Association of the Adaptor TANK with the IκB Kinase (IKK) Regulator NEMO Connects IKK Complexes with IKKε and TBK1 Kinases*

Canonical activation of NF-κB is mediated via phosphorylation of the inhibitory IκB proteins by the IκB kinase complex (IKK). IKK is composed of a heterodimer of the catalytic IKKα and IKKβ subunits and a presumed regulatory protein termed NEMO (NF-κB essential modulator) or IKKγ. NEMO/IKKγ is indispensable for activation of the IKKs in response to many signals, but its mechanism of action remains unclear. Here we identify TANK (TRAF family member-associated NF-κB activator) as a NEMO/IKKγ-interacting protein via yeast two-hybrid analyses. This interaction is confirmed in mammalian cells, and the domains required are mapped. TANK was previously shown to assist NF-κB activation in a complex with TANK-binding kinase 1 (TBK1) or IKKε, two kinases distantly related to IKKα/β, but the underlying mechanisms remained unknown. Here we show that TBK1 and IKKε synergize with TANK to promote interaction with the IKKs. The TANK binding domain within NEMO/IKKγ is required for proper functioning of this IKK subunit. These results indicate that TANK can synergize with IKKε or TBK1 to link them to IKK complexes, where the two kinases may modulate aspects of NF-κB activation.

dation of the inhibitors liberates the previously bound NF-B proteins to localize to the nucleus and bind to so-called B DNA binding elements located within many promoters/enhancers. The IB inhibitors are phosphorylated on specific serine residues (5,6) by kinases residing in a large complex referred to as the IB kinase complex (IKK). 1 IKKs are composed of two catalytic subunits, IKK␣ and IKK␤ (7)(8)(9)(10)(11), as well as a regulatory protein, named NEMO (NF-B essential modulator)/ IKK␥/FIP-3 (12)(13)(14). More recently, it has been demonstrated that the IKK kinases target not only the so-called small IB inhibitors, of which the IB␣ is the prototype, but that they also similarly phosphorylate and regulate the p105/NF-B1 and p100/NF-B2 precursors, leading either to their proteolytic degradation or to their processing to p50 and p52, respectively (15)(16)(17)(18). In addition to these functions, IKK kinase activity may also modulate the transactivation potential of the NF-B proteins liberated by the degradation of the inhibitors; activated IKK kinases have been shown to phosphorylate a transactivation domain of RelA, thereby promoting its ability to transcriptionally transactivate genes (19).
NEMO/IKK␥ is an essential component of the IKK complex, as evidenced for example by the inability of many signals, including TNF and interleukin-1, to induce NF-B activity in NEMO/IKK␥-deficient cells (13,20,21). It has been suggested that NEMO/IKK␥ may be required for the correct assembly of the IKK complex and/or for the recruitment of upstream activators of the IKK complex (12,13). However, the functions and mechanisms of NEMO/IKK␥ remain to be determined. If this essential component does indeed connect to a variety of different upstream signaling mediators, these would be important to identify, since they may be signal-specific mediators of NF-B activation and thus more specific potential targets for therapies intended to delimit NF-B activation.
We have used NEMO/IKK␥ as bait in a yeast two-hybrid screening to identify potential mediators of select upstream signaling pathways. Previously, we reported on the identification of one NEMO/IKK␥-interacting protein identified in this way and termed CIKS (connection to IKK and SAPK/JNK) (22) (also known as Act-1 (23)). Here we describe the identification of an additional NEMO/IKK␥-interacting protein, termed TANK (TRAF family member-associated NF-B activator). We show that TANK interacts with NEMO/IKK␥ (and the IKKs) in mammalian cells. TANK had previously been shown to be potentially involved in both positive and negative regulation of NF-B activity (24 -26). Positive regulation reportedly occurs via an association of TANK with two kinases, termed inducible IB kinase (also known as IKK⑀) and TBK1 (also known as T2K and NF-B-activating kinase) (27)(28)(29), although the mechanisms involved remain unknown. We demonstrate here that TANK synergizes with IKK⑀ and TBK1 to form a complex with NEMO/IKK␥ and thus with the IKKs. This links IKK⑀ and TBK1 with at least a subset of IKK complexes and suggests potentially direct effects on IKK-associated functions. We also provide evidence that the TANK-binding domain of NEMO may be important in transmitting signals.

