Complete Reconstitution of Human IκB Kinase (IKK) Complex in Yeast

The IκB kinase (IKK) complex, composed of two catalytic subunits (IKKα and IKKβ) and a regulatory subunit (IKKγ), is the key enzyme in activation of nuclear factor κB (NF-κB). To study the mechanism and structure of the complex, we wanted to recombinantly express IKK in a model organism that lacks IKK. For this purpose, we have recombinantly reconstituted all three subunits together in yeast and have found that it is biochemically similar to IKK isolated from human cells. We show that there is one regulatory subunit per kinase subunit. Thus, the core subunit composition of IKKα·β·γ complex is α1β1γ2, and the core subunit composition of IKKβ·γ is β2γ2. The activity of the IKK complex (α+β+γ or β+γ) expressed in yeast (which lack NF-κB and IKK) is 4–5-fold higher than an equivalent amount of IKK from nonstimulated HeLa cells. In the absence of IKKγ, IKKβ shows a level of activity similar to that of IKK from nonstimulated HeLa cells. Thus, IKKγ activates IKK complex in the absence of upstream stimuli. Deleting the γ binding domain of IKKβ or IKKα prevented IKKγ induced activation of IKK complex in yeast, but it did not prevent the incorporation of IKKγ into IKK and large complex formation. The possibility of IKK complex being under negative control in mammalian cells is discussed.

The IB kinase (IKK) complex, composed of two catalytic subunits (IKK␣ and IKK␤) and a regulatory subunit (IKK␥), is the key enzyme in activation of nuclear factor B (NF-B). To study the mechanism and structure of the complex, we wanted to recombinantly express IKK in a model organism that lacks IKK. For this purpose, we have recombinantly reconstituted all three subunits together in yeast and have found that it is biochemically similar to IKK isolated from human cells. We show that there is one regulatory subunit per kinase subunit. Thus, the core subunit composition of IKK␣⅐␤⅐␥ complex is ␣ 1 ␤ 1 ␥ 2 , and the core subunit composition of IKK␤⅐␥ is ␤ 2 ␥ 2 . The activity of the IKK complex (␣؉␤؉␥ or ␤؉␥) expressed in yeast (which lack NF-B and IKK) is 4 -5fold higher than an equivalent amount of IKK from nonstimulated HeLa cells. In the absence of IKK␥, IKK␤ shows a level of activity similar to that of IKK from nonstimulated HeLa cells. Thus, IKK␥ activates IKK complex in the absence of upstream stimuli. Deleting the ␥ binding domain of IKK␤ or IKK␣ prevented IKK␥ induced activation of IKK complex in yeast, but it did not prevent the incorporation of IKK␥ into IKK and large complex formation. The possibility of IKK complex being under negative control in mammalian cells is discussed.
Nuclear factor B (NF-B) 1 comprises a family of dimeric transcription factors that regulate the expression of over 150 genes involved in immune, stress, and antiapoptotic processes (1)(2)(3)(4). Under normal circumstances, NF-B is tightly regulated so as to prevent inappropriate inflammation while allowing a rapid response to infection or stress. In unstimulated cells, NF-B is found predominantly in the cytoplasm in a complex with IB proteins (a family of inhibitory subunits including IB␣, IB␤, IB␥, IB⑀, and Bcl-3), which sequester NF-B and prevent its migration to the nucleus (5,6). Diverse stimuli, including cytokines, bacterial and viral products, oxidants, and mitogens, lead to phosphorylation of two regulatory serine residues on IBs, which targets it for polyubiquitination and proteolytic degradation. This frees NF-B to move to the nucleus, where it binds to and stimulates the transcription of target genes (7). This phosphorylation is catalyzed by a large kinase complex, IB kinase (IKK) (8 -10). IKK is composed of two homologous kinase subunits, IKK␣ and IKK␤ (85 and 87 kDa, respectively) and a 52-kDa regulatory subunit IKK␥ (8,10,11), also called NEMO (NF-B essential modulator) (12). IKK␥ is required for activation of IKK in response to TNF and other stimuli (13). IKK␣ and IKK␤ each contain an N-terminal protein kinase domain (containing a canonical mitogen-activated protein kinase kinase activation loop (9)), a leucine zipper, and a helixloop-helix motif toward the C terminus (10). The catalytic subunits are associated with each other via their leucine zippers (11), and the helix-loop-helix domains are required for full IKK activation (14,15). It has been suggested that intramolecular interaction of the helix-loop-helix with the kinase domain is involved in IKK activation (14,15). Studies of recombinant IKK␣ and IKK␤ in insect cells indicate that the catalytic subunits are capable of forming both homodimers and heterodimers (11).
