The Toll receptor is a type I transmembrane receptor, evolutionarily conserved, and its homologs are found in insects, plants and mammals. The extracellular domain of Toll receptors contains leucine-rich repeat (LRR) whereas the cytoplasmic domain shows striking homology with that of IL-Rs, and is referred to as Toll/IL-1R (TIR) domain. The LRR domains are composed of different numbers of repeats of 24–29 amino acids in length, and each repeat contains the motif xxLxLxx as well as other conserved leucines. The Toll gene was initially identified as a gene essential for the dorsal-ventral development in the Drosophila embryo. Subsequently Toll signaling has been shown to play an essential role in the immune response to antifungal infection [1]. Toll-mutant flies show compromised induction of antifungal peptide, and are highly sensitive to fungal infection. This finding triggered search for Toll homologues in mammals through EST and genomic sequence database searches. So far, ten members of TLR family have been identified in mammals [2–4]. The essential role of Drosophila Toll in the antifungal response suggested that mammalian Toll-like receptors also participate in innate immunity. In fact, Toll-like receptors have been found to be essential for microbial recognition (Fig. 1).

2.1 LPS signaling and TLR4

2.2 Recognition of lipoproteins and peptidoglycan by TLR2

2.3 Recognition of lipid configuration of lipoproteins by TLR2/6 and TLR2/1 heterodimers

All membrane-anchored lipoproteins contain a lipolyated amino-terminal residue. The immunostimulatory activity of lipoproteins is attributed to the lipid portion. Fig. 2(a) shows structural difference between MALP-2 and PAM₃CSK₄. PAM₃CSK₄ contains a triacylated cysteinyl residue at the N-terminus, whereas the cysteinyl residue in MALP-2 is only diacylated. The response to MALP-2 is completely abolished in both TLR2-deficient and TLR6-deficient macrophages, whereas the response to PAM₃CSK₄ is normal in TLR6-deficient macrophages in spite of no response in TLR2-deficient macrophages [9]. The replacement of lipid portion of MALP-2 with that of PAM₃CSK₄ activated TLR6-deficient macrophages, demonstrating that TLR6 discriminates the subtle difference in the lipid portion of these lipopeptides (our unpublished data). Reciprocally, the response to PAM₃CSK₄ is significantly impaired in TLR1-deficient macrophages while the response to MALP-2 is normal in the mutant cells, indicating that TLR1 preferentially recognizes triacylated configuration of the lipid moiety [10].

2.4 Double stranded RNA recognition by TLR3

2.5 Flagellin recognition by TLR5

2.6 Anti-viral chemical compounds and TLR7

2.7 CpG DNA and TLR9

IL-1R and TLR family utilize the same signaling molecules. These include the adaptor molecule MyD88, the IL-1R-associated protein kinases (IRAKs), the TGF-β-activated kinase (TAK1), TAK1-binding protein 1 (TAB1) and 2 (TAB2), the tumor necrosis factor receptor-associated factor 6 (TRAF6) [22]. Triggering of IL-1R or TLR recruits the adaptor protein MyD88 to the receptor complex, which in turn enables the association with the IL-1R-associated kinases IRAK-4 and IRAK-1. During the formation of this complex, IRAK-4 is activated, and leads to the hyperphosphorylation of IRAK-1, which then induces interaction with TRAF6. The formation of IRAK-4/IRAK-1/TRAF6 causes conformational changes, and leaves the receptor complex. The IRAK-4/IRAK-1/TRAF6 then interacts with membrane-bound TAB2, and then the TAB2/TAK1/TAB1/TRAF6 complex translocates to the cytoplasm, leaving IRAKs. TAK1 is subsequently activated in the cytoplasm, leading to the activation of IκB kinase (IKK). Phosphorylated IRAK-1 stays at the membrane, and is eventually degraded. Activated IKK causes the activation of NF-κB through the phosphorylation and degradation of IκBs, which sequester NF-κB in the cytoplasms prior to activation. Activation of TAK1 also results in the activation of MAP kinases.

