Review

The Interleukin-1 Receptor/Toll-like Receptor Superfamily: Signal Transduction During Inflammation and Host Defense

+ See all authors and affiliations

Science's STKE  08 Aug 2000:
Vol. 2000, Issue 44, pp. re1
DOI: 10.1126/stke.442000re1

Abstract

The signal transduction pathways activated by the proinflammatory cytokine interleukin-1 (IL-1) have been the focus of much attention because of the important role that IL-1 plays in inflammatory diseases. A number of proteins have been described that participate in the post-receptor activation of the transcription factor NF-κB and stress-activated protein kinases such as p38 mitogen-activated protein kinase (MAPK). It has also emerged that the type I IL-1 receptor (termed IL-1RI) is a member of an expanding receptor superfamily. These related receptors all have sequence similarity in their cytosolic regions. The family includes the Drosophila melanogaster protein Toll, the IL-18 receptor (IL-18R), and the Toll-like receptors TLR-2 and TLR-4, which bind molecules from Gram-positive and Gram-negative bacteria, respectively. Because of the similarity of IL-1RI to Toll, the conserved sequence in the cytosolic region of these proteins has been termed the Toll-IL-1 receptor (TIR) domain. The same proteins activated during signaling by IL-1RI also participate in signaling by IL-18R and TLR-4. The receptor superfamily is evolutionarily conserved; members occur in plants and insects and also function in host defense. The signaling proteins activated are also conserved across species. This receptor superfamily therefore represents an ancient signaling system that is a critical determinant of the innate immune and inflammatory responses.

Introduction

During inflammation--which results from tissue injury, in response to infection, or in autoimmune diseases such as rheumatoid arthritis--cells release inflammatory mediators that give rise to the symptoms of inflammation. These include vascular changes such as increased blood flow, as well as extravasation and activation of leukocytes. During chronic inflammation, tissue remodeling and the production of acute-phase proteins by the liver occur. Cytokines are the key orchestrators of these processes, and the cytokine interleukin-1 (IL-1) plays a central role. Upon binding its receptor on the cell surface, IL-1 induces the expression of a large number of proinflammatory proteins. These include enzymes such as inducible cyclooxygenase (which leads to the production of prostaglandins, themselves important inflammatory mediators), adhesion molecules, chemokines, tissue-degrading enzymes, and acute-phase proteins such as serum amyloid A. The molecular basis for these effects has been studied in great detail, and many of the components in the signaling pathways culminating in changes in gene expression have been worked out. The effects of IL-1 are mediated via the type I IL-1 receptor (IL-1RI) (1). There are two types of IL-1, IL-1α and IL-1β, which both bind IL-1RI. A third protein, called IL-1 receptor antagonist (IL-1RA), also binds IL-1RI but acts as a true receptor antagonist.

IL-1RI has been recognized as part of an IL-1R/TLR (interleukin-1 receptor/Toll-like receptor) superfamily whose members encode various receptors involved in host defense and inflammation (2). These include the receptors for IL-18 (another important proinflammatory cytokine) (3) and for bacterial products such as Gram-negative-derived lipopolysaccharide (LPS) (4) and Gram-positive-derived lipoproteins (5). These receptors also appear to engage signaling pathways similar to those activated by IL-1. This would be predicted because the receptors share sequence similarity in the cytosolic region of each protein. Also in the superfamily are receptors in plants and insects, which again are critical for host responses to infection. Thus, the common theme in the IL-1 system is that of inflammation and defense. The signaling region of these proteins triggers the increased expression of genes important for the elimination of infection and subsequent repair. The overactivation of this system appears to contribute to the pathogenesis of inflammatory diseases such as rheumatoid arthritis and (in the case of bacterial infection) septic shock.

Members of the IL-1R/TLR Superfamily

Defining members of the superfamily

The defining motif of the IL-1R/TLR superfamily is a domain termed the Toll-IL-1R (TIR) domain (Fig. 1A). This domain was named because of the sequence similarity between IL-1RI and the Drosophila melanogaster protein Toll (6). At the time, Toll was thought to participate only in the establishment of dorsoventral polarity in the developing Drosophila embryo (7). It is now known that in the adult fly, Toll is also involved in antifungal responses (8). Although all members of the superfamily have a TIR domain, subtle differences occur between members, the importance of which is as yet unknown.

Fig. 1.

(A) The currently known members of the IL-1R/TLR superfamily. All members contain a conserved cytosolic region termed the TIR domain (Toll-IL-1R domain). The superfamily can be divided into two subgroups. The members of subgroup 1, the immunoglobulin (Ig) subgroup, all contain extracellular Ig domains and include receptors and accessory proteins for IL-1, IL-18, and the orphan receptor T1/ST2. Subgroup 2, the leucine-rich repeat (LRR) subgroup, includes the signaling receptors for LPS (TLR-4) and molecules from Gram-positive bacteria such as peptidoglycan and lipoproteins (TLR-2). Several plant members are also in this subgroup. MyD88 is exclusively cytosolic and is a signaling adapter for IL-1RI, IL-18R, TLR-2, and TLR-4, if not the entire family. Superfamily members may also occur in Caenorhabditis elegans and Streptomyces coelicor (not shown). (B) Three well-conserved regions in the TIR domain, as derived from an alignment of 31 family members (18). Single-letter abbreviations for amino acid residues are as follows: A, Ala; C, Cys; D, Asp; F, Phe; G, Gly; I, Ile; K, Lys; L, Leu; P, Pro; R, Arg; S, Ser; W, Trp; and Y, Tyr. The entire domain spans about 200 amino acids with varying numbers of amino acids separating the boxes, depending on the particular receptor. There are additional regions within the domain that exhibit high similarity. Subconsensus sequences for IL-1R, TLR, and plant members have also been calculated (20), but the three boxes shown are the most highly conserved regions. A Pro Æ His mutation in box 2 of the gene encoding TLR-4 renders TLR-4 unable to signal in C3H/HeJ mice (4). Studies using site-directed mutagenesis indicate that Arg and Asp in box 2, and Phe and Trp in box 3, are essential for IL-1RI function (33). Arg431 in IL-1RI, which lies outside the three conserved boxes, is also required for signaling (33).

