Activation of the I k B Kinases by RIP via IKK g /NEMO-mediated Oligomerization*

To understand the mechanism of activation of the I k B kinase (IKK) complex in the tumor necrosis factor (TNF) receptor 1 pathway, we examined the possibility that oligomerization of the IKK complex triggered by ligand-induced trimerization of the TNF receptor 1 complex is responsible for activation of the IKKs. Gel filtration analysis of the IKK complex revealed that TNF a stimulation induces a large increase in the size of this complex, suggesting oligomerization. Substitution of the C-terminal region of IKK g , which interacts with RIP, with a truncated DR4 lacking its cytoplasmic death domain, produced a molecule that could induce IKK and NF- k B activation in cells in response to TRAIL. Enforced oligomerization of the N terminus of IKK g or truncated IKK a or IKK b lacking their serine-cluster domains can also induce IKK and NF- k B activation. These data suggest that IKK g functions as a signaling adaptor between the upstream regulators such as RIP and the IKKs and that oligomerization of the IKK complex by upstream regulators is a critical step in activation of this complex. In unstimulated cells, the transcription factor NF- k B is se-questered in the cytoplasm

To understand the mechanism of activation of the IB kinase (IKK) complex in the tumor necrosis factor (TNF) receptor 1 pathway, we examined the possibility that oligomerization of the IKK complex triggered by ligandinduced trimerization of the TNF receptor 1 complex is responsible for activation of the IKKs. Gel filtration analysis of the IKK complex revealed that TNF␣ stimulation induces a large increase in the size of this complex, suggesting oligomerization. Substitution of the Cterminal region of IKK␥, which interacts with RIP, with a truncated DR4 lacking its cytoplasmic death domain, produced a molecule that could induce IKK and NF-B activation in cells in response to TRAIL. Enforced oligomerization of the N terminus of IKK␥ or truncated IKK␣ or IKK␤ lacking their serine-cluster domains can also induce IKK and NF-B activation. These data suggest that IKK␥ functions as a signaling adaptor between the upstream regulators such as RIP and the IKKs and that oligomerization of the IKK complex by upstream regulators is a critical step in activation of this complex.
In unstimulated cells, the transcription factor NF-B is sequestered in the cytoplasm through interaction with inhibitory proteins known as IBs (1). Upon stimulation by proinflammatory cytokines like TNF␣, IBs undergo phosphorylation at specific serine residues by kinases known as IB kinases (IKKs). 1 Phosphorylation marks IBs for ubiquitination and degradation via the proteosome pathway (2)(3)(4)(5). Degradation of IBs allows the liberated NF-B to translocate to the nucleus and activate the transcription of target genes (6).
Purified recombinant IKK␣ and IKK␤ are both able to phosphorylate IB␣ and IB␤ and can form homo-and heterodimers through their LZ motifs. Mutations interfering with this dimerization abolish kinase activity (11,24). No IKK activity can be elicited in vivo in IKK␥-deficient cells after treatment with TNF␣ or interleukin-1 (12,16,17). Moreover, IKK complexes assembled in vivo in cells expressing a truncated IKK␥ lacking its C-terminal LZ were not responsive to cytokine-mediated activation (13). These results clearly highlight the importance of IKK␥ in the activation of NF-B and suggest that the C-terminal region of this protein is necessary for engagement by upstream activators. However, the mechanism by which the IKK complex is activated remains to be elucidated.
TNF␣ induces NF-B activation through binding to two distinct receptors, p55 TNF-R1 and p75 TNF-R2, which are expressed in almost all cell types (25). Stimulation of the TNF receptors by TNF␣ promotes the assembly of two distinct signaling complexes via recruitment of different signaling proteins to the cytoplasmic tails of TNF-R1 and TNF-R2. Among these proteins TRADD, TRAF2, RIP, and FADD (26 -28) are recruited by TNF-R1, whereas TRAF2, TRAF1, and TRIP (29,30) are recruited by TNF-R2. Other proteins such as IAP1, IAP2, TANK, A20, NIK, and MEKK1 (21,(31)(32)(33)(34)(35) associate with the TNF receptor complexes and have also been suggested to play a role in regulation of NF-B activation by these receptors. However, the events following assembly of the TNF-R1 or TNF-R2 complexes that lead to activation of the IKK complex need to be addressed.
Recently, NEMO/IKK␥ has been shown to interact with RIP (15,36,37), an adapter protein that associates with the p55 receptor. RIP contains three major domains: a kinase domain at its N terminus, an intermediate domain, and a death domain at its C terminus. The use of RIP-deficient cell lines has highlighted the indispensable role of RIP in the p55 TNF␣ receptorinduced NF-B activation (38,39). Whereas RIP possesses a serine/threonine kinase domain capable of autophosphorylation, the kinase activity of RIP is not required for NF-B activation (27,38) Indeed, expression of a RIP mutant consist-ing of only its intermediate domain is sufficient to activate NF-B (27). Interestingly, Zhang and co-workers (37) have demonstrated that the intermediate domain of RIP is the region involved in the binding of this protein to IKK␥. It seems therefore possible that the role of RIP could be that of a signaling molecule in the TNF␣ signaling pathway, recruiting and channeling the signal from the TNF-R complex to the IKK complex.
