Modulation of the Interferon Antiviral Response by the TBK1/IKKi Adaptor Protein TANK*

Induction of type I interferons can be triggered by viral components through Toll-like receptors or intracellular viral receptors such as retinoic acid-inducible gene I. Here, we demonstrate that the TRAF (tumor necrosis factor receptor-associated factor) family member-associated NF-κB activator (TANK) plays an important role in interferon induction through both retinoic acid-inducible gene I- and Toll-like receptor-dependent pathways. TANK forms complexes with both upstream signal mediators, such as Cardif/MAVS/IPS-1/VISA, TRIF (Toll-interleukin-1 receptor domain-containing adaptor inducing interferon-β), and TRAF3 and downstream mediators TANK-binding kinase 1, inducible IκB kinase, and interferon regulatory factor 3. In addition, it synergizes with these signaling components in interferon induction. Specific knockdown of TANK results in reduced type I interferon production, increased viral titers, and enhanced cell sensitivity to viral infection. Thus, TANK may be a critical adaptor that regulates the assembly of the TANK-binding kinase 1-inducible IκB kinase complex with upstream signaling molecules in multiple antiviral pathways.


Induction of type I interferons can be triggered by viral components through Toll-like receptors or intracellular viral receptors such as retinoic acid-inducible gene I. Here, we demonstrate that the TRAF (tumor necrosis factor receptor-associated factor) family member-associated NF-B activator (TANK) plays an important role in interferon induction through both retinoic acid-inducible gene I-and Toll-like receptor-dependent pathways. TANK forms complexes with both upstream signal mediators, such as Cardif/MAVS/IPS-1/VISA, TRIF (Tollinterleukin-1 receptor domain-containing adaptor inducing interferon-␤), and TRAF3 and downstream mediators TANKbinding kinase 1, inducible IB kinase, and interferon regulatory factor 3. In addition, it synergizes with these signaling components in interferon induction. Specific knockdown of TANK results in reduced type I interferon production, increased viral titers, and enhanced cell sensitivity to viral infection. Thus, TANK may be a critical adaptor that regulates the assembly of the TANK-binding kinase 1-inducible IB kinase complex with upstream signaling molecules in multiple antiviral pathways.
A striking consequence of viral infection as well as Toll-like receptor (TLR) 3 stimulation is the induction of type I IFNs that include IFN␤ and more than a dozen IFN␣ proteins. These cytokines bind to a common receptor, the type I IFN receptor (IFNAR), leading to induction of a large set of genes important in antiviral responses (1)(2)(3). Induction of IFN is triggered by microbial components termed pathogen-associated molecular patterns, such as viral RNA and DNA, which can be viewed as a molecular "signature" of the invading pathogens (4 -7). A major class of cellular proteins capable of recognizing pathogens is the family of membrane receptors termed Toll-like receptors (4 -6). In addition, many types of cells also have TLR-independent, intracellular detection systems to sense viral invasion and initiate innate immune responses (8 -10). Recent studies have identified an RNA helicase-CARD-containing protein RIG-I and its homologue, Helicard/Mda5, as major intracellular receptors for viral double-stranded RNA.