Based on gel filtration analysis, IKK predominantly forms a 700 -900-kDa complex containing IKK␣, IKK␤, and IKK␥, but some IKK also elutes at 230 kDa (6,8). The stoichiometry of IKK subunits in the large complex is still not known. The 230-kDa complex appears to be dimers containing only IKK␣ and IKK␤, because IKK␣ and IKK␤ expressed in insect cells and purified to homogeneity elute at 230 kDa (11) and because, in IKK␥-deficient cells, IKK␣ and IKK␤ elute at this size (12). The large IKK complex contains a roughly stoichiometric amount of IKK␣ and IKK␤ and an unknown amount of IKK␥ (6,8,13).
IKK␥ is required for the stimulation of IKK activity by upstream signals such as TNF, Tax, lipopolysaccharide, phorbol 12-myristate 13-acetate, and interleukin 1 (12,13). An ␣-helical region toward the N terminus of IKK␥ interacts with six amino acids at the very C terminus of IKK␣ and IKK␤ (24); interfering with this interaction by means of a peptide inhibitor * 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  in cells diminishes stimulation of IKK by TNF␣ (24).
The effect of IKK␥ on basal IKK activity is less clear. One report indicated that IKK␤ (lacking the C-terminal region, where it binds to IKK␥) was able to activate NF-B 1.5-2 times more than wild-type IKK␤, and expression of IKK␤ that contains point mutations to prevent IKK␥ binding was able to activate NF-B to a greater extent than IKK␤ that is capable of binding IKK␥ (24 (24). From these experiments, the authors suggested that interfering with the interaction of IKK␥ and IKK␤ increases basal intrinsic activity of IKK (24). By contrast, another report indicated that expressed IKK␤ in COS cells alone had low activity but that its activity was stimulated by co-expression of IKK␥, suggesting that IKK␥ stimulates IKK␤ (in the absence of stimuli) (25). To better understand whether the presence of IKK␥ has a stimulatory or inhibitory effect on IKK␤ in the absence of stimulation and to ascertain the role of the IKK␥ binding domain (␥BD) on basal IKK activity, reconstitution of the full IKK complex in a model system lacking endogenous IKK and its upstream signaling pathways would be very helpful.
In this paper, we demonstrate that human IKK can be reconstituted in yeast and forms a complex that is the same size as IKK isolated from human cells. The activity of this complex was 4 -5-fold higher than the IKK activity from nonstimulated HeLa cells. We used this reconstituted system to study the role of the interaction of IKK␥ with IKK␣ and IKK␤ (on the level of kinase activity) and also to study the stoichiometry of subunits.

EXPERIMENTAL PROCEDURES
Cloning and Expression of IKK in Yeast-All IKK subunits were expressed with an influenza hemagglutinin (HA) tag at the N terminus. HA-IKK␥ was subcloned into the p425 methionine-inducible yeast expression vector, which contains a LEU2 selection marker (26). The promoter regions of pESC-ura and pESC-trp (Stratagene) were removed and replaced with the promoter, multiple cloning site, and transcription termination sequence from p425, and HA-IKK␣ and HA-IKK␤ were subcloned into these vectors, respectively, to generate pESC ura met HA-IKK␣ and pESC trp met HA-IKK␤. The mutant IKK␤ ⌬␥BD was generated by PCR using Pfu polymerase (Stratagene) and the primers 5Ј-GTTAAATGAGGGCCACACATTGG and 5Ј-TCATGAGGCCTGCTC-CAGGCAGCTGTGCTCTTCTTCTTCCGTCTGGGCCG TGAAACTCTG to loop out the 18 nucleotides corresponding to the ␥BD (24); the PCR product was digested and subcloned into the vector pESC trp met HA-IKK␤. IKK␣ ⌬␥BD was constructed using PCR to truncate the last 8 amino acids using the primers 5Ј GGATCAGATTATGTCTTTGCATGC and 5Ј CCCGTTAACTCAATTCATCATACT and subcloned into the pESC ura met HA-IKK␣ vector. The deleted regions were verified by sequencing.
Plasmids were transformed into Saccharomyces cerevisiae strain YPH499 (Stratagene) using lithium acetate as described (Stratagene pESC Yeast Epitope Tagging Vectors Instruction Manual). A 2-ml overnight culture of yeast was grown in selective drop-out medium (Q-Biogene) containing 4 mM methionine (Q-Biogene) to suppress expression of the IKK and then amplified in 400 ml of selective noninducing drop-out medium. The yeast were grown at 30°C with shaking at 300 rpm for 30 h before being transferred to inducing medium (without methionine) for 10 -12 h (at 30°C with shaking).
