Detection of IKKβ-IKKγ Subcomplexes in Monocytic Cells and Characterization of Associated Signaling*

The IκB kinase (IKK) complex is one major step in the regulation of the NF-κB/Rel system that is involved in inflammatory and immune responses as well as in proliferation and apoptosis. At present it is not clear whether besides the “classical” IKKα-IKKβ-IKKγ configuration additional complexes exist in vivo that solely contain IKKβ and IKKγ (without IKKα). In the current study we were able to demonstrate in monocytic cells that endogenous complexes, which only include IKKβ as the kinase-active molecule do indeed exist in vivo and that these complexes contain IKKγ as an additional component. Furthermore, we showed that these IKKβ-IKKγ complexes are involved in mainstream NF-κB activation cascades because they can be activated by tumor necrosis factor. In contrast, these subcomplexes appear not to participate in NIK-dependent pathways. As a next step we showed that exogenous IKKβ-IKKγ complexes can be formed in an intact cell by overexpression and that these artificial complexes fulfill the requirement for participation in regular signaling. Finally, in the absence of IKKα we found a retarded proteolysis of IκBα, but not of IκB∈, which is associated with a reduced IKK activity. Differential pathways represented by various IKK subcomplexes may open attractive possibilities in treatment of inflammation or cancer allowing specific therapeutic intervention.

One major step in the regulation of the NF-B/Rel system is the IKK 1 complex (1,2). Activation of this high molecular weight assembly complex is achieved via signaling pathways induced by numerous molecules or conditions, e.g. cytokines like TNF, bacterial, and viral products as well as several forms of cellular stress (1,3,4). Such stimulation leads to the phosphorylation of proteins of the NF-B inhibitor family IB, for example IB␣, -␤, and -⑀, which subsequently undergo ubiquitindependent proteasomal degradation (5)(6)(7). This frees the NF-B dimer, most commonly p50/p65 (RelA), which is trapped in an inactive state in the cytosol by IB (5)(6)(7). Following stimulation, NF-B translocates to the nucleus, where it binds to promoters or enhancers of genes, e.g. coding for cytokines, chemokines, adhesion molecules, proteases, and procoagulatory proteins involved in inflammatory and immune responses as well as in proliferation and apoptosis (7,8). Differential signaling pathways leading to the activation of NF-B/Rel transcription factors as well as targeted gene expression are only partly understood and may be attractive targets for specific therapeutic approaches for inflammation or cancer (9 -12).
The core IKK complex is composed of three subunits: IKK␣, IKK␤, and IKK␥ (also called NF-B essential modulator, NEMO; IKK-associated protein 1, IKKAP1) (1,2,13,14). IKK␤ and IKK␣ are the two kinase-active components of this complex. IKK␤ is essential for activation of the IKK complex by various proinflammatory stimuli (15)(16)(17)(18), whereas IKK␣ appears to be involved in specific signaling associated with proliferation/differentiation (19 -22). It should also be mentioned that it is now evident that IKK␣, although dispensable for IKK activation in response to proinflammatory stimuli, is essential for IKK activation by an alternative set of signals that do not affect the IKK␤ subunit (1). Furthermore, IKK␥ has been suggested to represent an adaptor or scaffold protein, which stabilizes the high molecular weight complex and/or regulates its kinase activity (1,2,(23)(24)(25)(26).
The upstream signaling pathways, which lead to the activation of the IKK complex, are still only partly understood (2,27). A relatively well studied example is the TNF signaling pathway: Engagement of TNF with TNF receptor (TNFR) 1 results in the formation of a multiprotein signaling complex containing the cytosolic adaptor proteins TNFR-associated death domain protein, TNFR-associated factor 2, and receptor-interacting protein (3,28,29). These adaptor proteins in turn recruit IKK␤ to the TNFR1 complex where it becomes activated, finally initiating downstream processes leading to NF-B activation (1,28). Additional kinases have been suggested to activate the IKK complex such as TAK1, which is associated with Toll-like receptor/IL-1 receptor-associated signaling pathways (1,30,31). The IKK signaling event may also be mediated, dependent on the stimulus, by protein kinase R, mitogen-activated protein kinase/ERK kinase kinase-1 (MEKK1) or MEKK3 (1,27). One of the first kinases proposed to activate the IKK complex was the NF-B-inducing kinase (NIK), but further studies indicated that it plays a function in specific pathways, e.g. processing of p100 and depends on the catalytic activities of downstream acting IKK␣ (13,32,33).