Upon recognition of different subsets of pathogen-associated molecular patterns, individual TLRs trigger distinct innate immune responses via recruitment of different MyD88 adaptor family members, primarily MyD88 and Toll-interleukin-1 receptor domain-containing adaptor inducing IFN-␤ (TRIF) (5,(11)(12)(13)(14). These adaptors function as a platform to organize downstream molecules into signaling complexes, leading to activation of multiple signal cascades, particularly the NF-B and IFN pathways, eventually resulting in specific cellular responses against different types of pathogens. Activation of NF-B pathway requires the IB kinase (IKK) signalosome, a signaling complex composed of two closely related kinase subunits (IKK␣ and IKK␤) and a regulatory subunit NF-B essential modulator (NEMO, also called IKK␥) (15)(16)(17)(18). IFN induction by both TLR and RIG-I pathways requires the TANK-binding kinase 1 (TBK1) or inducible IB kinase (IKKi), two non-canonical members of the IKK family (19 -23) that have been shown to function as the primary kinases capable of phosphorylating interferon regulatory factor 3 and 7 (IRF3/ IRF7). TBK1-and IKKi-mediated phosphorylation and activation of IRF3 and IRF7 are critical steps for IFN production during TLR stimulation or viral infection (19 -21, 24). Among 11 known TLR family members (1,2,6,8,13,19,25), TLR3 and TLR4-dependent IRF3/IRF7 phosphorylation and IFN induction require the adaptor protein TRIF, whereas TLR7/8 and TLR9-mediated activation of IFNs depends on MyD88 (2,6,13,25,26). However, the connection between TRIF or MyD88 and TBK1⅐IKKi in TLR-mediated signal transduction pathways for type I interferon production is not yet fully understood. In RIG-I-mediated intracellular viral detection pathway, the helicase domain of RIG-I can recognize the viral double-stranded RNA. The N-terminal CARD regions can induce NF-B and IRF3/ IRF7 activation (8 -10, 27-30). Recent reporters also demonstrated that Cardif/MAVS/IPS-1/VISA functions as a primary adaptor protein for RIG-I in IFN induction pathways (31)(32)(33)(34), but it is not yet clear how the Cardif/MAVS/IPS-1/VISA induces IFN activation. Recently, we found that TRAF3 plays a critical role in TLR as well as viral-induced IFN activation (35,36). Our studies also show that TAF3 can interact with MyD88, TRIF, and Cardif (35,37).
Because both TBK1 and IKKi are able to interact with an adaptor protein TANK (38,39), it is possible that TANK may participate in IFN induction pathways by organizing TBK1⅐IKKi signaling complexes. TANK was originally identified as a protein associated with TRAF3 (40,41), but its function has remained elusive. No studies have been reported on the function of either TANK itself or the TANK⅐TBK1⅐IKKi complex in IFN induction. Recent findings on the role of TBK1⅐IKKi as well as TRAF3 in IFN induction shed new light on the function of TANK. Results from this study indicate that TANK might be an important modulator of type I IFN induction in both TLR-dependent and intracellular detection mechanisms during viral infections.
Murine macrophage cell line RAW264.7, mouse embryonic fibroblasts (MEFs) and 293T cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Murine bone marrowderived macrophages (BMMs) were generated by culturing the bone marrow cells for 7 days in media containing macrophage colony-stimulating factor.
Reporter Gene Assays-Luciferase assays were preformed as previously described (42,43). Briefly, 293T cells seeded on 12-well plates were transiently transfected with a luciferase reporter plasmid, and various expression plasmids were indicated. In each experiment total amounts of DNA in each sample were kept constant by supplementation with the appropriate empty parental expression vector(s). For normalization of transfection efficiency, each transfection included 50 ng of a pCMV-LacZ control plasmid, which constitutively expresses ␤-galactosidase. 36 h after transfection cells were harvested, and firefly luciferase activities were analyzed using a luciferase assay kit (Promega Corp.) according to the manufacturer's protocol. -Fold induction was calculated for each sample by dividing the luciferase activity, normalized to ␤-galactosidase activity, by that observed in the sample containing only empty parental expression vectors. To measure luciferase activity in RAW cells, RAW cells seeded on 6-well plates were transfected with plasmids indicated, and 24 h after transfection cells were stimulated with LPS or poly(IC) or infected with virus for 24 h, and luciferase activities were measured.
Immunoblotting and Immunoprecipitation-293T cell transfection was performed as described (42,43). For Western analyses, protein samples in modified radioimmune precipitation assay buffer were separated on 10% SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted using standard methods. For immunoprecipitation, total cell lysates were precleared with Sepharose protein G beads for 1 h at 4°C. The cell lysates were immunoprecipitated with 2 g of specific antibody or control IgG for 16 h. The immunoprecipitates were washed 3 times with the lysis buffer and analyzed by Western blot.
RNA Interference-In siRNA experiments, double-stranded RNA duplexes composed of 21-nucleotide oligonucleotides were synthesized by Invitrogen. The siRNA oligonucleotide used for targeting murine TANK was GAC UUU CUG GGA CCU UAA ATT. The siRNA oligonucleotide targeting human TANK was UCA CUU CAA CAG ACU AUU ATT. siRNA duplexes were transfected into cells in 6-well plates using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. GFP siRNA was used as a control. 48 h after transfection cells were used for further experiments. To examine the effect of silencing TANK on IFN reporter gene activity in RAW cells, the oligonucleotides targeting murine TANK sequences were cloned into expression plasmid pBS-hH1-3 as described (44) and transfected into RAW cells.