Here we investigated the potential role of oligomerization in activation of the IKK complex. We show that the IKK complex undergoes oligomerization in response to TNF␣ stimulation and that RIP may regulate the activity of the IKK complex through oligomerization of IKK␥. Other evidence presented here demonstrate that IKK␥-mediated oligomerization of the IKK complex or enforced oligomerization of any of the IKK components is sufficient to activate the IKK kinase complex and the NF-B pathway. Expression Vectors and Antibodies-The full-length open reading frame of human IKK␥ was cloned from an HeLa cell line cDNA library by PCR and subcloned in pcDNA3 vector (Invitrogen). cDNAs for wild type IKK␣ and IKK␤ as well as the kinase-inactive mutants are gift from W. C. Green and have been described. The mouse RIP cDNA was a gift from B. Seed. Constructs encoding full-length RIP, IKK␥, IKK␣, or IKK␤ or truncated mutants were generated by PCR using modified complementary PCR adapter primers. The FKBP12 fusion of RIP intermediate domain (residues 286 -579), IKK␥, IKK␣, or IKK␤ were constructed in pcDNA3 by fusing three tandem repeats of FKBP12 cDNA in frame with the cDNAs of RIP, IKK␥, IKK␣, or IKK␤. The cDNA of the three tandem repeats of FKBP12 was excised from the pFkp3-HA vector by digestion with XhoI and subcloned into an XhoI cut pcDNA3 or a modified pcDNA3-T7 vector that contains a T7 tag sequence. Site-directed mutagenesis was done using overlapping PCR. FLAG and T7 epitope tagging was done by cloning the PCR-generated cDNAs of the respective genes in-frame into pFLAG CMV-2 (IBI Kodak) and pcDNA3 (Invitrogen) vectors, respectively. FLAG-M5 antibody was from Eastman Kodak Co. T7-HRP conjugate antibody was from Novagen. IKK␣, TRAF2, TRADD, and GFP polyclonal antibodies were from Santa Cruz. Monoclonal anti-RIP antibody was from PharMingen. NEMO/IKK␥ antibody was a gift from A. Israel.
Gel Filtration of Cellular Extracts on Superose 6 Column-100 ϫ 10 6 HeLa cells were washed in prewarmed phosphate-buffered saline and treated with TNF␣ for 5 min. Cells were then collected, washed in cold phosphate-buffered saline, and lysed in a buffer containing 25 mM Tris-HCl, pH 7.6, 100 mM NaCl, 1 mM EGTA, 1 mM dithiothreitol, 0.2% Nonidet P-40, 5% glycerol, 1 mM Na 3 Vo 4 , 10 mM ␤-glycerophosphate, 5 mM NaF, 10 g⅐ml Ϫ1 leupeptin and aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride. Lysates were incubated on ice for 30 min and centrifuged at 20,000 ϫ g for 15 min, and the supernatants were collected and recentrifuged for1 h at 100,000 g. 0.25 mg of the S-100 extracts (0.1 ml) were loaded onto a Superose-6 column (Amersham Pharmacia Biotech) pre-equilibrated with the lysis buffer, and proteins were eluted from the column at the flow rate of 0.25 ml/min. 0.2 ml of the 0.5-ml fractions were precipitated with 10% trichloroacetic acid and analyzed by SDS-polyacrylamide gel electrophoresis followed by Western blotting for IKK␣, IKK␥, TRAF2, TRADD, and RIP using appropriate antibodies.
Immunoprecipitation-Cells were lysed in 50 mM Tris-HCl, pH 7.6, 150 mM NaCl containing 0.5% Nonidet P-40, 1 mM Na 3 Vo 4 , 10 mM ␤-glycerophosphate, 5 mM NaF, 10 g⅐ml Ϫ1 leupeptin and aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride and clarified by centrifugation at 15,000 ϫ g for 15 min. The clarified lysates were preabsorbed on protein G-Sepharose (Amersham Pharmacia Biotech) and then incubated with antibody for 2 h, followed by protein G-Sepharose agarose IgG beads. Immune complexes were washed extensively in the lysis buffer and eluted by boiling in SDS sample buffer.
Reporter Gene Assays-293T cells (2.10 5 cells/well) in 12-well plate were transfected with 5XB-luciferase reporter plasmid together with each expression vector, as indicated. The total amount of DNA (0.7 g) was kept constant by inclusion of empty vector DNAs. The luciferase activity was determined with a Luciferase Assay System (Promega). A LacZ-expressing plasmid was used for normalizing transfection efficiencies.
IKK Kinase Assay-HeLa cells were transfected with epitope-tagged IKK␣ or IKK␤ expression plasmids together with other indicated plasmids. 24 h after transfection, cells were lysed in a buffer containing 50 mM Tris, pH 7.6, 137 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 1 mM Na 3 VO 4 , 10 mM ␤-glycerophosphate, 5 mM NaF protease inhibitors, and 1 mM dithiothreitol. Cell lysates were immunoprecipitated with the anti-epitope monoclonal antibody conjugated on agarose beads, washed three times with the lysis buffer, and washed twice with the kinase buffer (40 mM Tris, pH 7.5, 0.2 mM EDTA, 10 mM MgCl 2 , 10 mM ␤-glycerophosphate, 200 M Na 3 VO 4 , and 1 mM dithiothreitol). The immuoprecipitates were resuspended in 40 l of the kinase buffer containing 5 Ci of [␥-32 P]ATP (3000 Ci/mmol) and 1 g of recombinant glutathione S-transferase-IB␣ proteins as exogenous substrates of IKK. After incubation at room temperature for 30 min, the reactions were stopped by adding SDS sample buffer. The samples were analyzed by SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes, followed by autoradiography.
Confocal Microscopy-293T cells were grown on coverslips and then transfected with the GFP-tagged (SCD) of IKK␣ or IKK␤ together with the indicated vectors. 24 h after transfection, cells were left untreated or incubated with AP1510 for 30 min and then fixed with 4% paraformaldehyde in phosphate-buffered saline for 30 min. The coverslips were mounted on a glass slide, and the fluorescence of GFP was detected by confocal microscopy using excitation wavelength of 488 nm and detection wavelength of 522 nm. Images were Kalman-averaged.

