Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

IκB kinase complexes: gateways to NF-κB activation and transcription


Transcription factors of the NF-κB family regulate hundreds of genes in the context of multiple important physiological and pathological processes. NF-κB activation depends on phosphorylation-induced proteolysis of inhibitory IκB molecules and NF-κB precursors by the ubiquitin-proteasome system. Most of the diverse signaling pathways that activate NF-κB converge on IκB kinases (IKK), which are essential for signal transmission. Many important details of the composition, regulation and biological function of IKK have been revealed in the last years. This review summarizes current aspects of structure and function of the regular stoichiometric components, the regulatory transient protein interactions of IKK and the mechanisms that contribute to its activation, deactivation and homeostasis. Both phosphorylation and ubiquitinatin (destructive as well as non-destructive) are crucial post-translational events in these processes. In addition to controlling induced IκB degradation in the cytoplasm and processing of the NF-κB precursor p100, nuclear IKK components have been found to act directly at the chromatin level of induced genes and to mediate responses to DNA damage. Finally, IKK is engaged in cross talk with other pathways and confers functions independently of NF-κB.


The NF-κB transcription factor family comprises NFKB1 (p50/p105), 1209934 (p52/p100), RelA (p65), c-Rel and RelB, which maintain key functions in various biological and pathological processes (Gilmore, 2006). In most resting cells, the cytosolic inhibitor molecules IκBα, β and ɛ and the precursors p100 and p105 inhibit the release of active NF-κB. Two major types of signaling pathways have been implicated in NF-κB activation (Figure 1). A variety of stimuli induce the canonical NF-κB pathway with rapid phosphorylation and subsequent degradation of IκBs and p105 (Hayden and Ghosh, 2004). The IκB kinase (IKK) complex, containing the catalytic kinases IKKα and IKKβ and the regulatory non-enzymatic scaffold protein NEMO (NF-κB essential modifier; also called IKKγ) phosphorylates its substrates at conserved destruction boxes (Table 1) (Karin and Ben-Neriah, 2000). IKK-mediated phosphorylation targets IκBs and p105 for polyubiquitination by the Skp1, Cdc53/Cullin1, F-box proteinβtransducin repeat-containing protein (SCFβTrCP) E3 ligase complex and proteasomal destruction, causing the release of p50-, p65- and c-Rel-containing heterodimers.

Figure 1

Canonical and non-canonical NF-κB signaling pathways. A large spectrum of membrane receptors stimulates the canonical pathway via adaptors (e.g. TRAFs), which results in IKK complex activation. This involves non-degradative ubiquitination of TRAFs, RIP and NEMO/IKKγ (Ub K63), binding of ubiquitin chains by TABs or NEMO/IKKγ and activation loop phosphorylation of IKKα and β (P). UbLys63 modification can be reverted by deubiquitinating enzymes (DUB). IKK phosphorylates IκBs and p105, which are poly-ubiquitinated with UbLys48 chains (Ub K48) by SCFβTrCP and degraded by the proteasome. This liberates heterodimers of p50 with p65, c-Rel (not shown) or p50 homodimers. Several ligands, like LTβ or BAFF, additionally trigger the p100-processing pathway through TRAFs, the kinase NIK and selectively IKKα-containing complexes. Phosphorylation of p100 results in its poly-ubiquitination and processing to p52. The latter forms heteromers with RelB. A fraction of IKKα is recruited to chromatin and is thought to affect transcription of a subset of genes by phosphorylation of histone H3 and other proteins.

Table 1 Substrates of IKK

With much slower kinetics, a subset of inducers stimulates the non-canonical NF-κB pathway, where IKKα-mediated phosphorylaton of p100 promotes C-terminal processing of the precursor to generate p52-containing complexes (primarily p52/RelB), which also involves ubiquitination and proteasome activity (Pomerantz and Baltimore, 2002; Bonizzi and Karin, 2004) (Figure 1).

While many basic details of the canonical IKK pathway are well understood, fundamental questions concerning the IKK structure, composition, mechanisms of activation and localization have yet to be answered. Several potential activation mechanisms have been proposed. These include conformational changes by induced protein interactions, oligomerization-dependent auto-phosphorylation and phosphorylation by upstream IKK kinases (IKKKs), as well as catalytic, non-destructive Lys-63-linked polyubiquitination (Hayden and Ghosh, 2004; Chen, 2005). Many studies have supported the assumption that one major cytoplasmic IKK complex exists, which is constituted by the IKKα, β and NEMO subunits. However, IKKβ and NEMO on one hand and IKKα on the other underlie disparate regulation and transmit distinct signals (Hatada et al., 2000). This is exemplified by the primary requirement of IKKβ and NEMO, but not of IKKα, for proinflammatory signal transduction and prevention of embryonic hepatic apoptosis. IKKα, in turn, appears to be essential to stimulate p100 precursor processing and it regulates NF-κB-independent developmental processes. The dissimilar functions of the IKKs are also reflected by the different phenotypes of the corresponding knockout mice (see Gerondakis et al., 2006). Striking examples of the distinct regulation of IKK components are the nuclear roles proposed exclusively for IKKα and NEMO (Chen and Greene, 2004; Yamamoto and Gaynor, 2004). Nuclear IKKα may contribute to the transcription activation of target genes at the level of histone phosphorylation, while nuclear NEMO is implicated in sensing genotoxic stress.

Structural components of the IκB kinase core complex

The activation of NF-κB by all physiological stimuli investigated requires phosphorylation of IκB proteins. Thus, serine/threonine-specific IκB kinases are the bottleneck for all pathways that converge on NF-κB. For about 10 years, a large 700–900 kDa kinase complex with specificity for the serine residues in the destruction box of IκBα has been known and this complex was first partially purified from unstimulated HeLa cells (Chen et al., 1996). Activation of this IκB kinase complex was shown to require ubiquitination, which was independent of the proteasome, and thus provided a first hint to a non-degradative role for ubiquitin in the NF-κB pathway. However, the kinase molecules and ubiquitinated components in those experiments remained unknown. Thereafter, several groups identified two highly related catagolytic components of an inducible IκB kinase that were termed IKKα (or IKK1) and IKKβ (or IKK2) (DiDonato et al., 1997; Mercurio et al., 1997; Regnier et al., 1997; Woronicz et al., 1997; Zandi et al., 1997) (Figure 2). Both IKK enzymes contain an N-terminal kinase domain (KD), and extended regions C-terminal to the KD, which have conserved leucine zipper (LZ) and putative helix-loop-helix (HLH) motifs. Dimerization of the kinases is required for kinase activity and is mediated by the leucine zipper (Karin, 1999) (Figure 2). The C-terminal HLH domain is required for full IKK kinase activity and its intramolecular association with the KD was also proposed to play a role in the post-inductive downregulation of kinase activity (Delhase et al., 1999, Karin, 1999; Hayden and Ghosh, 2004, and see below).

Figure 2

Schematic outline of functional domains in IKKα, IKKβ and NEMO/IKKγ. Numbers refer to amino acid positions in the human proteins. Lysine residues in NEMO which undergo SUMOylation or non-degradative ubiquitination upon stimulation, UbK63 at Lys399 by MALT1 (Zhou et al., 2004) and at Lys285 by Nod2/RIP2 (Abbott et al., 2004); SUMO-1 and Ub at Lys277 and Lys309 by genotoxic stress (Huang et al., 2003a) and serine residues in NEMO/IKKγ that are phosphorylated upon Tax or TNFα stimulation (Ser-31, Ser-43, Ser-376) (Carter et al., 2003) or by ATM (Ser-85) (Wu et al., 2006b) are indicated. In IKKα and β, the T-loop serines are shown. del, deleted in splice variant IKKγΔ.

The kinase activity of both IKKα and IKKβ can be inactivated by a mutation of Lys-44, within the predicted ATP-binding site. However, IKKα and IKKβ are not biochemically equivalent. Mutational analysis of the activation loop (T-loop) serines in the KD (Figure 2) and in vitro experiments revealed that IKKβ is the primary target for proinflammatory stimuli (Delhase et al., 1999). In vitro, IKKβ has a higher catalytic activity towards IκBα than does IKKα, which in turn is a more proficient kinase for p100. IKKβ contains an ubiquitin-like domain (ULD) that is absolutely critical for its functional activity. The mutation or deletion of the ULD results in catalytic inactivation of IKKβ, while a deletion of the equivalent region in IKKα does not affect its kinase activity (May et al., 2004). The precise function of the ULD is not yet known. Only IKKα bears a functional nuclear localization sequence (NLS) (Sil et al., 2004).

The third IKK core component is the non-catalytic regulatory molecule named NEMO (aka IKKγ, IKKAP1 or Fip-3), which is essential for the function of the kinases in cells (Rothwarf et al., 1998; Yamaoka et al., 1998; Mercurio et al., 1999; Li et al., 1999b). NEMO is highly conserved, and structural predictions indicate an extended α-helical content with three coiled coil regions (CC1-3), a LZ motif and a C-terminal zinc finger (ZF) (Figure 2).

Based on gel filtration chromatography, the holocomplex formed by IKKα, β and NEMO has an apparent mass of 700–900 kDa. In cells lacking NEMO, IKKα and β migrate as a 300 kDa species and cannot be activated by any of the classical NF-κB inducers (tumor necrosis factor (TNF)α, Interleukin-1 (IL-1), Lipopolysaccharide (LPS) or dsRNA) (Yamaoka et al., 1998). Reconstitution of recombinant IKK complexes and reconstitution in yeast suggested that the holocomplex contains roughly equimolar amounts of NEMO and kinase molecules (Krappmann et al., 2000; Miller and Zandi, 2001).

Within the holocomplex and when assayed alone NEMO forms oligomers; however, there is some controversy about the number of NEMO monomers present in IKK complexes. One series of studies proposed that NEMO forms monomers and trimers (Agou et al., 2002, 2004). In contrast, a tetrameric oligomerization, as a dimer of NEMO dimers, was suggested by hydrodynamic and chemical crosslinking analyses of endogenous and highly purified recombinant NEMO (Tegethoff et al., 2003). A minimal oligomerization domain (MOD) for NEMO has been defined to extend from amino acids 246–365; this domain by itself only forms dimers, but full-length NEMO cannot form tetramers of upon its deletion. A second weak dimerization domain resides in the N-terminal region of NEMO. Thus, a tetramer of NEMO could be formed by a dimer of dimers (Tegethoff et al., 2003). Tetrameric NEMO could in theory bind two kinase dimers. This could be a plausible model for induced proximity allowing cross-phosphorylation of the kinase dimers after conformational changes of the NEMO scaffold upon docking to signalosome complexes. However, further studies are needed to unequivocally prove the oligomeric state of the holocomplex.

The interaction regions required for stable complex formation of NEMO with IKKα and IKKβ have been delineated. IKKα and β both bind to NEMO via a short conserved sequence (containing Leu-Asp-Trp-Ser-Trp-Leu, NEMO-binding domain, NBD) at their C-termini (May et al., 2000). There may also be weaker NEMO interactions conferred by additional, less defined regions in IKKα or β (Miller and Zandi, 2001). In NEMO, the kinase-binding region (KBD) has been mapped to various, in part non-overlapping N-terminal sequences between amino acids 50–93 (May et al., 2000), 105–200 (Poyet et al., 2000) or 1–196 and 65–196 for IKKα and IKKβ, respectively (Tegethoff et al., 2003). A recently described frequent NEMO splice variant, IKKγΔ (Hai et al., 2006), lacking amino acids 174–224 (Figure 2), binds both IKK kinases, and thus would delimit the KBD to sequences N-terminal to residue 174. The different regions in NEMO required for binding to IKKα or IKKβ suggest non-equivalent interactions. Competition experiments using the NBD peptide indicate that IKKβ binds to NEMO with considerably higher affinity than IKKα. Furthermore, a mutational analysis of the NBDs indicated that the binding of IKKα is less stringent (May et al., 2002). The exact interaction regions for IKKα and β in NEMO thus remain to be more precisely determined.

Given the essential role of NEMO for the activation of cellular IKK (Hayden and Ghosh, 2004), any disruption of its interaction with the catalytic components or of regions important for its oligomerization or interaction with activators can impair IKK function. This has been shown by binding competition with separately expressed subdomains. The C-terminal 108 and 118 residues of IKKα or IKKβ, respectively, containing the NBD, bind efficiently to NEMO and act as dominant-negative inhibitors to suppress induced kinase activity in transfected cells (Poyet et al., 2000) The NBD peptide containing only amino acids 735–745 of IKKβ (May et al., 2000), when fused to the antennapedia or HIV-Tat protein cell transduction domains, blocks IKK and NF-κB activation in transformed tumor cells, primary human cells and in mice (May et al., 2000; Choi et al., 2003; Jimi et al., 2004).

Likewise, the MOD inhibits induced IKK activity in intact cells when expressed separately (Tegethoff et al., 2003) and each of the MOD subregions (40 and 43 amino acids, respectively) can efficiently inhibit IKK activation when introduced into cells (Agou et al., 2004). The functional importance of the MOD was also demonstrated by mutation of two leucines within the LZ, which abrogates the ability of NEMO to confer kinase activation in response to TNFα or LPS (Makris et al., 2002). As one explanation, these mutations might disrupt the oligomeric state of NEMO. Very recently, however, it was shown that the MOD contains the binding site for Lys-63-linked polyubiquitin, which is important in the activation process (Figure 2 and see below). Thus, it remains to be determined to what extent functional impairment of IKK activity by competition with MOD peptides or by mutations in the MOD primarily affect ubiquitin-binding, oligomerization or both.

