Journal list menu

Volume 579, Issue 15 p. 3330-3335
Minireview
Free Access

Adaptor usage and Toll-like receptor signaling specificity

Aisling Dunne

Corresponding Author

Aisling Dunne

Department of Biochemistry, Trinity College Dublin, Dublin 2, Ireland

Corresponding author. Fax: +353 6772400Search for more papers by this author
Luke A.J. O‧Neill

Luke A.J. O‧Neill

Department of Biochemistry, Trinity College Dublin, Dublin 2, Ireland

Search for more papers by this author
First published: 26 April 2005
Citations: 115

Abstract

It is now well established that Toll-like receptors (TLRs) act as primary sensors of microbial compounds. Details of the molecular mechanisms governing TLR responses are emerging steadily and our understanding of the signaling pathways activated these receptors has improved greatly over the last few years. Differences in adaptor usage, cellular localisation and signaling cascades have been elucidated. In this review we will summarize the current understanding of TLR signaling and its regulation.

1 Introduction

It is now well established that Toll-like receptors (TLRs) represent a primary line of defence against invading pathogens in mammals, plants and insects. Recognition of microbial components by these receptors triggers the initial innate immune response that ultimately leads to inflammatory gene expression and clearance of the infectious agent. Among the structures recognized by TLRs are lipids, carbohydrates, nucleic acids and various proteins, collectively referred to as pathogen associated molecular patterns or PAMPs [1]. The receptors themselves have evolved with the capacity to distinguish different PAMPs, although this remains one of the more complex features of TLR activation (in many cases direct ligand binding has not yet been demonstrated). Eleven TLRs have been identified in humans while 13 can be found in searches of the mouse genome. TLRs 1-9 are conserved between both species. All share a common cytosolic TIR (Toll/interleukin-1 receptor) domain (also found in members of the interleukin-1 (IL-1) receptor family) and an extracellular leucine rich repreat region that mediates recognition of microbial by-products.

TLR4 was the first human TLR to be identified and was for some time considered the prototypic TLR. Differences are now beginning to emerge particularly in the signaling pathways emanating from the various receptors. Some of these differences arise due to the differential utilization of a group of cytosolic proteins referred to as adaptor molecules. They are so called because they function to couple TLRs to downsteam signaling cascades and the activation of transcription factors. These molecules also possess the signature TIR domain which serves as a point of contact between the receptors and the adaptors themselves. Another notable difference results from the differential activation of kinase cascades that serve as the link between membrane proximal events and gene expression in the nucleus. Finally, the expression pattern of TLRs differs both inter-and intra-cellularly. Positive staining of the cell surface with antibodies to specific TLRs reveals that TLR1, TLR2, TLR4, TLR5 and TLR6 are all localized to the plasma membrane whereas TLR3, TLR7, TLR8 and TLR9 are preferentially expressed in intracellular compartments such as endosomes (see Fig. 1 ) [1]. The nature of the ligand recognized by individual TLRs seems to determine their expression pattern. For example, TLR3, TLR7, TLR8 and TLR9 all recognize nucleic acid structures whereas TLR1, TLR2, TLR4, TLR5 and TLR6 generally recognize cell wall components. It should be noted however that surface exposed TLRs are not completely restricted in their expression pattern, for example, TLR2 can localize to phagosomes following exposure to certain ligands [2].

figure image
Cellular localisation of TLRs. TLR1, TLR2, TLR4, TLR5 and TLR6 reside mainly in the plasma membrane whereas those TLRs recognizing nucleic acid derivatives are localized to intracellular compartments such as endosomes.

In this review, we will summarize some of the signaling events triggered following TLR occupation. We will also highlight the differences that are now emerging between TLRs and classify them based on their ability to signal from the cell membrane or intracellular compartments.

2 Plasma membrane associated TLRs

2.1 TLR4

The recognition of lipopolysaccharide (LPS) or endotoxin by TLR4 remains one of the most extensively studied aspects of TLR signaling. The most compelling evidence that LPS was acting through TLR4 came from studies on a strain of mice called the C3H/HeJ mice. These mice have a point mutation in the gene encoding TLR4 that renders them hyporesponsive to LPS challenge [3]. It later emerged that TLR4 is also a receptor for endogenous ligands such as fibrinogen, heat shock proteins (HSP 60 and HSP 70), fibronectins and hyaluronic acid [4]. However, the doses at which these compounds were found to activate TLR4 far exceeded those required for the activation of TLR4 by LPS and it is also suspected that these preparations may have been contaminated with traces of LPS. It therefore remains to be seen if these compounds are true TLR4 ligands. At the molecular level, great progress has been made in understanding the signal transduction pathways activated following TLR4 stimulation. As will be discussed, many features are common to other TLR signaling pathways.

