Signaling to NF-kB

7/27/2019 Signaling to NF-kB

1/31

10.1101/gad.1228704Access the most recent version at doi:2004 18: 2195-2224Genes Dev.

Matthew S. Hayden and Sankar Ghosh

BSignaling to NF-

References

http://genesdev.cshlp.org/content/18/18/2195.full.html#related-urlsArticle cited in:

http://genesdev.cshlp.org/content/18/18/2195.full.html#ref-list-1This article cites 355 articles, 170 of which can be accessed free at:

service

Email alerting

click heretop right corner of the article or

Receive free email alerts when new articles cite this article - sign up in the box at the

CollectionsTopic

(90 articles)Signal Transduction(16 articles)Immunology

(142 articles)Chromatin and Gene ExpressionArticles on similar topics can be found in the following collections

http://genesdev.cshlp.org/subscriptionsgo to:Genes & DevelopmentTo subscribe to

Cold Spring Harbor Laboratory Press

Cold Spring Harbor Laboratory Presson July 11, 2012 - Published bygenesdev.cshlp.orgDownloaded from
http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1228704http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1228704http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1228704http://genesdev.cshlp.org/content/18/18/2195.full.html#related-urlshttp://genesdev.cshlp.org/content/18/18/2195.full.html#related-urlshttp://genesdev.cshlp.org/content/18/18/2195.full.html#related-urlshttp://genesdev.cshlp.org/content/18/18/2195.full.html#ref-list-1http://genesdev.cshlp.org/content/18/18/2195.full.html#ref-list-1http://genesdev.cshlp.org/content/18/18/2195.full.html#ref-list-1http://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;18/18/2195&return_type=article&return_url=http://genesdev.cshlp.org/content/18/18/2195.full.pdfhttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;18/18/2195&return_type=article&return_url=http://genesdev.cshlp.org/content/18/18/2195.full.pdfhttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;18/18/2195&return_type=article&return_url=http://genesdev.cshlp.org/content/18/18/2195.full.pdfhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/immunologyhttp://genesdev.cshlp.org/cgi/collection/chromatin_and_genehttp://genesdev.cshlp.org/cgi/collection/immunologyhttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://www.cshlpress.com/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/cgi/collection/signal_transductionhttp://genesdev.cshlp.org/cgi/collection/immunologyhttp://genesdev.cshlp.org/cgi/collection/chromatin_and_genehttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;18/18/2195&return_type=article&return_url=http://genesdev.cshlp.org/content/18/18/2195.full.pdfhttp://genesdev.cshlp.org/content/18/18/2195.full.html#related-urlshttp://genesdev.cshlp.org/content/18/18/2195.full.html#ref-list-1http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1228704


2/31

REVIEW

Signaling to NF-B

Matthew S. Hayden and Sankar Ghosh1

Section of Immunobiology and Department of Molecular Biophysics & Biochemistry, Yale University School of Medicine,New Haven, Connecticut 06520, USA

The transcription factor NF-B has been the focus ofintense investigation for nearly two decades. Over thisperiod, considerable progress has been made in determin-ing the function and regulation of NF-B, although thereare nuances in this important signaling pathway thatstill remain to be understood. The challenge now is toreconcile the regulatory complexity in this pathway

with the complexity of responses in which NF-B familymembers play important roles. In this review, we pro-vide an overview of established NF-B signaling path-ways with focus on the current state of research into themechanisms that regulate IKK activation and NF-Btranscriptional activity.

Inducible transcription factors regulate immediate andlong-lived cellular responses necessary for organismaladaptation to environmental plasticity. Such responsesare mediated to a large degree through changes in geneexpression. One transcription factor that serves as a keyresponder to changes in the environment is NF-B, anevolutionarily conserved signaling module that plays a

critical role in many biological processes. Understandinghow the transcriptional potential, activity, and selectiv-ity of NF-B are regulated is therefore a topic of intenseinvestigation in numerous laboratories.

The biological system in which NF-B plays the mostimportant role is the immune system (for reviews, seeGhosh et al. 1998; Li and Verma 2002; Bonizzi and Karin2004). Careful regulation of the transcriptional responsesto many different stimuli is crucial to the proper func-tioning of the mammalian immune system. NF-B regu-lates the expression of cytokines, growth factors, andeffector enzymes in response to ligation of many recep-tors involved in immunity including T-cell receptors(TCRs) and B-cell receptors (BCRs), TNFR, CD40,BAFFR, LTR, and the Toll/IL-1R family (for reviews,

see Ghosh et al. 1998; Silverman and Maniatis 2001;Bonizzi and Karin 2004). NF-B also regulates the expres-sion of genes outside of the immune system and, hence,can influence multiple aspects of normal and diseasephysiology. Recent work has highlighted the role of NF-B in embryonic development and in the developmentand physiology of tissues including mammary gland,

bone, skin, and central nervous system. However, suchvaried biological roles for NF-B raise the intriguingquestion of whether one common mechanism regulatessignaling to NF-B in all systems or whether discreteinputs create a diversity of transcriptional responses thatare tailored to particular tissues and organs. Understand-ing how NF-B integrates multiple stimuli in multiplesystems to generate a unified outcome suitable for spe-cific situations is a challenge that faces researchers inthis area. In keeping with the enormous progress that hasbeen made in the study of NF-B, there has been a veri-table explosion of review articles that have elegantlysummarized progress in different aspects of NF-B regu-lation and biology (Ghosh and Karin 2002; Kucharczak etal. 2003; Ruland and Mak 2003; Ben-Neriah and Schmitz2004; Bonizzi and Karin 2004; Chen and Greene 2004;Karin et al. 2004). Therefore, to avoid duplication, wehave decided to focus in this review on a few areas ofcurrent activity. The main question that will be dis-cussed in this review is how the different inducers acti-vate NF-B and the mechanisms that underlie the regu-lation of NF-B transcriptional activity. The choice ofthese areas for discussion is, of course, idiosyncratic and

we apologize for the narrow focus of this review. How-ever, to help an uninitiated reader delve right into theseareas of current research, we have provided a brief over-view of the current state of knowledge about this tran-scription factor. We hope that interested readers will finda sufficiently comprehensive listing of the relevant lit-erature in this article such that they will be able to go onand explore the biology of this fascinating transcriptionfactor in depth.

Overview of the NF-B pathway

The five members of the mammalian NF-B family, p65(RelA), RelB, c-Rel, p50/p105 (NF-B1), and p52/p100(NF-B2), exist in unstimulated cells as homo- or het-erodimers bound to IB family proteins. NF-B proteinsare characterized by the presence of a conserved 300-amino acid Rel homology domain (RHD) that is locatedtoward the N terminus of the protein and is responsiblefor dimerization, interaction with IBs, and binding toDNA (Fig. 1). Binding to IB prevents the NF-B:IBcomplex from translocating to the nucleus, therebymaintaining NF-B in an inactive state. NF-B signalingis generally considered to occur through either the clas-sical or alternative pathway (for review, see Bonizzi andKarin 2004). In the classical pathway of NF-B activa-tion, for example, upon stimulation by the proinflamma-

[Keywords: NF-B; I-B; signal transduction; transcription factors; pro-teinserinethreonine]1Corresponding author.E-MAIL [email protected]; FAX (203) 737-1764.Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/gad.1228704.

GENES & DEVELOPMENT 18:21952224 2004 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/04; www.genesdev.org 2195

http://www.cshlpress.com/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/http://www.cshlpress.com/http://genesdev.cshlp.org/


3/31

tory cytokine tumor necrosis factor (TNF), signalingpathways lead to activation of the subunit of the IBkinase (IKK) complex, which then phosphorylates IBproteins on two N-terminal serine residues (Fig. 2A). Inthe alternative pathway, IKK is activated and phos-phorylates p100 (Fig. 2B). Phosphorylated IBs are recog-nized by the ubiquitin ligase machinery, leading to theirpolyubiquitination and subsequent degradation, or pro-cessing in the case of p100, by the proteasome (for re-view, see Karin and Ben-Neriah 2000). The freed NF-Bdimers translocate to the nucleus, where they bind tospecific sequences in the promoter or enhancer regionsof target genes. Activated NF-B can then be down-regu-lated through multiple mechanisms including the well-characterized feedback pathway whereby newly synthe-sized IB protein binds to nuclear NF-B and exports itout to the cytosol.

There are seven IB family membersIB, IB,BCL-3, IB, IB, and the precursor proteins p100 andp105which are characterized by the presence of five toseven ankyrin repeats that assemble into elongated cyl-inders that bind the dimerization domain of NF-Bdimers (Fig. 1; Hatada et al. 1992). The crystallographicstructures of IB and IB bound to p65/p50 or p65/c-Rel dimers revealed that the IB proteins mask only thenuclear localization sequence (NLS) of p65, whereas theNLS of p50 remains accessible (Huxford et al. 1998; Ja-cobs and Harrison 1998; Malek et al. 2001, 2003). The

presence of this accessible NLS on p50 coupled withnuclear export sequences (NES) that are present on IBand p65 results in constant shuttling of IB/NF-Bcomplexes between the nucleus and the cytoplasm, al-though the steady-state localization is in the cytosol(Johnson et al. 1999; Huang et al. 2000). The dynamicbalance between cytosolic and nuclear localization is al-tered upon IB degradation, because it removes the con-tribution of the IB NES and exposes the masked NLS ofp65, resulting in predominantly nuclear localization ofNF-B.