The ZF in NEMO was found to be required for activation of IKK following genotoxic stress, but was first considered as largely dispensable for activation by TNFα or LPS (Huang et al., 2002). However, later studies suggested an additional role for the ZF in inflammatory signaling. It was claimed that TNFα-induced IKK activity does require an intact ZF (Tang et al., 2003). A NEMO ZF mutation (C417R) found in the genetic disorder anhydrotic ectodermal dysplasia with immunodeficiency (see Courtois and Gilmore, 2006) abrogated both LPS and TNFα induced IKK activity and phosphorylation of the activation loop of IKKβ (Yang et al., 2004). It was also shown that a combined mutation of Cys-289 and Cys-393 reduces kinase responsiveness to TNFα-, but not to IL-1β-stimulation (Makris et al., 2002). The ZF is further needed for IKKβ-mediated phosphorylation of NEMO at more N-terminal residues (Carter et al., 2003) and for TNFα-stimulated ubiquitination of NEMO (Tang et al., 2003). An intact ZF was suggested to be required for the activation of IKK by CD40 ligand as well (Jain et al., 2001). Together, these findings suggest a more general role of the NEMO ZF in IKK complex activation and that the different surfaces of this structure bind to distinct pathway-specific activators.

Mutations in the MOD or in the Zn finger of NEMO are frequently found in patients with incontinentia pigmenti and related inherited diseases of the epidermis and further support the high functional relevance of these domains (see Courtois et al., 2001; Courtois and Gilmore, 2006).

Other stoichiometric components of the IKK complex

A further protein proposed as an essential regulatory subunit of the IKK complex is protein rich in amino acids E, L, K and S (ELKS) (Ducut Sigala et al., 2004). The 105 kDa ELKS protein copurified with IKKβ, and knockdown of ELKS expression by siRNA impaired NF-κB activation and the induction of NF-κB target genes by proinflammatory cytokines. Based on immuno-depletion experiments, it was suggested that ELKS is stoichiometrically associated with IKKα, IKKβ and NEMO. As a primary function, ELKS was proposed to recruit IκBα to the IKK complex. More recently, ELKS was also implicated in the activation of IKK by the DNA damage-responsive kinase ataxia telangiectasia mutated (ATM) following genotoxic stress (Wu et al., 2006b) (see below). Although all of these observations are intriguing, a component with the size of ELKS has not been seen in other purified IKK fractions or as an IKK-co-precipitated species from metabolically labeled cells (Rothwarf et al., 1998; Krappmann et al., 2000; Chen et al., 2002). Furthermore, there are five different ELKS mRNA isoforms (Nakata et al., 2002) and the expression of these in the cell lines used and the specificity of siRNA targeting are critical issues. Thus, the significance of ELKS in IKK and NF-κB activation awaits further verification, for example by a genetic proof.

The heat shock protein (Hsp)90 and its cochaperone cell division cycle 37 (Cdc37) copurify with the IKK complex (Chen et al., 2002; Field et al., 2003; Bouwmeester et al., 2004) and were suggested to be stoichiometric components (Chen et al., 2002). Hsp90 together with Cdc37 directly interacts with the kinase domains of IKKα and β. Geldanamycin (GA), a drug that specifically blocks the ATPase of Hsp90, inhibits signal-induced and constitutive IKK activation (Chen et al., 2002; Broemer et al., 2004). Furthermore, it was suggested that GA causes dissociation of Hsp90 and Cdc37 from IKK and inhibits TNFα-induced recruitment of the IKK complex to TNF receptor 1 (TNF-R1) (Chen et al., 2002). However, Hsp90 is often required to prevent misfolding of substrates and subsequent proteasomal degradation. In fact, GA causes receptor interacting protein 1 (RIP1) degradation and might therefore inhibit IKK activation specifically in the TNF pathway (Lewis et al., 2000). While Hsp90 inhibition over several hours causes ubiquitination and proteasomal degradation of IKKα and β during their de novo synthesis, short time inhibition impaired induced or constitutive IKK kinase activation without affecting their steady-state amounts (Broemer et al., 2004). This suggests that Hsp90 is not only required for IKK activation, but also for its homeostasis. Proteasomal degradation of IKK components after extended exposure to GA was also observed by Pittet et al. (2005). The precise role Hsp90 and Cdc37 in the IKK activation process thus remains to be determined. As all functional studies on Hsp90 and Cdc37 so far have been carried out with pharmacological inhibitors, direct manipulation of Hsp90 and Cdc37 levels in cells will be required to further define their role in IKK activity.

Many other proteins have been shown to associate with IKK components in the context of various signaling pathways and conditions, as demonstrated by various methods, including co-immunoprecipitation from transfected cells or immunoprecipitation of endogenous components (Table 2). Further IKK-interacting candidates have been suggested by genomic or proteomic approaches (Bouwmeester et al., 2004; Rual et al., 2005). By far, most of these interactions are likely involved in transitory binding processes in individual functional contexts with the IKK core complex and do not represent stoichiometric components.

Table 2 IKK-interacting proteins

NF-κB activation by two distinct IκB kinase pathways

Two types of NF-κB activation pathways have been discriminated in the past years, which differ in respect to the types of stimuli, the IKK components involved and the targeted NF-κB subunits. The canonical pathway is activated by all physiological NF-κB-inducing agents, including inflammatory cytokines, pathogen-associated molecules and antigen receptors and involves IKKβ- and NEMO-dependent degradation of IκBs and nuclear translocation of mostly RelA containing heterodimeric NF-κB (Silverman and Maniatis, 2001) (Figure 1). The non-canonical or p100 processing pathway is activated by a limited number of stimuli, and selectively requires IKKα, which, dependent on the protein kinase NF-κB-inducing kinase (NIK), induces processing of the precursor p100 to p52 (Beinke and Ley, 2004; Hayden and Ghosh, 2004). Heterodimers of p52, often containing RelB, are the nuclear effectors of the non-canonical pathway and activate a selective group of target genes (Bonizzi et al., 2004) (Figure 1).

The non-canonical, p100 processing pathway

NIK plays a central role in the activation of the non-canonical NF-κB pathway. That is, NIK phosphorylates the T-loop serines of IKKα, which then (without requiring IKKβ or NEMO) phosphorylates p100 at C-terminal serines, to trigger ubiquitination and proteasomal processing of p100 (Senftleben et al., 2001; Xiao et al., 2001b) (Table 1). Several physiological inducers of the non-canonical pathway are known, which mostly belong to the TNF receptor and ligand multi-gene families. Survival and maturation of splenic B cells requires induction of the non-canonical pathway by B cell-activating factor (BAFF) (Claudio et al., 2002; Kayagaki et al., 2002). Processing of p100 is further induced by the Lymphotoxin β (LTβ) receptor (Dejardin et al., 2002; Mordmuller et al., 2003; Muller and Siebenlist, 2003; Yilmaz et al., 2003). The similar knockout mice phenotypes of LTβ receptor and ligand, of NIK, IKKα, p100 and RelB underscore that these form a common signaling pathway in stromal cells, which regulates peripheral lymphoid tissue organogenesis (Weih and Caamano, 2003; Bonizzi and Karin, 2004; Siebenlist et al., 2005). Processing of p100 is further stimulated by CD40 ligation (Coope et al., 2002) or by expression of the related latent membrane protein-1 (LMP1) of Epstein-Barr virus (Atkinson et al., 2003; Eliopoulos et al., 2003; Saito et al., 2003). Receptor activator of NF-κB ligand (RANKL) promotes p52 generation during osteoclastogenesis (Novack et al., 2003). Other examples are TNF family member with weak apoptosis-inducing activity (TWEAK) and Cd27 (Saitoh et al., 2003; Ramakrishnan et al., 2004). In fact, 12 different TNF receptor-associated factor (TRAF)-binding receptors of the TNF receptor superfamily, namely RANK, CD30, LTβR, B cell maturation antigen (BCMA), Herpes virus entry mediator (HVEM), p75TNF-R, transmembrane activator and calcium modulator and cyclophilin ligand interactor (TACI), CD40, CD27, glucocorticoid-induced TNF receptor (GITR), 4-1BB and OX40 have been shown to activate the non-canonical NF-κB pathway upon overexpression (Hauer et al., 2005).

However, LPS or Helicobacter pylori infection can also stimulate the non-canonical pathway in B lymphocytes and immature dendritic cells (DC) (Mordmuller et al., 2003; Saccani et al., 2003; Ohmae et al., 2005). Although a biological significance for the induction of p100 processing in response to LPS remains to be proven, one function could reside in the maturation process of DCs. The non-canonical pathway is also induced by the human T-cell leukemia virus (HTLV-1) Tax and Kaposi's sarcoma-associated herpes virus (KSHV) viral FADD-like interleukin-1-β-converting enzyme (FLICE/caspase 8)-inhibitory protein (vFLIP) proteins (Xiao et al., 2001a; Matta and Chaudhary, 2004; see Hiscott et al., 2006). These two viral proteins can physically recruit IKKα to p100 and require IKKα, but not NIK, to induce p100 processing. Tax, in fact, recruits the entire IKKα–IKKβ–NEMO complex by binding to NEMO (Xiao et al., 2001a). In contrast, induction of processing through the TNF receptor family-members (above) requires NIK in each case investigated. There, p100 processing is not simply the result of activation of IKKα by NIK, since many NF-κB inducers activate both IKKα and IKKβ, but do not induce p100 processing. It is thought that NIK first activates IKKα and then functionally cooperates with IKKα in the induction of p100 processing, for example, by mediating the docking of IKKα to p100 (Xiao et al., 2004). Phosphorylation of p100 causes recruitment of the SCFβTrCP ubiquitin ligase, ubiquitination at Lys-855 (Amir et al., 2004) and proteasomal degradation of the C-terminal part of p100, to generate p52 (Beinke and Ley, 2004).

The inducers of the p100 pathway also stimulate the canonical pathway and both cascades are linked, since the canonical pathway drives expression of p100 and RelB. A characteristic of the non-canonical branch is its slow kinetics of the onset of p100 to p52 conversion, which takes several hours compared to the fast IκB degradation that occurs within several minutes of activation of the canonical pathway. Furthermore, the non-canonical pathway usually results in long-lasting NF-κB activation and is sensitive to ribosomal inhibition, while the canonical pathway is not.

A requirement for de novo protein synthesis indicated that an unstable factor is required and/or that p100 processing is somehow coupled to its translation per se. There are observations supporting both scenarios. The production of p52 by LTβ- or LPS-stimulated p100 processing occurs only during the synthesis phase of p100, when visualized by pulse-chase analysis (Mordmuller et al., 2003) and may thus be confined to nascent, translating p100 molecules, or to a short-lived pool of the de novo synthesized, ribosome-released, but not yet completely matured p100 molecules. Liao et al. (2004) proposed that TRAF3 targets NIK for degradation by the proteasome, by physically associating with NIK. Induction of non-canonical signaling, for example, through CD40 or BAFF receptor, involves the degradation of TRAF3 and the concomitant enhancement of NIK expression. Thus, rescue of NIK from TRAF3-mediated degradation may be an essential step in the non-canonical pathway, and would require de novo synthesis of NIK. In fact, transfected TRAF3 can block or downregulate p52 generation induced by the 12 different TRAF-binding TNF receptors tested (Hauer et al., 2005). TRAF2 has also been shown to negatively regulate the non-canonical pathway in B cells (Grech et al., 2004), while TRAF3 is implicated as a positive regulator of the p100 pathway as well. Thus, further studies are needed to resolve precisely how TRAF2 and TRAF3 control the NIK expression level and p100 processing (see Hauer et al., 2005; Xia and Chen, 2005, for discussion).

The helix-loop-helix protein NF-κB activator 1 (Act1 aka connection to IKK and stress activated protein kinase (SAPK)/c-Jun N-terminal kinase (JNK) (CIKS)) may play an important role in the homeostasis of B cells by attenuating both canonical and non-canonical CD40 and BAFFR signaling. Act1 was originally shown to bind to NEMO (Table 2) and ectopic expression suggests that Act1 activates both IKK and JNK (Leonardi et al., 2000; Li et al., 2000; Mauro et al., 2003). In contrast, gene ablation of Act1 suggests a negative regulation of BAFF and CD40 (Qian et al., 2004). Act1 interacts with TRAF3 and CD40 or BAFFR upon stimulation. Act1 may inhibit CD40 and BAFFR signaling through its interaction with TRAF3 by affecting TRAF3 communication with TRAF2 or by affecting stability of NIK. Interestingly, Act1 and TRAF3 mRNAs are induced by BAFF, CD40L and LPS in B cells, suggesting that both regulators cooperate as a feedback control to dampen signaling for B-cell survival and activation (Qian et al., 2004).

Very little information is available on the biochemical nature of the IKK complex that mediates activation of the non-canonical pathway (Figure 1). There is no direct evidence for the existence of a separate IKK complex, which exclusively contains IKKα homodimers, although it is clear from the analysis of IKKβ or NEMO knockout cells that these two subunits are mostly dispensable for the non-canonical pathway (Bonizzi and Karin, 2004).

IKK regulation by upstream kinases

As one of the mechanisms of IKK activation, upstream kinases can phosphorylate the T-loops of IKKα and/or IKKβ. Alternatively, IKKs could undergo trans-autophosphorylation at these sites upon binding to activating interactors. Several MAP kinases, including NF-κB-inducing kinase (NIK), mitogen-activated protein/extracellular signal-regulated kinase (ERK) kinase kinase 1 (MEKK1), MEKK3, TGFβ-activating kinase 1 (TAK1) and NF-κB activating kinase (NAK) all phosphorylate IKKs and can induce NF-κB activation under in vitro or overexpression conditions. Other proposed upstream kinases include Cot/Tpl-2, the novel protein kinase C (PKCs) isoforms PKCθ, ζ or λ and others (see Ghosh and Karin, 2002; Hayden and Ghosh, 2004, for review). But only a few have been functionally verified by gene ablation experiments. NIK is essential in the non-canonical pathway (above), while inactivation of MEKK3 (Yang et al., 2001) or TAK1 (Sato et al., 2005; Shim et al., 2005, and see below) severely impairs IKK activation by several stimuli in the canonical pathway. However, functional redundancy and possible compensatory deregulations may obscure important functions in knockout studies.