EXPERIMENTAL PROCEDURES
Cell Culture and Biological Reagents-Human embryonic kidney 293 and HeLa cells were maintained as described (22,30). NEMO-deficient Jurkat cells were a generous gift from Dr. Shao-Cong Sun, (Pennsylvania State University College of Medicine) and were maintained in RPMI supplemented with 10% fetal bovine serum and 10% penicillin/ streptomycin.
Polyclonal anti-TANK rabbit antibodies were raised against the first 20 and the last 19 amino acids of human TANK. Anti-NEMO/IKK␥ and anti-Myc antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), as were anti-HA and anti-IKK␣ beads. Anti-FLAG beads were purchased from Sigma. Monoclonal anti-IKK⑀ and anti-IKK␣ antibodies were from Imgenex (San Diego, CA) and BD PharMingen (San Diego, CA), respectively. Mouse NEMO/IKK␥ and human TANK were both cloned by PCR from a mouse or human liver cDNA library, respectively (CLONTECH, Palo Alto, CA). Truncation mutants of NEMO/IKK␥ and TANK were generated by PCR. The FLAG-NEMO ⌬TANK construct was made by first cloning a PCRgenerated fragment encompassing the region from amino acid 250 to the stop codon into pcDNA3.1 FLAG; subsequently, a PCR-generated fragment harboring amino acids 2-200 was inserted in frame. The HA-NEMO⌬TANK construct was made by subcloning the NEMO⌬TANK coding sequence into pcDNA3.1 HA. The FLAG-TANK ⌬IKK⑀ construct was made by cloning a PCR-generated fragment encoding amino acids 2-110 into pcDNA3.1 FLAG ⌬N169-TANK. Fulllength CIKS, NEMO/IKK␥, IKK␣, and IKK␤ have been previously described (22,30). TBK1 and IKK⑀ were PCR-amplified from a liver cDNA library. cDNAs encoding TANK and IKK⑀ were cloned into pcDNA3.1 FLAG (Invitrogen) for expression in mammalian cells, and TBK1 and IKK⑀ were cloned into pcDNA3.1 Myc. Both TBK1 and IKK⑀ mutant (K38A) constructs were generated by site-directed mutagenesis (Stratagene, La Jolla, CA). A construct encoding full-length NEMO/ IKK␥ fused to GST was generated by subcloning full-length NEMO/ IKK␥ into pGEX-2T (Amersham Biosciences).
Yeast Two-hybrid Analysis-DNA encoding NEMO/IKK␥ (amino acids 1-339) was cloned into the GAL4 DNA-binding vector pGBT9 (CLONTECH, Palo Alto, CA) and used as bait in a two-hybrid screen of a human liver cDNA library (CLONTECH) in Saccharomyces cerevisiae Y190; positive clones were selected as described (22).
In Vitro Translation and GST Pull-down Assays-In vitro transcription and translation were carried out with 1 g of HA-TANK, as described (22). Both the full-length GST-NEMO/IKK␥ fusion protein and the wild type GST were produced and purified as described (22). Protein-protein interactions were performed by incubating an aliquot of GST-NEMO/IKK␥ or GST bound to the glutathione-Sepharose beads with 5 l of in vitro translated HA-TANK as described (31). Beads were washed five times with 1 ml of phosphate-buffered saline, 1% Triton, protease inhibitors; resuspended into migrating buffer; and run on an SDS-polyacrylamide gel before autoradiography.
Immunoprecipitations-For immunoprecipitations involving overexpressed proteins, 293 or HeLa cells (3 ϫ 10 6 ) were transfected via LipofectAMINE (Invitrogen) with expression vectors as indicated in Figs. 2 (B, C, and D), 3 (B and D), and 4 (B, C, and D). 24 h after transfection, cells were then washed with phosphate-buffered saline and lysed in 0.5% Triton lysis buffer. Ectopically expressed proteins were immunoprecipitated by using anti-FLAG or anti-HA antibodies bound to agarose beads or by using anti-NEMO antibodies (as indicated). The immunoprecipitate was washed five times with 0.5% lysis buffer and subjected to SDS-PAGE.