Earlier studies indicate that IKK is a large complex with a size varying between 700 and 900 kDa, suggesting the presence of additional components (1,13,14,34). One possibility that has been suggested is that some of the kinases or adaptors mentioned above in the previous paragraph, as well as additional signaling molecules, temporarily associate with the core complex (1). Alternatively, it may be possible that besides the "classical" components, different subunits are present as permanent components in the IKK complex. In this context it should be mentioned that two recent reports identified cell division control (Cdc) 37 and heat shock proteins (Hsp) 90 as well as 27 as further components of a TNF-inducible IKK complex (35,36). In addition, it has been shown that Hsp70 is able to form a supramolecular structure with the IKK␥ subunit (25).
At present it is not clear if besides the classical IKK␣-IKK␤-IKK␥ configuration additional complexes exist in vivo that solely contain IKK␤ and IKK␥ (without IKK␣) as suggested earlier (24). Interestingly, in genetic experiments IKK␣ Ϫ/Ϫ mice display a significant IKK activation in response to proinflammatory stimuli and also induction of NF-B DNA binding activity, presumably because of formation of artificial (since IKK␣ is missing), but still functional IKK␤-IKK␥ complexes (19 -22). Recombinant IKK␣ and IKK␤ are able to form homoor heterodimers, which suggests that similar IKK complexes may also exist in vivo (34). In addition, complete recombinant reconstitution of the IKK in yeast revealed that the core IKK complex is composed of ␣1␤1␥2, but also ␤2␥2 (37).
The current study was designed to examine in vivo in monocytic cells whether endogenous complexes do indeed exist which solely contain IKK␤ and no IKK␣ and if these complexes contain IKK␥. If so, we wanted to investigate whether these IKK␤-IKK␥ complexes participate in regular signaling and can be activated by certain stimuli, e.g. TNF, and to investigate which IB proteolysis patterns are induced during this activation.
PAGE and Western Blot Analysis-Cytosolic and nuclear extracts were isolated as described (4,39), and electrophoresis was performed with 12.5% polyacrylamide gels. The proteins were transferred to nitrocellulose membranes using the wet blot technique. After transfer, the membranes were incubated with antibodies against IKK␣ (Alexis Biochemicals, Carlsbad), IKK␤ (Biocarta, San Diego), IKK␥ (BD Biosciences, Palo Alto, CA), IB␣, IB⑀, NIK (Santa Cruz Biotechnology, Heidelberg, Germany), phospho-IB␣ (Calbiochem, Darmstadt, Germany), FLAG (Stratagene, Amsterdam, Netherlands), or actin (Sigma). This was followed by the appropriate horseradish peroxidaseconjugated secondary antibody (Dianova, Hamburg, Germany). The proteins were visualized on x-ray film using the Chemiluminescent Reagent Plus (PerkinElmer Life and Analytical Sciences). The obtained signals were analyzed densitometrically, and the values corrected for the loading control for each protein.
Immunodepletion-For immunodepletion cytosolic extracts were subjected to serial immunoprecipitation 4ϫ for 1 h at 4°C with 20 g of specific non-cross-reactive anti-IKK␣ or IKK␤ antibody and 250 l of 6% protein A-agarose in TNT buffer as described above. This procedure was followed by a normal immunoprecipitation for 1 h at 4°C with 1 g of antikinase antibody as described above.
Kinase Assay-IP was carried out as described above and followed by the kinase assay (11). The kinase reaction was performed in kinase buffer for 30 min at 30°C in the presence of 5 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences) and 500 ng of the substrate GST-IB␣ (Santa Cruz Biotechnology). Proteins were analyzed on 12.5% polyacrylamide gels (0.1% SDS), and following Western blot the transferred proteins were visualized by autoradiography. Densitometric analysis was performed for the obtained signals, and the values corrected for the loading control for each protein.
Overexpression Experiments-The plasmids coding for IKK␣, IKK␤ (both wild type, FLAG-tagged) as well as NIK wild type and mutated (KK429-430AA) were obtained from Tularik Inc., South San Francisco, and the overexpression plasmid for IKK␥ (NEMO) was a gift from Prof. Alain Israël and co-workers, Institut Pasteur, Paris, France. RcCMV (Invitrogen) containing no insert was used as a negative control. These plasmids were transiently co-transfected into HeLa cells or IKK␣ knockout MEFs using Superfect (Qiagen, Hilden, Germany) (38). After overnight culture, cells were left untreated or stimulated with TNF at 2 ng/ml, and cytosolic or nuclear extracts were isolated.