For harvesting and lysing the yeast, all steps were performed at 4°C unless otherwise indicated. They were first washed in 400 mM (NH 4 ) 2 SO 4 , 200 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 10% glycerol containing protease inhibitors (2.5 g/ml leupeptin, 20 g/ml aprotinin, 2.5 g/ml antipain, 2 g/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride, 0.1 g/ml chymostatin, and 1.1 g/ml phosphoramidon). 1-2 g of yeast pellet was resuspended in 2 ml of lysis buffer (20 mM Tris (pH 7.6), 20 mM NaF, 20 mM ␤-glycerophosphate, 0.5 mM Na 3 VO 4 , 2.5 mM sodium metabisulfite, 5 mM benzamidine, 1 mM EDTA, 0.5 mM EGTA, 10% glycerol, 300 mM NaCl, 1% Triton X-100, 2 mM dithiothreitol with protease inhibitors) in a capped 15-ml conical tube, frozen at Ϫ80°C, and thawed on ice. Acid-washed 425-600-m glass beads from Sigma (equal in volume to the yeast pellet) were added to the yeast, and the suspension was vortexed 3 times for 1 min each (with 1-min incubation on ice between mixings). Then the suspension was centrifuged at 3000 ϫ g for 3 min, and the supernatant was collected. To extract more protein, 1 ml of additional lysis buffer was added to the yeast, and the vortexing and centrifugation and resuspension procedure was repeated an additional eight times. To remove particulate material, the crude supernatant was centrifuged at 65,000 ϫ g for 1.5 h, and the supernatant was collected and stored at -80°C. For some experiments, IKK was partially purified by gel filtration.
Immunoprecipitation, Kinase Assays, and Western Blotting-Lysates (S100 supernatants) from nonstimulated or TNF-stimulated HeLa cells were prepared as previously described (8). IKK␥ with a hexahistidine tag was expressed in Escherichia coli and purified by nickel affinity chromatography as described previously (13). IKK␤ with a hexahistidine tag was expressed in Sf9 cells and purified by nickel affinity chromatography as described (11).
Extracts or fast protein liquid chromatography fractions from HeLa cells and yeast were immunoprecipitated using 1 g of monoclonal anti-IKK␣ antibodies (B78 -1, Pharmingen) followed by binding to protein G-Sepharose beads (Amersham Pharmacia Biotech). Immune complexes were pelleted and washed once with lysis buffer and once with 20 mM Tris (pH 7.6), 20 mM MgCl 2 .
For kinase assays, 5-15 l of fast protein liquid chromatography fraction or washed immune complexes was incubated for 30 min at 30°C with a 30-l reaction mixture containing 20 mM Tris (pH 7.6), 20 mM MgCl 2 , 20 M cold ATP, 2 mM dithiothreitol, 33 g/ml GST-IB␣ 1-54 , and ␥-32 P ATP (ICN). (GST-IB␣ 1-54 , expressed in bacteria and purified using glutathione-Sepharose beads (Amersham Pharmacia Biotech), was used as the substrate because it contains the regulatory serines but lacks other residues that could be phosphorylated nonspecifically (27).) The reaction was terminated by the addition of SDS-PAGE sample buffer and heating for 5 min at 97°C. After SDS-PAGE and transfer (see below) radiolabeling was detected by PhosphorImager (Molecular Dynamics).
Extracts, immunoprecipitated proteins, and fast protein liquid chromatography fractions were electrophoresed by SDS-PAGE and transferred to polyvinylidene difluoride (Bio-Rad). Blots were probed using monoclonal antibodies directed against IKK␣, IKK␤, or IKK␥ (Imgenex) or against HA (USC/Norris core facility) followed by horseradish peroxidase-linked anti-mouse IgG antibodies (Amersham Pharmacia Biotech) and then detected by chemiluminescence (Pierce SuperSignal reagent). For experiments to quantify the ratio of catalytic to regulatory subunits, proteins were transferred for 2 h at 300 mA; to verify that the transfer of IKK proteins was complete, any remaining proteins in the gel were transferred to a second polyvinylidene difluoride membrane. The results indicated that 99% of IKK␤ and 100% of IKK␥ were transferred under our conditions. Densitometry was performed using a Bio-Rad Fluor-S Max quantification system.

RESULTS
Reconstitution of Human IKK in Yeast-Endogenous and recombinantly expressed IKK has been characterized from mammalian cells as well as insect cells (Sf9 cells with the baculovirus system), but the yeast system may have some advantages for biochemical studies. The baculovirus expression system in Sf9 cells has been successfully used to reconstitute catalytic subunits (11,19). However, a complete reconstitution has not been shown in Sf9 cells and is not practical due to the complications associated with multiple viral infection in Sf9 cells. Mechanistic analysis is also complicated in Sf9 and mammalian cells by the presence of endogenous proteins because expressed mutated forms of IKK are directed into heterocomplexes containing endogenous proteins (11). Recent development of IKK knockout cell lines partially resolves this problem, but there are also newly discovered IKK homologs that may have some redundant and overlapping functions.
