Direct Phosphorylation of IκB by IKKα and IKKβ: Discrimination Between Free and NF-κB-Bound Substrate

See allHide authors and affiliations

Science  28 Aug 1998:
Vol. 281, Issue 5381, pp. 1360-1363
DOI: 10.1126/science.281.5381.1360


A large protein complex mediates the phosphorylation of the inhibitor of κB (IκB), which results in the activation of nuclear factor κB (NF-κB). Two subunits of this complex, IκB kinase α (IKKα) and IκB kinase β (IKKβ), are required for NF-κB activation. Purified recombinant IKKα and IKKβ expressed in insect cells were used to demonstrate that each protein can directly phosphorylate IκB proteins. IKKα and IKKβ were found to form both homodimers and heterodimers. Both IKKα and IKKβ phosphorylated IκB bound to NF-κB more efficiently than they phosphorylated free IκB. This result explains how free IκB can accumulate in cells in which IKK is still active and thus can contribute to the termination of NF-κB activation.

The IKK complex, isolated from extracts of HeLa cells treated with the proinflammatory cytokine TNF (tumor necrosis factor), phosphorylates two regulatory serine residues at the NH2-termini of the NF-κB inhibitors IκBα and IκBβ (1). This phosphorylation event triggers the polyubiquitination of IκBs followed by their degradation through the 26S proteasome, and thereby leads to NF-κB activation (2). IKK is a large, 900-kD, protein complex that is composed of multiple subunits. Two of these subunits, IKKα and IKKβ, are serine kinases (1, 3–6). Epitope-tagged IKKα and IKKβ, when produced by cell-free translation in reticulocyte lysates or by transient transfection of mammalian cells, are incorporated into the IKK complex, which can be isolated by immunoprecipitation of either IKKα or IKKβ (1, 4). The IκB kinase activity of the entire complex is rapidly stimulated by TNF or interleukin 1 (IL-1), with kinetics matching those of IκBα phosphorylation and degradation (1,4).

Because expression and immunoprecipitation of catalytically inactive IKKα or IKKβ mutants still resulted in isolation of a cytokine-inducible IκB kinase activity (4), the question was raised as to whether IKKα and IKKβ are directly responsible for IκB phosphorylation (7). Indeed, no IκB kinase activity was obtained after translation of IKKα or IKKβ in wheat germ extracts or after expression in Escherichia coli (1, 4, 8). Although transient expression of catalytically inactive IKKα or IKKβ mutants inhibits NF-κB activation (1, 3–6), these results are entirely compatible with IKKα and IKKβ being upstream kinases that activate another subunit of the IKK complex, which in turn phosphorylates IκB (7).

We used a baculovirus expression system to express (His)6-HA–tagged IKKα and (His)6-FLAG–tagged IKKβ separately or together in Sf9 insect cells (9). Recombinant IKKα and IKKβ were purified to apparent homogeneity from baculovirus-infected insect cells (10). These purified proteins (Fig. 1A) (11) phosphorylated IκBα specifically on Ser32and Ser36 and IκBβ on Ser19 and Ser23 (Fig. 1B) (11). Mutants of the NH2-terminal regulatory domains of both IκBs in which threonine was substituted for these serine residues were not phosphorylated (Fig. 1B) (11). Mutants of IKKα and IKKβ that are defective in generation of IKK activity in mammalian cells (4) were also expressed in Sf9 cells and purified. These mutants showed little or no IκB kinase activity (Fig. 1C). Hence, IKKα and IKKβ appear to be direct IκB kinases. Comparison of purified IKKβ to a panel of seven other kinases indicated that it was the only kinase that specifically phosphorylated IκBα at Ser32 and Ser36, as revealed by the differences in 32P incorporation into wild-type IκBα and the IκBα(A32/36) mutant (Fig. 1D). Even though some protein kinases such as casein kinase II (CKII) are efficient IκB kinases, the sites they phosphorylate are located in the COOH-terminal domain and not in the NH2-terminal regulatory domain (12).

