the regulatory logic of the nf- b signaling system2006). the canonical ikk complexes act on the...

The Regulatory Logic of the NF-kBSignaling System

Ellen O’Dea and Alexander Hoffmann

Signaling Systems Laboratory, Department of Chemistry and Biochemistry, University of California,San Diego, La Jolla, California 92093

Correspondence: [email protected]

NF-kB refers to multiple dimers of Rel homology domain (RHD) containing polypeptides,which are controlled bya stimulus-responsive signaling system that mediates the physiologicalresponses to inflammatory intercellular cytokines, pathogen exposure, and developmentalsignals. TheNF-kBsignalingsystemoperatesontransientor short timescales, relevant to inflam-mation and immune responses, and on longer-term timescales relevant to cell differentiationand organ formation. Here, we summarize our current understanding of the kinetic mechan-isms that allow for NF-kB regulation at these different timescales. We distinguish betweenthe regulation of NF-kB dimer formation and the regulation of NF-kB activity. Given thenumber of regulators and reactions involved, the NF-kB signaling system is capable of integrat-ing a multitude of signals to tune NF-kB activity, signal dose responsiveness, and dynamiccontrol. We discuss the prevailing mechanisms that mediate signaling cross talk.

How can the regulatory logic of a signalingsystem be understood? The regulatory

logic refers to the properties of a system thatare not evident by studying one molecular com-ponent in isolation, or even the interactionbetween two; hence, they are sometimes referredto as emergent system properties. Straight-forward emergent properties of a signalingsystem are signal-dose responses (which maybe linear or sinusoidal) and dynamic control ofthe response (which may be fast or slow ramp-ing, transient or oscillatory). More complexemergent properties may pertain to the inte-gration of different signals (synergistically orantagonistically), memory functions, or contin-gencies for prior stimulus exposures to signal

transduction. Those properties are mediatedby the molecular components arranged in a par-ticular networktopology.Quantitative measure-ments of the relevant biochemical reactions is aprerequisite for a molecular understanding ofthe regulatory logic, and tracking a multitudeof reactions via a computational model is aneffective and practical strategy.

Understanding the function of a networkmust begin with comprehensive accountingof the parts list, the list of molecular com-ponents. The NF-kB signaling system consistsof two protein families, the NF-kBs (activators)and the IkBs (inhibitors) (Fig. 1). The NF-kBtranscription factors are the result of combina-torial dimerization of five monomers that can

Editors: Louis M. Staudt and Michael Karin

Additional Perspectives on NF-kB available at www.cshperspectives.org

Copyright # 2010 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/cshperspect.a000216

Cite this article as Cold Spring Harb Perspect Biol 2010;2:a000216

1

Spring Harbor Laboratory Press at UCLA BIOMED LIBRARY/SERIALS on July 26, 2017 - Published by Coldhttp://cshperspectives.cshlp.org/Downloaded from

http://cshperspectives.cshlp.org/

produce 15 possible dimers. IkB activities act asstoichiometric inhibitors of the DNA bindingactivities of these dimers, and all combinationsare, in principle, possible. Two classes of IkBkinase (IKK) complexes (the canonical andnoncanonical IKK complexes) control thehalf-life and therefore abundances of the IkBactivities. The canonical IKK are defined asthose complexes bound and regulated byNF-kB essential modulator (NEMO), and thenoncanonical complexes require both IKKaand NF-kB inducing kinase (NIK) activitiesbut are independent of NEMO (Scheidereit2006). The canonical IKK complexes act onthe classical IkB proteins IkBa, -b, -1, and theIkBg activity mediated by p105, whereas thenoncanonical IKK complex acts on the IkBdactivity mediated by p100. Both of the latter

nonclassical IkB activities (IkBg and IkBd)reside in a high-molecular weight complex(Savinova et al. 2009), which we term theIkBsome. In addition, both IKK activities canaffect NF-kB dimer generation by controllingthe processing of precursor proteins during orshortly after their translational synthesis.Canonical IKK may enhance processing ofp105 to p50, thereby increasing the availabilityof p50-containing dimers, whereas noncanoni-cal IKK controls processing of p100 to p52,thereby generating p52-containing dimers.

Considering the regulatory logic of the NF-kB signaling system, we distinguish regulatorymechanisms on two different timescales atwhich they operate: NF-kB dimer formationvia NF-kB protein synthesis and dimerization,and the regulation of NF-kB dimer activity via

IKK aIKK aIKK bIKK b

p105

RelA p50

p52

RelB

cRel

RelA

cRel

RelB

p52

p50

50 p100 52

IKKbKinases

IkBactivities

IkBsome

Precursorprocessing

NF-kBdimers

kB sites (G–5)G–4G–3R–2N–1N0N+1(N+2)(Y+3C+4C+5)

NEMO NEMO NEMO

a b e g d

IKKa IKKa IKKa

Figure 1. Molecular components of the IKK–IkB–NF-kB signaling system. The IkB kinases form canonicalNEMO-containing (green) complexes and noncanonical IKKa complexes (blue), which control thedegradation of IkB proteins as well as precursor processing. IkBa, IkBb, IkB1, p105 (IkBg), and p100(IkBd) are able to bind and sequester NF-kB dimers (“IkB activity”). The p50 and p52 NF-kB proteins areinitially synthesized as the precursor proteins p105 and p100, respectively. The five NF-kB family memberscan potentially form 15 possible dimers, which may bind to a large family of kB sites in DNA.

E. O’Dea and A. Hoffmann

2 Cite this article as Cold Spring Harb Perspect Biol 2010;2:a000216



IkB degradation and resynthesis. The former(NF-kB protein synthesis and dimerization) islargely associated with longer-term cell differ-entiation processes, whereas the latter (IkBdegradation and resynthesis) controls reversibleresponses to often transient inflammatorystimuli. Hence, we first consider the regulatorylogic of NF-kB dimer generation and thendiscuss the mechanisms that regulate dimeractivity.

