recent insights into the structure of toll-like receptors ... · recent insights into the structure...

Biochem. J. (2009) 422, 1–10 (Printed in Great Britain) doi:10.1042/BJ20090616 1

REVIEW ARTICLERecent insights into the structure of Toll-like receptors andpost-translational modifications of their associated signalling proteinsSusan CARPENTER1 and Luke A. J. O’NEILLSchool of Biochemistry and Immunology, Trinity College Dublin, Ireland

TLRs (Toll-like receptors) are essential modulators of the innateimmune response through their ability to respond to a diverserange of conserved structures within microbes. Recent advanceshave been made in our understanding of the initiation of TLRsignals as a result of the elucidation of crystal structures ofTLRs interacting with their ligands. Most notably the structureof TLR1/2 with triacylated lipopeptide and TLR4 in a complexwith LPS (lipopolysaccharide) and MD2 has been solved. Theseexplain the basis for TLR dimerization which initiates signalling.Modifications of TLRs and their receptor proximal signallingproteins have also been uncovered. Phosphorylation of adaptor

proteins and ubiquitination (both Lys48- and Lys63-linked) ofTLRs, IRAKs (interleukin-1 receptor-associated kinase), Pellinosand TRAF6 (tumour-necrosis-factor-receptor-associated factor6) have been described, which promote signalling and lead tosignal termination. A detailed molecular account of the initiationand termination of TLR signalling is presented.

Key words: innate immune system, phosphorylation, post-translational modification, signalling, Toll-like receptor (TLR),ubiquitination.

INTRODUCTION

TLRs (Toll-like receptors) play a critical role in the innateimmune system, acting as pathogen-recognition receptors againstconserved structures within microbes [1]. Ten TLRs have beenidentified in humans. They are type 1 transmembrane glyco-proteins characterized by the presence of an LRR (leucine-rich repeat domain) and a TIR [Toll/IL (interleukin)-1 receptor]domain [2]. LRRs are found in a diverse number of proteinsand are involved in ligand recognition and signal transduction[3]. TLR4 was the first TLR to be identified as an orthologue ofDrosophila Toll [4,5]. TLR4 recognizes LPS (lipopolysaccharide)from Gram-negative bacteria. TLR2 can form a complex withTLR1 or TLR6 and responds to lipopeptides from a widevariety of microbes. The TLR1–TLR2 dimer responds to triacyl-ated lipopeptides, whereas TLR2–TLR6 responds to diacylatedlipopeptides. TLR3 is responsible for the recognition of dsRNA(double-stranded RNA), TLR5 recognizes flagellin, single-stranded RNA is recognized by TLR7 and TLR8, and TLR9recognizes DNA which can be host- or pathogen-derived [6].The intracellular TIR domain portion consists of approx. 200amino acids. Three particular boxes are highly conserved amongfamily members: box 1 is considered the signature sequence ofthe family, whereas boxes 2 and 3 contain amino acids criticalfor signalling. TLRs are proposed to dimerize following ligandbinding, resulting in the recruitment of TIR domain-containingadaptor molecules to initiate downstream signalling throughinteractions within the TIR regions [7]. Four such signallingadaptors have been described: MyD88 (myeloid differentiationfactor 88), Mal (MyD88 adaptor-like protein), TRIF (TIR domain-

containing adaptor inducing interferon-β) and TRAM (TRIF-related adaptor protein). The fifth TIR adaptor SARM (sterile-αand HEAT/Armadillo motifs-containing protein) has been shownto inhibit TRIF [8]. Structures of TLR3, TLR2 and TLR4with their ligands have been elucidated recently and provide anunderstanding of ligand-induced activation of TLRs [9–12].

Tight control of the TLR pathway is essential to maintainhomoeostasis, since overactivation of TLRs has been linked tovarious infectious and inflammatory diseases. TLR engagementleads to the activation of the transcription factor NF-κB (nuclearfactor κB) which regulates the induction of pro-inflammatorycytokines such as TNF (tumour necrosis factor) α, IL-1 and IL-6.It can also activate members of the MAPK (mitogen-activatedprotein kinase) family including p38 and JNK (c-Jun N-terminalkinase) [6]. These kinases are involved in the transcription ofgenes and they also regulate mRNA stability. Tight regulationof these pathways results from post-translational modificationprocesses. Phosphorylation has long been studied as an essentialmodification process within TLRs. Ubiquitination is emergingas being equally important as phosphorylation for maintainingcontrol of the TLR signalling pathways. In the present review,we discuss recent advances in structural elucidation, and also inpost-translational modification of proteins in the TLR system.

LRRs AND THE TIR DOMAIN

The two core domains of the TLRs are the LRR modules (whichare involved in ligand binding) and the TIR domain which isrequired for signal transduction. Obtaining the crystal structuresof the LRRs for any of the TLRs has been long awaited and

Abbreviations used: Btk, Bruton’s tyrosine kinase; dsRNA, double-stranded RNA; FHA, forkhead-associated; HOIL-1L, haem-oxidized iron-regulatoryprotein 2 ubiquitin ligase-1; IκB, inhibitor of nuclear factor κB; IKK, IκB kinase; IL, interleukin; IRAK, IL-1 receptor-associated kinase; IRF, interferon-responsefactor; LPS, lipopolysaccharide; LRR, leucine-rich repeat domain; LUBAC, linear ubiquitin-chain-assembly complex; MyD88, myeloid differentiation factor88; Mal, MyD88 adaptor-like protein; MAPK, mitogen-activated protein kinase; NF-κB, nuclear factor κB; NEMO, NF-κB essential modulator; Pam3CSK4,tripalmitoyl-cysteinyl-seryl-(lysyl)4; PI3K, phosphoinositide 3-kinase; RIP, receptor-interacting protein; TAK1, transforming growth factor β-activated kinase1; TAB, TAK1-binding protein; TIR, Toll/IL-1 receptor; TLR, Toll-like receptor; TNF, tumour necrosis factor; TRAF, TNF receptor-associated factor; TIFA,TRAF-interacting protein with a forkhead-associated domain; TRIAD3A, two RING fingers and double-RING-finger-linked 3A; TRIF, TIR domain-containingadaptor inducing interferon-β; TRAM, TRIF-related adaptor protein; Ubc13, ubiquitin-conjugating enzyme 13; UBD, ubiquitin-binding domain.

1 To whom correspondence should be addressed (email [email protected]).

c© The Authors Journal compilation c© 2009 Biochemical Society

2 S. Carpenter and L. A. J. O’Neill

very difficult to achieve. This was overcome by the developmentof the so-called ‘hybrid LRR technique’ [10,13]. This involvedfusing hagfish variable lymphocyte receptors at the conservedLXXLXLXXN motifs within the LRR regions. Intriguingly, ithas been shown that these fusions do not alter the structure or thefunction of the TLRs and therefore allowed the structures to beobtained. Some hybrids failed to form soluble proteins becauseof the formation of atomic collisions or the hydrophobic coresat the fusion sites were exposed. Therefore only hybrids whichproduced soluble proteins which formed stable heterodimers andcould bind ligands were used for the crystallographic studies[10,13]. LRR-containing proteins can be divided into sevensubfamilies, and the TLRs belong to the ‘typical’ subfamily.Each LRR region consists of 24 amino acids and they possessthe conserved motif XLXXLXXLXLXXNXLXXLPXXXFX anddisplay the unique horseshoe shape associated with LRR proteins.Interestingly, the LRR regions of TLR1, TLR2 and TLR4 havebeen shown to deviate in their conformation when compared withother ‘typical’ family members [10,13]. Their LRRs have beenshown to be divided into an N-terminal region, a central regionand a C-terminal region. Their central regions lack the conservedasparagine ladder normally associated with conferring stabilityon the horseshoe structure. This anomaly could allow these TLRsto vary their structural conformations and this could explain theirability to bind a variety of ligands as well as co-receptors whichare essential for them to signal.