Figure 1

Direct phosphorylation of IκB by IKKα and IKKβ. (A) IκBα phosphorylation by purified recombinant IKKβ. The protein was expressed using a baculovirus vector in Sf9 (9) cells and purified as described (10). The indicated amounts of highly purified IKKβ were separated by SDS-PAGE and visualized by silver staining (SS). These were also examined for phosphorylation of GST-IκBα(1–54) using a standard IκB kinase assay (KA). The phosphorylated GST-IκB(1–54) and IKKβ (Auto) bands are indicated. The migration positions of molecular size markers (in kilodaltons) are also indicated. Similar results were obtained for recombinant IKKα (11). (B) Substrate specificity of purified recombinant IKKβ was examined using GST-IκBα FL (full length) wild-type (WT), GST-IκBα FL with Ala at positions 32 and 36 (AA) or with single Ala substitutions at positions 32 (32A) or 36 (36A), GST-IκBβ FL, GST-IκBα(1–54) WT, GST-IκBα(1–54) AA, GST-IκBα(1–54) TT (Thr substitutions at positions 32 and 36), GST-IκBβ(1–44) WT, and GST-IκBβ(1–44) AA. (C) Effect of mutations on the kinase activities (KA) of recombinant IKKα and IKKβ. Autophosphorylation (Auto) and IκB kinase activities [with GST-IκBα(1–54) as a substrate] of purified recombinant wild type (WT), kinase defective [KM for IKKα and KA for IKKβ (4)], leucine zipper defective [LZ (4)], and helix-loop-helix defective [HLH (4)] mutants of IKKα and IKKβ, were determined. The bottom panels show immunoblots (IB) using anti-HA to detect WT IKKα and its derivatives (left) and anti-FLAG (M2) to detect WT IKKβ and its derivatives. (D) Phosphorylation of IκBα by different protein kinases. Wild-type GST-IκBα FL and GST-IκBα(A32/36) were incubated with similar amounts of the indicated protein kinases in standard kinase buffer and [γ-32P]ATP for 30 min. The reactions were terminated and separated by SDS-PAGE, and the extent of substrate phosphorylation was revealed by autoradiography. The bands corresponding to the autophosphorylated kinases are indicated by a dot.

Recombinant IKKα and IKKβ expressed alone or together (IKKα-IKKβ) migrated on a Superose 6 gel filtration column with an apparent molecular size of 230 to 250 kD (Fig. 2A). The native IKK complex from HeLa cells has an apparent size of 900 kD (1). Thus, the Sf9 cells may lack components that are present in the large IKK complex. Chemical cross-linking (13) of recombinant IKKα, IKKβ, or IKKα-IKKβ complexes resulted in appearance of a single species migrating around 200 kD (Fig. 2B). Mutations in the leucine zipper (LZ) motif of IKKα abolished formation of the dimeric 200-kD cross-linked species (Fig. 2B). This mutant does not phosphorylate IκB proteins, indicating that dimerization of IKKα may be necessary for formation of a functional IκB kinase. Mutations in the helix-loop-helix (HLH) motifs of IKKα or IKKβ did not abolish their homodimerization, as judged by the appearance of the 200-kD cross-linked species (Fig. 2B). However, these mutations did result in loss of kinase activity (Fig. 1C). Thus, the HLH motifs of IKKα and IKKβ are necessary for catalytic activity even in the absence of other regulatory proteins.

Figure 2

Formation of dimers by IKKα and IKKβ. (A) The Superose 6 elution profile of the purified IKK complex from TNF-α–treated HeLa cells was compared to the elution profiles of purified recombinant IKKα, IKKβ, and IKKα-IKKβ (α+β) expressed in Sf9 cells (1,4). The kinase activities toward GST-IκBα(1–54) are shown. The numbers at the top indicate the elution positions of molecular size standards (in kilodaltons). The elution profiles of IKKα- and IKKβ-immunoreactive materials determined by immunoblotting were identical to those of the kinase activities (11). (B) Purified IKK from TNF-α–treated HeLa cells, recombinant IKKα-IKKβ (α+β), IKKα (α), IKKβ (β), an HLH mutant of IKKβ, and HLH and LZ mutants of IKKα were cross-linked with EGS as described (13). Cross-linked proteins were either directly separated by SDS-PAGE or immunoprecipitated as indicated below before separation by SDS-PAGE. In either case, proteins were detected by immunoblotting. For the IKK complex from HeLa cells (HeLa), a representative blot shows an immunoblot (with anti-IKKα) of proteins that were immunoprecipitated with anti-IKKβ. An identical pattern of bands was observed when the blot was stripped and reprobed with anti-IKKβ (13). For the (HA)-IKKα–(FLAG)-IKKβ complex (α+β), the cross-linked proteins were immunoprecipitated with anti-HA and immunoblotted with anti-FLAG (M2). For (HA)-IKKα (α), (FLAG)-IKKβ (β), and the various mutants, the cross-linked proteins were directly resolved by SDS-PAGE and probed with anti-HA for IKKα and anti-FLAG (M2) for IKKβ. (C) Association of (HA)-IKKα and (FLAG)-IKKβ in coinfected Sf9 cells. (HA)-IKKα and (FLAG)-IKKβ were expressed either separately or together as indicated. The antibodies used for immunoprecipitation (IP) and immunoblotting (IB) were anti-HA and anti-FLAG (M2), as indicated.

Immunoprecipitation experiments with HA-tagged IKKα and FLAG-tagged IKKβ revealed formation of stable heterodimers when the two proteins were expressed together in Sf9 cells (Fig. 2C). We examined whether heterodimers in which one subunit was catalytically inactive would still catalyze the phosphorylation of IκBα (Fig. 3). Complexes of catalytically inactive IKKα and wild-type IKKβ exhibited similar IκB kinase activity to that of wild-type IKKα-IKKβ heterodimers. A heterodimer composed of wild-type IKKα and catalytically inactive IKKβ was also active, albeit not as active as wild-type IKKα homodimer. These results explain why immunoprecipitation of catalytically inactive IKKα or IKKβ transiently expressed in cultured mammalian cells results in formation of partially active IKK complexes (4).