REGULATION OF NF-kB DIMERGENERATION

NF-kB refers to homo- and heterodimeric DNAbinding complexes that consist of Rel homologydomain (RHD) containing polypeptides. Ofthe 15 potential dimers (Fig. 1), three do bindDNA but lack transcriptional activity (p50:p50,p52:p52, and p50:p52), and three are notknown to bind DNA (RelA:RelB, cRel:RelB,and RelB:RelB), leaving nine potential transcrip-tional activators (Hoffmann and Baltimore2006). Dimer formation and DNA bindingoccur through the RHD. Upon dimerization,NF-kB dimers can interact with IkB proteins,as the ankyrin repeat domain (ARD) of IkBforms a large interaction surface around thedimeric interface (Hayden and Ghosh 2004;Hoffmann et al. 2006). IkB binding biases cellu-lar localization of the NF-kB dimer to the cyto-plasm, thereby inhibiting the DNA bindingactivity of the NF-kB dimers and allowing forstimulus-responsive activation by cytoplasmicIKK complexes.

Several biochemical reactions control thegeneration and therefore availability of NF-kBdimers. Synthesis of the RHD polypeptidesis the first step, and is regulated at the levelof transcription in a cell-type-specific manner.Furthermore, RHD polypeptide genes (exceptrela) are inducible by NF-kB activity (seenf-kb.org for a list of NF-kB target genes),potentially resulting in positive feedback aswell as functional interdependence betweenthem. In addition, the generation of p50 andp52, the two RHD polypeptides that lack tran-scriptional activation domains (TADs) but actas binding partners for cRel, RelB, and RelA, is

controlled by proteolysis of p105 and p100.These proteolytic events may be responsive tocanonical and noncanonical signals.

RHD polypeptides require dimerization toavoid degradation, presumably via unfoldedprotein degradation pathways. NF-kB dimerformation and stability may not only be deter-mined by intrinsic association and dissociationrate constants between RHD polypeptides, butmay also involve interactions with IkB proteins.One intriguing possibility is that associationwith the IkBs stabilizes the NF-kB dimers byslowing their dissociation into monomers, andpossibly also the degradation of the intactdimers. If this hypothesis is found to be valid,IkB interactions with NF-kB dimers wouldnot only inhibit NF-kB function, but wouldalso contribute to generation of NF-kB dimersin the first place.

Equilibrium States

The generation of the cell-type-specific NF-kBdimer repertoire is a function of cell-type-specific homeostatic regulation that may besubject to signals conditioning the cell in aparticular microenvironment. Homeostaticregulation may be tuned at several differentmechanistic levels: (1) stimulus-responsive andcell-type-specific RHD polypeptide synthesis;(2) dimerization specificities governing theassociation and dissociation of each of the 15monomer pairs; (3) IkB-NF-kB interactionscontributing to dimer stability (Fig. 2A,B). AsIkBs may have differential affinities for NF-kBdimers, the relative abundances of various IkBproteins may affect the repertoire of latentdimers available for activation.

Different cell types have indeed beenreported to contain different repertoires ofNF-kB dimers. Murine embryonic fibroblasts(MEFs), just like many of the transformedcell lines commonly used (HeLa, HEK293),contain primarily latent RelA:p50 heterodimer,as well as RelA:RelA and p50:p50 homodimers.cRel and RelB expression can be detected inthese cells, but appears functionally negligible.In contrast, B-cells are abundant with readilyactivatable cRel and p50 containing dimers

The Regulatory Logic of the NF-kB Signaling System

Cite this article as Cold Spring Harb Perspect Biol 2010;2:a000216 3



(Sen 2006). Dendritic cells (DCs) depend onboth cRel and RelB for maturation (Ouaazet al. 2002; Gerondakis et al. 2006), althoughthe regulation of these cRel and RelB-containingdimer activities remain to be characterized.

Although cell-type-specific expression ofRHD polypeptides and NF-kB dimers hasbeen established, the molecular regulation thatunderlies the shifts in the NF-kB dimer reper-toire during cell differentiation remains to bedelineated. An early study examined the ex-pression of NF-kB monomers and dimers intransformed cell lines that represent B-cell sub-types along the B-cell differentiation pathway(Liou et al. 1994). This work tracked differentialNF-kB dimer availability, but what molecularmechanisms mediate these changes in thehomeostatic NF-kB dimer repertoire hasremained unclear. First insights about thesemolecular mechanisms come from the detailedanalysis of NF-kB dimer misregulation inknockout cells deficient in one or more NF-kB orIkB genes (Basak et al. 2008). The mechanisms

by which compensation or interdependence ofNF-kB dimer regulation is mediated can there-by be characterized and the insights may beapplicable to normal differentiation processes.

There are several levels in which the gener-ation of different NF-kB dimers is interdepen-dent, allowing for cross talk and/or robustnessthrough compensation. Cross talk may be medi-ated by the fact that synthesis of the four RHDpolypeptides RelB, cRel, p50, and p52 is depen-dent on RelA-containing dimers at the level ofgene expression (Fig. 2A). Hence, phenotypesresulting from RelA deficiency may (in part)be mediated by the consequently diminishedavailability of other RelA-dependent NF-kBdimers. Second, potential competition in dimer-ization interactions between limited pools ofNF-kB monomers may mediate cross-regulationin dimer generation when specific RHDpolypeptide synthesis rates change. However,a primary binding partner for the TAD-containing RelA, RelB, and cRel proteins isp50, which is generally produced in excess by

cRel:cRel

A BIkBsome

syn1p105

c

A B

p100

50

52

NF kB

IkB

syn2

deg1f

f

f

f

f

deg2

deg3

deg4IkBsome

p50:p50

RelB:p50

RelB:p52

cRel:p50

cRel:RelA

RelA:p50

RelA:RelA

RelA:p52

Figure 2. Mechanisms determining NF-kB dimer generation. (A) Synthesis of RHD polypeptides and theirdimerization affinities control the generation of NF-kB dimers, whose relative abundances in MEFs areindicated by their relative size. Green arrows indicate regulation by canonical IKK and the blue arrowregulation by noncanonical IKK. The amount of processing of the p105 and p100 proteins alters the amountof p50 and p52 available for dimer interaction, and abundances of certain subunits is influenced by theamount of others. (B) IkB-NF-kB interactions may play a role in dimer generation. A theoretical model ofNF-kB dimer metabolism indicates that dimerizaton may reduce monomer degradation (if deg3,deg2 and/or deg1), and further, that IkB interactions with the dimer may not only block dimer dissociation intomonomers, but also dimer degradation (if deg4,deg3).





