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

See allHide authors and affiliations

Science's STKE  25 Feb 2003:
Vol. 2003, Issue 171, pp. re3
DOI: 10.1126/stke.2003.171.re3


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 nuclear factor κB (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 (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 10 Toll-like receptors (TLRs), TLR-1 to TLR-10, which bind to microbial products, activating host defense responses. 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 other receptors with TIR domains. The receptor superfamily is evolutionarily conserved; members also occur in plants and insects, where they also function in host defense. The signaling proteins that are activated are also conserved across species. Differences are, however, starting to emerge in signaling pathways activated by different receptors. This receptor superfamily, therefore, represents an ancient signaling system that is a critical determinant of the innate immune and inflammatory responses.


During inflammation, which results from tissue injury, infection, or autoimmune diseases, such as rheumatoid arthritis (RA), cells release inflammatory mediators that give rise to the symptoms of inflammation. These symptoms include vascular changes, such as increased blood flow, and 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 stimulates 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 that culminate in changes in gene expression have been delineated. The effects of IL-1 are mediated by the type I IL-1 receptor (IL-1RI) (1). There are two types of IL-1, IL-1α and IL-1β, both of which 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 interleukin-1 receptor/Toll-like receptor (IL-1R/TLR) superfamily, whose members comprise the various receptors involved in host defense and inflammation (2). These include the receptors for the proinflammatory cytokine IL-18 (3), bacterial products such as Gram-negative-derived lipopolysaccharide (LPS) (4), and lipoproteins derived from all bacteria, including gram-positive bacteria (5). These receptors also appear to engage signaling pathways similar to those activated by IL-1, which is expected because all of these receptors share sequence similarity in their cytosolic regions. The superfamily also includes receptors in plants and insects that are critical for host responses to infection. Thus, the common theme in the IL-1/TLR system is that of inflammation and defense. The signaling region of these proteins triggers the increased expression of genes important for defense against infection and subsequent tissue repair. Excessive activation of this system appears to contribute to the pathogenesis of inflammatory diseases, such as RA 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 cytosolic 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). Originally, Toll was thought to participate only in the establishment of dorsoventral polarity in the developing Drosophila embryo (7); however, 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 exist between members; the importance of these differences is as yet unknown, but they may give rise to specific signaling pathways, as has been shown so far for the LPS receptor TLR-4.

Fig. 1.

(A) The currently known members of the IL-1R/TLR superfamily (mammalian, Drosophila, plant, and Xenopus receptors are shown in blue, red, green, and magenta, respectively). All members contain a conserved cytosolic region termed the TIR domain. The superfamily can be divided into three subgroups. The members of subgroup 1, the 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 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 (shown in green). The adaptor subgroup contains MyD88, Mal, and TRIF, which are exclusively cytosolic. MyD88 is a signaling adaptor for IL-1RI, IL-18R, TLR-2, and TLR-4, whereas Mal acts as an adaptor for TLR-2 and TLR-4. Superfamily members may also occur in Caenorhabditis elegans and Streptomyces coelicor (not shown). (B) Three well-conserved regions (box 1, box 2, and box 3) 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 → His mutation in box 2 of 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. Arg431 in IL-1RI, which lies outside the three conserved boxes, is also required for signaling (38).

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). All members of this subgroup contain extracellular immunoglobulin (Ig) domains, 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); and the IL-18 receptor (IL-18R) (3) and its accessory protein, AcPL (12). Also included are IL-1Rrp2 (13), T1/ST2 (14), and the most recently discovered member, IL-1 receptor accessory protein-like (IL-1RAPL) (15). The ligands for these last three receptors have yet to be discovered, but the latter two have important roles in type 2 T helper (TH2) cell responses and cognitive process in the brain, respectively.

Members of the second subgroup also have the cytosolic TIR domain, but have a series of leucine-rich repeats (LRRs) instead of Ig domains extracellularly. The founding member of this group is Toll (6, 7). There are nine additional Drosophila members based on analysis of the genome, the best characterized of which are 18-Wheeler (16), MstProx (17), and Tehao (18). The putative ligand for Toll is Spätzle (19); ligands for MstProx and Tehao proteins are as yet unknown. Ten mammalian TLRs are in this subgroup, TLR-2 and TLR-4 being the most extensively studied. TLR-2 is required for responses to molecules from Gram-positive bacteria (4, 20-24), and TLR-4 is the long-sought-for signaling component for LPS (4).

The plant members of the superfamily are also in this second subfamily. Although they are exclusively cytosolic receptors, they have LRRs and are therefore more closely related to Toll and the TLRs. All are involved in disease resistance in plants (25-27).

Adaptor subgroup

The third subgroup comprises cytosolic proteins with TIR domains, which act as adaptors. Until recently, MyD88 was the only known mammalian cytosolic TIR domain-containing protein (28). It is an adaptor for IL-1RI, IL-18R, and all TLRs so far tested. Database searches of the human genome have led to the identification of MyD88-adaptor-like (Mal), which is also known as TIRAP (29, 30). In vitro studies indicated that Mal is a specific adaptor in TLR-4 signaling; however, recent studies with Mal-deficient mice have revealed that Mal also acts as an adaptor for TLR-2 (31, 32). The newest addition to the adaptor subgroup is TIR domain-containing adaptor inducing interferon (IFN)-β (TRIF) (33). TRIF activates nuclear factor κB (NF-κB) is response to ligands for TLR-2, -4, and -7. TRIF is also involved in the activation of the IFN-β promoter in response to the TLR-3 ligand polyinosine-polycytosine (polyI:C). These molecules are discussed in more detail below. Two vaccinia virus proteins, A46R and A52R, also share amino acid sequence similarity with the consensus TIR domain (34). A46R inhibits NF-κB activation by IL-1, whereas A52R inhibits signaling by TLR-4 and IL-1R1. These proteins probably function to suppress immune activation during the host response.