Vesicular Stomatitis Virus (VSV) Infection and Plaque Assay-For
TANK-induced anti-viral response, 293T cells in 12-well plate were transfected with empty vector or pCMV-FLAG-TANK plasmid. 24 h after transfection cells were infected with VSV at a multiplicity of infection of 0.1 for 24 h. To examine the effect of silencing TANK on VSV infection, MEFs were transfected with siRNA oligo duplexes for TANK or GFP using Lipofectamine 2000 (Invitrogen), and 48 h after siRNA treatment MEFs were infected with VSV at a multiplicity of infection of 0.1. Cell viability was determined by the trypan blue exclusion method. Virus yield was measured in culture supernatants collected from VSV-infected 293T cells or MEFs by standard plaque assay. Briefly, Vero cells were infected with serial dilutions of recovered viruses for 1 h and were overlaid with Dulbecco's modified Eagle's medium containing 0.5% low melting agarose. After 24 h of incubation, plates were stained with crystal violet, and plaques were counted.
IFN Production-To measure IFN production, 293T cells transfected with plasmids or infected with viruses were cultured for 24 h. Culture supernatants were collected, and IFN␤ levels were determined by IFN ELISA kits according to the manufacturer's instructions (PBL Biomedical Laboratories).
Neutralizing Type I IFN Activities-To block IFN activities in culture supernatants, an anti-human IFN␣ antibody and an anti-human IFN␤ antibody (PBL Biomedical Laboratories)

FIGURE 2. Knockdown of endogenous TANK attenuates IFN induction triggered by viral infection.
A, 293T cells in 6 wells were treated with TANK siRNA or GFP siRNA oligos for 48 h, then cells were either transfected with 6 g of poly(IC) or infected with Sendai virus (SeV) for 16 h. Induction of IFN␤ promoter activity was analyzed. B, 293T cells treated with siRNA were transfected with 6 g of poly(IC) or infected with Sendai virus as in A. After 24 h, culture supernatant was collected and analyzed by ELISA for production of IFN␤. Data are represent as the means Ϯ S.D. of duplicate samples. C, efficiency of silencing TANK expression by siRNA. 293T cells or RAW cells were transfected with siRNA oligos targeting human or murine TANK sequence, respectively, and the level of endogenous TANK protein expression after 48 h of siRNA treatment was measured by immunoblot (WB) using anti-TANK antibody. The blots were also re-probed with an anti-actin antibody for loading control. D, TANK is involved in the intracellular detection system against viral infection in fibroblasts. 293T cells treated with TANK siRNA or GFP siRNA duplexes were infected with Sendai virus for the times indicated, and STAT1 phosphorylation was measured by Western blot. E, neutralizing type I IFN by anti-IFNAR antibody impairs TANK-mediated anti-viral response. 293T cells were transfected with siRNA oligos targeting TANK or GFP. 48 h after transfection cells were pretreated with anti-IFNAR antibody (anti-IFNAR2) or control antibody (IgG2a) for 30 min and then infected with Sendai virus for 4 h. Cells were lysed, and STAT1 phosphorylation (pSTAT1) was analyzed by Western blot. F, TANK is required for anti-viral response in macrophages. BMMs in 6-well treated with TANK siRNA oligo duplexes were infected with VSV for the time indicated; cell lysates were analyzed by immunoblot for STAT1 phosphorylation. All data are representative of at least three independent experiments.
were mixed together as anti-IFN␣/␤ antibodies at a final concentration of 20 g/ml. Culture supernatant from 293T cells transfected with TANK was treated with anti-IFN␣/␤ antibodies or nonspecific IgG for 1 h. Then conditioned media were incubated with fresh 293T cells for 1 h. Cells were lysed, and STAT1 phosphorylation was analyzed by Western blot.