The TIR domains of TLR1, TLR2 and TLR10 have been struc-turally resolved [7,14]. The TIR domain consists of a five-strandedβ-sheet surrounded by five α-helices. Key residues within theTIR domain are the BB-loop, the DD-loop and the αC-helix.Experimental work has indicated that the BB-loop is essential forTIR dimerization and subsequent adaptor recruitment. Mutationswithin this region abolish signalling [7,15]. Studying TIR–TIRinteractions using X-ray crystallography has been very difficult,and the only TIR–TIR interface whose structure has been resolvedis that of the TLR10 dimer. It shows that the BB-loop andαC-helix on each TIR domain are the key to dimer formation. Thedimer interface has been shown to be made up of a hydrophobiccore which is made up of four residues of the BB-loop (Tyr668,Phe672, Pro674 and Ile678). These residues are not only involvedin dimer contact, but also critical for providing stability to theBB-loop. Phe710 and Cys706 of the αC-helix also contribute tothe hydrophobic interactions. These hydrophobic interactions arethen surrounded by a hydrogen-bonding network. The structurealso shows that there is a large region of the BB-loop which isnot involved in the dimer formation, but is instead exposed to thesurrounding solvent. This region is believed to be involved inthe recruitment of other TIR domain-containing adapter proteins[12]. We have had to rely on molecular modelling and dockingstudies to gain insights into these interactions. It has beensuggested that Mal and MyD88 bind to different regions in TLR4,allowing the formation of a heterotetrameric receptor–adaptorcomplex [16]. Gautam et al. [17] have suggested that the DD-loop of TLR2 forms a close contact with the BB-loop of TLR1and a mutation within the BB loop of TLR1 can interfere withsignalling. It has been suggested that peptide and peptide mimeticsof the BB-loop region could function as possible inhibitors of TLRsignalling [14]. A key goal is to determine a TIR–TIR/adaptor–TIR structure, as this could be useful for drug design.

TLRs AND THEIR LIGANDS

TLR3 and dsRNA

The first structure of TLR3 was described independently by twogroups [18,19]. Both groups showed that the LRR region of

TLR3 displays a heavily glycosylated horseshoe-shaped solenoidstructure. However, each group differed in where they predictedthe dsRNA would bind. Choe et al. [19] reasoned that dsRNAwould bind to the convex surface on the extracellular domainas it is glycan-free and therefore the dsRNA could bind to thepositively charged residues in this region. Bell et al. [18] proposedthat the nucleotide-binding site is in the concave surface thatpossesses four potential asparagine-linked glycosylation sites andhas an overall negative charge, and that it is the glycosylation ofthe concave surface which controls dsRNA binding. The crystalstructure of TLR3 bound to dsRNA was then resolved by Luiet al. [10]. They showed that TLR3 ectodomains are monomericin solution and that dimerization occurs upon ligand binding.Binding occurs on the N-terminal (LRR1–LRR3) and C-terminal(LRR19–LRR21) sites on the lateral side of the convex surface ofTLR3 ectodomains which lie on opposite sides of the dsRNA. Itis the positively charged residues on the termini of TLR3 whichmakes the major contributions to the interaction with the sugar-phosphate backbones of dsRNA. There is binding between the twoTLR3 C-terminal domains of the ectodomain and this providesstability to the dimer and is also believed to bring the TIR domainsinto contact, thus allowing for the downstream signalling cascadeto occur.

TLR1/2

The crystal structure of the extracellular domains of TLR2and TLR1 and a synthetic triacylated lipopeptide Pam3CSK4

[tripalmitoyl-cysteinyl-seryl-(lysyl)4] has been described [11].Similarly to that observed for TLR3, Jin et al. [11] could show that,without bound ligand, TLR1 and TLR2 resolved as monomers;however, when Pam3CSK4 was added, the TLRs formed stableheterodimers. Pam3CSK4 consists of three lipid chains; two werefound to insert into a hydrophobic pocket in TLR2 and the otherchain inserts into a narrow hydrophobic channel in TLR1. Bothpockets are located between the central and C-terminal domainsof the convex region. This was found to be quite unusual asmost ligand-binding sites on LRR proteins to date have beenfound to be on the concave surfaces. Further stability withinthe TLR1–TLR2 heterodimer and Pam3CSK4 is achieved byhydrophobic, hydrogen-bonding and ionic interactions betweenTLR1 and TLR2 [11].

TLR4

A crystal structure of the TLR4–MD2 complex binding theantagonist Eritoran has been described [20]. TLR4 requires itsco-receptor MD2 in order to recognize LPS [21]. LPS is madeup of a lipid A portion, which is the endotoxic component, acore oligosaccharide and a so-called O-antigen which is alsoa carbohydrate. The lipid A portion consists of phosphorylateddiglucosamine and four to seven acyl chains. Synthethic lipidA moieties consisting of six lipid chains and two phosphategroups are optimal for endotoxicity [22,23]. Lipid A containingfive or seven acyl chains is 100-fold less active. Eritoran is asynthetic molecule derived from the lipid A component of a non-pathogenic LPS from Rhodobacter sphaeroides [24]. It functionsas an antagonist of TLR4 signalling. Eritoran consists of fourlipid chains. The crystal structure obtained shows that all fouracyl chains of Eritoran bind to a hydrophobic pocket in MD2and it does not bind to TLR4 [20]. This observation suggestedthat LPS only binds to MD2 and no direct interaction betweenEritoran and TLR4 occurs. This has been shown recently notto be the case [12]. The much anticipated crystal structure of


Toll-like receptors 3

Figure 1 The interactions of LPS in the TLR4–MD2–LPS complex

(A) The structure of TLR4–MD2 complex binding to LPS is taken from the crystal structure [12]. The lipid A component of LPS is in red. LRRCT, C-terminal LRR; LRRNT, N-terminal LRR. (B) Five ofthe acyl chains of LPS bind to MD2 (in green) while the remaining chain (R2) is exposed to the surface and it binds TLR4 (in blue). Two phosphate groups of lipid A (1-phosphate and 4-phosphatehighlighted in red) are key to the TLR4 dimer formation through binding to positively charged lysine residues and an arginine residue on both TLR4 and MD2. (C) When TLR4–MD2 binds Eritoran,the Phe126 loop is exposed to the solvent area. (D) When TLR4–MD2 binds LPS, the Phe126 loop forms hydrophobic interactions with lipid chains R2 and R3 and with TLR4. These interactions causea structural shift in the Phe126 loop, which allows for the correct positioning of the R2 lipid chain to interact with TLR4 as well as allowing dimerization with TLR4 to occur.

