Roles for Homotypic Interactions and Transautophosphorylation in IκB Kinase (IKKβ) Activation*

The nuclear factor κB (NF-κB)/Rel family of transcription factors participates in a wide range of biological activities including inflammation, immunity, and apoptosis. NF-κB is kept inactive in the cytoplasm in unstimulated cells by virtue of the masking of its nuclear localization sequence by bound IκB protein. Cellular stimuli trigger the destruction of IκB proteins and the liberation of NF-κB to enter the nucleus and activate gene expression. A multisubunit IκB kinase complex (IKK) phosphorylates IκB proteins and mediates the activation of NF-κB by proinflammatory stimuli such as tumor necrosis factor α. Phosphorylation of IκB proteins triggers their polyubiquitination and their subsequent recognition and degradation by the proteasome. The IKK complex contains two catalytic subunits, IKKα and IKKβ, and a noncatalytic subunit, NF-κB essential modifier/IKKγ. IKK activation depends upon the phosphorylation of residues in the activation loop of IKKβ and the subsequent activation of IKKβ kinase activity. However, the events contributing to IKKβ phosphorylation are not well understood. Here, we present evidence that the activation of IKKβ depends on its ability to form homotypic interactions and to transautophosphorylate. We find that an intact leucine zipper in IKKβ is necessary for homotypic interactions, kinase activation, and phosphorylation on its activation loop. Enforced oligomerization of an IKKβ mutant defective in forming homotypic interactions restores kinase activation. Homotypic interactions allow IKKβ molecules to transautophosphorylate one another on their activation loops. Finally, the oligomerization of IKKβ is stimulated by tumor necrosis factor α in cultured cells. Our findings support a model whereby ligand-induced homotypic interactions between IKKβ molecules result in IKKβ phosphorylation and consequently IKK activation.


The nuclear factor B (NF-B)/Rel family of transcription factors participates in a wide range of biological activities including inflammation, immunity, and apoptosis. NF-B is kept inactive in the cytoplasm in unstimulated cells by virtue of the masking of its nuclear localization sequence by bound IB protein. Cellular stimuli trigger the destruction of IB proteins and the liberation of NF-B to enter the nucleus and activate gene expression. A multisubunit IB kinase complex (IKK) phosphorylates IB proteins and mediates the activation of NF-B by proinflammatory stimuli such as tumor necrosis factor ␣. Phosphorylation of IB proteins triggers their polyubiquitination and their subsequent recognition and degradation by the proteasome.
The IKK complex contains two catalytic subunits, IKK␣ and IKK␤, and a noncatalytic subunit, NF-B essential modifier/IKK␥. IKK activation depends upon the phosphorylation of residues in the activation loop of IKK␤ and the subsequent activation of IKK␤ kinase activity. However, the events contributing to IKK␤ phosphorylation are not well understood. Here, we present evidence that the activation of IKK␤ depends on its ability to form homotypic interactions and to transautophosphorylate. We find that an intact leucine zipper in IKK␤ is necessary for homotypic interactions, kinase activation, and phosphorylation on its activation loop. Enforced oligomerization of an IKK␤ mutant defective in forming homotypic interactions restores kinase activation. Homotypic interactions allow IKK␤ molecules to transautophosphorylate one another on their activation loops. Finally, the oligomerization of IKK␤ is stimulated by tumor necrosis factor ␣ in cultured cells. Our findings support a model whereby ligand-induced homotypic interactions between IKK␤ molecules result in IKK␤ phosphorylation and consequently IKK activation.
The IB kinase complex (IKK) 1 functions as a mediator of NF-B activation in response to multiple stimuli by phospho-rylating IB inhibitor proteins and causing their degradation (1,2). The IKK complex contains three distinct subunits, namely IKK␣, IKK␤, and NEMO (or IKK␥) (1,2). Subunit reconstitution experiments in yeast and mammalian cells suggest that IKK is composed of a NEMO homodimer bound together with either an IKK␣/IKK␤ heterodimer or an IKK␤ homodimer (3). Several lines of evidence have indicated that IKK␣ and IKK␤ have distinct biochemical functions. Gene disruption experiments in mice have demonstrated that IKK␤ plays an essential role in NF-B activation in response to most stimuli, whereas IKK␣ seems to be largely dispensable (1). Also, IKK␤ and IKK␣ have been shown to have different substrate preferences in vitro. IKK␤ has been shown to have a higher specific activity for phosphorylating IB␣, whereas IKK␣ has a preference for using the NF-B2 (p100) precursor as a substrate (4).