abundant synthesis and processing of p105,leading to system robustness. When p50 islacking in nfkb12/2, then the mature nfkb2gene product p52 compensates. In this scenario,however, the demand for p52 depletes the poolof p100, which is required for the noncanonicalIkBd activity, thereby attenuating noncanonicalsignal responsiveness. This demonstrates thecapacity for one means of cross talk regulationbetween the canonical and noncanonical path-ways (Basak et al. 2008).

One motivation for understanding the regu-lation of the cell-type-specific NF-kB dimerrepertoire is that disease-associated cells maywell exhibit an altered equilibrium state. Suchaltered states may be caused by chronic exposureto external signals within the disease microen-vironment. Inflammatory signals such as TNF,IL-1, IL-6, or IFNg are associated with inflam-matory conditions or tumors that can skew theNF-kB dimer repertoire and NF-kB responsive-ness of macrophages. Such changes may miti-gate or potentiate the pathology. In otherpathological conditions, altered NF-kB dimerrepertoires may be a cause of the pathology. Insome subsets of B-cell lymphomas, an enhancedpro-proliferative NF-kB-cRel gene expressionsignature can be traced back to cRel gene ampli-fications (Shaffer et al. 2002). Enhanced cRelexpression may in other subsets be the result ofcell intrinsic alterations that are not immediatelyobvious, such as an altered chromatin homeo-stasis caused by misregulated HAT (histoneacetyltransferase) and HDAC (histone deacetyl-ase) activities. The effects of disease-associatedaltered NF-kB dimer repertoires continue topose important questions for research.

Stimulus-responsive Alterations

As four RHD polypeptides are encoded byknown NF-kB-RelA response genes, the NF-kB dimer repertoire is generally thought to bereadily alterable in response to inflammatorystimulation or other RelA-dimer-inducingstimuli (Fig. 3). However, it is unclear howdynamic or reversible the resulting changesare, whether they are occurring within hoursor days, or whether they are transient or long

lasting. Most observations suggest that changesin the NF-kB dimer repertoire because ofstimulus-induced transcription are occurringon long timescales and are long-lasting; wetherefore suggest that transcriptional control ofRHD polypeptide expression determines thehomeostatic state of the NF-kB signaling system.

Stimulus-responsive alterations in the NF-kB dimer pool do, however, occur as a resultof stimulus-responsive processing of p105 andp100 to p50 and p52. Both processing eventsoccur cotranslationally or at least shortly follow-ing translation, before the precursors formhigher order complexes (including IkBsomes)through homo- and heterotypic interactionsthat render them processing incompetent.Processing of p105 to p50 is mediated by long-term constitutive canonical IKK activity thatensures elevated expression of p50. The result-ing p50:p50 homodimers have been observedin TLR-stimulated macrophages as well asother cell types, but their physiological role isunclear. Processing of p100 to p52 is mediatedby noncanonical IKK activity and has beenstudied in detail. This processing event gener-ates primarily the RelB:p52 dimer, which regu-lates expression of organogenic chemokines andis implicated in survival of certain differentiat-ing and maturing cell types. Thus, noncanonicalpathway activating stimuli, such as BAFF, LTb,RANKL, and CD40L, can be thought of as alter-ing the NF-kB dimer repertoire to mediate theirfunctional effects (in addition to regulating theactivities of NF-kB dimers via IkB regulation, asdiscussed later).

REGULATION OF NF-kB ACTIVITY VIA IkBDEGRADATION AND RESYNTHESIS

The three canonical IkB proteins, IkBa, IkBb,and IkB1, and the noncanonical IkBd activitybind and inhibit NF-kB dimers and therebyallow for their stimulus-responsive activation.The degradation and resynthesis of IkBs iscoordinately regulated to produce dynamicNF-kB activities that are stimulus-specific.For example, the ubiquitous RelA:p50 dimeris bound and inhibited by the canonical IkBproteins (IkBa, IkBb, and IkB1), as well as the





IkBd activity. RelB-containing dimers, how-ever, are known to have strong binding speci-ficity only for IkBd. In this manner, differentIkB-NF-kB dimer associations allow for specificdimers to be activated in response to specificstimuli.

The regulated metabolism (synthesis anddegradation) of the canonical IkB proteins,especially IkBa, has been studied in detail.Two degradation pathways control steady-statelevels of canonical IkBs and NF-kB activity.