Oligomerization of the IKK Complex in Response to TNF␣
Stimulation-Because TNF␣ is a trimer, ligation of TNF-R1 should result in a sequential oligomerization of the receptorinteracting protein RIP, followed by oligomerization of the IKK complex, which interacts with RIP through IKK␥ (see below and Refs. 15, 36, and 37). Therefore, if such a process occurs, this should result in an increase in the molecular mass of the IKK complex. To test this possibility, we analyzed the IKK complex elution profile and activity following chromatographic fractionation of 293T cellular extracts on a Superose-6 fast protein liquid chromatography column. We reasoned that if TNF␣ treatment induces oligomerization of the IKK complex, we should be able to detect a shift in the elution profile of the IKK components in extracts prepared from TNF-treated cells compared with untreated cells. Cellular extracts were prepared under mild condition to prevent dissociation of the complexes and then loaded onto a Superose-6 column. Fractions eluted from the column were analyzed for IKK activity and the presence of specific components of the IKK and TNF-R1 complexes using Western blot analysis. In extracts of unstimulated cells, the majority of IKK␣ and IKK␥ were eluted in a peak centered around fraction 25 (relative molecular mass, ϳ650,000; M r , ϳ650 kDa) (Fig. 1). In addition to this peak, a smaller amount of IKK␣ and IKK␥ was also detected in a minor peak of approximately 150 kDa. No kinase activity was detected in any of these fractions. Of note, both IKK␣ and IKK␥ migrated as doublets or triplets, which might be corresponding to different states of phosphorylation (9,40). Among these, the faster migrating bands were the predominant species. Interestingly, a dramatic shift in the IKK complex elution profile was observed in extracts of TNF␣-stimulated cells. IKK␣ and IKK␥ eluted in a major peak of approximately 1.3 MDaЈ, twice the size of the unstimulated complex. The smaller pool of IKK was still detected, although it shifted to a fraction of a molecular mass of approximately 200 kDa. Interestingly, after TNF␣ stimulation, the intensity of the slower migrating band of IKK␣ and IKK␥ increased dramatically compared with the faster migrating bands. This modification could be the result of an increase in phosphorylation of IKK␣ and IKK␥ after stimulation with TNF␣ and activation of the IKK complex. Indeed, biochemical analyses have indicated that the three subunits of the IKK complex undergo phosphorylation after TNF stimulation (40). Importantly, the IKK activity was associated predominantly with the fraction containing the 1.3-MDa complex, whereas little activity was detected in the 650-kDa fraction.
None of the downstream components of the TNF-R1, namely TRADD, TRAF2, and RIP, co-eluted with the activated IKK complex. Moreover, unlike IKK␣ or IKK␥, TNF␣ stimulation did not affect the distribution of TRADD, TRAF2, or RIP. These proteins were detected in fractions 27-31, which correspond to sizes of ϳ200 to ϳ500 kDa, in both unstimulated and stimulated extracts. RIP was less abundant in the fractions from the unstimulated cells. Perhaps this could be due to its association with the TNF-R1 complex in the nonsoluble membrane fraction before stimulation. Of note, prior to TNF stimulation, a small amount of the IKK components was also detected in fraction 27, thus co-eluting with TRAF2. However, TNF treatment of the cells clearly separated the elution profiles of TRADD, TRAF2, and RIP from that of the IKK components. Consequently, the molecular mass increase observed for the IKK complex following TNF treatment cannot be explained by a stable association between the IKK complex and components of the activated TNF-R1. Instead, our results suggest that the small IKK complex (M r , ϳ650 kDa) is dynamically recruited to the TNF-R1, activated via oligomerization, and then released as a larger complex containing the activated kinases.
IKK␥ Mediates the Association of RIP with the IKK Complex-Recent observations suggest that RIP recruits the IKK complex to the TNF-R1 in a stimulus-dependent manner (37). Moreover, RIP and IKK␥ has been shown to interact in a yeast two-hybrid system (37). To confirm which component of the IKK complex serves as the primary target of RIP, human 293T cells were transfected with expression constructs for FLAGtagged RIP and T7-IKK␤ with or without T7-IKK␥. In the absence of RIP, IKK␤ and IKK␥ were not precipitated with the FLAG antibody ( Fig. 2A). In the absence of ectopic IKK␥, a small amount of IKK␤ was co-immunoprecipitated with RIP. Interestingly, in the presence of ectopic IKK␥, a remarkably higher amount of IKK␤ co-immunoprecipitated with RIP. The ectopic T7-IKK␥ was also detected in these complexes. This result shows that IKK␥ mediates the interaction of RIP with the IKK complex, thus confirming the direct interaction of RIP with IKK␥.  (41). Transient transfection of the RIP ID-FKBP12 construct into 293T cells resulted in expression of a protein of the expected size (Fig. 2D). In the absence of AP1510, only a modest NF-B activation was observed in cells expressing the RIP ID-FKBP12 chimera (Fig. 2C). However, incubation of the RIP ID-FKBP12 transfected cells, but not the empty vector transfected cells, with AP1510 resulted in a large increase in NF-B activity (Fig. 2C). Very little NF-B activation was detected when the RIP ID-FKBP12 chimera was transfected with kinase-inactive IKK␤ in the presence of AP1510 (Fig. 2C). AP1510-induced oligomerization also resulted in activation of IKK activity to a magnitude similar to that observed with TNF␣ (Fig. 2D).