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 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 (35). Three particular boxes can be identified, however, that 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. The crystal structures of the TIR domains of human TLR1 and TLR2 have been solved (36), and they reveal substantial structural diversity, which may be necessary for mediating distinct signaling pathways. Central to the two structures is a five-stranded parallel β-sheet surrounded by a total of five α-helices on both sides. Also common to both structures is a large surface patch containing the so-called "BB" loop. At the tip of this loop lies a conserved proline residue (Pro681) that is mutated to a histidine in LPS-hyporesponsive C3H/HeJ mice (see below) (4). It is likely that Pro681 forms a point of contact with downstream signaling components, such as MyD88.

IL-1RI-Like Members


Along with the TIR domain, the hallmark of the IL-1RI subfamily is the presence of three extracellular Ig domains. In human IL-1RI, six amino acids are essential for signaling: Arg431, Lys515, Arg518, Phe513, Trp514, and Tyr519 (37). In addition, Pro521 is required for maximal signaling capacity. Phe513 and Trp514 are in the conserved box 3 in the TIR domain of IL-1RI (Fig. 1B). The precise role of each of these amino acids is still unknown. Presumably, they are required for correct recognition of the signaling domain by signaling proteins. Some may also be involved in trafficking of the receptor (38) to the plasma membrane. It is possible that the association among the TIR regions of IL-1RI, IL-1RAcP, and MyD88 [which are present in a complex after IL-1 binding (28)] is required for signaling. This kind of trimerization also occurs in the p55 tumor necrosis factor (TNF) receptor system; 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, both cytokines ultimately activate similar pathways. However, there are also differences between IL-1 and TNF, in that TNF is toxic to many cell types. This toxicity may be due to TNF's ability to recruit the Fas-associated death domain (FADD) protein, which triggers apoptosis (39).

One amino acid in the cytosolic region of IL-1RI, Tyr479, has been suggested to interact directly with a signaling protein (40). The p85 subunit of phosphatidylinositol 3-kinase (PI3K) may associate with this tyrosine when phosphorylated, thereby recruiting p110, the catalytic subunit of PI3K, to the membrane. Tyr479 occurs in a Tyr-Glu-X-Met motif recognized by p85. The p85 subunit may associate with IL-1RAcP rather than IL-1RI (40); however, IL-1RAcP does not contain a tyrosine homologous to the Tyr479 of IL-1RI (9). PI3K stimulates the phosphorylation of the p65 subunit of the transcription factor NF-κB, thereby promoting its transactivating capabilities (41).

In all systems in which it has been tested, IL-1RI is important for IL-1 effects. Antibodies to IL-1RI block IL-1 responses in several cell systems. In IL-1RI-deficient mice, IL-1 responses such as increased IL-6 production and fever are impaired (42). Also impaired were the acute-phase response, delayed-type hypersensitivity, and the ability to combat infection by Listeria. These results indicate the importance of IL-1RI in mediating inflammation and fighting 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 (43, 44). 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 interacts with the receptor at two regions, one with the first two Ig domains and the other with domain 3. The binding of IL-1β induces a conformational change that results in domain 3 moving so that the receptor wraps around the ligand (43). IL-1RA, unlike IL-1β, does not induce movement in domain 3 (44), possibly because IL-1RA contacts IL-1RI only at domains 1 and 2. Such a movement in domain 3 triggered by IL-1β 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 (45). A complete account of the molecular changes that occur upon binding IL-1 in IL-1RI and IL-1RAcP extracellularly, and indeed intracellularly, awaits the full elucidation of all the structures involved both in the absence and presence of ligand.


IL-1RAcP acts as the second chain of the IL-1R complex and is required for signaling initiated by IL-1 (9). Despite having three Ig domains, IL-1RAcP does not bind IL-1 (9, 46). However, IL-1RAcP appears to increase the affinity of IL-1RI for ligand by a factor of about five (9). IL-1RAcP is widely expressed, and on the whole its expression correlates with that for IL-1RI. Brain is the exception, where in rat the two proteins appear not to colocalize (47).

On the basis of 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. Reconstitution experiments in cells lacking IL-1RAcP have identified regions in the cytoplasmic tail that are required for IL-1 signaling (48). Truncation analysis revealed that box 3 of the TIR domain of IL-1RAcP (amino acids 538 to 542) is critical for NF-κB activation, IL-2 production, and stabilization of IL-2 mRNA through c-Jun NH2-terminal kinase (JNK), whereas optimal IL-2 promoter activity depends on amino acids 527 to 534. It has been suggested that a putative adaptor molecule binds to residues 527 to 534 resulting in the recruitment of MyD88 to box 3. One such candidate is the recently described scaffolding protein, Toll-interacting protein (Tollip) (49). Tollip was found to associate with the TIR-domain of the IL-RAcP and, more recently, TLRs 2 and 4 (50). In resting cells, Tollip forms a complex with IL-1R-associated kinase (IRAK) and inhibits IL-1- or LPS-induced signaling by binding to and blocking IRAK phosphorylation. Upon activation, Tollip-IRAK complexes are recruited to the activated receptor, permitting the association of IRAK and MyD88. IRAK then undergoes rapid autophosphorylation and, in turn, phosphorylates Tollip, resulting in the dissociation of IRAK-Tollip complexes and the activation of downstream signaling components (49).