When associated with NF-kB, IkBa, -b, and-1 are degraded through a canonical IKK-dependent mechanism. IkB proteins that arenot associated with NF-kB, however, are intrin-sically unstable with a 5–10 minute half-life,being degraded in an IKK- and ubiquitin-independent manner (O’Dea et al. 2007;Mathes et al. 2008). Accumulating evidenceindicates that the 20S proteasome can mediatedegradation of free IkBa, whereby weaklyfolded regions in carboxy-terminal ankyrin

Canonical

NF-kB IkBsomeIkB

NF-kB

Cytoplasm

Nucleus

Target genes

p-elF2a p50cRel p105 p100

RelB

a b e g

d

Stress

Noncanonical

Figure 3. Mechanisms controlling stimulus-responsive NF-kB activities. Canonical signals activateNEMO-containing IKK complexes (green), which degrade the canonical IkB proteins (IkBa, b, and 1) andthe IkBg activity (composed of asymmetric p105 dimers) associated with NF-kB dimers. Released NF-kBdimers move to the nucleus to activate gene expression programs, including the expression of IkBa, IkB1,p105, p100, cRel, and RelB proteins. Noncanonical signals activate IKKa complexes, which degrade IkBdcomplexes associated with NF-kB dimers. The resulting increase in synthesis of p100 and RelB, concomitantwith IKKa activity, causes increased p100 processing to p52 and dimerization with RelB, to generate activeRelB:p52 dimers to the nucleus. Stress signals can activate the eIF2a kinases, causing phosphorylation ofeIF2a and resulting in inhibited translation. A block in IkB synthesis, in combination with constitutive IKKactivity, results in the loss of IkB proteins and subsequent NF-kB dimer activation.





repeats and other carboxy-terminal sequencesmediate its instability (Alvarez-Castelao andCastano 2005; Truhlar et al. 2006). NF-kB asso-ciation triggers complete IkB protein foldingand removes IkB from the 20S proteasomepathway. In addition to determining the IkBdegradation pathway, NF-kB activity also drivesthe transcription of IkBa and IkB1, thusproviding another self-inhibitory mechanismto prevent constitutive NF-kB activity.

IkBd activity is the result of the dimerizationof two p100 molecules (through interactionbetween their RHDs), whereby the ARD of onep100 molecule folds back onto the RHD–RHD dimer interface to effect self-inhibition(“in cis”), leaving the ARD of the secondp100 molecule capable of binding latent NF-kB dimers (primarily RelA:p50, but alsoRelB:p50) “in trans” (Basak and Hoffmann2008). Subsequent studies with processing-defective mutants of p105 provided evidencefor an analogous activity, termed IkBg (Sris-kantharajah et al. 2009). NF-kB also drives thesynthesis of p105/IkBg and p100/IkBd, whichmay, like IkBa and IkB1, contribute to tightcontrol over constitutive NF-kB activity.Recent studies revealed that these nonclassicalIkB activities, IkBg and IkBd, reside withinhigh molecular weight complexes, whose ass-embly is mediated by the helical dimerizationdomains present within p100 and p105 inaddition to RHD dimerization and ARD inter-action surfaces (Savinova et al. 2009). Thesecomplexes may be termed IkBsomes as theytrap a variety of NF-kB dimers for slow homeo-static or stimulus-responsive release at latertimes.

IkB Signaling in Response toInflammatory Signals

Endogenous inflammatory stimuli (e.g., cyto-kines-TNF, IL-1b) or pathogen-derived sub-stances (e.g., LPS or CpG DNA) activate thecanonical NF-kB pathway. Engagement ofthe TNF receptor (TNFR), IL-1b receptor(IL-1bR), or TLRs causes phosphorylation-dependent activation of a NEMO-containingkinase complex. Once activated, the canonical

IKK complex phosphorylates IkBa, -b, and -1at two specific serine residues, as a signal forK48-linked ubiquitination at two specific lysineresidues, leading to degradation by the 26Sproteasome (Karin and Ben-Neriah 2000). Deg-radation of IkB releases NF-kB, allowing it tolocalize to the nucleus to bind DNA and acti-vate gene expression. Interestingly, IkBa andIkB1 proteins are among the large number ofNF-kB response genes, thus functioning as nega-tive-feedback regulators. Indeed, NF-kB activityinduced by inflammatory stimuli shows com-plex and diverse temporal or dynamic profiles.

A mathematical model of the IkB–NF-kBsignaling module recapitulates the signalingevents triggered by TNF stimulation observedexperimentally in MEFs (Kearns and Hoff-mann 2009; Cheong et al. 2008). Combined ex-perimental and computational studies showedthat the three IkB proteins IkBa, -b, and -1each have distinct roles in the dynamic controlof NF-kB activation and termination. Inresponse to TNF, IkBa is rapidly degradedand then rapidly resynthesized in an NF-kB-dependent manner. Continued TNF stimu-lation propagates a cycle of synthesis and degra-dation of IkBa that can result in oscillationsof nuclear NF-kB activity (Hoffmann et al.2002). IkB1 expression is also strongly inducedby NF-kB activity, but with a distinct 40-minutedelay (Kearns et al. 2006). As IkB1 protein ac-cumulates at later time points, this antiphasenegative-feedback loop acts to dampen theIkBa-driven oscillations.

Recent evidence indicates that the non-canonical IkB, IkBd provides negative feedbackon canonical NF-kB signaling when it is in-duced by pathogen-triggered TLR signals thatproduce longer lasting canonical IKK activitiesthan inflammatory cytokines (e.g., TNF orIL-1). The induction of NF-kB by either LPSor TNF drives increased synthesis of p100, buton a delayed and longer timescale than thatof IkBa. The slowly accumulating p100 canform into IkBsomes, within which IkBd trapsNF-kB dimers (Shih et al. 2009). WhereasTNF-induced signaling wanes within 8 hours,LPS-induced NF-kB remains elevated at latetime points. This late NF-kB activity becomes





subject to attenuation by IkBd, which is un-responsive to and cannot be degraded by cano-nical signals. Cells deficient in IkBd activity(nfkb2 gene knockout), show similar NF-kBactivation profiles in responses to TNF, butshow significantly higher LPS-induced lateNF-kB activity as compared to wild-type(IkBd containing) cells (Shih et al. 2009). Thisstudy also emphasized that the prominentIkBa negative feedback is actually specific fortransient cytokine signals. Because its inhi-bition is reversible by canonical IKK activity, itplays little role in shaping NF-kB temporalprofiles in response to long lasting TLR-initiated signals. In fact, perinatal lethality ofthe ikba2/2 mouse was rescued by compounddeficiency with the tnf gene (Shih et al. 2009).