IL-1R and TLR subgroups

On the basis of sequence similarities in the extracellular domain, it is possible to divide the IL-1R/TLR superfamily into two major subgroups (Fig. 1A). The first contains the founding member, IL-1RI (1). Members of this subgroup all contain immunoglobulin (Ig) domains extracellularly and include the IL-1 receptor accessory protein (IL-1RAcP) (9), the type II IL-1 receptor (IL-1RII) (10) and its vaccinia virus homolog B15R (11) (which are included here because of their three Ig domains), and the IL-18 receptor (IL-18R) (3) and its accessory protein AcPL (accessory protein-like) (12). Also included are IL-1Rrp2 (13), T1/ST2 (14), and the most recently discovered member, IL-1RAPL (IL-1 receptor accessory protein-like) (15). The ligands for these three receptors have yet to be discovered, but the latter two have important roles in type 2 T helper (TH2) cell responses and in the brain, respectively.

Members of the second subgroup also have the TIR domain, but have a series of leucine-rich repeats (LRRs) instead of Ig domains. The founding member of this group is Toll (6, 7). There are three further Drosophila members, termed 18-Wheeler (16), MstProx (17), and Tehao (18). The putative ligand for Toll is Spätzle (19); ligands for the other two proteins are as yet unknown. Six mammalian TLRs are in this subgroup (20, 21). The ligands for most of these are not known, but TLR-2 is required for responses to molecules from Gram-positive bacteria (e.g., lipoproteins) (4,22-26), and TLR-4 is the long-sought-for signaling component for LPS (4). The plant members of the superfamily are also in this subfamily. Even though they are exclusively cytosolic, they have LRRs and are therefore more closely related to Toll and the TLRs. All are involved in disease resistance in plants (27-29).

Apart from these two subfamilies, there are two other proteins with TIR domains. One of these, termed SIGGIR (single Ig IL-1 receptor relative), is a transmembrane protein but has only one Ig domain (30). Its ligand has not been identified, and its function has yet to be determined. The final known protein with a TIR domain is MyD88 (31, 32). This is an exclusively cytosolic protein that also contains a death domain. It functions as an adapter, triggering activation of IL-1 receptor-associated kinase (IRAK) (31, 32).

The high degree of sequence similarity in each of these groups identifies the TIR domain as a protein motif involved in inflammation and host defense. An alignment of 31 family members reveals three particularly conserved regions in the TIR domain (Fig. 1B) (18). The TIR domain can be defined as a core module present in all IL-1Rs and TLRs, including plant members and MyD88. It spans about 200 amino acids, with varying degrees of sequence similarity among superfamily members (20). Three particular boxes can be identified, however, which are highly conserved among family members (Fig. 1B) (18). Box 1 can be considered the signature sequence of the family; boxes 2 and 3 contain amino acids critical for signaling. A structural model of the TIR domain has yet to be provided; however, on the basis of its conserved amino acid sequence, the TIR domain structure might be similar to that of the bacterial chemotaxis regulator CheY (20). From this prediction it has been suggested that the TIR domain may act as a conformational trigger rather than as a molecular scaffold (20). Also of note is the occurrence of a TIR domain-containing protein in Streptomyces coelicor (18). This would suggest that the TIR domain was not only present in the common unicellular ancestor of plants and animals, but probably arose before the divergence of eukaryotes and prokaryotes. If this is correct, the TIR domain would be one of the first signaling domains to have evolved, attesting to its efficacy as a signaling cassette. Therefore, its role in defense may have been essential for the evolution of multicellular organisms.

IL-1RI-like Members

IL-1RI

Along with the TIR domain, the hallmark of this subfamily is the presence of three extracellular Ig domains. In IL-1RI, six amino acids (conserved in the sequences of those receptors whose genes had been cloned at that time: human, mouse, and chicken IL-1RIs and Toll) are essential for signaling: Arg431, Lys515, Arg518, Phe513, Trp514, and Tyr519 (33). In addition, Pro521 is also required for maximal signaling capacity. Phe513 and Trp514 are in the conserved box 3 in the TIR domain of IL-1RI. The precise role of each of these amino acids is still undetermined. Presumably they are required for correct recognition of the signaling domain by signaling proteins. Some may also be involved in trafficking of the receptor (34). It is possible that the association of this region among IL-1RI, IL-1RAcP, and MyD88 [which are present in a complex after IL-1 binding (31)] is required for signaling. Such a trimerization between TIR domains in IL-1RI, IL-1RAcP, and MyD88 also occurs in the p55 tumor necrosis factor (TNF) receptor system, in which trimeric TNF brings together three death domains in each p55 TNF receptor trimer. Receptor trimerization may be induced by both IL-1 and TNF, which is interesting because although the signaling proteins associated with these receptors differ, ultimately both cytokines activate similar pathways. However, there are also differences between IL-1 and TNF, in that TNF is toxic to many cell types. This may be because TNF can recruit the Fas-associated death domain (FADD) protein, which triggers apoptosis (35).