Immunoprecipitations of endogenous NEMO/IKK␥ or endogenous IKK␣ were performed with 293 or HeLa cells (6 ϫ 10 6 ) after cells had been transfected with expression vectors for various proteins as indicated in Figs. 5 and 6B. Harvested cells were washed and lysed as described above. Anti-NEMO/IKK␥ immunoprecipitations were performed by incubating the total cell extracts with polyclonal antibodies for 2 h at 4°C, followed by an overnight incubation with protein Aagarose conjugate (Santa Cruz Biotechnology) (see Figs. 5, A and C, and 6), whereas anti-IKK␣ immunoprecipitations were performed by incubating the cell extracts with monoclonal anti-IKK␣ beads overnight (see Fig. 5B).
For identification of ternary complexes by immunoprecipitation experiments (see Figs. 2D and 7 (A and B)), 293 cells (10 7 ) were transfected with the indicated expression vectors, as described above. 24 h after transfection, ectopically expressed TANK was immunoprecipitated with anti-FLAG for 2 h at 4°C. The immunoprecipitate was washed five times with the lysis buffer and incubated overnight with the FLAG peptide (Sigma), according to the protocol provided by the manufacturer. The supernatants were subsequently incubated with anti-NEMO antibodies and protein A-agarose conjugate overnight. The resulting immunoprecipitates were washed with the lysis buffer and subjected to SDS-PAGE.
For endogenous coimmunoprecipitations, 293 cells (6 ϫ 10 7 ) were left untreated or were stimulated with 40 ng/ml of PMA and 2 M ionomycin (Sigma) for the indicated period of time and subsequently lysed in the lysis buffer. An anti-TANK immunoprecipitation was then performed with the polyclonal antibodies for 2 h at 4°C, followed by incubation overnight with protein A-agarose conjugate. An immunoprecipitation with an aliquot of a prebleed rabbit serum was performed in parallel as a negative control. The immunoprecipitates were subjected to anti-IKK⑀ and -IKK␣ Western blots.

Identification of TANK as a NEMO/IKK␥-interacting Pro-
tein-To gain insights into how NEMO/IKK␥ may transmit signals to the IKKs, we screened for NEMO/IKK␥ -interacting proteins via yeast two-hybrid assays. Mouse NEMO/IKK␥ (aa 1-339) was used as bait in a fusion with the DNA binding domain of GAL4 to trap interacting proteins generated from a human liver expression library fused to the GAL4 activation domain. Positive clones isolated included portions of IKK␣, IKK␤, and CIKS (22). In addition, three independent and overlapping clones encoded parts of TANK (also known as TRAFinteracting protein or I-TRAF). TANK had previously been identified as a potential regulator in NF-B-activating pathways, although its precise role is controversial. TANK was originally discovered as a protein capable of binding to TRAF1, -2, and -3 (24 -26).
To delineate the region in NEMO/IKK␥ that is required for interaction with TANK, we tested various truncation mutants of NEMO/IKK␥ (fused to the GAL4 DNA binding domain) for binding to the isolated TANK-GAL4 activation domain fusion protein in yeast (Fig. 1B). Among the C-terminal deletion mutants of the 412-amino acid-long NEMO/IKK␥ protein, one lacking the last 100 amino acids and thus lacking the entire leucine zipper domain was still able to interact with TANK. Among the various N-terminal deletions of NEMO/IKK␥, those lacking any or all of the first 200 amino acids were still able to bind TANK, whereas one lacking the first 250 amino acids was not. Complementing this result, a NEMO/IKK␥ construct composed of amino acids 150 -250 was sufficient to mediate the interaction with TANK. Based on these findings, we conclude that the region between amino acids 200 and 250 of NEMO/ IKK␥ mediates binding to TANK. This domain of NEMO/IKK␥ is distinct from the one required for interaction with the IKKs (amino acids 50 -100) (32,33). 2 The interaction between NEMO/IKK␥ and TANK was also confirmed in vitro. An Escherichia coli-produced recombinant GST-NEMO/IKK␥ fusion protein (full-length) bound in vitro translated [ 35 S]HA-TANK, whereas recombinant GST alone did not ( Fig. 2A, top panel).