Enforced Oligomerization of the Central Domain of RIP Induces IKK Activation-Based
To address the physiological relevance of this finding, we transiently expressed the RIP ID-FKBP12 chimera in wild type or IKK␥-deficient Rat-1 cells (12). Although NF-B activation was detected in wild type Rat-1 cells, no activation was observed in the IKK␥-deficient 5R cells after incubation of the RIP ID-FKBP12 transfected cells with AP1510 or treatment with TNF␣ (Fig. 2E). This result provides genetic proof for the importance of IKK␥ in RIP-induced activation of NF-B, confirming its role as a molecular adaptor in the assembly of the RIP/IKK complexes. The inability of the RIP ID-FKBP12 chimera to induce NF-B activation in 5R cells cannot be attributed to defects in the NF-B pathway downstream of IKK␥ 293T cells were either mock-treated or treated with TNF␣ for 5 min. S-100 extracts prepared from these cells were fractionated by gel filtration chromatography on a Superose-6 column. Proteins in each alternative fractions were analyzed by Western blotting using the indicated antibodies and assayed for IB␣ kinase activity following immunoprecipitation with an anti-IKK␣ antibody. KA, kinase assay.
because transfection of these cells with IKK␥ can restore NF-B activation by Tax, which is expressed stably in this cell line (Ref. 12 and data not shown). Combined, these results suggest that oligomerization of the intermediate domain of RIP is sufficient to activate the IKK complex, and this activation requires IKK␥.

FIG. 2. Enforced oligomerization of the intermediate domain of RIP activates the IKK complex and NF-B.
A, IKK␥ mediates the assembly of the RIP/IKK complexes. 293T cells were transfected with expression constructs for T7-IKK␤, T7-IKK␥, and FLAG-RIP as indicated. 24 h after transfection, cells were lysed, and the lysates were immunoprecipitated (IP) with anti-FLAG antibody. The immunoprecipitates were immunoblotted (IB) with anti-T7 antibody. Expression of T7-IKK␤, T7-IKK␥ or FLAG-RIP was determined by immunoblotting with anti-T7 or anti-FLAG antibodies, respectively. B, schematic representation of the RIP ID-FKBP12 chimera used in the oligomerization experiments. C, RIP ID-FKBP12 chimera activates NF-B in response to drug-induced oligomerization. 293T cells were transfected with 5XB-luciferase reporter together with "empty" vectors or the RIP ID-FKBP12 expression construct as indicated. 24 h after transfection, cells were either left untreated or incubated with either TNF␣ for 5 h or with AP1510 for 6 h as indicated. Cells were then collected and lysed, and the luciferase activity in the cell lysates was determined. pRSC-LacZ was included in all transfection reactions to normalize the transfection efficiency. Mean values Ϯ S.E. are shown from three independent experiments performed in duplicate. D, the RIP ID-FKBP12 chimera activates IKK␤ in response to drug-induced oligomerization. HeLa cells were transfected with FLAG-IKK␤ expression construct together with either empty vectors or the T7-RIP ID-FKBP12 expression construct. 24 h after transfection, cells were treated as above except that TNF␣ treatment was for 15 min, and AP1510 treatment was for 30 min and then lysed. The lysates were immunoprecipitated with anti-FLAG antibody, and the IKK activity associated with FLAG-IKK␤ was determined by immune complex kinase assay (KA). Expression of FLAG-IKK␤ and the T7-RIP ID-FKBP12 chimera was determined by immunoblotting with anti-FLAG or anti-T7 antibodies, respectively. KI, kinase-inactive. E, IKK␥ is required for RIP ID-induced NF-B activation. Rat-1 cells or the IKK␥-deficient Rat-1 derivative (5R) cells were transfected with 5XB-luciferase reporter together with either empty vector or the RIP ID-FKBP12 expression construct. 24 h after transfection, cells were either left untreated or incubated with AP1510 for 6 h. The luciferase activity in the transfected cell lysates was assayed and normalized as in C.

FIG. 3. NF-B and IKK inducing activities of the truncated IKK␥-FKBP12 chimeras.
A, IKK␥ interacts with IKK␤ via its N terminus. 293T cells were transfected with T7-IKK␤ and different FLAG-tagged full-length, N-terminally or C-terminally truncated IKK␥ expression constructs. 24 h after transfection, cells were lysed, and the lysates were immunoprecipitated (IP) with anti-FLAG antibody. The immunoprecipitates were immunoblotted (IB) with anti-T7 antibody. The expression of T7-IKK␤ and the different FLAG-IKK␥ truncated mutants was determined by immunoblotting with anti-T7 or anti-FLAG antibodies, respectively. B, schematic representation of IKK␥-FKBP12 chimeras used in the oligomerization experiments. C, IKK␥-FKBP12 chimeras activate NF-B in response to drug-induced oligomerization. 293T cells were transfected with 5XB-luciferase reporter together with empty vector or the indicated IKK␥-FKBP12 expression constructs. 24 h after transfection, cells were either left untreated or incubated with either TNF␣ for 5 h or with AP1510 for 6 h as indicated. The luciferase activity in the transfected cell lysates was assayed and normalized as described in the legend to Fig. 1C. D, IKK␥-FKBP12 chimeras activate IKK␤ in response to drug-induced oligomerization. HeLa cells were transfected with FLAG-IKK␤ expression construct together with either empty vector or the indicated T7-IKK␥-FKBP12 expression constructs. 24 h after transfection, cells were treated as above, except that TNF␣ treatment was for 15 min and AP1510 treatment was for 30 min and then lysed. The lysates were immunoprecipitated with anti-FLAG antibody, and the IKK activity Enforced Oligomerization of the N-terminal Half of IKK␥ Induces NF-B Activation-Several studies have highlighted the crucial importance of IKK␥ in the activation of NF-B in response to several stimuli (12,16,17). In addition, the use of truncated mutants indicates that the C-terminal region of IKK␥ seems to be necessary for the recruitment of this protein to upstream activators (13). Tax and RIP, two potent activators of NF-B, have been shown to bind to a region located in the C-terminal half of IKK␥ (36,37). IKK␥ is known to form a stable complex with IKK␣ and IKK␤ in vivo (12)(13)(14)(15)42). To determine precisely which regions of IKK␥ interact with IKK␣ and IKK␤, we transfected 293T cells with expression constructs encoding T7-tagged IKK␣ or IKK␤ and several FLAGtagged full-length or truncated IKK␥ constructs. Western blot analysis of immunoprecipitates from the transfected cells revealed that full-length or C-terminally truncated IKK␥ proteins that contain at least the first 200 amino acids were able to bind to IKK␤ (Fig. 3A). No association was observed between IKK␤ and the truncated mutants lacking the first 200 amino acids. Similar observations were obtained when the interactions were performed with IKK␣ (data not shown). We conclude that it is the N-terminal domain of IKK␥ that interacts with IKK␣ and IKK␤.