In several instances, it has been recognized that upstream kinases with established functional roles in IKK activation cascades, such as RIP1, interleukin-1 receptor-associated kinase 1 (IRAK1) or protein kinase double-stranded RNA-dependent (PKR) (Hsu et al., 1996; Knop and Martin, 1999; Li et al., 1999a; Bonnet et al., 2000), do not require their kinase activity, but rather act by recruiting other proteins. Against previous expectations, examples for bona fide IκB kinase kinases (IKKKs) with a genetically verified function are now surprisingly limited, despite the large number of different pathways that activate IKK.

Ubiquitin-mediated IKK regulation

The attachment of ubiquitin-polymers to substrates by the sequential action of ubiquitin activating (E1) enzyme, ubiquitin conjugating (E2) enzymes and ubiquitin ligases (E3) determines the fate of the modified protein, depending on the type of isopeptide bond used for ubiquitin-polymerization (Haglund and Dikic, 2005). Lys-48-linked oligomers cause proteasomal destruction, while Lys-63-linked chains usually do not destabilize the substrate and rather serve to modify its activity. Much of the recent interest in the regulation of IKK has focused on non-degradative ubiquitination of IKK and of upstream signalosome components with Lys-63-linked polyubiquitin (UbLys63). Regulation occurs at the level of attachment, binding and degradation of UbLys63 chains (for recent reviews, see Chen, 2005; Krappmann and Scheidereit, 2005). As it turns out, several TRAF molecules, which contain RING finger domains, act as E3 enzymes and synthesize UbLys63 chains. Activation of the E3 activity of TRAFs, thought to be induced by oligomerization triggered upon ligand-receptor interaction, results in trans-auto-ubiquitination of TRAFs as well as in modification of RIP1 and NEMO (Figure 3). Polyubiquitin binding domains (UBDs) then mediate the recruitment of TAK1 and IKK to the ubiquitin-bearing receptor proximal signalosomes, where activation of IKK either by TAK1 or by trans-auto-phosphorylation occurs (Chen, 2005; Krappmann and Scheidereit, 2005). Inhibitory deubiquitinating enzymes (DUBs), such as A20 and cylindromatosis protein (CYLD), which can degrade UbLys63 chains, counteract this activation process. A20 is a dual-function molecule: after removal of UbLys63 chains from RIP1 with its deubiquitinating activity, A20 then ubiquitinates RIP1 with UbLys48 chains by virtue of its ZF domain, which has E3 activity (Wertz et al., 2004). This triggers proteasomal destruction of RIP1 (see below).

Figure 3

Regulation of IKK and NF-κB signaling cascades by non-degradative ubiquitin modification. Ub-ligase (E3) components (red), cellular deubiquitinating components (A20, CYLD, red), Ub-binding domains (UBD) and their presumed interactions with polyubiquitinated targets (grey arrows), as well as activated kinases (blue) are indicated. Other components, which may contribute by either affecting ubiquitination indirectly or by conferring protein interactions are indicated in ovals. Poly-UbLys63 chains and mono-ubiquitination are indicated with black dots. IPA, intracellular pathogens. See text for details.

Ubiquitin ligase activities of receptor-activated signaling molecules

A UbLys63-specific E3 activity was first shown for TRAF6 and specifically requires the ubiquitin-conjugation E2 enzyme Ubc13/Uev1A (Deng et al., 2000). TRAF6 oligomerization greatly stimulates its E3 activity and is enhanced by the TRAF6-interacting protein with a forkhead-associated domain (TIFA) (Ea et al., 2004). The importance of TRAF6 oligomerization has also been demonstrated by the recent identification of β-arrestins as negative regulators of Toll-like receptor (TLR) and IL-1 signaling (Wang et al., 2006). β-Arrestins 1 and 2 bind selectively to TRAF6 after TLR or IL-1R activation and dampen oligomerization and auto-ubiquitination of TRAF6 and the subsequent activation of NF-κB and AP-1. Consistent with an inhibitory role in the NF-κB pathway, LPS-treated β-arrestin 2-deficient mice show an increase in both, inflammatory cytokine expression and susceptibility to endotoxic shock. However, β-arrestins also directly interact with IκBα and block its phosphorylation and degradation (Gao et al., 2004; Witherow et al., 2004). The relative contribution at these two levels upstream and downstream of IKK and the physiological regulation of β-arrestins in NF-κB pathways are not yet fully explored. Finally, sequestosome 1/p62, a scaffolding protein that also binds to TRAF6 and facilitates TRAF6 oligomerization and UbLys63 modification has been shown to be required for the nerve growth factor-induced IKK/NF-κB pathway (Wooten et al., 2005).

While TRAF6 ubiquitinates NEMO in the TLR and IL-1 pathways, TRAF2 is UbLys63-modified and catalyzes the ubiquitination of RIP1 in the TNFα pathway (Wang et al., 2001; Shi and Kehrl, 2003; Kanayama et al., 2004; Wertz et al., 2004) (Figure 3). Gene ablation studies have demonstrated essential functions for TRAF6 in the IL-1, TLR and CD40 pathways and for the functionally redundant molecules TRAF2 and TRAF5 in the TNFα pathway (Hayden and Ghosh, 2004). The E3 function of these TRAF molecules has thus to be considered as an important clue to a signaling role for ubiquitin, although direct genetic evidence is still missing, for example by mutation of the catalytically essential residues in TRAFs or of UbLys63-modified residues, once they are mapped.

In T-cell receptor (TCR) and B-cell receptor (BCR) signaling (see Hayden et al., 2006), all components of the caspase-recruitment-domain-containing membrane-associated guanylate (CARMA1)–Bcl10–mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) (CBM) complex are essential for IKK and NF-κB activation (Thome, 2004) (Figure 3). MALT1 has a UbLys63-specific E3 activity, which is activated by B cell lymphoma 10 (Bcl-10)-mediated oligomerization. As a consequence, MALT1 ubiquitinates NEMO. Mutation of the modified lysine in NEMO abolishes IKK activation (Zhou et al., 2004). Ubiquitination by MALT1 is apparently direct, although MALT1 does not reveal any discernible conserved E3 domain. Interestingly, the IKK complex limits the duration of its activation by phosphorylating Bcl10, which disrupts the Bcl10-MALT1 association and Bcl10-triggered NEMO ubiquitination (Wegener et al., 2006). It has also been shown that Bcl10 and MALT1 can induce the E3 activity of TRAF6. In fact, silencing of TRAF6 and TRAF2 impairs TCR-mediated IKK activation (Sun et al., 2004). According to this model, MALT1 triggers ubiquitination of NEMO through stimulation of the E3 activity of TRAF6 (Figure 3). However, evidence for such a role of the TRAF proteins in TCR signaling by knockout studies is yet missing.

The activation of IKK by TCR further requires caspase-8 to link IKK to the CBM complex (Su et al., 2005). In the absence of caspase-8, ubiquitination of NEMO still occurs, however, without T-loop phosphorylation of IKKα and β. Thus, NEMO ubiquitination per se is not sufficient for IKK activation. Also, 3-phosphoinositide-dependent kinase 1 (PDK1) has been shown to bridge the IKK complex to Bcl10 and MALT1 to facilitate NEMO ubiquitination (Lee et al., 2005). Another regulator that may contribute to the clustering of IKK and CBM is hematopoietic progenitor kinase 1 (HPK1) (Brenner et al., 2005). This kinase can stimulate IKKβ activity by direct phosphorylation and rapidly dissociates from IKK upon TCR stimulation. Interestingly, in primary T cells sensitive to activation-induced cell death, HPK1 is proteolytically processed into a C-terminal cleavage product, HPK1-C. This variant sequesters the inactive IKK complex, but fails to dissociate upon TCR stimulation, resulting in a block of IKK and NF-κB activation. This block is thought to sensitize T cells towards apoptosis upon restimulation (Brenner et al., 2005). TCR-mediated IKK activation is thus a highly dynamic and complex process, and the interplay of the various regulators is just starting to be unraveled.

Site-specific ubiquitination of NEMO seems to be a general phenomenon in IKK signaling: the C-terminal region of NEMO is ubiquitinated at distinct lysines not only by MALT1, but also in the DNA damage pathway (Huang et al., 2003a), or in the course of IKK activation elicited by the intracellular bacterial sensor Nucleotide-binding and oligomerization domain 2 (Nod2) in a RIP2-dependent manner (Abbott et al., 2004) (Figure 3). While these lysines (Figure 2) in NEMO are functionally essential for IKK activation by the respective stimuli, the functional changes triggered by these modifications are yet unknown. The non-degradative Ub chains seem to act as a scaffold for assembling a competent IKK signaling complex.

Ubiquitin-binding domains associated with TAK1 and IKK complexes

In the NF-κB pathways, receptor domains specific for UbLys63 chains were first identified in the homologous transforming growth factor-β activated kinase 1 binding protein (TAB)2 and TAB3 proteins (Kanayama et al., 2004), which together with the structurally unrelated component TAB1 form complexes with TGFβ-activated kinase 1 (TAK1) (Figure 3). A number of observations made with in vitro assays and transfected cells indicated that TAK1 plays a role in the activation of IKK and NF-κB by TNFα, IL-1 or LPS. TAK1 can directly phosphorylate the T-loop serines of IKKβ, is recruited to TRAF2 or TRAF6 upon TNFα or IL-1 stimulation, and activates IKK. Furthermore, combined down modulation of TAB2 and TAB3 with siRNAs impairs TNFα and IL-1 activation of IKK and JNK/p38 (reviewed by Krappmann and Scheidereit, 2005). Genetic evidence for an important function for TAK1 in IKK signaling has been provided recently (Sato et al., 2005; Shim et al., 2005). TAK1-deficient cells show a severe defect in TNFα, IL-1 and LPS induced activation of IKK and JNK. Some residual inducible IKK activation remains in these cells, however, suggesting that TAK1 is partially redundant with other kinases. In contrast, TAK1 is not required for LTβ activation of p100 processing in the non-canonical pathway (Shim et al., 2005), or for B cell receptor induced NF-κB activation (Sato et al., 2005). Interestingly, cells lacking TAB1 or TAB2 do not show any defect in IKK signaling (Shim et al., 2005). This could be explained by functional complementation at least of the related TAB2 and TAB3 proteins, but the requirement of the TAB proteins in IKK signaling clearly requires further investigations.

The ubiquitination sites in the TRAFs have not yet been identified. Recently, it was shown that, in TNFα-stimulated cells, RIP1 is modified with UbLys63 selectively at a single lysine (Lys-377). Mutational analysis indicated that Lys-377 is essential for the recruitment of TAK1 and of the IKK complex to the TNF receptor and for TNFα-induced NF-κB activation (Ea et al., 2006; Li et al., 2006). Moreover, it turned out that NEMO itself is a polyubiquitin binding protein and has a high preference for UbLys63 as compared to UbLys48 chains (Ea et al., 2006; Wu et al., 2006a). Interestingly, the UbLys63-chain on RIP1 mediates independent recruitment of TAB2- and NEMO-containing complexes upon TNFα stimulation (Figure 3) (Ea et al., 2006). Thus, the ubiquitin chain seems to act as a scaffolding interface, which assembles activating kinase and effector kinase in close proximity. However, mutation of Lys-377 in RIP1 abolishes the prior recruitment of RIP1 to the TNF receptor as well and therefore this residue may confer additional functions.

The mapped ubiquitin-binding domain (UBD) in NEMO largely overlaps with the minimal oligomerization domain MOD. The mutation of four single residues (Y308, F312, D311 and L329) in the NEMO CC3/LZ region (Figure 2) impaired UbLys63 chain binding, recruitment of NEMO to ubiquitinated RIP1 and the TNF receptor, as well as IKK activation (Ea et al., 2006; Wu et al., 2006a). Owing to the overlap of UBD and MOD, the IKK inhibiting effects of MOD peptides and of MOD deletions (Tegethoff et al., 2003; Agou et al., 2004) might be explained by their interference with ubiquitin binding. Conversely, ubiquitin binding to NEMO may be dependent on or even affect the oligomeric state of this region. Direct binding of NEMO to Ub was also indicated by the isolation of different Ub species with NEMO in yeast two-hybrid systems (Rual et al., 2005; Wu et al., 2006a). The UBD of NEMO binds preferentially to Lys63-linked chains as compared with Lys48-linked chains, although the degree of discrimination differed between the two studies (Ea et al., 2006; Wu et al., 2006a).

An apparent problem is the specificity of the interaction between UBDs and their ubiquitin-modified targets, if only polyubiquitin chains were recognized. However, additional Ub-independent binding contributions may enhance the specificity. It is at present difficult to reconcile the findings of site-specific UbLys63 modifications of NEMO on the one hand with UbLys63 binding by NEMO on the other, each shown to be functionally essential. Nevertheless, the presence of a ubiquitin binding domain in NEMO, the central regulator of canonical NF-κB signaling, strongly suggests that UbLys63 conjugation may be a generally used regulatory modification in these pathways. The same concept was supported by the prior identification of UbLys63-specific proteases that inhibit IKK activation (see below).

Negative control of IκB kinase signaling

Regulation of IKK by auto-phosphorylation and phosphatases

An important aspect of IKK regulation is the mechanism that determines the rapid inactivation following initial stimulation. While IKK activation is correlated with phosphorylation at the T-loop serines, post-inductive attenuation involves C-terminal phosphorylation (Karin, 1999). Once activated, IKKβ progressively phosphorylates a serine cluster C-terminal to the HLH, which results in inactivation of IKKβ. It was proposed that this inactivating phosphorylation results in a conformational change that may loosen an intramolecular interaction of the C-terminal HLH region with the KD (Delhase et al., 1999). C-terminal phosphorylation in IKKβ can be subdivided into autophosphorylation of serines close to the HLH and phosphorylation of serines in and around the NEMO binding domain (Schomer Miller et al., 2006). According to this recent study, phosphorylation of the serine cluster as whole has no inactivating effect, while NBD phosphorylation inhibits T-loop phosphorylation. Therefore, NBD phosphorylation might influence the precise mode of interaction between IKKβ and NEMO (May et al., 2002). So far, it is not clear whether IKKβ or another kinase regulates NBD phosphorylation.