One amino acid in the cytosolic region of IL-1RI, Tyr479, has been suggested to interact directly with a signaling protein. The p85 subunit of phosphatidylinositol 3-kinase (PI 3-kinase) may associate with this tyrosine when phosphorylated, thereby recruiting p110, the catalytic subunit of PI 3-kinase, to the membrane (36). Tyr479 occurs in a motif recognized by p85, Tyr-Glu-X-Met. The p85 subunit has been shown to associate with IL-1RAcP rather than IL-1RI (37); however, IL-1RAcP does not contain a homologous tyrosine (9). In that study, the role of PI 3-kinase was to induce the phosphorylation of the p65 subunit of the transcription factor NF-κB, thereby promoting its transactivating capabilities.

In those systems in which it has been tested, IL-1RI is clearly important for IL-1 effects. Antibodies to IL-1RI block IL-1 responses in several cell systems. In IL-1RI-deficient mice, such IL-1 responses as increased IL-6 production and fever are impaired (38). Also impaired were the acute-phase response, delayed-type hypersensitivity, and the ability to combat infection by Listeria (38). These results all indicate the importance of IL-1RI for inflammation and infection.

Progress has also been made in our understanding of the structural basis for IL-1RI recognition by IL-1. The crystal structures of IL-1RI in a complex with IL-1β or IL-1RA have been solved to 2.7 Å resolution (39, 40). The structures revealed that Ig domains 1 and 2 are tightly linked, whereas domain 3 is separate and connected by a flexible linker. IL-1β contacts the receptor via two regions, one of these interacting with the first two domains and the other with domain 3. The binding of IL-1β actually induces a conformational change, which results in domain 3 moving such that the receptor wraps around the ligand (39). IL-1RA, unlike IL-1β, does not induce movement in domain 3 (40). This may be because IL-1RA only contacts IL-1RI at domains 1 and 2. Such a movement in domain 3 may allow interaction between IL-1RI and IL-1RAcP, which then triggers the signal. Two hydrophilic domains have been identified in IL-1RAcP that are critical for signaling and hence possible interaction with IL-1RI (41). A full account of the molecular changes that occur upon binding IL-1 in IL-1RI and IL-1RAcP extracellularly, and indeed intracellularly, will await the full elucidation of all the structures involved, both in the absence and presence of ligand.

IL-1RII

The existence of a second IL-1 receptor, IL-1RII (10), was suggested from cross-linking studies, and its gene sequence was later determined. It has three Ig domains and exhibits a high degree of sequence similarity to IL-1RI. Intracellularly, however, it proved very different, having only a short cytoplasmic tail of 29 amino acids and containing no TIR domain. Its expression is more restricted than that of IL-1RI, with high expression on lymphoid and myeloid cells (10). Certain cell types express both receptors, and even in cells in which IL-1RII appears to predominate, small amounts of IL-1RI can be detected (42). IL-1RII is shed from cells and acts as a decoy receptor, preventing IL-1 from binding to IL-1RI (43). This provides another means (along with IL-1RA) to limit IL-1 action. The shedding of IL-1RII from neutrophils can be induced by glucocorticoids and IL-4, which may account for part of their anti-inflammatory effects. IL-1RII also interacts with, and thereby limits the availability of, IL-1RAcP (44). This may prevent IL-1RI from recruiting IL-1RAcP, thus providing another means of inhibiting IL-1 action. IL-1RII has a very low affinity for IL-1RA (45), which is consistent with the role of both proteins as inhibitors of IL-1 binding to IL-1RI.

B15R

A search of the GenBank database with IL-1RII revealed an IL-1RII homolog in several viruses of the pox family, most notably vaccinia virus. The vaccinia B15R gene encodes a protein very similar to IL-1RII (11, 46). Deletion of B15R gives rise to a much more virulent virus. This indicates that the role of B15R in vaccinia is to limit the damage to the host caused by the overproduction of IL-1 during infection. The virus lacking B15R is much more pyrogenic than wild-type virus (11), providing further proof that IL-1 is a key pyrogenic cytokine, and, in the case of viral infection, probably the most important endogenous pyrogen.

IL-1RAcP

The discovery of IL-1RI and IL-1RII appears to account for all of the binding capacity of cells for IL-1, at least in the cell types studied. However, IL-1RAcP is also required for signaling, acting as the second chain of the IL-1R complex (9). Despite having three Ig domains, IL-1RAcP does not bind IL-1 (9, 47). IL-1RAcP is widely expressed, and on the whole its expression correlates with that for IL-1RI. The exception to this is in brain, where in rat the two proteins appear not to colocalize (48). IL-1RAcP also appears to increase the affinity of IL-1RI for ligand by a factor of about 5 (9).

It is still not clear whether IL-1RI and IL-1RAcP are in a complex before the addition of ligand. IL-1 may cause a conformational change in IL-1RI that recruits IL-1RAcP to the complex. Another more likely scenario is that the addition of ligand increases the affinity of IL-1RI for IL-1RAcP. From the structural studies described above, it is likely that the Ig domains are involved in the assembly of IL-1RI and IL-1RAcP into a heterodimer. This has important consequences for signaling studies in which chimeras are constructed between extracellular IL-1RI and the intracellular region of IL-1R/TLR superfamily members whose ligands are unknown (described below). Such chimeras would recruit IL-1RAcP, which may not occur with intact receptors.