Our observations suggest that IKK␥ is an adaptor molecule, which links upstream regulators such as RIP to the IB kinases ( Fig. 2A). By using its C-terminal region to interact with the upstream regulators and its N-terminal region to associate with the IB kinases, IKK␥ can transmit the NF-B activation signals from the upstream regulators to the effector kinases. One way by which the activation signals are transmitted is through oligomerization of the upstream regulators (43). Indeed, we have shown that artificially induced oligomerization of the isolated central domain of RIP, which interacts with IKK␥, is sufficient to activate NF-B independent of the kinase and the death domains of RIP (Fig. 2C). Because the central domain of RIP interacts with the C-terminal half of IKK␥, we hypothesized that, upon TNF stimulation, RIP recruits and oligomerizes IKK␥, which in turn passes the oligomerization signal from RIP to the effector kinases, resulting in their activation. If this hypothesis is correct, then enforced oligomerization of the IKK␥ N-terminal region, which links IKK␥ to the IB kinases, should activate these kinases and induce NF-B, independent of upstream signals. To test this hypothesis, we fused full-length and several truncated IKK␥ to a 3-fold repeat of the FKBP12 polypeptide (Fig. 3B). Transient tranfection of these constructs into 293T cells resulted in expression of proteins of the expected size at similar levels (Fig. 3D). In the absence of AP1510, none of the transfected chimeric proteins induced significant NF-B activation as measured by the LUC reporter assay (Fig. 3C). However, all chimeric proteins with the exception of IKK␥ (1-105)-FKBP12 or IKK␥ (200 -419)-FKBP12, which do not interact with IKK␣/␤ (Fig. 3A), induced  (Fig. 3D). This effect was similar or superior in magnitude to that obtained after stimulation of cells with TNF␣. These experiments show that the first 200 residues of IKK␥ constitute the minimal sequence for activation of NF-B. No NF-B activation was detected when the FKBP12 constructs were co-transfected with kinase-inactive IKK␤ (data not shown) nor after treatment of empty vectortransfected 293T cells with AP1510 (Fig. 3C). These observations suggest that oligomerization of the N-terminal domain of IKK␥ activates the associated IKK␣ and IKK␤, which indicates that oligomerization of the kinases themselves may induces their kinase activity (see below).
To rule out the possibility that the IKK␥-FKBP12 chimeras function through interaction with the endogenous IKK␥ protein, we examined their ability to activate NF-B in the IKK␥deficient 5R cells. These cells have been stably transformed with Tax, which induces activation of the IB kinases by inter- FIG. 5. IKK␣ and IKK␤ are recruited to the IKK complex via their SCDs. a, the SCD of the IKK kinases mediates their association with IKK␥. 293T cells were transfected with FLAG-IKK␥ and either T7-tagged full-length or SCD-truncated (⌬SCD) IKK␣ or IKK␤ or SCD domain of IKK␣ or IKK␤ expression constructs as indicated. ⌬SCD constructs contain amino acids 1-640 of IKK␣ and amino acids 1-643 of IKK␤; SCD constructs contain the last 108 or 113 amino acids of IKK␣ and IKK␤, respectively. 24 h after transfection, cells were lysed, and the lysates were immunoprecipitated (IP) and immunoblotted (IB) as described in the legend to Fig. 2A acting with the C-terminal domain of IKK␥ (12,44,45). As expected, transfection of these cells with the IKK␥ (1-200)-FKBP12 chimera, which does not interact with Tax, did not activate NF-B in the absence of AP1510 (Fig. 3E). As observed with 293T cells, however, incubation of the IKK␥ (1-200)-FKBP12 transfected cells with the AP1510 oligomerizer resulted in a significant increase in NF-B activity (Fig. 3E). These results confirm that enforced oligomerization of the ectopicaly expressed IKK␥-FKBP12 chimeric proteins are responsible for the NF-B activation observed in 293T cells.
Physiological Oligomerization of the IKK Complex Can Also Induce NF-B Activation-To test the IKK oligomerization model in a more physiological context, we decided to use a physiological stimulus to induce oligomerization of the IKK complex directly without passing through the adaptor molecules TRADD, TRAF2, and RIP. To do that we linked the IKK complex directly to DR4 (TRAIL-R1), a member of the TNF-R family that can be trimerized by its physiological ligand TRAIL (46 -48) by replacing the cytoplasmic death domain (DD) of DR4 with the first 200 residues of IKK␥. To test the ability of the DR4⌬DD-IKK␥ (1-200) chimera to activate the IB kinases in response to physiological stimulation with TRAIL, we transfected it into the IKK␥-deficient 5R cells and then incubated the cells with or without TRAIL. In the absence of TRAIL, no NF-B activation was elicited after expression of the DR4⌬DD-IKK␥ (1-200) chimera in 5R cells (Fig. 4A). However, incubation of cells expressing the chimera, but not cells expressing IKK␥  or DR4⌬DD, or cells transfected with an empty vector, with TRAIL resulted in a remarkable increase in NF-B, as well as IKK activities (Fig. 4). These results demonstrate that physiological trimerization of the N-terminal half of IKK␥, following death receptor ligation, is sufficient to fully activate the IKK complex.