Detection of Endogenous IKK␤-IKK␥ in Vivo in Monocytic
Cells-Initial experiments were performed with monocytic cells to demonstrate the presence of endogenous complexes, which contain IKK␤ as well as IKK␥ but no IKK␣. For this purpose THP-1 cells were incubated in the absence or presence of TNF (2 ng/ml) for 5 min, and cytosolic extracts were isolated. IKK␣-containing complexes or free IKK␣ were removed by immunodepletion using an antibody against the IKK␣ protein, and this procedure was repeated up to 4ϫ for each sample. The remaining IKK complexes were then precipitated by anti-IKK␥, and the nature of the precipitated IKK proteins was monitored by Western blot analysis. These experiments demonstrated that virtually no IKK␣ (associated with IKK␥) remained in the cellular extracts following the IKK␣ depletion procedure (Fig. 1A). In contrast, in the IKK␣-depleted samples a significant level of IKK␤ and, as expected, IKK␥ was detected in the precipitate (Fig. 1A). Comparable results were observed in the absence or presence of TNF. The densitometric analysis of three experiments showed that during repeated immunoprecipitations with the anti-IKK␣ antibody the level of IKK␣ went down essentially to zero (Fig. 1B). Under the same conditions, a decrease of IKK␤ down to ϳ46% of the undepleted control was observed, which was comparable to the decrease in IKK␥ (31%) at the same time (Fig. 1B). Essentially the same results were obtained when, following the IKK␣ immunodepletion, the anti-IKK␤ antibody was used instead of anti-IKK␥ to precipitate the remaining complexes for further analysis (data not shown). These experiments demonstrate that IKK complexes exist in vivo in monocytic cells, whose major components are IKK␤ as well as IKK␥. In the present article these complexes are designated as IKK␤-IKK␥ complexes.
Activation of Endogenous Monocytic IKK␤-IKK␥ Complexes by TNF-In the following we wanted to evaluate whether the endogenous IKK␤-IKK␥ complexes play a role in the regular NF-B signaling pathway, and whether these can be activated by known stimuli of this system. For this purpose, THP-1 monocytic cells were incubated with TNF (2 ng/ml) and the cytosolic extracts were subjected to an immunodepletion of IKK␣ as described above. The supernatants containing the IKK␤-IKK␥ complexes as well as undepleted samples were subjected to immunoprecipitation with an anti-IKK␥ antibody, and the obtained precipitate was added to a kinase assay. The experiments demonstrated a reduced baseline of IKK activity in unstimulated IKK␣-depleted cells ( Fig. 2A). However, exposure to TNF still significantly increased IKK activity in IKK␣depleted cells, albeit with a lower absolute peak level compared with normal, undepleted cytosolic extracts. Following the evaluation of the kinase activity the complete depletion of IKK␣ protein as well as the remaining amount of IKK␤-and IKK␥ proteins was verified by Western blot analyses of the kinase assay precipitate (Fig. 2A). These experiments were further extended by a kinetic study in which THP-1 cells were incu-bated with TNF for 2.5, 5, 7.5, and 10 min. As described above, kinase activity of IKK␥-precipitated complexes was determined followed by densitometric analysis. These data show a transient activation of IKK␤-IKK␥ complexes by TNF, with fold induction values of kinase activity lowered to ϳ60% compared with non-depleted samples (Fig. 2B). Taken together, these results indicate that the IKK␤-IKK␥ complexes in monocytic cells can be stimulated by a classical stimulus such as TNF and are therefore likely to be involved in mainstream NF-B activation cascades potentially displaying a slightly different signaling dynamic.