Of relevance to this point is whether IL-1RI and IL-1RAcP interact with the same signaling components. There are subtle differences in their respective TIR domains, which might lead to the recruitment of different proteins during signaling. It has been suggested from immunoprecipitation studies that IRAK is recruited via IL-1RAcP, with IRAK-2 being recruited to IL-1RI (49, 50). Both IRAKs, however, also appear to be recruited via MyD88 (31, 32, 50). Whether IL-1RAcP has a subtly different role from IL-1RI in signaling will await further clarification. A plausible scenario, however, is that the interaction between IL-1RI and IL-1RAcP creates a surface that allows MyD88 to be recruited via its TIR domain.

IL-18R

Two other IL-1R-related proteins, IL-1Rrp1 and IL-1Rrp2, have been identified (13, 51). IL-1Rrp1 is the receptor for IL-18 (3) and has therefore been renamed IL-18R. IL-1Rrp2 is still without a ligand and has no known function.

The cytokine IL-18 was first described as interferon-γ inducing factor (52). It is a key inducer of TH1 cell development and function, activating both TH1 cells and natural killer (NK) cells. The processing of the precursor form of IL-18, like that of IL-1β, is catalyzed by IL-1 converting enzyme (caspase 1). IL-18 also activates signaling pathways similar to those activated by IL-1 (see below).

IL-18R is detectable in lung, spleen, heart, testis, peripheral blood T cells, and NK cells (3). It is absent from brain, although again this may be a detection problem. A role for IL-18R in TH1 cells is borne out from studies with IL-18R knockout mice (53). Such mice have TH1 cells that are unresponsive to IL-18, have defective cytolytic responses in their NK cells, and show defects in TH1 cell development.

The accessory protein for IL-18R was identified through its sequence similarity to IL-1RAcP and was named AcPL (accessory protein-like) (12). Its expression appears to be more restricted than that of IL-18R, in that it is not detectable in heart or testis. It seems likely at this stage that AcPL will function as IL-1RAcP does in IL-1 signaling.

T1/ST2

A gene named T1 or ST2 was cloned as a delayed early response gene that responds to proliferative signals (54, 55). It encodes a soluble protein similar to the extracellular domain of IL-1RI. A transmembrane form was subsequently described (56). The similarity of T1/ST2 to IL-1RI in the transmembrane form extends throughout the molecule, and it contains Ig domains extracellularly and a TIR domain intracellularly. Soluble and membrane-bound versions are expressed in different tissues from different promoters (57). The soluble form is produced mainly by fibroblasts and possibly mast cells, whereas the membrane-bound form is predominantly expressed on hematopoietic cells (57). Hematopoietic cells can, however, be induced to express the soluble form upon stimulation with proinflammatory agents including IL-1. Expression of T1/ST2 is increased during inflammation in vivo (58). It may be that the putative ligand, when bound to the receptor, prevents proliferation, and soluble receptor might block this effect. This may explain why high concentrations of soluble T1/ST2 correlate with proliferation.

T1/ST2 is still an orphan receptor, but two putative ligands have been described. The first protein was purified, and the gene that encodes it has been cloned. However, the expressed gene product was unable to activate NF-κB (59). The second protein was partially purified and was found to activate p38 mitogen-activated protein kinase (MAPK), but not NF-κB (60). This was an ~18-kD protein with a possible precursor of 32 kD; these sizes are similar to those of the mature and precursor forms of IL-1β. The TIR domain in T1/ST2 is capable of activating NF-κB, as demonstrated in studies with chimeras between extracellular IL-1RI and intracellular T1/ST2 (14). However, such a chimera would recruit IL-1RAcP, which may mediate the NF-κB signal by recruiting IRAK. It is possible that T1/ST2 has its own specific accessory protein, yet to be described, which does not recruit IRAK and thus does not activate NF-κB.

Differential display analysis of TH1 or TH2 cells revealed T1/ST2 to be strongly expressed on TH2 cells (61). T1/ST2 was found to play a critical role in TH2 cell function, in that a neutralizing antibody to T1/ST2 attenuated TH2-driven responses in vivo. Both soluble T1/ST2 and the antibody to T1/ST2 were particularly effective in limiting airway hyperreactivity in a model of allergic asthma (62). These data indicate that T1/ST2 is critical for TH2 cell function. Conflicting data have emerged from two separate studies on T1/ST2-deficient mice; in one case TH2 cell development and function was normal (63), whereas in the other it was impaired (64). This may be because different challenges were used on the mice to test their phenotype.

The uncovering of the ligand for T1/ST2 will clearly enhance our understanding of the biology of TH1 and TH2 cells. Furthermore, from studies on IL-18R, it is clear that IL-18R and T1/ST2 have divergent roles in TH subtype development and activation. However, the presence of a TIR domain in both receptors implies that the signaling pathways in TH1 and TH2 cells during activation are initially similar. Alternatively, it is possible that the IL-18 effect requires NF-κB activation, whereas that of T1/ST2 does not. In support of this, a recent study has indicated that NF-κB is required for TH1 but not TH2 cell function (65).

Newer members: SIGIRR and IL-1RAPL

Two new members of this subfamily have been described. SIGIRR (single Ig IL-1R-related molecule) is ubiquitously expressed and has a TIR domain and one extracellular Ig domain (30). A chimera between the IL-1RI extracellular domain and SIGIRR intracellular domain was unable to signal NF-κB in response to IL-1. The basis for this lack of effect is not clear, and the function of SIGIRR has yet to be determined. Furthermore, the ligand for SIGIRR is still unknown.