To block binding of IFN␣/␤ to the type I IFN receptor (IFNAR), we utilized a mouse monoclonal antibody against human IFN␣/␤ receptor (anti-IFNAR2) (PBL Biomedical Laboratories) or nonspecific isotype control mouse-IgG2a at final concentrations of 25 g/ml. Cells were pretreated with anti-IFNAR2 antibodies or control antibody for 30 min and then infected with Sendai virus for 4 h. Cells were lysed, and STAT1 phosphorylation was analyzed by Western blot.

RESULTS
TANK Is Able to Induce IFN Activation-As an initial step to determine whether TANK is involved in the type I IFN activation pathway, we analyzed the effect of TANK on activation of IFN promoter. As shown in Fig. 1, A-C, overexpression of TANK alone was able to induce considerable IFN transcription in a dosage-dependent manner as measured by a reporter plasmid carrying either an IFN␤ or IFN-stimulated response element promoter, which is regulated by IRFs. In accordance with early publications, TANK itself could slightly induce NF-B transcription. We also confirmed the role of TANK in the induction of type I IFN at protein level. Production of endogenous IFN␤ could be detected in the culture supernatants of 293T cells transfected with TANK in a dosage-dependent manner, although we could not detect IFN with lower concentrations of TANK, probably because the amount of IFN produced might be below the detection limit of the ELISA system. As an alternative way to examine the type I IFNs in the culture supernatants, we investigated whether type I IFNs induced by TANK could lead to the activation of STAT1. We conducted experiments using a mixture of anti-IFN␣/␤ antibodies to neutralize IFN activities in the culture supernatants. As shown in Fig. 1E, the addition of conditional media (CM) collected from TANK-transfected cells led to STAT1 phosphorylation, as opposed to CM from cells transfected with empty vector. Treatment using anti-IFN␣/␤ antibodies but not nonspecific control IgG clearly reduced STAT1 activation, indicating that type I IFNs in the culture supernatants were responsible for this effect.
Because type I IFNs provide immediate innate antiviral defense, we then analyzed the effect of TANK-induced IFN on VSV infection. VSV is a cytopathic virus and is very sensitive to IFNs and has been widely used in IFN antiviral assays (19,33,34,45). 293T cells were transfected with an empty vector or a plasmid encoding TANK. Subsequently, cells were infected with VSV for 24 h, and viral titer was measured by plaque assay. As shown in Fig.  1F, overexpression of TANK reduced virus growth as supernatant from TANK-expressing cells had much lower viral titer. These results indicate that TANK has the ability to furnish antiviral responses through induction of type I IFN.
Involvement of TANK in Multiple Viral Detection Pathways-To more directly investigate the involvement of TANK in virally induced type I IFN production, we examined the effect of silencing of TANK on IFN␤ induction by viral infection. As fibroblast cells, 293T cells lack most of the TLRs; therefore, IFN response is likely mediated through intracellular detection pathways. Knockdown of endogenous TANK protein in 293T cells decreased IFN␤ transcription induced by Sendai virus infection or transfection of poly(IC), which has been shown to mimic viral double-stranded RNA ( Fig. 2A). Consequently, 293T cells treated with TANK siRNA, but not GFP siRNA, produced much less IFN protein in response to poly(IC) transfection and Sendai virus infection (Fig. 2B). The efficiency of siRNA to reduce TANK protein levels in human and murine cells is shown in Fig. 2C. We further elucidated TANK-mediated STAT1 phosphorylation trigged by viral infection. Although viral infection induced strong STAT1 phosphorylation in cells with control siRNA, silencing TANK expression led to impaired STAT1 phosphorylation during viral infection (Fig.  2D). To further confirm the role of type I IFN in TANK-dependent phosphorylation of STAT1, we utilized a neutralizing antibody against IFNAR to block binding of type I IFNs with their receptor. The addition of the neutralizing anti-human IFNAR antibody but not isotype control significantly inhibited viral induced STAT1 phosphorylation (Fig. 2E). These data together demonstrate that type I IFNs induced by TANK was responsible for TANK-mediated STAT1 phosphorylation.
