IKK (IκB Kinase) Complex
Identification of the kinase specifically responsible for IκB phosphorylation remained a challenge in the NF-κB field for some time before various complementary approaches allowed full definition of its nature and composition. Instrumental in this was first the characterization of the phosphorylated residues of IκBα responsible for its degradation (serine 32 and 36 located in a short conserved motif) (Whiteside and Israël 1997). This allowed the search by cell extracts fractionation of a kinase able to phosphorylate a polypeptidic sequence including them. Using this strategy, Karin’s lab initially identified a protein of 85 kD that appeared to be catalytically active only after stimulation of cells with TNF-α, a classical NF-κB activator (DiDonato et al. 1997). This inducible feature, combined with the specific recognition of Ser32/36 residues, demonstrated that it was indeed a bona fide IκB kinase. Sequence of this kinase proved identical to CHUK, a previously described kinase with no assigned function (Connelly and Marcu 1995). Given its demonstrated participation in NF-κB activation, CHUK was coined as IKBα kinase α (IKKα). The same group co-purified with IKKα an 87 kD protein presenting high sequence similarity and a catalytic activity against IκBα also (Zandi et al. 1997). It was called IKKβ. At the same time, using a similar purification approach, Mercurio et al. also identified these two kinases and called them IKK1 for the one identical to IKKα and IKK2 for the one identical to IKKβ (Mercurio et al. 1997). Finally, using a yeast two-hybrid screen aimed at identifying interacting partners of NIK, a kinase activating NF-κB through a poorly defined mechanism at that time, Goeddel’s lab also identified CHUK (Régnier et al. 1997). As previously reported by DiDonato et al., this kinase was shown to exhibit activity toward the IκBα polypeptide in an in vitro kinase assay. A search for IKKα homologs through database screening by the same laboratory subsequently identified IKKβ (Woronicz et al. 1997).
Very quickly following their identification, IKKα and IKKβ were shown to form homo- and heterodimers, suggesting that they may act in concert to phosphorylate the two serine residues of IκBα and could represent the catalytic core of a so-called IKB kinase (IKK) complex. A support for the existence of such a protein complex came from gel filtration experiments revealing that IKKα and IKKβ were present in identical species migrating with a molecular weight of 400–600 kD. This suggested that (many?) other components may participate within the same entity in recognition/degradation of IκBα during the NF-κB activation process.
Using an unbiased approach based on the genetic complementation of mutant cells unable to activate NF-κB in response to a large set of stimuli, Yamaoka et al. were the first to identify a non-catalytic component of murine IKK, interacting with more affinity to IKKβ than IKKα. It was called NF-κB essential modulator (NEMO) (Yamaoka et al. 1998). A human protein of the same size (50 kD) and exhibiting a very similar sequence was found in the purified preparation of IKK of DiDonato et al. (see above) and called IKKγ (Rothwarf et al. 1998). Independently, Mercurio et al. used immunoprecipitation of IKK catalytic subunits in HeLa cells to identify associated proteins and found again the human homologue of NEMO that they called IKKAP1 (Mercurio et al. 1999).
Since then, no other components of IKK have been identified as genuine core components of this kinase despite the claim that ELKS (Ducut Sigala et al. 2004) and chaperones Cdc37 and Hsp90 (Chen et al. 2002) may represent additional subunits. As said above, gel filtration analysis initially suggested that many additional subunits might exist. Nevertheless, this view was somehow tempered following the observation that the simple mixing of purified IKKα, IKKβ, and NEMO proteins could produce upon chromatographic analysis a species with a much larger molecular weight than predicted, mostly because of the very elongated structure of NEMO (see below) causing an aberrant delay in gel elution.
Structure and Mechanisms of Activation
IKK is currently viewed as a heterodimer of IKKα/IKKβ associated with a dimer of NEMO. It is not excluded that IKKβ homodimers associated with NEMO also exist.
The recent crystallization of IKKβ (Xu et al. 2011; Polley et al. 2013) has revealed several additional and sometimes unpredicted features deserving comments. First, the overall structure of the dimer resembles a pair of shears (Fig. 2b) with the dimerization interface (the « blade ») appearing much more complex than initially thought. It consists in an α-helical scaffold/dimerization domain (SDD) formed by a large sequence composed of six α-helices and encompassing the previously described LZ and HLH, which do not exist as such. The SDD adopts an elongated conformation with the α2 and α6 helices (70 aa and 77 aa, respectively) running parallel on each IKK2β promoters and providing the hydrophobic residues that allow dimerization. The “handle” includes the kinase domain (KD), which has a typical bilobal form, and a previously poorly characterized domain, the ubiquitin-like domain (ULD), which by interacting with both the SDD and the KD allows full structuration of the catalytic moiety. This domain was originally identified by May et al. (May et al. 2004) who proposed its participation in substrate recognition.
