min). Cell lysates were prepared and incubated with p65-GST fusion protein bound to glutathione-agarose beads (pulldown assays). The proteins were resolved by SDS-PAGE, blotted onto PVDF membranes, and probed with anti-p52 antibody. Blots were visualized by enhanced chemiluminescence. B, cell lysates prepared from WT or TRAF2-KO MEFs treated with rat IFN␤ (1,000 units/ml) were resolved by SDS-PAGE, blotted onto PVDF membranes, probed with anti-IB␣ or anti-actin antibody for a loading control, and visualized by enhanced chemiluminescence. C, nuclear extracts were prepared from Daudi cells treated with IFNCon1 (1,000 units/ml) or TNF (10 ng/ml) for varying times and subjected to EMSA. D, nuclear extracts were prepared from Daudi cells treated with IFNCon1 (1,000 units/ml, 30 min) and incubated with anti-p50 or anti-p65 antibody or a 50-fold excess of unlabeled B oligonucleotide probe (cold) prior to EMSA.
formed, were clearly different from those induced by IFN treatment. In addition, supershift assays with Rel antisera demonstrate the induction of p50-containing NFB complexes in nuclear extracts of IFN-treated TRAF2-KO MEFs (Fig. 3D). Moreover, anti-p52 Ab caused no supershift of IFN-induced complexes in TRAF2-KO cells (data not shown). Taken together, these results demonstrate that TRAF2 is not required for the activation of the classical NFB signaling pathway by IFN.
To directly examine the contribution of TRAF2 to the alternative NFB pathway, IFN-promoted formation of p52-p65 complexes was assessed by p65-GST pulldown assays with extracts prepared from WT and TRAF2-KO MEFs treated with IFN for varying times. As shown in Fig. 4A, whereas IFN induced the formation of p52-p65 complexes in WT MEFs, these NFB complexes were not detected in IFN-treated TRAF2-KO MEFs. Immunoblotting with p50 did not detect the formation of p50-p65 complexes. To confirm the role of TRAF2 in this pathway, TRAF2-KO MEFs were reconstituted with TRAF2. As shown in Fig. 4A, the level of TRAF2 expression in TRAF2-reconstituted TRAF2-KO MEFs was similar to that observed in WT MEFs. Most importantly, reconstitution of TRAF2-KO cells with TRAF2 restored the IFN-induced formation of p52-p65 complexes. These results demonstrate that formation of p52-p65 complexes induced by IFN is TRAF2-dependent.
The JAK/STAT pathway is critical in IFN signaling. To determine whether TRAF2 affects JAK/STAT signaling, we also assessed STAT activation by IFN in WT and TRAF2-KO MEFs. As shown in Fig. 4B, IFN-induced STAT1, STAT2, and STAT3 activation as determined by tyrosine phosphorylation was indistinguishable between WT and TRAF2-KO MEFs. Moreover, the kinetics of IFN-induced STAT1 activation were similar in WT and TRAF2-KO MEFs (Fig. 4C). Similar kinetics of IFN-induced STAT2 and STAT3 activation were also observed in WT and TRAF2-KO MEFs (data not shown). Taken together, these results illustrate the selective role that TRAF2 plays in IFN signaling: TRAF2 is required for the induction of the alternative NFB pathway by IFN but is not required for either the IFN-induced JAK/STAT pathway or the classical NFB pathway.
Role of TRAF2 in Biological Effects Induced by IFN-We next assessed the role of TRAF2 in the induction of the biological actions of IFN. To determine the role of TRAF2 in the antiproliferative activity of IFN, WT and TRAF2-KO MEFs were cultured for 3 days in the presence of varying concentrations of IFN. As shown in Fig. 5A, the proliferation of WT and TRAF2-KO MEFs was inhibited by IFN treatment to a similar extent, with an ϳ30% inhibition observed at 100 units/ml IFN. We then assessed the sensitivity of WT and TRAF2-KO MEFs to the antiviral action of IFN by determining the ability of IFN to reduce VSV replication. As shown in Fig. 5B, whereas 10 units/ml IFN markedly protected WT MEFs from virus-induced cell death, it was not effective in TRAF2-KO MEFs. VSV induced cell death to a similar extent in untreated WT and TRAF2-KO MEFs. Moreover, the ability of IFN to inhibit virus replication was markedly impaired in TRAF2-KO MEFs at the IFN concentrations examined (Fig. 5C). To substantiate whether TRAF2 plays an important role in the ability of IFN to inhibit virus replication, we performed additional experiments in which the antiviral action of IFN against EMCV was assayed . Role of TRAF2 in the activation of the alternative NFB pathway and the STAT signaling pathway by IFN. A, cell lysates prepared from WT, TRAF2-KO, or TRAF2-KO MEFs reconstituted with TRAF2 and treated with rat IFN␤ (1,000 units/ml) were subjected to immunoblotting for TRAF2 or actin or subjected to p65-GST pulldown assays and immunoblotted for p50 or p52. B, cell lysates prepared from WT, TRAF2-KO, or TRAF2-KO MEFs reconstituted with TRAF2 and treated with rat IFN␤ (1,000 units/ml, 15 min) were subjected to immunoblotting for phosphorylated (p) STAT1, STAT2, or STAT3 and visualized by enhanced chemiluminescence. C, cell lysates prepared from WT or TRAF2-KO MEFs were treated with rat IFN␤ (1,000 units/ml) for varying times, subjected to immunoblotting for phosphorylated STAT1, and visualized by enhanced chemiluminescence.