Many of these potential pitfalls can be overcome by using a reconstituted system. Current knowledge indicates that S. cerevisiae lacks NF-B activity (28) and therefore is unlikely to contain NF-B or its upstream signaling molecules. Therefore, exogenously expressed proteins (such as IKK subunits) probably would not be affected by yeast signaling pathways.
The three subunits of IKK were subcloned into plasmids (each with a different selection marker: uracil, tryptophan, or leucine) containing HA tags and methionine-inducible promoters and transformed into yeast. (The inducible system was used in order to grow the yeast to a sufficient density before induction in case the expressed proteins were toxic). The yeast were grown in selective liquid media prior to induction. After 10 -12 h induction, the yeast were washed and lysed, and the S100 was obtained (see under "Experimental Procedures").
As indicated by Western blot (Fig. 1A), yeast that were not transformed did not contain IKK␣, IKK␤, or IKK␥ (see the far right lane (YPD)); however, these yeast do contain a protein recognized by the ␣HA antibody that runs below IKK␥ (data not shown). Yeast were transformed with IKK␣, IKK␤, or IKK␥ in various combinations, and clones expressing the IKK proteins at high levels were chosen for further study. In most clones transformed with multiple subunits, the IKK␥ expression was higher than the expression of ␣ or ␤ (as assessed by Western analysis with their identical HA tag). The level of IKK␣ was slightly lower than the level of IKK␤ in the IKK␣⅐␤⅐␥ clone shown (which was used for further studies).
Because IKK expressed in bacteria forms large aggregates that are not native (data not shown), we needed first to determine whether IKK reconstituted in yeast formed a complex that was similar in size to IKK isolated from human cells. Extracts from untransformed yeast and yeast expressing human IKK␤ or IKK␣⅐␤⅐␥ or mutant IKK␤ KA ⅐␥ were fractionated on a Superose 6 gel filtration column, and IKK activity toward GST-IB␣ 1-54 was assessed in each fraction. As shown in Fig.  1B, IKK␤ (alone) produced in yeast runs at 158 -300 kDa; this is the same size as dimers of IKK␤ (without IKK␥) from mammalian or Sf9 cells (11). The predominant peak of IKK from TNF-stimulated HeLa cells elutes at about 900 kDa. IKK␣⅐␤⅐␥ produced in yeast produces two peaks, one the size of the full IKK complex from human cells and the other around 158 -300 kDa (the size of the catalytic subunit dimers). Extracts from untransformed yeast and from yeast expressing mutant IKK␤ KA ⅐␥ do not have significant IKK activity in any fraction (compared with an equal amount of fractions 10 -11 taken from yeast expressing IKK␣⅐␤⅐␥). Similar results were obtained when IKK was isolated from each fraction by immunoprecipitation for kinase assay. These results indicate that the IKK that we have expressed in yeast is native and that, most likely, the 900-kDa complex contains no additional proteins. To demonstrate that it behaves the same as in mammalian cells, we expressed a mutant of IKK in which the critical lysine in the catalytic site is mutated to alanine (␤ KA ); this IKK was inactive as assessed by immunoprecipitation/kinase assay (Fig. 1C).
Stoichiometry of the IKK Complex-The IKK␣, IKK␤, and IKK␥ that were used for yeast expression have identical HA tags at their N termini. This allowed us to determine the ratio of regulatory to catalytic subunits in the complex. Supernatant from yeast co-expressing human IKK␤ and IKK␥ was partially purified by gel filtration to remove any subunits that were not incorporated into the large complex. The 900-kDa fraction was analyzed by Western blot using antibodies directed against HA. As shown in Fig. 2A, there is roughly an equal amount of IKK␤ and IKK␥ in this complex. Densitometric analysis indicates that the ratio of ␥ to ␤ is between 1.2 and 1.5.