The overal structure of IKKα is very similar to the one of IKKβ, in particular concerning the dimerization interface which also involves a SDD making contacts with an ULD that was originally thought to be present only in IKKβ (Polley et al. 2016). Interestingly, a higher-order organization of IKKα dimers has also been observed in crystals, a hexameric structure formed by a trimer of dimers. This peculiar arrangement is not observed with IKKβ questioning its putative relevance in the autophosphorylation of IKK (IKKα/IKKβ associated with NEMO). Instead it may represent the active species required for RelB/p52 phosphorylation in the noncanonical pathway of NF-κB signaling which only involves IKKα dimers (see below).
Controversy still exists regarding the real affinity of the NUB domain, the ZF, and the full-length NEMO protein toward K-63- versus linear-linked polyubiquitin chains (Rahighi et al. 2009, Laplantine et al. 2009). Further complicating the picture, mixed chains of polyubiquitins have been shown to be produced in cells and could actually represent the genuine target of NEMO. In any case, this intricate mode of ubiquitin recognition represents the heart of NEMO function. Through its interaction with polyubiquitinated partners, NEMO triggers the recruitment of IKK and its catalytic activation either by upstream kinase TAK1, the catalytic component of the TAK complex which is also recruited to ubiquitinated partners by TAB2/TAB3 subunits (see below), or by proximity-induced autophosphorylation (see above).
NEMO posttranslational modifications also regulate its function during the IKK activation process. In addition to being an ubiquitin-binding protein, NEMO is also a target for ubiquitination. Several lysine residues such as Lys285, Lys309, and Lys399 have been shown to be modified by K63- and/or linear-linked chains (Abbott et al. 2004; Tokunaga et al. 2009). This may help strengthening interaction of multiprotein complexes and favor IKK activation. NEMO has also been reported to be phosphorylated at several residues, but the function of these modifications in IKK activity remains less clear. For instance, residue Ser68 can be phosphorylated by IKKα/IKKβ (Palkowitsch et al. 2008), whereas Ser8, Ser17, Ser31, and Ser43 represent targets of GSK-3β (Medunjanin et al. 2016). In both cases this may modulate IKK activity.
Finally, it is worth mentioning that NEMO is often targeted by viral- or bacterial- derived proteins to block the IKK/NF-κB activation process. In several cases, it is degraded by specific proteases, such as members of the 3C-like proteases family (Zhu et al. 2016), and its degradation by the proteasome following K27-linked ubiquitination has also been reported (Ashida et al. 2010).
NF-κB-Related Functions of IKK
As mentioned above IKK is the master regulator of the NF-κB activation process. It is required in every situations in which NF-κB is activated through the so-called canonical pathway. This has been formally demonstrated by studying various cellular systems or by preparing mouse knockouts and further validated through the identification of IKK-related genetic diseases.
Due to space constraints, we will not describe here the many situations in which IKK is required but will provide a short overview of its main physiological functions, only pointing to specific classes of IKK/NF-κB activators. How dysregulation of these functions impact on human health, through the identification of IKK-related genetic diseases, will also be briefly summarized.
A main function of IKK is to allow NF-κB activation when cells are exposed to bacteria- or virus-derived products. In this case, IKK activation is triggered by a collection of receptors belonging to the classes of TLRs, NLRs, or RLRs, which specifically recognize pathogen-associated molecular patterns (PAMPs) such as LPS, flagellin, dsRNA, etc. These receptors, which are present on dendritic cells or macrophages, represent the first line of defense participating in innate immunity (Kaisho et al. 2008). Activated dendritic cells are then able to process and present antigens to T cells, triggering adaptive immunity.
IKK has also various functions at this level. First, it allows activation of NF-κB by the TCR and the BCR. In both cases similar transduction pathways, requiring a members of the CARMA family of protein, CARMA1, and the Bcl10/MALT1 complex connect to IKK (Lin et al. 2004). Moreover, IKK is a key participant in B-cell differentiation and function, leading to synthesis of antibodies (Sasaki et al. 2016). By acting at all these different levels, IKK shapes the immune response. As a consequence, its dysfunction can produce immunodeficiencies or autoimmunity.
IKK is a pivotal player during inflammation (Karin and Greten 2005). On one end, it is activated by a large set of pro-inflammatory cytokines or chemokines, among them TNF-α, IL-1β, οr ΙL-6. On the other, its activation allows synthesis of numerous mediators of the inflammation process, even its own activators. Finally, IKK, more specifically IKKα, controls the resolution of the inflammation. Through this unique property, IKK represents a critical hub in the intricate inflammatory response, and its activation needs to be tightly regulated.
Cell Death and Proliferation
Deregulation of IKK activity, most often constitutive activation of IKK, is a frequent hallmark of cancer cells. This is due to the role that NF-κB can play in cell proliferation, through activation of cyclin A, D1, or CDK6, for instance, or cell survival, as a brake in TNF-α-induced cell death (Fig. 5). By favoring the establishment and maintenance of an inflammatory environment (see above), NF-κB can also influence tumor growth (Bollrath and Greten 2009).