Degradation of IB is a tightly regulated event that isinitiated upon specific phosphorylation by activatedIKK. The IKK activity in cells can be purified as a 700900-kDa complex, and has been shown to contain twokinase subunits, IKK (IKK1) and IKK (IKK2), and aregulatory subunit, NEMO (NF-B essential modifier) orIKK (for reviews, see Rothwarf and Karin 1999; Ghoshand Karin 2002). In the classical NF-B signaling path-way, IKK is both necessary and sufficient for phos-phorylation of IB on Ser 32 and Ser 36, and IB on Ser19 and Ser 23. The role of IKK in the classical pathwayis unclear, although recent studies suggest it may regu-late gene expression in the nucleus by modifying thephosphorylation status of histones. The alternative path-way, however, depends only on the IKK subunit, whichfunctions by phosphorylating p100 and causing its in-ducible processing to p52 (Fig. 2B). The alternative path-

Figure 1. Schematic representation of NF-B, IB, andIKK proteins family of proteins. Members of the NF-B,

IB, and IKK proteins families are shown. The numberof amino acids in each protein is indicated on the right.Presumed sites of cleavage for p100 (amino acid 447)and p105 (amino acid 433) are shown. Phosphorylationand ubiquitination sites on p100, p105, and IB proteinsare indicated. (RHD) Rel homology domain; (TAD)transactivation domain; (LZ) leucine zipper domain onIKK/ and Rel-B; (GRR) glycine-rich region; (HLH) he-lixloophelix domain; (Z) zinc finger domain; (CC1/2)coiled-coil domains; (NBD) NEMO-binding domain; ()-helical domain.

Hayden and Ghosh

2196 GENES & DEVELOPMENT



4/31

way is activated in response to a subset of NF-B induc-ers including LT and BAFF.

Upon phosphorylation by IKKs, IB proteins are recog-nized and ubiquitinated by members of the Skp1CulinRoc1/Rbx1/Hrt-1F-box (SCF or SCRF) family of ubiqui-tin ligases (for review, see Ben-Neriah 2002). TrCP(E3RS or Fbw1a), the receptor subunit of the SCF familyubiquitin ligase machinery, binds directly to the phos-phorylated E3 recognition sequence (DS*GXXS*) on IB(Yaron et al. 1997, 1998; Fuchs et al. 1999; Hatakeyamaet al. 1999; Kroll et al. 1999; Spencer et al. 1999; Suzukiet al. 1999; Winston et al. 1999; Wu and Ghosh 1999).Recognition of IB leads to polyubiquitination at con-served residues, Lys 21 and Lys 22 on IB, by the E3SCF-TrCP and the E2 UbcH5 (Alkalay et al. 1995; Schereret al. 1995; DiDonato et al. 1996). Although it is com-monly believed that degradation of IB is a cytoplasmicevent, TrCP1 is almost exclusively nuclear with its re-ceptor site occupied by hnRNP-U (Davis et al. 2002).This finding has led to the suggestion that phosphory-lated IB must out-compete hnRNP-U for binding toTrCP1. However, TrCP2, a highly homologous iso-form of TrCP1, is localized in the cytoplasm and canbind IB, although it has lower ligase efficiency than

TrCP1 (Suzuki et al. 2000; Davis et al. 2002). Cells fromTrCP1-deficient mice display partially decreased ratesof IB and IB degradation, indicating that thisTrCP1 function can be partially compensated for byTrCP2 (Nakayama et al. 2003). Determination of theactual role of -TrCP isoforms in IB degradation willhave to await the generation of mice lacking both iso-forms.

The precursor protein p105 undergoes constitutiveprocessing, as opposed to degradation, via the protea-some, through a cotranslational mechanism (Fan andManiatis 1991; Palombella et al. 1994; L. Lin et al. 1998).Limited proteolysis of the precursor protein, which gen-erates p50, is dependent on the presence of a glycine-richregion (GRR) between amino acids 376 and 404 thatserves as a stop signal for proteolysis (Lin and Ghosh1996; Orian et al. 1999). Whether p105 can also undergoinducible processing remains contentious. Multiple re-ports have demonstrated IKK-dependent, and IKK-in-dependent, phosphorylation of p105 C-terminal serinesincluding Ser 923 and Ser 927, followed by inducible deg-radation, although there is no definitive evidence thatthis degradation of p105 has a functional role (Fujimotoet al. 1995; MacKichan et al. 1996; Heissmeyer et al.

Figure 2. Classical and alternative pathways ofNF-B activation. A model depicting the two sig-naling pathways to NF-B. (A) One pathway isthe classical pathway mediated by IKK andleading to phosphorylation of IB. Inputs for theclassical pathway include TNFR1/2, TCR andBCR, TLR/IL-1R, and many others. (B) The alter-native pathway involves NIK activation of IKKand leads to the phosphorylation and processingof p100, generating p52:RelB heterodimers. Inputsignals for the alternative pathway follow liga-tion of LTR, BAFFR, and CD40R. These alter-native pathway stimuli also activate the classicalpathway.

Signaling to NF-B

GENES & DEVELOPMENT 2197



5/31

1999, 2001; Orian et al. 2000; Salmeron et al. 2001; Langet al. 2003; Cohen et al. 2004). It has been suggestedthat similar to IB, phosphorylation of p105 leads toSCF-TrCP binding, resulting in polyubiquitination anddegradation of the protein. Polyubiquitination of p105leading to degradation involves multiple lysine residues,

and is dependent on an acidic region between residues445 and 453 (Harhaj et al. 1996; Orian et al. 1999; Cohenet al. 2004). In contrast, partial processing leading to p50generation may be carried out by a different ubiquitinligase machinery (Amir et al. 2002; Lang et al. 2003; Co-hen et al. 2004). It was recently demonstrated thatSCFTrCP is not responsible for signal-induced process-ing, and it has been suggested that this event is indepen-dent of ubiquitination (Cohen et al. 2004). The p105 pre-cursor protein forms multiple heterodimeric and ho-modimeric complexes (Ghosh 2004), and this associationwith NF-B proteins inhibits constitutive processing(Harhaj et al. 1996; Cohen et al. 2001). Most likely, pro-cessing of p105 is partially determined by the milieu of

NF-B dimers present in any given tissue.Processing of p100 bears some similarity to both IB

degradation and p105 processing. The phosphorylation ofp100 has, so far, only been shown to be catalyzed byIKK acting downstream from NF-B inducing kinase(NIK; Fig. 2B; Senftleben et al. 2001a). Unlike p105, p100processing is a tightly regulated event, with only mini-mal constitutive processing in unstimulated cells(Heusch et al. 1999). Interestingly, it appears that NIKcan act both as an IKK-activating kinase as well as adocking protein linking IKK to p100 (Xiao et al. 2004).The C-terminal death domain (DD) of p100 has beenshown to function as a processing inhibitory domain(PID), and expression of a p100 lacking this domain re-sults in increased constitutive processing in a mannerdependent on nuclear shuttling (Xiao et al. 2001; Liaoand Sun 2003). Similar to IB, phosphorylation of p100leads to the recruitment of SCFTrCP, polyubiquitinationof Lys 855 in a region with sequence homology to Lys 22of IB, and subsequent processing to p52 (Fong and Sun2002; Amir et al. 2004; Cohen et al. 2004). Like p105, thep100 GRR is also required for partial processing yieldingp52 and release of active p52:NF-B complexes (Heuschet al. 1999).

Following degradation of IB, the released NF-B isable to bind promoter and enhancer regions containingB sites with the consensus sequence GGGRNNYYCC(N = any base, R = purine, and Y = pyrimidine). The crys-tal structures of NF-B dimers bound to the B enhancerreveal how both immunoglobulin-like domains thatmake up the RHD contact the DNA of the B site. TheN-terminal Ig-like domain is primarily responsible forsequence specificity of NF-B, whereas hydrophobic resi-dues within the C-terminal domain form the dimeriza-tion interface (Ghosh et al. 1995; Muller et al. 1995; F.E.Chen et al. 1998; Y.Q. Chen et al. 1998; Huxford et al.1998; Huang et al. 2001). RelB, c-Rel, and p65 contain atransactivation domain (TAD) located toward the C ter-minus that is necessary for transactivation by these pro-teins. Homodimers of p52 and p50 lack TADs and hence

have no intrinsic ability to drive transcription. In fact,binding of p52 or p50 homodimers to B sites of restingcells leads to repression of gene expression (Zhong et al.2002). The repressive function of p50 or p52 homodimersmay provide a threshold for NF-B transactivation thatcan be regulated through the expression and process-

ing of the p100 and p105 precursors. The TADs on p65,c-Rel, and RelB promote transcription by facilitating therecruitment of coactivators and the displacement of re-pressors. The function of TADs is enhanced through di-rect modifications of NF-B including phosphorylation,and represents another layer of regulation of the NF-B-mediated transcriptional response (for review, see Chenand Greene 2004).

Biological roles of NF-B and IB proteins

All of the mammalian Rel and IB family members havebeen knocked out in mice, and many conditional knock-outs and multigene knockouts have also been gener-

ated. The phenotype of these various mice with regard totheir immune system function and roles in the regula-tion of apoptosis has recently been reviewed in detail(Kucharczak et al. 2003; Bonizzi and Karin 2004). There-fore, we provide here a succinct overview of the functionof each pathway component based on the relevant ge-netic experiments in support of such a role.

Genetic studies first revealed the crucial role of p65 inmediating protection from apoptosis during TNF sig-naling. p65/ mice exhibit lethality caused by liver de-generation at gestational day 1516 (Beg et al. 1995b).The insult leading to the observed liver apoptosis wasshown to be TNF signaling in the developing liver, ascrossing p65/ mice with either TNF/ (Doi et al.1999) or TNFR/ (Alcamo et al. 2001) mice rescues theliver phenotype and allows the study of the effects of p65knockout on other tissues. p65/TNFR1 double knock-outs have increased susceptibility to bacterial infection,highlighting the role of p65 in innate immune responsesand the initiation of innate immune responses by non-hematopoietic cells (Alcamo et al. 2001). The knockoutsof p65 were also used to show its importance in classswitching in B cells and lymphocyte proliferation follow-ing various stimuli in chimeric mouse models (Doi et al.1997).

In contrast to p65-deficient mice, p50 (NF-B1)-defi-cient mice do not show any developmental defects (Shaet al. 1995). Instead, p50/ mice exhibit decreased im-munoglobulin production and defective humoral im-mune responses. B cells from these mice do not respondefficiently to LPS, underscoring the role of the classicalp65/p50 heterodimer in Toll/IL-1 signaling pathways,whereas p50 is redundant for antiapoptotic pathways.