Similarly, when IKK␣⅐␤⅐␥ was partially purified by gel filtration and analyzed by Western using antibodies against HA, the ratio of IKK(␣ϩ␤) to IKK␥ was 1:1. We attempted to use the HA immunoblot to quantify the ratio of IKK␣ to IKK␤, but unfortunately, the tagged proteins are inseparable, even with a large 7.5% SDS-PAGE gel (Fig. 2B). It was previously shown (by Coomassie Blue staining) that the IKK complex contains ). The yeast were grown and the protein was expressed as described under "Experimental Procedures." A, the immunoblots indicate that IKK subunits were only expressed in strains transformed with these genes. B, IKK␣⅐␤⅐␥ reconstituted in yeast forms a 900-kDa complex as assessed by gel filtration. Extracts from yeast expressing IKK␤, IKK␣⅐␤⅐␥, IKK␤ KA ⅐␥, or no IKK, as well as TNF-stimulated HeLa cell extract, were fractionated by Superose 6 gel filtration, and the kinase activity toward GST-IB␣ 1-54 was assessed in each fraction. The results indicate that IKK␣⅐␤⅐␥ forms both large (ϳ900 kDa) smaller (158 -300 kDa) complexes, whereas IKK␤ alone forms a smaller, 158 -300-kDa, complex. The samples were not standardized for equal amounts of IKK in each gel filtration; this experiment is intended to show the size of complexes containing IKK activity. The bottom panels show that there is negligible IKK activity in yeast that were not transformed with IKK and in yeast transformed with kinase-defective IKK (IKK␤ KA ⅐␥) (compared with an equal amount of wild-type IKK␣⅐␤⅐␥ expressed in yeast fractions 10 -11). C, extracts from yeast expressing IKK␤ KA ⅐␥ and IKK␤⅐␥ (as well as untransformed yeast) were chromatographed by Superose 6 gel filtration, and IKK was immunoprecipitated from the 900-kDa fraction (fractions 10 -11) using antibodies directed against HA. The complexes were assessed for kinase activity, and the level of IKK␤ was assessed on the same blot by Western blotting using antibodies against IKK␤. roughly equal amounts of IKK␣ and IKK␤ (13). Therefore, the core subunit composition of IKK␣⅐␤⅐␥ complex is ␣ 1 ␤ 1 ␥ 2 , and the core subunit composition of IKK␤⅐␥ is ␤ 2 ␥ 2 .
Activity of Human IKK Expressed in Yeast-In terms of activity, we predicted two possible scenarios: 1) that the complex would be low activity (similar to or lower than IKK activity from nonstimulated HeLa cells), or 2) that the complex would have high activity (similar to IKK from TNF-stimulated cells). IKK activity from yeast expressing IKK␣⅐␤⅐␥ (partially purified by gel filtration) was compared with nonstimulated and TNFstimulated HeLa cell extracts (S100); for these studies, the complexes were all immunoprecipitated using specific antibodies against IKK␣ (the subunit that was limiting in the yeast). The results (Fig. 3A) indicate that the activity of yhIKK␣⅐␤⅐␥ is intermediate to nonstimulated and TNF-stimulated HeLa cells. The activity of TNF-stimulated HeLa cells was ϳ15-20fold higher than the activity from nonstimulated HeLa cells, and the activity of IKK␣⅐␤⅐␥ expressed in yeast was ϳ4-fold higher than the activity from nonstimulated HeLa cells (Fig.  3B). To verify that the IKK complex reconstituted in yeast is specific for the regulatory serines in IB␣, we tested the activity of this enzyme toward a mutant form of IB␣ in which the regulatory serines are substituted with alanines (AA). Similar to the enzyme from HeLa cells, IKK␣⅐␤⅐␥ made in yeast phosphorylates wild-type IB␣ 1-54 but not the AA mutant (Fig. 3C).
We also compared the various recombinant IKK complexes expressed in yeast to each other. Fig. 3D compares the activities of IKK␣, IKK␣⅐␥, and IKK␣⅐␤⅐␥. The 900-kDa complexes of IKK␣⅐␥ and IKK␣⅐␤⅐␥ were partially purified by gel filtration before immunoprecipitation to eliminate complexes not containing ␥, whereas IKK␣ was immunoprecipitated directly from the S100. (Samples were adjusted to contain similar amounts of IKK␣ in this experiment. Because the stoichiometry of IKK␣⅐␤⅐␥ is 1:1:2, and the stoichiometry of ␣⅐␥ is 2:2, the IKK␣⅐␤⅐␥ sample contained approximately twice as many total IKK complexes as IKK␣ or IKK␣⅐␥.) The results indicate that IKK␣ and IKK␣⅐␥ have very low kinase activity toward GST-IB␣ 1-54 whereas IKK␣⅐␤⅐␥ has much higher kinase activity. The activity of IKK␣⅐␥ was over twice the activity of IKK␣ alone. The activity of IKK␣⅐␤⅐␥ was 10 -13-fold higher than that of IKK␣⅐␥.
Next, we compared the activities of IKK␤, IKK␤⅐␥, and IKK␣⅐␤⅐␥ complexes reconstituted in yeast and partially purified by gel filtration. As shown in Fig. 3E, the activity of IKK␤ was lower than the IKK activity of the complexes containing IKK␤⅐␥ or IKK␣⅐␤⅐␥. The activity of IKK␤⅐␥ and IKK␣⅐␤⅐␥ was ϳ7-15-fold higher than that of IKK␤ alone. These data suggest that IKK␥ plays a role in allowing the kinase to self-activate. The kinase-stimulating effect of co-expression of IKK␥ with IKK␤ was observed in completely different yeast clones and

FIG. 2. IKK has a 1:1 ratio of regulatory to catalytic subunits.