Dephosphorylation of the T-loops of IKKβ by phosphatases like PP2A or PP2Cβ is another post-inductive mechanism for IKK deactivation (DiDonato et al., 1997; Prajapati et al., 2004). In fact, IKK forms cellular complexes with the phosphatases PP2A and PP2Cβ (Prajapati et al., 2004; Kray et al., 2005). The PP2A binding site is close to the kinase-binding region of NEMO (Kray et al., 2005). Deletion of the site impairs Tax- and TNF-induced activation of IKK, suggesting a positive role. However, it cannot be excluded that the effect of the internal deletion was caused by altering the conformation of NEMO. Silencing of PP2Cβ expression altered the kinetics of TNFα-induced IKKβ activity, in a way consistent with a role in signal termination (Prajapati et al., 2004). It is not clear yet whether PP2Cβ acts directly on IKKβ or on an upstream kinase, such as TAK1. The precise roles of auto-phosphorylation and dephosphorylation of IKK thus remain to be further determined.

Negative regulation of IKK by chaperones and Hsps

While Hsp90, Cdc37 and FK506-binding protein 51 kDa (FKBP51) are positively implicated in IKK regulation (Chen et al., 2002; Bouwmeester et al., 2004; Broemer et al., 2004) (see above), several other Hsps have been suggested as negative regulators of IKK. Heat shock or increased cellular Hsp70 levels inhibit NF-κB activation, although the precise mechanism is enigmatic. Ran et al. (2004) showed that Hsp70 specifically binds to NEMO and that its overexpression inhibits IKK activity. In turn, Hsp70 siRNA reduces heat-induced IKK inhibition. Hsp70 binds to the MOD in NEMO and disrupts the tetrameric scaffold, resulting in formation of NEMO-Hsp70 heteromers and downregulation of IKK activity (Ran et al., 2004). However, it remains to be seen whether NEMO oligomerization can be disrupted by Hsp70 upon heat shock in vivo. As an interesting further possibility, by binding to NEMO, Hsp70 might interfere with ubiquitin binding or ubiquitin modification, both of which involve the MOD (see above).

Another Hsp, Hsp27, was reported to inhibit IKK activity as well (Park et al., 2003). Hsp27 interacts with endogenous IKKα and IKKβ. TNFα treatment increases the association of Hsp27 with IKKβ, curiously without changing its association with IKKα. Hsp27 overexpression inhibits, while small interfering RNA (siRNA) enhances TNFα induced IKK and NF-κB activity.

hTid-1 is a human homologue of the Drosophila tumor suppressor l(2)Tid and a novel DnaJ protein, which can interact with Hsp70. hTid-1 associates with both IκBs and the IKK complex and expression inhibits, while knockdown enhances TNFα-induced IKK and NF-κB activities in transfected cells (Cheng et al., 2005). Possibly hTid-1 acts at the level of IKK-substrate interaction and its precise mechanism of action is yet to be established.

From these studies it is not yet clear which of the different Hsps and chaperones are associated with endogenous IKK complexes at a given time. Also, positively and negatively acting chaperones might work through a common mechanism, for example, by mutual competition for substrate (IKK) binding.

Negative regulation of IKK and NF-κB pathways by deubiquitinating enzymes

Another potent mechanism to downregulate IKK activation is by deubiquitination of UbLys63 conjugated TRAFs or NEMO. The Lys63-specific ubiquitin C-terminal deubiquitinase CYLD was identified via its interaction with NEMO in a two hybrid screen and by searching for DUBs which interfere with NF-κB signaling (Brummelkamp et al., 2003; Kovalenko et al., 2003; Trompouki et al., 2003). Structural analyses have shown that CYLD binds with one of its cytoskeleton-associated protein-glycine conserved CAP-Gly domains to one of the two proline-rich sequences located between the MOD/UBD and ZF domains of NEMO (Saito et al., 2004). In transfected cells, CYLD interacts with NEMO and TRAFs, selectively degrades UbLys63 chains from NEMO, TRAF2, TRAF6 or TRAF7, and represses activation of IKK and NF-κB by various stimuli. Conversely, RNAi knockdown of CYLD enhances the induction of NF-κB and TRAF-ubiquitination (Brummelkamp et al., 2003; Kovalenko et al., 2003; Regamey et al., 2003; Reiley et al., 2005; Trompouki et al., 2003; Yoshida et al., 2005). However, CYLD has also been shown to negatively regulate MAPK activation. Its relative contribution to IKK/NF-κB versus MAPK signaling is a controversial issue (Reiley et al., 2004; Yoshida et al., 2005).

Interestingly, signal-induced TRAF2-ubiquitination is associated with site-specific phosphorylation of CYLD that prevents CYLD from inhibiting the ubiquitination of TRAF2 (Reiley et al., 2005). Phosphorylation of cellular CYLD by diverse IKK-inducing stimuli is dependent on cellular NEMO, can be stimulated by transfected IKKα or β and is apparently direct. The sites phosphorylated by IKK were mapped to a serine-rich cluster in CYLD (Table 1). It is thus an interesting notion that IKK negatively regulates the DUB activity of CYLD, its own inhibitor. Despite such observations with endogenous CYLD and the results of RNAi experiments, much of the evidence for an IKK-regulating function of CYLD has been obtained from experiments with transfected cells. Thus, it is unclear to what extent CYLD regulates IKK/NF-κB in vivo. Recently, one knockout of CYLD revealed an unexpected function specific for T cells (Reiley et al., 2006): CYLD regulates proximal TCR signaling in thymocytes by selectively binding to and deubiquitinating the active form of the kinase Lck. Interestingly, both Lys 48- and Lys 63-linked polyubiquitin chains were degraded. In line with an essential role in TCR signaling, CYLD-deficient mice display reduced numbers of CD4(+) and CD8(+) single-positive thymocytes and peripheral T cells (Reiley et al., 2006). However, surprisingly, activation of IKK and NF-κB or of ERK, JNK and p38 kinases by Toll-like receptors or TNF receptors was not affected in the absence of CYLD. As these pathways were analysed only in bone marrow-derived macrophages, CYLD may regulate IKK signaling in other cell types or in a receptor-specific manner. Functional redundancy of CYLD with other DUBs may be a further explanation.

Originally CYLD was identified as a tumor suppressor mutated in familial cylindromatosis (Bignell et al., 2000; Courtois and Gilmore, 2006). Affected patients develop benign tumors originating from the skin appendages. There, NF-κB is expected to be activated and to drive proliferation through induction of sonic hedgehog (Shh) and cyclin D1 expression (Schmidt-Ullrich et al., 2006). Thus, a tumor suppressor function for CYLD by limiting NF-κB activation in these cell types is plausible. However, cylindromatosis patients do not show any other pathologies as would be expected for a systemic, unbalanced activation of IKK or MAPK signaling. A second knockout study indeed revealed a tumor suppressor function for CYLD in skin, which surprisingly affects B cell lymphoma 3 (Bcl-3), rather than NF-κB p50/RelA activation (Massoumi et al., 2006). CYLD is needed to prevent high susceptibility to chemically induced skin tumors and hyperproliferation of keratinocytes exposed to phorbolester or UV irradiation. These treatments trigger the translocation of CYLD from the cytoplasm to the nuclear periphery, where CYLD binds to and deubiquitinates UbLys63-modified Bcl-3 (Figure 3). Nuclear Bcl-3 is known to recruit the histone acetyltransferase Tip60 to p50 or p52 homodimers and to activate genes like KAI1 or cyclin D1 (Dechend et al., 1999; Westerheide et al., 2001; Baek et al., 2002) (Figure 1) (see Basseres and Baldwin, 2006). By reversing Bcl-3 ubiquitination, CYLD prevents nuclear accumulation of Bcl-3, stimulation of cyclin D1 expression and proliferation (Massoumi et al., 2006). While CYLD-deficient keratinocytes reveal enhanced IKK- and NF-κB-responses towards TNFα, the regulation of Bcl-3 seems to be IKK-independent. Thus, the study shows that UbLys63-modification also directly regulates the nuclear effectors in the NF-κB system. But it remains to be clarified, how UbLys63-modification of Bcl-3 is achieved and how it promotes nuclear translocation and perhaps nuclear activity.

A potent feedback inhibitor of NF-κB signaling in the TNFα pathway is A20. The knockout mouse has established that A20 limits inflammatory responses by termination of TNFα-induced activation of IKK and NF-κB (Lee et al., 2000). The DUB activity of A20 negatively regulates TNF-R and TLR signaling by proteolysis of UbLys63-chains on RIP1 and TRAF6 (Boone et al., 2004; Wertz et al., 2004). While this process is mediated by the N-terminal cysteine protease domain, a Lys48-specific E3 ligase activity of the C-terminal ZF domain attaches degradative chains onto RIP1, which is thereupon destroyed by the proteasome (Wertz et al., 2004). UbLys63-chains have to be lysed, before UbLys48-chains can be conjugated (Wertz et al., 2004) and a possible reason might be that both chains are synthesized on Lys-377 of RIP1 (Ea et al., 2006). Interestingly, NEMO can protect RIP1 against TNFα-induced proteasomal degradation (Wu et al., 2006a). It is possible that by binding to UbLys63-modified RIP1, NEMO prevents access of A20.

Finally, bacteria have evolved very efficient strategies to circumvent innate immune reactions of infected organisms. The virulence factor and cysteine protease Yersinia outer protein J (YopJ) of Yersinia bacteria is known to inhibit host cell proinflammatory reactions by blocking both MAPK and NF-κB pathways (Orth et al., 2000). As it turns out, YopJ is a deubiquitinating enzyme, too. In transfected cells, YopJ displays broad target specificity and removes not only UbLys63-conjugates from TRAF2 and TRAF6, but also UbLys48-chains from IκBα and thereby inhibits NF-κB activation at two levels, activation of IKK and proteasomal degradation of IκBα (Zhou et al., 2005).

In summary, several mechanisms may account for post-induction attenuation of IKK activity. Phosphatases like PP2A and PP2C may play a role, or Hsps like Hsp70 or hTid-1. The expression of A20 or CYLD is induced by NF-κB and this may provide an autoregulatory negative feedback mechanism to limit the duration of IKK activation (Jono et al., 2004). However, this cannot easily explain the fast inactivation (Cheong et al., 2006). Autophosphorylation of IKK at its C terminus or another yet unknown mechanism associated with the ubiquitin-dependent activation process may be the prevailing determinant of fast downregulation of IKK activity.

Distinct physiological functions of IKKα and IKKβ

From gene targeting studies, it is well documented that IKKα, IKKβ and NEMO underlie different regulation and carry out distinct physiological functions (see Ghosh and Karin, 2002; Bonizzi and Karin, 2004; Gerondakis et al., 2006, for detailed discussion). Ample evidence supports the model that both IKKβ and NEMO are both essential for NF-κB activation by inflammatory cytokines and microbial pathogens, as well as for preventing TNFα-induced hepatic apoptosis, as expected for canonical NF-κB signaling (see Gerondakis et al., 2006). In contrast, IKKα has several unique physiological roles, which involve NF-κB-independent, as well as NF-κB-dependent processes. IKKα controls epidermal differentiation and skeletal morphogenesis independent of its kinase activity and of NF-κB activation. In contrast, proliferation of mammary epithelium during lactation, activation of the non-canonical p100 pathway and negative control of inflammatory signaling involve activation of the IKKα kinase function and modulation of NF-κB activity.

The IKKα knockout mouse shows a skin abnormality that is caused by defective keratinocyte differentiation and proliferation (Hatada et al., 2000, for review). This phenotype can be rescued with a catalytically inactive IKKα mutant and involves the NF-κB-independent production of a not well-defined soluble factor (Hu et al., 2001). IKKα knockout mice also display abnormal skeletal and craniofacial morphogenesis, and these defects could be largely rescued upon re-expression of IKKα or of a kinase-dead IKKα variant in skin (Sil et al., 2004). In a way independent of its kinase activity, IKKα inhibits the expression of the morphogen fibroblast growth factor 8 (FGF8). This is thought to be one of the mechanisms by which epidermal IKKα controls mesoderm-derived morphogenesis. Intriguingly, an intact NLS in IKKα appears to be critical for its function in keratinocyte differentiation (Sil et al., 2004).

In tooth development, morphogenic functions of IKKα are exerted by NF-κB- dependent as well as NF-κB independent processes (Ohazama et al., 2004). IKKα mediates cusp formation of molars through the activation of the NF-κB pathway. However, at an early step of incisor formation, IKKα determines the direction of epithelial in-growth in an NF-κB-independent fashion. In this process, IKKα affects the spatial and temporal expression of Notch1 and 2, Wnt7b and Shh (Ohazama et al., 2004).

There are a number of fundamental questions emerging from these studies describing kinase-independent modes of action of IKKα. The nature of both its upstream regulators and immediate downstream effectors is unknown. Also, it is not established whether IKKα operates as part of a signaling complex, or by directly controlling nuclear gene transcription.

Specific functions for IKKα in p100 processing (see above) and mammary gland differentiation, which depend on its kinase activity, were revealed in mice expressing a non-responsive T-loop serine mutant of IKKα (IKKαAA/AA) (Cao et al., 2001; Senftleben et al., 2001). In mammary gland development during pregnancy, this IKKα mutation results in the same defects as disruption of RANK signaling or of cyclin D1 expression. IKKα may function as part of the classical IKK complex to confer RANK ligand-induced NF-κB activation, which in turn activates cyclin D1 expression and proliferation of mammary epithelial cells (Cao et al., 2001).