Following exposure of cells to LPS, TLR4 homodimerizes and recruits two sets of adaptor molecules. MyD88 and MyD88 adaptor-like (Mal) (also known as TIRAP) function together to activate the early phase of NF-κB activation which results in the production of pro-inflammatory cytokines, dendritic cell maturation and the up-regulation of co-stimulatory molecules [5-7]. TRIF [8] and TRAM [9, 10] activate the late phase of NF-κB activation and a second pathway leading to the production of the type I interferon, IFN-β. The Mal/MyD88 pathway was the first to be characterized and in many ways resembles that seen following exposure of cells to IL-1. Unlike MyD88, Mal is not required for IL-1 signaling and its role in TLR4 signaling appears to be structural one, acting as a bridge to couple MyD88 to TLR4. MyD88 differs from Mal in that it also possesses an N-terminal death domain. This domain serves to recruit members of the IL-1 receptor associated kinase (IRAK) family of which there are four. IRAK-1 and IRAK-4 possess intrinsic serine/threonine kinase activity [11, 12] whereas IRAK-2 and IRAK-M are catalytically inactive due to the absence of key residues in their respective kinase domains [13, 14]. IRAK-4 is thought to function upstream of IRAK-1 and has been shown to phosphorylate IRAK-1 in vitro [15]. Upon activation, IRAK-1 engages with tumor necrosis factor (TNF)-receptor-associated factor 6 (TRAF6), a member of the TRAF family of proteins [16]. TRAF6 functions in both TNF-receptor and IL-1R/TLR signaling and regulates an array of physiological processes including inflammatory responses and bone metabolism [17, 18]. It has recently been demonstrated that TRAF6 undergoes ubiquitination following cell stimulation and that TRAF6 itself is a ubiquitin ligase [19]. This modification does not prime TRAF6 for degradation but rather activates the protein and subsequent downstream kinase cascades. Once ubiquitinated, TRAF6 activates IKK through a TAK1-TAB1-TAB2 kinase complex [20]. Ea and co-workers [21] have also demonstrated that TAB2 binds to Lys-63 linked polyubiquitin chains through a conserved novel zinc finger domain which is essential for TAK1 activation. Once activated, TAK1 phosphorylates the activation loop of IKKβ, thereby activating the IKK complex and the pleiotropic transcription factor, NF-κB. TAK1 also phosphorylates MKK6 and 7, which in turn activate the p38 and JNK kinase pathways, respectively [20].

TLR4 utilizes the adaptors TRIF and TRAM independently of Mal and MyD88 to initiate the late phase of NF-κB activation and also to induce the expression of IFN-β and other IFN-inducible genes via the transcription factor IRF-3 [22]. TRAM, like Mal, acts as a bridge to couple TRIF to TLR4 and is absolutely required for TLR4 mediated responses. TRAM−/− mice are completely impaired in their responses to LPS while Mal or MyD88 deficient mice do respond albeit with delayed kinetics [23-25]. TRIF is the largest of the adaptor molecules and structurally can be subdivided into three distinct regions. The N-terminus of TRIF contains binding sites for TRAF6 and the non-canonical kinases IKKi and TBK1, the TIR domain presumably tethers TRIF to its binding partner TRAM and finally the C-terminus of TRIF contains a RIP-1 binding site. Both TRAF6 and RIP-1 are thought to mediate TRIF induced NF-κB activation whereas IKKi and TBK1 are required for TRIF-induced IRF-3 activation. A more detailed description of IRF-3 activation is presented below in the context of TLR3 signaling which also requires the adaptor molecule TRIF.

2.2 TLR1, TLR2, TLR5 and TLR6

TLR2 recognizes a vast array of microbial components including lipoproteins from various pathogens, peptidoglycan from gram positive bacteria, glycophosphatidylinositol (GPI) anchors from malaria causing parasites, zymosan from fungi and forms of LPS that are structurally distinct from those recognized by TLR4 [1]. The ability of TLR2 to recognize such a wide range of compounds has been attributed to the ability of TLR2 to heterodimerize with other TLRs, namely TLR1 and TLR6. The association of TLR2 with TLR1 permits the recognition of triacyl lipopeptides whereas TLR2/TLR6 heterodimers recognize diacyl lipopeptides [26, 27].