NIK Cannot Activate IKK␤-IKK␥ Complexes-Recently, it has been shown that NIK together with IKK␣-or IKK␣-containing complexes is involved in specific pathways, e.g. p100 processing (32). In our experiments, we wanted to test if the IKK␤-IKK␥ complexes also participate in NIK-induced signaling. To investigate this we overexpressed either wild type or the dominant negative form of NIK together with the CMV vector alone as a control in HeLa cells. Western blot analysis of cytosolic extracts of these cells revealed an overexpression of NIK both in the wild-type and mutated form (Fig. 3A). NIKinduced IKK activity was monitored in cytosolic extracts of these cells by kinase assay following an immunoprecipitation A, sequential immunodepletion (0 to 4 steps with protein Aagarose bound to specific antibody) was carried out with cytosolic extracts of either unstimulated or TNF-stimulated (2 ng/ml, 5 min) THP-1 cells using a specific antibody against IKK␣ (ID IKK␣). This was followed by immunoprecipitation with an anti-IKK␥ antibody, and the amount of IKK␣, -␤, and -␥ proteins in the precipitate was tested using Western blot analysis. B, the level of IKK␣, -␤, and -␥ from experiments as described in A was analyzed by densitometric measurement. The amount of the respective protein in untreated samples (0ϫ ID) was defined as 100%. The experiments were repeated independently three times (mean Ϯ S.E.).
using an antibody against IKK␣ or IKK␥. These experiments demonstrated significantly increased kinase activity in HeLa cells overexpressing wild-type NIK with a more intense absolute peak level following the anti-IKK␣ immunoprecipitation compared with the immunoprecipitation using an antibody against IKK␥ (Fig. 3B). As a next step to investigate the role of IKK␣ the above samples were subjected to an immunodepletion using an anti-IKK␣ antibody. Following two and four steps of depletion, the samples were added to an immunoprecipitation using an anti-IKK␥ antibody and subjected to a kinase assay. Remarkably, NIK-induced kinase activity was reduced after two steps of immunodepletion and totally abolished following four steps of IKK␣ immunodepletion (Fig. 3C). Western blot analysis revealed, similar as described above, that IKK␣ was completely depleted while there were still IKK␤-IKK␥ complexes left. These results indicate that the signaling induced by NIK overexpression cannot be mediated by IKK␤-IKK␥ complexes and that the removal of IKK␣ abolishes the inducbility by NIK.
Formation of IKK␤-IKK␥ by Overexpression and Activation of These Complexes-In these experiments we aimed to reconstitute exogenous IKK␤-IKK␥ complexes by overexpression experiments. HeLa cells were used to overexpress different ratios of IKK␤ and IKK␥ or CMV vector alone. It should be mentioned that in this case only IKK␤ has a FLAG tag whereas IKK␥ is untagged. A total amount of 10 g of DNA with different ratios was transfected, and the overexpression of the different proteins was detected by Western blot analysis (Fig. 4A). As a next step, the association of IKK␤ with IKK␥ was tested. For this, cytosolic extracts of HeLa cells overexpressing IKK␤ and IKK␥ in different ratios were subjected to an immunoprecipitation for either kinase, and the FLAG-or IKK␥-precipitated proteins were detected by Western blot analysis. The results demonstrate that the overexpressed IKK␤ and IKK␥ associate with each other and that exogenous IKK␤-IKK␥ complexes can be reconstituted (Fig. 4B). For all of the following experiments a plasmid DNA ratio of 1:1 was used. To find out whether endogenous IKK␣ protein binds to this artificial complex, the cytosolic extracts of either unstimulated or TNF-stimulated (2 ng/ ml) HeLa cells overexpressing both IKK␤ and IKK␥ were immunoprecipitated using an anti-FLAG antibody, and the precipitates were analyzed. The experiments revealed that both IKK␤ and IKK␥ could be isolated under this condition whereas no significant amount of IKK␣ is present in this artificial complex, and that TNF does not seem to have any effect on the formation of the complexes (Fig. 4C). Finally, we wanted to check the possibility of stimulating these complexes using cytosolic extracts of HeLa cells overexpressing IKK␤ and IKK␥ FIG. 2. Activation of endogenous monocytic IKK␤-IKK␥ complexes by TNF. A, THP-1 cells were incubated in the absence or presence of TNF (2 ng/ml, 5 min). Anti-IKK␥ was used to precipitate IKK complexes from non-depleted or depleted (ID IKK␣) extracts. Kinase activity was measured by adding GST-IB␣ as the substrate. The levels of IKK␣, IKK␤, and IKK␥ in the precipitate used for kinase assay was subsequently determined by Western blot analysis. B, cytosolic extracts of monocytic THP-1 cells incubated with TNF in time course experiments were treated as described in A to compare IKK activity of non-depleted or depleted samples. Data of three independent experiments were analyzed by densitometry and normalized to IKK␥ loading. Results are depicted as fold induction above the unstimulated control (mean Ϯ S.E.). The inset shows representative kinase assays for both undepleted (no ID) and depleted samples.
or CMV vector alone. Immunoprecipitations were carried out using an anti-FLAG antibody, and the precipitate was added to kinase assays. In these experiments no signal in cytosolic extracts of cells transfected with the CMV vector alone could be measured (Fig. 4D). In contrast, in cells overexpressing IKK␤ as well as IKK␥ a relatively high basal activity was detected as observed earlier (40), which was further increased by stimulation with TNF (Fig. 4D). Equal loading was confirmed by Western blot analyses following the kinase assay using FLAG-antibodies. These experiments revealed that exogenous IKK␤-IKK␥ complexes appear to fulfill the minimal requirements to participate in regular signaling pathways.