The ability of the IkB-NF-kB signalingmodule to mediate complex temporal controlover NF-kB activity has led to research intothe functionality of NF-kB dynamics. Differentinflammatory stimuli elicit different IKK acti-vation profiles, which induce distinct temporalprofiles of NF-kB activity. For example, TNFstimulation induces a transient spike in IKKactivity in which the second phase is rathersmall because of the short TNF half-life andthe inhibitory functions of the deubiquitinaseA20 (Werner et al. 2008). In contrast, LPS acti-vates gene expression of cytokines that providefor positive autocrine feedback, amplifyinglate IKK activity (Werner et al. 2005). Thesestudies have demonstrated that temporal cont-rol of NF-kB activity is important for stimulus-specific gene expression programs, yet it remainsunclear how gene promoters decipher temporalprofiles. If the temporal profile of NF-kBactivity conveys information about the stim-ulus and therefore represents a code (i.e.,“temporal code”), it remains unclear how thisinformation is decoded. Also, given that tem-poral control depends on kinetic rate constants,it is likely that different cells encode stimulusinformation differently. Further experimentalinvestigation and expansion of the currentNF-kB mathematical model to include othercell types will be necessary to decipher howNF-kB dynamics plays a role in mammalianphysiology.

IkB Signaling in Response toDevelopmental Signals

A group of noninflammatory stimuli havebeen shown to activate NF-kB through thenoncanonical NF-kB signaling pathway. Thesedevelopmental signals of the TNF-receptorsuper-family such as B-cell activation factorreceptor (BAFFR), lymphotoxin b receptor(LTbR), and receptor activator of NF-kB(RANK), have been described to activate NF-kB activity at a low level for hours or days. Thenoncanonical pathway is not transduced by acanonical NEMO-containing kinase complex,but rather by an IKKa-containing complex,whose activity is also dependent on NIK. Inaddition to the noncanonical NIK/IKKa-dependent NF-kB activation, these stimuli mayalso, in certain cellular conditions and contexts,activate the canonical NEMO/IKKb-dependentNF-kB activation pathway (Pomerantz andBaltimore 2002).

Initial studies of NF-kB activation by thesedevelopmental stimuli focused on the gener-ation of the RelB:p52 dimer by cotranslationalproteolytic processing of de novo synthesizedp100 to p52. The p100 protein contains car-boxy-terminal serines whose phosphorylationby IKKa is critical for stimulus-responsive pro-cessing. More recently, it was shown that thesame pathway also activates latent RelA:p50NF-kB complexes not through NEMO-depen-dent degradation of classical IkB, but rathervia the degradation of IkBd activity.

IkBd activity and the IkBsomes, whichmediate it, were first discovered in cells lackingall three classical IkB proteins, IkBa, -b, and -1(Tergaonkar et al. 2005; Basak et al. 2007). Inthese cells, the ubiquitous RelA:p50 dimer wasshown to be associated with cytoplasmic p100proteins. Further biochemical analysis showedthat all cells contain significant amounts ofNF-kB associated with high molecular weightcomplexes of p100 and p105 that mediate IkBgand IkBd activities (Savinova et al. 2009). Thefunctional importance of these complexes wasrevealed by the discovery that noncanonicalsignals, through NIK- and IKKa-dependent deg-radation of IkBd, release associated RelA:p50





and RelB:p50 to the nucleus (Basak et al. 2007).Furthermore, conditions or specific stimuli thatcontrol the abundance of IkBd are therefore ableto tune the responsiveness of RelA:p50 acti-vation through noncanonical signals. WhenIkBd is highly abundant, usually weak develop-mental signals that engage the noncanonicalpathway are able to activate inflammatorygenes via strong RelA:p50 activation.

IkB Signaling in Response toCellular Stresses

Recent work has uncovered homeostatic mech-anisms of regulation of NF-kB in resting cells(i.e., in the absence of a stimulus) that predeter-mine the responsiveness of the NF-kB system toinducers, including stress stimuli.

Homeostatic regulation of IkB synthesisand degradation has emerged as an importantcharacteristic that renders the NF-kB signalingsystem surprisingly insensitive to a variety ofperturbations. First, the differential degra-dation rates of free and bound IkB allowsfor compensation between IkBs, evident bystudies in IkB knockout cells (O’Dea et al.2007). The loss of IkBa, for example, causesexcess NF-kB to bind and thus stabilize IkB1,in addition to enhancing its synthesis. Second,the very short half-life of free IkB necessitatesa high rate of constitutive IkB synthesis to main-tain the small cellular free IkB pool (estimatedat 15% of total [Rice and Ernst 1993]) that iscritical for keeping basal NF-kB activity levelslow. One consequence of the very high constitu-tive IkB synthesis/degradation flux is that theNF-kB system is remarkably resistant to tran-sient alterations in translation rates that are ahallmark of metabolic stress agents. Indeed,UV, UPR (unfolded protein response), or otherribotoxic stress that cause partial inhibition ofIkB synthesis rates were found to activate NF-kB only modestly (O’Dea et al. 2008).

Understanding homeostatic control of theIkB-NF-kB system also resolved some see-mingly conflicting observations with regardsto NF-kB activation by UV. Although UV doesnot induce IKK activity, basal IKK activitywas found to be required (Huang et al. 2002;

O’Dea et al. 2008). The systems model revealedhow basal IKK activity is critical for slowlydegrading NF-kB-bound IkBa, allowing foraccumulation of nuclear NF-kB in response totranslational inhibition. Whereas stimuli thatinduce canonical IKK activity cause the deg-radation of NF-kB-bound IkB proteins, UVirradiation was shown to cause the depletionof the free IkB pool. Indeed, mutations thatstabilize free IkB (but allow for IKK-dependentdegradation of NF-kB-bound IkB) abolishedactivation of NF-kB by UV (O’Dea et al.2008). UV-induced CK2 activity and phos-phorylation of the carboxyl terminus of IkBamay further accelerate IkBa degradation butthe molecular mechanism of this pathwayremain unclear (Kato et al. 2003).