Next we investigated the interaction of NEMO/IKK␥ with TANK in mammalian cells. FLAG-TANK was transiently coexpressed in 293 cells together with HA-NEMO/IKK␥ or a NEMO/IKK␥ mutant lacking the first 250 amino acids (HA-NEMO/IKK␥ ⌬N250). Cell extracts were immunoprecipitated with anti-FLAG antibodies (Fig. 2B, top panel) and HA-NEMO/ IKK␥ was co-immunoprecipitated, but only if FLAG-TANK had also been co-transfected (lane 5). The specificity of the interaction between NEMO/IKK␥ and TANK was confirmed by the fact that the NEMO/IKK␥ deletion mutant lacking the first 250 amino acids (HA-NEMO/IKK␥ ⌬N250) was not co-immunoprecipitated with TANK (lane 6), in agreement with the data in yeast. Similar results were obtained when the NEMO/IKK␥ was immunoprecipitated to look for TANK (data not shown). Note that when both NEMO/IKK␥ and TANK were co-expressed, a shift in the migration of the TANK protein was detected by Western blot, even in the absence of any overexpressed kinases (middle panel, lane 5). To further confirm the interaction between TANK and NEMO/IKK␥, 293 cells were transfected with both HA-NEMO/IKK␥ and FLAG-TANK, and extracts were immunoprecipitated with anti-HA (Fig. 2C). TANK was co-immunoprecipitated with NEMO/IKK␥ (Fig. 2C,  top panel, lane 2). A shift in the TANK protein was detected when co-expressed with NEMO/IKK␥ (Fig. 2C, middle panel; lane 2, FLAG-TANK*), most likely due to phosphorylation. Interestingly, it is this slower migrating form of TANK that preferentially co-immunoprecipitated with HA-NEMO/IKK␥ (Fig. 2C, top panel, lane 2).
Because FLAG-TANK could also be shown to co-immunoprecipitate with HA-IKK␤, especially if NEMO/IKK␥ was cotransfected (data not shown), we asked whether these three proteins might be able to form a ternary complex. To test this, 293 cells were transfected either with HA-IKK␤ or with FLAG-TANK or both (Fig. 2D). Anti-FLAG immunoprecipitations were carried out, and the immunoprecipitates were released from the beads by incubating them with a FLAG peptide. The released material was immunoprecipitated with antibodies to the endogenous NEMO/IKK␥ and an anti-HA Western analysis was per-formed, revealing the presence of IKK␤ (lane 2). Therefore, a ternary complex of TANK, NEMO/IKK␥, and IKK␤ must have been formed in 293 cells. Importantly, this complex was formed with endogenous NEMO/IKK␥, demonstrating that endogenous levels of NEMO/IKK␥ were sufficient to mediate the interaction between transfected TANK and IKK␤.
Two Distinct Regions of TANK Are Required for Interaction with NEMO/IKK␥-To delineate the domain in TANK required for interaction with NEMO/IKK␥ in mammalian cells, various truncated TANK proteins were generated (Fig. 3, A and C) and tested for their ability to co-immunoprecipitate with NEMO/ IKK␥ in 293 cells (Fig. 3, B and D). A TANK protein lacking the first N-terminal 30 aa (Fig. 3A) was able to co-immunoprecipitate with NEMO/IKK␥ (Fig. 3B, lane 7), but TANK proteins lacking the first N-terminal 70 aa or more were not (lanes 3-5, 8, and 9). All C-terminal deletions of TANK tested (Fig. 3C) failed to co-immunoprecipitate with NEMO/IKK␥ (Fig. 3D, lanes [3][4][5][6][7][8]. This suggests that an N-terminal TANK domain (between aa 30 and 70) and a C-terminal TANK domain (between aa 248 and 425) are both required for interaction with the regulatory subunit of the IKK complex in mammalian cells. (The same results were obtained with a NEMO/IKK␥ construct lacking the C-terminal 72 aa; data not shown). By contrast, the C-terminal domain of TANK was sufficient in yeast (see Fig.  1A). The reason for this is not clear, but the assay for the interaction in yeast may be more sensitive than the one in mammalian cells.