IKK␥ Interacts with the C-terminal Serine Cluster Domain of IKK␣ and IKK␤-To understand how IKK␥ regulates the activity of the IB kinases, it is important to know which region of these kinases interact with IKK␥. IKK␣ and IKK␤ have very similar primary structures with an N-terminal protein kinase domain, a central LZ domain, and a C-terminal region containing a HLH motif and a SCD (see Fig. 7a). Because the Cterminal region of IKK␤ has been recently shown to play a regulatory role in the function of the enzyme (40), we decided to delete the SCD portion of this region and then determine whether the truncated kinases can still interact with IKK␥. To this end, FLAG-tagged IKK␥ was transiently co-expressed in 293T cells together with either T7-tagged full-length or Cterminally truncated IKK␣ or IKK␤. Western blot analysis of immunoprecipitates from the transfected cells revealed that only the full-length but not the C-terminally truncated IKK␣ or IKK␤ associated specifically with IKK␥ (Fig. 5a). Interestingly, the last 108-or 113-amino acid region of IKK␣ or IKK␤, respectively, which contains the SCD, was also able to interact with IKK␥, suggesting that this region is sufficient for interaction of IKK␥ with the two kinases.
To verify this observation in a more physiological context, we constructed GFP-tagged SCDs of IKK␣ and IKK␤ and examined them for possible dominant negative inhibitory activity. Expression of the GFP-tagged SCDs drastically inhibited the activation of NF-B by TNF␣ in a dose-dependent manner (Fig.  5b). Moreover, NF-B activation induced by enforced oligomerization of the IKK␥ (1-300)-FKBP12 and the RIP ID-FKBP12 chimeras was also blocked by the GFP-tagged SCDs of IKK␣ or IKK␤ (Fig. 5b). Combined, these results show that the SCDs of IKK␣ and IKK␤ play an important role in mediating the interaction of these kinases with IKK␥ and in transmitting the upstream activation signals to these kinases.
Combined with previous observations, our results suggest that RIP associates with IKK␥ and mediates its oligomerization. This in turn leads to oligomerization and activation of the associated IB kinases. The use of IKK␥-FKBP12 chimeras bypasses the TNF signaling pathway, because drug-induced oligomerization of the N-terminal part of IKK␥ results in the activation of the IKK kinases, probably via recruitment and oligomerization of their SCDs. To monitor this process in vivo, we used the GFP-tagged SCDs of IKK␣ and IKK␤ as reporters. Transient expression of the GFP-tagged SCDs revealed a uniform cytoplasmic localization for both chimeras (Fig. 6a). Cotransfection with the IKK␥ (1-300)-FKBP12 chimeric protein in the absence of AP1510 did not alter this distribution (Fig.  6c). However, incubation of the cells expressing the GFPtagged SCDs and IKK␥ (1-300)-FKBP12 with AP1510 resulted in a dramatic alteration of fluorescence from uniform to punctate fluorescence (Fig. 6d). The punctate fluorescence was not observed when cells expressing the GFP-tagged SCDs were treated with AP1510 in the absence of IKK␥ (1-300)-FKBP12 (Fig. 6b). These results represent a clear in vivo demonstration of the ability of the N-terminal part of IKK␥ to induce oligomerization of the IB kinases through engagement of their SCDs. Finally, the punctate fluorescence was also observed when cells expressing a GFP-tagged IKK␥ (1-200)-FKBP12 chimera were incubated with AP1510 but not in the absence of the drug (Fig.  6, a and b), demonstrating that AP1510 indeed induces oligomerization of the IKK␥-FKBP chimeras in vivo.
Enforced Oligomerization of SCD-truncated IKK␣ or IKK␤ Induces NF-B Activation-Following confirmation of the IKK␥ interaction with the SCD of IKK␣ and IKK␤, we proceeded to test directly whether SCD-truncated IKK␣ and IKK␤ fused to the FKBP12 oligomerization domain can be activated by oligomerization. Different C-terminally truncated IKK␣ or IKK␤ modules were fused to the inducible FKBP12 cassette and transfected in 293T cells (Fig. 7a). Treatment of cells expressing full-length or SCD-truncated IKK␣ and IKK␤ fused to FKBP12 with AP1510 resulted in large increases in NF-B activity (Fig. 7b). This, however, was not observed in cells transfected with constructs encoding HLH-SCD-truncated FKBP12 or LZ-HLH-SCD-truncated FKPB12 fusion proteins, as well as in cells transfected with a construct encoding kinaseinactive-FKBP12 fusion protein ( Fig. 7b and data not shown). Consistently, immunoprecipitation of the T7-tagged FKBP12 constructs after treatment of the expressing cells with AP1510 resulted in the precipitation of a high IB kinase activity for full-length and SCD truncated IKK␣ or IKK␤ (Fig. 7c). These results confirm that direct oligomerization of IKK␣ or IKK␤ induces their activation.