TNF-␣ and interleukin-1 stimulate the phosphorylation of IKK␤ on two serine residues, Ser 179 and Ser 181 , found in its activation loop. Phosphorylation on both of these sites is required for IKK activation (5). However, the regulatory mechanisms governing this phosphorylation event are not understood. It has been proposed that the activation of IKK depends upon its recruitment to the TNF-␣ receptor signaling complex upon ligand binding (6,7). Although it seems that only a small fraction of total IKK is recruited to the membrane, activation of this fraction of IKK may be sufficient to stimulate the activation of other remaining inactive IKK complexes through transautophosphorylation events.
One possible mechanism whereby IKK may initially be activated is through the activity of an IKK␤ kinase. Although many candidate IKK␤ kinases (based on overexpression transient transfection studies) have been proposed in the literature, only a few of these kinases have been shown genetically in cultured cells or animal models to be important for IKK activation. Two of these IKK␤ kinases, MEKK3 and PKC, have been implicated (based upon gene disruption experiments) to be upstream regulators of IKK (8,9). MEKK3 was shown to be required for IKK activation in response to TNF-␣ and interleukin-1 signaling in mouse embryonic fibroblasts (8). In contrast, PKC was found to be important in mice for IKK activation in response to TNF-␣ in the mouse lung but not in mouse embryonic fibroblasts (9). Recently, TAK1 was found to be important for IKK activation in HeLa cells in experiments using small interfering RNAs (10). It seems likely that different cell types may rely on distinct IKK␤ kinases to activate IKK.
Another possible mechanism whereby the initial pool of IKK may be activated is through an induced proximity mechanism. In such a scenario, the oligomerization of an IKK activator and interactor such as receptor interacting protein may induce the proximity of IKK complexes and facilitate mutual transautophosphorylation (11,12). Of note, we and others have shown * This work was supported by grants from the National Institutes of Health and the Walther Cancer Institute (to K. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ § To whom correspondence may be addressed. E-mail: kunliang@ umich.edu. 1 The abbreviations used are: IKK, IB kinase; TNF-␣, tumor necrosis factor ␣; NF-B, nuclear factor B; IB, inhibitor of NF-B; MEKK3, mitogen-activated protein kinase kinase 3; TAK1, transforming growth factor ␤-activated kinase 1; PKC, zeta isoform of protein kinase C; GST, glutathione S-transferase; HA, hemaglutinin; NEMO, NF-B essential modifier. that enforced oligomerization of IKK subunits including IKK␤ by artificial chemical agents can lead to enhanced IKK and NF-B activity (11,12). Although it has also been shown that the overexpression of active IKK␤ leads to its autophosphorylation, the ability of IKK␤ molecules to transautophosphorylate serines on its activation loop has not been yet demonstrated. Furthermore, the oligomerization of any IKK subunits in vivo in response to stimuli has also not been shown.
In this report we provide evidence that IKK␤ activation requires the ability of IKK␤ molecules to oligomerize and to phosphorylate one another on their activation loops. An intact leucine zipper in IKK␤ is required for homotypic interactions, kinase activation, and activation loop phosphorylation. Furthermore, enforced oligomerization of an IKK␤ mutant defective in making homotypic interactions restores kinase activation. Homotypic interactions permit IKK␤ molecules to transphosphorylate one another in vivo on their activation loops. Last of all, TNF-␣ stimulates the homotypic interactions between IKK␤ proteins in cultured cells. Our results suggest a model whereby ligand-induced IKK␤ homotypic interactions lead to IKK␤ transautophosphorylation and consequently IKK␤ activation.

EXPERIMENTAL PROCEDURES
Cell Culture-Human embryonic kidney 293T cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum. All cells were grown in a 37°C incubator with 5% CO 2 . Transfections were performed with LipofectAMINE (Invitrogen), according to the manufacturer's instructions.
Immunoprecipitations-Transiently transfected 293T cells were lysed in IP lysis buffer (50 mM Tris-HCl, pH 7.6, 250 mM NaCl, 1% Triton X-100, 0.5% Nonidet P-40, 3 mM EGTA, 3 mM EDTA, 10% glycerol, 0.1 mM Na 3 VO 4 , 10 M N-ethylmaleimide) supplemented with a mixture of protease inhibitors (Roche Applied Science). Lysates were incubated with antibody and protein G-Sepharose, and immunoprecipitates were washed three times with IP lysis buffer. All immunoprecipitates were boiled in SDS-PAGE sample buffer and resolved on a 7.5% polyacrylamide gel. Proteins were transferred to polyvinylidene difluoride membrane and probed with FLAG M2, phospho-IKK␤ (Cell Signaling), or phospho-IB␣ antibody (Cell Signaling). To minimize the detection of the IgG heavy chain, protein A conjugated to horseradish peroxidase (Zymed Laboratories Inc.) was used as a secondary antibody for anti-NEMO immunoblots.