Proteolysis of IB␣ but Not of IB⑀ Is Retarded in the Absence of IKK␣-The next question was to find out if the absence of IKK␣ affects the proteolysis pattern of the inhibitory IB proteins. To examine this we used the model of IKK␣ Ϫ/Ϫ MEF cells (41) in time course experiments in which the cells were stimulated with TNF (2 ng/ml). The proteolysis of IB␣ and -⑀ was determined by Western blot analysis in wild type as well as in IKK␣ knockout cells. As expected, TNF induced a rapid degra-dation of IB␣ in wild-type cells (Fig. 5A). Interestingly, Western blot analysis indicated that, following TNF stimulation, IB␣ was degraded more slowly in IKK␣ Ϫ/Ϫ cells compared with the wild type. This effect could be reverted when an IKK␣ overexpression plasmid was transfected into the IKK␣ Ϫ/Ϫ cells (Fig. 5A). As a control, the level of actin was monitored, which remained unchanged during this experiment (data not shown). As a next step, data were analyzed by densitometry and IB␣ values were normalized to actin levels. This analysis revealed that in wild-type cells a significant amount of proteolysis of IB␣ was detected at 5 min following exposure to TNF with a maximal degradation at 15 min (Fig. 5B). In contrast, in knockout fibroblasts a weak TNF-initiated degradation of IB␣ was detected at 10 min, and a time point of maximal degradation was established at 20 min. When IKK␣ was overexpressed in IKK␣ knockout cells a significant amount of IB␣ proteolysis could be already detected at 10 min after TNF stimulation with a maximal degradation at 15 min. In contrast, IB⑀ proteolysis seems to be similar in wild-type and IKK␣ Ϫ/Ϫ cells (Fig. 5C). In both cases, initial IB⑀ degradation could be detected at 5-10 FIG. 3. IKK␤-IKK␥ complexes cannot be activated by NIK. A, HeLa cells were transfected with 2 g of the wildtype (Wt) or of the mutated (mut) form of NIK or CMV as a control. Overexpression was checked by Western blot analysis of the cytosolic extract. B, following immunoprecipitation of the extract described above with an antibody against IKK␣ or IKK␥, kinase activity was determined using GST-IB␣ as a substrate. C, kinase activity of non-depleted or depleted (ID IKK␣) extracts of cells with overexpressed NIK (wild type, mutated) or CMV as a control was measured. The amount of IKK␣, -␤, and -␥ in this precipitate was determined by Western blot analysis. min after TNF stimulation reaching a maximum at 60 min (Fig. 5D).
Reduced IKK Activity and IB␣ Phosphorylation in the Absence of IKK␣-Because phosphorylation is the first step that marks IB proteins for degradation we decided to analyze the IKK activity as well as the phosphorylation status of IB␣ in IKK␣ knockout cells compared with wild-type MEFs. For this we performed kinase assays following IKK␤ immunoprecipitation with extracts of wild-type or IKK␣ Ϫ/Ϫ cells stimulated with TNF for 5 or 10 min, respectively. As a loading control, the level of immunoprecipitated IKK␤ was detected by Western blot analysis. Autoradiography followed by densitometry revealed a significant TNF-induced IKK activity in both wild-type and knockout cells (Fig. 6A). However, in IKK␣ Ϫ/Ϫ cells the IKK activity was reduced by ϳ50% compared with the wild type (Fig. 6A), which is consistent with the data shown in Fig. 2B. Similar results were obtained when an anti-IKK␥ antibody was used as the precipitation antibody for kinase assay (data not shown). As there seem to be differences in the kinase activity between wild-type and IKK␣ knockout cells we decided to examine the phosphorylation status of IB␣ following stimulation with TNF. Western blot analyses were performed using an anti-phospho-IB␣ antibody or as a control for equal loading an anti-actin antibody. The data revealed a reduced phosphorylation of IB␣ in the IKK␣ Ϫ/Ϫ cells, which confirms the data obtained by kinase assays (Fig. 6B).