Another important function specific for IKKα is to limit the duration and amplitude of pathogen-induced IKK->NF-κB signaling. IKKα dampens the induction of proinflammatory and antiapoptotic genes in LPS-exposed macrophages. This contributes to the resolution of inflammation, as was shown with mice expressing the IKKαAA/AA mutant (Lawrence et al., 2005). Activated IKKα triggers C-terminal phosphorylation of RelA (at Ser536) and of c-Rel, resulting in increased proteasomal degradation rates of these factors. As a consequence, the recruitment time of RelA or c-Rel to target genes is shortened and gene expression alleviated. The degradation of RelA and c-Rel promoted by IKKα presumably occurs in the nucleus (Lawrence et al., 2005). Similarly, it has been proposed that chromatin-bound RelA is subject to ubiquitination and proteasomal degradation in the course of TNFα-stimulated NF-κB activation (Ryo et al., 2003; Saccani et al., 2004). Enhanced NF-κB activation and cellular function were also seen in macrophages lacking IKKα (Li et al., 2005). However, in this study a different mechanism was suggested. LPS-treated IKKα-deficient macrophages revealed accelerated IκBα degradation, but no alteration of RelA stability or of phosphorylation at Ser-536. The reason for the discrepancies regarding RelA phosphorylation and degradation is not clear. They may in part be due to inherent differences between the complete absence of IKKα and the inactivation exclusively of its T-loop serines in the two studies. An inhibitory IKKα function was also proposed for its homologue in zebrafish, Ikk1, the downmodulation or overexpression of which inversely correlated with NF-κB activity and expression of NF-κB-regulated genes (Correa et al., 2005).

Thus, IKKα contributes by additional mechanisms to the transient nature of canonical IKK->NF-κB signaling. Apart from controls that mediate the fast post-inductive decline of IKKβ activity by trans-autophosphorylation and by other negative regulators (see above), IKKα directly limits the time course of NF-κB activity. This regulation complements the potent negative feed-back established by autoregulated IκBα expression (Hoffmann et al., 2002).

Cross talk of IKK components with NF-κB-independent pathways

One emerging concept is that IKKs regulate substrates other than IκBs, NF-κBs or precursors and mediate additional functions beyond NF-κB activation. As discussed above, genetic studies have shown that IKKα controls epidermal differentiation and skeletal or tooth morphology in processes independent not only of NF-κB, but also of its kinase activity. Examples, where the kinase activity of IKK is involved, are nuclear substrates, such as estrogen receptor α (ERα), histone H3, silencing mediator for retinoic acid and thyroid hormone receptor (SMRT) or steroid receptor co-activator 3 (SRC-3) (Table 1). However, the genes affected were largely known NF-κB targets (see below). A potential cross talk of NEMO or IKKs with other NF-κB-independent pathways has been indicated in several recent studies.

The hypoxia-inducible factors 1α (HIF-1α) and 2α (HIF-2α) are transcription factors that mediate cellular responses to low oxygen. Specifically, HIF-2α has been shown to bind to NEMO, and NEMO knockdown diminishes transcriptional activity of HIF-2α (Bracken et al., 2005). NEMO appears to facilitate CREB binding protein (CBP)/p300 recruitment to HIF-2α and stimulates HIF-2α transcription activity. This NF-κB pathway-independent function presumably involves nuclear NEMO. However, functional data with endogenous regulators, including NEMO-dependent gene expression profiling under hypoxic conditions, are needed to confirm this hypothesis.

Another example for an NF-κB-unrelated function of IKK is the inactivation of the FOXO3A forkhead transcription factor. FOXO3A attenuates proliferation and is thought to be normally under the control of protein kinase Akt. Activation of the Akt kinase in the phosphoinositide 3-kinase (PI3K) pathway results in the phosphorylation of FOXO proteins, causing their nuclear exclusion and degradation. Hu et al. (2004) have found that the nuclear FOXO3A expression level inversely correlates with IKKβ expression in primary breast cancer cells. The activated IKK complex binds to and phosphorylates FOXO3A. The substrate sites have similarity to the destruction boxes in IκBs (Table 1) and phosphorylation results in ubiquitination and proteasomal degradation of FOXO3A (Hu et al., 2004). The targeting of FOXO3A is another way by which activated IKK may positively affect cellular proliferation. The precise cellular regulation will likely involve cell type specific controls and a more complex interaction between the FOXO and IKK/NF-κB systems. Indeed, FOXO3A is also engaged in a negative feed-back control to attenuate NF-κB activity by stimulating the expression of IκB family members in T cells (Lin et al., 2004).

NF-κB-independent actions of IKKs were also described for mRNA stability regulation or tyrosine kinase signaling: an AU-rich element of the mRNA of the β1,4-galactosyltransferase 1 (β4GalT1) gene is bound by a complex of tristetraprolin (TTP) and the chaperone 14-3-3β, resulting in mRNA destabilization. In TNFα-stimulated cells, both IKKβ and PKCδ phosphorylate 14-3-3β on serines, which causes release of TTP and 14-3-3β and mRNA stabilization (Gringhuis et al., 2005). Docking protein 1 (Dok1) is an abundant Ras-GTPase-activating protein-associated tyrosine kinase substrate that promotes cell migration and attenuates cell growth. In response to TNF, IL-1 or γ irradiation, IKKβ phosphorylates Dok1 at several serines and modifies Dok1 function (Lee et al., 2004). In insulin receptor signaling, phosphorylation of insulin receptor substrate-1 (IRS-1), another Dok family member, by IKK (Gao et al., 2002) is thought to inhibit tyrosine phosphorylation of IRS-1 by the insulin receptor and to impair metabolic insulin signaling pathways.

It is yet unknown which of these proposed functions of NEMO or IKKs can be substantiated by physiological evidences. However, given the multiple facets of NEMO/IKK regulation that have become apparent in the past years, more indications for an involvement in NF-κB-independent processes will be no surprise.

Nuclear functions for IKKs

Post-translational modifications of NF-κB subunits play an essential role in the control of transcription activity in the nucleus and especially phosphorylation and acetylation of RelA is of crucial importance (Chen and Greene, 2004) (see also Perkins, 2006). Likewise, phosphorylation and acetylation of histone proteins correlates with the recruitment of NF-κB to its target genes and is required for the transcriptional activation of proinflammatory genes (Saccani et al., 2002). Several reports have proposed an additional layer of regulation through which IKKα has an impact on inducible gene regulation by affecting chromatin dynamics. A first indication consistent with a nuclear role was the observation that IKKα accumulates in the nucleus upon treatment of cells with the export inhibitor leptomycin B, suggestive of nuclear shuttling of IKKα (Birbach et al., 2002). It was then shown that TNFα induces nuclear accumulation of IKKα (Anest et al., 2003; Yamamoto et al., 2003). In TNFα-induced cells, histone H3 phosphorylation and acetylation was rapidly induced, but this was completely abolished in cells lacking IKKα. Using chromatin immunoprecipitation (ChIP), both studies showed that TNFα stimulates a rapid recruitment of IKKα to the IκBα promoter region and further to the IL-6 (Anest et al., 2003) and IL-8 genes (Yamamoto et al., 2003), respectively. IKKα is recruited independently of RelA, but appears to act in concert with RelA and CBP, which are recruited with the same kinetics upon TNFα stimulation. The studies suggest that IKKα directly phosphorylates histone H3, although it is not known whether other kinases may be involved in vivo. Inherent differences were seen when comparing IKKα and IKKβ: IKKβ neither underwent nuclear shuttling and TNFα-induced nuclear import, nor was it required for H3 phosphorylation (Birbach et al., 2002; Anest et al., 2003; Yamamoto et al., 2003).

The biological significance of the nuclear IKKα function is controversial, primarily because it is not understood why IKKα contributes to TNFα-induced gene expression in the cultured cells used in these or other studies (Li et al., 2002; Massa et al., 2005), whereas both in IKKα knockout mice or mice with catalytically inactive IKKα the induction of proinflammatory gene responses by TNFα, IL-1 or LPS in general appears to be normal (Cao et al., 2001). IKKα even suppresses activation of inflammatory gene expression, as was shown recently with mice expressing an activation-deficient IKKα variant (Lawrence et al., 2005).

However, nuclear functions for IKKα have also been indicated in other contexts. The induction of keratinocyte differentiation was associated with elevated levels of nuclear IKKα. A functional nuclear localization signal present in IKKα, but not IKKβ (Sil et al., 2004) (Figure 2) was required to induce differentiation of complemented Ikka−/− keratinocytes. Although this indicates that IKKα acts in the nucleus to induce keratinocyte differentiation, in this case it is independent of the kinase activity of IKKα (see above) (Sil et al., 2004).

Some NF-κB-regulated genes may be in a repressed state in unstimulated cells, where binding sites are occupied by p50 homodimers, acting as a platform to assemble corepressor complexes, containing histone deacetylases (HDAC). Upon stimulation, a remodeling of these complexes relieves repression and allows access of activators, such as Bcl-3 or RelA and recruitment of histone acetyltransferases (HAT) (Baek et al., 2002). IKKα has been implicated in chromatin remodeling at the inhibitor of apoptosis protein (cIAP2) and IL-8 gene promoters by phosphorylation of SMRT, which triggers an exchange of corepressor for co-activator complexes (Hoberg et al., 2004). Laminin/integrin signaling stimulated the nuclear accumulation of IKKα, and to some extent that of IKKβ, but not of NEMO. Laminin rapidly induced recruitment of IKKα and IKKβ to the cIAP2 and IL-8 genes, which inversely correlated with chromatin binding of SMRT and HDAC3. According to the proposed model, activated IKKα directly phosphorylates SMRT on the promoter, which results in gene derepression by displacement of the SMRT–HDAC3 complex, allowing subsequent recruitment of RelA, p300 and PolII. However, in this case histone H3 phosphorylation was not affected upon IKKα depletion (Hoberg et al., 2004). It was also shown by this group that, simultaneously with SMRT-phosphorylation, chromatin-bound, activated IKKα phosphorylates RelA at serine 536 in the transactivation domain (Table 1), which contributes to corepressor release and allows subsequent acetylation of RelA by p300 (Hoberg et al., 2006). Interestingly, the laminin- or TNFα-induced recruitment of RelA and IKKα to the target genes is not static, rather these molecules cycle on and off during the course of NF-κB-driven transcription.

It is not understood which processes determine whether RelA phosphorylation at Ser-536 results in an increased transcription activation function or triggers RelA degradation (Lawrence et al., 2005).

As a puzzling observation, IκBα was recruited to the cIAP2 gene upon stimulation, but this was unrelated to SMRT-mediated repression. IκBα was also proposed to be constitutively chromatin-associated with the Notch-regulated hes1 gene, which is not a known NF-κB target gene (Aguilera et al., 2004). There, it was suggested that TNFα stimulation triggers recruitment of both, IKKα and IKKβ and release of IκBα, correlating with increased histone acetylation and transcription. It remains to be seen how the transcriptional regulation of genes like cIAP2 or hes1 is affected under physiological conditions in IKK-deficient mice and to what extent the observed effects are stimulus- or cell type-dependent.

An early evidence that IKKs could be engaged in the regulation of chromatin dynamics was indicated by the copurification of all three IKK components with steroid receptor co-activator 3 (SRC-3) (Wu et al., 2002). Experiments with transfected cells indicated that IKKs phosphorylate SRC-3 in vitro and synergize with SRC-3 in reporter gene activation (Wu et al., 2002, 2004). IKKα has then been shown to cooperate with estrogen receptor α (ERα) and SRC-3 in the transcription activation of estrogen-responsive genes, including cyclin D1 and c-myc (Park et al., 2005). IKKα can be coimmunoprecipitated with ERα and SRC-3 in solution. Estrogen treatment stimulated the rapid recruitment of IKKα (but not of IKKβ), ERα and SRC-3 to the responsive genes and increased IKKα phosphorylation of ERα, SRC-3 and histone H3. The IKKα-mediated phosphorylations of SRC-3 and ERα occur at residues, which are required for enhancing the transcriptional activity of these molecules (Table 1). A contribution of IKKα to estrogen-induced proliferation is at least in part compatible with the previous demonstration that mice with a kinase-defective IKKαAA knock-in allele have impaired proliferation of mammary epithelium (Cao et al., 2001). But in these mice, IKKα activity is needed for RANK ligand-induced cyclin D1 expression by activation of NF-κB, while for estrogen-induced gene expression the IKKα effects are NF-κB-independent. Perhaps IKKα regulates cyclin D1 transcription by distinct effector transcription factors, depending on the cell type or stimulus. A completely unexpected level of control was proposed by the observations that IKKα physically associates with cyclin D1 protein, phosphorylates cyclin D1 and enhances its turnover and nuclear export (Kwak et al., 2005). How these effects are linked to a proproliferative role of IKKα is not understood. Another mechanism by which IKKα may contribute to estrogen-induced cell-cycle progression is through its regulation of E2F1 expression, apparently independent of cyclin D1 gene regulation (Tu et al., 2006). Estrogen treatment can promote the association of endogenous IKKα and E2F1 and the corecruitment of both to the E2F1 or thymidine kinase 1 (TK1) gene promoters.

Growth factor stimulation was shown to regulate nuclear IKKα as well. Recruitment of IKKα to the c-fos gene promoter region is enhanced upon EGF stimulation, and EGF-induced H3 phosphorylation requires the presence of IKKα (Anest et al., 2004). RelA is constitutively associated with the c-fos gene and contributes to its inducibility. However, induced IKKα recruitment occurs in the absence of RelA.