Mice lacking p52 (NF-B2) fail to develop normal B-cell follicles and germinal centers and have additionaldefects in their splenic architecture and Peyers patchdevelopment (Caamano et al. 1998; Franzoso et al. 1998;Paxian et al. 2002). These mice display normal B-cellmaturation and undergo class-switch recombination,but generate inadequate humoral responses to various

Hayden and Ghosh




6/31

T-cell-dependent antigens. In addition, p52 knockoutsfail to generate appropriate class-switched antibodies fol-lowing influenza infection. Rescue of the T-cell-depen-dent response by adoptively transferring p52-deficientcells into rag-1/ mice indicates that this deficit is notintrinsic to the B cells (Franzoso et al. 1998). Overall, the

p52 knockout mice have a phenotype that is consistentwith a role for p52 in LTR signaling, particularly withregard to development of secondary lymphoid organs (forreviews, see Mebius 2003; Bonizzi and Karin 2004).

RelB is unique in that it does not homodimerize, andfurther, is unable to heterodimerize with c-Rel or p65.RelB forms heterodimers with p100, p52, and p50(Ryseck et al. 1992; Dobrzanski et al. 1994) and does notinteract with any IB molecules other than p100 (Solanet al. 2002). Therefore, RelB/p52 heterodimers exhibitconstitutive nuclear localization (Ryseck et al. 1992;Dobrzanski et al. 1993, 1994; Lernbecher et al. 1994).RelB has been implicated in constitutive NF-B activityin multiple tissues, and RelB-deficient mice have de-

creased baseline NF-B activity in the thymus andspleen. Interestingly, loss of RelB results in increasedinflammatory infiltration in multiple organs as well assevere deficits in adaptive immunity (Weih et al. 1995).Loss of cellular immune responses in these mice is con-sistent with the observed deficit in thymic medullaryepithelial cells and the loss of functional dendritic cells(Burkly et al. 1995). This phenotype is exaggerated whenp50 is also knocked out, indicating that p50 and RelBcooperate in the regulation of genes that limit inflam-mation (Weih et al. 1997). relB/ B cells, althoughcrippled in their proliferative response, undergo normalIgM secretion and class switching in response to variousstimuli (Snapper et al. 1996). The importance of RelB inlymphoid organogenesis is evident in RelB knockouts,particularly in the development of Peyers patches(Paxian et al. 2002; Yilmaz et al. 2003). These findingshighlight the role of RelB/p52 dimers (probably in re-sponse to LTR signaling) during lymphoid organogen-esis.

c-Rel forms homodimers and heterodimers with p50and p65. c-Rel-deficient mice fail to generate a produc-tive humoral immune response, and peripheral T and Bcells fail to respond to typical mitogenic stimuli (Kont-gen et al. 1995). Mice lacking the c-Rel TAD, but with anintact RHD display a more severe phenotype, withmarked hyperplasia in secondary lymphoid organs andhypoplastic bone marrow, and support a role for c-Rel inB-cell class switching (Zelazowski et al. 1997; Carrascoet al. 1998).

Transcriptional specificity is partially regulated by theability of specific NF-B dimers to preferentially associ-ate with certain members of the IB family. For example,the classical NF-B heterodimer (p65/p50) is predomi-nantly, although not exclusively, regulated by IB.Therefore, IB-deficient mice display increased but notconstitutive p65/p50 DNA-binding activity, demonstrat-ing that additional mechanisms regulate the activity ofthis dimer (Beg et al. 1995a; Klement et al. 1996). IB isknown to participate in a feedback loop where newly

synthesized IB inhibits the activity of NF-B. Consis-tent with such a role, in the absence of IB the termi-nation of NF-B activation in response to TNF- is sig-nificantly delayed. In fact, if one places IB, which oth-erwise does not participate in the early down-regulationof NF-B activity, under control of the IB promoter,

the normal kinetics of NF-B activation and down-regu-lation are restored (Cheng et al. 1998). Analysis of celllines in which multiple isoforms of IBs have beenknocked out demonstrates that the functional charac-teristics of IB, IB, and IB are primarily the resultof temporal differences in their degradation and re-synthesis that is attributable to differential transcrip-tional regulation of their promoters (Hoffmann et al.2002). Therefore, although IB is the primary regulatorof p65:p50 complexes, it remains to be seen whetherthere are intrinsic properties of the IB protein, in ad-dition to its transcriptional regulation, that contribute tothis central function in classical NF-B signaling path-ways.

The role of the other IB family members is less wellunderstood. The phenotype of IB-deficient mice hasnot been published, and, consequently, the biologicalrole of this family member remains to be clarified. Earlystudies suggested that BCL-3 interacts with and removesp50 and p52 homodimers from B sites, thereby remov-ing the repressive effects of dimers lacking TADs(Hatada et al. 1992; Kerr et al. 1992; Wulczyn et al. 1992;Franzoso et al. 1993; Naumann et al. 1993). However,subsequent reports suggest BCL-3 forms transcription-ally active complexes with p52 and p50 homodimers(Bours et al. 1993; Fujita et al. 1993). Although thesereports have demonstrated that the ability of BCL-3 com-plexes to promote transcription depends on BCL-3 phos-phorylation, the upstream components of this path-way and their regulation have not been well charac-terized (Nolan et al. 1993). The BCL-3 knockout mouseis viable but is unable to generate an appropriate hu-moral immune response to multiple pathogens. Thesemice lack spleen germinal centers, and although theyexhibit normal serum antibody levels, they fail to de-velop antigen-specific responses (Franzoso et al. 1997;Schwarz et al. 1997). This phenotype is partially sharedby p52/ mice, supporting the idea that BCL-3 cooper-ates with p52 to form a transcriptionally active complex(Sha et al. 1995; Caamano et al. 1996; Franzoso et al.1997; Schwarz et al. 1997; Brasier et al. 2001; Kuwataet al. 2003).

IB has a selective role in the regulation of p65 ho-modimers and c-Rel:p65 heterodimers (Li and Nabel1997; Simeonidis et al. 1997; Whiteside et al. 1997). IBdegradation occurs in response to many NF-B inducers,albeit with considerably slower kinetics than that forIB. Similar to IB, IB has some nuclear-cytoplas-mic shuttling ability, although markedly less than IB,and hence, IB-containing complexes are predomi-nantly cytoplasmic (Simeonidis et al. 1997; Tam et al.2000; Lee and Hannink 2002). IB has a noncanonicalNES that is necessary for the cytoplasmic localization ofc-Rel and IB itself (Lee and Hannink 2002). As IB is

Signaling to NF-B




7/31

induced in response to NF-B activation (Simeoni-dis et al. 1997; Whiteside et al. 1997), it appears that itmay function in regulating later phases of NF-B geneactivation by p65:c-Rel complexes. IB is actually theC-terminal region of mouse p105 (Inoue et al. 1992;Gerondakis et al. 1993; Grumont and Gerondakis 1994).

It is still unclear whether IB, which is synthesizedas an alternate mRNA using a separate promoter,has a defined biological role. Finally, IB (also calledMAIL/INAP), perhaps the eighth member of the mam-malian IB family, is found in the nucleus and is ex-pressed in response to IL-1 and TLR stimulation ofNF-B but not TNF (Kitamura et al. 2000; Haruta et al.2001; Yamazaki et al. 2001; Eto et al. 2003; Muta et al.2003). Although IB has weak homology to BCL-3and the other IBs, there is not yet any direct evidencethat it is functionally an IB family member, as it hasnot yet been shown to interact with any Rel familymembers.

Signaling pathways to NF-B

Numerous pathways lead to the activation of NF-B. Al-most universally, these pathways proceed via activation

of IKK, degradation of IB, and enhancement of the tran-scriptional activity of NF-B. However, there is signifi-cant variability in fundamental aspects of these path-ways, for example, TNF signaling via TNFR1 results inthe rapid activation of IKK and nearly complete degrada-tion of IB within 10 min, whereas signal-induced deg-

radation of IB through the TCR takes nearly 45 min.Furthermore, individual NF-B responses can be charac-terized as consisting of waves of activation and inactiva-tion of the various NF-B family members (Hoffmann etal. 2002, 2003). This is because of sustained activation ofthe IKK complex, and its selectivity for different IBs, aswell as the differential regulation of IB expression byNF-B dimers. The baseline complement of NF-B com-plexes present in a cell can therefore influence the natureof the transcriptional response to a given stimulus. Thetranscriptional activity and specificity of NF-B are alsodifferentially regulated. Summation of these effects re-sults in unique transcriptional responses to a givenstimulus in a particular cell type. A careful examination

of the most-studied pathwaysTNFR, TLR/IL-1R, TCR,and BCRhighlights some of these differences, but alsoreinforces the commonalities that characterize signalingto NF-B in general (Fig. 3).

Figure 3. Major signaling pathways that lead to NF-B activation. Signal transduction pathways emanating from TNF receptor (A),Toll/IL-1 receptor (B), the BCR (C), and intermediary proteins involved. (Lower left) An outline of pathways leading to NF-B as aconsequence of cell stress and DNA damage is also indicated. The activation of the TCR is shown using cross-linking anti-CD3/CD28antibodies.

Hayden and Ghosh




8/31

TNF: a model NF-B stimulus

The TNFR superfamily consists of at least 19 ligands and29 receptors and exhibits a remarkable diversity in tissuedistribution and physiology (for review, see Aggarwal2003). These receptor/ligand pairs initiate a variety of

biological responses, primarily through activation of in-ducible transcription factors such as NF-B and AP-1 (forreview, see Wajant et al. 2003). TNF is probably themost widely studied member of this family of cytokinesand performs multiple roles in innate and adaptive im-mune responses. Most notably, through activation ofNF-B, TNF signaling directly regulates the expressionof antiapoptotic genes such as cIAP1/2 and Bcl-XL (forreview, see Kucharczak et al. 2003). If NF-B signaling isblocked, then exposure to TNF induces rapid apoptosisin most cell types. Aberrant TNF signaling has been im-plicated in many disease states, and anti-TNF antibodiesare in clinical use for the treatment of inflammatory con-ditions such as rheumatoid arthritis (for review, see Ag-

garwal 2003). Therefore, a detailed understanding of IKKactivation through this pathway may provide an oppor-tunity to develop targeted inhibitors with the potentialto treat such conditions. However, despite the tremen-dous clinical relevance of this pathway, and enormouseffort in numerous laboratories, there are still significantgaps in our understanding of the mechanisms that un-derlie TNF-mediated activation of NF-B.