A, HA-IKK␤ and HA-IKK␥ were co-expressed in yeast, and the 900-kDa complex was isolated by gel filtration. There is an equal amount of both subunits in this complex as assessed by Western blot against their identical HA tag. Similarly, when HA-IKK␣, HA-IKK␤, and HA-IKK␥ were co-expressed in yeast and isolated by gel filtration, the total amount of catalytic subunit (HA-IKK␣ ϩ HA-IKK␤) was equal to the total amount of regulatory subunit (HA-IKK␥). Therefore, the ratio of regulatory to catalytic subunits is 1:1. B, HA-IKK␣⅐␤⅐␥ complex (partially purified by gel filtration) was electrophoresed through a large 7.5% SDS-PAGE gel and transferred to polyvinylidene difluoride, and parallel lanes were probed using antibodies directed against IKK␤, IKK␣, and HA. Because the Western bands for HA-IKK␣ and HA-IKK␤ directly overlap, it is not possible to discern the ratio of HA:IKK␣ to HA:IKK␤.

FIG. 3. Kinase activity of IKK reconstituted in yeast.
A, IKK␣⅐␤⅐␥ reconstituted in yeast (and partially purified by gel filtration) as well as S100 extracts from nonstimulated and TNF-stimulated HeLa cells were immunoprecipitated using antibodies against IKK␣, and then IKK activity was assessed. The same blot was probed to assess the amounts of IKK␣ and IKK␤. B, the kinase activity in each lane in A was quantified by phosphorimager. C, to examine substrate specificity, IKK␣⅐␤⅐␥ reconstituted in yeast (and partially purified by gel filtration) as well as S100 extracts from nonstimulated and TNF-stimulated HeLa cells were immunoprecipitated using antibodies against IKK␣, and then IKK activity toward wild-type (WT) substrate or GST-IB␣ 1-54 , in which the two regulatory serines (32 and 36) are mutated to alanines (AA), was assessed. The same blot was probed to assess the amount of GST-IB␣ 1-54 substrate and the amount of IKK␤. D, IKK␣⅐␥ and IKK␣⅐␤⅐␥ (900-kDa complexes partially purified by gel filtration) and IKK␣ (S100) were immunoprecipitated using antibodies against IKK␣, and kinase activity was assessed. The same blot was probed to assess the amount of IKK␣. E, IKK␤, IKK␤⅐␥, and IKK␣⅐␤⅐␥ were partially purified by gel filtration, and IKK activity was assessed in the various fractions. The same blot was probed to assess the amount of IKK␤. F, IKK␤, IKK␤⅐␥, and IKK␣⅐␤⅐␥ were partially purified by gel filtration, and the amount of IKK activity was assessed using varying amounts of substrate (GST-IB␣ 1-54 ). G, IKK␤⅐␥ expressed in yeast and partially purified by Superose 6 gel filtration was immunoprecipitated using antibodies against HA, IKK␤ expressed in Sf9 cells was purified by nickel chromatography and immunoprecipitated using antibodies against FLAG, and the kinase activity was assessed. The same blot was probed with antibodies against IKK␤ to assess the amount of IKK␤. preparations, indicating that the effect is a general phenomenon (data not shown). Moreover, the higher activities of IKK␤⅐␥ and IKK␣⅐␤⅐␥ than IKK␤ alone was observed over a range of IB␣ concentrations, indicating that the substrate was not limiting (Fig. 3F). Finally, we compared the activity of IKK␤⅐␥ expressed in yeast to IKK␤ expressed in Sf9 by immunoprecipitation/kinase assay; the results (Fig. 3G) indicate that the enzyme expressed in Sf9 cells is over twice as active as the enzyme expressed in yeast.
Role of IKK␥ and ␥ Binding Domain in IKK Activity-To further explore the role of IKK␥ on the activity of IKK, we generated IKK␣ and IKK␤ constructs in which the ␥BD at the C terminus (24) has been deleted. IKK␤ ⌬␥BD was transformed alone and along with IKK␥ and IKK␥ plus IKK␣ ⌬␥BD into S. cerevisiae and the interaction of IKK␥ with these mutants was assessed by immunoprecipitation and by gel filtration. As previously shown by affinity pull-down analysis (24), the interaction of IKK␥ with IKK␤ ⌬␥BD was very weak compared with the interaction of IKK␥ with wild-type IKK␤ as assessed by immunoprecipitation (data not shown). However, the interaction of IKK␥ with IKK␤ ⌬␥BD or with IKK␣ ⌬␥BD ϩ IKK␤ ⌬␥BD was not entirely abolished as assessed by gel filtration. As shown in Fig. 4A, IKK␤ ⌬␥BD expressed alone elutes from the Superose 6 gel filtration column at 158 -300 kDa (the same as wild-type IKK␤). However, when co-expressed with IKK␥ in the yeast, some of the IKK␤ ⌬␥BD forms a complex with IKK␥ and elutes as a high molecular weight complex. Similarly, some of the IKK␣ ⌬␥BD ϩIKK␤ ⌬␥BD forms a Ͼ700-kDa complex with IKK␥.