TLR4–MD2 binding to LPS has finally been solved [12](Figure 1A). Park et al. [12] elegantly show that, withoutLPS, TLR4 and MD2 associate, but dimerization of the TLR4–MD2 complex with another TLR4–MD2 complex only occursfollowing binding of LPS. The structure shows LPS bindingto two copies of TLR4 and MD2 arranged symmetrically. Thestructure of MD2 shows that it consists of a β-cup fold structureconsisting of two antiparallel β-sheets, and it is this structurewhich forms the hydrophobic pocket to allow LPS to bind. Inorder for dimerization to occur, TLR4 forms hydrophobic andhydrophilic interactions with LPS which also associates with thePhe126 and Leu87 loops of MD2. Five of the lipid chains of LPSare buried within the hydrophobic pocket of MD2 in a mannersimilar to Eritoran. The remaining lipid chain (referred to as R2,Figure 1B) is on the surface of MD2 where it forms a hydrophobicinteraction with phenylalanine residues on TLR4 as shown inFigure 1. The binding of LPS causes structural changes in MD2mainly on the Phe126 loop, leading to hydrophilic interactionsbetween MD2 and TLR4, stabilizing the complex further. Itappears that the Phe126 loop of MD2 is critical for the activationof the TLR4 complex in response to LPS stimulation. Phe126 is atthe core of the dimerization interface following LPS stimulation.The crystal structure of TLR4–MD2 bound to the antagonistEritoran shows that the Phe126 loop of MD2 was exposed to thesurrounding solvent [20] (Figure 1C). This would suggest thatthis is the key to understanding why Eritoran is an antagonist.Clearly the structural change which occurs to the Phe126 loop ofMD2 following LPS stimulation is essential for the formation

of the dimers and subsequently the initiation of downstreamsignalling [12] (Figure 1D). Mutational analysis of the Phe126 loopof MD2 also supports this finding. If Phe126 is mutated, it preventsdimerization and downstream signalling does not occur [20,25].

The phosphate groups which are known to be critical for LPSactivity form ionic interactions with positively charged residues onboth TLR4 and MD2. The extra two lipid chains in LPS comparedwith Eritoran results in a displacement of the phosphorylatedglucosamines and this allows the phosphate groups to associatewith TLR4 and MD2.

Obtaining the crystal structure of TLR4 and LPS has helped usto understand the anomaly that lipid IVa which acts as an antag-onist in humans could act as a mild agonist for murine or equineTLR4. Walsh et al. [26] show that changing Arg385 in equine TLR4to glycine abolishes the agonist activity of lipid IVa [26]. Fromthe crystal structure of TLR4 and LPS, the 1-phosphate of LPSand the positively charged arginine residue in equine TLR4 couldbe enough to allow stronger interactions with the 1-phosphateof lipid IVa, thereby conferring agonistic activity of lipid IVain horse. These data again show the importance of the correctpositioning of the phosphate groups in LPS for activation of theTLR4 complex.

All structures of TLRs bound to their ligands reveal a com-mon ‘M’-shaped architecture which is shown in Figure 2. TheC-termini of the extracellular domains converge, which shouldtherefore allow the interaction between the TIR domains tooccur and initiate downstream signalling events [27]. Additionalstructural studies around the TIR interactions should improve



Figure 2 TLRs and their ligands

The ectodomains of TLR3 are monomeric in solution until they bind to dsRNA and dimerization occurs. The dsRNA interacts at both the N- and C-termini of each TLR3 ectodomain. The bindingoccurs on the glycan-free surface of each ectodomain [10]. The binding of dsRNA causes the C-terminal domains to associate and this adds stability to the dimer and is proposed to bring the twoTIR domains into contact to initiate downstream signalling. Two downstream signalling pathways are subsequently triggered, leading to activation of NF-κB and IRF3. TLR4 and MD2 interact beforestimulation with LPS. When LPS is added, it leads to dimerization of two TLR4–MD2 complexes. Five acyl chains of LPS bind the hydrophobic pocket of MD2, while the sixth acyl chain (in red)associates with TLR4 [12]. Dimerization of the tetrameric complex brings the TIR domains together and initiates downstream signalling through the adaptor molecules MyD88, Mal, TRIF and TRAM,which leads to the activation of the transcription factor NF-κB and the release of pro-inflammatory cytokines. TLR2 can heterodimerize with TLR1 and respond to triacylated lipopeptides or it canbind to TLR6 and respond to diacylated lipopeptides. Both complexes utilize the adaptors MyD88 and Mal to initiate downstream signalling, which leads to the activation of NF-κB and the releaseof pro-inflammatory cytokines. IFN, interferon; NAP1, nucleosome assembly protein 1; RANTES, regulated upon activation, normal T-cell expressed and secreted; TAK1, transforming growth factorβ-activated kinase 1.

our understanding of the downstream signalling events followingligand interactions.

POST-TRANSLATIONAL MODIFICATIONS IN TLRs

The other important biochemical aspect of TLR signalling wherethere has been recent progress is in post-translational modific-ation. As mentioned above, tight control of TLR signallingpathways is essential to maintain homoeostasis. Control is pro-vided by phosphorylation and ubiquitination.

Phosphorylation

Tyrosine phosphorylation of TLRs

Experimental evidence to date suggests that phosphorylation isan essential part of the formation of multiprotein complexes in

TLR signalling at the cell membrane. Tyrosine phosphorylationof TLR2, TLR3, TLR5 and TLR4 has been shown to be essentialfor initiation of downstream signalling events [28–31]. Tyr616 andTyr761 are essential for TLR2 signalling, as a double mutant ofthese residues abolishes NF-κB-dependent reporter activity [28].A study of the phosphorylation of TLR3 revealed that Tyr759 andTyr858 are important as mutations resulted in loss of signallingin response to dsRNA [28]. A key kinase linked to most TLRdownstream signalling is PI3K (phosphoinositide 3-kinase). Thep85 subunit of PI3K is capable of binding to phosphotyrosineresidues via its SH2 (Src homology 2) domain. Mutants ofTyr616 and Tyr761 in TLR2 prevent binding to the p85 subunitof PI3K and therefore results in an inhibition of NF-κB signalling[28]. Phosphorylation of Tyr759 on TLR3 has been shown to berequired to recruit PI3K which binds to its downstream kinaseAkt, and this is needed for IRF (interferon-response factor) 3phosphorylation and activation [28]. A putative PI3K-binding site



in TLR5 has been localized to Tyr798. Mutation of this residueinhibits signalling in response to flagellin. Tyr674 and Tyr680 aretwo phosphorylation sites in TLR4. If these sites are mutated toalanine they inhibit signalling, presumably by altering the TLR4–MyD88 interaction [28].

Phosphorylation of adaptor molecules

Much work has been carried out showing the post-translationalmodifications of the adaptor molecules Mal and TRAM. Mal isutilized by both TLR4 and TLR2 and it is subject to a numberof modifications. Mal has been shown to possess a PtdInsP2-binding domain, which enables Mal to be localized to the plasmamembrane where it can act as a bridging adaptor to MyD88[32]. Mal undergoes tyrosine phosphorylation by Btk (Bruton’styrosine kinase) [33]. Tyr86 and Tyr187 are important in Mal asmutant forms acted as dominant-negative inhibitors of NF-κBactivation following LPS stimulation [28]. An additional studyhas shown that Tyr86, Tyr106 and Tyr159 are important residues,as mutagenesis of these residues impaired Mal phosphorylation,affected its interaction with Btk and also impaired downstreamsignalling [34].

TRAM acts as a bridging adaptor for TRIF in the TLR4 sig-nalling complex and, like Mal, it undergoes a number of modific-ations [35]. TLR4 has been shown to be capable of signallingfrom the plasma membrane and from the endosome. TRAM hasbeen shown to traffic to early endosomes where it binds TRIFand results in the activation of TBK-1 {TANK [TRAF (TNF-receptor-associated factor)-associated NF-κB activator]-bindingkinase 1} which phosphorylates IRF3, leading to the inductionof interferon-β [32]. Recently, a splice variant of TRAM hasbeen described termed TRAM adaptor with gold domain (TAG),which displaces TRIF from TRAM in the late endosome to limitsignalling [36]. TRAM has been shown to undergo myristoylationat its N-terminus, allowing it to localize to the plasma membrane[37]. TRAM has been shown to be phosphorylated on Ser16 byprotein kinase Cε following LPS stimulation [38].