TLR2 is similar to TLR4 in that it utilizes the adaptors Mal and MyD88 to transmit signals to NF-κB however there is no evidence for a TRIF/TRAM pathway post ligand binding. Biochemical studies with heat-killed Staphylococcus aureus (HKSA) have demonstrated that phosphorylation of TLR2 on specific tyrosine residues facilitates the recruitment of PI3K, an effector of the small G-protein, Rac-1 [28]. The p85 subunit of PI3K binds to phospho-tyrosine residues via a src-homology domain. Phosphorylation of TLR2 by an as yet unidentified kinase induces the formation of an active TLR2–Rac1–PI3K complex. The adaptors Mal and MyD88 are presumably recruited to the receptor at the same time and initiate a signaling pathway leading to IκB degradation and nuclear translocation of active p50/p65 NF-κB heterodimers. Rac-1 participates in the transactivation of NF-κB which occurs following the binding of NF-κB to its consensus response elements within the nucleus [29]. PI3K has also been implicated in the TLR4 signaling pathway, again acting as a mediator of NF-κB transactivation [30]. In this case, PI3K does not bind directly to TLR4 but rather interacts with TLR4 through MyD88.

TLR5 recognizes flagellin, a monomeric constituent of bacterial flagella [31]. A stop codon polymorphism in the ligand-binding domain of TLR5 is associated with susceptibility to legionnaires’ disease [32] thus highlighting the importance of TLR5 in microbial recognition particularly at the mucosal surface.

3 Intracellular TLRs

Among the nucleic acid recognizing TLRs, TLR3, TLR7 and TLR8 have been implicated in the context of viral infection whereas TLR9 is involved in the recognition of both bacterial and viral DNA [33-36]. TLR3 is highly expressed in macrophages, dendritic cells and epithelial cells [37] and induces the expression of IFN-β following exposure to double stranded (d/s) RNA. TLRs 7 and 9 are expressed in a subset of DCs called plasmacytoid DCs or pre-DC2 cells. Exposure of these cells to ssRNA or DNA from viral genomes results in the secretion of large amounts of IFN-α. These receptors also differ in their signaling profiles.

3.1 TLR3

TLR3 functions independently of MyD88 requiring only the adaptor TRIF to transmit signals to the nucleus whereas TLR7, TLR8 and TLR9 all signal via MyD88. TLR3 can activate NF-κB to some extent and this is thought to be mediated by the interaction of RIP1 with TRIF. As described for TLR4, TRIF also recruits the kinases TBK1 and IKKi to activate IRF-3. It was originally believed that TBK1 and IKKi associate with TANK (TRAF family member-asssociated NF-κB activator), however a recent study by Sasai et al. [22] has revealed that it is the TANK homolog, NAP1 (NF-κB-activating kinase (NAK)-associated protein 1) that forms a functional complex with TBK1 and IKKi rather than TANK itself.

A detailed study examining the phosphorylation status of TLR3 has been carried out by Sakar and colleagues [38, 39]. These workers demonstrated that phosphorylation of TLR3 on Tyr759 facilitates the recruitment of PI3K. Mutation of this residue to phenylalanine not only prevented TLR3/PI3K complex formation but also resulted in only a partial phosphorylation of IRF3 and a failure to induce target genes. Phosphorylation on Tyr858 was also found to be critical for full IRF-3 activation. In a simplified scenario, phosphorylated TLR3 recruits both TRIF and PI3K. TRIF in turn recruits the kinase complex NAP1/IKKi /TBK1 leading to partial IRF3 phosphorylation. At the same time, PI3K (presumably acting through AKT) initiates a kinase cascade that results in full IRF3 phosphorylation and transcriptional activation. This pathway mimics the one seen with TLR2 and 4 where PI3K mediates the transactivation of NF-κB (see Fig. 2 ).

figure image
Summary of TLR signaling pathways. A common feature of TLR signaling pathways is the recruitment of specific adaptor molecules into receptor complexes, the initiation of kinase cascades and the induction of gene expression following the activation of transcription factors such as NF-κB and IRF family members.