DNA damage, caused by irradiation orchemotherapeutic drugs, however, has beenreported to induce IKK activation. Untilrecently, it was unclear how a nuclear signalcould relay back to the cytoplasm. It wasfound that DNA damage not only initiates theactivation of the nuclear kinase ataxia telangiec-tasia mutated (ATM), the primary regulator ofthe tumor suppressor and transcription factorp53, but it also initiates the sumoylation ofNEMO by the sumo ligase PIASy, promotingits nuclear localization (Mabb et al. 2006). Itwas known that activated ATM was requiredfor NF-kB activation by DNA damage, andrecent work has uncovered the connectionbetween ATM activity and IKK activation.Wu et al. (2006) showed that nuclear sumoy-lated NEMO associates with and is phos-phorylated by the activated ATM, promotingmono-ubiquitination of NEMO that triggersits export to the cytoplasm. The ATM-NEMOcomplex associates with the protein ELKS,facilitating ATM-dependent activation of thecanonical IKK complex, leading to IkBa degra-dation and NF-kB activation (Wu et al. 2006).

THE NF-kB SIGNALING SYSTEM AS ANINTEGRATOR OF SIGNALS

The NF-kB signaling system mediates the phys-iological effects of a variety of signals. It doesso via a large number of biochemical reactions





determining the abundance and kinetic regu-lation of two families of proteins (NF-kB andIkB) and their interactions with each other.Many NF-kB-inducing stimuli control theactivity of the two IKK complexes, but othersimpact many other reactions to alter NF-kBactivity, or modulate the responsiveness of theNF-kB signaling system. When stimuli affectdifferent reaction rates within the same signal-ing system, there is potential for signaling crosstalk. As such, the NF-kB signaling system hasgreat potential for integrating diverse signals.In the following section, we summarize just afew of many possible examples.

Integrating Signals that Control IkBDegradation and Synthesis

Ribotoxic Stresses and Inflammatory SignalsCross Talk via IkBa

Whereas ribotoxic stress-inducing stimuli suchas UV irradiation or the unfolded proteinresponse (UPR) that act to inhibit the trans-lation of IkBa can rapidly deplete the cellularpool of free, intrinsically unstable IkBa, NF-kB-bound IkBa remains associated with NF-kBuntil either constitutive or induced canonicalIKK-mediated phosphorylation triggers itsdegradation. The concept of this differentialdegradation of IkBa, combined with compu-tational simulations suggesting that the levelof constitutive IKK activity predetermines theresponsiveness of NF-kB activity to transla-tional inhibition, led to the prediction thatcells chronically exposed to low levels of inflam-matory signals may have significantly enhancedresponses to stress stimuli that inhibit proteinsynthesis (O’Dea et al. 2008).

Indeed, it was found that cells treated withlow doses of inflammatory signals that onlymarginally enhance IKK and NF-kB activities,synergize with UV or ER stress agents thatinhibit protein synthesis (via eIF2a phosphor-ylation) to produce a super-additive increasein NF-kB activity and NF-kB-dependent in-flammatory gene expression (O’Dea et al.2008). The combined effect of the metabolicstress reaction (eIF2a phosphorylation) and

IKK activation in response to inflammatorysignals or immune receptors may be a morecommon mechanism to activating NF-kB thanpreviously recognized. A potential function ofsuper-activating NF-kB during metabolic stres-ses may be to override the concomitant decreasein translation rates for NF-kB controlled survi-val and immune response genes.

Inflammatory Signals and DevelopmentalSignals Cross Talk via IkBd

Inflammatory signaling can interact and coor-dinate responses with developmental cues. Forexample, lymph node development and homeo-stasis appears to require not only the develop-mental regulator lymphotoxin b (LTb) butalso the inflammatory cytokine TNF (Rennertet al. 1998). Conversely, inflammation canderail normal developmental or cellular homeo-stasis and be a major factor in cancer progression(Karin and Greten 2005). Over the last few years,it has become apparent that the NF-kB signalingsystem may be a primary integrator of thesediverse signals to mediate cross talk in bothphysiological and pathological processes.

Following the identification of the fourthIkBd activity that mediates noncanonicalsignals, its NF-kB-induced expression bycanonical signals was explored as a cross talkmechanism (Basak et al. 2007). Inflammatorystimuli drive the expression of p100, whichmediates inhibition of RelA:p50 complexesthrough its IkBd activity. Because IkBd isnot degraded by canonical NEMO-mediatedsignals, the cellular pool of RelA:p50 bound toIkBd increases, thereby amplifying RelA:p50responsiveness to subsequent developmentalsignals (which degrade IkBd) to the extentthat developmental stimuli may cause inflam-matory gene expression. Combined compu-tational modeling and experimental studiesdemonstrated the potential for developmentalsignals such as LTb to result in inflammatoryresponses. This cross talk mediated by IkBdhas been speculated to play a role in regulatingcancer-associated inflammation (Basak andHoffmann 2008).





Integrating Signals that Control NF-kBSynthesis and IkB Degradation

Several observations suggest that there may besynergistic cross talk between canonical andnoncanonical IKK pathways via the NF-kBsignaling system. RelA-containing dimers areknown to increase the synthesis of RelBand p100/p52 polypeptides (www.nf-kb.org).Indeed, basal RelA NF-kB activity was shownto be required for LTb-responsive RelB:p52dimer generation (Basak et al. 2008). In B-cells, basal or tonic canonical IKK activity wasshown to control the generation of the p100substrate for noncanonical BAFFR signals(Stadanlick et al. 2008). By the same rationaleand molecular mechanism, inflammatorysignals such as TNF and LPS may potentiallyenhance RelB:p52 generation in the presence ofdevelopmental stimuli. However, at this time,no compelling physiological or molecular evi-dence has been presented, and the functionalsignificance of the RelB versus RelA dependentgene expression remains to be elucidated.