Reduced Proteolysis of IB␣ in the Presence of Overexpressed IKK␤-IKK␥-Finally, we reconstituted exogenous IKK complexes by overexpression (see also Fig. 4), which contained IKK␣, IKK␤, and IKK␥ or solely IKK␤ and IKK␥, respectively, and examined proteolysis of endogenous IB␣ protein. The overexpression of the different proteins was monitored by Western blot analyses (Fig. 6C, data not shown). Densitometric analyses were calculated following normalization to actin. In the presence of both reconstituted exogenous complexes a significant proteolytic removal of IB␣ was detected (Fig. 6C). Remarkably, in the presence of exogenous IKK␤-IKK␥, a low but clearly detectable level of endogenous IB␣ remained (15 or 4% of control, dependent on the amount of transfected DNA), whereas almost no IB␣ appeared to be left in the presence of the reconstituted classical IKK␣-IKK␤-IKK␥ complexes (Fig.  6C). This demonstrates a reduced proteolysis of endogenous IB␣ in the presence of overexpressed IKK␤-IKK␥ compared with the situation when IKK␣-containing complexes are reconstituted and indicates differences between the two complexes in mediation of IB␣ proteolysis. DISCUSSION The current study demonstrates the presence of endogenous IKK␤-IKK␥ complexes in vivo in monocytic cells. Consistent with our data, earlier experiments indicated that heterogenous IKK complexes exist in HeLa cells (24), whereas in the SLB T-cell line no IKK␤-only pool was found, which suggests cell type-specific differences. Recombinant IKK␣ and IKK␤ are known to form homo-or heterodimers, which also implies that similar IKK complexes may indeed exist in vivo (34). Complete recombinant reconstitution of the IKK in yeast reveals that the stoichiometric architecture of the core IKK complex is either ␣1␤1␥2 or, remarkably, ␤2␥2 (37).
Our data show that these monocytic IKK␤-IKK␥ complexes can be activated by TNF, a potent stimulus for NF-B (3,28). In our experiments, following IKK␣ immunodepletion, a significant TNF-induced increase of IKK activity was detected in the anti-IKK␥ precipitate, albeit with a lower absolute basal and peak kinase activity as well as a reduced fold induction value when compared with non-depleted cells. These data agree with a previous experiment, which demonstrates that after sequen- Overexpression in HeLa cells of IKK␤ and IKK␥ in different ratios (10 g of total plasmid DNA) or CMV as a control was monitored by Western blot analysis. It should be mentioned that the overexpressed IKK␤ protein is FLAG-tagged. Please note that in some of the CMV controls a weak unspecific band was detected. B, association of expressed IKK␤ with IKK␥. The overexpressed proteins were immunoprecipitated (IP) by an antibody against IKK␥ or the FLAG tag, respectively. The proteins in the precipitate were detected by Western blot analysis (WB). C, IKK␣ is not present in overexpressed IKK␤-IKK␥ complexes. IKK␤ and IKK␥ were overexpressed (ratio: 1:1) and IKK complexes of cytosolic extracts from unstimulated or TNF stimulated (2 ng/ml, 5 min) cells were immunoprecipitated using an antibody against the FLAG tag. The different proteins of the precipitate were monitored by Western blot analysis. D, activation of overexpressed IKK␤-IKK␥ complexes. IKK␤ as well as IKK␥ (1.25 g each) were overexpressed and the cells stimulated by TNF (2 ng/ml). The IKK␤-IKK␥ complexes were immunoprecipitated with anti-FLAG (IP), and the precipitate was added to a kinase assay using GST-IB␣ as a substrate. tial immunoprecipitation with an IKK␣ antibody there is some kinase activity remaining in the HeLa extract (24). The levels of IKK␤ or IKK␥ were not changed by TNF treatment in our study suggesting that no dynamic rearrangement occurs during this stimulation. In earlier studies using IKK␣ Ϫ/Ϫ mice, the fibroblast cells display significant IKK activation in phosphorylating IB␣ and IB␤ in vitro in response to the proinflammatory stimuli TNF or IL-1, measured by kinase assays with antibodies to either IKK␤ or IKK␥ (19 -22). The latter effect is presumably due to the presence of endogenous or, alternatively, formation of new (since IKK␣ is missing) but still func-tional IKK␤-IKK␥ complexes (21). In accordance with functional IKK␤-IKK␥ complexes, several previous data including ours demonstrate that, when wild-type or dominant negative mutant forms of IKK␣ are expressed or fibroblasts of IKK␣ AA/AA knock in mice were stimulated, the TNF-or LPSinduced NF-B activation or B-dependent transcription in monocytic cells is at most partly affected whereas a dominant negative mutant of IKK␤ is much more effective (4,18,42). Taken together, these data suggest that IKK␤-IKK␥ complexes participate in NF-B-associated signaling.