The second new member, termed IL-1RAPL (IL-1 receptor accessory protein-like), was found in a study of the genetic basis of a nonspecific form of X-linked mental retardation (15). The gene responsible for this condition was shown to encode a protein most similar to IL-1RAcP, particularly in its TIR domain. In the disease, nonoverlapping mutations and a nonsense mutation in the gene were found in patients with cognitive impairment. The normal role of the gene is likely to be in memory and learning, where IL-1 has been shown to be important (66, 67). IL-1RAPL is expressed in brain regions associated with memory and learning (particularly the dentate gyrus in the hippocampus). This raises the possibility that in brain, some of the effects of IL-1 (possibly the physiological effects with regard to memory) are mediated by this receptor (or, if IL-1RAPL is an accessory protein, by the receptor for IL-1RAPL). The cognate ligand has not been described.

Toll-like Members

The sequence similarity between IL-1RI and Drosophila Toll provided the first description of the TIR domain (6), although the term TIR has only recently been coined. Toll-like members differ from those in the IL-1RI subgroup in that they contain LRRs instead of Ig domains. They include the mammalian TLRs (of which six have been described, termed TLR-1 through TLR-6), other insect members (including 18-Wheeler), and proteins involved in disease resistance in plants (notably N protein, L6, RPP1, and RPP5).

Toll

In the developing Drosophila embryo, Toll drives the establishment of dorsoventral polarity. The ligand that activates Toll in this process is suggested to be Spätzle (19), although direct binding of Spätzle to Toll has not been described. Spätzle has a cysteine knot structure, which also occurs in members of the nerve growth factor (NGF) family (68). The structure of Toll has been modeled and it resembles that of platelet glycoprotein 1b (69), the receptor for Von Willebrand factor. In a further similarity, Spätzle, like Von Willebrand factor, is generated as a result of a protease cascade activated in Drosophila during development.

The reported similarity between Toll and IL-1RI was intriguing, because at the time the only known role for Toll was in development. It then emerged that there were similar signaling pathways activated by both receptors. In the case of Toll, it activates a protein kinase termed Pelle through interaction with Tube, an adapter protein. Pelle then leads to the phosphorylation of Cactus, which dissociates from the transcription factor Dorsal and translocates to the nucleus, where it increases the expression of target genes involved in dorsoventral polarity (Fig. 2). In an analogous signaling mechanism, IL-1 activates the protein kinase IRAK, which is related to Pelle; MyD88 acts as the adapter but is not related to Tube. IRAK functions in a protein kinase cascade that leads to phosphorylation of IκB (inhibitor of κB). When phosphorylated, IκB dissociates from NF-κB, which in turn translocates to the nucleus and increases inflammatory and immune gene expression. IκB is a homolog of Cactus, and NF-κB comprises Rel family members homologous to Dorsal (Fig. 2). This signaling pathway, starting with the TIR domain, is thus being used in two different contexts to alter gene expression. It is possible that the TIR domain may have evolved to activate Rel family members.

Fig. 2.

Conserved mechanisms for signaling in the IL-1R/TLR superfamily. The signaling pathways that lead to the activation of Rel family proteins, through Spätzle (in Drosophila melanogaster), LPS, IL-1, or IL-18, are quite similar. Spätzle is the ligand for Toll and is generated as a result of a protease cascade. It is not clear whether LPS is the direct ligand for TLR-4 (or CD14 in a complex with MD-2), or whether LPS activates a protease cascade through a pattern recognition receptor that produces a ligand for TLR-4. Inflammation and stimulation of the immune system lead to increased expression of IL-1 and IL-18, which then bind and activate IL-1RI and IL-18R, respectively. MyD88 and Tube are adapter proteins that do not share sequence similarity. There are high sequence similarities between Pelle and IRAK, between dTRAF and TRAF-6, between Drosophila and mammalian ECSIT, and between Cactus and IκB. Dorsal and Dif are both Rel family members. NF-κB-inducing kinase (NIK) provides an alternative route to the IκB kinase complex in mammals. NIK is downstream of two further kinases, TAB-1 and TAK-1 (not shown). A role for NIK in NF-κB activation by IL-1 has recently been disputed. Other signaling pathways in mammals activate the stress kinase p38 (not shown). See text for further details.

In adult Drosophila, Toll has an additional role in the response to infection. Toll activates another Rel or NF-κB family member, Dif, which regulates the expression of antifungal peptides such as drosomycin (Fig. 2) (8, 70). The role of Toll and IL-1 in host defense is therefore conserved between insects and mammals. Such conservation can also be seen in the human TLRs.

In terms of the ligand for Toll, in the adult it was postulated that a Spätzle-like protein would have this role. Indeed, Spätzle in adult flies is cleaved during immune challenge with Escherichia coli (71). The active ligand was shown to be present in mutant flies lacking the serine protease inhibitor (serpin) Spn43Ac. Expression of Spn43Ac increased in wild-type flies during infection, indicating the presence of a negative feedback loop. These results indicate that Toll is not the receptor for microbial products; instead, an unknown upstream receptor activates a protease cascade, kept in check by the serpin Spn43Ac, which blocks the protease that results in Spätzle production.

18-Wheeler, MstProx, and Tehao

Three other proteins homologous to Toll have been described in Drosophila: 18-Wheeler, MstProx, and Tehao (16, 17, 72). Like Toll, 18-Wheeler functions in development and then contributes to the response to infection in adults. The 18-Wheeler protein signals increased production of antibacterial peptides, such as attacin, by activating Dif (72, 73). Thus, in Drosophila, Toll responds to fungal pathogens, whereas 18-Wheeler responds to bacteria (8, 72, 73). The roles of MstProx and Tehao are as yet unknown.