MyD88 is an adapter molecule that functions to recruit IRAK to the IL-1 or TLR4 receptor complexes after IL-1 or LPS stimulation, respectively. Indeed, MyD88-deficient mice are unresponsive to IL-1, LPS and other microbial cell wall components such as peptidoglycan and lipopeptides [23–25]. However, there is a difference in the signaling pathways triggered by LPS and by these latter stimuli. Mycoplasmal lipopeptide activation of NF-κB and MAP kinases, which is mediated by TLR2, is completely abolished in TLR2-deficient or MyD88-deficient macrophages. However, LPS activation of MAP kinases and NF-κB remains intact in MyD88-deficient macrophages, although activation is delayed compared with wild-type mice [24]. This indicates that the LPS response is mediated by both MyD88-dependent and -independent pathways, each of which leads to the activation of MAP kinases and NF-κB. Nevertheless, the MyD88-dependent pathway is essential for the inflammatory response mediated by LPS. In an attempt to characterize the MyD88-independent pathway, we tried to identify the genes induced in a MyD88-independent manner [26]. We generated cDNA library from MyD88-deficient macrophages before and after LPS stimulation and performed a subtractive suppression hybridization. This screening identified several genes induced in LPS-stimulated MyD88-deficient macrophages. Most were a CXC chemokine, IP-10, and others included GARG16 and IRG-1. These are so-called interferon inducible genes. Consistent with the lack of MyD88-independent pathway in the TLR2 signaling, wild-type macrophages did not induce IP-10 mRNA in response to mycoplasmal lipopeptide. It is shown that IFN inducible genes are induced by type I interferons. Analyses with IFN-$\tilde{\alpha }\beta$ R KO mice and STAT1 KO mice showed that IP-10 induction is secondary to IFN-β induction in response to LPS [27]. In contrast, GARG16 was induced independently of MyD88 and IFN-β. A recent paper has shown the same conclusion [28].

Previous studies show that interferon inducible genes including IP-10 are regulated by IRF-3. After viral infection, treatment with dsRNA, or LPS stimulation, IRF-3 undergoes phosphorylation and translocates from the cytoplasm to the nucleus, resulting in transcriptional activation of interferon-inducible genes. As expected, the ISRE binding activity and IRF-3 activation were induced in a MyD88-independent manner [24]. Fig. 3 summarizes the difference in the signaling pathway and gene expression between LPS and mycoplasmal lipoprotein.

In addition to the induction of IFN-inducible genes, the MyD88-independent pathway is also involved in the LPS-mediated maturation of DCs, which accompanies the induction of costimulatory molecules and production of inflammatory cytokines [29]. When MyD88-deficient bone marrow-derived DCs were cultured with LPS, they augmented surface expression of co-stimulatory molecules such as CD40, CD80 and CD86, and activated T cells to proliferate, whereas TLR4-deficient DCs failed to maturate in response to LPS, indicating that DC maturation proceeds without MyD88 in response to LPS. Further studies showed that activation of either MyD88-dependent or MyD88-independent pathway is enough for induction of costimulatory molecules. Nevertheless, MyD88-independent induction of costimulatory molecules is mainly secondary to IFN-β induction as is the case of interferon-inducible genes [27].

Another TIR domain-containing protein (TIRAP, also named Mal) has been identified by a database search. TIRAP was shown to associate with TLR-4 but not other TLRs [30,31]. Overexpression of a dominant negative form of TIRAP inhibited LPS-induced NF-κB activation or DC maturation. Blockage of TIRAP using a cell-permeable inhibitory peptide also prevented induction of IP-10 by LPS [28]. Thus, it was expected that TIRAP/Mal might participate in the MyD88-independent pathway. However, studies with TIRAP-deficient mice have revealed that TIRAP/Mal is not involved in the MyD88-independent pathway, and acts as an adaptor in the MyD88-dependent signaling pathways via TLR2 and TLR4 [32,33]. Recently, another adaptor molecule named TRIF or TICAM-1 has been identified [34,35]. Overexpression of TRIF, but not MyD88 or TIRAP, preferentially activates the promoter of the interferon β. A dominant negative form (DN) of TRIF, but not MyD88DN or TIRAPDN, blocked poly I:C-mediated TLR3 response, indicating the specific role of this novel adaptor in TLR3 signaling. TRIFDN also blocked the signaling pathways activated by other TLR family, which indicates some role of TRIF in MyD88-dependent pathway. Further analyses including generation of TRIF knockout mice will be required for the precise role of TRIF in the TLR signaling.