The above data also provide supporting evidence for the importance of the HLH motif in the kinase activity of both IKK␣ and IKK␤, because deletion of this motif prevents oligomerization-induced kinase activation. Recently, it has been shown that co-expression of an isolated C-terminal fragment containing the HLH and SCD motifs together with a C-terminally truncated IKK␤ lacking this portion restores the ability to the C-terminally truncated IKK␤ to be activated by TNF␣ (40). To test whether an isolated HLH-SCD-FKBP12 chimera can activate the C-terminally truncated IKK␤ in trans after enforced oligomerization with the AP1510 drug, we transfected constructs expressing these proteins into HeLa cells and then treated the cells with TNF␣ or AP1510. Interestingly, both TNF␣ and AP1510 were able to induce the kinase activity of the truncated IKK␤ in the presence but not the absence of the T7-IKK␤ (HLH-SCD)-FKBP12 chimera (Fig. 7d). These observations suggest that the HLH-SCD motif can activate the kinase domain in trans after it receives the activation/oligomer-  Fig. 2C. KI, kinase-inactive. Insets represent immunoblots of the expressed proteins. c, ⌬SCD-IKK␣/␤-FKBP12 chimeras phosphorylate IB␣ in response to drug-induced oligomerization. HeLa cells were transfected with the indicated T7-tagged expression constructs. 24 h after transfection, cells were either left untreated or incubated with TNF␣ for 15 min or with AP1510 for 30 min as indicated. Cells were lysed, the lysates were immunoprecipitated (IP) with anti-T7 antibody, and the IKK activity was determined by immune complex kinase assay (KA). Expression of the different T7-IKK␣/␤ constructs was determined by immunoblotting (IB) with anti-T7 antibody. d, the C-terminal region of IKK␤ can activate in trans its kinase domain upon drug-induced oligomerization. HeLa cells were transfected with the indicated expression constructs. T7-IKK␤ (HLH-SCD)-FKBP12 construct contains amino acids 559 -756 of IKK␤. 24 h after transfection, cells were either left untreated or incubated with either TNF␣ ization signal from an upstream regulator such as TNF␣. However, the requirement for TNF␣ can be bypassed if the HLH-SCD motif is linked physically to the FKBP12 oligomerization motif. In this case, the activation/oligomerization signal can be relayed directly to the kinase domain by AP1510 independent of the upstream regulators. Moreover, the activation signal must pass through the SCD motif to activate the kinase domain. This is because in the absence of the SCD, the remaining portion of IKK␣ or IKK␤ cannot be activated by upstream signals such as those triggered by TNF␣ (Fig. 7c, right panel). However, AP1510 can activate the SCD-truncated kinases if they are physically linked to the FKBP oligomerization motif (Fig. 7, b and c).
Phosphorylation of the T-loop Is Required for Oligomerization-induced Activation of IKK␣ or IKK␤-Recent studies showed that phosphorylation of two specific serines (Ser 177 and Ser 181 ) in the T-loop of IKK␤ is required for its activation. Mutation of these two serines to alanine residues prevents IKK activation by TNF␣, whereas their replacement with phosphomimetic glutamate residues causes constitutive activation of IKK␤ (8,40). Because oligomerization of the IKKs triggers their activation, it is possible that oligomerization induces phosphorylation of the T-loop serines resulting in full activation of the kinases. If this hypothesis is correct, then substitution of Ser 177 and Ser 181 of IKK␤ or Ser 176 and Ser 180y of IKK␣ in our SCD truncated FKBP12 IKK␣ or IKK␤ constructs with alanine residues should prevent oligomerization-induced NF-B activation. As expected, no NF-B activation was detected after treatment of the cells expressing the T-loop Ser to Ala mutants with AP1510 (Fig. 7e). Because kinase-inactive IKK␣-or IKK␤-FKBP12 with wild type T-loop also failed to be activated following enforced oligomerization (Fig. 7b), these results strongly suggest that oligomerization-induced autophosphorylation of the T-loop serines causes activation of the IB kinases.

DISCUSSION
The cellular response to TNF␣ depends on cell type and the presence of specific signaling molecules. Engagement of the TNF receptors, TNF-R1 and TNF-R2, leads to the activation of two competing pathways: a pro-apoptotic pathway and an antiapoptotic pathway (49). The anti-apoptotic pathway, which involves the activation of NF-B via the IKK complex, is known to be initiated by recruitment of adaptor and effector molecules like TRADD, RIP, and TRAF proteins on the trimerized TNF receptors. Very recently RIP has been shown to bind to IKK␥ in a stimulus-dependent manner (37). However, its is not yet clear how interaction of RIP with IKK␥ could result in activation of the IKK complex.
A potential model that could explain how the IKK complex is activated after TNF-R1 ligation is activation by oligomerization. In support of this model, we have shown that following TNF␣ stimulation, the IKK complex undergoes a structural change that leads to a significant increase in its molecular mass, as observed using gel filtration. Interestingly, only the highest molecular weight complexes possessed IKK activity. This molecular mass increase could not be attributed to a stimulus-dependent, stable association of the IKK complex with the TNF receptor components, because these eluted in later fractions. Moreover, although a small amount of the IKK complex co-eluted with the activated TNF-R1 components in some fractions, no IKK activity was associated with these fractions. A previous study also reported that the activity of the IKK kinases found in the TNF-R1 complex after immunoprecipitation with a TNF-R1-specific antibody was significantly less than that found in the IKK complex after direct immunoprecipitation with an IKK␥ specific antibody after TNF stimulation (37), suggesting that the released and oligomerized IKK complex is the active form of IKK. Thus, prior to TNF stimulation, the IKK components exist as a ϳ650-kDa heteromeric for 15 min or AP1510 for 30 min as indicated. Cells were lysed, and the lysates were immunoprecipitated with anti-FLAG antibody, and the IKK activity was determined by immune complex kinase assay. Expression of the different T7-or FLAG-IKK␤ constructs was determined by immunoblotting with anti-T7 or anti-FLAG antibodies, respectively. e, mutation of the T-loop serines inhibits oligomerization-induced activation of IKK␣/␤. The SCD of IKK␣ and IKK␤ was replaced with an inducible FKBP12 oligomerization cassette to generate the wild type (⌬SCD) IKK␣-FKBP12 and IKK␤-FKBP12 constructs. The T-loop serines were mutated to alanines to generate the SS-AA mutant (⌬SCD AA) IKK␣-FKBP12 and (⌬SCD AA) IKK␤-FKBP12 constructs. 293T cells were transfected with the 5XB-luciferase reporter and the indicated IKK␣-FKBP12 or IKK␤-FKBP12 constructs. 24 h after transfection, cells were either left untreated or incubated with AP1510 for 6 h as indicated. The luciferase activity in the transfected cell lysates was assayed and normalized as described in the legend to Fig. 2C. complex of IKK␥, IKK␤, and IKK␣. Following TNF stimulation, TNF-R1-bound RIP recruits this IKK complex through interaction with IKK␥ and induces its oligomerization. This oligomerization results in activation of the IB kinases. The activated, high molecular mass ϳ1.3 MDaЈ IKK complex is then released from the TNF receptor and is able to phosphorylate the IB proteins.