Toll homologs have also been described in other insects. The dipteran Clognia albipunctata and the beetle Tribolium castaneum both have Tolls (termed clToll and trToll, respectively) that regulate dorsoventral polarity (74). They probably also have a role in host defense.

Mammalian Toll-like receptors: TLRs 1 to 6

Six TLRs have been found in mammals (20, 21). TLRs are more closely related to Toll than to IL-1R because of the extracellular LRR structure.

Soon after the report on TLRs 1 to 5 (20), a human Toll (termed hToll) was described that was identical to TLR-4 (75). When expressed as a fusion protein with CD2, which traffics TLR-4 to the membrane, TLR-4 increases the expression of B7.1 in antigen-presenting cells, and of inflammatory cytokines (including IL-1). B7.1 is the ligand for CD28 that is expressed on T cells and is required for T cell activation. These responses are part of the initial innate response to infection. It therefore appears that the TLR family, using the TIR domain to signal, is a key participant in innate immunity (and therefore inflammation) that is conserved across species.

TLR-1 is ubiquitously expressed and more abundant than the other TLRs (20, 21). TLR-2 is expressed in brain, heart, muscle, and lung. TLR-3 is expressed in a pattern similar to that of TLR-2, but with high expression in the pancreas and placenta. In contrast, TLR-4 and TLR-5 are less abundant. Large amounts of TLR-4 are expressed in placenta and peripheral blood leukocytes, whereas ovaries and monocytes express large amounts of TLR-5. TLR-6 is expressed in spleen, thymus, ovary, and lung (21). In addition, the size of mRNA transcripts for certain TLRs varies in tissues (20). None of these expression studies examined the presence of TLR protein.

A protein termed RP105 can also be included in the TLR subfamily. The RP105 sequence is similar to that of the extracellular domain of Toll (76); however, it has a short cytoplasmic sequence that lacks the TIR domain. RP105 is expressed predominantly on B cells, and when cross-linked, RP105 drives proliferation (77).

TLR-2 and TLR-4 as signaling receptors for microbial products

The most immediate question concerning TLRs is the nature of their ligands. There may be Spätzle-like ligands in mammals, but these have yet to be described. The demonstration that TLR-4 could signal required its expression as a membrane-localized fusion protein (75). An important breakthrough came from work on LPS. It was known that the protein CD14, which occurs on the surface of macrophages, was required for LPS action (78), although because CD14 was a glycophosphatidylinositol-anchored protein, it was suspected that a second chain was needed to generate the signal. Evidence that this second chain might be TLR-2 came from experiments that showed full reconstitution of LPS responses in 293 cells cotransfected with CD14 and TLR2 (79, 80) in the presence of serum (79) or LPS-binding protein (80).

However, TLR-4 was shown to be required for LPS signaling in studies from LPS-resistant mice. The mutation in the C3H/HeJ mouse, which is insensitive to LPS, had long been sought and was found to be in the gene encoding TLR-4 (4). A single point mutation (Pro712 to His) in box 2 of the TIR domain was found in the gene for TLR-4 in these mice, which rendered TLR-4 unable to signal. Similarly, in the C57BL/10ScCr mouse, which is also resistant to LPS, there is a null mutation in the gene for TLR-4 (4). TLR-2 was intact in both these strains of mice. Work involving crosses between mouse strains normal and refractory to LPS demonstrated that the Pro Æ His point mutation exerts a dominant negative effect on LPS signaling (4, 81). The mutant may act as a decoy that competes for binding of LPS, or alternatively may disrupt the formation of signaling complexes. However, another mutation has been reported in the C3H/HeJ mouse, in the gene encoding the guanosine triphosphatase (GTPase) Ran (82). Retroviral transfer of wild-type Ran cDNA rendered the mice sensitive to LPS, and adenoviral transfer of mutant Ran into normal mice made them less responsive. Ran may therefore also be important for LPS responses. The relationship of Ran to TLR-4 signaling remains unclear.

Further confusion as to whether TLR-2 or TLR-4 is the receptor for LPS in mammals came from studies in Chinese hamsters. The gene encoding TLR-2 is mutated in Chinese hamsters, leading to a truncation of the protein (83); TLR-4 is normal, however, and mediates the effect of LPS. It is therefore possible that the functions of TLR-2 are specific to humans and those of TLR-4 are specific to mouse and hamster. Both C3H/HeJ and C57BL/10ScCr mice are immunologically normal and are impaired only in their responses to Gram-negative bacteria; this implies that mammalian TLRs, like Drosophila TLRs, may discriminate between pathogens. TLR-2 but not TLR-4 functions in the response to Gram-positive bacteria or their products, such as lipoproteins, lipoteichoic acid, and peptidoglycan (22-26).

Two recent studies (5, 84) show clearly that TLR-2 is required for responses to Gram-positive bacteria and yeast, whereas TLR-4 mediates the response to Gram-negative bacteria and LPS. TLR-2 is recruited to phagosomes that contain yeast. The Pro681 to His mutation results in a TLR-2 that does not internalize or induce responses. The effect of LPS is not impaired, however. On the other hand, an identical mutation at Pro712 in TLR-4 [i.e., the same as that in C3H/HeJ mice (4)] abrogated responses to LPS but not to Gram-positive bacteria. Finally, dominant negative MyD88 blocked the response to TLR-2, providing further evidence for its importance in TIR domain signaling. These studies appear to resolve, in mice at least, the roles of TLR-2 and TLR-4. It is likely that the roles of these receptors in humans will be the same. The early observation that TLR-2 was important for the effect of LPS (70, 71), although it provided the first indication for a role for TLRs in LPS action, may have resulted from overexpression of TLR-2 and thus may not reflect a physiological role for TLR-2.

