Identification of TIFA as an Adapter Protein That Links Tumor Necrosis Factor Receptor-associated Factor 6 (TRAF6) to Interleukin-1 (IL-1) Receptor-associated Kinase-1 (IRAK-1) in IL-1 Receptor Signaling*

Tumor necrosis factor receptor-associated factor 6 (TRAF6) transduces signals from members of the Toll/interleukin-1 (IL-1) receptor family by interacting with IL-1 receptor-associated kinase-1 (IRAK-1) after IRAK-1 is released from the receptor-MyD88 complex upon IL-1 stimulation. However, the molecular mechanisms underlying regulation of the IRAK-1/TRAF6 interaction are largely unknown. We have identified TIFA, a TRAF-interacting protein with a forkhead-associated (FHA) domain. The FHA domain is a motif known to bind directly to phosphothreonine and phosphoserine. In transient transfection assays, TIFA activates NFκΒ and c-Jun amino-terminal kinase. However, TIFA carrying a mutation that abolishes TRAF6 binding or mutations in the FHA domain that are known to abolish FHA domain binding to phosphopeptide fails to activate NFκΒ and c-Jun amino-terminal kinase. TIFA, when overexpressed, binds both TRAF6 and IRAK-1 and significantly enhances the IRAK-1/TRAF6 interaction. Furthermore, analysis of endogenous proteins indicates that TIFA associates with TRAF6 constitutively, whereas it associates with IRAK-1 in an IL-1 stimulation-dependent manner in vivo. Thus, TIFA is likely to mediate IRAK-1/TRAF6 interaction upon IL-1 stimulation.

Interleukin-1 (IL-1) 1 is a principal cytokine responsible for the induction of mediators that orchestrate immune and inflammatory responses by activating transcription factors NFB and AP-1 (1). The Toll-like receptors (TLRs) have also recently emerged as critical molecules in the establishment of innate immune and inflammatory responses due to their ability to recognize various pathogen-associated molecules (2). Thus, the Toll/IL-1 receptor family members play a critical role in the switch from innate to adaptive immunity. The signal transduction pathways of the Toll/IL-1 receptor family members consist of common intermediates including MyD88, members of IL-1 receptor-associated kinase (IRAK) family, and tumor necrosis factor receptor-associated factor 6 (TRAF6), whereas TLR3 and TLR4 have MyD88-independent pathways (3). Furthermore, the signaling cascade is evolutionarily conserved with that initiated by the Drosophila Toll receptor (2). Therefore, the MyD88/IRAK/TRAF6 pathway is an essential protein linkage required for the immune and inflammatory systems.
IL-1 signaling is initiated by ligand-induced formation of a receptor complex that consists of the IL-1 receptor and IL-1 receptor accessory protein. Then, the cytosolic protein MyD88 (4,5) and Tollip (6) are recruited to this complex. MyD88 in turn recruits members of the IRAK family including IRAK-1, IRAK-2, IRAK-M (4,7,8), and IRAK-4 (9) via interaction between their death domains. IRAK-1 is activated presumably via phosphorylation by IRAK-4, leaving the receptor complex to interact with TRAF6. Subsequently, TRAF6 activates transforming growth factor ␤-activated kinase 1 (TAK1), which activates a number of downstream signaling cascades, including those of IB kinase, p38, and c-Jun amino-terminal kinase (JNK), leading to the activation of transcription factors such as NFB and AP-1 (10). All members of the IRAK family activate NF〉 when overexpressed in cells. However, only IRAK-4 requires kinase activity to activate NF〉, and IRAK-4 phosphorylates IRAK-1 in vitro (9). Furthermore, introduction of IRAK-1, IRAK-2, or IRAK-M can restore IL-1-induced NF〉 activation in IRAK-1-deficient cells, whereas IRAK-4 cannot compensate for the loss of IRAK-1. Thus, IRAK-4 may act upstream of IRAK-1 and phosphorylate IRAK-1 in a stimulation-dependent manner. TRAF6-deficient mice are defective in IL-1 signaling (11,12). Furthermore, TRAF6 activates TAK1 when TRAF6 was artificially oligomerized without IL-1 stimulation (13,14). Although these recent data indicate that IRAK-1 and TRAF6 play pivotal roles in the Toll/IL-1 receptor signaling, the molecular mechanisms underlying the interaction of TRAF6 with IRAK-1 released from the receptor complex and how the IRAK-1⅐TRAF6 complex activates downstream signals remain to be elucidated. In this study, we identified an adapter protein that is likely to link IRAK-1 to TRAF6 and activates TRAF6 upon IL-1 stimulation.