In summary, while IKKα recruitment has been proposed to functionally correlate with the induced expression of several genes (IκBα, IL-6, IL-8, cIAP2, cyclin D1, E2F1, hes1, c-fos), IKKβ recruitment has not and its chromatin-association has been seen only on a subset of the genes investigated (IκBα, IL-8, cIAP2, hes1). IKKα recruitment to the genes was stimulated by TNFα, laminin, estrogen or EGF, but intriguingly, was not yet shown for any of the numerous other well-known IKK stimuli. NEMO was rarely analysed for recruitment to chromatin and theoretically its detection may be limited by accessibility of these proteins to the available antibodies in the chromatin context or under the cross-linking conditions used for the chromatin immunoprecipitation protocols.

It has been shown that a fraction of cellular NEMO is present in the nucleus and that leptomycin B treatment indicates nuclear-cytoplasmic shuttling of NEMO (Verma et al., 2004). According to this study, NEMO can compete with IKKα or p65 for binding to the co-activator CBP in vitro and inhibit reporter gene activation in transfected cells. Such a repressor function for NEMO, however, needs to be verified in vivo.

With the discovery of functional roles of IKKs in the nucleus, a number of important questions are emerging. There is confusion about whether or not nuclear abundance of IKKα is stimulated by TNFα and how much nuclear IKKα is present in unstimulated cells (Anest et al., 2003; Yamamoto et al., 2003; Verma et al., 2004). Inconsistent effects were seen with leptomycin B under the different conditions used, indicating nuclear shuttling of IKKα (Birbach et al., 2002) or of NEMO, but not of IKKα (Verma et al., 2004). Thus, it is not understood, how the cytoplasmic signals elicited by TNFα or other stimuli are relayed to nuclear IKKs and how these are activated. Future studies will have to address the composition of soluble nuclear IKK complexes and their relation to the cytoplasmic IKKs. Also, it remains to be investigated whether IKKs are recruited to discrete gene regions, how gene-specific association of IKKs is achieved and through which DNA-bound or indirectly recruited components sequestration of IKKs occurs. Some of the studies implicate IKK recruitment along with transcription factors or co-factors (e.g. with ERα, E2F1 or SRC3, see above) and in fact several other proteins that have been suggested to associate with IKK subunits are nuclear transcription regulators (e.g., CBP, FOXO3A, HIF-2α, interferon regulatory factor-7 (IRF-7), see Table 2). It will be interesting to see what influence nuclear IKKs can have on stimulus-specific gene expression profiles and most importantly, whether these functions can be confirmed at a genetic level.

IKK activation by nuclear signals in the DNA damage response

An important nuclear function for NEMO has been described in the activation of IKK signaling in response to DNA damage. It is well documented that genotoxic stresses, which are caused by irradiation or topoisomerase I- or II-inhibiting drugs, activate cytoplasmic IKK and NF-κB. The signals that are induced upon generation of DNA double-strand breaks by these treatments originate from the nucleus and have to be processed by the cytoplasmic IKK complex. A crucial sensor for DNA double-strand breaks is the nuclear kinase ATM, which is rapidly activated by genotoxic stress and activates IKK and NF-κB (Piret et al., 1999; Li et al., 2001). How nuclear DNA-damage signaling may be relayed to the cytoplasm has been revealed by Huang et al. (2003a). It turned out that genotoxic stress causes modification of NEMO with small ubiquitin-like modifier (SUMO-1), a ubiquitin-like polypeptide often conjugated to nuclear substrates (Hay, 2005). SUMO-1 attachment results in nuclear retention of NEMO, but not of IKKα or IKKβ (Huang et al., 2003a). The responsiveness to DNA-damage signals specifically requires the ZF of NEMO (Huang et al., 2002). However, the SUMOylation sites of NEMO are at Lys-277 and Lys-309, outside of the ZF (Figure 2) (Huang et al., 2003a). In the nucleus, NEMO is subsequently de-SUMOylated and ubiquitinated at the same residues. Ubiquitination requires the prior phosphorylation of NEMO by ATM. According to their model, ubiquitinated NEMO is then re-exported to the cytoplasm, where it activates IKKα- and β-containing complexes by an unknown mechanism. The authors claim an involvement of ELKS in this process (Wu et al., 2006b). ATM phosphorylates NEMO selectively at serine 85 (Figure 2) and this is required for subsequent mono-ubiquitination, but not for SUMO-modification of NEMO (Wu et al., 2006b). NEMO and ATM associate in an inducible manner and a small fraction of ATM is exported along with NEMO to the cytoplasm, indicating a cytoplasmic function for ATM in IKK activation.

A role in cytoplasmic IKK activation by genotoxic stress has also been described for RIP1. Fibroblasts deficient in RIP1 are refractory to DNA damage-induced IKK activation and RIP1 forms a complex containing IKKβ in response to genotoxic agents (Hur et al., 2003). RIP1-mediated IKK activation is independent of TRAF2 and TRAF5 or TNF receptor-associated death domain (TRADD) in the TNF receptor complex, but is abrogated in ATM-deficient cells. Although this indicates a specific role for RIP1 in DNA-damage signaling, in addition to its function in TNF signaling, the connection to nucleus-derived stress signals is enigmatic. How RIP1 might play a role in a SUMOylation-controlled nucleus-to-cytoplasm signaling process was shown with the recent identification of a function for the death domain-containing protein p53-inducible death-domain-containing protein (PIDD) in genotoxic stress signaling. After induction of DNA damage, PIDD accumulates in the nucleus and forms a complex with RIP1 and NEMO, which are both found in significant amounts in the nucleus before stimulation (Janssens et al., 2005). In this complex, RIP1 bridges PIDD to NEMO (Janssens et al., 2005) and facilitates SUMOylation and ubiquitination of NEMO (Figure 3).

Together, these studies thus identified a nuclear NEMO complex as a recipient for DNA damage-induced modifications, which are essential for subsequent IKK and NF-κB activation. Nevertheless, it is difficult to envision how exported modified NEMO activates cytoplasmic IKK complexes. Also, it will be interesting to see which ATM-independent signals trigger SUMOylation of NEMO and which process activates PIDD. The enzymes that modify NEMO with SUMO-1 and ubiquitin, once identified, may provide tools to substantiate these findings.


It has become clear that the complementary investigation of IKK signaling by biochemical methods and by gene targeting has provided consistent views of many of its molecular and physiological aspects. However, a number of important fundamental questions concerning the composition, structure and activation mechanisms of IKK complexes still must be answered.

One of the most pressing problems concerns the structure and composition of IKK complexes. Many studies on canonical IKK signaling are consistent with the existence of only one tripartite IKKα–IKKβ–NEMO complex, which undergoes multiple, transient interactions with scores of regulators and effectors. However, distinct IKKα subcomplexes have been proposed for the non-canonical pathway or have to be postulated for IKKα, IKKβ and NEMO, given the suggested non-equivalent nuclear engagements of these components. An additional complication emerges from the proposed presence of further stoichiometrically associated molecules in IKK complexes, including chaperones and additional scaffolds. Likewise, the expression of IKKα, β and NEMO in early development or in other situations may not be equivalently regulated, perhaps giving rise to unique complexes. It is not known which processes control the homeostasis and the fate of IKK complexes. More insight into the structure of IKK complexes may also be required to understand the molecular mechanisms by which non-destructive ubiquitin modification and ubiquitin-binding regulates IKK activation. Findings from the last years have revealed an unanticipated diversity of IKK- and NF-κB-regulation by ubiquitin and ubiquitin-like molecules. How the specificity underlying these modifications and their recognition is reached in the different pathways is not yet clear. The provision of supportive unequivocal genetic evidences for a role of ubiquitin- of SUMO-1-modifications in IKK signaling will remain an important task.

Other open questions concern the mechanisms that account for the duration and amplitude of IKK activation. Although several scenarios have been described, including trans-autophosphorylation, the action of phosphatases and others, no unifying concept has yet been provided. Elucidation of this issue is crucial for understanding the basis for transient versus persistent IKK and NF-κB activation in inflammatory signaling or in tumor cells, respectively.



NF-κB activator 1


Ataxia telangiectasia mutated


β transducin repeat-containing protein


B cell-activating factor


B cell lymphoma 3


B cell lymphoma 10


B cell maturation antigen


Caspase-recruitment-domain-containing membrane-associated guanylate kinase


CARD adaptor inducing IFN-β


CREB binding protein


Cell division cycle 37


Inhibitor of apoptosis protein


connection to IKK and SAPK/JNK


COP9 signalosome component 3


cylindromatosis protein


docking protein 1


Equine herpesvirus


Protein rich in amino acids E, L, K and S


Epithelial Na+ channel


estrogen receptor α


extracellular signal-regulated kinase


Fanconi anemia complementation group A


Fibroblast growth factor 8


FK506-binding protein 51 kDa


Forkhead box transcription factor


Glucocorticoid-induced TNF receptor


Hepatitis C virus


Histone deacetylase


Hypoxia-induced factor 2α


Hematopoietic progenitor kinase 1


Heat shock protein


Human tumorous imaginal disc 1 protein


Human T-cell leukemia virus-1


Huntingtin protein


Herpes virus entry mediator


Inhibitor of kappaB binding


IκB kinase




Interleukin-1 receptor-associated kinase 1


Interferon regulatory factor-7


Insulin receptor substrate-1


c-Jun N-terminal kinase


Kaposi's-sarcoma-associated herpesvirus


Latent membrane protein




Lymphotoxin β


Lymphotoxin β receptor


Mucosa-associated lymphoid tissue lymphoma translocation protein 1


Mitogen-activated protein/ERK kinase kinase


NF-κB essential modifier


Nuclear factor-kappaB


NIK-binding protein


NF-κB-inducing kinase


Nucleotide-binding and oligomerization domain 2


PYRIN and NACHT domain protein 1


3-phosphoinositide-dependent kinase 1


p53-inducible death-domain-containing protein


Protein kinase C


Protein kinase double-stranded RNA-dependent


Phorbol myristate acetate


Receptor activator of NF-κB


Receptor activator of NF-κB ligand


Ret finger protein


Receptor interacting protein 1


Ribosomal S6 kinase


Stress activated protein kinase


Skp1, Cdc53/Cullin1, F-box protein


Sonic hedgehog


Silencing mediator for retinoic acid and thyroid hormone receptor


Steroid receptor co-activator 3


Signal-transducing adaptor protein-2


Small ubiquitin-like modifier


Transforming growth factor-β activated kinase 1 binding protein


Transmembrane activator and calcium modulator and cyclophilin ligand interactor


Transforming growth factor-β activated kinase 1


TNF receptor associated factor family member-associated NF-κB activator


T cell receptor


Tyrosine receptor kinase-fused gene


TRAF-interacting protein with a forkhead-associated domain


Toll-like receptor


Tumor necrosis factor


TNF receptor


TNF receptor-associated death domain


TNF receptor-associated factor


TNF receptor-associated ubiquitous scaffolding and signaling protein


TNF family member with weak apoptosis-inducing activity


Viral FADD-like interleukin-1-β-converting enzyme (FLICE/caspase 8)-inhibitory protein