For coimmunoprecipitation assays, cells were lysed in EBC150 (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 50 mM NaF, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol) supplemented with a mixture of protease inhibitors (Roche Applied Science), and lysates were incubated with the appropriate antibody for 2 h at 4°C and with protein G-Sepharose for an additional 1 h. All immunoprecipitates were boiled in SDS-PAGE sample buffer and resolved on a 7.5% polyacrylamide gel.

RESULTS AND DISCUSSION
It has been observed previously that recombinant IKK␤ purified from mammalian or insect cells is an active kinase (13)(14)(15)(16)(17). This activity has been attributed to phosphorylation of Ser 179 /Ser 181 in the activation loop because a purified IKK␤ mutant with alanine substitutions at Ser 179 /Ser 181 is inactive when purified under the same conditions (5,13). This result could be attributed to the endogenous activity of an IKK␤ kinase, autoactivation by IKK␤ molecules, or a combination of the two. To study the mechanism whereby IKK␤ autoactivation may proceed, we first sought to construct a version of IKK␤ that would not autoactivate. To achieve this goal, we chose to disrupt the leucine zipper motif in IKK␤. Previous reports have suggested that the leucine zipper of IKK␣ is important for homotypic interactions with itself and heterotypic interactions with IKK␤ (15). However, the leucine zipper of IKK␤ has yet to be characterized functionally. To generate a mutant of IKK␤ with a nonfunctional leucine zipper, we substituted the methionine residue at position 472 within the predicted leucine zipper motif of IKK␤ with a serine residue (M472S). This mutation rendered IKK␤ defective in binding to wild-type IKK␤ but not NEMO in coimmunoprecipitation experiments ( Fig. 1A and data not shown). Furthermore, this mutation significantly impaired the kinase activity of IKK␤. The M472S mutant demonstrated significantly reduced autophosphorylation activity and kinase activity toward recombinant IB␣ (Fig. 1B). These results suggest that autoactivation by IKK␤ in cells requires homotypic IKK␤ interactions.
To determine whether the M472S mutation affected the ability of IKK␤ to be phosphorylated on its activation loop, we immunoprecipitated epitope-tagged wild-type and mutant IKK␤ proteins from transfected cells and performed immunoblotting using an antibody that specifically recognizes Ser 181 of IKK␤. We compared wild-type IKK␤ with the M472S mutant and a mutant of IKK␤ with both Ser 179 and Ser 181 mutated to Ala (S179A/S181A). We found that in contrast to wild-type IKK␤, both the M472S and S179A/S181A mutants were not detectably phosphorylated on Ser 181 (Fig. 2). Thus oligomerization of IKK␤ seems to be required for the phosphorylation of residues in its activation loop, thereby leading to kinase activation. If the impaired kinase activity of the M472S mutant was caused by impaired phosphorylation of the activation loop, mimicking the phosphorylation of these residues in this mutant should be able to restore kinase activity. To mimic phosphorylation of IKK␤, we mutated the two serines in the activation loop of the M472S mutant to glutamic acid residues (M472S/S179E/S181E). Similar glutamic acid substitutions alone have been shown to yield a constitutively active kinase (17). We found that introduction of the glutamic acid substitutions restored the kinase activity of the M472S mutant, as wild-type IKK␤ (WT), IKK␤(M472S) (M472S), IKK␤(S179E/ S181E) (EE), or IKK␤(M472S/S179E/ S181E) (M472S/EE), as indicated. Lysates were prepared, and anti-FLAG immunoprecipitates (IP) were treated as in Fig. 1B. B, 293T cells transfected with HA-IB␣ and pcDNA3 (Ϫ), FLAG-tagged wild-type IKK␤, IKK␤(S179E/S181E), IKK␤(K44M/S179E/S181E) (K44M/EE), IKK␤(M472S), or IKK␤(M472S/S179E/ S181E), as indicated. Anti-HA immunoprecipitates were probed with phospho-IB␣ or HA antibody. Alternatively, lysates were probed with FLAG M2 antibody to detect levels of IKK␤ proteins (bottom). IB, immunoblot. assessed by in vitro kinase assays (Fig. 3A). We wanted to see whether similar results were also seen with IKK␤ kinase activity in vivo as well. Expression of the M472S mutant was able to elicit IB␣ phosphorylation in vivo only when glutamic acid substitutions were made in the activation loop (Fig. 3B). Thus, the M472S mutant is inactive because of its inability to stimulate phosphorylation in its activation loop.