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening and Northern Blotting-Full-length mouse TRAF6 was subcloned into pBD-GAL4-Cam (Stratagene, La Jolla, CA) and used as bait in a two-hybrid screening of an ICR (8-weekold female) bone marrow cDNA library fused to the activation domain of Gal4 in pAD-GAL4 -2.1 (Stratagene). The cDNA encoding human TIFA was identified from a human B-cell cDNA library. Sequence data for mouse, and human TIFA have been submitted to DDBJ/EMBL/Gen-Bank TM databases under accession number AB062111 and AB062110, respectively. A mouse multiple tissue RNA blot (Clontech, Palo Alto, CA) was incubated with 32 P-labeled full-length mouse TIFA cDNA and ␤-actin cDNA at 65°C. The filter was washed with 0.5ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate), 0.2 (w/v)% SDS at 65°C for 30 min.
GST Pull-down Assay, Western Blotting, and Immunoprecipitation-For the GST pull-down assay, HEK293T cells were cotransfected with pME-GST-TIFA and either pME-FLAG-TRAF6, pME-FLAG-TIFA, or pEF-IRAK-1. Thirty-six hours after transfection, cells were lysed in 500 l of TNE buffer (50 mM Tris-HCl, pH 8.0, 1% Nonidet P-40, 1 mM EDTA, 150 mM NaCl, 0.5 mM dithiothreitol) and centrifuged to remove cellular debris. The resulting supernatant was used as a cell lysate. Cell lysates were incubated with glutathione-Sepharose (Amersham Biosciences), and the GST fusion protein complexes were separated on 8.5% polyacrylamide/SDS gels. For analysis of the IRAK-1⅐TIFA⅐TRAF6 complex, HEK293T cells were cotransfected with the indicated combinations of pEF-IRAK-1, pME-Myc-TIFA, and pME-FLAG-TRAF6. Cell lysates were subjected to immunoprecipitation by the addition of anti-FLAG antibody and protein G-Sepharose (Amersham Biosciences). The resulting immunoprecipitates were separated on 10% or 12.5% polyacrylamide/SDS gels. Cell lysates were also used to analyze the expression level of each protein. Establishment of mouse embryonic fibroblasts (MEFs) and TRAF6-deficient MEFs and introduction of FLAG-TIFA into MEFs by retrovirus vector were performed as described (17). Mouse 70Z and MEF cells were either untreated or treated with IL-1 (20 ng/ml) for 5 min. Cell lysates were subjected to immunoprecipitation with 1 g of the appropriate antibody and 20 l of protein G-Sepharose (Amersham Biosciences). Immunoprecipitates or whole-cell lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA). Membranes were immunoblotted with specific antibodies and visualized with appropriate horseradish peroxidase-conjugated second antibody and the ECL Western blotting system (Amersham Biosciences). A mouse multiple-tissue Northern blot (Clontech) was probed with a cDNA fragment of mouse TIFA (upper) or ␤-actin (lower). C, structure of TIFA and its mutants and alignment of TRAF6 binding sites. D, identification of the TRAF6 binding site in TIFA. HEK293T cells were cotransfected with pME-GST-TIFA or pME-GST-TIFA mutant and pME-FLAG-TRAF6. Thirty-six hours after transfection, cell lysates were subjected to GST pull-down assay. WB, Western blot.

Separation of Recombinant TIFA Protein by Gel Filtration Column
Chromatography-MBP-tagged TIFA protein was expressed in Escherichia coli BL21 and affinity purified with amylose resin (New England BioLabs, Beverly, MA). MBP-TIFA (24 g) was cleaved with 2.4 units of PreScission Protease (Amersham Biosciences) at 4°C for 4 h. Fifty microliters of the reaction mixture was applied to a Superdex 75 column (0.32 ϫ 30 cm), and fractions (20 l) were collected. The fractions were analyzed on 10% polyacrylamide-SDS gel, and proteins were visualized by Coomassie Brilliant Blue Staining. Molecular weight marker proteins were analyzed under the column conditions described above.