Whether IL-1RI and IL-1RAcP are in a complex before the addition of ligand is still unclear. IL-1 may cause a conformational change in IL-1RI that recruits IL-1RAcP to the complex. A more likely scenario is that the addition of ligand increases the affinity of IL-1RI for IL-1RAcP. As mentioned above, the extracellular Ig domains are probably involved in the assembly of IL-1RI and IL-1RAcP into a heterodimer, with full IL-1 responsiveness requiring the synergistic actions of box 3 within IL-1RAcP and boxes 1 and 2 in IL-1R1.

Decoy receptors: IL-1RII and B15R

The existence of a second IL-1 receptor, IL-1RII (10), was suggested from cross-linking studies, and its gene sequence was later determined. The extracellular domain has three Ig domains and exhibits a high degree of amino-acid sequence similarity to IL-1RI. Its intracellular domain, however, is 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 in 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 (51). IL-1RII is shed from cells and acts as a decoy receptor, preventing IL-1 from binding to IL-1RI (52). This provides another way (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 (53). 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 (54), which is consistent with the role of both proteins as inhibitors of IL-1 binding to IL-1RI.

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, 55) with apparently similar IL-1 inhibitory action. Deletion of B15R gives rise to a more virulent virus, suggesting 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 also 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.


The cytokine IL-18 was first described as IFN-γ-inducing factor (56). It is a key inducer of T helper 1 (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). IL-18R is not detectable in brain, possibly due to expression that is below the limits of detectability rather than complete lack of expression. A role for IL-18R in TH1 cells is borne out by studies with IL-18R knockout mice (57). 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 by its sequence similarity to IL-1RAcP and was named accessory protein-like (AcPL) (12). Its expression appears to be more restricted than that of IL-18R, in that it is not detectable in heart or testis. It is likely that AcPL functions as IL-1RAcP does in IL-1 signaling.

An IL-18 binding protein (BP) has also been identified and may be functionally similar to IL-1RII. It lacks a transmembrane domain and acts as a soluble decoy receptor by binding to IL-18 and preventing interaction with the IL-18R (58). IL-18BP is highly homologous to a family of poxvirus proteins that can also bind to IL-18 and inhibit IFN-γ production and NK cell or cytotoxic T lymphocyte (CTL) activity in a dose-dependent manner (59, 60).


A gene named T1 or ST2 was cloned as a delayed early response gene that is stimulated in response to proliferative signals (61, 62). Both a soluble protein similar to the extracellular domain of IL-1RI and a transmembrane form have been described (63). The similarity of the transmembrane form of T1/ST2 to IL-1RI extends throughout the molecule, including the extracellular Ig domains and an intracellular TIR domain. Soluble and membrane-bound versions are expressed in different tissues from different promoters (64). The soluble form is produced mainly by fibroblasts and possibly mast cells, whereas the membrane-bound form is predominantly expressed on hematopoietic cells (64). 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 (65). 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 candidate protein was purified and the gene that encodes it was cloned. However, the expressed gene product was unable to activate NF-κB (66). The second protein was partially purified and was found to activate p38 mitogen-activated protein kinase (MAPK), but not NF-κB (67). 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 mediating NF-κB activation, 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, that 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 (68). 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 (69). 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 (70), whereas in the other it was impaired (71). This discrepancy may be due to differences in challenges used on the mice to test their phenotype.

The identification of the ligand for T1/ST2 will 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 the latter hypothesis, new data indicate that NF-κB is required for TH1, but not TH2, cell function (72).

Newer members

Four new members of this IL-1RI-like subfamily have been described. Single Ig IL-1R-related molecule (SIGIRR) is ubiquitously expressed and has a TIR domain and one extracellular Ig domain (73). A chimera between the IL-1RI extracellular domain and SIGIRR intracellular domain was unable to stimulate NF-κB in response to IL-1. The basis for this lack of effect is not clear. The function of SIGIRR and its ligand have yet to be determined.

The second new member, termed IL-1 receptor accessory protein-like (IL-1RAPL), 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. 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 (74, 75). 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 that forms a functional complex with IL-1RAPL). The cognate ligand has not been described.

Sequence database searches have also identified TIGIRR and IL-1Rrp2 (76). TIGIRR shows a high degree of sequence similarity with IL-1RAPL and is likewise located on human chromosome X. IL-1Rrp2 is thought to be the receptor for the novel IL-1 family member IL-1F9 (77). Transient transfection of Jurkat cells with the putative receptor renders them permissive to NF-κB activation upon IL-1F9 treatment. A second novel IL-1 homolog, IL-1F5, shows striking similarity with the IL-1 receptor antagonist and was shown to inhibit the IL-1F9-induced response (77). Binding of IL-1F9 or IL-1F5 to IL-1Rrp2 has not yet been demonstrated and it is unclear at present whether this association is direct.

Toll-Like Members

The sequence similarity between IL-1RI and Drosophila Toll provided the first description of the TIR domain (6). 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 10 have been described, termed TLR-1 through TLR-10), other insect members (including 18-Wheeler), and proteins involved in disease resistance in plants (notably N protein, L6, RPP1, and RPP5).