in WT and TRAF2-KO MEFs. MEFs were treated with various IFN concentrations and infected with EMCV, and virus produced was assayed for the ability to induce cytopathicity in MEFs derived from STAT1-KO mice. STAT1-KO MEFs were employed to eliminate the possible induction of the IFN system during EMCV infection, as they are highly resistant to the antiviral action of IFN (24). As shown in Fig. 5D, IFN inhibited the replication of EMCV to a greater extent in WT MEFs than in TRAF2-KO MEFs. For example, treatment with only 3 units/ml IFN reduced EMCV replication in WT-MEFs to an extent equivalent to IFN treatment with 30 units/ml in TRAF2-KO MEFs, representing a 10-fold difference in IFN sensitivity. It is of interest that the resistance of TRAF2-KO MEFs to the antiviral action of IFN was overcome at high IFN concentrations. These results indicate that TRAF2 plays an important role in the antiviral activity of IFN but is dispensable in the antiproliferative action of IFN.

DISCUSSION
TRAF2 is an important signal transducer for a wide range of TNF receptor superfamily members (TNFR1, TNFR2, p75 NTR , CD40, etc.), as well as lymphocyte costimulatory receptors (LMP1, CD27, 4-1BB, etc.) (25). Type I IFNs bind to a ubiquitously expressed receptor that comprises two subunits, FIGURE 5. Role of TRAF2 in the induction of antiproliferative and antiviral activities by IFN. A, TRAF2-KO and WT MEFs were treated with rat IFN␤ as indicated, and after 3 days, the cell counts were enumerated as described previously (19). B, TRAF2-KO and WT MEFs were treated overnight with rat IFN␤ (10 units/ml) and then infected with VSV (0.1 plaque-forming unit/cell), and at 24 h post-infection, photomicrographs were taken. C, TRAF2-KO and WT MEFs were treated with rat IFN␤ as indicated and infected with VSV, and at 24 post-infection, the viral titer was assayed on indicator Vero cells as described previously (11). D, TRAF2-KO and WT MEFs were treated overnight with rat IFN␤ as indicated and infected with EMCV, and serial dilutions of virus-containing supernatants (24 h post-infection) were assayed for their effect on cell viability of STAT1-KO MEFs by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide uptake. A photograph of a representative assay is shown, which starts with a 1,000-fold dilution of viral supernatants. Three independent experiments gave similar results. IFNAR1 and IFNAR2. Previous studies have demonstrated that IFNAR1 does not directly bind IFN but functions as a speciesspecific signal transducer for type I IFNs (22). Although the role of IFNAR1 in the JAK/STAT pathway in IFN signal transduction has been characterized (7,26,27), its role in integrating IFN-induced NFB activation has not been defined. In this study, we have shown that IFN induces the binding of TRAF2 to the intracellular domain of the IFNAR1 chain by coimmunoprecipitation, transfection of epitope-tagged IFNAR1 constructs, and GST pulldown assays. Our finding that TRAF proteins were not associated with the IFNAR2 ligand-binding subunit of the IFN receptor evidenced the specificity of the interaction of TRAF2 with IFNAR1.