Because IKK␥ has been shown to interact with the intermediate domain of RIP (37) (36). Based on these observations, the N-terminal half of IKK␥ constitutes the effector domain of this molecule, whereas the Cterminal domain is likely to function as an oligomerization domain whose major role, after connection to upstream activator like RIP, is to induce clustering-dependent activation of IKKs (Fig. 8). This hypothesis was tested using several Cterminally truncated IKK␥ fused to a 3-fold repeat of the oligomerization cassette FKBP12. The first 200 N-terminal residues constituted the minimal sequence required for IKK and NF-B activation following oligomerization with AP1510. Thus, clustering of the N-terminal effector domain of IKK␥ is sufficient for activation of the downstream IKKs.
Because AP1510 may induce extensive oligomerization of the IKK complex, which might be taken as a nonphysiological stimulus, it was therefore important to validate our observations, to examine the effect of a more "physiological" oligomerization of IKK␥. For this purpose, we constructed a DR4⌬DD-IKK␥ (1-200) chimera, where the cytoplasmic DD of DR4 (TRAIL-R1) was replaced with the first 200 residues of IKK␥, and tested its ability to induce IKK and NF-B activation following TRAIL stimulation. Like TNF␣, the biologically active form of TRAIL is a homotrimer as suggested by its crystal structure (46). It is therefore reasonable to assume that trimerization of DR4 by its ligand constitutes a physiological transmembrane signaling. Interestingly, ligation of the DR4⌬DD-IKK␥ (1-200) chimera induced NF-B as well as IKK activation in the IKK␥-deficient 5R cells (Fig. 4). Importantly, this activation was comparable or superior in magnitude to the one obtained using the IKK␥-FKBP12 chimeras. This result shows that regulated clustering of the N-terminal effector domain of IKK␥ is sufficient for activation of the downstream IKKs and that the upstream activation signal must be relayed to the IKKs through the N terminus of IKK␥.
Previous studies showed that IKK␥ is associated with IKK␣ and IKK␤ in unstimulated cells (12)(13)(14)(15)42). Our data show that IKK␣ and IKK␤ interact with IKK␥ via their SCDs. Because IKK␥ may oligomerize by binding to RIP, this interaction induces activation of the kinases perhaps via oligomerization.
Interestingly, direct oligomerization of either IKK␣ or IKK␤ by SCD deletion mutants fused to the inducible FKBP12 oligomerization cassette activated both kinases. Based on this we postulate that the function of IKK␥ is to transmit the upstream oligomerization signal to the IKKs through its interaction with their SCDs. Given that IKK␥ is physically associated with the SCD of IKKs, any stimulus that could induce oligomerization of IKK␥ would also lead to oligomerization of the IKKs themselves. Consistent with this, we found that the SCD of IKK␣ or IKK␤ are potent dominant negative inhibitors of TNF␣-and RIP-induced NF-B activation, perhaps because of their ability to block the sites of interactions between IKK␥ and the IKKs. Moreover, SCD-truncated IKK␣ or IKK␤ were unable to be activated following TNF stimulation, demonstrating that the SCD plays an essential role in transmitting the activation signal to the kinases.
Activation of the IKK complex depends on phosphorylation of IKK␣/␤, because treatment of the purified activated IKK complex with the protein phosphatase 2A results in its inactivation (9). Phospho-peptide mapping of IKK␤ revealed that phosphorylation occurs at serines 177 and 181 located in the T-loop and in the SCD (40). In the case of IKK␤, conversion of the two phosphoacceptors Ser 177 and Ser 181 to alanines prevents its activation, whereas conversion to phospho-mimetic glutamate residues causes its constitutive activation (40), suggesting that phosphorylation of these serines is essential for activation. However, it is not known whether this phosphorylation is due to the action of an upstream kinase or to autophosphorylation. Three kinases, NIK, MEKK1, and Akt (19 -22, 50, 51), have been proposed to regulates NF-B activation. However, very recent observations raise doubts about their involvement in the physiological regulation of IKK activation (40,52,53). Because direct oligomerization of the SCD-truncated IKKs in the absence of any other stimuli induces activation of these kinases, we believe that oligomerization triggers autophosphorylation of the T-loop serines. This process could be explained by the proximity model of activation. IKKs have low basal kinase activity in the unphosphorylated ϳ650-kDa IKK complex. Oligomerization could change the conformation of IKKs bringing the T-loop serines of one precursor molecule in close proximity to the kinase active site of an adjacent molecule resulting in autophosphorylation of the T-loop in trans (Fig. 8). Once the T-loop serines are phosphorylated, the IKKs become fully active. In support of this model, we demonstrated that mutations of the T-loop serines to alanine inhibit oligomerization-induced activation of the IKKs. Furthermore, overexpression of native IKK␣ or IKK␤ in baculovirus-infected Sf9 cells has been shown to lead to their activation in the absence of extracellular stimuli, and this activity was abolished when mutations preventing LZ-mediated dimerization were introduced (11,24). This result can be explained by postulating that overexpression of IKK␣ or IKK␤ results in their oligomerization and autophosphorylation, thereby mimicking IKK␥-induced oligomerization after cytokine treatment. In our case, direct oligomerization of the IKK kinases bypasses the initial steps in TNF␣ signaling, resulting in IKK autophosphorylation, activation, and induction of NF-B.
Although we have provided strong evidence that oligomerization of the IKK complex plays an important role in its activation, we cannot ignore the possibility that other mechanisms of activation might also be important. Direct phosphorylation of IKK␣/␤ by other kinases such NIK, MEKK1, or Akt could also activate the IKK complex, providing a second level of regulation to this important survival pathway.