In the developing Drosophila embryo, Toll drives the establishment of dorsoventral polarity. The ligand that activates Toll in this process may 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 (78). The structure of Toll has been modeled, and it resembles that of platelet glycoprotein 1b (79), 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. Toll activates a protein kinase termed Pelle by interacting with adaptor proteins dMyD88 and Tube. This pathway leads to the phosphorylation of Cactus, which dissociates from the transcription factor Dorsal, allowing Dorsal to translocate to the nucleus and increase the expression of target genes involved in dorsoventral polarity (Fig. 2). In an analogous signaling mechanism, IL-1 activates the protein kinase IRAK-1, which is related to Pelle; MyD88, which is homologous to dMyD88, acts as the adaptor. A human homolog of Tube has not yet been identified. IRAK-1 functions in a protein kinase cascade that leads to phosphorylation of inhibitor of κB (IκB). When phosphorylated, IκB dissociates from NF-κB, allowing NF-κB to translocate to the nucleus and increase expression of inflammatory and immune response genes. IκB is a homolog of Cactus. Mammalian NF-κB is a family (NF-κB/Rel) that includes members homologous to Dorsal (Fig. 2). This signaling pathway, starting with the TIR domain, is thus used in two different contexts to alter gene expression.

Fig. 2.

Conserved mechanisms for signaling in the IL-1R/TLR superfamily. The signaling pathways that lead to the activation of NF-κB/Rel family proteins, through Spätzle (in Drosophila melanogaster), LPS, IL-1, or IL-18, are quite similar. During development in Drosophila, Spätzle is the ligand for Toll and is generated as a result of a protease cascade. Spätzle activation during immune challenge in Drosophila may require a peptidoglycan recognition protein (PGRP) and activation of the protease Persephone. In mammals, 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 PRR 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. There are high sequence similarities between dMyD88 and MyD88, between Pelle and IRAK-4, between dTRAF and TRAF-6, and between Cactus and IκB. Dorsal and Dif are both NF-κB/Rel family members. Human MyD88 and Mal also share a high degree of sequence similarity. A Drosophila homolog of Mal has not yet been identified. Other signaling pathways in mammals activate the stress-activated 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 NF-κB/Rel family member, Dif, which regulates the expression of antifungal peptides such as Drosomycin (Fig. 2) (8, 80). It was postulated that a Spätzle-like protein would be the Toll ligand in the adult in response to infection. Indeed, Spätzle in adult flies is cleaved during immune challenge with several Gram-positive and fungal pathogens (81). The active ligand was present in mutant flies lacking the serine protease inhibitor (serpin) Necrotic (encoded by Nec). Expression of Necrotic increased in wild-type flies during infection, indicating the presence of a negative feedback loop. These findings suggested that Toll is not the receptor for microbial products, but rather that Toll is triggered (as in the embryo) by cleaved Spätzle, which is generated in a serpin-controlled proteolytic cascade. Mutations in the persephone (psh) gene, which encodes a novel serine protease, block the induction of Drosomycin following fungal infection (82). This phenotype was rescued by overexpression of Spätzle, indicating that Persephone acts upstream of Spätzle. Direct activation of Spätzle by Persephone has not yet been demonstrated. Persephone itself does not contain a microbial pattern recognition-binding domain and is presumably activated by an as-yet-unidentified upstream fungal pattern recognition receptor (PRR).

Searches for fly mutations that block antimicrobial gene expression have also led to the identification of peptidoglycan recognition proteins (PGRPs) that are probable PRRs. Gram-positive bacterial infections activate the Toll pathway through PGRP-SA, an extracellular recognition factor circulating in the fly hemolymph (blood) (83). Gram-negative and gram-positive bacteria activate the Imd pathway through the putative transmembrane receptor PGRP-LC (84). The Imd pathway differs from the Toll pathway in that it has no TIR domain members. The ultimate target of this pathway is Relish, an NF-κB/Rel family member closely related to mammalian p105. Relish is not inhibited by Cactus, but contains intrinsic inhibitory sequences in the form of ankyrin repeat domains. Activation of Relish depends on signal-induced endoproteolytic cleavage that frees the Rel homology domain, allowing for its translocation to the nucleus.

Other putative members of the Drosophila TIR family have been identified, including Tehao, 18-Wheeler, MstProx, Tollo, CG7520 (Tlr6), and CG18241 (Tlr4) (16, 17, 85). Both Toll and Tehao are involved in the innate immune response, whereas the functions of the others remain to be determined. Although, initially, 18-Wheeler was thought to be involved in host defense in the fly (85), subsequent data do not support this function (86). 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 (87). They probably also have a role in host defense.

Mammalian Tol-Llike receptors

Soon after the report on TLRs 1 to 5 from Drosophila (20), a human Toll (termed hToll) was described that was identical to TLR-4 (88). When expressed as a fusion protein with CD2, which renders it constitutively active, TLR-4 increased the expression of B7.1 in antigen-presenting cells, and stimulated the production of inflammatory cytokines, including IL-1. B7.1 is the ligand for CD28, which 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.

Human TLR-1 is ubiquitously expressed and more abundant than the other TLRs (35). 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 cells in ovaries and monocytes express large amounts of TLR-5. TLR-6 is expressed in spleen, thymus, ovary, and lung (89) TLR-9 and TLR-10 are highly expressed on B cells, whereas TLR-7 is selectively expressed on plasmacytoid dendritic cells (DCs), whose key feature is the production of type I IFNs (90). In addition, the size of mRNA transcripts for certain TLRs varies in tissues, suggesting the presence of alternative splice forms (35).

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 (91); however, it has a short cytoplasmic sequence that lacks the TIR domain. RP105 is expressed predominantly on B cells, and absence of the molecule results in impaired responses to LPS, suggesting that RP105 surface expression modulates B cell responsiveness to endotoxin (92).