TRAF2 binds to the intracellular domains of TNF receptor superfamily members as well as other proteins to mediate NFB activation. We have demonstrated that IFN induces TRAF2 binding to the amino-terminal half of the intracellular domain of IFNAR1. Interestingly, this is the first interaction of a signaling protein in the IFN system to dock to this portion of IFNAR1. We found previously that the carboxylterminal half of IFNAR1 was involved in STAT3 binding to the IFNAR1 chain (7). IFN activation of the classical NFB pathway involves STAT3, which acts as an adaptor for the PI3K/Akt pathway that results in classical NFB activation (10). It is tempting to speculate that the IFNAR1 chain directs both classical and alternative NFB pathways. The alternative pathway is directed through the amino-terminal half, whereas the carboxyl-terminal half directs the classical pathway. Although IFNAR1 knock-out mice have been derived, the role of IFNAR1 in NFB activation has not yet been examined. An SXXE motif has been identified as a TRAF2-binding consensus sequence in several TNF receptor superfamily members (28). A single conserved SXXE motif (SIDE at amino acids 477-480) is present in the amino-terminal intracellular domain of human IFNAR1 that is present in multiple IFNAR1 homologs (mouse, bovine, ovine, and human chains). In studies with the TNF receptor superfamily, TRAF2 association with receptors is constitutive, but we found that TRAF2 association with IFNAR1 association is dependent on IFN treatment, which is blocked by tyrosine kinase and PI3K inhibitors. Our results are consistent with the hypothesis that IFN-induced serine/threonine phosphorylation of TRAF2 through the PI3K/Akt pathway is required for TRAF2 binding to the IFNAR1 subunit of the IFN receptor. IFN induces the tyrosine and serine phosphorylation of IFNAR1 as well as other signaling intermediates, i.e. STAT proteins, that affect IFN action (7,22,26,29). Moreover, the phosphorylation of TRAF2 at Thr 117 is induced by TNF and is required for TRAF2 function (30). However, preliminary studies using a TRAF2(T117A) mutant do not implicate a role for the phosphorylation of this TRAF2 residue in IFN signaling. Therefore, it will be important in future studies to elucidate the role of IFN-induced phosphorylation of TRAF2 in binding to the IFNAR1 subunit of the IFN receptor.
Many cytokines promote the dissociation of the cytosolic inactive NFB-IB complexes via the serine phosphorylation and degradation of IB, leading to NFB translocation to the nucleus and DNA binding (12), which is denoted as the classical NFB pathway. The degradation of IB proteins requires the activation of the IB kinase (IKK), a multiprotein complex consisting of IKK␣ and IKK␤ catalytic subunits and the IKK␥/ NFB essential modulator regulatory subunit (31). Targeted gene disruption of individual IKK proteins has determined that IKK␤ and IKK␥ (but not IKK␣) are the major mediators of the classical NFB pathway (32)(33)(34)(35)(36). TRAF2-mediated ubiquitinylation of the TAK1 kinase, an upstream regulator of IKKs, is important in activating the classical NFB pathway (37). Recent studies have identified that the LMP1 protein of Epstein-Barr virus, B cell-activating factor, lymphotoxin-␤, and lipopolysaccharide induce NFB activation through an NFB signaling pathway that does not involve IB degradation (38 -46). This alternative pathway of NFB activation involves the linkage of TRAF2 to the activation of the MAP3K-related kinase, NFBinducing kinase, which results in the ubiquitinylation and proteolytic processing of the p100/NFB2 protein and nuclear translocation of p52/Rel dimers to regulate NFB target genes (47). Although TRAF2 plays a critical role in the alternative NFB pathway, both positive and negative regulatory roles of TRAF2 have been described (48,49).
We have shown by several independent means (TRAF2-KO MEFs and expression of human IFNAR1 in heterologous mouse L cells) that TRAF2 is required for IFN-induced activation of the alternative NFB pathway, but not of the classical pathway. These results are consistent with our previous finding that a dominant-negative TRAF2 construct blocked the formation of p52/p65 dimers but had no effect on IFN-induced IB degradation (8). In contrast to the requirement of TRAF2 for the activation of the alternative NFB pathway, we found that TRAF2 does not affect the activation of the JAK/STAT signaling pathway by IFN, as IFN activates STAT1, STAT2, and STAT3 in TRAF2-KO MEFs.
These results also demonstrate the distinct role that TRAF2 plays in the biological activities of IFN. Although the ability of IFN to inhibit the proliferation of WT and TRAF2-KO MEFs was similar, the induction of the antiviral activity of IFN against two different viruses, VSV and EMCV, was markedly reduced in TRAF2-KO MEFs. Moreover, in preliminary studies, we found that TRAF2 also plays a complex role in regulating IFN-induced gene expression, with some genes induced at higher levels in TRAF2-KO MEFs and others unaffected or induced at lower levels (data not shown). We showed previously that NFB plays an important and complex role in regulating IFNinduced gene expression (11,50). Because IFN-induced STAT activation is unaffected in TRAF2-KO MEFs, the differences in gene expression observed probably reflect the binding of different NFB complexes to gene promoters. However, it is presently unknown which genes are regulated by p52-containing NFB complexes.
In conclusion, our studies add a new facet to our understanding of how IFNs generate signals. TRAF2 is bound to the aminoterminal domain of the IFNAR1 chain upon IFN addition. TRAF2 mediates the IFN-induced alternative NFB pathway. Although TRAF2 is dispensable for STAT activation and induction of antiproliferative activity, TRAF2 is important in the induction of the antiviral activity of IFN.