TNF family receptors lack intrinsic enzymatic activ-ity. Instead, signaling is achieved by recruitment of in-tracellular adapter molecules that associate with the cy-toplasmic tail of the TNFR in a signal-dependent manner(for review, see Aggarwal 2003; Wajant et al. 2003). Therecruitment of TNFR1 to membrane microdomains, re-ferred to as lipid rafts, with subsequent assembly of thesignaling complex, is necessary for signaling to NF-Band prevention of apoptosis (Hueber 2003; Legler et al.2003). Ligation of TNFR1 by trimeric TNF causes ag-gregation of the receptor and dissociation of Silencer ofdeath domain (SODD), an endogenous inhibitor of TNFreceptor activity, allowing binding of the TNFR-associ-ated death domain protein (TRADD; Fig. 3A; Jiang et al.1999). SODD may, however, play a redundant role, asstudies on SODD/ mice reveal marginal effects onTNF signaling to NF-B (Endres et al. 2003; Takadaet al. 2003). TRADD subsequently recruits downstreamadapters that result in IKK, p38, JNK, and caspase acti-vation. One critical set of adapter molecules that arerecruited to TRADD through direct interaction is theTNF-receptor-associated factor (TRAF) family (for re-view, see Bradley and Pober 2001; Chung et al. 2002;Dempsey et al. 2003).

Multiple members of the TRAF family includingTRAF2, TRAF3, and TRAF5 have been implicated inTNF signaling, however elucidating their functionalroles has been complicated because of diverse homotypicand heterotypic interactions that can occur betweenmembers of this family. TRAF2 is inducibly recruited tothe TNFR and interacts efficiently with TRADD (Hsu etal. 1996). Surprisingly, TRAF2-deficient mice have intact

TNF signaling to NF-B; instead, TRAF2 appears to me-diate signaling to AP-1 (Yeh et al. 1997). TRAF5 knock-outs also exhibit normal NF-B activation by TNF; how-ever, TRAF2/5 double knockout cells have substantiallyreduced TNF-induced IKK activation (Yeh et al. 1997;Nakano et al. 1999; Tada et al. 2001). Therefore, TRAF2

and TRAF5 appear to play a redundant role in TNF sig-naling to NF-B, although the nature of this role remainsto be determined. It has been suggested that TRAF2 re-cruits the IKK complex following TNF treatmentthrough a direct interaction with the leucine zipper ofIKK or IKK (Devin et al. 2001). TRAF2 also interactswith the serine/threonine kinase, Receptor interactingprotein 1 (RIP1), which can also be independently re-cruited to TRADD. RIP1 is essential for TNF-inducedNF-B activation (Hsu et al. 1996), and RIP-deficientcells exhibit increased apoptosis following TNF stimu-lation owing to their inability to activate NF-B (Kelliheret al. 1998). Intriguingly, RIP may function as a scaffoldmolecule in TNF signaling because its kinase activity is

dispensable for NF-B activation (Hsu et al. 1996; Ting etal. 1996; Devin et al. 2000). RIP1 can bind directly toNEMO and thereby recruit IKK to the TNFR1 signalingcomplex independent of TRAF2 (Zhang et al. 2000). In-terestingly, IKK recruitment in the absence of RIP, pre-sumably through TRAF2, is not sufficient for IKK acti-vation (Devin et al. 2001), demonstrating that RIP has arole in addition to or independent of simple recruitmentof the IKK complex. RIP may nucleate the assembly of asignaling complex that induces IKK activation througholigomerization of NEMO and subsequent autophos-phorylation of IKK (Delhase and Karin 1999). Alterna-tively, the assembly of a signaling complex may facili-tate NF-B signaling by bringing the IKK complex intoclose proximity with an IKK kinase.

An intriguing feature of signaling by TNF is its abilityto stimulate both death and survival. As noted above, thebalanced induction of both pathways usually leads toactivation of cells; however, if activation of NF-B, andconsequently the survival pathways, are blocked, TNFbecomes a potent apoptosis-inducing factor. In general,direct up-regulation by NF-B of factors such as IAPs isresponsible for the antiapoptotic effect of NF-B (for re-view, see Kucharczak et al. 2003). It appears, however,that the two main target transcription factors of TNFsignaling, NF-B and AP-1, also participate directly inarbitrating the choice between survival and death. It hasbeen observed that prolonged activation of JNK/AP-1leads to apoptosis (Guo et al. 1998; De Smaele et al. 2001;Tang et al. 2001). NF-B induction can lead to the syn-thesis of certain effectors that inhibit the JNK pathway,thereby limiting JNK/AP1 activation. One such JNK-in-hibiting factor is GADD45, which inhibits JNK signal-ing by blocking the upstream kinase MKK7 (De Smaeleet al. 2001; Jin et al. 2002; Papa et al. 2004). Although amore recent study showed that activation of JNK is simi-lar to wild-type cells in GADD45/ MEFs (Jin et al.2002), it is possible that GADD45 may cooperate withsome other NF-B-induced inhibitor of JNK. For ex-ample, X-chromosome-linked inhibitor of apoptosis

Signaling to NF-B




9/31

(XIAP) is induced by NF-B, and, in addition to inhibit-ing multiple proapoptotic caspases, it also blocks JNKactivation (Tang et al. 2001). Another NF-B-regulatedgene, cFLIP (Casper), inhibits JNK at the level of TRAF/RIP (Shu et al. 1997; Kreuz et al. 2001). Furthermore, theroles of JNK and NF-B are not entirely antagonistic, as

expression of the antiapoptotic protein cIAP1 is coregu-lated by JNK and NF-B, and consequently TNF-in-duced apoptosis is increased in JNK double-knockoutcells (Lamb et al. 2003; Ventura et al. 2003). Interest-ingly, TRAF-2-deficient cells are more sensitive toTNF-induced cell death although NF-B signaling re-mains intact. This may reflect the requirement forTRAF-2 binding to cIAP in order for it to be localized toits site of function or the requirement of JNK-mediatedJunD activation for appropriate expression of cIAP (Shuet al. 1996; Wang et al. 1998; Lamb et al. 2003).

Lymphotoxin-R, BAFF-R, and CD40: TNFR family

members that signal to NF-B through thealternative pathway

The recent discovery of the alternative signaling path-way leading to inducible p100 processing has provided anovel addition to the well-studied pathway of NF-B ac-tivation through the degradation of IB proteins (for re-view, see Bonizzi and Karin 2004). This represents animportant step forward in understanding how NF-Bfamily members might be differentially regulated. Thealternative pathway is unique in that it is independent ofIKK and NEMO (Claudio et al. 2002; Dejardin et al.2002). Instead, the functional IKK is thought to be IKKhomodimers, which selectively phosphorylate p100 as-sociated with RelB. Therefore, processing of p100 re-leases a subset of transcriptionally active NF-B dimers,consisting mainly of p52:RelB (Dejardin et al. 2002; Xiaoet al. 2004). In contrast to the other pathways discussedhere, in the alternative pathway the mechanism of IKKactivation is known. NF-B inducing kinase (NIK) is re-sponsible for directly phosphorylating and activatingIKK (Senftleben et al. 2001a; Xiao et al. 2001). However,events that occur upstream of NIK are unclear. The keyquestion concerning signaling by these stimuli is howthey are channeled to NIK and IKK, even though theirreceptor signaling domains resemble those of other TNFfamily members. Because LT, BAFF, and CD40L alsoactivate the classical pathway, it would appear that theintracellular signaling domains of these receptors pos-sess additional sequence motifs that allow their couplingto NIK and the alternative pathway. The creation andsystematic analysis of chimeras between these receptorsand those that only signal through the classical pathwaywill probably be necessary to elucidate the mechanismby which the alternative pathway is activated.

Toll/IL-1 receptor signaling to NF-B

Signaling to NF-B mediates multiple aspects of innateand adaptive immunity (for review, see Ghosh et al.1998; Silverman and Maniatis 2001; Bonizzi and Karin

2004). NF-B plays an essential role in early events ofinnate immune responses through Toll-like receptor(TLR) signaling pathways. TLRs are evolutionarily con-served Pattern recognition receptors (PRRs) that recog-nize conserved Pathogen-associated microbial patterns(PAMPs) present on various microbes (for review, see

Janeway and Medzhitov 2002; Kopp and Medzhitov2003; Takeda et al. 2003). The role of TLRs as arbitratorsof the selfnon-self decision means that they play a cen-tral role in innate immunity as well as in the initiationof the adaptive immune responses. To date, 11 mamma-lian TLRs have been described, and each of these signalsto NF-B. These receptors have varied tissue distributionand recognize many different PAMPs including LPS,dsRNA, nonmethylated CpG DNA, and flagellin. Somemembers of the TLR family are also capable of het-erodimerization, thereby further expanding the reper-toire of molecules that are recognized. Significantprogress has been made over the past couple of years indeciphering the relevant signaling pathways that operate

downstream of TLRs (for review, see Barton and Medzhi-tov 2003). The intracellular domain of Toll-like recep-tors bears strong homology with the intracellular do-main of the IL-1 receptor, and it is this shared Toll-IL-1R(TIR) domain that mediates interaction with down-stream signaling adapters that lead to activation of threekey transcription factors, NF-B, AP-1, and IRF3.