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

By far the most extensively characterized members of the TLR family are the proteins TLR-2 and TLR-4. TLR-2 acts as a PRR for various microbial products, including peptidoglycan, bacterial lipoproteins, lipoarabinomannan (LAM), glycosylphosphotidylinositol lipid from Trypanosoma cruzi, and zymosan, a component of yeast cell walls (22, 24, 93, 94). TLR-2 does not recognize these pathogen-associated molecular patterns (PAMPs) independently, but rather functions as a heterodimer with either TLR-1 or TLR-6 (95). A dimer of TLR-1 and TLR-2 (TLR-1/TLR-2) appears to recognize bacterial lipopeptides, whereas a TLR-2/TLR-6 dimer responds to mycobacterial lipopeptides. These lipopeptides differ in the number of acyl groups they contain, attesting to the specificity in recognition by different TLRs. To date, no other TLR pairs have been identified, and it is possible that receptors might function as homodimers, as does TLR-4.

TLR-4 has been identified as the receptor for LPS, lipoteichoic acid (LTA), fibronectin, F protein from syncytial virus, and taxol, a plant diterpene structurally unrelated to LPS but possessing potent LPS-mimetic effects on murine cells (4, 96-99). TLR-4 was initially 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, was in the gene encoding TLR-4 (4). A single point mutation (Pro712 → His712) 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. The LPS used in these studies was derived from Escherichia coli; however, researchers have demonstrated that responses to Porphyromonas gingivalis-derived LPS are only partially impaired in TLR-4-deficient macrophages (100), suggesting that a receptor other that TLR-4 is involved in recognition of some forms of LPS. It was subsequently shown that TLR-2 recognizes LPS from P. gingivalis and the eubacteria Leptospira interogans (101). The molecular basis for why different types of LPS are recognized by different TLRs has recently been suggested (102). If the lipid A portion is conical in shape (as in E. coli LPS), it will be recognized by TLR-4, whereas cylindrical lipid A (as in P. gingivalis LPS) is recognized by TLR-2. A difference between fly and mammalian Toll is apparent here, where evidence suggests that LPS binds TLR-4 directly, whereas in flies, Spätzle is generated in response to fungal products through the activation of Persephone. In humans, LPS action on TLR-4 can be blocked by the serpin antithrombin-III, indicating a role for a serine protease in mammals also (103).

It is becoming increasingly clear that, although a core set of genes are induced by both TLRs, differences in gene expression are also evident. For example, peptidoglycan, which activates the TLR-2/TLR-6 dimer, is a strong inducer of IL-8 and IL-12p40 homodimers, which inhibit the production of active IL-12 (104). In contrast, TLR-4 activation leads to the production of IL-12p70, the active form of IL-12. In vivo, E. coli LPS is a strong inducer of IFN-γ with little or no IL-5, IL-13, or IL-10 being produced, whereas LPS from Porphyromonas induces high levels of IL-5, IL13, and IL-10, but little or no IFN-γ. Taken together, these observations suggest that TLR-4 induces a response that is more polarized toward TH1 cells. Another important difference between TLR-2 and TLR-4 concerns recruitment of the PI3K adaptor p85, which is dependent on phosphorylation of a tyrosine within the TIR domain of TLR-2 (105). This tyrosine does not occur in TLR-4; therefore, activation of PI3K by LPS activation of TLR-4 must occur by a different mechanism.

Before binding TLR-4, LPS is captured by the serum protein LPS-binding protein (LBP) (106). LBP catalyzes the transfer of LPS to soluble CD14 or GPI-anchored CD14. Recent evidence suggests that LPS is then briefly released into the lipid bilayer, where it interacts with a complex composed of TLR-4, MD-2 (a small protein found in association with TLR-4 on the cell surface), heat shock protein (HSP) 70, HSP90, chemokine receptor 4, and growth differentiation factor-5 (107). The integrins CD11b and CD18 are further recruited to the complex. Indeed, certain responses to LPS are impaired in CD11b- and CD18-deficient macrophages (108). The composition of such activation clusters may vary with stimulus or from cell-type to cell-type thereby providing an additional level of specificity in TLR signaling.

A recently characterized polymorphism in human TLR-4 (Asp299Gly), which renders individuals more susceptible to severe bacterial infections, has been associated with decreased risk of atherosclerosis (109). In this study, patients heterozygous for the Asp299Gly allele exhibited lower levels of circulating inflammatory mediators and a smaller intima-media thickness in the common carotid artery. Conversely, subjects with wild-type TLR-4 produced higher levels of proinflammatory cytokines and exhibited a significantly greater risk of developing carotid artery disease.

Other members of the TLR family

In mammals, TLR-3 is thought to contribute to antiviral recognition (110). This receptor recognizes double-stranded RNA (dsRNA), which is produced by most viruses during the infection cycle. TLR-3-deficient mice have a profound defect in their ability to respond to the synthetic dsRNA analog polyI:C, and transfection of HEK293 cells with TLR-3 renders them sensitive to polyI:C. Microarray analysis of DCs exposed to either LPS or dsRNA has identified a common set of genes regulated by both stimuli, presumably through the activation of the transcription factor NF-κB (111). There are, however, marked differences in the two responses, indicating that each stimulus activates distinct signaling pathways. MyD88 does not appear to be required for TLR-3 signaling (see below). Also, TLR-3 does not contain the conserved proline residue found at the position equivalent to Pro712 in murine TLR-4. However, a hydrophobic amino acid, which TLR-3 does have, may suffice in this position. There is, however, a possibility that this feature may account for those differences seen and possibly results in the recruitment of distinct signaling components to the activated TLR-3 complex.