Whereas wild-type IKK␤⅐␥ and wild-type IKK␣⅐␤⅐␥ elute predominantly in fractions 10 and 11 (ϳ900 kDa), the IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥ and IKK␤ ⌬␥BD ⅐␥ complexes eluted predominantly in fractions 11 and 12, suggesting that the size or shape of the complex may be slightly different from wild-type IKK.
To investigate the role of the ␥BD in IKK activity, we compared the activity of these mutant forms to the corresponding wild-types (Fig. 4B). IKK␤ ⌬␥BD alone had a level of activity similar to that of IKK␤ wild-type, and as shown previously, the activity of IKK␤ alone was much lower than that with IKK␥. We looked at two gel filtration fractions from the IKK␤ ⌬␥BD ⅐␥ extract, fraction 11, in which IKK␤ ⌬␥BD was complexed with IKK␥, and fraction 14, which was devoid of IKK␥. Fraction 11 had very low activity, indicating that the association of IKK␤ ⌬␥BD with IKK␥ was not enough for IKK␥ to allow IKK to self-activate, suggesting that the ␥BD is required for the selfactivation of IKK␤ in the absence of stimulation. Fraction 14 had a level of activity that was similar to that of wild-type IKK␤ and that of IKK␤ ⌬␥BD alone.
Similar effects were observed when we compared the activity of IKK␣⅐␤⅐␥ wild-type to IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥. Association of IKK␥ with the IKK␣ ⌬␥BD and IKK␤ ⌬␥BD mutants was not sufficient to allow the complex to self-activate. It appears that the presence of the ␥BD is needed for IKK␥ to allow IKK to self-activate even in the absence of upstream signaling. This may suggest that this interaction is inhibited in resting mammalian cells.
Finally, we wanted to investigate whether we could activate IKK␤ by the addition of purified IKK␥ in vitro. IKK␤ (partially FIG. 4. IKK␥ is required for IKK to self-activate. A, IKK␤ ⌬␥BD , IKK␤ ⌬␥BD , IKK␥ (IKK␤ ⌬␥BD ⅐␥), IKK␤ ⌬␥BD , IKK␣ ⌬␥BD , and IKK␥ (IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥) were co-expressed in yeast, and complex formation was assessed by Superose 6 gel filtration followed by Western blot. IKK␤ ⌬␥BD forms only a small complex, similar in size to wild-type IKK␤. IKK␤ ⌬␥BD ⅐␥ and IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥ form complexes that are over 700 kDa. B, co-expression of IKK␥ with IKK␤ lacking a ␥BD does not facilitate self-activation. Coexpression of IKK␥ with wild-type IKK␤ or with wild-type IKK␤ and IKK␣ forms complexes that are partially activated. However, expression of IKK␥ with IKK␤ ⌬␥BD (IKK␤ ⌬␥BD ⅐␥ or IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥) forms complexes that are not activated. C, incubation of IKK␤ or IKK␤ ⌬␥BD with IKK␥ in vitro does not allow IKK to self-activate. IKK␤ and IKK ␤⌬␥BD (partially purified by gel filtration) were incubated with 0, 10, and 50 ng of purified IKK␥ for 30 min at 4°C prior to kinase assay. Comparing the kinase activity to a similar amount of IKK␤⅐␥ (extract reconstituted in yeast), binding of IKK␥ to IKK␤ in vitro did not allow IKK to self-activate. purified by gel filtration) was incubated with 0, 10, and 50 ng of pure IKK␥ for 30 min on ice before assessment of IKK activity. As shown in Fig. 4C, addition of IKK␥ could not activate the kinase. Similarly, incubation of IKK␥ with IKK␤ ⌬␥BD did not change the kinase activity. This suggests that the IKK␥ must form a complex with IKK␤ in vivo in order to facilitate self-activation. DISCUSSION Previous research indicated that S. cerevisiae lacks NF-B activity (28), and this report indicates that yeast do not contain IKK subunits as assessed by Western blot and also lack the ability to phosphorylate the regulatory serines on IB␣. Reconstitution of IKK complex containing ␣, ␤, and ␥ subunits turned out to be a useful tool because it allowed production of a large quantity of native complex for structural and mechanistic studies. Similar to mammalian and insect cells, IKK catalytic subunits expressed alone in yeast form relatively small 158 -300-kDa complexes, whereas the catalytic subunits co-expressed with IKK␥ elute at ϳ900 kDa. This indicates that the IKK reconstituted in yeast is native and most likely contains no additional proteins. Through the use of the identical HA tag on each subunit, we were able to show that there is approximately a 1:1 ratio of IKK catalytic subunits to IKK␥. Therefore, the core subunit composition of IKK␣⅐␤⅐␥ is ␣ 1 ␤ 1 ␥ 2 . Both IKK␣ and IKK␣⅐␥ reconstituted in yeast had a much lower level of kinase activity toward GST-IB␣ 1-54 than IKK␣⅐␤⅐␥ when adjusted for equal amounts of IKK␣. This was a predicted result because it was previously shown that IKK␤ is a more effective kinase for IB␣ than IKK␣ (19).