Figure 3

Kinase activity of IKK heterodimers. Wild-type (FLAG)-IKKβ was expressed together with either wild-type (HA)-IKKα or kinase-defective (HA)-IKKα (KM) in Sf9 cells. Wild-type IKKα was expressed by itself or together with kinase-defective FLAG-IKKβ (KA). Extracts of cells infected with various combinations of the corresponding viruses, as indicated, were subjected to immunoprecipitation (IP) with either anti-HA or anti-FLAG (M2) as indicated. The kinase activity (KA) of each immune complex was examined with GST-IκBα(1–54) as a substrate. Amounts of wild-type (FLAG)-IKKβ or (HA)-IKKα proteins in the immune complexes were determined by immunoblotting (IB) with anti-FLAG (M2) or anti-HA, respectively. Exposure times for left and right panels were 1.5 and 5 hours, respectively.

IKKβ phosphorylated a fusion protein of glutathione transferase with the first 54 amino acids of IκB [GST-IκBα(1–54)] more efficiently than did IKKα (Fig. 4). However, full-length IκBα is phosphorylated by IKKα and IKKβ with similar efficiencies. The Michaelis constants (K m) of IKKα and IKKβ toward full-length GST-IκBα were 2.1 and 2.2 μM, respectively (14).

Figure 4

Kinase activities of recombinant IKKα and IKKβ. (HA)-IKKα and (HA)-IKKβ were expressed separately in Sf9 cells (10) and immunoaffinity-purified on immobilized anti-HA. Immunoblot (IB) analysis shows that similar amounts of each protein were used to compare their kinase activities (KA) toward purified GST-IκBα(1–54) and GST-IκBα(FL) (full length). Coomassie blue (CB)–stained gels show the relative amounts of the two substrates.

In resting cells, cytoplasmic IκB proteins are associated with NF-κB dimers, and phosphorylation does not result in dissociation of this complex (2). Thus, the actual substrate for IKK is the IκB–NF-κB complex rather than free IκB. Some stimuli—such as IL-1, which is one of the strongest NF-κB activators—lead to prolonged activation of IKK that lasts 2 hours or more (4). Once IκBα is phosphorylated and degraded (within 3 to 10 min), NF-κB translocates to the nucleus and activates gene transcription (15), including that of the IκBα gene (16). Newly synthesized free IκBα accumulates within 60 min of the initial stimulus and translocates to the nucleus, where it binds DNA-bound NF-κB to induce its shuttling to the cytoplasm (17). This process is important for termination of the NF-κB response. But given the prolonged activation of IKK, it is puzzling how newly synthesized IκBα escapes the IKK-induced degradation pathway. We therefore compared the efficiency with which IKK phosphorylated free and NF-κB–bound IκBα (18). At similar concentrations, IκBα complexed with NF-κB was a better substrate for IKKα or IKKβ than was free IκBα (Fig. 5). Kinetic analysis indicated that in the presence of NF-κB, theK m for IκBα phosphorylation by IKKβ decreased from 2.2 μM to 1.4 μM and the relative maximum initial velocity V max was increased by a factor of 5 (14). Although IκBβ is not involved in rapid feedback inhibition of NF-κB activity (19), its phosphorylation by IKKα or IKKβ was also strongly enhanced by binding of NF-κB (Fig. 5). These results, which show that IKK prefers NF-κB–bound IκB proteins, explain why newly synthesized free IκBα is not phosphorylated and degraded in cells in which IKK remains active after the initial inductive stimulus.

Figure 5

Preferred phosphorylation by IKKs of IκBs bound to NF-κB. The kinase activities (KA) of purified (HA)-IKKα (left panel) or (FLAG)-IKKβ (right panel) toward free or NF-κB–prebound GST-IκBα(FL) and GST-IκBβ(FL) were measured. The NF-κB heterodimer was composed of recombinant p50 and the Rel homology domain of p65 (18). Shown are the autoradiograms of the kinase assays (KA) and Coomassie blue (CB)–stained gels depicting the various substrates.

The IKKα and IKKβ protein kinases, and not another component of the IKK complex, appear to be directly responsible for IκB phosphorylation. The minimal active IKK complex is apparently a dimer composed of IKKα or IKKβ (or both). These experiments also provide an explanation for the mechanism underlying the termination of the NF-κB activation response. Such inactivation is important because prolonged or chronic NF-κB activation can result in inflammatory disorders (20).

  • * These authors contributed equally to this report.

  • To whom correspondence should be addressed. E-mail: karinoffice{at}


View Abstract

Navigate This Article