The ability of different mammalian TLRs to recognize different microbial products is analogous to the situation in flies, where Toll is involved in antifungal responses and 18-Wheeler responds to bacteria (70, 73). The different receptors also induce the expression of different genes and hence may activate differing signaling pathways. The roles of TLR-1, TLR-3, TLR-5, and TLR-6 are not well understood. Chimeras of the extracellular domain of Fas fused to the transmembrane and cytosolic domain of TLR-2, TLR-5, or TLR-6 activate NF-κB (85). A chimera of the extracellular domain of IL-1RI with the intracellular domain of TLR-1 was inactive, however (14). Apart from TLR-3, all TLRs contain the conserved Pro residue shown to be important for recruitment of TLR-2 to phagosomes (5). TLR-3 may therefore have a somewhat different role. Expression of TLR-4 increases in cardiomyocytes during ischemia (86). This raises the intriguing possibility that TLRs may also participate in inflammation in the absence of infection, perhaps being activated by endogenous ligands.

As stated above, the putative ligand for Toll is Spätzle, which, during an immune response in the fly, is activated by a protease cascade (71). The situation could be similar in humans, because no studies have demonstrated direct binding of LPS or microbial products from Gram-positive bacteria to TLRs. A more upstream pattern recognition receptor may therefore bind microbial products, activate a protease cascade, and thus generate ligands for TLRs. This is further suggested from the well-known ability of LPS to drive coagulation and the complement cascade, and from the model of the structure of Toll, which resembles platelet glycoprotein Ib (69). Furthermore, coagulation products such as thrombin activate NF-κB (87, 88), an important signal triggered by LPS. Whatever the protease cascade, it would most likely be membrane associated, because the only soluble factors so far shown to be required for LPS responsiveness in cell culture are LPS binding protein and CD14 (89).

Plant members of the IL-1R/TLR superfamily

Members of the IL-1R/TLR superfamily are also present in plants. The first to be described was the tobacco mosaic virus resistance gene product N protein (27, 90). N protein is more related to Toll than to IL-1R, in that along with the TIR domain it also has four LRRs. The role of N protein in tobacco is to activate an enhanced response in the plant during infection, which leads to necrosis of infected plant tissue, thereby halting pathogen growth and spread. Several other plant proteins with similarity to TLRs are involved in disease resistance, including L6 in flax and Rpp5 in Arabidopsis (28, 91). Comparison of these plant members with the vertebrate receptors indicates that the Toll subgroup of the family probably arose first, with Ig domains replacing LRRs in the IL-1RI subgroup at a later stage. The ligands for the plant receptors are wholly unknown but are most likely cytosolic.

Apart from receptors, there are also plant homologs of IRAK that are involved in disease resistance, including Pto and Fen in tomato. This implies that signaling components are also conserved between species. NF-κB family members have yet to be described in plants, although p38 MAPKs are strongly conserved in all eukaryotic species so far examined (92), which implies that, like IRAK, Pto and the other plant homologs may be involved in activation of p38. The IL-1-TLR receptor system and certain of the signaling proteins it activates appear to be pan-genomic.

Signal Transduction

The goal of research into IL-1 signaling has been to detail the events triggered by IL-1RI that culminate in increased expression of target genes, most of which encode proteins involved in immunity and inflammation. Two of the key signaling molecules activated by IL-1 are the transcription factor NF-κB and p38 MAPK. Both are also activated by IL-18 (acting through IL-18R) and LPS (acting through TLR-4) (93, 94). This indicates the conserved nature of the signaling pathways elicited by the TIR domain. IL-1 actually activates four protein kinase cascades. The best characterized involves NF-κB, and the three others activate the MAPKs p38, p42/p44, and JNK (c-Jun NH2-terminal kinase), respectively. The activation of NF-κB leads to increased transcription of target genes. The p42/p44 MAPK also appears to regulate transcription, whereas p38 and JNK promote the stabilization of induced mRNA.