TLR3 signaling mainly depends on the MyD88-independent pathway. Poly I:C induce activation of NF-κB and MAP kinases in MyD88-deficient macrophages almost to the same extent as wild-type macrophages (our unpublished data). Dominant negative form of MyD88 or TIRAP/Mal did not abolish the poly I:C-mediated TLR3 response. There is no any interaction detected between MyD88 or Mal and TLR3 upon poly I:C stimulation. Recently, a paper reported that TLR3-mediated activation of NF-κB and MAP kinses is through an IRAK-independent pathway, employing signaling components TLR3-TRAF6-TAK1-TAB2-PKR [36]. TRIF is a likely candidate adaptor to recruit TRAF6-TAK-1-TAB2 to TLR3. In fact, TRIF harbors three TRAF6 binding motifs in the N-terminus. More recently, two papers independently showed that non-canonical IκB kinases, IKKε (also called IKKi) and TBK1 are involved in IRF3 activation, and that they may function downstream of adaptor TRIF, which specifically activates the IFN-β promoter.

A database search of TIR domain-containing adaptors has revealed two additional adaptor molecules (our unpublished data). These findings suggest that several TIR domain-containing adaptor molecules are involved in the TLR-mediated signaling pathways and differential utilization may provides the specificity in the TLR signaling (Fig. 4).

The exposure of macrophages to LPS induces a hypo-responsive state to a second challenge with LPS that is termed LPS tolerance [37]. LPS tolerance in macrophages/monocytes has been investigated extensively and shown to be characterized by a decline in the production of inflammatory cytokines including TNF-α, IL-6 and IL-12 and alterations to the LPS-induced activation of signaling cascades including protein kinase C, PI3K, mitogen-activated protein (MAP) kinases, and IκB kinases. However, the precise mechanism responsible for LPS tolerance remains unclear. We and others previously showed that down-regulation of the surface expression of the TLR4-MD2 complex and TLR4 caused LPS-induced LPS tolerance in mouse peritoneal macrophages and THP-1 human monocytic cell lines, respectively [38,39]. However, LPS tolerance induced by lipopeptides and LTA (cross tolerance), both of which are recognized by TLR2, has been found to occur through a distinct mechanism from LPS-induced LPS tolerance [40,41]. These findings imply that LPS tolerance was differentially induced by each TLR ligand. We analyzed the induction of LPS tolerance by several microbial components in mouse peritoneal macrophages [40,42]. LPS (TLR4) signaling is composed of the MyD88-dependent cascade that is essential for production of inflammatory cytokines and the MyD88-independent cascade that mediates the expression of IFN-inducible genes, whereas TLR2 signaling involves the MyD88-dependent cascade alone. TLR3 signaling is composed of the MyD88-independent pathway. Pre-exposure to LPS led to impaired activation of both the pathways. In contrast, pre-treatment with mycoplasmal lipopeptides (TLR2 ligand) did not affect the MyD88-independent pathway, but impaired the MyD88-dependent signaling by inhibiting LPS-mediated activation of IRAK-1. Pre-treatment with double-stranded RNA (TLR3 ligand), led to defective activation of the MyD88-independent, but not the MyD88-dependent, pathway. Imidazoquinoline compounds, TLR7 ligand, had no effect on the MyD88-independent pathway, but inhibited LPS-induced activation of MyD88-dependent signaling through down-regulation of IRAK-1 expression. Thus, cross-tolerance is induced in such a way that a given pathway, once activated, is refractory to subsequent activation, which may explain the complicated nature of LPS tolerance (Fig. 5).