Yersinia outer protein J


  1. Abbott DW, Wilkins A, Asara JM, Cantley LC . (2004). Curr Biol 14: 2217–2227.

  2. Agou F, Traincard F, Vinolo E, Courtois G, Yamaoka S, Israël A et al. (2004). J Biol Chem 279: 27861–27869.

  3. Agou F, Ye F, Goffinont S, Courtois G, Yamaoka S, Israël A et al. (2002). J Biol Chem 277: 17464–17475.

  4. Aguilera C, Hoya-Arias R, Haegeman G, Espinosa L, Bigas A . (2004). Proc Natl Acad Sci USA 101: 16537–16542.

  5. Albanese C, Wu K, D’Amico M, Jarrett C, Joyce D, Hughes J et al. (2003). Mol Biol Cell 14: 585–599.

  6. Amir RE, Haecker H, Karin M, Ciechanover A . (2004). Oncogene 23: 2540–2547.

  7. Anest V, Cogswell PC, Baldwin Jr AS . (2004). J Biol Chem 279: 31183–31189.

  8. Anest V, Hanson JL, Cogswell PC, Steinbrecher KA, Strahl BD, Baldwin AS . (2003). Nature 423: 659–663.

  9. Atkinson PG, Coope HJ, Rowe M, Ley SC . (2003). J Biol Chem 278: 51134–51142.

  10. Baek SH, Ohgi KA, Rose DW, Koo EH, Glass CK, Rosenfeld MG . (2002). Cell 110: 55–67.

  11. Basseres D, Baldwin Jr AS . (2006). Oncogene this issue.

  12. Beinke S, Ley SC . (2004). Biochem J 382: 393–409.

  13. Bignell GR, Warren W, Seal S, Takahashi M, Rapley E, Barfoot R et al. (2000). Nat Genet 25: 160–165.

  14. Birbach A, Gold P, Binder BR, Hofer E, de Martin R, Schmid JA . (2002). J Biol Chem 277: 10842–10851.

  15. Bonizzi G, Bebien M, Otero DC, Johnson-Vroom KE, Cao Y, Vu D et al. (2004). EMBO J 23: 4202–4210.

  16. Bonizzi G, Karin M . (2004). Trends Immunol 25: 280–288.

  17. Bonnet MC, Weil R, Dam E, Hovanessian AG, Meurs EF . (2000). Mol Cell Biol 20: 4532–4542.

  18. Boone DL, Turer EE, Lee EG, Ahmad RC, Wheeler MT, Tsui C et al. (2004). Nat Immunol 5: 1052–1060.

  19. Bouwmeester T, Bauch A, Ruffner H, Angrand PO, Bergamini G, Croughton K et al. (2004). Nat Cell Biol 6: 97–105.

  20. Bracken CP, Whitelaw ML, Peet DJ . (2005). J Biol Chem 280: 14240–14251.

  21. Brenner D, Golks A, Kiefer F, Krammer PH, Arnold R . (2005). EMBO J 24: 4279–4290.

  22. Broemer M, Krappmann D, Scheidereit C . (2004). Oncogene 23: 5378–5386.

  23. Bruey JM, Bruey-Sedano N, Newman R, Chandler S, Stehlik C, Reed JC . (2004). J Biol Chem 279: 51897–51907.

  24. Brummelkamp TR, Nijman SM, Dirac AM, Bernards R . (2003). Nature 424: 797–801.

  25. Cao Y, Bonizzi G, Seagroves TN, Greten FR, Johnson R, Schmidt EV et al. (2001). Cell 107: 763–775.

  26. Carter RS, Pennington KN, Ungurait BJ, Ballard DW . (2003). J Biol Chem 278: 19642–19648.

  27. Chariot A, Leonardi A, Muller J, Bonif M, Brown K, Siebenlist U . (2002). J Biol Chem 277: 37029–37036.

  28. Chen G, Cao P, Goeddel DV . (2002). Mol Cell 9: 401–410.

  29. Chen LF, Greene WC . (2004). Nat Rev Mol Cell Biol 5: 392–401.

  30. Chen ZJ . (2005). Nat Cell Biol 7: 758–765.

  31. Chen ZJ, Parent L, Maniatis T . (1996). Cell 84: 853–862.

  32. Cheng H, Cenciarelli C, Nelkin G, Tsan R, Fan D, Cheng-Mayer C et al. (2005). Mol Cell Biol 25: 44–59.

  33. Cheong R, Bergmann A, Werner SL, Regal J, Hoffmann A, Levchenko A . (2006). J Biol Chem 281: 2945–2950.

  34. Choi M, Rolle S, Wellner M, Cardoso MC, Scheidereit C, Luft FC et al. (2003). Blood 102: 2259–2267.

  35. Choi SH, Park KJ, Ahn BY, Jung G, Lai MM, Hwang SB . (2006). Mol Cell Biol 26: 3048–3059.

  36. Claudio E, Brown K, Park S, Wang H, Siebenlist U . (2002). Nat Immunol 3: 958–965.

  37. Coope HJ, Atkinson PG, Huhse B, Belich M, Janzen J, Holman MJ et al. (2002). EMBO J 21: 5375–5385.

  38. Correa RG, Matsui T, Tergaonkar V, Rodriguez-Esteban C, Izpisua-Belmonte JC, Verma IM . (2005). Curr Biol 15: 1291–1295.

  39. Courtois G, Gilmore TD . (2006). Oncogene this issue.

  40. Courtois G, Smahi A, Israël A . (2001). Trends Mol Med 7: 427–430.

  41. Dechend R, Hirano F, Lehmann K, Heissmeyer V, Ansieau S, Wulczyn FG et al. (1999). Oncogene 18: 3316–3323.

  42. Dejardin E, Droin NM, Delhase M, Haas E, Cao Y, Makris C et al. (2002). Immunity 17: 525–535.

  43. Delhase M, Hayakawa M, Chen Y, Karin M . (1999). Science 284: 309–313.

  44. Deng L, Wang C, Spencer E, Yang L, Braun A, You J et al. (2000). Cell 103: 351–361.

  45. Devin A, Lin Y, Yamaoka S, Li Z, Karin M, Liu Z . (2001). Mol Cell Biol 21: 3986–3994.

  46. DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E, Karin M . (1997). Nature 388: 548–554.

  47. Ducut Sigala JL, Bottero V, Young DB, Shevchenko A, Mercurio F, Verma IM . (2004). Science 304: 1963–1967.

  48. Ea CK, Deng L, Xia ZP, Pineda G, Chen ZJ . (2006). Mol Cell 22: 245–257.

  49. Ea CK, Sun L, Inoue J, Chen ZJ . (2004). Proc Natl Acad Sci USA 101: 15318–15323.

  50. Eliopoulos AG, Caamano JH, Flavell J, Reynolds GM, Murray PG, Poyet JL et al. (2003). Oncogene 22: 7557–7569.

  51. Evans JD, Seeger C . (2006). Hepatology 43: 615–617.

  52. Field N, Low W, Daniels M, Howell S, Daviet L, Boshoff C et al. (2003). J Cell Sci 116: 3721–3728.

  53. Fu DX, Kuo YL, Liu BY, Jeang KT, Giam CZ . (2003). J Biol Chem 278: 1487–1493.

  54. Gao H, Sun Y, Wu Y, Luan, B, Wang Y et al. (2004). Mol Cell 14: 303–317.

  55. Gao Z, Hwang D, Bataille F, Lefevre M, York D, Quon MJ et al. (2002). J Biol Chem 277: 48115–48121.

  56. Gerondakis S, Grumont R, Gugasyan R, Wong L, Isomura I, Ho W et al. (2006). Oncogene this issue.

  57. Ghosh S, Karin M . (2002). Cell 109: S81–S96.

  58. Gilmore TD . (2006). Oncogene this issue.

  59. Grech AP, Amesbury M, Chan T, Gardam S, Basten A, Brink R . (2004). Immunity 21: 629–642.

  60. Gringhuis SI, Garcia-Vallejo JJ, van Het Hof B, van Dijk W . (2005). Mol Cell Biol 25: 6454–6463.

  61. Haglund K, Dikic I . (2005). EMBO J 24: 3353–3359.

  62. Hai T, Yeung ML, Wood TG, Wei Y, Yamaoka S, Gatalica Z et al. (2006). J Virol 80: 4227–4241.

  63. Hatada EN, Krappmann D, Scheidereit C . (2000). Curr Opin Immunol 12: 52–58.

  64. Hauer J, Puschner S, Ramakrishnan P, Simon U, Bongers M, Federle C et al. (2005). Proc Natl Acad Sci USA 102: 2874–2879.

  65. Hay RT . (2005). Mol Cell 18: 1–12.

  66. Hayden MS, Ghosh S . (2004). Genes Dev 18: 2195–2224.

  67. Hayden MS, West AP, Ghosh S . (2006). Oncogene this issue.

  68. Heissmeyer V, Krappmann D, Hatada EN, Scheidereit C . (2001). Mol Cell Biol 21: 1024–1035.

  69. Hiscott J, Nguyen T-LA, Arugello M, Nakhaei P, Paz S . (2006). Oncogene this issue.

  70. Hoberg JE, Popko AE, Ramsey CS, Mayo MW . (2006). Mol Cell Biol 26: 457–471.

  71. Hoberg JE, Yeung F, Mayo MW . (2004). Mol Cell 16: 245–255.

  72. Hoffmann A, Levchenko A, Scott ML, Baltimore D . (2002). Science 298: 1241–1245.

  73. Hong X, Xu L, Li X, Zhai Z, Shu H . (2001). FEBS Lett 499: 133–136.

  74. Hoshino K, Sugiyama T, Matsumoto M, Tanaka T, Saito M, Hemmi H et al. (2006). Nature 440: 949–953.

  75. Hsu H, Huang J, Shu HB, Baichwal V, Goeddel DV . (1996). Immunity 4: 387–396.

  76. Hu MC, Lee DF, Xia W, Golfman LS, Ou-Yang F, Yang JY et al. (2004). Cell 117: 225–237.

  77. Hu WH, Pendergast JS, Mo XM, Brambilla R, Bracchi-Ricard V, Li F et al. (2005). J Biol Chem 280: 29233–29241.

  78. Hu Y, Baud V, Oga T, Kim KI, Yoshida K, Karin M . (2001). Nature 410: 710–714.

  79. Huang J, Teng L, Li L, Liu T, Chen D, Xu LG et al. (2004). J Biol Chem. 279: 16847–16853.

  80. Huang TT, Feinberg SL, Suryanarayanan S, Miyamoto S . (2002). Mol Cell Biol 22: 5813–5825.

  81. Huang TT, Wuerzberger-Davis SM, Wu ZH, Miyamoto S . (2003a). Cell 115: 565–576.

  82. Huang WC, Chen JJ, Inoue H, Chen CC . (2003b). J Immunol 170: 4767–4775.

  83. Hur GM, Lewis J, Yang Q, Lin Y, Nakano H, Nedospasov S et al. (2003). Genes Dev 17: 873–882.

  84. Jain A, Ma CA, Liu S, Brown M, Cohen J, Strober W . (2001). Nat Immunol 2: 223–228.

  85. Janssens S, Tinel A, Lippens S, Tschopp J . (2005). Cell 123: 1079–1092.

  86. Jimi E, Aoki K, Saito H, D’Acquisto F, May MJ, Nakamura I et al. (2004). Nat Med 10: 617–624.

  87. Jono H, Lim JH, Chen LF, Xu H, Trompouki E, Pan ZK et al. (2004). J Biol Chem 279: 36171–36174.

  88. Joo M, Hahn YS, Kwon M, Sadikot RT, Blackwell TS, Christman JW . (2005). J Virol 79: 7648–7657.

  89. Kanayama A, Seth RB, Sun L, Ea CK, Hong M, Shaito A et al. (2004). Mol Cell 15: 535–548.

  90. Karin M . (1999). Oncogene 18: 6867–6874.

  91. Karin M, Ben-Neriah Y . (2000). Annu Rev Immunol 18: 621–663.

  92. Kayagaki N, Yan M, Seshasayee D, Wang H, Lee W, French DM et al. (2002). Immunity 17: 515–524.

  93. Khoshnan A, Ko J, Watkin EE, Paige LA, Reinhart PH, Patterson PH . (2004). J Neurosci 24: 7999–8008.

  94. Knop J, Martin MU . (1999). FEBS Lett 448: 81–85.

  95. Kovalenko A, Chable-Bessia C, Cantarella G, Israël A, Wallach D, Courtois G . (2003). Nature 424: 801–805.

  96. Krappmann D, Hatada EN, Tegethoff S, Li J, Klippel A, Giese K et al. (2000). J Biol Chem 275: 29779–29787.

  97. Krappmann D, Scheidereit C . (2005). EMBO Rep 6: 321–326.

  98. Kray AE, Carter RS, Pennington KN, Gomez RJ, Sanders LE, Llanes JM et al. (2005). J Biol Chem 280: 35974–35982.

  99. Kwak YT, Li R, Becerra CR, Tripathy D, Frenkel EP, Verma UN . (2005). J Biol Chem 280: 33945–33952.

  100. Lamberti C, Lin KM, Yamamoto Y, Verma U, Verma IM, Byers S et al. (2001). J Biol Chem 276: 42276–42286.

  101. Lawrence T, Bebien M, Liu GY, Nizet V, Karin M . (2005). Nature 434: 1138–1143.

  102. Lebowitz J, Edinger RS, An B, Perry CJ, Onate S, Kleyman TR et al. (2004). J Biol Chem 279: 41985–41990.

  103. Lee EG, Boone DL, Chai S, Libby SL, Chien M, Lodolce JP et al. (2000). Science 289: 2350–2354.

  104. Lee KY, D’Acquisto F, Hayden MS, Shim JH, Ghosh S . (2005). Science 308: 114–118.

  105. Lee S, Andrieu C, Saltel F, Destaing O, Auclair J, Pouchkine V et al. (2004). Proc Natl Acad Sci USA 101: 17416–17421.

  106. Leonardi A, Chariot A, Claudio E, Cunningham K, Siebenlist U . (2000). Proc Natl Acad Sci USA 97: 10494–10499.

  107. Lewis J, Devin A, Miller A, Lin Y, Rodriguez Y, Neckers L et al. (2000). J Biol Chem 275: 10519–10526.

  108. Li H, Kobayashi M, Blonska M, You Y, Lin X. (2006). J Biol Chem 281: 13636–13643.

  109. Li N, Banin S, Ouyang H, Li GC, Courtois G, Shiloh Y et al. (2001). J Biol Chem 276: 8898–8903.

  110. Li Q, Lu Q, Bottero V, Estepa G, Morrison L, Mercurio F et al. (2005). Proc Natl Acad Sci USA 102: 12425–12430.

  111. Li X, Commane M, Burns C, Vithalani K, Cao Z, Stark GR . (1999a). Mol Cell Biol 19: 4643–4652.

  112. Li X, Commane M, Nie H, Hua X, Chatterjee-Kishore M, Wald D et al. (2000). Proc Natl Acad Sci USA 97: 10489–10493.

  113. Li X, Massa PE, Hanidu A, Peet GW, Aro P, Savitt A et al. (2002). J Biol Chem 277: 45129–45140.

  114. Li Y, Kang J, Friedman J, Tarassishin L, Ye J, Kovalenko A et al. (1999b). Proc Natl Acad Sci USA 96: 1042–1047.

  115. Liao G, Zhang M, Harhaj EW, Sun S-C . (2004). J Biol Chem 279: 26243–26250.

  116. Lin L, Hron JD, Peng SL . (2004). Immunity 21: 203–213.

  117. Lin X, Cunningham Jr ET, Mu Y, Geleziunas R, Greene WC . (1999). Immunity 10: 271–280.

  118. Makris C, Roberts JL, Karin M . (2002). Mol Cell Biol 22: 6573–6581.

  119. Massa PE, Li X, Hanidu A, Siamas J, Pariali M, Pareja J et al. (2005). J Biol Chem 280: 14057–14069.

  120. Massoumi R, Chmielarska K, Hennecke K, Pfeifer A, Fassler R . (2006). Cell 125: 665–677.

  121. Matta H, Chaudhary PM . (2004). Proc Natl Acad Sci USA 101: 9399–9404.

  122. Mauro C, Vito P, Mellone S, Pacifico F, Chariot A, Formisano S et al. (2003). Biochem Biophys Res Commun 309: 84–90.