TLR signaling is initiated by the recruitment of cyto-solic adapters that all share the TIR domain (for review,see Dunne and ONeill 2003). MyD88 was the first TIRdomain containing adapter protein characterized andwas shown to interact with the TIR domain on TLR/IL-1R cytoplasmic tails by homotypic interaction. MyD88is crucial for normal NF-B induction in response to IL-1,IL-18, and LPS (TLR4; Adachi et al. 1998; Kawai et al.1999). MyD88 recruitment to TLR4 following receptoraggregation leads to recruitment of another TIR-domain-containing adapter, TIRAP or Mal (Fig. 3B). TIRAP me-diates NF-B activation downstream of TLR2 and TLR4,but not IL-1R or other TLRs (Horng et al. 2002; Yama-moto et al. 2002). Currently it is not known how TIRAPselectively acts only in a subset of MyD88-mediated sig-naling pathways (for review, see McGettrick and ONeill2004). MyD88 also contains an N-terminal death domain(DD) that recruits downstream DD-containing signalingmolecules such as the serine/threonine kinase IRAK (forreview, see Janssens and Beyaert 2003). The exact roleplayed by IRAK family members is somewhat enigmaticbecause like RIP, IRAK-1 kinase function is also dispens-able for its role in TLR/IL-1 signaling (Knop and Martin1999; X. Li et al. 1999). IRAK-4 is also required for sig-naling from TLR and IL-1, likely upstream of IRAK-1(Suzuki et al. 2002). IRAK recruitment and activation arenecessary to bring TRAF6 into the signaling complex,although exactly how this is achieved remains mysteri-ous (Qian et al. 2001; Takaesu et al. 2001). TRAF6 isrecruited to TLR/IL-1R and is required for MyD88-de-pendent activation of NF-B (Cao et al. 1996; Wescheet al. 1997; Wu and Arron 2003). TRAF6-deficient cellsexhibit a complete loss of NF-B DNA binding induced

Hayden and Ghosh




10/31

by IL-1R as well as loss of the early NF-B responsedownstream from TLR4 (Lomaga et al. 1999).

The link between TRAF6 and the IKK complex hasremained controversial. Two sets of adapters have beenproposed to link TRAF6 with IKK. The first involves theTransforming growth factor activated kinase 1 (TAK1),

and two associated adapter proteins TAB1 and TAB2(Takaesu et al. 2000). The major support for this signal-ing module came from studies that indicated that theycould be copurified with TRAF6 in cells (Wang et al.2001). The presence of a RING finger domain in TRAF6led to the suggestion that TRAF6 could function as aubiquitin ligase (Deng et al. 2000). Upon stimulation,TRAF6 would ubiquitinate itself or components of theTAK1/TAB1/TAB2 complex, thereby priming them forIKK activation (Sun et al. 2004). However, as of yet thereis no genetic evidence in mammalian systems support-ing such a critical role for TAK1 in TLR/IL-1 signaling.Intriguingly, a recent study indicated that RNAi-medi-ated knock-down of TAK1 inhibited signaling from both

IL-1 and TNF receptors, which was surprising becauseTRAF6 is not involved in the TNF pathway (Takaesu etal. 2003). One possibility is that TAK1 function is notlimited to TRAF6, but rather may extend to other TRAFsthat also possess RING finger domains. A more recentreport has implicated TAK1 as an essential componentin TCR signaling, raising the possibility that contrary tocurrent belief, TRAF proteins are also involved in anti-gen-receptor signaling (Sun et al. 2004). However, thephenotypes of TAB1 and TAB2 knockouts fail to supportany role for these proteins in TLR/IL1 and TNF signal-ing. Knockout of TAB1 and TAB2 both lead to embry-onic lethality at embryonic days 17 (E17) and 12.5(E12.5), respectively (Komatsu et al. 2002; Sanjo et al.2003). Fibroblasts isolated from the embryos displaycompletely normal signaling in response to TNF and IL-1(Sanjo et al. 2003; M.S. Hayden and S. Ghosh, unpubl.).Knocking out TAK1 causes embryonic lethality at E10,thus preventing signaling studies. Therefore, although itremains possible that future development and character-ization of conditionally knocked out TAK1 cells willsupport the proposed role in TLR/IL-1, TNF, and TCRsignaling, it is clear that there are significant complexi-ties yet to be resolved.

Besides the TAK1/TAB1/TAB2 proteins, another pro-tein that has been reported to act as a bridge betweenTRAF6 and the IKK complex is ECSIT (evolutionarilyconserved signaling intermediate in toll pathways).ECSIT was identified in a yeast two-hybrid screen as abinding partner of TRAF6 and was shown to be requiredfor TLR and IL-1 signaling, but not TNF-signaling, usingRNAi knock-down techniques (Kopp et al. 1999; Xiao etal. 2003). Knockout of ECSIT leads to very early embry-onic lethality, thereby preventing isolation of cells suit-able for signaling studies. However, the knockouts ofECSIT display phenotypes that are identical to thoseseen in knockouts of Bmp receptor-1, a member of theTGF-receptor family that is known to function in earlydevelopment. In fact, subsequent analysis showed thatECSIT can function as a coactivator for effectors of Bmp/

TGF signaling, namely, the Smad transcription factors(Xiao et al. 2003). This surprising finding raises the pos-sibility that ECSIT may regulate both TLR and TGF/Bmp pathways, and hence may help provide an explana-tion for why these pathways cross-repress one another(Moustakas and Heldin 2003). It is also important to

point out that the TAK/TAB proteins were also initiallyidentified and characterized as intermediates in TGFsignaling, and hence it is possible that this link betweenadapters in TLR and TGF signaling pathways may bemore extensive than currently imagined.

The initial studies with ECSIT suggested that it mightfunction in TLR signaling by recruiting and activatingthe kinase MEKK1, which is also involved in TGF sig-naling (Kopp et al. 1999; Zhang et al. 2003). However,knockouts of MEKK1 do not display any overt phenotypein TLR or TNF signaling, suggesting that some otherkinase might be involved (Xia et al. 2000; Yujiri et al.2000). More recently, MEKK3 has been strongly impli-cated in TLR signaling. MEKK3-deficient cells do not

transcribe IL-6 following TLR4 or IL-1R stimulation andexhibit delayed and weak NF-B DNA binding followingLPS stimulation (Huang et al. 2004). Although TRAF6and MEKK3 inducibly associate in TLR4 signaling, themechanisms that regulate this process are not known(Huang et al. 2004). Therefore, it is possible, althoughnot proven, that ECSIT exerts its role in TLR signalingby somehow modulating the function of MEKK3.

Multiple TLRs are also capable of signaling in the ab-sence of MyD88. LPS stimulation in MyD88/ cells re-sults in NF-B activation with slower kinetics than fromnormal TLR signaling and leads to expression of only asubset of target genes (Kawai et al. 1999). TRIF (TICAM-1), a TIR-domain-containing adapter, mediates activa-tion of NF-B in the absence of MyD88 when cells arestimulated through TLR3 and TLR4 (Oshiumi et al.2003). TRIF expression is regulated by NF-B and istherefore induced by TLR and IL-1R signaling (Hardy etal. 2004). Studies using cells from TRIF-deficient micedemonstrate that TRIF is required for early and late NF-B responses and IRF3 responses to LPS, but not for JNKactivation (M. Yamamoto et al. 2003a; Akira 2004). Re-constitution of TRIF/ cells with a mutant form lackingthe TRAF-binding domain restores induction of IFN-responsive genes, via activation of IRF3, but not NF-Bactivation, indicating that TRIF is the point of diver-gence in signaling to NF-B and IRF3 by TLR4. Recentlyit has been reported that TRIF binds RIP1 and thatRIP1/ MEFs have decreased NF-B signaling fromTLR3-poly(I:C) (Meylan et al. 2004). Finally, anotherTIR-domain-containing adapter, TRAM (TRIF-relatedadapter molecule), functions upstream of TRIF inMyD88-independent signaling from TLR4. TRAM is re-quired for IRF-3 activation and for the delayed phase ofNF-B activation following TLR4 engagement. TLR4-in-duced IRAK activation by MyD88, however, is unaf-fected by the absence of TRAM (M. Yamamoto et al.2003b). TRAM does not function in TLR3 or IL-1R sig-naling pathways (Fitzgerald et al. 2003; M. Yamamoto etal. 2003b).

Signaling to NF-B




11/31

Two divergent members of the IKK family, IKKi (IKK)and TBK1 (T2K), have been implicated downstream ofTRIF in signaling to IRF-3 and IRF-7, following engage-ment of TLR3 and TLR4 (Hemmi et al. 2004; Sharmaet al. 2003). Although initially both of these kinases wereimplicated in regulation of NF-B activity, recent results

have cast doubt on those conclusions. Studies usingknockout cells indicate that neither is required for NF-B activation by LPS or TNF (Akira 2004; Hemmi et al.2004). Also in contrast to the study reporting the knock-out of TBK1, a recent study reported that there was nodiscernible effect on the transcriptional activity of NF-B in TBK1 knockout cells following stimulation withmultiple PAMPs (McWhirter et al. 2004). The reason forthis discrepancy remains to be resolved. IKKi/ cellshave normal induction of IRF3 following stimulationwith LPS (hence TLR4), whereas TBK1/ cells do not(Hemmi et al. 2004). IKKi expression is regulated by NF-B, and it appears that IKKi may be constitutively activeonce expressed (Mercurio 2004). IKKi may facilitate

CCAAA/enhancer-binding protein (C/EBP) pathwaysthat contribute to the expression of a subset of genes thatare induced by IL-1, LPS, TNF, or PMA (Kravchenko etal. 2003). Therefore, TBK1 but not IKKi mediates TLRsignaling to IRF-3, likely through direct phosphorylationof IRF-3, and neither kinase appears to be directly in-volved in TLR signaling to NF-B.