The role of IRAKs (IL-1 receptor-associated kinases)

Following receptor activation and adaptor recruitment, there are anumber of signalling molecules involved downstream which areregulated by phosphorylation. Four IRAKs have been identified:IRAK-1, IRAK-4, IRAK-2 and IRAK-M. IRAK-1 and IRAK-4are serine/threonine kinases. IRAK-4 has been shown to phos-phorylate IRAK-1, and this plays a key role in TLR signallingcascades. IRAK-1 has been shown to activate NF-κB, notthrough the classical pathway involving degradation of IκB(inhibitor of NF-κB), but through facilitating the phosphorylationof p65/RelA. This was confirmed by studies on IRAK-1-knockoutcells, which showed impaired phosphorylation of p65/RelA,whereas IκB degradation was normal [39]. IRAK-1 alsoappears to be important for phosphorylation of IRF5–IRF7,as IRAK-1-knockout cells show reduced activation of IRF5–IRF7 following stimulation with TLR7, TLR8 and TLR9 ligands[40,41]. As mentioned above, IRAK-4 phosphorylates IRAK-1. Once phosphorylated, IRAK-1 can then continue to undergoautophosphorylation followed by degradation [42]. IRAK-4 ismore important for the activation of NF-κB than IRAK-1.This was demonstrated by examining IRAK-4-knockout micewhich showed impaired NF-κB activation and reduced cytokineproduction in response to TLR ligands [43]. The kinase activityof IRAK-4 has also been shown to be essential for signalling, asoverexpression of the kinase-dead form of IRAK-4 resulted in areduction in LPS-induced NF-κB activation [44]. Children with

an IRAK-4 deficiency suffer from persistent pyogenic bacterialinfections and do not respond to a number of TLR ligands[45]; however, intriguingly, children with this deficiency return tonormal as adults [21]. Clearly, IRAK-4 is essential for protectionin children, but is expendable in adults. IRAK-2 and IRAK-Mare pseudokinases, and IRAK-M has been shown to act as anegative regulator of TLR signalling [46]. IRAK-M-knockoutmacrophages displayed higher NF-κB activation and increasedpro-inflammatory cytokines in response to TLR ligands [46].IRAK-2 can interact with TRAF6, MyD88 and Mal and canactivate NF-κB. Studies carried out on IRAK-2-knockout miceindicate that IRAK-2 is needed for optimal responses to TLRligands and the mice produce less IL-6 and TNFα [47]. However,the mice show normal MAPK and early-phase NF-κB activation,which is probably due to redundancy with IRAK-1. IRAK-2/IRAK-1 double knockouts are completely resistant to LPSor CpG-induced septic shock and macrophages show reducedresponses to all TLR ligands except TLR3. These results indicatethat IRAK-1 and IRAK-2 work together to mount appropriateresponses to TLR stimulation.

Ubiquitination

Ubiquitination, like phosphorylation, is a reversible covalentmodification involved in the regulation and localization of targetproteins. Ubiquitination is emerging as a key process required forregulating the immune system. Phosphorylation has often beenshown to precede ubiquitination, and a good example is IκB,which initially gets phosphorylated in order to release NF-κB tothe nucleus. IκB is subsequently degraded by the proteasometo terminate the signal. For the remainder of this review, wediscuss the various types of ubiquitination and the proteinsinvolved in regulating TLR signalling.

Ubiquitin

Ubiquitin consists of 76 amino acids and is highly conservedacross species. It covalently links to one or more lysine residueson target proteins through the following three steps: initially, aubiquitin-activating enzyme (E1) processes an inactive ubiquitinprecursor molecule. The activated molecule is then transferred toa cysteinyl group on a ubiquitin-conjugating enzyme (E2). Thefinal step involves a ubiquitin–protein ligase (E3) whose job isto transfer the ubiquitin from the E2 on to a lysine residue ona substrate. Since ubiquitin contains seven lysine residues (Lys6,Lys11, Lys27, Lys29, Lys33, Lys48 and Lys63), and each of these canbe conjugated to the C-terminus of another ubiquitin, it enablesthis process to be repeated to form polyubiquitin chains. Thiscan result in Lys48-linked or Lys63-linked ubiquitination. Lys48-linked ubiquitination results in degradation of the protein bythe 26S proteasome, whereas Lys63-linked ubiquitination leadsto activation of the substrate in the absence of degradation[48]. Proteins can also undergo mono-ubiquitination and linearubiquitination in more detail. Both processes are covered inmore detail below. Figure 3 illustrates the four known types ofubiquitination including examples of each in the TLR system.

Mono-ubiquitination

Proteins which can bind to ubiquitin contain UBDs (ubiquitin-binding domains) and can be referred to as ubiquitin receptors.Ubiquitin receptors can interact directly with mono-ubiquitin orthey can interact with polyubiquitin chains on proteins whichhave undergone ubiquitination [49]. Often, ubiquitin receptorsare themselves mono-ubiquitinated, and this is known as coupledmono-ubiquitination. It has also been proposed that ubiquitin



Figure 3 Ubiquitination modifications, including examples within the TLR signalling pathway

The process of ubiquitination occurs when ubiquitin covalently links to one or more lysine (K) residues on target proteins through the use of the following three proteins: a ubiquitin-activatingenzyme (E1), a ubiquitin-conjugating enzyme (E2) and a ubiquitin–protein ligase (E3). Since ubiquitin (Ub) contains seven lysine residues and each of these can be conjugated to the C-terminus ofanother ubiquitin, it enables this process to be repeated to form polyubiquitin chains. This can result in Lys48- or Lys63-linked ubiquitination. Lys48-linked ubiquitination results in degradation of theprotein by the 26S proteasome, whereas Lys63-linked ubiquitination leads to activation of the substrate in the absence of degradation. Mono-ubiquitination occurs when a single ubiquitin is addedto a substrate. This can affect protein conformation, binding to other proteins, localization, target protein lysosomal degradation or inhibition of the target protein. Linear ubiquitinated proteins canact as scaffolds to recruit other downstream signalling components.

receptors can recruit ubiquitin machinery to promote coupledmono-ubiquitination. The process of mono-ubiquitination of atarget protein can affect its conformation and ability to interactwith other proteins. It can also affect the target protein’ssubcellular localization and activity and can cause the protein toundergo lysosomal degradation [50]. In relation to TLR signalling,the mono-ubiquitination of histone H2A has been linked to therepression of specific chemokines in macrophages affecting cellmigration in response to TLR activation [51]. It has been shownthat ubiquitin receptors which undergo mono-ubiquitination areinvolved in intramolecular interactions between mono-ubiquitinand their UBDs and are therefore inhibited as a mechanism forcontrol within the pathway [52].