3.2 TLR7, TLR8 and TLR9

Details of the molecular mechanisms governing induction of IFNs by TLR7, TLR8 and TLR9 are beginning to emerge. Unlike TLR3, these receptors signal in a MyD88 dependent manner to activate IRF-3 related transcription factors. Kawai and colleagues have recently demonstrated that MyD88 and TRAF6 directly associate with IRF7 to induce IFN-α following stimulation of cells with ligands for TLR7 and TLR9 [40]. Silencing of the ubiquitin-conjugating enzyme, UBC13, reduces the activation of IFN-α promoters indicating that the ubiquitin ligase activity of TRAF6 is required for the activation of IRF-7. In a follow up study, Uematsu and coworkers demonstrated that IRAK-1 also participates in the activation of IRF-7 possibly through direct phosphorylation as suggested by in vitro kinase assays [41]. It should be noted however that TBK1 and IKKi have also been shown to phosphorylate IRF-7 [42] and further studies will be required to identify the direct effectors of this transcription factor.

IRF-5 is the latest member of the IRF family found to be activated by TLR ligands [43], however its activation is much more restricted when compared to IRF-3 and IRF-7. Indeed many viruses that activate IRF-3 and IRF-7 fail to activate IRF-5. Recent studies have demonstrated that IRF-5, like IRF-7, is activated by TLR7 and TLR8 [44] and complexes with both MyD88 and TRAF6 [43]. Hence IRF-5 displays characteristics comparable to IRF-7.

4 Negative regulation of TLR signaling

As discussed above, activation of TLRs by pathogens of bacterial or viral origin typically leads to the production of pro-inflammatory mediators such as (IL-1), cyclooxygenase, nitric oxide, adhesion molecules and chemokines. Excessive production of these molecules contributes to the pathogenesis of inflammatory diseases such as rheumatoid arthritis and in the case of bacterial LPS, septic shock. It is therefore critical that checkpoints are put in place to maintain tissue homeostasis and prevent excessive damage to the host. Several molecules have been described recently that function in this particular capacity. For example, IRAK-M, a kinase inactive IRAK family member, prevents the disssociation of IRAK-1 from MyD88 and the subsequent association of IRAK-1 with the downstream adaptor TRAF6. IRAK-M deficient mice are hyper-responsive to LPS challenge and exhibit reduced tolerance to endotoxin when compared to wild-type mice. A splice variant of MyD88, termed MyD88s, also inhibits LPS/IL-1 induced NF-κB activity by preventing the association of IRAK-1 with IRAK-4 and the subsequent activation of IRAK-1. At the receptor level, two proteins termed SIGGIR and T1/ST2 (which are themselves orphan receptors) have been shown to be involved in the negative regulation of TLR signaling. SIGGIR-deficient mice show enhanced responses to IL-1 and TLR ligands and are more susceptible to inflammatory disorders [45] while T1/ST2 deficient mice show defective induction of LPS tolerance [46].

More recently, the conjugation of ubiquitin to intracellular signaling proteins has been described as an important mechanism regulating TLR responses. As mentioned above, TRAF6 undergoes ubiquitination on activation. Jensen and Whitehead [47] first observed that TRAF6 is recycled via deubiquitination following treatment of cells with IL-1. Boone et al. [48] then went on to show that the deubiquitinating enzyme, A20, returns TRAF6 to an unmodified state and in doing so terminates TLR responses. Indeed A20 is induced by a variety and TLR ligands and macrophages deficient in the enzyme show enhanced NF-κB activity following cell stimulation. In addition, A20 knockout mice are more susceptible to endotoxin enduced shock when compared to wild-type mice.

Triad3A, a novel RING-type E3 ubiquitin-ligase was recently shown to promote the degradation of TLR3, TLR4, TLR5 and TLR9 but not TLR2 [49]. Depletion of Triad3A from cells using small interfering RNAs enhanced responses to LPS, flagellin and CpG DNA and led to the increased expression of TLR4 and TLR9. As well as ubiquitinating the TLRs, Triad3A was also found to undergo autoubiqutination similar to TRAF6. It will be interesting to discover if a novel E3 ubiquitin ligase exists to regulate TLR2 levels or whether TLR2 exhibits prolonged stability when compared to other TLRs.

5 Concluding remarks

The repertoire of molecules modulating TLR signaling has expanded greatly since the discovery that these receptors are responsible for the recognition of microbial components. Differences are also beginning to emerge in the signaling pathways activated by different TLRs. Collectively, these differences ensure that the host response is effective in eradicating a plethora of infectious agents given the limited number of receptor molecules.