IKK activation has been shown to participate in the protection of cells exposed to DNA-damaging agents (McCool and Miyamoto 2012). In this case, the activation mechanism differs greatly from standard activation mechanisms described above. It has been proposed that in the resting situation, a pool of “free” NEMO, whose characteristics remain poorly defined, shuttles between the cytoplasm and the nucleus. Following exposure to DNA-damaging agents, such as etoposide or doxorubicin, free NEMO gets sumoylated and accumulates in the nucleus where it can associate with ATM, the key kinase activated by recognition of double-strand DNA breaks. This induces NEMO phosphorylation by ATM and translocation of both proteins in the cytoplasm where they activate IKK/NF-κB. At this level, a participation of ELKS has been demonstrated (Wu et al. 2010), suggesting that this protein indeed plays a role in IKK activation but in a very specific setting.
IKK-Related Genetic Diseases
Identification of IKK-related inherited diseases has confirmed in humans several previously identified functions of IKK and also revealed new ones (Senegas et al. 2015). As predicted, NEMO and IKBKB mutations generate complex and severe immunodeficiencies. TLR, TCR, and BCR signaling can be impaired leading to attenuated innate and acquired immunity, which results in a broad range of infections caused by bacterial, viral, or parasitic agents. In addition, it has been shown that NEMO deficiency, causing a lack of NF-κB activation, can trigger skin inflammation. Since skin inflammation can also result from excessive NF-κB activity, this demonstrates that skin homeostasis tightly depends on NF-κB signaling. Finally, another important function of IKK/NF-κB in the skin has been revealed through the identification of patients carrying hypomorphic mutation of NEMO. They exhibit impaired development of skin appendages (sweat gland, teeth, hair).
IKKα as a Key Regulator of the Noncanonical Pathway of NF-κB Activation
Other Functions of IKK and Its Individual Subunits
This review describes the structure and function of the IKK complex, which is primarily involved in NF-κB activation through the canonical pathway. Nevertheless, as a whole the IKK complex can also fulfill few NF-κB-independent functions and the same applies to its individual components.
In two situations that may actually be linked, it has been proposed that IKK regulates processes that do not involve NF-κB. First, it has been shown that IKK regulates the autophagy process upon various forms of cellular stress such as nutrient deprivation, inhibition of mTOR, or endoplasmic reticulum stress (Criollo et al. 2010). Independently, it has been shown that IKK is involved in the response of cells to glutamine removal (Reid et al. 2016). In this specific case, it phosphorylates PFKFB3, inducing its destabilization. Since PFKFB3 is a key component in aerobic glycosis, this contributes to downregulating this metabolic process.
Individually, IKKβ can induce the phosphorylation of a collection of substrates that are not linked to NF-κB signaling (Chariot 2009). NEMO has also a scaffold function during the activation of TBK1/IKKε, two IKK-like enzymes controlling IRF3/IRF7, the kinases that mediate the synthesis of type 1 interferon proteins during viral infection (Zhao et al. 2007). Finally, IKKα have a specific function in the nucleus as an H3 kinase (Sun 2012) and plays a key role in epidermis development during embryogenesis by controlling periderm formation (Richardson et al. 2014).
In most biological processes involving NF-κB, which are many, a participation of IKK has been firmly demonstrated. This indicates the key function of this kinase complex in channeling the effect of a wide range of stimuli to allow the proper coordinated synthesis of many proteins. The core structure of IKK appears quite simple, with two catalytic subunits (IKKα and IKKβ) associated with a single regulatory one (NEMO). As such, this provides an attractive target to efficiently inhibit the NF-κB signaling pathway. Nevertheless, this broad involvement of IKK also precludes acting on NF-κB in a given setting without affecting other of its essential functions. Therefore, the jury is still out regarding the true value of directly targeting IKK for therapeutic purposes. To solve this uncertainty, several specific fields of research should be investigated more thoroughly. Among them are the exact catalytic relationship that exists between IKKα and IKKβ, with the two enzymes working in concert within IKK despite exhibiting a quite different level of activity. This may provide a way to act distinctly on NF-κB-controlled signaling pathways. Moreover, the characterization of NEMO posttranslational modifications (ubiquitination, phosphorylation, etc.) remains a neglected field that may provide strategies to target IKK context specifically.
I thank Dr. Jérémie Gautheron for preparing Fig. 5.
- Criollo A, Senovilla L, Authier H, Maiuri MC, Morselli E, Vitale I, Kepp O, Tasdemir E, Galluzzi L, Shen S, Tailler M, Delahaye N, Tesniere A, De Stefano D, Younes AB, Harper F, Pierron G, Lavandero S, Zitvogel L, Israel A, Baud V, Kroemer G. The IKK complex contributes to the induction of autophagy. EMBO J. 2010;29:619–31.PubMedCrossRefGoogle Scholar
- Polley S, Huang DB, Hauenstein AV, Fusco AJ, Zhong X, Vu D, Schröfelbauer B, Kim Y, Hoffmann A, Verma IM, Ghosh G, Huxford T. A structural basis for IκB kinase 2 activation via oligomerization-dependent trans auto-phosphorylation. PLoS Biol. 2013;11(6):e1001581.PubMedCrossRefPubMedCentralGoogle Scholar