TANK Binding-deficient NEMO/IKK␥ Mutant Impaired in Mediating PMA and Ionomycin (P/I)-induced NF-B Activation-We next explored the possible relevance of the interaction of TANK with NEMO/IKK␥ in mediating activation of NF-B. A NEMO/IKK␥ mutant was constructed in which the TANK-binding domain was specifically deleted (NEMO⌬TANK) (Fig. 4A). When overexpressed in 293 cells, this NEMO mutant failed to interact with TANK (Fig. 4B, top panel, lane 6), as predicted by the results obtained in yeast (see Fig. 1B). However, this NEMO mutant still interacted with transfected IKK␤ (Fig. 4C, top panel, lane 3). Moreover, NEMO⌬TANK also interacted with CIKS, another NEMO/ IKK␥-interacting protein (22) (Fig. 4D, top panel, lane 3). NEMO/IKK␥ thus interacts with TANK via a domain not required for interaction of NEMO/IKK␥ with the IKKs or with CIKS. We then tested the ability of the NEMO⌬TANK mutant to restore NF-B activation in NEMO-deficient Jurkat cells (34) in response to stimulation with P/I. Whereas transfection of wild-type NEMO/IKK␥ led to significant P/I-induced B reporter activity, the NEMO⌬TANK mutant was largely unable to transmit this signal (Fig. 4E). Although this does not prove that interaction with TANK is critical for the function of NEMO, given that as yet unknown functions of NEMO may have been impaired in this particular mutant, the data are nonetheless consistent with the notion that NEMO/IKK␥ normally has to bind to proteins such as TANK to be fully functional.
IKK⑀ and TBK1 Promote the Interaction of TANK with the IKK Complex-Two reports identified murine inducible IB kinase (35) (human homolog termed IKK⑀ (36)) and TBK1 (also named NF-B-activating kinase (28) and T2K (37)) as two TANK-interacting kinases (27,35) capable of activating NF-B in transfection experiments. Inducible IB kinase and TBK1 were shown to interact with the N-terminal half of TANK and to cause TANK phosphorylation in cotransfection experiments in the C-terminal half (27,35). Nevertheless, mechanisms for activation of NF-B by these kinases remained uncertain. We confirmed and extended the published work on the interaction and phosphorylation of TANK with IKK⑀ and TBK1. Both kinases interacted with TANK in the region between amino acids 111 and 169 (just C-terminal to the first of two domains required for interaction with NEMO/IKK␥), and they phosphorylated TANK between amino acids 192 and 247, dependent on the interaction (data not shown).
To investigate whether IKK⑀ may be involved in regulating the ability of TANK to interact with NEMO/IKK␥, 293 cells were transfected with Myc-tagged IKK⑀ (Fig. 5A, lane 1 3, top panel). Similarly, coexpression of transfected TANK resulted in a readily detectable co-immunoprecipitation of IKK⑀ and NEMO/IKK␥ (lane 3, second panel from top). Interestingly, the K38A IKK⑀ mutant also promoted the interaction between TANK and NEMO/IKK␥, albeit it to a lesser degree, suggesting that the kinase activity of IKK⑀ is not absolutely required for this effect (lane 4, top two panels).
An analogous experiment was performed in which endogenous IKK␣ was immunoprecipitated instead of endogenous NEMO/IKK␥ (Fig. 5B). As expected, little of the exogenously introduced TANK was found in association with IKK␣ in the absence of transfected IKK⑀, but the presence of IKK⑀ strongly promoted the interaction of TANK with IKK␣, presumably via NEMO/IKK␥ (top panel, lanes 2 and 3, respectively). Again, this effect of IKK⑀ was largely independent of its kinase activity (lane 4). Taken together, the results suggest that TANK can link IKK⑀ to the IKK complex and that TANK and IKK⑀ synergize to promote this interaction, largely independent of IKK⑀ kinase activity.
We obtained similar results when TBK1 was tested in these types of experiments. As with IKK⑀, exogenously introduced TBK1 could be readily found in association with endogenous NEMO/IKK␥, but only in the presence of exogenously introduced TANK (Fig. 5C, lane 3).