Luciferase Reporter Assay and in Vitro Kinase Assay-HEK293T cells were transfected with 1 ng of 3xB-luc or 3xMB-luc (mutant B sites), 10 ng of ␤-actin-␤-galactosidase and the indicated amounts of various expression plasmids. Thirty-six hours after transfection, luciferase activity was measured with the PicaGene luciferase assay system (TOYO INK, Tokyo), and ␤-galactosidase activity was used to standardize transfection efficiency. For the in vitro JNK kinase assay, HEK293T cells were transfected with 10 ng of pEF-T7-JNK and the indicated amounts of pME-FLAG-TIFA. Thirty-six hours after transfection, His-JNK was immunoprecipitated with anti-T7 antibody. Immunoprecipitates were incubated with 2 g of GST-c-JUN (1-89) fusion protein (New England BioLabs) in 20 l of kinase buffer (20 mM HEPES, pH 7.5, 20 mM MgCl 2 , 5 Ci of [␥-32 P]ATP (3000 Ci/mmol)) at 30°C for 30 min.

Identification of TIFA as a TRAF6-interacting Protein-To
identify the protein(s) that regulate association of IRAK-1 with TRAF6, we performed yeast two-hybrid screening of a mouse bone marrow cDNA library using murine TRAF6 as bait. We identified a cDNA encoding a 184-amino acid protein, identical to T2BP previously identified as a TRAF2-binding protein, that activates NFB and JNK in transient transfection assays (Ref. 18 and data not shown). Because the present study showed that T2BP associates with TRAF6 and a forkhead-associated (FHA) domain of the T2BP (Fig. 1A) is likely to play a critical role in the signaling (see "Discussion"), we designated T2BP as a TRAF-interacting protein with an FHA domain, TIFA. TIFA mRNA was expressed highly in spleen and moderately in other tissues, whereas it was not detectable in skeletal muscle (Fig.  1B). FHA domains are conserved sequences of 60 -100 amino acids found mainly in eukaryotic nuclear proteins (19) that participate in establishing or maintaining cell-cycle checkpoints, DNA repair, or transcriptional regulation (20). More importantly, some of them were shown to bind directly to phosphoserine/phosphothreonine (Ser(P)/Thr(P)) residues (21)(22)(23)(24) in the same way that SH2 domains interact with phosphotyrosine residues (20,25).
To determine whether interaction of TIFA with TRAF6 is required for TIFA-induced NF〉 and JNK activation, we identified the TRAF6 binding site in TIFA. Wild-type TIFA and various mutants were expressed as GST fusion proteins in HEK293T cells (Fig. 1C) in conjunction with FLAG-TRAF6. Subsequent GST pull-down assays revealed that the carboxylterminal portion of TIFA (C domain) is sufficient for TRAF6 binding and that the amino-terminal portion (N domain) and the FHA domain are not necessary (Fig. 1D). Recent analysis of the crystal structure of the TRAF-C domain of TRAF6 in complex with TRAF6-binding peptides from CD40 and RANK revealed that eight amino acid residues of the TRAF6-binding peptide contact the TRAF-C domain of TRAF6 and identified XXPXEXX-(aromatic/acidic) as a consensus sequence for TRAF6 binding (26) (Fig. 1C). Substitution of Ala for Glu at the fifth amino acid position of the TRAF6 binding site in CD40 abolished binding to TRAF6 (27,28). A consensus TRAF6 binding motif is present in the C domain of both human and mouse TIFA (Fig. 1C). Substitution of Ala for Glu-178 (E178A) abolished binding of TIFA to TRAF6 (Fig. 1D), which is consistent with the notion that the TRAF domain of TRAF6 is responsible for binding to both CD40 (15) and TIFA (data not shown). We then analyzed the ability of various TIFA mutants to activate NFB and JNK. Both the FHA and the C domain were required, whereas the N domain was not necessary for NFB activation ( Fig. 2A). GST-E178A did not activate NFB or JNK (Fig. 2, A and B), indicating that interaction of TIFA with TRAF6 is essential for TIFA-mediated NF〉 and JNK activation. To clarify the role of the FHA domain of TIFA in NF〉 and JNK activation, Gly-50 and Ser-66 were replaced with Glu and Ala, respectively (G50ES66A, Fig. 1, A and C), since identical mutations in the FHA domain of the Arabidopsis kinaseassociated protein phosphatase abolished its binding to the phosphorylated receptor-like protein kinase (22). Interestingly, GST-G50ES66A, which binds to TRAF6 (Fig. 1D), did not activate NFB or JNK (Fig. 2, A and B), indicating that interaction of TIFA with TRAF6 is necessary but not sufficient for activation. Thus, the FHA domain of TIFA plays a critical role in the NFB and JNK activation.