When we overexpressed IKK␤ together with IKK␥ we were FIG. 5. IB␣ but not IB⑀ is degraded more slowly in the absence of IKK␣. A, wild-type cells (Wt), IKK␣ Ϫ/Ϫ MEFs, or IKK␣ Ϫ/Ϫ cells in which IKK␣ was expressed (Re, transfection of 2 g DNA) were incubated with TNF (2 ng/ml) in a time course experiment. The existence of IKK␣ in these cells was monitored by Western analysis in the absence (Ϫ) or presence (ϩ) of TNF (5 min), and actin was used as a loading control (left panels). After stimulation with TNF the level of IB␣ in cytosolic extracts was examined by Western blot analysis (right panels). B, densitometric analysis was used to compare the rate of IB␣ degradation between wild-type MEFs and IKK␣ Ϫ/Ϫ cells (without and with IKK␣ overexpression marked as Re). An IB␣/actin ratio was calculated using the amount of actin as a loading control. The value in unstimulated cells was defined as 100%. The data of three independent experiments were analyzed and mean Ϯ S.E. is depicted. C, wild-type (Wt) or IKK␣ Ϫ/Ϫ MEFs were incubated with TNF in a time course experiment, and I〉⑀ degradation in the cytosolic extracts was determined by Western blot analysis. D, rate of IB⑀ degradation was compared between wild-type cells and IKK␣ Ϫ/Ϫ MEFs by densitometric analysis similar as described in B.
able to demonstrate that these proteins associate with each other as was expected from earlier studies (26,37). This appears to be independent of IKK␣, since in these newly formed complexes no endogenous IKK␣ could be detected. Most importantly, we showed that these exogenous IKK␤-IKK␥ complexes can be activated by TNF. This indicates that the IKK␤-IKK␥ complex formed by overexpression itself fulfills the requirement to be connected upstream to regular signaling pathways.
It has been recently demonstrated that IKK␣ is a specific target for NIK and that once activated it may lead to phosphorylation-dependent processing of NF-B/p100 (1,32,33). Furthermore, it has been suggested that the classical IKK complex is not involved in this pathway that rather depends on a different IKK␣-containing complex (1). Therefore, to evaluate the participation of IKK␤-IKK␥ we tested if the removal of IKK␣-containing complexes would affect the kinase activity induced by overexpression of the NIK protein. Remarkably, following IKK␣ immunodepletion, no NIK-induced kinase activity could be detected in the remaining IKK␤-IKK␥ complexes, which implies that the latter complexes do not participate in NIK signaling. It should also be mentioned that these experiments may serve as an indirect functional approach to demonstrate the efficiency of the IKK␣ immunodepletion procedure applied in our studies.
To study signaling in the absence of IKK␣ we used the model of IKK␣ Ϫ/Ϫ fibroblasts (41), which were stimulated with TNF, and the proteolysis pattern of the inhibitory IB proteins was examined. Interestingly, following TNF treatment, IB␣ was degraded more slowly in IKK␣ Ϫ/Ϫ cells compared with the wild type. This effect could be reverted when IKK␣ was overexpressed in the IKK␣ Ϫ/Ϫ cells demonstrating that the presence of IKK␣ itself directly affects the degradation of IB␣. Furthermore, we found that the TNF-induced IKK activity and phosphorylation of IB␣ were reduced in knock out cells compared with wild-type fibroblasts. These findings suggest that the retarded proteolysis of IB␣ in IKK␣ knockout cells is at least partially caused by signaling associated with reduced IKK activity and indicates a slightly different signaling dynamic between IKK␤-IKK␥ and the IKK␣-containing core complex. In contrast, the absence of IKK␣ did not affect IB⑀ proteolysis since a comparable degree of TNF-induced IB⑀ degradation was detected in both knockout and wild-type cells. This may suggest that the reduced but still significant IKK activity in IKK␣ knockout cells is sufficient to mediate degradation of IB⑀ (which is known to occur at a slower rate) (7) to a similar extent as wild-type cells.