We also monitored viral-induced STAT1 phosphorylation in murine BMMs, which may use TLR-dependent and -independent mechanisms to detect viral infection. As shown in Fig. 2F, knocking down TANK protein severely reduced virally induced STAT1 phosphorylation. Next, we investigated the involvement of TANK in intracellular anti-viral response in MEFs. MEFs treated with siRNA duplexes specific for TANK were infected with VSV. TBK1-deficient MEFs were also included as a positive control. As shown in Fig. 3, almost all of TBK1-deficient cells died post-VSV infection because of defects in IFN induction. However, IFN pretreatment could reduce such VSVinduced cell death. As for TBK1-deficient cells, but with a lesser extent, MEFs treated with siRNA duplexes specific for TANK also exhibited accelerated cell death upon VSV infection (Fig. 3,  A and C). Accordingly, MEFs with knockdown of TANK had much higher viral titer (Fig. 3B). These data point out that TANK may play an important role in innate immunity against viral infection.
Association of TANK with the Adaptor Protein Cardif-Induction of type I interferons in fibroblasts is triggered by intracellular viral receptors such as RIG-I. Recent studies demonstrate that Cardif/MAVS/IPS-1/VISA is a critical adaptor protein for the RIG-I-mediated NF-B and IFN activation. To further elucidate the mechanism of TANK-mediated IFN induction by viral infection, we examined the involvement of TANK in RIG-I pathways. We observed that knockdown of TANK protein inhibited activation of type I IFN promoters by RIG-I (Fig. 4A). We also found that TANK could synergize with Cardif to induce IFN promoter transcription (Fig. 4B), Conversely, knockdown of TANK reduced IFN␤ production induced by Cardif (Fig. 4C). Furthermore, TANK and Mda5/ Helicard, a homologue of RIG-I, together had a strong synergy with TBK1 in IFN induction (data not shown). These data suggest that the TANK may participate in cytoplasmic detection pathways.
The involvement of TANK in the RIG-I pathway promoted us to examine the interaction between TANK and signaling molecules in this pathway. Although we were unable to detect an interaction between TANK and RIG-I in our overexpression system, we did find interaction of TANK with the adaptor protein Cardif/MAVS/IPS-1/VISA. When cell lysates from 293T cells transfected with FLAG-TANK and Myc-Cardif were immunoprecipitated with anti-FLAG and probed with antimyc antibody, Cardif was detected in TANK immunoprecipitates. Similarly, TANK was also found associated with Cardif immunoprecipitates (Fig. 4D). As discussed below, we also have detected endogenous TANK and Cardif interaction in MEFs infected with Sendai virus (see Fig. 7D).
Knockdown of TANK Protein Attenuates Induction of IFN by TLRs-In addition to intracellular viral detection pathways, another major mechanism responsible for IFN activation is through TLRs. We then examined the function of TANK in TLR-induced IFN activation in macrophages. The murine macrophage cell line, RAW264.7, was co-transfected with a TANK construct and either an NF-B or an IFN␤ luciferase reporter plasmid. As expected, stimulation of RAW cells with TLR ligands led to induction of NF-B and IFN activities (Fig. 5A). Interestingly, introduction of TANK into macrophages enhanced IFN␤ transcription in response to TLR3 or TLR4 stimulation, which uses TRIF-dependent pathways. Furthermore, we also noticed that changes occurred to TANK protein upon TLR activation. When BMMs were treated with LPS or poly(IC), slow migrating and diffused high molecular weight bands appeared, indicating possible modification of TANK protein (data not shown). These results implicate the involvement of TANK in TLR signaling.
To more directly evaluate the physiological role of TANK in TLR-dependent IFN production, we used siRNA to knock down endogenous TANK protein in macrophages. RAW cells were treated with TANK siRNA or control GFP siRNA, then stimulated with LPS, CpG, poly(IC), or R848 to activate TLR4, -9, -3, or -7 respectively. Knocking down TANK protein led to a significant reduction in IFN␤ promoter transcription induced by various TLRs (Fig. 5B). Consequently, silencing TANK protein expression also resulted in impaired STAT1 phosphorylation in RAW cells stimulated with LPS (Fig. 5C). Compared with the IFN pathway, TLR-induced extracellular signal-regulated kinase activation was generally not affected.