Possible Implication of TIFA Self-association in TRAF6 Activation-TRAF6 transduces signals when oligomerized (13,14). Because a part of the expressed TIFA is phosphorylated (data not shown), homo-oligomerization of TIFA via the FHA domain/phosphopeptide interaction may facilitate or induce oligomerization of TRAF6. To address whether TIFA forms a homo-oligomer, the binding of FLAG-TIFA to various GST-TIFA mutants was analyzed (Fig. 3A). GST-TIFA, but not GST, associated with FLAG-TIFA, indicating that TIFA forms a homo-oligomer. Both the FHA domain and the C domain were required, whereas the N domain was not necessary for the TIFA self-association. The structural requirement for the selfassociation is similar to that for NFB and JNK activation, suggesting that the self-association of TIFA may be a prerequisite for activation of NFB and JNK. However, GST-G50ES66A associated with FLAG-G50ES66A, suggesting that the FHA domain/phosphopeptide interaction is not likely to be involved in the self-association. More importantly, because the G50ES66A mutant does not activate downstream signals, homo-oligomer of wild-type TIFA and that of the G50ES66A mutant could have different physical properties. To address this question, recombinant TIFA was generated as an MBPtagged protein with a cleavage site for human rhinovirus type-14 3C protease (29) at the junction (MBP-TIFA). Purified MBP-TIFA was digested almost completely by a fusion protein of GST with the 3C protease (GST-3C), and the reaction mixture was analyzed by Superdex 75 gel filtration chromatography (Fig. 3B, upper). According to the elution profile of the marker proteins, MBP (43 kDa) was present as a monomer, and GST-3C (46 kDa) formed a dimer. TIFA (24 kDa) was eluted at around fraction 14, which corresponds to TIFA trimer. However, when recombinant TIFA-G50ES66A protein was subjected to a similar experiment, the mutant protein was eluted at around fraction 7, which corresponds to TIFA pentamer or hexamer (Fig 3B, lower). These results indicate that introduction of the G50ES66A mutation result in loss of the ability to form a trimer, presumably by altering the conformation of the FHA and the C domains, which are responsible for the selfassociation. Taken together, the self-association of TIFA may be required for activation of downstream signals, but the size or the conformation of homo-oligomer could be important.
TIFA Enhances the Association of TRAF6 and IRAK-1-We next addressed whether TIFA interacts with IRAK-1. Coexpression of GST-TIFA with IRAK-1 followed by GST pull-down assay revealed that TIFA associates with IRAK-1 (Fig. 4A). Both GST-G50ES66A and GST-E178A bind to IRAK-1 as efficiently as wild-type TIFA, indicating that the IRAK-1/TIFA interaction is not mediated by recognition of the phosphopeptide by the FHA domain. Because TIFA binds both TRAF6 and IRAK-1, we hypothesized that TIFA may affect the IRAK-1/ TRAF6 interaction. To test this, the interaction between TRAF6 and IRAK-1 was analyzed in both the absence and presence of increasing expression of TIFA. Transfection experiments were carried out using 0.2 g of TRAF6 and IRAK-1 expression vectors to minimize the IRAK-1/TRAF6 interaction in the absence of TIFA expression. The amount of IRAK-1 coprecipitated with TRAF6 was increased dramatically at the optimal expression of TIFA (Fig. 4B), indicating that TIFA, when overexpressed, can link TRAF6 to IRAK-1.