NEMO (NF-κB essential modulator) and linear ubiquitination

An additional form of ubiquitination known as linear ubiquitin-ation has of late been linked to NF-κB activation. Linear ubi-quitination chains are unique in that this linkage does not involvelysine linkages; instead, the linkages are between the N- andC-terminus of each ubiquitin [53]. NEMO has been known toundergo ubiquitination, and it has been shown recently thatthis is linear ubiquitination, which is essential for its ability toactivate NF-κB and not the MAPK pathway [54–56]. NEMO hadbeen shown previously to undergo Lys63-linked polyubiquitin-ation, resulting in the recruitment of the TAK1 (transforminggrowth factor β-activated kinase 1)–TAB (TAK1-binding protein)1–TAB2 complex. Ubc13 (ubiquitin-conjugating enzyme 13)encodes an E2 enzyme which is essential for generating Lys63-linked chains, and, when the Ubc13 conditional knockout

mouse was generated, it had very little effect on NF-κB activ-ation, thus suggesting the presence of a Ubc13-independent NF-κB activation pathway which now appears to involve linear ubi-quitination [37]. Tokunaga et al. [54] have shown that the LUBAC(linear ubiquitin-chain-assembly complex) which consists of twoRING finger proteins HOIL-1L (haem-oxidized iron-regulatoryprotein 2 ubiquitin ligase 1) and HOIP (HOIL-1L-interactingprotein) are involved in forming head-to-tail linear polyubiquitinchains on NEMO. They have shown that it is Lys285 andLys309 in NEMO that are linearly ubiquitinated by LUBAC. Asmentioned above, two additional papers have been publishedrecently showing that the UBAN [ubiquitin binding in ABIN(A20-binding inhibitor of NF-κB) and NEMO] motif present inNEMO binds linear ubiquitin chains with stronger affinity thanLys63-linked chains [55,56]. However, the exact physiologicalfunctions for linear ubiquitination are unknown. It is proposedthat the linear ubiquitin chains may act as scaffolds to recruitactivating complexes. All results indicate that linear ubiquitinationof NEMO is essential for NF-κB activation, but is not requiredfor the activation of MAPK pathway.

Ubiquitin ligases

The final area that we consider are the key ubiquitin ligasesassociated with the TLRs: TRAF6, TRIAD3A (two RING fingersand double-RING-finger-linked 3A), A20 and, the most recentlydescribed, members of the Pellino family.

TRAF6. TRAF6 is a RING-domain-containing E3 lig-ase. TRAF6 can cause Lys63-linked polyubiquitination of itself



through the E2 enzyme complex containing Ubc13 and Uev1A(ubiquitin-conjugating enzyme E2 variant 1A) [57]. TAB2 andTAB3 are capable of recognizing Lys63-linked polyubiquitinchains on TRAF6. TAK1 is recruited to TRAF6 via TAB2 andTAB3 [28]. The IKK (IκB kinase) complex is then also recruited.Cross-talk between phosphorylation and ubiquitination occurs atthis point in the pathway. IκBα is phosphorylated on Ser32 or Ser36,it then undergoes Lys48-linked ubiquitination of Lys21 or Lys22,releasing NF-κB to the nucleus [58,59]. Much work has beencarried out to determine what molecule is involved in promotingTRAF6 ligase activity. TIFA [TRAF-interacting protein with anFHA (forkhead-associated) domain] has been shown to induceTRAF6 oligomerization and polyubiquitination [60]. Mutationswithin the TRAF6-binding site of TIFA inhibited activation ofIKK in vitro. These results would suggest that TIFA is neededfor TRAF6 oligomerization, and this could be essential for theactivation of the E3 ligase activity of TRAF6. Recently, Keatinget al. [61] have shown that IRAK-2 is the key IRAK and notIRAK-1 in mediating TRAF6 polyubiquitination in TLRsignalling. They showed that A52, which is a known inhibitorof TLR signalling, achieves this inhibition by targeting IRAK-2.They also showed that mutants of IRAK-2 were unable to activateNF-κB or promote TRAF6 ubiquitination. Clearly, TRAF6 Lys63-linked polyubiquitination is an essential part of TLR signalling;however, much work is still needed to determine whether thereare more proteins than TIFA and IRAK-2 involved in TRAF6oligomerization and subsequent activation.

TRIAD3A. TRIAD3A is a ubiquitin ligase which has beenshown to be involved in the ubiquitination and degradationof TLRs [62]. The protein contains a TRIAD domain oftenfound in E3 ligases. TRIAD3A has been shown to promote theubiquitination and degradation of TLR9, TLR4 and, to a lesserextent, TLR3 and TLR5. TRIAD3A did not affect TLR2 as thetwo proteins are unable to interact. Knockdown of TRIAD3Aresulted in enhanced responses to LPS, flagellin and CpG DNA,suggesting that, overall, TRIAD3A can function as a negativeregulator of TLR signalling.

Pellinos. The Pellinos are the most recent family of proteins toemerge as key players in TLR signalling. There are three membersof the Pellino family: Pellino1, Pellino2 and Pellino3. Thereare also splice variants of Pellino 3: Pellino3L and Pellino3S.They have all been shown to be E3 ligases. Pellinos possessa C-terminal RING-like domain which confers E3 ubiquitinligase activity. Recently, the crystal structure of Pellino2 hasbeen resolved and it shows that the Pellinos also possess aphosphothreonine-binding FHA domain [63]. The main functionof the Pellinos within TLR signalling is that they bind and causeLys63-linked polyubiquitination of IRAK-1. IRAK-1 and IRAK-4have also been shown to phosphorylate the Pellino proteins whichthen undergo Lys48-linked polyubiquitination, resulting in theirdegradation [64–66]. The Pellino proteins have also been shownto bind to TRAF6 and TAK1. This is an interesting observation,since Pellinos can activate IRAK-1 which is known to bind toTRAF6, suggesting that the Pellinos are essential proteins neededfor the activation of this complex [67].

The literature to date on the role of each family member of thePellinos have resulted in conflicting data. The role of Pellino1 inthe activation of NF-κB has been questioned; however, it doesappear to have a role in IL-1-induced NF-κB, as knockdown ofPellino1 impairs this activation [44]. Pellino1 has been shown notto be involved in the MAPK pathway [44]. Again, the role ofPellino2 seems quite unclear. Overexpression of Pellino2 drivesboth the NF-κB and the MAPK pathways [68].

A number of papers have reported on the function of Pellino3.Butler et al. [64] show that of the three family membersPellino3 has the most E3 ligase activity and it promotesLys63-linked polyubiquitination of IRAK-1. Previous reportssuggested that IRAK-1 is degraded following stimulation [42,44].However, the recent work on Pellinos would suggest thatIRAK-1 is Lys63- and not Lys48-linked ubiquitinated. Theactivation of IRAK-1 is believed to bring in the IKK complexthrough recruitment of NEMO to Lys63-linked chains onIRAK-1. TAK1 is recruited to the Lys63-linked polyubiquitinchains on TRAF6, and these linkages allow the IKK complexand TAK1 to come into close proximity. TAK1 can thenactivate the IKKs and stimulate the downstream signalling events[44]. Interestingly, a recent paper has been published whichsuggests that Pellino3 can act as a negative regulator of TAK1-mediated activation of NF-κB [69]. This suggests that there isone ubiquitination site on IRAK-1 (Lys134) and that this sitecan be Lys48- or Lys63-linked polyubiquitinated. The authorsshow that Pellino3b-mediated Lys63-linked polyubiquitinationcan compete with Lys48-linked polyubiquitination to preventIRAK-1 degradation and, in doing so, inhibit IL-1-inducedTAK1 activation by preventing the TRAF6–TAK1–TAB3–TAB3complex from translocating from the membrane to the cytosol.This all reverts to the role of IRAK-1 and whether degradationor activation is required for the subsequent activation of TAK1and the IKK complex. Clearly, more work is needed to clarify therole of the Pellinos. The generation of Pellino-knockout mice willhopefully shed some much needed light on the exact functions ofthese family members within IL-1 and TLR signalling.

A20. A20 can act as both a deubiquitinating and a ubiquitinatingenzyme. A20 can remove ubiquitin from TRAF6, allowingTRAF6 to be recycled as well as terminating TLR signals[63,70]. Results from A20-knockout cells show enhanced NF-κB activation in response to TLR2, TLR4 and TLR9 ligands.A20 has also been shown to be capable of simultaneouslyremoving Lys63-linked polyubiquitin chains from RIP (receptor-interacting protein) and adding Lys48-linked chains resulting inRIP degradation [71].