We further extended these studies to primary macrophages. BMMs were transfected with siRNA to knock down TANK protein expression, and STAT1 phosphorylation was determined after stimulation with LPS or poly(IC). Knockdown of TANK significantly impaired TLR3-and TLR4-induced STAT1 phosphorylation whiling showing little or no effects on either p38 or c-Jun NH 2 -terminal kinase activation (Fig. 5, D  and E, and data not shown). These results imply that TANK might play an important role in the type I IFN induction in macrophages in response to multiple TLR ligands.
TANK Cooperates with TRAF3 and TRIF in IFN Induction-Induction of type I interferons triggered by LPS or poly(IC) is mediated by the adaptor protein TRIF. We, therefore, evaluated the contribution of TANK to the TRIF-dependent IFN pathway. Fig. 6A shows that TANK could enhance TRIF-mediated IFN activation. Conversely, knockdown of TANK diminished TRIF-induced IFN␤ and IFN-stimulated response element promoter activation (Fig. 6B). These results indicate that TANK and TRIF could synergistically stimulate IFN expression in TLR pathways.
Because TANK was originally identified via yeast two-hybrid screening as a TRAF3-interacting protein (42,46) and TRAF3-deficient cells have defective IFN response to TLR stimulation and viral infection (35,36), we hypothesized that TANK in association with TRAF3 may provide a molecular link between TRIF and downstream signaling molecules. Even though either TANK or TRAF3 alone is a weak inducer for IFN␤ production, TANK and TRAF3 exhibited a great synergy in enhancing TRIF-mediated IFN␤ induction (Fig. 6C). The requirement for TANK and TRAF3 in IFN activation and their cooperation with TRIF suggested that these signaling molecules may interact with each other. Therefore, we performed immunoprecipitation studies to evaluate this possibility. Myc-tagged TRIF was co-expressed with TRAF3 and TANK and was immunoprecipitated using anti-Myc antibody. Then immunoprecipitates were blotted with anti-TANK-or anti-TRAF3-specific antibodies. TANK and TRAF3 could be detected in TRIF immunoprecipitates. Interestingly, the amount of TANK associated with TRIF evidently increased when TRAF3 was co-expressed (Fig. 6D). These results suggest these signaling proteins may form a supramolecular complex.
TANK Interacts with TBK1 and IKKi in IFN Induction Pathways-Both TLR and RIG-I pathways utilize TBK1 and/or IKKi kinases for IFN induction. TBK1 was initially identified through its interaction with TANK in yeast two-hybrid screening. Subsequent studies also revealed the interaction between TANK and IKKi (38,39,47). The ability of TANK to interact with TBK1⅐IKKi and the involvement of TANK in TRIF-and Cardif-mediated IFN activation imply that TANK may interact FIGURE 5. Knockdown of endogenous TANK abrogates TLR-mediated IFN induction. A, RAW cells were transfected with an expression plasmid encoding TANK and an IFN␤ luciferase plasmid. 24 h after transfection cells were treated with LPS (100 ng/ml) or poly(IC) (1 g/ml) for 24 h. The luciferase activities were measured and normalized to ␤-galactosidase activity as described under "Experimental Procedures." Data are shown as -fold induction relative to the unstimulated vector control in each experiment and mean values Ϯ S.D. of duplicate samples. B, RAW cells were transfected with plasmids expressing RNA interference specific for murine TANK or GFP and a luciferase reporter plasmid, and 48 h after transfection cells were stimulated with LPS (100 ng/ml), CpG (100 nM), poly(IC) (1 g/ml), or R848 (1 ng/ml) to stimulate TLR4, -9, -3, and -7/8, respectively, and induction of IFN␤ promoter activity was measured by luciferase assay as described under "Experimental Procedures." Data are shown as -fold induction relative to the unstimulated vector control in each experiment and mean values Ϯ S.D. of duplicate samples. C, RAW cells treated with TANK siRNA or GFP siRNA were stimulated with 100 ng/ml LPS for indicated times and analyzed for STAT1 phosphorylation by Western blot. As the control, cell lysates were also blotted for STAT1 protein, extracellular signal-regulated kinase (ERK) phosphorylation, and ERK protein. D, knockdown of TANK in primary macrophages. siRNA oligos targeting TANK or GFP were transfected into BMMs to silence the expression of endogenous TANK protein. 