Interaction of TIFA with TRAF6 and IRAK-1 in Vivo-For investigation of the physiological significance of the IRAK-1⅐TIFA⅐TRAF6 complex, the interactions of endogenous proteins were analyzed. For identification of endogenous TIFA protein, rabbit anti-TIFA polyclonal antibody was generated. When lysates prepared from 70Z mouse pre-B cells were subjected to immunoprecipitation followed by Western blotting with anti-TIFA antibody, two bands at ϳ24 kDa that comigrated with protein expressed from TIFA cDNA in HEK293T cells were identified (Fig. 5A, left four lanes). These two bands were not observed when control IgG was used for immunoprecipitation, indicating that they represent endogenous TIFA. We FIG. 4. TIFA enhances the association of TRAF6 and IRAK-1. A, interaction of TIFA with IRAK-1. HEK293T cells were cotransfected with pME-GST-TIFA, pME-GST-TIFA mutant, or pME-GST and pEF-IRAK-1. Thirty-six hours after transfection, cell lysates were subjected to GST pull-down assay. WB, Western blot. B, TIFA links TRAF6 to IRAK-1. HEK293T cells were cotransfected with 0.2 g of pME-FLAG-TRAF6 and/or pEF-IRAK-1 and increasing amounts of pME-Myc-TIFA. FLAG-TRAF6 was immunoprecipitated with anti-FLAG antibody. next addressed whether TIFA interacts with TRAF6. Endogenous TRAF6 was immunoprecipitated from either unstimulated 70Z cells or from cells stimulated with IL-1 for 5 min, and coprecipitated TIFA was analyzed by Western blotting with anti-TIFA antibody. Similar amounts of TIFA were detected irrespective of IL-1 stimulation (Fig. 5A, right two lanes), indicating that TIFA interacts with TRAF6 constitutively in the 70Z pre-B cell line. To determine whether TIFA interacts with TRAF6 in a different cell type, wild-type MEF and TRAF6deficient MEF (17) were analyzed. Because the reactivity of anti-TIFA antibody was relatively weak, wild-type MEF and TRAF6 Ϫ/Ϫ MEF expressing FLAG-TIFA were generated by retrovirus vector-mediated gene transfer. To avoid overexpression of FLAG-TIFA, infection with retrovirus carrying FLAG-TIFA and a puromycin resistance gene was performed with a low titer of virus followed by puromycin selection. The amount of FLAG-TIFA expressed in wild-type MEF and TRAF6 Ϫ/Ϫ MEF was ϳ3 times that of endogenous TIFA protein in MEF and, because of this low expression of FLAG-TIFA, the basal activity of NF〉 and JNK in MEF was not affected (data not shown). Thus, we may characterize the FLAG-TIFA as endogenous TIFA. When lysates from wild-type or TRAF6 Ϫ/Ϫ MEF were subjected to immunoprecipitation with anti-TRAF6 antibody followed by Western blotting with anti-FLAG antibody, TIFA was precipitated only in the presence of TRAF6 (Fig. 5B). Therefore, anti-TRAF6 antibody did not immunoprecipitate TIFA directly due to its cross-reactivity. From these data, we conclude that TIFA binds TRAF6 constitutively both in 70Z pre-B cells and in MEF in vivo.
We next investigated the interaction of TIFA and IRAK-1 in vivo. When FLAG-TIFA was immunoprecipitated with anti-FLAG antibody from wild-type MEF in the absence or presence of IL-1 stimulation for 5 min followed by Western blotting with anti-IRAK-1 antibody, phosphorylated IRAK-1 was coprecipitated only in the presence of IL-1 stimulation (Fig. 5C). Because activated phospho-IRAK-1 is reported to be released from the receptor-MyD88 complex (5), it is possible that IRAK-1 associated with TIFA after release from the receptor complex, consistent with the previous finding that only activated phospho-IRAK-1 is coprecipitated with TRAF6 upon IL-1 stimulation (30). IRAK-1 was not precipitated from MEFs infected with control virus. These results clearly indicate that TIFA interacts with IRAK-1 in an IL-1 signal-dependent manner in vivo and suggest that TIFA links TRAF6 to IRAK-1 upon IL-1 stimulation.