As already mentioned IKK␣ is not required for most proinflammatory signaling pathways (18 -22), and our experiments also indicate that IKK␤-IKK␥ complexes without IKK␣ function properly in an intact cell. Furthermore, recent studies suggest that IKK␣ plays a role in specific signaling pathways and differentiation processes as well as B-cell-mediated responses (19 -22, 42, 43). So one may ask why did IKK␣ remain in the classical IKK core complex assembly at all. Our experiments indicate that in the classical complex IKK␣ may function as a "turbo-like molecule" to accelerate certain signaling pathways such as IB␣ proteolysis. One possibility is that IKK␣ facilitates the association of amounts of FLAG-IKK␣, FLAG-IKK␤, and IKK␥ expression plasmid (ϩ) in the case of reconstitution of the classical IKK complex or with a two-thirds amount of IKK␤ plasmid (ϩϩ) compared with IKK␥ to achieve the proportion of the IKK␤-IKK␤-IKK␥ complex. As a control CMV alone was transfected. The level of endogenous IB␣ was determined by Western blot analysis (upper panel). Furthermore, the expression of IKK␣ and IKK␤ was monitored using an anti-FLAG antibody. The data were analyzed by densitometry and normalized to actin (lower panel). The IB␣/actin value in CMV-transfected cells was defined as 100%, and the data of three independent experiments are shown (mean Ϯ S.E.).
FIG. 6. IKK activity and IB␣ proteolysis in the absence of IKK␣. A, IKK activity is reduced in IKK␣ Ϫ/Ϫ cells. Wild-type (Wt) as well as IKK␣ Ϫ/Ϫ MEFs were incubated with TNF (2 ng/ml) for 0, 5, and 10 min and kinase activity was determined by adding GST-IB␣ as substrate (upper panel). Anti-IKK␤ antibody was used to precipitate IKK complexes. The levels of IKK␤ in the precipitate were subsequently determined by Western blot analysis. IKK activity of three independent experiments was analyzed by densitometry and normalized to the loading (lower panel). Results are depicted as fold induction above the unstimulated control (mean Ϯ S.E.). B, phosphorylation of IB␣ is reduced in IKK␣ Ϫ/Ϫ cells. The phosphorylation status of IB␣ in extracts treated as in A was determined by Western blot analysis with an anti-phospho-IB␣ antibody, and actin was used as a loading control. C, reduced proteolysis of endogenous IB␣ in the presence of overexpressed IKK␤-IKK␥ complexes compared with the situation when IKK␣-containing complexes are reconstituted. IKK␤ and IKK␥ were overexpressed in HeLa cells in the presence or absence of IKK␣ to reconstitute IKK␣-IKK␤-IKK␥ or IKK␤-IKK␥ complexes, respectively. A total amount of either 4.5 or 9 g of DNA was transfected in equal the core complex with additional proteins such as upstream kinases or adapters that enhance IKK activity and IB␣ phosphorylation (1,3). Alternatively, IKK␣ could potentially regulate IKK activity by serving itself as an upstream kinase that can phosphorylate other IKK molecules (1,3,44). The observation that in our overexpression studies (i.e. without an exogenous stimulus) IB␣ proteolysis was reduced in the presence of IKK␤-IKK␥ compared with the situation in which IKK␣-IKK␤-IKK␥ was reconstituted could indicate that trans-autophosphorylation between the IKK molecules is involved but also that constitutive signaling is affected by the absence of IKK␣ (41,45). Taken together, the ability of IKK␣ to increase IKK activity is possibly one of the reasons why this kinase, despite its different specific functions, remained a participant of the classical core complex during evolution.
The existence of IKK␤-IKK␥ besides IKK␣-IKK␤-IKK␥ complexes may open attractive therapeutic avenues for inflammation and cancer (9,10). For example, in sepsis hyperinflammation could be reduced by specifically blocking the latter IKK␣containing complexes, which would leave the IKK␤-IKK␥ subcomplexes still intact to mediate important basic processes as part of a minimal response capacity, e.g. to orchestrate the immune response (1,8,46). Alternatively, the blocking of IKK␤ may allow the recovery of the postulated IKK␣-specific (possibly IKK␤-independent) pathways (1,2,32,47). An alternate use of both inhibitory strategies during therapy would not allow exhaustion of both sides of the pathways but still effectively reduce inflammation.