Given that IKK⑀ (and TBK1) promote the association of TANK with the IKKs, although they do not interact with the IKKs by themselves, we tested whether or not a TANK construct lacking the IKK⑀-interacting domain (FLAG-TANK ⌬IKK⑀; Fig. 6A) could still be promoted by IKK⑀ to coimmunoprecipitate with NEMO/IKK␥. We first demonstrated with transfection experiments in 293 cells that such a mutant of TANK indeed failed to interact with IKK⑀ and failed to be phosphorylated by IKK⑀ in an in vitro kinase assay but continued to co-immunoprecipitate well with co-transfected NEMO/ IKK␥ and IKK␤, as predicted (data not shown). Such a mutant allowed us to ask whether IKK⑀ promoted the association of TANK with NEMO/IKK␥ by a direct association with TANK or whether this might occur indirectly via an effect of IKK⑀ on the IKK complex. As shown in Fig. 6B (lane 5), the TANK mutant lacking the IKK⑀ binding domain was also no longer promoted by this kinase to interact with endogenous NEMO/IKK␥, whereas wild-type TANK was, regardless of whether IKK⑀ was wild-type or kinase-inactive (Fig. 6B, lanes 2 and 3, respectively). These results suggest that the direct association of IKK⑀ with TANK allows these proteins to cooperatively interact with NEMO/IKK␥. It is possible, for example, that binding of IKK⑀ changes the conformation of TANK such that it more readily interacts with the IKK complex.
The results also suggest that TANK might be part of a ternary complex with both IKK⑀ and NEMO/IKK ␥ (and thus the IKK complex). To test such a hypothesis directly, we transfected 293 cells with FLAG-tagged TANK and either wild-type (WT) or K38A mutant (DN) Myc-tagged IKK⑀ (Fig. 7A, lanes 2  and 3, respectively). An anti-FLAG immunoprecipitation was carried out, followed by incubation with a FLAG peptide to elute the immunoprecipitated material so that it could be reimmunoprecipitated with antibodies to endogenous NEMO/ IKK␥. These final immunoprecipitates were subjected to an anti-Myc Western analysis. In such experiments, we detected both WT IKK⑀ and the K38A (DN) mutant (upper panel, lanes   2 and 3, respectively), indicative of the existence of a ternary complex that includes TANK, IKK⑀, and endogenous NEMO/ IKK␥. The same results were obtained in a similar experiment in which endogenous IKK␣ was immunoprecipitated instead of endogenous NEMO/IKK␥ (Fig. 7B). These experiments suggest that ectopically expressed IKK⑀ can be part of a ternary complex with TANK and the IKK complex.
Endogenous TANK Associates with Endogenous IKK, Independent of P/I-IKK⑀ has been described as part of a PMAinducible IKK-like complex that contains an unknown IKKlike kinase activity (36). We therefore investigated whether stimulation of 293 cells with P/I could modulate the ability of TANK to interact with IKK⑀ or with IKK␣, a representative component of the IKK complex (Fig. 8). 293 cells were either left unstimulated or were treated from 15 min to 8 h with P/I prior to harvest, and total cell extracts were subjected to an anti-TANK immunoprecipitation, followed by an anti-IKK⑀ or anti-IKK␣ Western analysis. As expected, we detected a strong coimmunoprecipitation between endogenous TANK and endogenous IKK⑀, but this association was not modulated by the P/I treatment (second panel from top, lanes 2-7). We also observed a much weaker but very reproducible association of endogenous TANK with endogenous IKK␣ (top panel; overnight exposure; IKK⑀ was detected within minutes). Again, the association of TANK with the IKK complex (as demonstrated for IKK␣) was not modulated by the P/I stimulation (top panel, lanes 2-7). These results suggest that P/I treatment, which was hypothesized to activate IKK⑀ has apparently no effect on the ability of TANK to associate with IKK⑀ or to associate with the IKK complex. This is at least consistent with results above that demonstrated that the kinase activity of IKK⑀ is not required for association with the IKK complex via TANK. DISCUSSION We have shown here with experiments in yeast, in vitro and in transfected cells, that TANK can physically associate with NEMO/IKK␥ and thus the IKK complex. Two domains of TANK are required for the interaction with an N-terminal domain of NEMO/IKK␥ in mammalian cells. An association of TANK with IKK complexes could also be demonstrated in untransfected cells. Although TANK has been previously implicated in regulation of NF-B activity, a direct link to NEMO/ IKK␥ or to the IKK complex has not been reported. This discovery supports the previously suggested notion that NEMO/IKK␥ serves an adapter function to link upstream signal mediators with the IKK complex (12). In addition to TANK, other proteins such as the previously identified CIKS (22) (also known as Act1) may also interact with NEMO/IKK␥ to link IKKs to select upstream signaling pathways. Some IKKs may be dedicated to specific signaling pathways.