TIFA Mediates IL-1-induced NFB Activation via IRAK-1 and TRAF6 -To further investigate the notion that TIFA functions by interacting with IRAK-1 and TRAF6, the effect of a dominant-negative mutant (DN) of TIFA and DN of previously identified proteins involved in IL-1 signaling on NFB-mediated transcription was analyzed. GST-G50ES66A suppressed IL-1-and TLR4-induced NFB activation but not TNF␣-induced activation (Fig. 6A), suggesting that TIFA functions in TRAF6-mediated signaling. MyD88-and IRAK-1-induced NFB activation was also suppressed by GST-G50ES66A, whereas TAK1/TAB1-induced activation (10) was not affected (Fig. 6B, left, luciferase activity without GST-G50ES66A was set to 100). Unexpectedly, TRAF6-induced NFB activation was further augmented by GST-G50ES66A (Fig. 6B, right, luciferase activity with reporter alone was set to 1), which binds to TRAF6 but does not activate NF〉 without expressing FIG. 5. Interaction of TIFA with TRAF6 and IRAK-1 in vivo. A, interaction of TRAF6 with TIFA in 70Z pre-B cells. Lysates prepared from 70Z cells with (ϩ) or without (Ϫ) IL-1 stimulation (20 ng/ml, 5 min) were immunoprecipitated (IP) with either anti-TIFA, control rabbit IgG, or anti-TRAF6 antibody. Immunoprecipitates were analyzed by Western blotting (WB) with anti-TIFA, anti-IRAK-1, and anti-TRAF6 antibodies. Lysates prepared from HEK293T cells transfected with pME-TIFA or control vector were analyzed on identical gels. Open and solid arrowheads indicate the Ig light chain and nonspecific bands, respectively. B, interaction of TRAF6 with TIFA in MEF cells. Both wild-type MEF (ϩ/ϩ) and TRAF6-deficient MEF (Ϫ/Ϫ) were infected with retrovirus expressing FLAG-TIFA. Anti-TRAF6 antibody immunoprecipitates were analyzed by Western blotting with anti-FLAG antibody. Open and solid arrowheads indicate the Ig heavy chain and Ig light chain, respectively. Lysates were also subjected to immunoblotting directly to analyze FLAG-TIFA expression. C, IL-1 stimulation-dependent interaction of TIFA and IRAK-1 in MEF cells. Wild-type MEFs were infected with retrovirus expressing FLAG-TIFA or control virus.
Both types of MEFs were unstimulated (Ϫ) or stimulated (ϩ) with IL-1 (20 ng/ml, 5 min). Anti-FLAG antibody immunoprecipitates were analyzed by Western blotting with anti-IRAK-1 antibody. The solid arrowhead indicates Ig light chain. Lysates were also subjected to immunoblotting directly to analyze IRAK-1 expression. exogenous TRAF6. Transfection of a suboptimal dose of TRAF6 leads to the activation of NF〉 to some extent, suggesting that exogenously expressed TRAF6 can be partially activated presumably by its conformational change or by its oligomerization under this transient transfection condition. GST-G50ES66A may be able to activate partially activated exogenous TRAF6.