The activity of reconstituted IKK␣⅐␤⅐␥ was higher than an equivalent amount of IKK from nonstimulated HeLa cells but lower than an equivalent amount of IKK from TNF-simulated HeLa cells. In mammalian cells, IKK is regulated by phosphorylation and dephosphorylation, but the exact mechanisms of regulation are still not known. IKK activity is inhibited by PP2A in vitro, indicating that the kinase is activated by phosphorylation (8). Phosphorylation of two sites in the activation loop of IKK␤ is essential for activation of IKK by itself or after stimulation with TNF or interleukin 1, although the kinase responsible is unknown (14). Putative upstream kinases of IKK include NF-B-inducing kinase (29), mixed-lineage kinase (30), NF-B-activating kinase (20), and DNA-dependent protein kinase (31). There is also evidence to suggest that the phosphorylation of T-loop residues may occur through autophosphorylation (15), indicating that IKK can self-activate. The partial activation of IKK reconstituted in yeast could be explained if yeast contains a true IKK activator (such as an upstream kinase like mitogen-activated protein kinase kinase kinase) that is only partially active under conditions in which the IKK was being made. Alternatively, the yeast may contain a kinase that is homologous to the true IKK activator but far less capable of activating IKK.
On the other hand, it is possible that the partial activity of IKK reconstituted in yeast is due to lack of negative regulation in the yeast. An enzymatic activity or negative regulator using a different mechanism may be preventing IKK self-activation in mammalian cells. This negative regulation may occur through regulation of IKK␣⅐␤ interaction with IKK␥ (see below). IKK reconstituted in yeast will provide a useful system for analyzing putative positive and negative regulators of IKK. The yeast reconstitution system was used to assess the role of IKK␥ on IKK activity and to assess the importance of the ␥BDs found in IKK␤ and IKK␣. IKK␤ expressed in yeast in the absence of IKK␥ had much lower IKK activity than IKK␤⅐␥ or IKK␣⅐␤⅐␥. There are two alternative reasons for lower kinase activity in the absence of IKK␥. First, it is possible that IKK␥ is needed to allow homologous signaling proteins present in yeast to activate the expressed IKK (to the small extent it was activated). Second, it is also possible that IKK needs to form a large complex in order to autophosphorylate and self-activate and does so through IKK␥. The fact that IKK␤ ⌬␥BD ⅐␥ and IKK␣ ⌬␥BD ⅐␤ ⌬␥BD ⅐␥ formed large but inactive complexes indicates that IKK␥ interacts with different regions of IKK␣ and IKK␤ to hold the complex together, and the interaction of IKK␥ with the ␥BD of IKK␣ and IKK␤ is a dynamic interaction required for activation. The yeast data suggest that this dynamic interaction is somehow prevented in resting mammalian cells. In addition, the 4 -5-fold higher activity from TNF␣stimulated cells over IKK expressed in yeast suggests that interaction of IKK␥ with the C terminus of IKK␣ and IKK␤, although required for activation, is not sufficient for full activation of IKK. This in turn suggests that IKK may be regulated by a multistep mechanism. A multistep activation mechanism would provide IKK with the regulatory potential, e.g. being activated at different intensities and kinetics, to respond to the great diversity of NF-B inducers. The yeast reconstitution system will provide a useful tool for further structural and mechanistic analyses of IKK. Human IKK expressed in yeast can be used for clean mechanistic analysis because there is no background of endogenous IKK proteins. It is also useful for biochemical and regulatory studies, because when the IKK is expressed in yeast and isolated, it is simple to test whether a single molecule or subcellular fraction changes the activity of the enzyme. Finally, it can be used to study the structure and composition of the IKK complexes.