FINAL CONCLUSIONS

The TLR signalling pathways are highly complex. Tight controlis needed in order to maintain homoeostasis. Solving the crystalstructures of TLRs binding to their ligands has allowed usto understand further the initiation of TLR signals. The threestructures (TLR3, TLR1–TLR2 and TLR4) clearly elucidate themolecular basis for TLR dimerization and subsequent signaltransduction. There are a number of neutralizing antibodiesavailable to block signalling and, more recently, there aresmall molecules such as Eritoran for TLR4, which are capableof inhibiting signalling. Understanding how ligands bind theirreceptors through solving crystal structures will allow for betterinhibitors to be designed to block this process.

Part of the complexity within TLR signalling comes from themany post-translational modifications which occur and are sum-marized in Figure 4. The role of mono-ubiquitination andlinear ubiquitination is emerging as an exciting area of regul-ation for TLR signalling. Further studies are needed to fullyunderstand all the proteins responsible for the phosphorylationand ubiquitination events involved in the regulation of TLRsignalling. Much work has already been done in an attempt todevelop drugs to exploit proteins involved in phosphorylationand ubiquitination, since both processes when dysregulated canlead to disease. Targeting components of phosphorylation has



Figure 4 Phosphorylation and ubiquitination in TLR signalling

Dimerization of TLRs occurs following ligand binding, and this is proposed to bring the TIR domains into contact to initiate downstream signalling through the adaptor molecules MyD88,Mal, TRIF and TRAM. A series of post-translational modifications occur, including phosphorylation and Lys48- (K48) and Lys63- (K63) linked and linear ubiquitination. Downstream signallingpathways are subsequently triggered, leading to activation of NF-κB, IRF3 and IRF7, which results in the release of pro-inflammatory cytokines. An animated version of this Figure can be seen athttp://www.BiochemJ.org/bj/422/0001/bj4220001add.htm. AP1, activator protein 1; IFN, interferon; NAP1, nucleosome assembly protein 1; PKC, protein kinase C; TAK1, transforming growth factorβ-activated kinase 1.

yielded clinical success. Ubiquitin ligases have been a targetfor potential drugs to treat cancer; however, to date, none ofthe approaches has yielded significant clinical advances [72].Targeting ubiquitination may prove difficult owing to the fact thatthese processes occur alongside many other signalling pathwaysand E3 ligases may not be specific enough as targets [72].However, further elucidation of the structural and molecular basisfor the signalling should help in the ultimate goal of targetingthese pathways therapeutically.

FUNDING

The work performed in our laboratory is supported by Science Foundation Ireland.

REFERENCES

1 Janeway, Jr, C. A. (1992) The immune system evolved to discriminate infectious nonselffrom noninfectious self. Immunol. Today 13, 11–16

2 Rock, F. L., Hardiman, G., Timans, J. C., Kastelein, R. A. and Bazan, J. F. (1998) A familyof human receptors structurally related to Drosophila Toll. Proc. Natl. Acad. Sci. U.S.A.95, 588–593

3 Kobe, B. and Deisenhofer, J. (1995) A structural basis of the interactions betweenleucine-rich repeats and protein ligands. Nature 374, 183–186

4 Gay, N. J. and Keith, F. J. (1991) Drosophila Toll and IL-1 receptor. Nature 351,355–356

5 Medzhitov, R., Preston-Hurlburt, P. and Janeway, Jr, C. A. (1997) A human homologueof the Drosophila Toll protein signals activation of adaptive immunity. Nature 388,394–397

6 Takeda, K. and Akira, S. (2007) Toll-like receptors. Curr. Protoc. Immunol., Unit 14.127 Xu, Y., Tao, X., Shen, B., Horng, T., Medzhitov, R., Manley, J. L. and Tong, L. (2000)

Structural basis for signal transduction by the Toll/interleukin-1 receptor domains.Nature 408, 111–115

8 Carty, M., Goodbody, R., Schroder, M., Stack, J., Moynagh, P. N. and Bowie, A. G. (2006)The human adaptor SARM negatively regulates adaptor protein TRIF-dependent Toll-likereceptor signaling. Nat. Immunol. 7, 1074–1081

9 Leonard, J. N., Ghirlando, R., Askins, J., Bell, J. K., Margulies, D. H., Davies, D. R. andSegal, D. M. (2008) The TLR3 signaling complex forms by cooperative receptordimerization. Proc. Natl. Acad. Sci. U.S.A. 105, 258–263

10 Liu, L., Botos, I., Wang, Y., Leonard, J. N., Shiloach, J., Segal, D. M. and Davies, D. R.(2008) Structural basis of Toll-like receptor 3 signaling with double-stranded RNA.Science 320, 379–381

11 Jin, M. S., Kim, S. E., Heo, J. Y., Lee, M. E., Kim, H. M., Paik, S. G., Lee, H. and Lee, J. O.(2007) Crystal structure of the TLR1–TLR2 heterodimer induced by binding of atri-acylated lipopeptide. Cell 130, 1071–1082



12 Park, B. S., Song, D. H., Kim, H. M., Choi, B. S., Lee, H. and Lee, J. O. (2009) Thestructural basis of lipopolysaccharide recognition by the TLR4–MD-2 complex.Nature 458, 1191–1195

13 Kim, H. M., Oh, S. C., Lim, K. J., Kasamatsu, J., Heo, J. Y., Park, B. S., Lee, H., Yoo, O. J.,Kasahara, M. and Lee, J. O. (2007) Structural diversity of the hagfish variable lymphocytereceptors. J. Biol. Chem. 282, 6726–6732

14 Nyman, T., Stenmark, P., Flodin, S., Johansson, I., Hammarstrom, M. and Nordlund, P.(2008) The crystal structure of the human Toll-like receptor 10 cytoplasmic domainreveals a putative signaling dimer. J. Biol. Chem. 283, 11861–11865

15 Underhill, D. M., Ozinsky, A., Hajjar, A. M., Stevens, A., Wilson, C. B., Bassetti, M. andAderem, A. (1999) The Toll-like receptor 2 is recruited to macrophage phagosomes anddiscriminates between pathogens. Nature 401, 811–815

16 Dunne, A., Ejdeback, M., Ludidi, P. L., O’Neill, L. A. and Gay, N. J. (2003) Structuralcomplementarity of Toll/interleukin-1 receptor domains in Toll-like receptors and theadaptors Mal and MyD88. J. Biol. Chem. 278, 41443–41451

17 Gautam, J. K., Ashish, Comeau, L. D., Krueger, J. K. and Smith, Jr, M. F. (2006) Structuraland functional evidence for the role of the TLR2 DD loop in TLR1/TLR2 heterodimerizationand signaling. J. Biol. Chem. 281, 30132–30142

18 Bell, J. K., Botos, I., Hall, P. R., Askins, J., Shiloach, J., Segal, D. M. and Davies, D. R.(2005) The molecular structure of the Toll-like receptor 3 ligand-binding domain.Proc. Natl. Acad. Sci. U.S.A. 102, 10976–10980

19 Choe, J., Kelker, M. S. and Wilson, I. A. (2005) Crystal structure of human Toll-likereceptor 3 (TLR3) ectodomain. Science 309, 581–585

20 Kim, H. M., Park, B. S., Kim, J. I., Kim, S. E., Lee, J., Oh, S. C., Enkhbayar, P.,Matsushima, N., Lee, H., Yoo, O. J. and Lee, J. O. (2007) Crystal structure of theTLR4–MD-2 complex with bound endotoxin antagonist Eritoran. Cell 130,906–917