48 h after transfection cells were stimulated with LPS for indicated time. Cell lysates were immunoblotted with antibodies against phospho-STAT1, phospho-p38, p38 protein, and STAT1 protein. E, similar to D except that BMMs were stimulated with poly(IC). Cell lysates were blotted for phosphor-STAT1 and STAT1 protein. All data are representative of three-four independent experiments. and modulate TBK1⅐IKKi activities. To test this, we examined whether TANK cooperates with TBK1 and IKKi in the type I IFN induction pathway. 293T cells were transiently transfected with TANK and TBK1 or IKKi expression plasmids, and induction of IFN␤ was measured by reporter gene assay. TANK was found to be able to increase TBK1-and IKKi-induced IFN␤ promoter activities (Fig. 7, A and B). In agreement with previous overexpression studies, we found TANK could interact with TBK1 and IKKi in the 293T overexpression system (Fig. 7C and data not shown). Our data further demonstrate that TANK could form a signaling complex with TBK1 and IRF3 (Fig. 7C). When FLAG-tagged TANK was immunoprecipitated with anti-FLAG antibody, co-expressed TBK and IRF3 were detected by Western blot analysis. Of note, immunoblotting displayed that endogenous TBK1 protein could be immunoprecipitated with TANK in 293T cells (Fig. 7C, lane 2). Our results imply that TANK could potentially function as an adaptor molecule to organize and collaborate with TBK1 and IKKi in IFN activation.
TANK Is Recruited to Upstream Signaling Complexes in Response to Viral Infection or TLR Stimulation-The results described above promoted us to examine interaction of TANK with these signaling molecules under physiological conditions. First, we analyzed endogenous TANK signaling complexes in viral infected cells. MEFs were infected with Sendai virus, and TANK signaling complexes were immunoprecipitated with an anti-TANK antibody. A number of signaling molecules including Cardif, TRAF3, and TBK1 were found to associate with TANK. Interestingly, increased association between TANK and Cardif was observed in viral-infected MEFs (Fig. 7D). Together, our results demonstrate the formation of Cardif/TRAF3/TANK/TBK1 signaling complex during viral infection.
We then analyzed endogenous TANK signaling complexes in macrophages. The signaling complex was immunoprecipitated from untreated or poly(IC)-stimulated macrophages using anti-TRAF3 antibody. TANK was detected in TRAF3 immunoprecipitates from both stimulated and non-stimulated RAW cells (Fig. 7E). No association was observed using control antibodies. The relative amount of TBK1 in TRAF3 immunoprecipitates increased upon TLR stimulation. We further investigated the recruitment of signaling molecules to TRIF. A number of signaling molecules including TANK, TBK1, and IRF3 were found to associate with the adaptor TRIF, and such association was increased upon poly(IC) stimulation in macrophages (Fig. 7F). These results suggest that TLR stimulation leads to further recruitment of these proteins to form a TRIF signaling complex essential for IFN activation. Collectively, our data imply that TANK could function as an adaptor molecule to link and recruit downstream TBK1⅐IKKi to the upstream signaling complex in multiple IFN induction pathways.

DISCUSSION
Induction of type I IFNs is triggered by microbial components through TLR or a variety of TLR-independent intracellular pattern recognition receptors such as RIG-I. A critical step essential for both recognition systems is TBK1-and IKKi-induced phosphorylation and activation of IRF3 and IRF7. However, the molecular mechanisms responsible for activation of TBK1⅐IKKi kinase activities are not fully understood. In this study we demonstrate that TANK might provide the link between upstream molecules, such as TRIF or Cardif, and the downstream signaling complex TBK⅐IKKi⅐IRF3. The specific role played by TANK in TBK1⅐IKKi activation and IFN induction may resemble the critical role of NEMO in regulating IKK␣⅐IKK␤ and NF-B activation. Thus, two IKK complexes, the classical NEMO⅐IKK␣⅐IKK␤ complex and the non-canonical TANK⅐TBK1⅐IKKi complex, control key transcription factors NF-B and IRFs, respectively, in response to TLR stimulation and viral infection (Fig. 8).
