Consistent with their role in pathogen recognition, TLR family members are expressed by cells involved in the first line of host defense, including neutrophils, macrophages, dendritic cells, dermal endothelial cells and mucosal epithelial cells. TLR2 and TLR4, which are the major receptors for bacterial lipoproteins and LPS, respectively, are also expressed on B and T cells, which mediate the more complex adaptive immunity via a large repertoire of antigen-specific immunoglobulin-class receptors [12]. With the exception of TLR9, which is an intracellular receptor, most TLRs are expressed on the cell surface. Interestingly, the subcellular localization of TLR4 has been shown to differ in macrophages and intestinal epithelial cells. In macrophages, TLR4 is expressed on the cell surface and is internalized following engagement of the ligand, whereas in epithelial cells it resides in the Golgi apparatus [13].

The first indication that TLRs may function as pattern-recognition receptors in mammals came from the discovery that the phenotype of lps mice, which do not show the systemic inflammatory response usually triggered by LPS, resulted from loss-of-function mutation in the TLR4 gene, thus establishing TLR4 as a receptor for the LPS of Gram-negative bacteria [14]. Subsequent studies using TLR4 knockout mice corroborated this finding [15]. Ex vivo studies, later confirmed by gene-knockout experiments, showed that there are additional agonists for TLRs. A variety of chemically diverse pathogen-associated molecular patterns are now known to be TLR agonists, as indicated in Figure 1. Studies using TLR2 and TLR4 knockout mice have revealed that these receptors mediate cellular responses to cell-wall components of Gram-positive and Gram-negative bacteria, respectively [16]. These components include LPS, other bacterial glycolipids, peptidoglycan, and bacterial lipoproteins. TLR4 has also been implicated in the recognition of coat proteins (for example, protein F) of the respiratory syncytial virus (RSV), which is a major cause of bronchiolitis in infants [17]. TLR3, TLR5 and TLR9 have been implicated in the recognition of viral double-stranded RNA, bacterial flagellin and bacterial CpG DNA, respectively [18,19,20]. Recently, TLR7 was shown to be involved in cellular activation by small anti-viral compounds, including imidazoquinoline compounds, although the natural ligand for this receptor remains unknown [21]. In addition to their role in pathogen recognition, recent evidence suggests that some TLRs respond to endogenous factors produced by stressed or damaged cells. These factors include heat-shock proteins (Hsps) [22,23,24], fragmentation products of the extracellular matrix components fibronectin and hyaluronan [25,26], and mammalian chromatin [27]. The putative capacity of TLRs to heterodimerize may explain, at least in part, how TLRs are capable of recognizing so many chemically diverse agonists.

Little is known about how microbial products activate TLRs. In Drosophila, Toll does not directly bind microbial products, but instead binds a proteolytic fragment of the secreted growth factor Spaetzle. The current view is that microbial antigens interact with a yet unidentified recognition receptor upstream of Toll to activate zymogens in the fly's hemolymph, ultimately leading to cleavage of Spaetzle to its active Toll ligand form [12]. In contrast, TLR4-dependent recognition of LPS by mammalian cells requires two accessory proteins: the high-affinity LPS acceptor protein CD14, a glycosylphosphatidylinositol (GPI)-anchored protein, and the small secreted protein MD2, which associates with the extracellular domain of TLR4 [28,29,30]. Physical contact between LPS and TLR4 was recently demonstrated [31,32], but to date, there is no evidence for direct ligand binding and recognition for all other TLRs.

Collectively, TLRs function to alert the immune system to the presence of microorganisms. Engagement of TLRs with their ligands leads to the production of various pro-inflammatory cytokines, chemokines, and effector molecules, depending on the cell type that is activated [33,34,35]. TLR knockout mice have been used to study the roles of TLRs in the immune response against different pathogens in vivo. For example, TLR2-deficient mice are highly susceptible to lethal infections with the sepsis-causing pathogen Staphylococcus aureus [36]. In contrast, infections with Haemophilus influenzae or with RSV persist longer in the lungs of TLR4-deficient mice than in those of control animals [17,37].

Activation of signaling through TIR domains results in recruitment of the cytoplasmic adaptor proteins MyD88 and TOLLIP (Toll interacting protein) [38] (Figure 2). The IL-1 receptor-associated kinases IRAK-1 and IRAK-2 interact with the death domain of MyD88, a motif found in many apoptosis-inducing signaling molecules, and are recruited to the TLR complex. TOLLIP can also recruit IRAKs to the complex, albeit with different kinetics. Upon recruitment, IRAK-1 and IRAK-2 associate with TNF receptor-associated factor 6 (TRAF6), another adaptor protein. Very recently, two groups [39,40] have identified another serine-threonine kinase, termed IRAK-4, as an early component of the TLR-signaling cascade, possibly acting upstream of IRAK-1. TRAF6 induces activation of TGF-β-activated kinase (TAK1) and the mitogen-activated protein (MAP) kinase kinase MKK6, which, in turn, activate the transcription factor NF-κB, the c-Jun N-terminal kinase (JNK) and the p38 MAP kinase. The importance of MyD88 and TRAF6 in TLR signaling has been confirmed by targeted gene disruption. Homologs of many molecules involved in TLR signaling in mammals have been identified in Drosophila.

An additional TLR4 signaling pathway was revealed through the observation that certain LPS-induced responses did not require MyD88 [41]. Subsequently, two groups [39,40] identified a molecule called TIR domain-containing adapter protein (TIRAP) or MyD88-adapter-like (Mal), which interacts with TLR4 and mediates MyD88-independent TLR4 signaling. Although downstream components of this pathway remain to be identified, engagement of TLR4 with its ligand was recently shown to induce the secretion of the anti-viral interferon-β (IFN-β) via a TIRAP/Mal-dependent, but MyD88-independent, mechanism [42]. Autocrine or paracrine production of IFN-β was also shown to be required for the expression of selected genes that could not be induced via engagement of TLR2, including inducible nitric oxide synthase (iNOS) or interferon inducible protein 10 (IP10; Figure 2).