The Toll-signaling pathway can also be negativelyregulated by proteins that are induced or activated uponTLR signaling and therefore may help to limit signalingfrom these receptors. For example, Tollip is an adapterprotein found in association with IRAK in the IL-1R sig-naling pathway (Burns et al. 2000; Zhang and Ghosh2002). Tollip binds to IRAK, and to TLR2, TLR4, or IL-1Rupon signaling. Activation of IRAK by MyD88 results inautophosphorylation of IRAK and phosphorylation ofTollip, causing their dissociation (Bulut et al. 2001;Zhang and Ghosh 2002). Tollip can suppress TLR/IL-1Rsignaling when overexpressed and, therefore, functionsas a negative regulator of IRAK activity. It is not knownwhether Tollip functions in down-regulation of TLR sig-naling or acts as a constitutive inhibitor of IRAK. It hasbeen suggested that high levels of Tollip in the intestinalepithelium help to prevent inflammation in response tocommensals (Melmed et al. 2003). Another well-charac-terized negative regulator is a member of the IRAK fam-ily, IRAK-M. IRAK-M is inducibly expressed in myeloidcells and was initially shown to induce NF-B signalingwhen overexpressed (Wesche et al. 1999). However, inknockout mice, TLR and IL-1R signaling to NF-B isenhanced, and the cells do not become tolerant of re-peated LPS stimulation (Kobayashi et al. 2002). IRAK-Mprevents the release of IRAK-1 and IRAK-4 from theTLR/MyD88 signaling complex, and therefore inhibitsthe association between IRAKs and TRAF6. Further up-stream, alternative splice variants of MyD88 have beenreported to inhibit TLR signaling by competitively bind-ing IRAK4 (Janssens et al. 2002, 2003; Burns et al. 2003).Finally, at the level of the membrane, SIGIRR (TIR8), amember of the IL-1R family that contains an extracellu-

lar Ig domain, has been shown to decrease TLR/IL-1Rsignaling (Thomassen et al. 1999; Wald et al. 2003). InSIGIRR/ cells, treatment with either LPS or IL-1 leadsto enhanced and prolonged activation of both NF-B andJNK. SIGIRR binds to other Toll/IL-1 receptors and in-teracts with IRAK and TRAF6 (Wald et al. 2003). It re-

mains to be seen whether it inhibits signaling by com-peting for adapters or preventing the dissociation ofIRAK from the receptor in a manner analogous to IRAK-M. Interestingly, SIGIRR is most abundantly expressedin epithelial cells, leading one to speculate that it maysuppress TLR signaling at sites of exposure to commen-sal bacteria.

BCR and TCR signaling to NF-B

Signaling from BCRs and TCRs is critical for mountingadaptive immune responses and has consequently beenof great interest to numerous laboratories. Activation ofNF-B downstream of BCR and TCR allows antigen-spe-

cific proliferation and maturation of lymphocytes intoeffector cells. Signaling through these two antigen recep-tors is functionally analogous, although the moleculardetails differ. From a clinical standpoint, understandinghow these signaling pathways operate may provide noveltargets for therapies directed at autoimmunity, inflam-matory disorders, and transplant rejection.

Signaling through the TCR, or analogously throughthe BCR, and various costimulatory molecules leads toNF-B activation (for review, see Ruland and Mak 2003).TCR ligation induces phosphorylation of key residues onITAMs present on TCR and CD3 (Fig. 3C). Phosphory-lated ITAMs recruit SH2-domain-containing adapters,most notably the Syk family tyrosine kinases Lck andZAP70. However, the link between the receptor proxi-mal tyrosine kinases and NF-B is poorly defined. Recentstudies have implicated several potential signaling inter-mediates that appear to comprise a novel signaling path-way, including PKC, CARMA1/CARD11, BCL10, andMALT1 (for reviews, see Lucas et al. 2004; Simeoni et al.2004; Thome 2004). The IKK complex is rapidly re-cruited to the immunological synapse through an un-known mechanism and can be colocalized to the TCR byconfocal immunofluorescence analysis (Khoshnan et al.2000). In ZAP-70-deficient cells, NF-B activation in-duced by TCR stimulation can be rescued with a NEMO-(SH2)2 chimera that targets NEMO to the immunologi-cal synapse, indicating that events downstream fromZAP-70 are only required for IKK recruitment (Weil et al.2003). One intriguing hypothesis that has been proposedis that recruitment of the IKK complex is dependent onchanges in cytoskeletal structure during formation of theimmunological synapse (Weil et al. 2003). Although theimportance of scaffolding molecules in the MAPK path-way has been established, little is known thus far abouthow these processes affect NF-B signaling downstreamof antigen receptors.

It is clear from knockout studies that PKC is essentialfor activation of NF-B by TCR (Sun et al. 2000), but thespecific role played by PKC in T cells, and PKC1/2 in

Hayden and Ghosh




12/31

B cells (Guo et al. 2004), remains to be elucidated. PKCis specifically recruited to the immunological synapse,although how this specificity for PKC is achieved re-mains unclear because T cells express most of the otherPKC isoforms. The exact role of PKC in the immuno-logical synapse is also unclear, as is the mechanism re-

sponsible for recruiting PKC. PKC is capable of di-rectly interacting with the IKK complex in primary Tcells (Khoshnan et al. 2000), and it is possible that PKCmight act as a scaffold/adapter to link events in the syn-apse with the other essential components in this path-way, namely, CARMA1/CARD11, BCL10, and MALT1.Knockouts of each of these genes leads to a specific blockin NF-B activation in response to antigen-receptor sig-naling, and therefore current efforts are focused on de-termining how these proteins are linked to the IKK com-plex (Ruland et al. 2001; Hara et al. 2003; Ruefli-Brasseet al. 2003).

The MAGUK family protein CARD11/CARMA-1 isrequired for PKC-mediated activation of NF-B follow-

ing TCR ligation (Gaide et al. 2002; Hara et al. 2003).BCL10, which interacts with MALT1 and cIAPs, is alsocritical for NF-B activation via the BCR (Ruland etal. 2001). BCL10 interacts with CARMA-1 and under-goes CARMA-1-dependent phosphorylation, althoughCARMA-1 lacks kinase activity (Bertin et al. 2001; Gaideet al. 2001). BCL10 knockout mice are unable to activateNF-B in response to antigen-receptor stimulation (Ru-land et al. 2001). Interestingly, genetic evidence of a rolefor RIP2 has recently been reported in T-cell signaling(Ruefli-Brasse et al. 2004). RIP2 associates with BCL10and is necessary for TCR-induced BCL10 phosphoryla-tion and IKK activation. It is not yet clear how upstreammediators of TCR signaling may regulate RIP2 and, inturn, how RIP2 might affect IKK activity. BCL10 oligo-merization has been implicated in IKK activationthrough a process that involves ubiquitination of NEMO(Zhou et al. 2004). This ubiquitination event appears tobe mediated by MALT1 and possibly CARMA-1. A sur-prising recent finding is that TCR-induced ubiquitina-tion of NEMO may depend on TRAF6, which has notpreviously been thought to play any role in TCR signal-ing, as T-cell deficits have not been reported in TRAF6-deficient mice (Lomaga et al. 1999; Sun et al. 2004). It isinteresting to note that RIP2 associates with TRAF6, andone could speculate that this interaction has a functionalrole in IKK activation downstream of antigen receptorsignaling (McCarthy et al. 1998).

Cell stress

Various intracellular stressors, both physiological andpathological, induce the activation of NF-B (Fig. 3). Un-derstanding the pathways responsible will require a sig-nificant shift in thinking from traditional signaling para-digms, as there is not an initiating receptor ligationevent. Although less conventional, the physiological rel-evance of these pathways is undeniable. A properly or-chestrated response to genotoxic stress is a fundamentaldefense against transformation and cancer. However, the

biological function of NF-B in this response is not im-mediately clear. Although it is generally hypothesizedthat activation of NF-B provides an antiapoptotic sig-nal, thereby providing a window of opportunity for thecell to repair DNA damage, this has not been formallyproven. In fact, under some circumstances, NF-B sig-

naling can play a proapoptotic role in response to stimulisuch as UV irradiation (Campbell et al. 2004). Neverthe-less, cell irradiation and DNA damage have long beenknown to activate NF-B (Brach et al. 1991). However,considerable discrepancies exist in the experimental sys-tems used to investigate cell stress, and this has hinderedprogress in this area. Many of the agents used to inducegenotoxic stress do so through diverse perturbations ofcell physiology and could potentially activate NF-Bthrough multiple mechanisms. Even when selectivelyinduced, DNA damage elicits a highly complex cellularresponse that has been the subject of considerable re-search. The product of the gene mutated in ataxia-telan-giectasia (ATM), a phosphoinositide 3-kinase-related

kinase (PIKK), has a central role in sensing DNA dou-ble-strand breaks (DSBs) and causing the consequent ac-tivation of the DNA repair machinery (for review, seeYang et al. 2004). It has been shown using ATM knock-out cells, and cells from patients with ataxia-telangiec-tasia, that ATM is required for NF-B activation follow-ing ionizing radiation (Li et al. 2001). Topoisomerase in-hibitors, such as camptothecin, can also activate NF-B,presumably through the generation of DSBs (Piret andPiette 1996). Although NF-B activation in response tolow-dose radiation could be due to production of reactiveoxygen species (ROS; Mohan and Meltz 1994), such amodel seems inconsistent with a requirement for ATMin this process.

Activation of IKK following treatment with topoisom-erase inhibitors is dependent on the zinc finger domainin NEMO (Huang et al. 2002). A recent report showedthat modification of NEMO with the small ubiquitin-like modifier (SUMO-1) plays a mechanistic role in theactivation of IKK following treatment with topoisomer-ase inhibitors (Huang et al. 2003). Sumolation of NEMOmay be necessary for the recruitment of NEMO to thenucleus resulting in ATM-dependent ubiquitination ofNEMO and subsequent activation of IKK. Other studieshave suggested that RIP is required for the activation ofIKK in response to DNA damage in a manner that isindependent of TRAF family members (Hur et al. 2003).It is unclear how these data can be reconciled with theproposed role for SUMO-1 in this process.

NF-B has also been shown to be activated followingUV-C irradiation; however, this response is not depen-dent on DNA damage (Devary et al. 1993). NF-B acti-vation following UV-C irradiation has been shown to beindependent of IB Ser 32/36 phosphorylation and is,therefore, IKK-independent (Li and Karin 1998; Katoet al. 2003; Tergaonkar et al. 2003). Instead, it mightoperate through activation of p38 pathways leading tothe activation of casein kinase 2 (CK2). CK2 then di-rectly phosphorylates IB on C terminus serine resi-dues. Whereas mutation of these serines abolishes the

Signaling to NF-B




13/31

ability of CK2 to induce IB degradation following UVirradiation (Barroga et al. 1995; Lin et al. 1996; Mc-Elhinny et al. 1996; Kato et al. 2003), some early tran-sient activation of NF-B appears intact in CK2 knock-down cells, p38 knockout cells, or IB knockout cellsreconstituted with IB in which the CK2 phosphoac-

ceptor sites were mutated (Kato et al. 2003). A similarpathway of NF-B activation has been proposed in cellstreated with doxorubicin (Tergaonkar et al. 2003). Doxo-rubicins mechanism of action is multipartite, and there-fore it is still unclear whether the NF-B activation ob-served is caused by generation of free radicals and ROSvia effects similar to UV-C, or caused by topoisomeraseinhibition mediated through DNA intercalation. Morerecently, these findings have been called into questionby work showing that doxorubicin and UV-C induce ap-optosis through release of NF-B dimers that can sup-press transcription of antiapoptotic genes by recruitinghistone deacetylases (Ashikawa et al. 2004; Campbell etal. 2004). It is not clear whether these findings are gen-

erally applicable or are cell-type-specific. In summary,whereas nonlethal UV-C irradiation leads to activationof CK2 downstream from ROS, and consequent degrada-tion of IB, the response to DNA damage is mediatedthrough ATM, and perhaps RIP, to cause IKK activation.In the future, clarification of these issues will require asystematic investigation in multiple cell types usingcarefully controlled induction of various types of cellstress. In particular, the issue of ROS effects on NF-Bsignaling merits detailed examination, given that this islikely the source of much confusion in the investigationof multiple NF-B signaling pathways (Zhang and Chen2004).