TIFA-induced NFB activation was blocked by TRAF6-DN and TAK1-DN (Fig. 6C). An inactive form of Ubc13 (Ubc13(C87A)) also suppressed TIFA-induced activation (Fig.  6C), suggesting that the TIFA signal leads to Lys-63-linked polyubiquitination of TRAF6, which is required for TAK1-mediated activation of IB kinase and MKK6 (13,31). TIFAinduced activation was not affected by MyD88-DN, but it was blocked by IRAK-1-DN (Fig. 6C). Because TIFA binds IRAK-1 when overexpressed (Fig. 4A), it is possible that IRAK-1-DN inhibited TIFA-mediated NF〉 activation by interfering with the activation process of TIFA even if IRAK-1 functions upstream of TIFA in vivo. Taken together, these findings support the notion that TIFA links TRAF6 to IRAK-1. DISCUSSION We report the identification of TIFA as a TRAF6-binding protein that can mediate IL-1 signaling. In transient transfection assays, TIFA binds both TRAF6 and IRAK-1. Furthermore, IRAK-1 is efficiently coprecipitated with TRAF6 under conditions of optimal TIFA expression. Thus, TIFA, when overexpressed, can link TRAF6 to IRAK-1 in the absence of IL-1 stimulation. By analyzing the interaction of endogenous TIFA with TRAF6 in two different types of cells, we showed that TIFA can bind TRAF6 irrespective of IL-1 stimulation. In contrast, interaction of TIFA and IRAK-1 is dependent on IL-1 stimulation in vivo. These results strongly suggest that TIFA links TRAF6 to IRAK-1 in an IL-1 stimulation-dependent manner in vivo. TIFA may play a similar role in the signaling from members of the TLR family, because the TLR signal transduction pathways are also mediated by IRAK-1 and TRAF6 (3,32). In fact, TLR4-mediated NFB activation was inhibited by expression of the DN of TIFA. Thus, TIFA may function in innate immunity. Although T2BP, identical to TIFA, was shown to associate with TRAF2, when they overexpressed, the T2BP/ TRAF2 interaction in vivo has never been demonstrated (18). Nevertheless, TIFA may regulate both TRAF2 and TRAF6 in vivo.
The association of signal transducers including kinases and adapter proteins is largely regulated through protein phosphorylation, allowing dissociation as the balance shifts from kinase to phosphatase activity. Phosphorylation of proteins on serine and threonine residues has traditionally been recognized as a way to regulate enzyme conformation. Recent studies have revealed that serine/threonine phosphorylation can play critical roles in the formation of multimolecular signaling complexes through specific interaction of phosphorylated peptides and Ser(P)/Thr(P) binding modules including FHA domains, 14-3-3, WW domains, and WD40 and leucine-rich repeat domains (25). Among the identified FHA domain-containing proteins, a specific target protein has been determined for some. In each case, phosphorylation-dependent interaction of the FHA domain with the target protein is thought to be responsible for the physiologically significant regulation of cell function (20). The FHA2 domain of Rad53p, a protein kinase involved in the DNA damage response and in cell cycle arrest in Saccharomyces cerevisiae, binds Rad9p when Rad9p is phosphorylated by DNA damage signals. Mutation of the FHA2 domain of Rad53p abolishes DNA damage-induced G 2 /M cell cycle arrest, indicating the biological relevance of the Rad53p-Rad9p interaction (21). In addition, the FHA domain of Arabidopsis kinase-associated protein phosphatase interacts with a plasma membraneintegrated receptor, leucine-rich repeat receptor-like protein kinase, when receptor-like protein kinase is autophosphorylated upon ligand binding (22). Receptor-like protein kinase is a product of the CLAVATA1 (CLV1) gene, mutations that result in plants with enlarged shoots and floral meristems (33). These observations and the fact that the G50ES66A mutant of TIFA is not able to activate NF〉 or JNK suggest that a FIG. 6. TIFA activates NFB via TRAF6 and IRAK-1. A, HEK293T cells or HepG2 cells were transfected with 3xB-luc and increasing amounts (0, 3, or 10 g) of pME-GST-G50ES66A. Thirty hours after transfection, cells were either stimulated with TNF␣ (10 ng/ml, HEK293T) or IL-1 (20 ng/ml, HepG2) for 6 h. HEK293T cells were transfected with pEF-TLR4 and pEF-MD-2 together with 3xBluc and increasing amounts of pME-GST-G50ES66A. B, HEK293T cells were transfected with an expression plasmid encoding one of the NFB activators including MyD88, IRAK-1, TAK1/TAB1, or TRAF6 in the presence of increasing amounts (0, 3, or 10 g) of pME-GST-G50ES66A. C, HEK293T cells were