21 Shimazu, R., Akashi, S., Ogata, H., Nagai, Y., Fukudome, K., Miyake, K. and Kimoto, M.(1999) MD-2, a molecule that confers lipopolysaccharide responsiveness on Toll-likereceptor 4. J. Exp. Med. 189, 1777–1782

22 Erridge, C., Bennett-Guerrero, E. and Poxton, I. R. (2002) Structure and function oflipopolysaccharides. Microbes Infect. 4, 837–851

23 Rietschel, E. T., Seydel, U., Zahringer, U., Schade, U. F., Brade, L., Loppnow, H., Feist, W.,Wang, M. H., Ulmer, A. J., Flad, H. D. et al. (1991) Bacterial endotoxin: molecularrelationships between structure and activity. Infect. Dis. Clin. North Am. 5, 753–779

24 Mullarkey, M., Rose, J. R., Bristol, J., Kawata, T., Kimura, A., Kobayashi, S., Przetak, M.,Chow, J., Gusovsky, F., Christ, W. J. and Rossignol, D. P. (2003) Inhibition of endotoxinresponse by e5564, a novel Toll-like receptor 4-directed endotoxin antagonist.J. Pharmacol. Exp. Ther. 304, 1093–1102

25 Kobayashi, M., Saitoh, S., Tanimura, N., Takahashi, K., Kawasaki, K., Nishijima, M.,Fujimoto, Y., Fukase, K., Akashi-Takamura, S. and Miyake, K. (2006) Regulatory rolesfor MD-2 and TLR4 in ligand-induced receptor clustering. J. Immunol. 176,6211–6218

26 Walsh, C., Gangloff, M., Monie, T., Smyth, T., Wei, B., McKinley, T. J., Maskell, D., Gay, N.and Bryant, C. (2008) Elucidation of the MD-2/TLR4 interface required for signaling bylipid IVa. J. Immunol. 181, 1245–1254

27 Jin, M. S. and Lee, J. O. (2008) Structures of the Toll-like receptor family and its ligandcomplexes. Immunity 29, 182–191

28 Arbibe, L., Mira, J. P., Teusch, N., Kline, L., Guha, M., Mackman, N., Godowski, P. J.,Ulevitch, R. J. and Knaus, U. G. (2000) Toll-like receptor 2-mediated NF-κB activationrequires a Rac1-dependent pathway. Nat. Immunol. 1, 533–540

29 Sarkar, S. N., Smith, H. L., Rowe, T. M. and Sen, G. C. (2003) Double-stranded RNAsignaling by Toll-like receptor 3 requires specific tyrosine residues in its cytoplasmicdomain. J. Biol. Chem. 278, 4393–4396

30 Sarkar, S. N., Peters, K. L., Elco, C. P., Sakamoto, S., Pal, S. and Sen, G. C. (2004) Novelroles of TLR3 tyrosine phosphorylation and PI3 kinase in double-stranded RNA signaling.Nat. Struct. Mol. Biol. 11, 1060–1067

31 Ivison, S. M., Khan, M. A., Graham, N. R., Bernales, C. Q., Kaleem, A., Tirling, C. O.,Cherkasov, A. and Steiner, T. S. (2007) A phosphorylation site in the Toll-like receptor 5TIR domain is required for inflammatory signalling in response to flagellin. Biochem.Biophys. Res. Commun. 352, 936–941

32 Kagan, J. C. and Medzhitov, R. (2006) Phosphoinositide-mediated adaptor recruitmentcontrols Toll-like receptor signaling. Cell 125, 943–955

33 Gray, P., Dunne, A., Brikos, C., Jefferies, C. A., Doyle, S. L. and O’Neill, L. A. (2006)MyD88 adapter-like (Mal) is phosphorylated by Bruton’s tyrosine kinase during TLR2 andTLR4 signal transduction. J. Biol. Chem. 281, 10489–10495

34 Piao, W., Song, C., Chen, H., Wahl, L. M., Fitzgerald, K. A., O’Neill, L. A. and Medvedev,A. E. (2008) Tyrosine phosphorylation of MyD88 adapter-like (Mal) is critical for signaltransduction and blocked in endotoxin tolerance. J. Biol. Chem. 283, 3109–3119

35 Kenny, E. F. and O’Neill, L. A. (2008) Signalling adaptors used by Toll-like receptors:an update. Cytokine 43, 342–349

36 Palsson-McDermott, E. M., Doyle, S. L., McGettrick, A. F., Hardy, M., Husebye, H.,Banahan, K., Gong, M., Golenbock, D., Espevik, T. and O’Neill, L. A. (2009) TAG, a splicevariant of the adaptor TRAM, negatively regulates the adaptor MyD88-independent TLR4pathway. Nat. Immunol. 10, 579–586

37 Rowe, D. C., McGettrick, A. F., Latz, E., Monks, B. G., Gay, N. J., Yamamoto, M., Akira, S.,O’Neill, L. A., Fitzgerald, K. A. and Golenbock, D. T. (2006) The myristoylation ofTRIF-related adaptor molecule is essential for Toll-like receptor 4 signal transduction.Proc. Natl. Acad. Sci. U.S.A. 103, 6299–6304

38 McGettrick, A. F., Brint, E. K., Palsson-McDermott, E. M., Rowe, D. C., Golenbock, D. T.,Gay, N. J., Fitzgerald, K. A. and O’Neill, L. A. (2006) Trif-related adapter molecule isphosphorylated by PKCε during Toll-like receptor 4 signaling. Proc. Natl. Acad.Sci. U.S.A. 103, 9196–9201

39 Li, L., Su, J. and Xie, Q. (2007) Differential regulation of key signaling molecules ininnate immunity and human diseases. Adv. Exp. Med. Biol. 598, 49–61

40 Uematsu, S., Sato, S., Yamamoto, M., Hirotani, T., Kato, H., Takeshita, F., Matsuda, M.,Coban, C., Ishii, K. J., Kawai, T. et al. (2005) Interleukin-1 receptor-associated kinase-1plays an essential role for Toll-like receptor (TLR)7- and TLR9-mediated interferon-αinduction. J. Exp. Med. 201, 915–923

41 Schoenemeyer, A., Barnes, B. J., Mancl, M. E., Latz, E., Goutagny, N., Pitha, P. M.,Fitzgerald, K. A. and Golenbock, D. T. (2005) The interferon regulatory factor, IRF5,is a central mediator of Toll-like receptor 7 signaling. J. Biol. Chem. 280,17005–17012

42 Yamin, T. T. and Miller, D. K. (1997) The interleukin-1 receptor-associated kinase isdegraded by proteasomes following its phosphorylation. J. Biol. Chem. 272,21540–21547

43 Suzuki, N., Suzuki, S., Duncan, G. S., Millar, D. G., Wada, T., Mirtsos, C., Takada, H.,Wakeham, A., Itie, A., Li, S. et al. (2002) Severe impairment of interleukin-1 andToll-like receptor signalling in mice lacking IRAK-4. Nature 416, 750–756

44 Jiang, Z., Ninomiya-Tsuji, J., Qian, Y., Matsumoto, K. and Li, X. (2002) Interleukin-1(IL-1) receptor-associated kinase-dependent IL-1-induced signaling complexesphosphorylate TAK1 and TAB2 at the plasma membrane and activate TAK1 in the cytosol.Mol. Cell. Biol. 22, 7158–7167