CONCLUDING REMARKS

The regulatory logic of the NF-kB signalingsystem suggests regulation at two levels. Thefirst pertains to the generation of the tran-scriptional activators, the NF-kB dimers, viamonomer expression and dimerization reac-tions. The second is the regulation of the me-tabolism of the IkB inhibitors via synthesisand degradation control. Inflammatory sig-nals largely involve the latter, and producecomplex dynamic control of NF-kB activity.Homeostatic and cell differentiation associatedmechanisms control NF-kB dimer generation.Interestingly, signals engaging the noncanon-ical IKK complex engage both mechanisticlevels; such developmental signals are able toproduce inflammatory responses (though gener-ally weakly) and they affect the available NF-kBdimer repertoire. Recognizing the two levelsalso suggests that interesting, nonadditive ornonlinear signaling may result when the twointerface. With multiple interdependenciesapparent within the NF-kB signaling system,

these cross-regulatory connections must be fur-ther characterized mechanistically and quan-tified to allow for a better understanding ofphysiological regulation and pathological mis-regulation of the NF-kB signaling system. Bio-physical descriptions at the molecular andcellular level are likely to result in further insightsabout a signaling system that has tremendoushuman health relevance.

ACKNOWLEDGMENTS

We thank the many researchers whose workinformed the conceptual framework describedin this article and apologize if it was not explic-itly described or cited due to space restrictions.Research on NF-kB in our laboratory wasfunded by National Institutes of Health (NIH)grants GM071573 and GM071862. We thank S.Basak and B. Schroefelbauer for frequent discus-sions and critical reading.

REFERENCES

Alvarez-Castelao B, Castano JG. 2005. Mechanism of directdegradation of IkBa by 20S proteasome. FEBS Lett 579:4797–4802.

Basak S, Hoffmann A. 2008. Crosstalk via the NF-kB signal-ing system. Cytokine Growth Factor Rev 19: 187–197.

Basak S, Shih VF, Hoffmann A. 2008. Generation andactivation of multiple dimeric transcription factorswithin the NF-kB signaling system. Mol Cell Biol 28:3139–3150.

Basak S, Kim H, Kearns JD, Tergaonkar V, O’Dea E, WernerSL, Benedict CA, Ware CF, Ghosh G, Verma IM, et al.2007. A fourth IkB protein within the NF-kB signalingmodule. Cell 128: 369–381.

Cheong R, Hoffmann A, Levchenko A. 2008. UnderstandingNF-kB signaling via mathematical modeling. Mol SystBiol 4: 192.

Gerondakis S, Grumont R, Gugasyan R, Wong L, Isomura I,Ho W, Banerjee A. 2006. Unravelling the complexities ofthe NF-kB signalling pathway using mouse knockout andtransgenic models. Oncogene 25: 6781–6799.

Hayden MS, Ghosh S. 2004. Signaling to NF-kB. Genes Dev18: 2195–2224.

Hoffmann A, Baltimore D. 2006. Circuitry of nuclear factorkB signaling. Immunol Rev 210: 171–186.

Hoffmann A, Natoli G, Ghosh G. 2006. Transcriptionalregulation via the NF-kB signaling module. Oncogene25: 6706–6716.

Hoffmann A, Levchenko A, Scott ML, Baltimore D. 2002.The IkB-NF-kB signaling module: Temporal controland selective gene activation. Science 298: 1241–1245.





Huang TT, Feinberg SL, Suryanarayanan S, Miyamoto S.2002. The zinc finger domain of NEMO is selectivelyrequired for NF-kB activation by UV radiation andtopoisomerase inhibitors. Mol Cell Biol 22: 5813–5825.

Karin M, Greten FR. 2005. NF-kB: Linking inflammationand immunity to cancer development and progression.Nat Rev Immunol 5: 749–759.

Karin M, Ben-Neriah Y. 2000. Phosphorylation meets ubi-quitination: The control of NF-kB activity. Annu RevImmunol 18: 621–663.

Kato T Jr, Delhase M, Hoffmann A, Karin M. 2003. CK2 Isa C-Terminal IkB Kinase Responsible for NF-kB Activa-tion during the UV Response. Mol Cell 12: 829–839.

Kearns JD, Hoffmann A. 2009. Integrating computationaland biochemical studies to explore mechanisms inNF-kB signaling. J Biol Chem 284: 5439–5443.

Kearns JD, Basak S, Werner SL, Huang CS, Hoffmann A.2006. IkB1 provides negative feedback to controlNF-kB oscillations, signaling dynamics, and inflamma-tory gene expression. J Cell Biol 173: 659–664.

Liou HC, Sha WC, Scott ML, Baltimore D. 1994. Sequentialinduction of NF-k B/Rel family proteins during B-cellterminal differentiation. Mol Cell Biol 14: 5349–5359.

Mabb AM, Wuerzberger-Davis SM, Miyamoto S. 2006.PIASy mediates NEMO sumoylation and NF-kB acti-vation in response to genotoxic stress. Nat Cell Biol 8:986–993.

Mathes E, O’Dea EL, Hoffmann A, Ghosh G. 2008. NF-kBdictates the degradation pathway of IkBa. Embo J 27:1357–1367.

O’Dea EL, Kearns JD, Hoffmann A. 2008. UVas an amplifierrather than inducer of NF-kB activity. Mol Cell 30:632–641.