transfected with 0.1 g of pME-TIFA in the absence or presence of increasing amounts (3 or 10 g) of expression vector encoding MyD88-(152-296), IRAK-1-(1-211), TRAF6-(275-530), TAK1(K63W), or Ubc13(C87A). Thirty-six hours after transfection, luciferase activity was measured. Relative values in which the fold activation in the absence of each dominant-negative mutant was set to 100 (A, B except TRAF6 columns, C) or in which the luciferase activity with reporter alone was set to 1 (TRAF6 columns in B) are shown. Results shown are the mean Ϯ S.D. of triplicate experiments and are representative of two independent experiments. specific target protein for the FHA domain of TIFA may exist and play a role in the regulation of TIFA-mediated signaling. In this study, we identified three TIFA-binding proteins including TRAF6, IRAK-1, and TIFA itself. However, they were able to bind the G50ES66A TIFA mutant as well as wild-type TIFA, which suggests that none of them is a candidate protein that binds to the FHA domain in a phosphorylation-dependent manner. Because TRAF6 transduces signals when oligomerized (13,14) and recombinant TIFA can form trimers, one may postulate that TIFA is involved in the oligomerization of TRAF6 in concert with IRAK-1 to activate NF〉 and JNK. Interestingly, recombinant TIFA G50ES66A mutant formed a pentamer or hexamer, not a trimer. Moreover, the G50ES66A mutant is able to activate NFB in the presence of the suboptimal dose of TRAF6 but not in concert with endogenous TRAF6 in the absence of exogenous TRAF6. Given that transient transfection of the suboptimal dose of TRAF6 results in the partial activation of TRAF6, the intact structure of the FHA domain of TIFA may be required for the initial step of the TRAF6 activation that may require specific oligomer formation of TRAF6. Further studies are required to clarify exact roles of the FHA domain of TIFA; that is, binding to the putative phosphoprotein, holding the TIFA protein in an active conformation, or some other functions.
Whether TIFA is essential for the IRAK-1/TRAF6 interaction or whether it stabilizes the interaction is not clear. One interesting possibility is that TIFA may augment the signal strength. IRAK-1 contains three consensus TRAF6 binding sites (26), and TIFA contains a single consensus TRAF6 binding site. Because TIFA and IRAK-1 interacts upon IL-1 stimulation, the number of TRAF6 molecules involved in signaling in cells abundantly expressing TIFA could be higher than that in cells with less TIFA expression. Furthermore, IRAK-1/TIFA interaction may induce their structural changes and increase their affinity for TRAF6. It is possible that the FHA domain of TIFA may bind to the substrate of IRAK-1 or IRAK-4, which leads to the stabilization of IRAK-1⅐TIFA⅐TRAF6 complex. It has been reported that kinase activity of IRAK-1 is not required for IL-1 signaling in a human embryonic kidney 293 cell line (34). Perhaps it can affect signal strength in other specific cell types. In this sense, it is interesting that TIFA is highly expressed in spleen but scarcely expressed in skeletal muscle. During the preparation of our manuscript, Pellino 1 was reported to be associated with IRAK-1⅐IRAK-4⅐TRAF6 complex in an IL-1 signal-dependent manner (35). Thus, Pellino 1 and TIFA may have similar roles. However, they have distinct tissue specificities; Pellino 1 is weakly expressed in spleen but significantly expressed in skeletal muscle.
TRAF6 is also involved in mediating signals from members of the TNF receptor superfamily, including CD40, RANK, Xlinked ectodysplatin-A2 receptor (XEDAR), and p75 neutrophin nerve growth factor receptor (11,12,36,37). Although a recent study showed that the p75 nerve growth factor receptor recruits IRAK-1 upon ligand stimulation (38), other receptors are thought to form trimers and do not require IRAK-1 (39). In our preliminary experiments, TIFA did not bind the cytoplasmic tail of CD40 or RANK, suggesting that TIFA may not be required for oligomerization of TRAF6 to transduce signals from members of the TNF receptor superfamily with the exception of the p75 nerve growth factor receptor. TIFA may regulate signals that are mediated by both TRAF6 and IRAK-1. To elucidate the molecular mechanism underlying TIFA signaling, especially to identify a role of the FHA domain, crystallographic studies of TIFA and identification of a target protein of the FHA domain are required. Further studies of TIFA may reveal a novel regulation of inflammation and innate immunity.