Other ligands recognized by TLRs include flagellin, a highly conserved protein found in bacterial flagella (112), which is recognized by TLR-5. Flagellin is unusual in that it does not possess any features that would distinguish it as a nonself pathogen. Unmethylated CpG-DNA occurs only in bacteria (because of the lack of methylation enzymes) and is recognized by TLR-9 (113). Occupation of these receptors results in the activation of macrophages, DCs, and B cells to secrete cytokines such as IL-12 and IL-18. The most recent putative TLR ligand to be identified is the antiviral compound imiquimod (114). After initially being found to be inactive in MyD88-knockout mice, the drug was tested on various TLR-knockout mice, leading eventually to the identification of TLR-7 as the possible receptor. TLR-8 has also been identified as a receptor for imiquimod (115).

Plant members of the IL-1R/TLR superfamily

Members of the IL-1R/TLR superfamily are also present in plants. These members of the family are structurally quite different from the mammalian and invertebrate members in that they are completely cytosolic, lacking any transmembrane or extracellular domains. The first to be described was the tobacco mosaic virus resistance gene product N protein (25, 116). Four LRRs, along with the TIR domain, make N protein more related to Toll than to IL-1R. The role of N protein in tobacco is to activate an enhanced inflammatory 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 (26, 117). Arabidopsis is actually predicted to contain 85 TLRs, making them the primary means of pathogen detection in this or, indeed, any plant. From an evolutionary perspective, 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. Although the ligands for the plant receptors are wholly unknown, they are most likely cytosolic, because the receptors lack transmembrane and extracellular domains.

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) (118, 119). All of these receptors possess a TIR domain, indicating 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 and p44 [p42/p44; also known as extracellular signal-regulated kinase (ERK) 1 and ERK2 (ERK1/2)], and JNK, respectively. The activation of NF-κB leads to increased transcription of target genes. The ERK1/2 MAPK also appears to regulate transcription, whereas p38 and JNK promote the stabilization of induced mRNA.

MyD88 and Mal

Activation of NF-κB by IL-1 begins with formation of a complex containing MyD88, IRAK, and IRAK-2 (28, 120, 121). MyD88 was first described as a protein expressed during myeloid differentiation (122). It contains a COOH-terminal TIR domain and an NH2-terminal death domain (DD) (28). MyD88 dimerizes through both the death and TIR domains. Dominant-negative constructs of MyD88 block NF-κB activation by IL-1, TLR-2, TLR-4, or TLR-9 when transfected into HEK293 cells along with the various receptors (123, 124). MyD88 associates with the activated IL-1R complex and TLR-4 in coimmunoprecipitation experiments (28). The importance of MyD88 in TIR domain signaling is highlighted by recent studies on MyD88-knockout mice (125). These mice are completely impaired in their ability to respond to ligands such as IL-1, IL-18, peptidoglycan, and CpG-DNA. A detailed analysis of macrophages and DCs from these mice revealed that, although cytokine production in response to LPS and polyI:C was completely abrogated, NF-κB and MAPK activation still occurred, albeit with delayed kinetics. This led to the conclusion that a pathway distinct from MyD88 was being triggered by TLR-4 and possibly by TLR-3. This hypothesis was supported by the finding that several genes are still induced in MyD88-deficient macrophages following LPS stimulation (126). These genes include IFN-inducible protein (IP)-10 and various IFN-sensitive genes, which all depend on the transcription factor IRF-3. A search of a human DC cDNA library led to the identification of a MyD88 homolog named MyD88-adaptor like (Mal) (also known as TIRAP) (29, 30). Mal also contains a COOH-terminal TIR domain, but it lacks a death domain in its NH2-terminus. Mal associates with the TIR domain of MyD88 and TLR-4, and dominant-negative forms of the protein block TLR-4, but not IL-1 or TLR-9 signaling. Further insight into the role played by Mal in TLR signaling has come from recent studies on Mal-deficient mice (31, 32). These mice are impaired in their responses not only to TLR-4 ligands, but also to TLR-2 ligands. LPS induction of interferon-inducible genes is normal in these mice, as is LPS-induced DC maturation, suggesting that Mal does not participate in a MyD88-independent pathway, as previously thought, but instead plays a role in the MyD88-dependent pathway shared by TLR-2 and TLR-4. A third adaptor molecule, TRIF, has also been identified (33). TRIF activates NF-κB in response to ligands for TLR-2, -4, and -7. In overexpression studies, TRIF preferentially activated the IFN-β promoter, whereas dominant-negative forms of the molecule blocked the polyI:C-mediated TLR-3 response, indicating that TRIF functions as an adaptor in the TLR-3 signaling pathway. In support of this, TRIF was found to associate with TLR-3 and IRF-3 in coimmunoprecipiation experiments (33). Whether or not TRIF is involved in the LPS-induced IFN response has yet to be determined and awaits further studies.


The key function of MyD88 in TLR and IL-1 signaling is to recruit members of the IRAK family. IRAK-1 was the first member of this family to be identified (120). It contains a DD and a Ser-Thr kinase domain, and is homologous to the Drosophila protein kinase Pelle. As mentioned previously, IRAK-1 is bound to Tollip in resting cells, where it is prevented from undergoing phosphorylation (49). Upon activation, IRAK-1 is phosphorylated, weakening its affinity for Tollip and making it more accessible for interaction with MyD88. The association of MyD88 with active IRAK-1 is disrupted upon further phosphorylation of IRAK-1. This secondarily phosphorylated form of IRAK-1, called hyperphosphorylated IRAK-1, is then released into the cytosol, where it interacts with the soluble tumor necrosis factor-α (TNF-α) receptor-associated factor 6 (TRAF-6). The interaction with TRAF-6 results in NF-κB activation (described below). Until recently, only two other members of the IRAK family were known, IRAK-2 (121) and IRAK-M (127). Both proteins contain a DD and an apparent kinase domain; however, neither is catalytically active. This is due to the absence of key residues within their putative kinase domains. Indeed, the kinase activity of IRAK-1 is dispensable for IL-1-induced NF-κB activation, because overexpression of a kinase-inactive mutant still stimulated NF-κB-dependent reporter gene expression (128).