Numerous studies indicated that, depending upon the nature of viruses or the cell types infected by viruses, host cells can recognize viral infection and produce type I interferons either by TLRs in innate immune cells such as macrophages and den- Interestingly, our data also indicate that TANK is involved in IFN␤ induction by TLRs. TRIF has been demonstrated to be the major adaptor molecule for TLR3 or TLR4-mediated type I IFN induction (5,(12)(13)(14). Our work suggests that the TANK⅐ TBK1⅐IKKi complex may be recruited to TRIF via another adaptor molecule, TRAF3. We found TANK could associate with TRIF and TRAF3 and collaborate with these molecules in inducing IFN␤ promoter activation. This finding is contradic- FIGURE 7. TANK provides molecular link between TBK1⅐IKKi and upstream signaling complexes in the IFN pathway. A and B, 293T cells were transfected with TBK1 (50 ng) or IKKi (50 ng) and increasing amounts of TANK. 36 h after transfection the luciferase activities were measured and normalized based upon ␤-galactosidase activity. Data are expressed as the -fold induction calculated for each sample by dividing the luciferase activity by that observed in the sample containing only empty expression vectors. Data represent the means Ϯ S.D. of duplicate samples and are representative of four independent experiments. C, 293T cells were transfected with FLAG-TANK, HA-TBK1, and HA-IRF3 as indicated. 36 h after transfection cell lysates were immunoprecipitated with anti-FLAG antibody. The immunoprecipitates (IP)were probed by Western blot (WB) with anti-TBK1, anti-HA for IRF3, and anti-TANK antibodies. D, MEFs were infected with Sendai virus for 2 and 6 h, and cell lysates were immunoprecipitated with control goat IgG or anti-TANK antibody. The immunoprecipitates were analyzed by Western blot with antibodies specific for Cardif, TANK, TRAF3, and TBK1. E, endogenous interactions of TRAF3, TANK, and TBK1 in macrophages. RAW cells were stimulated with or without 1 g/ml of poly(IC) for 4 h. The cell lysates were immunoprecipitated with anti-TRAF3 antibody or control rabbit IgG. The immunoprecipitates were analyzed by Western blot with antibodies specific for TANK, TBK1. and TRAF3. HC, heavy chain. F, endogenous association of TRAF3, TANK, TBK1, and IRF3 with TRIF. RAW cells were stimulated with poly(IC). The cell lysates were immunoprecipitated with anti-TRIF antibody. The immunoprecipitates were analyzed by Western blot with antibodies specific for TRAF3, TANK, TBK1, IRF3, and TRIF. All data in C-F are representative of three independent experiments. tory to those from Sasai et al. (48), who did not observe the co-immunoprecipitation of TANK and TRIF. Although the discrepancy between these studies is not clear, it might arise from different experimental conditions. Nevertheless, our data underscore the importance of TANK in IFN induction triggered by multiple TLRs. An intriguing question is how TANK regulates TBK1⅐IKKi activities. Our data support a model whereby TANK and TRAFs play a critical role in the IFN induction pathways by promoting the assembly of distinct signaling complexes such as TRIF or Cardif complexes, which may contain different adaptors, kinases, and effectors. The interaction of these molecules promotes the activation of TBK1⅐IKKi, events that are essential for IFN production in response to TLR stimulation or viral infection. Another possible mechanism is through modification and oligomerization of TANK. We found that endogenous TANK protein underwent modification upon TLR stimulation, and TANK protein has been shown to be able to form oligomers. Interestingly, it appears that TANK functions in the type I interferon pathway in a similar way as NEMO/IKK␥ in the NF-B activation pathway. The modification and oligomerization of NEMO have been proposed as one of possible mechanisms for IKK␣ and IKK␤ activation. How-ever, the mechanism of TANK function in TLR stimulation and viral infection still needs further study.
In summary, the results presented in this study implicate TANK as a key adaptor protein that converges signals generated from both membrane receptors, such as TLRs, and cytoplasm sensors, such as RIG-I, into the TBK1⅐IKKi signaling complex, eventually leading to phosphorylation of IRF3/7 and production of type I IFNs. These data suggest that TANK may be an important component in innate immune responses. This also raises the possibility that TLRs and intracellular detection receptors may modulate each other's signaling pathways through the TANK/TBK/IKK complex in response to viral infection.