45 Picard, C., Puel, A., Bonnet, M., Ku, C. L., Bustamante, J., Yang, K., Soudais, C., Dupuis,S., Feinberg, J., Fieschi, C. et al. (2003) Pyogenic bacterial infections in humans withIRAK-4 deficiency. Science 299, 2076–2079

46 Kobayashi, K., Hernandez, L. D., Galan, J. E., Janeway, Jr, C. A., Medzhitov, R. andFlavell, R. A. (2002) IRAK-M is a negative regulator of Toll-like receptor signaling. Cell110, 191–202

47 Kawagoe, T., Sato, S., Matsushita, K., Kato, H., Matsui, K., Kumagai, Y., Saitoh, T., Kawai,T., Takeuchi, O. and Akira, S. (2008) Sequential control of Toll-like receptor-dependentresponses by IRAK1 and IRAK2. Nat. Immunol. 9, 684–691

48 Dunne, A. and O’Neill, L. (2005) New insights into the post-translational modification ofToll-like receptor signaling molecules. J. Endotoxin Res. 11, 325–332

49 Hicke, L., Schubert, H. L. and Hill, C. P. (2005) Ubiquitin-binding domains. Nat. Rev. Mol.Cell Biol. 6, 610–621

50 Haglund, K. and Dikic, I. (2005) Ubiquitylation and cell signaling. EMBO J. 24,3353–3359

51 Zhou, W., Zhu, P., Wang, J., Pascual, G., Ohgi, K. A., Lozach, J., Glass, C. K. andRosenfeld, M. G. (2008) Histone H2A monoubiquitination represses transcriptionby inhibiting RNA polymerase II transcriptional elongation. Mol. Cell 29,69–80

52 Hoeller, D., Crosetto, N., Blagoev, B., Raiborg, C., Tikkanen, R., Wagner, S., Kowanetz, K.,Breitling, R., Mann, M., Stenmark, H. and Dikic, I. (2006) Regulation of ubiquitin-bindingproteins by monoubiquitination. Nat. Cell Biol. 8, 163–169

53 Kirisako, T., Kamei, K., Murata, S., Kato, M., Fukumoto, H., Kanie, M., Sano, S.,Tokunaga, F., Tanaka, K. and Iwai, K. (2006) A ubiquitin ligase complex assembles linearpolyubiquitin chains. EMBO J. 25, 4877–4887

54 Tokunaga, F., Sakata, S., Saeki, Y., Satomi, Y., Kirisako, T., Kamei, K., Nakagawa, T., Kato,M., Murata, S., Yamaoka, S. et al. (2009) Involvement of linear polyubiquitylation ofNEMO in NF-κB activation. Nat. Cell Biol. 11, 123–132

55 Rahighi, S., Ikeda, F., Kawasaki, M., Akutsu, M., Suzuki, N., Kato, R., Kensche, T.,Uejima, T., Bloor, S., Komander, D. et al. (2009) Specific recognition of linear ubiquitinchains by NEMO is important for NF-κB activation. Cell 136, 1098–1109

56 Lo, Y. C., Lin, S. C., Rospigliosi, C. C., Conze, D. B., Wu, C. J., Ashwell, J. D., Eliezer, D.and Wu, H. (2009) Structural basis for recognition of diubiquitins by NEMO. Mol. Cell33, 602–615

57 Chen, Z. J. (2005) Ubiquitin signalling in the NF-κB pathway. Nat. Cell Biol. 7,758–765

58 Scherer, D. C., Brockman, J. A., Chen, Z., Maniatis, T. and Ballard, D. W. (1995)Signal-induced degradation of IκBα requires site-specific ubiquitination. Proc. Natl.Acad. Sci. U.S.A. 92, 11259–11263



59 Chen, Z., Hagler, J., Palombella, V. J., Melandri, F., Scherer, D., Ballard, D. andManiatis, T. (1995) Signal-induced site-specific phosphorylation targets IκBα to theubiquitin–proteasome pathway. Genes Dev. 9, 1586–1597

60 Ea, C. K., Sun, L., Inoue, J. and Chen, Z. J. (2004) TIFA activates IκB kinase (IKK) bypromoting oligomerization and ubiquitination of TRAF6. Proc. Natl. Acad. Sci. U.S.A.101, 15318–15323

61 Keating, S. E., Maloney, G. M., Moran, E. M. and Bowie, A. G. (2007) IRAK-2 participatesin multiple Toll-like receptor signaling pathways to NF-κB via activation of TRAF6ubiquitination. J. Biol. Chem. 282, 33435–33443

62 Chuang, T. H. and Ulevitch, R. J. (2004) Triad3A, an E3 ubiquitin–protein ligaseregulating Toll-like receptors. Nat. Immunol. 5, 495–502

63 Lin, C. C., Huoh, Y. S., Schmitz, K. R., Jensen, L. E. and Ferguson, K. M. (2008) Pellinoproteins contain a cryptic FHA domain that mediates interaction with phosphorylatedIRAK1. Structure 16, 1806–1816

64 Butler, M. P., Hanly, J. A. and Moynagh, P. N. (2005) Pellino3 is a novel upstreamregulator of p38 MAPK and activates CREB in a p38-dependent manner. J. Biol. Chem.280, 27759–27768

65 Butler, M. P., Hanly, J. A. and Moynagh, P. N. (2007) Kinase-active interleukin-1receptor-associated kinases promote polyubiquitination and degradation of the Pellinofamily: direct evidence for PELLINO proteins being ubiquitin–protein isopeptide ligases.J. Biol. Chem. 282, 29729–29737

66 Moynagh, P. N. (2009) The Pellino family: IRAK E3 ligases with emerging roles in innateimmune signalling. Trends Immunol. 30, 33–42

67 Jiang, Z., Johnson, H. J., Nie, H., Qin, J., Bird, T. A. and Li, X. (2003) Pellino 1 is requiredfor interleukin-1 (IL-1)-mediated signaling through its interaction with the IL-1receptor-associated kinase 4 (IRAK4)–IRAK–tumor necrosis factor receptor-associatedfactor 6 (TRAF6) complex. J. Biol. Chem. 278, 10952–10956

68 Strelow, A., Kollewe, C. and Wesche, H. (2003) Characterization of Pellino2, a substrate ofIRAK1 and IRAK4. FEBS Lett. 547, 157–161

69 Xiao, H., Qian, W., Staschke, K., Qian, Y., Cui, G., Deng, L., Ehsani, M., Wang, X., Qian,Y. W., Chen, Z. J. et al. (2008) Pellino 3b negatively regulates interleukin-1-inducedTAK1-dependent NF-κB activation. J. Biol. Chem. 283, 14654–14664

70 Boone, D. L., Turer, E. E., Lee, E. G., Ahmad, R. C., Wheeler, M. T., Tsui, C., Hurley, P.,Chien, M., Chai, S., Hitotsumatsu, O. et al. (2004) The ubiquitin-modifying enzymeA20 is required for termination of Toll-like receptor responses. Nat. Immunol. 5,1052–1060

71 Wertz, I. E., O’Rourke, K. M., Zhou, H., Eby, M., Aravind, L., Seshagiri, S., Wu, P.,Wiesmann, C., Baker, R., Boone, D. L. et al. (2004) De-ubiquitination and ubiquitinligase domains of A20 downregulate NF-κB signalling. Nature 430,694–699

72 Hoeller, D. and Dikic, I. (2009) Targeting the ubiquitin system in cancer therapy. Nature458, 438–444

Received 21 April 2009/10 June 2009; accepted 12 June 2009Published on the Internet 29 July 2009, doi:10.1042/BJ20090616


recent insights into the structure of toll-like receptors ... · recent insights into the structure...

Documents