O’Dea EL, Barken D, Peralta RQ, Tran KT, Werner SL,Kearns JD, Levchenko A, Hoffmann A. 2007. A homeo-static model of IkB metabolism to control constitutiveNF-kB activity. Mol Syst Biol 3: 111.

Ouaaz F, Arron J, Zheng Y, Choi Y, Beg AA. 2002. Dendriticcell development and survival require distinct NF-kBsubunits. Immunity 16: 257–270.

Pomerantz JL, Baltimore D. 2002. Two pathways to NF-kB.Mol Cell 10: 693–695.

Rennert PD, James D, Mackay F, Browning JL, Hochman PS.1998. Lymph node genesis is induced by signalingthrough the lymphotoxin b receptor. Immunity 9:71–79.

Rice NR, Ernst MK. 1993. In vivo control of NF-kB acti-vation by I k B a. Embo J 12: 4685–4695.

Savinova OV, Hoffmann A, Ghosh G. 2009. The Nfkb1 andNfkb2 proteins p105 and p100 function as the core of het-erogeneous NF-kBsomes. Mol Cell 34: 591–602.

Scheidereit C. 2006. IkB kinase complexes: gateways to NF-kB activation and transcription. Oncogene 25: 6685–6705.

Sen R. 2006. Control of B lymphocyte apoptosis by the tran-scription factor NF-kB. Immunity 25: 871–883.

Shaffer AL, Rosenwald A, Staudt LM. 2002. Lymphoidmalignancies: The dark side of B-cell differentiation.Nat Rev Immunol 2: 920–932.

Shih VF, Kearns JD, Basak S, Savinova OV, Ghosh G,Hoffmann A. 2009. Kinetic control of negative feedbackregulators of NF-kB/RelA determines their pathogen-and cytokine-receptor signaling specificity. Proc NatlAcad Sci 106: 9619–9624.

Sriskantharajah S, Belich MP, Papoutsopoulou S, Janzen J,Tybulewicz V, Seddon B, Ley SC. 2009. Proteolysis ofNF-kB1 p105 is essential for T cell antigen receptor-induced proliferation. Nat Immunol 10: 38–47.

Stadanlick JE, Kaileh M, Karnell FG, Scholz JL, Miller JP,Quinn WJ 3rd, Brezski RJ, Treml LS, Jordan KA,Monroe JG, et al. 2008. Tonic B cell antigen receptorsignals supply an NF-kB substrate for prosurvival BLySsignaling. Nat Immunol 9: 1379–1387.

Tergaonkar V, Correa RG, Ikawa M, Verma IM. 2005.Distinct roles of IkB proteins in regulating constitutiveNF-kB activity. Nat Cell Biol 7: 921–923.

Truhlar SM, Torpey JW, Komives EA. 2006. Regions of IkBathat are critical for its inhibition of NF-kB. DNA inter-action fold upon binding to NF-kB. Proc Natl Acad Sci103: 18951–18956.

Werner SL, Barken D, Hoffmann A. 2005. Stimulus speci-ficity of gene expression programs determined by tem-poral control of IKK activity. Science 309: 1857–1861.

Werner SL, Kearns JD, Zadorozhnaya V, Lynch C, O’Dea E,Boldin MP, Ma A, Baltimore D, Hoffmann A. 2008.Encoding NF-kB temporal control in response to TNF:Distinct roles for the negative regulators IkBa and A20.Genes Dev 22: 2093–2101.

Wu ZH, Shi Y, Tibbetts RS, Miyamoto S. 2006. Molecularlinkage between the kinase ATM and NF-kB signalingin response to genotoxic stimuli. Science 311: 1141–1146.





October 7, 20092010; doi: 10.1101/cshperspect.a000216 originally published onlineCold Spring Harb Perspect Biol

Ellen O'Dea and Alexander Hoffmann

B Signaling SystemκThe Regulatory Logic of the NF-

Subject Collection NF-kB

BκRegulators of NF-Use of Forward Genetics to Discover Novel

Tao Lu and George R. Stark

BκOncogenic Activation of NF-Louis M. Staudt

B ResponseκSelectivity of the NF-Ranjan Sen and Stephen T. Smale System

B SignalingκThe Regulatory Logic of the NF-

Ellen O'Dea and Alexander HoffmannB in the Nervous SystemκNF-

Barbara Kaltschmidt and Christian Kaltschmidt Development and FunctionB Pathway in LymphocyteκRoles of the NF-

Steve Gerondakis and Ulrich SiebenlistB: Regulation by UbiquitinationκSignaling to NF-

Ingrid E. Wertz and Vishva M. Dixit ActivationBκThe IKK Complex, a Central Regulator of NF-

Alain Israël

BκNF-Ubiquitination and Degradation of the Inhibitors of

Schueler-Furman, et al.Naama Kanarek, Nir London, Ora

B in the Nervous SystemκNF-Barbara Kaltschmidt and Christian Kaltschmidt

Signaling ModuleBκA Structural Guide to Proteins of the NF-

Tom Huxford and Gourisankar GhoshInflammation

B Pathway inκThe Nuclear Factor NF-

Toby LawrenceDrosophilaB in the Immune Response of κNF-

Charles Hetru and Jules A. Hoffmann and CancerB as a Critical Link Between InflammationκNF-

Michael Karin

Responses by Chromatin OrganizationB-dependent TranscriptionalκControl of NF-

Gioacchino NatoliProteins

BκSpecification of DNA Binding Activity of NF-

Fengyi Wan and Michael J. Lenardo

http://cshperspectives.cshlp.org/cgi/collection/ For additional articles in this collection, see

Copyright © 2010 Cold Spring Harbor Laboratory Press; all rights reserved


http://cshperspectives.cshlp.org/cgi/collection/

http://cshperspectives.cshlp.org/cgi/collection/


the regulatory logic of the nf- b signaling system2006). the canonical ikk complexes act on the...

Documents