  123. May MJ, D’Acquisto F, Madge LA, Glockner J, Pober JS, Ghosh S . (2000). Science 289: 1550–1554.

  124. May MJ, Larsen SE, Shim JH, Madge LA, Ghosh S . (2004). J Biol Chem 279: 45528–45539.

  125. May MJ, Marienfeld RB, Ghosh S . (2002). J Biol Chem 277: 45992–46000.

  126. Mercurio F, Murray BW, Shevchenko A, Bennett BL, Young DB, Li JW et al. (1999). Mol Cell Biol 19: 1526–1538.

  127. Mercurio F, Zhu H, Murray BW, Shevchenko A, Bennett BL, Li J et al. (1997). Science 278: 860–866.

  128. Meylan E, Curran J, Hofmann K, Moradpour D, Binder M, Bartenschlager R et al. (2005). Nature 437: 1167–1172.

  129. Miller BS, Zandi E . (2001). J Biol Chem 276: 36320–36326.

  130. Miranda C, Roccato E, Raho G, Pagliardini S, Pierotti MA, Greco A . (2006). J Cell Physiol 208: 154–160.

  131. Mordmuller B, Krappmann D, Esen M, Wegener E, Scheidereit C . (2003). EMBO Rep 4: 82–87.

  132. Muller JR, Siebenlist U . (2003). J Biol Chem 278: 12006–12012.

  133. Nakata T, Yokota T, Emi M, Minami S . (2002). Genes Chromosomes Cancer 35: 30–37.

  134. Novack DV, Yin L, Hagen-Stapleton A, Schreiber RD, Goeddel DV, Ross FP et al. (2003). J Exp Med 198: 771–781.

  135. Ohazama A, Hu Y, Schmidt-Ullrich R, Cao Y, Scheidereit C, Karin M et al. (2004). Dev Cell 6: 219–227.

  136. Ohmae T, Hirata Y, Maeda S, Shibata W, Yanai A, Ogura K et al. (2005). J Immunol 175: 7162–7169.

  137. Orth K, Xu Z, Mudgett MB, Bao ZQ, Palmer LE, Bliska JB et al. (2000). Science 290: 1594–1597.

  138. Otsuki T, Young DB, Sasaki DT, Pando MP, Li J, Manning A et al. (2002). J Cell Biochem 86: 613–623.

  139. Panta GR, Kaur S, Cavin LG, Cortes ML, Mercurio F, Lothstein L et al. (2004). Mol cell Biol 24: 1823–1835.

  140. Park KJ, Gaynor RB, Kwak YT . (2003). J Biol Chem 278: 35272–35278.

  141. Park KJ, Krishnan V, O’Malley BW, Yamamoto Y, Gaynor RB . (2005). Mol Cell 18: 71–82.

  142. Perkins ND . (2006). Oncogene this issue.

  143. Piccolella E, Spadaro F, Ramoni C, Marinari B, Costanzo A, Levrero M et al. (2003). J Immunol 170: 2895–2903.

  144. Piret B, Schoonbroodt S, Piette J . (1999). Oncogene 18: 2261–2271.

  145. Pittet JF, Lee H, Pespeni M, O’Mahony A, Roux J, Welch WJ . (2005). J Immunol 174: 384–394.

  146. Pomerantz JL, Baltimore D . (2002). Mol Cell 10: 693–695.

  147. Poyet JL, Srinivasula SM, Alnemri ES . (2001). J Biol Chem 276: 3183–3187.

  148. Poyet JL, Srinivasula SM, Lin JH, Fernandes-Alnemri T, Yamaoka S, Tsichlis PN et al. (2000). J Biol Chem 275: 37966–37977.

  149. Prajapati S, Verma U, Yamamoto Y, Kwak YT, Gaynor RB . (2004). J Biol Chem 279: 1739–1746.

  150. Qian Y, Qin J, Cui G, Naramura M, Snow EC, Ware CF et al. (2004). Immunity 21: 575–587.

  151. Ramakrishnan P, Wang W, Wallach D . (2004). Immunity 21: 477–489.

  152. Ran R, Lu A, Zhang L, Tang Y, Zhu H, Xu et al. (2004). Genes Dev 18: 1466–1481.

  153. Regamey A, Hohl D, Liu JW, Roger T, Kogerman P, Toftgård R et al. (2003). J Exp Med 198: 1959–1964.

  154. Regnier CH, Song HY, Gao X, Goeddel DV, Cao Z, Rothe M . (1997). Cell 90: 373–383.

  155. Reiley W, Zhang M, Sun S-C . (2004). J Biol Chem 279: 55161–55167.

  156. Reiley W, Zhang M, Wu X, Granger E, Sun S-C . (2005). Mol Cell Biol 25: 3886–3895.

  157. Reiley WW, Zhang M, Jin W, Losiewicz M, Donohue KB, Norbury CC et al. (2006). Nat Immunol 7: 411–417.

  158. Rothwarf DM, Zandi E, Natoli G, Karin M . (1998). Nature 395: 297–300.

  159. Rual JF, Venkatesan K, Hao T, Hirozane-Kishikawa T, Dricot A, Li N et al. (2005). Nature 437: 1173–1178.

  160. Ryo A, Suizu F, Yoshida Y, Perrem K, Liou YC, Wulf G et al. (2003). Mol Cell 12: 1413–1426.

  161. Saccani S, Marazzi I, Beg AA, Natoli G . (2004). J Exp Med 200: 107–113.

  162. Saccani S, Pantano S, Natoli G . (2002). Nat Immunol 3: 69–75.

  163. Saccani S, Pantano S, Natoli G . (2003). Mol Cell 11: 1563–1574.

  164. Saito K, Kigawa T, Koshiba S, Sato K, Matsuo Y, Sakamoto A et al. (2004). Structure 12: 1719–1728.

  165. Saito N, Courtois G, Chiba A, Yamamoto N, Nitta T, Hironaka N et al. (2003). J Biol Chem 278: 46565–46575.

  166. Saitoh T, Nakayama M, Nakano H, Yagita H, Yamamoto N, Yamaoka S . (2003). J Biol Chem 278: 36005–36012.

  167. Sato S, Sanjo H, Takeda K, Ninomiya-Tsuji J, Yamamoto M, Kawai T et al. (2005). Nat Immunol 6: 1087–1095.

  168. Schmidt C, Peng B, Li Z, Sclabas GM, Fujioka S, Niu J et al. (2003). Mol Cell 12: 1287–1300.

  169. Schmidt-Ullrich R, Tobin DJ, Lenhard D, Schneider P, Paus R, Scheidereit C . (2006). Development 133: 1045–1057.

  170. Schomer Miller B, Higashimoto T, Lee YK, Zandi E . (2006). J Biol Chem 281: 15268–15276.

  171. Sekine Y, Yumioka T, Yamamoto T, Muromoto R, Imoto S, Sugiyma K et al. (2006). J Immunol 176: 380–389.

  172. Senftleben U, Cao Y, Xiao G, Greten FR, Krahn G, Bonizzi G et al. (2001). Science 293: 1495–1499.

  173. Seo T, Park J, Lim C, Choe J . (2004). Oncogene 23: 6146–6155.

  174. Shi CS, Kehrl JH . (2003). J Biol Chem 278: 15429–15434.

  175. Shim JH, Xiao C, Paschal AE, Bailey ST, Rao P, Hayden MS et al. (2005). Genes Dev 19: 2668–2681.

  176. Shirane M, Hatakeyama S, Hattori K, Nakayama K . (1999). J Biol Chem 274: 28169–28174.

  177. Siebenlist U, Brown K, Claudio E . (2005). Nat Rev Immunol 5: 435–445.

  178. Sil AK, Maeda S, Sano Y, Roop DR, Karin M . (2004). Nature 428: 660–664.

  179. Silverman N, Maniatis T . (2001). Genes Dev 15: 2321–2342.

  180. Soond SM, Terry JL, Colbert JD, Riches DW . (2003). Mol Cell Biol 23: 8334–8344.

  181. Stilo R, Liguoro D, Di Jeso B, Formisano S, Consiglio E, Leonardi A et al. (2004). J Biol Chem 279: 34323–34331.

  182. Su H, Bidere N, Zheng L, Cubre A, Sakai K, Dale J et al. (2005). Science 307: 1465–1468.

  183. Sun L, Deng L, Ea CK, Xia ZP, Chen ZJ . (2004). Mol Cell 14: 289–301.

  184. Takaesu G, Surabhi RM, Park KJ, Ninomiya-Tsuji J, Matsumoto K, Gaynor RB. . (2003). J Mol Biol 326: 105–115.

  185. Tanaka H, Fujita N, Tsuruo T . (2005). J Biol Chem 280: 40965–40973.

  186. Tang ED, Wang CY, Xiong Y, Guan KL . (2003). J Biol Chem 278: 37297–37305.

  187. Tegethoff S, Behlke J, Scheidereit C . (2003). Mol Cell Biol 23: 2029–2041.

  188. Thome M . (2004). Nat Rev Immunol 4: 348–359.

  189. Trompouki E, Hatzivassiliou E, Tsichritzis T, Farmer H, Ashworth A, Mosialos G . (2003). Nature 424: 793–796.

  190. Tu Z, Prajapati S, Park KJ, Kelly NJ, Yamamoto Y, Gaynor RB . (2006). J Biol Chem 281: 6699–6706.

  191. Vacca A, Felli MP, Palermo R, Di Mario G, Calce A, Di Giovine M et al. (2006). EMBO J 25: 1000–1008.

  192. Verma UN, Yamamoto Y, Prajapati S, Gaynor RB . (2004). J Biol Chem 279: 3509–3515.

  193. Wang C, Deng L, Hong M, Akkaraju GR, Inoue J, Chen ZJ . (2001). Nature 412: 346–351.

  194. Wang Y, Tang Y, Teng L, Wu Y, Zhao X, Pei G . (2006). Nat Immunol 7: 139–147.

  195. Wegener E, Oeckinghaus A, Papadopoulou N, Lavitas L, Schmidt-Supprian M, Ferch U et al. (2006). Mol Cell 23: 13–23.

  196. Weih F, Caamano J . (2003). Immunol Rev 195: 91–105.

  197. Wertz IE, O’Rourke KM, Zhou H, Eby M, Aravind L, Seshagiri S et al. (2004). Nature 430: 694–699.

  198. Westerheide SD, Mayo MW, Anest V, Hanson JL, Baldwin Jr AS . (2001). Mol Cell Biol 21: 8428–8436.

  199. Witherow DS, Garrison TR, Miller WE, Lefkowitz RJ . (2004). Proc Natl Acad Sci USA 101: 8603–8607.

  200. Wooten MW, Geetha T, Seibenhener ML, Babu JR, Diaz-Meco MT, Moscat J . (2005). J Biol Chem 280: 35625–35629.

  201. Woronicz JD, Gao X, Cao Z, Rothe M, Goeddel DV . (1997). Science 278: 866–869.

  202. Wu CJ, Conze DB, Li T, Srinivasula SM, Ashwell JD . (2006a). Nat Cell Biol 8: 398–406.

  203. Wu RC, Qin J, Hashimoto Y, Wong J, Xu J, Tsai SY et al. (2002). Mol Cell Biol 22: 3549–3561.

  204. Wu RC, Qin J, Yi P, Wong J, Tsai SY, Tsai MJ et al. (2004). Mol Cell 15: 937–949.

  205. Wu ZH, Shi Y, Tibbetts RS, Miyamoto S . (2006b). Science 311: 1141–1146.

  206. Xia ZP, Chen ZJ . (2005). Sci STKE 272: e7.

  207. Xiao G, Cvijic ME, Fong A, Harhaj EW, Uhlik MT, Waterfield M et al. (2001a). EMBO J 20: 6805–6815.

  208. Xiao G, Fong A, Sun S-C . (2004). J Biol Chem 279: 30099–30105.

  209. Xiao G, Harhaj EW, Sun S-C . (2001b). Mol Cell 7: 401–409.

  210. Yamamoto Y, Gaynor RB . (2004). Trends Biochem Sci 29: 72–79.

  211. Yamamoto Y, Verma UN, Prajapati S, Kwak YT, Gaynor RB . (2003). Nature 423: 655–659.

  212. Yamaoka S, Courtois G, Bessia C, Whiteside ST, Weil R, Agou F et al. (1998). Cell 93: 1231–1240.

  213. Yang F, Yamashita J, Tang E, Wang HL, Guan K, Wang CY . (2004). J Immunol 172: 2446–2452.

  214. Yang J, Lin Y, Guo Z, Cheng J, Huang J, Deng L et al. (2001). Nat Immunol 2: 620–624.

  215. Yilmaz ZB, Weih DS, Sivakumar V, Weih F . (2003). EMBO J 22: 121–130.

  216. Yoshida H, Jono H, Kai H, Li JD . (2005). J Biol Chem 280: 41111–41121.

  217. Zandi E, Rothwarf DM, Delhase M, Hayakawa M, Karin M . (1997). Cell 91: 243–252.

  218. Zha J, Han KJ, Xu LG, He W, Zhou Q, Chen D et al. (2006). J Immunol 176: 1072–1080.

  219. Zhang P, Chan J, Dragoi AM, Gong X, Ivanov S, Li ZW et al. (2005). EMBO Rep 6: 531–537.

  220. Zhang SQ, Kovalenko A, Cantarella G, Wallach D . (2000). Immunity 12: 301–311.

  221. Zhou H, Monack DM, Kayagaki N, Wertz I, Yin J, Wolf B et al. (2005). J Exp Med 202: 1327–1332.

  222. Zhou H, Wertz I, O’Rourke K, Ultsch M, Seshagiri S, Eby M et al. (2004). Nature 427: 167–171.

Download references


The author apologizes that many other relevant references could not be cited due to space constraints. The authors thank members of the laboratory, especially Drs Michael Hinz, Ruth Schmidt-Ullrich, Elmar Wegener and Buket Yilmaz, for critical comments on the manuscript. Work in the author's laboratory is supported by grants from the Bundesministerium für Bildung und Forschung and the Deutsche Forschungsgemeinschaft.

Author information



Corresponding author

Correspondence to Claus Scheidereit.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Scheidereit, C. IκB kinase complexes: gateways to NF-κB activation and transcription. Oncogene 25, 6685–6705 (2006).

Download citation


  • NF-kappaB
  • IkappaB kinase
  • phosphorylation
  • ubiquitin
  • chromatin
  • transcription

Further reading


Quick links