The fact that kinase-negative versions of IRAK-1 can still undergo phosphorylation in the absence of adenosine triphosphate (ATP) binding led to the search for an IRAK-1-activating kinase and the subsequent identification of IRAK-4 (129). Among IRAK family members, IRAK-4 is the one with a protein sequence most similar to Drosophila Pelle. IRAK-4-knockout mice are completely resistant to LPS, CpG-DNA, peptidoglycan, and polyI:C, indicating that IRAK-4 plays an essential role in innate immunity (130). Furthermore, IRAK-4 is indispensable for IL-1 signaling. Although overexpression of IRAK-2 or IRAK-M compensated for lack of IRAK-1 in IRAK-1-deficient cells, overexpression of IRAK-4 did not activate NF-κB in IRAK-1-deficient cells, suggesting that IRAK-4 functions upstream of IRAK-1. In agreement with this, in vitro IRAK-4 phosphorylates IRAK-1 on Thr387 and Ser376 (129). This primary phosphorylation event presumably facilitates the interaction of IRAK-1 with MyD88 and the receptor complex. Once activated, IRAK-1 undergoes hyperphosphorylation and leaves the complex to interact with TRAF-6.

IRAK-M is induced upon TLR stimulation and functions as a negative regulator of TLR signaling (131). Macrophages from IRAK-M-deficient mice showed enhanced proinflammatory cytokine production and impaired endotoxin tolerance. In addition, overexpression of IRAK-M in 293T cells prevented the formation of IRAK-1/TRAF-6 complexes upon IL-1β stimulation. It has been proposed that IRAK-M inhibits TLR action either by preventing the phosphorylation of IRAK-1 by IRAK-4 or by stabilizing the TLR/MyD88/IRAK-1/IRAK-4 complex. IRAK-2 may also serve as a negative regulator of signaling, because it shares similar features to IRAK-M, notably the lack of kinase activity. IRAK-2 is expressed ubiquitously, whereas IRAK-M expression is restricted to the monocyte and macrophage lineage. Clarification of the role of IRAK-2 in IL-1/TLR signaling awaits the generation of IRAK-2-knockout mice.


The role of IRAK-1 is to recruit TRAF-6, a member of the RING-finger family of proteins. TRAF-6 is the only TRAF family member to participate in both TNF and IL1/TLR signal transduction (132). TRAF-6 interacts with TAB-2, and this complex activates TGF-β-activated kinase (TAK)-1 (133). TAK-1 then serves as a branch point, leading to activation of the IκB kinase complex and NF-κB, and the upstream kinases that activate p38 and JNK. The initial insight into the role of TRAF-6 in IL-1 signaling came from studies in which a dominant-negative mutant lacking the NH2-terminal region of TRAF-6 inhibited IL-1-induced NF-κB activation (132). TRAF-6 was also found to associate with endogenous IRAK-1 in 293 cells stably expressing IL-1R1. The crystal structure of TRAF-6 alone and in complex with peptides from the TNF receptor (TNFR) superfamily members CD40 and RANK has recently been reported (134). A Pro-X-Glu-X-X-(aromatic or acidic residue) TRAF-6 binding motif was identified not only in CD40 and RANK, but also in IRAK family members. IL-1 stimulation triggers ubiquitination and oligomerization of TRAF-6, followed by association with the recently identified TRAF-6-binding protein (T6BP) (135). Polyubiquitinated TRAF-6 binds to TAB-2, which is membrane-bound in untreated cells and translocates to the cytosol following IL-1 stimulation. TAB-2 functions as an adaptor linking TRAF-6 to TAK-1 and TAB-1. TRAF-6 itself functions as a ubiquitin ligase in conjunction with cofactors Ubc13 and Uev1A [also known as TRAF-6-regulated IKK activator 1 (TRIKA1)] to mediate the assembly of polyubiquitin chains to activate IKK. The components of a second IKK activator complex, TRIKA2, have been identified as TAK-1, TAB-1, and TAB-2 (136, 137). The precise mechanism by which ubiquitinated TRAF-6 activates TAK-1 has not yet been elucidated; what is clear, however, is that ubiquitination is being observed more frequently as a prominent signaling event in many systems and is no longer considered to be solely a marker for destruction.


TAK-1 activation culminates in the assembly of a high molecular weight complex known as the signalosome. IKKα and IKKβ are components of this complex, which also contains the scaffold protein IKKγ [also known as NF-κB essential modulator (NEMO)] (138). The function of the signalosome is to phosphorylate a group of NF-κB-inhibitory proteins collectively termed IκBs. Phosphorylation of the IκBs results in their ubiquitination and subsequent degradation. IKKα and IKKβ each can phosphorylate IκBα on two key serines, Ser32 and Ser36, leading to the release of the NF-κB heterodimer. The kinases share 52% similarity at the amino acid level; however, studies from knockout mice have shown that IKKβ is the predominant kinase in the activation of NF-κB (139). The IKKs also phosphorylate and regulate p105 (also known as NF-κB1) and p100 (also known as NF-κB2) precursors, leading to their processing to p50 and p52, respectively (140, 141). NEMO is an essential component of the signalosome, because cells lacking this protein are unable to respond to NF-κB-inducing agents (138). NEMO also binds to various proteins other than the IKKs, including RIP, an adaptor involved in TNFα signaling, and A20, a phosphatase involved in modulating IKK activity (142). NEMO also binds to the adaptor molecule TANK (143), which synergizes with the IKK-related kinases, IKKϵ and TBK1, to positively regulate the IKK complex. Association of NEMO with TANK is thought to link the IKK complex to upstream signaling mediators. This complex set of protein-protein interactions culminates in NF-κB activation (Fig. 2).