Given that the M472S mutant is impaired both in oligomerization and kinase activity, a likely possibility is that oligomerization contributes to kinase activation. If the inactivity of the M472S mutant was caused primarily by the inability to oligomerize, enforced oligomerization of the M472S mutant should lead to its activation. To test this idea, we fused the M472S mutant to three tandem repeated oligomerization domains of Fpk (Fpk3). This protein tag can be induced to oligomerize by the cell-permeable artificial ligand AP1510 (18). We compared the kinase activity of Fpk3 fusion protein in the presence of AP1510 or vehicle treatment. For comparison, we also fused versions of the M472S mutant containing activation loop substitutions to alanine or glutamic acid residues. In the presence of the oligomerizer AP1510, the kinase activity of the M472S mutant was restored with respect to both IB␣ phosphorylation and autophosphorylation (Fig. 4). M472S mutants also containing activation loop substitutions to alanine or glutamic acid residues were inactive and constitutively active, respectively. Also, their activities were unaffected in the presence of oligomerizer, suggesting that the activating effect of AP1510 proceeded through phosphorylation of Ser 179 and Ser 181 (Fig. 4). Thus, IKK␤ oligomerization can result in IKK␤ phosphorylation and activation.
One can imagine that the oligomerization of IKK␤ might lead directly to IKK␤ activation through transautophosphory-lation or indirectly though an IKK␤-associated kinase. To examine whether transphosphorylation of IKK␤ molecules contributed to phosphorylation of the activation loop, we cotransfected the dead kinase K44M mutant along with different amounts of a constitutively active IKK␤/S179E/S181E mutant. Of note, the K44M mutant is not phosphorylated on Ser 181 and cannot autophosphorylate itself ( Fig. 1B and data not shown). After transfections, we immunoprecipitated the K44M mutant from cells and performed immunoblot analysis with phospho-IKK␤ antibody. We found that phosphorylation of Ser 181 in the K44M mutant was induced by IKK␤/S179E/ S181E. In contrast, a version of the K44M mutant carrying an M472S substitution as well could not be phosphorylated (Fig.  5). Thus, IKK␤ transphosphorylation contributes to IKK␤ activation loop phosphorylation.
To examine whether oligomerization of IKK␤ occurred in response to TNF-␣, we cotransfected two IKK␤ constructs with different epitope tags and performed coimmunoprecipitation assays after cell stimulation with TNF-␣ for various times. To perform this experiment, we had to use low levels of expression plasmids, as higher expression of IKK␤ led to constitutive oligomerization and activation. We found that TNF-␣ stimulation of cells resulted in the coimmunoprecipitation of IKK␤ proteins (Fig. 6). These data provide evidence that IKK␤ oligomerization can be induced by a physiological stimulus in vivo.
In this report, we provide evidence for a model of IKK activation whereby ligand-induced IKK␤ homotypic interactions lead to transautophosphorylation and consequently IKK activation. Although our results indicate that enforced oligomerization of an IKK␤ mutant defective in forming homotypic interactions is sufficient to restore activation, it is unclear from FIG. 6. TNF-␣ stimulates the oligomerization of IKK␤ in cultured cells. 293T cells were transfected with FLAG-IKK␤ and either pcDNA3(Ϫ) (vector), HA-IKK␤, or HA-IKK␣, as indicated. 24 h after transfection, cells were stimulated with TNF-␣ (10 ng/ml) for the various times indicated, and lysates were prepared. Anti-HA-immunoprecipitates were probed with FLAG to detect coimmunoprecipitated protein. Alternatively, immunoprecipitates (IP) were probed with HA antibody. Also, lysates were probed with FLAG antibody to detect total levels of FLAG-IKK␤ (bottom). IB, immunoblot; NS, not stimulated.
our experiments whether oligomerization of IKK␤ alone is sufficient to activate IKK␤. A model in which IKK␤ molecules autoactivate one another necessarily implies that an "initial pool" of active IKK␤ exists. One could imagine that this pool of active IKK␤ molecules might be derived from a preexisting fraction of IKK␤ in unstimulated cells. Alternatively, this pool of IKK␤ may be the product of the activity of an inducible kinase such as MEKK3, PKC, or TAK1 (8 -10). The biochemical mechanism underlying the phosphorylation of IKK␤ is likely to be complicated further by other events such as the phosphorylation and ubiquitination of NEMO (5, 19 -22). Another unresolved issue is whether oligomerization and phosphorylation of IKK␤ are separable or interrelated events. For instance, in addition to amplifying the pool of active IKK␤ molecules through transautophosphorylation, IKK␤ oligomerization may conceivably facilitate the recognition of inactive IKK␤ molecules by an IKK␤ kinase. Current investigations are focused on clarifying these and other important issues regarding the mechanism of IKK activation.