Regulation of IKK

With the possible exception of CK2, the common featureof all pathways leading to the activation of NF-B is theactivation of one of the IB kinases. Consequently, thisis the most important regulatory step in determining theNF-B response to a given stimulus. Current evidencepoints toward a mechanism that depends on the assem-bly of complex signalosomes for the activation of IKKwhere the key event is induced proximity allowingcross-phosphorylation between the IKK subunits. How-ever, a role for an activating IKK kinase cannot be com-pletely ruled out. In fact, activation of IKK in the alter-

native pathway by NIK is well established. Determiningwhether such an event also occurs in the activation ofthe classical pathway will require a more comprehensiveanalysis of the composition of the IKK complexes in vari-ous tissues and cells under different physiological con-ditions.

Structure and function of the IKK complex

The initial efforts to identify the IB kinase led to thepurification of a high-molecular-weight kinase complexfrom unstimulated cells that was capable of phosphory-

lating IB on the appropriate serine residues (Chen et al.1996; Lee et al. 1997). Subsequently, multiple groupsidentified a stimulus-dependent kinase activity that wastermed the IB kinase (DiDonato et al. 1997; Mercurioet al. 1997; Woronicz et al. 1997; Zandi et al. 1997). Thefirst component of this 700900-kDa complex was

shown to be CHUK, a serine/threonine kinase that, re-markably, had been proposed to be a regulator of genetranscription based solely on sequence homology (Con-nelly and Marcu 1995). CHUK, a 745-amino acid protein,has since been renamed IKK. IKK was discoveredshortly thereafter based on sequence homology and bio-chemical purification. IKK and IKK are serine/threo-nine kinases that are characterized by the presence ofan N-terminal kinase domain, a C-terminal helixloophelix (HLH) domain, and a leucine zipper domain (Fig. 1).NEMO (also known as IKK, IKKAP1, and Fip-3), theregulatory subunit of the IKK complex, was initially iso-lated through genetic complementation cloning in anNF-B-unresponsive cell line, and subsequently by affin-

ity purification and as a factor interacting with an ad-enoviral inhibitor of NF-B (Rothwarf et al. 1998; Ya-maoka et al. 1998; Y. Li et al. 1999; Mercurio et al. 1999).NEMO is a 48-kDa protein that is not related to IKKand IKK and contains a C-terminal zinc finger-like do-main, a leucine zipper, and N-terminal and C-terminalcoiled-coil domains (Fig. 1).

IKK and IKK share 52% overall sequence identityand 65% identity within the catalytic domain. BothIKK and IKK can be inactivated through the mutationof a single lysine, Lys 44, within the predicted ATP-bind-ing site (Fig. 1; Mercurio et al. 1997; Woronicz et al.1997; Zandi et al. 1997). IKK and IKK are capable ofphosphorylating multiple members of the IB family atmultiple sites. These kinases also have substrates out-side of the IB family, most notably p65. In vitro studiesindicate that the IKK/IKK heterodimers have highercatalytic efficiency than either homodimer (Huynh et al.2000). Although IKK and IKK display different speci-ficities for different IB family members, it is not knownwhether they are independently regulated. Althoughboth IKK and IKK are capable of phosphorylating IBat Ser 32 and Ser 36, and IB at Ser 19 and Ser 23, IKKis less efficient and cannot replace IKK in knockoutstudies (DiDonato et al. 1997; Mercurio et al. 1997; Reg-nier et al. 1997). Both IKK and IKK are less efficientkinases for IB than for IB, and this difference mayexplain the delayed degradation kinetics of IB follow-ing stimulation with TNF (Wu and Ghosh 2003). IBbound to NF-B is a better substrate for IKK than freeIB, which may prevent the degradation of newly syn-thesized IB, thus allowing it to accumulate and down-regulate NF-B activity (Zandi et al. 1998). It is unclear,however, how the enzymology of the IKKs is affected byheterodimerization and complex formation in vivo.

IKK and IKK dimerization is dependent on the leu-cine zipper domain, which is therefore required for ki-nase activity (Mercurio et al. 1997; Woronicz et al. 1997;Zandi et al. 1997). Available data indicate that IKK andIKK preferentially form heterodimers in vivo. However,

Hayden and Ghosh




14/31

the sequence features that determine this preferencehave not been elucidated. Studies using knockout cellsshow that both IKK and IKK are capable of ho-modimerization, although evidence for homodimers inunmanipulated cells is more controversial. Immuno-depletion studies have demonstrated the presence of

IKK/NEMO complexes activated by anti-CD3/CD28 inprimary T cells (Khoshnan et al. 1999). An IKK-onlycomplex was also shown to exist in HeLa cells by affinitypurification, anion exchange chromatography, and im-munodepletion, although this complex appeared to beonly weakly activated by TNF (Mercurio et al. 1999).Also, because in the alternative pathway, phosphoryla-tion of p100 in response to BAFF or LT stimulation canoccur in the absence of IKK and NEMO (Senftlebenet al. 2001a), it has been suggested that there is a discreteIKK-only complex. However, there is not yet definitiveevidence that IKK homodimers, which are not bound toNEMO, exist in vivo and mediate signaling via the al-ternative pathway.

The HLH domain is required for full IKK activity andis also involved in the down-regulation of kinase activ-ity. Deletion mutants lacking the HLH domain have di-minished kinase activity following overexpression orstimulation (Delhase et al. 1999). Interestingly, a C-ter-minal deletion mutant of IKK (1559) can be function-ally rescued through the coexpression of the HLH do-main of IKK (amino acids 558756), indicating that adirect interaction between the HLH domain and kinasedomain is necessary for activity. However, when the C-terminal domain of IKK is expressed in trans, or whenthe C-terminal serine residues are mutated, IKK activityis prolonged. More recent data have shown that remov-ing the HLH region also abrogates the binding of IKK toNEMO, and hence the role of the HLH domain may be infacilitating the assembly of functional IKKNEMO com-plexes (May et al. 2002).

IKK and IKK interact with NEMO through a C-ter-minal hexapeptide sequence (Leu-Asp-Trp-Ser-Trp-Leu)termed the NEMO binding domain (NBD; Fig. 1; Mayet al. 2000, 2002). A peptide encompassing this sequencewhen coupled to the cell-permeable antennapedia do-main blocks NF-B signaling by disrupting the IKK com-plex (May et al. 2000, 2002). NEMO binding to IKK re-quires residues 135231, located within the first coiled-coil motif, and can interact with the NBD of both IKKand IKK. Although in vitro assembly of the complexindicates that only IKK assembles with NEMO, datafrom IKK knockout mice clearly indicate otherwise, asIKK/NEMO complexes are readily formed (Rothwarfet al. 1998; Yamaoka et al. 1998; Mercurio et al. 1999).Competition experiments using the NBD peptide indi-cate that IKK binds to NEMO with considerably higheraffinity (May et al. 2002). Furthermore, IKK is less sen-sitive to mutation of two residues within the NBD (Asp749 and Trp 742) that abolish IKK binding to NEMO,indicating that the binding of IKK is less stringent. Al-though IKK or IKK produced in baculovirus or yeastsystems is catalytically active, association with NEMOis required for in vivo IB phosphorylation. In fact, ex-

change of C-terminal portions of IKK and IKK, whichcontain the NBD and HLH domain, confers some in-crease in IKK-like behavior on IKK (Kwak et al. 2000).Taken together, these results indicate that the C-termi-nal portion of IKK, either through the interaction withNEMO or through inducing a conformational change,

strongly affects kinase activity and specificity.NEMO has been shown to form tetramers in vitro, and

it appears that NEMO can be detected in several oligo-meric states in vivo (Tegethoff et al. 2003). In contrast, ithas been reported recently that NEMO multimerizationis cell-type-specific, with NEMO existing as dimer, tri-mer, or both, but never as tetramer (Israel 2004). Thisdiscrepancy is likely attributable to differences in experi-mental systems, such as the use of different homo-bifunctional cross-linking agents. It has also been ar-gued that NEMO exists as a monomer in resting cellsand undergoes regulated trimerizationNEMO/IKK/IKK NEMO3/IKK/IKK (Israel 2004). Upon trimer-ization, the second coiled-coil domain (CC2, amino acids

248291) and leucine zipper domain (LZ, amino acids294341) align in an antiparallel fashion with an overallstructure similar to the HIV-1 gp41 ectodomain (Agou etal. 2004). When recombinant NEMO peptides encom-passing the LZ and CC2 domains are mixed in vitro, theyself-assemble into a pseudohexameric state. These re-sults are consistent with previous reports that oligomer-ization of NEMO may occur through association of theC-terminal domain with RIP, and that forced oligomer-ization of NEMO activates IKK (Poyet et al. 2000). In-deed, peptide mimics of CC2 and the LZ domain havebeen found to inhibit NEMO oligomerization and NF-Bactivation without interfering with NEMO binding tothe IKKs (Israel 2004).