NF-κB

Remarkable progress has been made in our understanding of NF-κB (95), including the post-receptor pathway activated by IL-1 leading to NF-κB. Because of its importance in inflammatory gene expression, NF-κB has become the focus of much attention generally. Many other agents apart from IL-1 activate NF-κB, including LPS, TNF, and various cellular stresses such as ultraviolet irradiation. NF-κB participates in stress responses in cells. Activation of NF-κB by IL-1 begins with formation of a complex among MyD88, IRAK, and IRAK-2 (31, 32, 50). MyD88 was first described as a protein expressed during myeloid differentiation (96). Its COOH-terminal domain is similar to part of Toll, and the NH2-terminus of MyD88 contains a death domain (31, 32). MyD88 dimerizes through both the death and TIR domains. MyD88 also associates with the activated IL-1R complex and apparently recruits IRAK to the complex (31, 32). However, there is also evidence that IL-1RI can recruit IRAK-2 directly and that IL-1RAcP can recruit IRAK (49, 50). Although IRAK is a kinase, the role of its catalytic function is uncertain (97, 98). Upon activation, IRAK becomes phosphorylated. However, if the IRAK gene is mutated such that it is catalytically inactive, it functions normally in terms of signal transduction, which implies that the kinase activity is not required (97, 98). Like MyD88, IRAK contains a death domain, but its function is unknown. Despite these uncertainties, the role of IRAK is to recruit TNF receptor-associated factor 6 (TRAF-6), a member of the TRAF family of proteins. Downstream of TRAF-6 lie two kinases, TAB-1 and TAK-1 (99). These in turn activate NF-κB-inducing kinase (NIK). NIK is a point of convergence for IL-1 and TNF signaling. Its role is apparently to activate the signalsome, the multiprotein complex that contains the IκB kinases (IKKs). However, a role for NIK has recently been disputed, because analysis of the alymphoplasia (aly) mutation of mice showed the genetic defect to be a point mutation in NIK that resulted in the mice failing to develop lymphoid organs. However, NF-κB activation by TNF was normal (100). IKK-1 and IKK-2 each can phosphorylate IκBα on two key serines, Ser32 and Ser36, leading to the release of the NF-κB heterodimer. An important protein in the signalsome is the scaffold protein NEMO (NF-κB essential modulator) (101). This complex set of protein-protein interactions culminates in NF-κB activation (Fig. 2).

Another route to the signalsome involves the kinase MEKK1 (102). A novel protein linking TRAF-6 to MEKK1 has been described and is termed ECSIT (evolutionarily conserved signaling intermediate in Toll pathways) (103). As the name suggests, there is a Drosophila homolog.

Similar proteins are involved in LPS and IL-18 signaling (93, 94). However, LPS only requires IKK-2 for NF-κB activation (104). From transgenic studies, it appears that IKK-1 is involved in development, whereas IKK-2 is required for inflammation (105). It is therefore possible that the IL-1RI/TLR system will culminate in IKK-2 activation in the NF-κB pathway.

Mitogen-activated protein kinases

Details of the upstream regulation of the MAPKs p38, p42/p44, and JNK by IL-1 are not completely worked out. All three are activated by IL-1, IL-18, and TLRs, although to different extents in different cell types. Treatment of cells with IL-1 activates Raf-1, which lies upstream of MAPK kinase (106), as well as MAPK kinase itself (107), which lies upstream of p42/p44; MKK3 and MKK6 (108), which lie upstream of p38; and MKK7 (109), which lies upstream of JNK. Details of how the kinase cascades are activated are less clear. There may be a role for small GTPases, however. IL-1 causes increased GTP binding and hydrolysis in membranes (110), and direct evidence for Rho and Rac1 activation has been presented (111, 112). In addition, dominant negative mutants of Rac1 and Cdc42 block activation of JNK and p38 MAPK in response to IL-1 (113, 114), although in the case of JNK this has been disputed (115). Most recently, Ras has been shown to be critical for p38 activation, with Rap having an antagonistic effect (116).

MAPKAP (MAPK activating protein) kinase 2 is activated by p38 and phosphorylates heat shock protein 27 (hsp27) (117). Hsp27 is phosphorylated in IL-1-treated cells, but the physiological importance of this event is uncertain. The use of inhibitors of p38 and MEK indicates that the role of p38 is largely posttranscriptional for some genes (e.g., the gene encoding inducible cyclooxygenase) (118), whereas MEK appears to be involved in transcription of genes such as that encoding IL-2 (119).

Gene Knockout Studies

Mice deficient in MyD88, IRAK, or TRAF-6 have indicated the importance of these signal transducers for the actions of IL-1, IL-18, and LPS (120-124) (Table 1). In MyD88-deficient mice, responses to IL-1, IL-18, or LPS in vivo are impaired (120, 121). In cultured cells deficient in MyD88, neither IL-1 nor IL-18 activates NF-κB or JNK (120). However, both signals are activated by LPS, although the response is delayed (121). This implies that LPS can use signaling molecules other than MyD88 in vitro.

Table 1.

Phenotypes of MyD88-, IRAK-, and TRAF-6-deficient mice.

Cells from IRAK-deficient mice are unresponsive to IL-18 in terms of NF-κB and JNK activation, and in animal studies, responses to IL-1 and IL-18 are absent (122, 123). LPS was not tested in these mice. In TRAF-6-deficient mice, activation of NF-κB and JNK in response to IL-1 or LPS is impaired. These mice are also osteopetrotic, which implies that TRAF-6 has an important role in bone turnover (124).

Conclusions

The IL-1R/TLR superfamily has emerged as an important determinant for inflammation and host defense. The conserved TIR domain, which defines the family, is evolutionarily ancient, possibly arising in prokaryotes but certainly emerging in the common unicellular ancestor to plants and animals. In mammals, its importance can be seen in the critical role played by IL-1 and IL-18 in inflammation and by TLR-2 and TLR-4 in the host response to bacteria. The signaling pathways activated by the superfamily involve the transcription factor NF-κB and stress-activated protein kinases p38 and JNK. The NF-κB system is in all animals, with p38 and JNK also occurring in plants.

This area is likely to yield much new information on the host response to injury and infection. Immediate questions concern the identity of ligands for orphan members of the superfamily--most notably T1/ST2 (which appears to play a key role in allergic airway disease) and the TLRs--and whether there are subtleties in the signaling pathways activated by superfamily members. Furthermore, the role of TLRs in noninfectious inflammation is likely to be important. Further understanding of TLRs and their signaling mechanisms may facilitate novel therapies for both inflammatory and infectious diseases.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
View Abstract

Related Content

Navigate This Article