Transfected TANK was reported to negatively affect activation of NF-B in response to various stimuli (25,26). The mechanisms for this negative effect remain to be determined, although it was suggested that for some signals, TANK could inhibit by competing with members of the TNF receptor family for binding to TRAF2 (26). Transfected TANK was also reported to positively regulate activation of NF-B together with low levels of co-transfected TRAF2 (24). This effect was subsequently described to be mediated by the association of TANK with kinases distantly related to IKK␣/␤, namely TBK1 and IKK⑀ (27,35). However, what signals these kinases respond to and by what mechanism they may activate NF-B in concert with TANK has remained unclear. We have shown here that the association of TANK with NEMO/IKK␥ and the IKK complex is dramatically increased in the presence of transfected IKK⑀ or TBK1. The physical interaction of IKK⑀ and TBK1 with TANK is sufficient to promote the interaction of TANK with NEMO/IKK␥, whereas their kinase activities are largely dispensable for this effect. IKK⑀ was previously also reported to associate with and regulate an as yet unidentified IKK-like kinase (35). Whereas the present data do not address this issue, they do demonstrate an association of IKK⑀ with the classical IKK kinases, which could of course occur in addition to the association with an unknown IKK-like activity. Our data suggest that TANK may function as an adapter to mediate a direct influence of IKK⑀ and TBK1 on the IKK core complex or on other proteins directly associated with the core IKK complex. We speculate that at least a subset of IKK core complexes exist as part of more loosely assembled, larger signaling complexes that may serve to channel specific activation signals, possibly at special sites within cells. As part of such larger signaling complexes surrounding some IKK cores, TBK1 and IKK⑀ could be in a position not only to directly modulate the IKK␣/␤ kinase activity (27) but conceivably also to regulate other aspects, such as association of the IKK␣/␤ kinases with their substrates or the phosphorylation of NF-B proteins.
In attempts to find a functional requirement for the association of TANK with NEMO/IKK␥, we discovered a possible role in mediating activation of NF-B via P/I. NEMO-deficient Jurkat cells reconstituted with a NEMO mutant lacking the TANK-interacting site (but able to bind CIKS and IKK␣/␤) are significantly impaired in P/I-induced activation of NF-B as compared with NEMO-deficient Jurkat cells reconstituted with wild-type NEMO. This suggests that TANK or another protein binding NEMO in the same domain may be required for NEMO to properly channel signals to the IKKs, although alternative explanations cannot be ruled out as yet.
An association of TBK1 or IKK⑀ with the IKK complex could affect NF-B activity in several ways. In addition to the possibility that TBK1 and IKK⑀ activate the IKKs (28), the association with the IKK complex could also help these kinases modulate other functions, such as the transactivation potential of NF-B proteins. Such a hypothesis can be derived from T2K-deficient mice. T2K-deficient embryos succumbed to massive apoptosis in the liver, similar to IKK␤ and RelA-deficient mice (38 -41). In the two latter knockouts, the defect was shown to be due lack of activation of NF-B in response to TNF, which led to TNF-induced apoptosis, unopposed by the normally protective effects of NF-B. In T2K-deficient embryonic cells, however, inflammatory cytokine-induced liberation of NF-B from their IB inhibitors was shown to be largely intact, and some NF-B target genes were induced, whereas others were not. Therefore, it was speculated that promoter-specific transactivation functions of liberated NF-B proteins might be targeted by TBK1. Given the ready access TBK1 and IKK⑀ could have to NF-B dimers via their association with TANK and the IKK complex (as shown here), these two kinases could well be in position to modulate transactivation functions of NF-B proteins such as RelA.
In summary, our data provide direct evidence that at least some IKK core complexes can be linked to TANK or other potentially similarly acting proteins. TANK may function as an adapter for the IKK⑀ and TBK1 kinases. These kinases could be liberated and activated by as yet unknown signals so that they may, together with TANK, synergistically engage the IKKs to form a ternary complex. As part of such a hypothesized larger IKK complexes, IKK⑀ and TBK1 could directly modulate activities of the IKK complex. TANK and CIKS may belong to a larger family of adaptors dedicated to link specific signaling pathways to IKK complexes.