Thus far, IKK, IKK, and NEMO are the only com-ponents that have been conclusively demonstrated to bea part of the IKK complex. The stoichiometry of thesecomponents in the complex continues to be debated, al-though many researchers have supported the idea of aheterodimer of IKK and IKK associating with NEMO(Rothwarf and Karin 1999; Miller and Zandi 2001). Thisdebate about the composition of the complex has beendriven by the discrepancy between the apparent molecu-lar weight observed during gel filtration chromatography(700900 kDa) and the predicted size based on the actualmolecular weight of the components (200350 kDa).There are, however, certain difficulties in defining newcomponents of the IKK complex with any certainty. Oneproblem has been the assumption that there is only oneIKK complex. Most studies have, therefore, not madeefforts to distinguish whether all or only a subset of theIKK complexes contain a given constituent. Partial pu-rification efforts have identified proteins that likely as-sociate only with a small subset of IKK, if at all, and maydo so in a tissue-specific manner. Therefore, the onlydefinitive components of the complex remain IKK,IKK, and NEMO.

Perhaps not surprisingly, various members of the NF-B and IB family can associate with the IKK complex(Cohen et al. 1998; Heilker et al. 1999; Bouwmeester et

Signaling to NF-B




15/31

al. 2004). It has been shown more recently that the chap-erone HSP-90/Cdc37 is constitutively associated withthe IKK complex (G. Chen et al. 2002). A role for HSP-90in IKK was supported by functional data generated usingthe HSP-90 inhibitor geldanamycin (GA), which inhibitsactivation of IKK by TNF-. HSP-90, however, can asso-

ciate with multiple other kinases that are involved in theNF-B pathway, including RIP, MEKK3, MEKK1, NIK,AKT, and TBK1 (Fisher et al. 2000; Sato et al. 2000; Goesand Martin 2001). It has also been speculated thatthrough binding to NEMO, HSP-90 or HSP-70 may regu-late NEMOs oligomeric state (Agou et al. 2002; G. Chenet al. 2002). Therefore, it is difficult to ascribe a specificrole for HSP-90/Cdc37 as a bona fide component of theIKK complex. Instead, HSP-90 may function as a chap-erone during assembly of the complex or during its re-generation following signaling.

ELKS is a 105-kDa regulatory protein that has recentlybeen proposed to be an IKK-interacting protein (Sigala etal. 2004). ELKS was copurified with IKK by immunoaf-

finity purification, and has structural motifs reminiscentof NEMO with coiled-coil and leucine zipper domains.In unstimulated cells, ELKS appears to be a stoichiomet-ric component of the IKK complex based on immuno-depletion experiments. Knock-down of ELKS usingRNAi blocks the early activation of NF-B followingTNF or IL-1 stimulation. However, it appears that thelate phase of NF-B activation may be intact in thesecells. Furthermore, it was shown that when ELKS isknocked down, the IKK complex fails to associate withIB, suggesting that ELKS mediates interaction of theIKK complex with IB but not IB (Sigala et al. 2004).Therefore, it is tempting to speculate that ELKS providesthe link between MEKK3, as discussed below, and as-sembly of the IKK/IB complex necessary for activationof the early NF-B response. Surprisingly, whereas over-expression of full-length ELKS does not activate IKK, ex-pression of an N-terminal deletion mutant of ELKS does.The biological significance of this regulatory componenthas not yet been verified genetically.

There are significant emerging data arguing against theexistence of additional components for the IKK complex.Recombinant NEMO when mixed with either IKK orIKK can self-assemble into a complex of apparent mo-lecular weight that is similar to the purified complex(Krappmann et al. 2000). Coproduction of NEMO withIKK and IKK in yeast also yields an appropriately sizedcomplex (Miller and Zandi 2001). This may be becauseNEMO has a large Stokes radius, which explains whyNEMO trimers migrate with an apparent molecularmass of 550 kDa during gel filtration chromatography(Agou et al. 2004). Alternatively, this may simply reflectan IKK//NEMO complex of higher-order stoichiom-etrya dimer of trimers. An additional IKK complex, ofan apparent molecular mass of 300 kDa has been re-ported by multiple groups (Zandi et al. 1997; Yamaokaet al. 1998). This complex seems to consist of only IKKand IKK (Yamaoka et al. 1998; Zandi et al. 1998); how-ever, the physiological role of this IKK complex remainsto be determined.

Genetic analysis of the role of IKK subunits

Knockouts of each of the three IKK subunits have beengenerated. The IKK knockout phenotype is similar tothat observed for the p65 knockout, supporting an argu-ment for a central role for IKK in mediating NF-B sig-

naling via TNF (Q. Li et al. 1999b; Z.W. Li et al. 1999;Tanaka et al. 1999). Surprisingly, even ikk+/ heterozy-gous animals have decreased NF-B activity (Z.W. Liet al. 1999). The similarity of their phenotypes and theability to rescue both by knocking out TNRF1 demon-strate that IKK is required for TNF signaling to NF-Band the inhibition of TNF-induced apoptosis (Z.W. Li etal. 1999; Senftleben et al. 2001b). Of note, TNFR/IKKdouble knockouts show a more pronounced defect in in-nate immune responses to bacterial infection, only sur-viving a few days after birth, whereas the p65 knockoutscan survive for many months. These data indicate thatthere is no partial compensation for IKK by IKK or anyother kinase.

The ikk/ mice are born live, but die perinatallyfrom multiple morphological defects, particularly in epi-dermal and skeletal development (Hu et al. 1999; Q. Liet al. 1999a; Takeda et al. 1999). In contrast to theikk/ mice, ikk/ mice do not exhibit liver apoptosis,and cells derived from these mice have only mildly re-duced NF-B activation in response to TNF. These ini-tial characterizations indicated that IKK had little rolein NF-B signaling, and perhaps had some unrelated rolein development, as knockout mice exhibit altered limbbud morphology at E12.5 (Q. Li et al. 1999a; Takeda et al.1999). The epidermal layer of the skin in IKK-deficientmice fails to differentiate properly, and they exhibit mul-tiple skeletal abnormalities. Subsequent analysis hasshown that these defects are NF-B-independent. Theikk/ mouse can be rescued by knocking in an IKK inwhich both of the activation loop serines are mutated toalanines (IKKAA; Cao et al. 2001; Hu et al. 2001),indicating that the role of IKK in epidermal develop-ment is independent of its kinase activity. More re-cently, conditional expression of IKK or kinase deadIKK in basal keratinocytes has been shown to be suffi-cient to rescue the morphogenetic defects observed inthe IKK/ mice (Sil et al. 2004). The mechanism ofNF-B-independent control of keratinocyte differentia-tion by IKK is unknown; however, it is dependent onthe presence of a previously unrecognized NLS in IKK(amino acids 232240; Sil et al. 2004). IKK/ double-knockout mice show further defects in NF-B signaling.The mice are characterized by failure of neurulation dur-ing development, and MEFs from double-knockout em-bryos do not respond to TNF, IL-1, or LPS (Q. Li et al.2000). Interestingly, the neurulation defect is not foundin NEMO-deficient mice.

NEMO, like IKK, is required for signaling throughthe classical NF-B pathway. Studies using deletion mu-tants of NEMO demonstrate that the C-terminal portionis required for stimulus-induced activation of IKKthrough interaction with upstream adapters, whereasbinding to IKK and IKK occurs using sequences from

Hayden and Ghosh




16/31

the N terminus (Rothwarf et al. 1998; Makris et al. 2002).NEMO-deficient mice die at E12.5E13 from massiveliver apoptosis (Rudolph et al. 2000), and MEFs fromNEMO/ mice have a complete block in NF-B activa-tion by TNF, LPS, or IL-1. Therefore, based on studiesusing knockouts of the different IKK subunits, it was

concluded that only IKK and NEMO participate in sig-naling to NF-B in response to proinflammatory cyto-kines. One notable observation from the characteriza-tion of the NEMO/ MEFs is a slight increase in thebaseline IKK activity, although this activity is not induc-ible. Finally, the female NEMO+/ mouse replicates thefindings from human Incontinentia pigmenti patients,where female heterozygotes display a severe inflamma-tory skin disease that is probably the result of randomX-inactivation, as NEMO is located on the X chromo-some (Makris et al. 2000; Schmidt-Supprian et al. 2000).

Mechanism of IKK activation

Activation of the IKK complex is dependent on phos-phorylation, and can be decreased by phosphatase treat-ment of IKK purified from activated cells (DiDonatoet al. 1997). Most research has focused on the phosphory-lation events on IKK, as the importance of IKK hasbeen appreciated more recently. IKK is phosphorylatedon two sites within the activation loop of the kinasedomain: Ser 177 and Ser 181 (Delhase et al. 1999). IKKis also phosphorylated outside of the activation loop atmultiple serines that are C-terminal of the HLH motif,most likely through autophosphorylation following ac-tivation. Deletion mutants that lack the C-terminal ser-ine-rich segment are activated normally in response toTNF but remain active for longer periods of time, indi-cating that autophosphorylation or trans-autophos-phorylation of IKK is a negative regulatory mechanism(Delhase et al. 1999).

Multiple upstream kinases have been suggested to actas IKK kinases (IKK-K). However, most such candidatesfailed to survive the test of gene knockouts. Such accu-mulation of negative results has led to the increasingbelief that the IKK complex is primarily regulatedthrough trans-autophosphorylation. Indeed, many of theproposed IKK-Ks have been found to play the role ofadaptors instead of kinases. Trans-autophosphorylationcaused by induced proximity and/or conformationalchange is an appealing mechanism of IKK activation andcould be mediated through recruitment of the IKK com-plex by multimeric adapter/scaffolding complexes suchas RIP or TRAF family members (Delhase et al. 1999;Inohara et al. 2000; Poyet et al. 2000; Tang et al. 2003).To date, however, these studies have relied on overex-pression of IKK components that may have altered thestoichiometry of the IKK complex. Therefore, the role ofIKK-K or autophosphorylation must be evaluated on acase-by-case basis.

The strongest evidence for an IKK-K is in the alterna-tive pathway of NF-B signaling. NF-B inducing kinase(NIK) is a MAP3K family member that directly phos-phorylates and activates IKK (Regnier et al. 1997; X. Lin

et al. 1998; Ling et al. 1998; Uhlik et al. 1998) in responseto activation of the alternative pathway through selectmembers of the TNFR family including LTR andBAFFR. Although NIK has also been implicated in thecanonical signaling pathway, and does associate incom

Signaling to NF-kB

Documents

Transcript of Signaling to NF-kB