Another route to the signalosome involves the kinase MAPK or extracellular signal-regulated kinase (ERK) kinase kinase 1 (MEKK1) (144). A novel protein linking TRAF-6 to MEKK1 has been described and is termed ECSIT (evolutionarily conserved signaling intermediate in Toll pathways) (145). As the name suggests, there is a Drosophila homolog.

Similar proteins are involved in LPS and IL-18 signaling (118, 119). However, LPS requires only IKKβ for NF-κB activation (146). On the basis of transgenic studies, it appears that IKKα is involved in development, whereas IKKβ is required for inflammation (147). It is, therefore, possible that the IL-1RI/TLR system will culminate in IKKβ activation in the NF-κB pathway.


Details of the upstream regulation of the MAPKs p38, ERK1/2, 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 (i) Raf-1, which lies upstream of MAPK kinase (148); (ii) MAPK kinase itself (149), which lies upstream of ERK1 and ERK2; (iii) MAPK kinase (MKK) 3, MKK4, and MKK6 (150), which lie upstream of p38; and (iv) MKK4 and MKK7 (151), which lie upstream of JNK. Details of how the kinase cascades are activated are less clear; however, roles for small guanosine triphosphatases (GTPases) have been identified. IL-1 causes increased GTP binding and hydrolysis in membranes (152), and direct evidence for Rho and Rac1 activation has been presented (153, 154). In addition, dominant-negative mutants of Rac1 and Cdc42 block activation of JNK and p38 MAPK in response to IL-1 (155, 156); although in the case of JNK, this has been disputed (157). Ras is critical for p38 activation, with Rap having an antagonistic effect (158). Ras has been positioned downstream of TRAF-6 on the pathway to p38 (159).

Studies in MKK3-deficient mice have revealed a major role for p38 activation in the production of IL-12. Mice lacking the gene are seriously compromised in their ability to synthesize this cytokine, which is a key regulator of IFN-γ production (160). Further studies have revealed that MKK3-mediated p38 activity is important for TNF-induced cytokine production and production of TNFα itself (161). Studies in JNK2-deficient fibroblasts have shown that JNK is required for the production of multiple cytokines, including IL-6 and IFN-γ (162). JNK is activated by TRAF-2 upon TNF stimulation and by TRAF-6 in response to IL-1 and LPS (163). ERK1 and ERK2 activation is an important event in T cell activation and is required for TH2 differentiation in addition to IL-4-induced signal transducer and activator of transcription 6 (STAT6) phosphorylation (164).

Gene Knockout Studies

Some of the results from knockout mice are described above. This section summarizes the involvement of key components in the signaling pathways activated by various members of the IL-1/TLR superfamily. 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 (165-169) (Table 1).

Table 1.

Phenotypes of mice deficient for Mal, MyD88, IRAK-1, TRAF-6, IRAK-4, or Rip-2.

In MyD88-deficient mice, responses to IL-1, IL-18, or LPS in vivo are impaired (125, 165, 166). In cultured cells deficient in MyD88, neither IL-1 nor IL-18 activates NF-κB or JNK (125). However, LPS still activates both pathways, although the response is delayed (125). MyD88-deficient cells are also capable of activating IRF-3, as well as inducing genes containing IFN-stimulated regulatory elements in response to LPS (126). This implies that LPS can use signaling molecules other than MyD88. Two such candidates described here are the adaptors Mal and TRIF. Mal-deficient cells assume a similar phenotype to MyD88-deficient cells, in that responses to TLR-2 and TLR-4 ligands are impaired at the NF-κB level, whereas LPS-induced IRF-3 activation is unaffected (31, 32). TRIF directly interacts with and activates IRF-3 as part of the TLR-3 response to polyI:C (33). A full characterization of the pathways activated by this molecule awaits the generation of knockout mice.

Cells from IRAK-deficient mice are unresponsive to IL-18 and LPS in terms of NF-κB and JNK activation, and in animal studies, responses to IL-1, IL-18 and LPS are absent (167-169). Responses to IL-1 and various TLR ligands are severely impaired in IRAK-4 deficient mice and cultured cells, indicating that this member of the IRAK family is indispensable for TIR-dependent signaling. Studies in IRAK-M deficient mice have revealed a negative regulatory role for this kinase. IRAK-2 may have a similar role, although gene knockout studies have not yet been reported. In mice deficient for TRAF-6, activation of NF-κB and JNK in response to IL-1 or LPS is impaired. These mice are also osteopetrotic (having increased bone density), which implies that TRAF-6 has an important role in bone turnover (170). Finally, mice deficient in the TNF-receptor associated kinase RIP-2, do not respond to stimuli that activate TLR-2, TLR-3, and TLR-4, but still respond to stimuli that activate TLR-9, again attesting to specificity in signaling by TLRs (171).


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 having arisen 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 TLRs 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 present in all animals, with p38 and JNK pathways 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 and whether there are further 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.


  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.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
View Abstract

Stay Connected to Science Signaling

Navigate This Article