Review
Conservation and Divergence of Ligand Recognition and Signal Transduction Mechanisms in Toll-Like Receptors
2017 Volume 65 Issue 8 Pages 697-705
Details
2017 Volume 65 Issue 8 Pages 697-705
Toll-like receptors (TLRs) play a central role in innate immunity as pathogen sensors. During the last decade, structural analyses of TLRs have revealed the mechanisms of ligand recognition and signal transduction. Each TLR recognizes its cognate ligand in a different manner, whereas signal transduction is achieved by a common mechanism. In this review, the mechanisms of ligand recognition and signal transduction by TLRs are summarized based on recent structural information.
The innate immune system is critical for protection against constant challenges by foreign microorganisms. It rapidly senses the invasion of foreign bodies and activates primary immune responses and subsequent acquired immunity. Innate immune receptors, called pattern recognition receptors (PRRs), such as Toll-like receptors (TLRs), nucleotide-binding oligomerization domain-like receptors, and retinoic acid-inducible gene 1-like receptors, recognize pathogen-associated molecular patterns (PAMPs) as danger signals and transmit signals to downstream molecules to evoke innate immunity, such as inflammation and antiviral responses.1) Among PRRs, TLRs are the best studied; recent structural analyses have revealed their ligand recognition and signal transduction mechanisms.
TLRs are type-I transmembrane receptors, composed of an extracellular leucine-rich repeat (LRR) domain, a transmembrane region, and an intracellular Toll-interleukin-1 receptor (TIR) domain.2,3) The LRR is a structural motif in which one LRR unit with about 20 to 30 residues consists of a β-strand and an α-helix or loop structure and typically forms a horseshoe-shaped structure (Fig. 1). The LRR domain of TLRs is composed of 20 to 26 LRR units and is responsible for ligand binding, and the intracellular TIR domain is responsible for downstream signal transduction. In general, it is believed that the unliganded TLR presents as a monomer and exhibits dimerization in response to agonistic ligands; subsequently, the intracellular TIR domain associates with and activates downstream adaptor molecules for signal transduction.4)
Horseshoe structure of human TLR2 (PDB 2Z7X)24) (top) and structures of representative LRR units (LRR4 and LRR7) (bottom). The consensus sequence of the LRR motif is shown. Residues forming the hydrophobic core of the LRR structure are shown as stick representations.
Ten TLRs (TLR1 to TLR10) have been identified in humans and 12 (TLR1 to TLR9 and TLR11 to TLR13) have been detected in mice. Each TLR recognizes specific PAMPs derived from different pathogens as activating ligands5) (Table 1), including PAMPs in lipopeptides (as ligands for TLR1, TLR2, and TLR6),6–9) double-stranded (ds) RNA (for TLR3),10) lipopolysaccharide (LPS) (for TLR4 and its co-receptor MD-2),11–14) flagellin (for TLR5),15,16) single-stranded (ss) RNA (for TLR7 and TLR8),17–19) DNA with an unmethylated CpG dinucleotide motif (for TLR9),20) and 23S ribosomal RNA (rRNA) (for TLR13; mouse only).21,22)
| TLR | Dimerization | Ligand |
|---|---|---|
| TLR1 and TLR2 | Hetero- | Triacyl-lipopeptide |
| TLR2–TLR6 | Hetero- | Diacyl-lipopeptide |
| TLR3 | Homo- | dsRNA |
| TLR4–MD-2 | Homo- | LPS |
| TLR5 | Homo- | Flagellin |
| TLR7 | Homo- | ssRNA, guanosine, imidazoquinolines |
| TLR8 | Homo- | ssRNA, uridine, imidazoquinolines |
| TLR9 | Homo- | CpG ssDNA |
| TLR13 | Homo- | 23S ribosomal RNA |
Members of the TLR2 subfamily of TLRs (TLR1, TLR2, and TLR6)23) are involved in the recognition of lipopeptides derived from bacteria.6–9) These receptors uniquely recognize lipopeptides by forming 1 : 1 : 1 heterodimers, i.e., TLR1 and TLR2 form a heterodimer to recognize triacyl-lipopeptide24) and TLR2 and TLR6 form a heterodimer to recognize diacyl-lipopeptide25) (Figs. 2a, b). The first crystal structure of an agonist-bound activated TLR was determined for TLR1–TLR2–triacyl-lipopeptide24) (Fig. 2a). TLR1 and TLR2 associate in a tail-to-tail fashion in which their C-terminal regions face each other. This enables the subsequent association of the intracellular TIR domains, resulting in the activated form of the TLR dimer. The three acyl chains of the lipopeptide are inserted into the hydrophobic channel formed at the central region of the LRR domain of TLR1 and TLR2; one and two of the three acyl chains of the lipopeptide are recognized by TLR1 and TLR2, respectively24) (Fig. 3a).
(a) TLR1/TLR2/triacyl-lipopeptide (PDB 2Z7X),24) (b) TLR2/TLR6/diacyl-lipopeptide (PDB 3A79),25) (c) TLR3/dsRNA (PDB code 3CIY),27) (d) TLR4/MD-2/LPS (PDB 3FXI),38) (e) TLR5/FliC (PDB 3V47),45) (f) TLR7/ssRNA/guanosine (PDB 5GMF),60) (g) TLR8/ssRNA/uridine (PDB 4R07),62) (h) TLR9/CpG DNA (PDB 3WPC),64) and (i) TLR13/ssRNA13 (PDB 4Z0C).74) Agonistic ligands are displayed in magenta, and regions from fusion partners in gray.
In addition to the hydrophobic interactions, the peptide moieties of the lipopeptide interact with each TLR by hydrogen bonds and van der Waals forces. TLR2 also forms a heterodimer with TLR6 to recognize diacyl-lipopeptide25) (Fig. 2b). Because the hydrophobic channel of TLR6, which accommodates the acyl chain, is blocked by Phe343, the two acyl chains of the ligand bind to TLR2, and TLR6 exhibits additional hydrophobic bonding and van der Waals interactions with the peptide moiety of the ligands25) (Fig. 3b). Interestingly, Streptococcus pneumoniae lipoteichoic acid and the synthetic phospholipid derivative 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-diethylenetriaminepentaacetic acid (PE-DTPA), which lack the peptide moiety at the head group, exhibit high affinity for TLR2 but are not able to induce TLR2–TLR1 or TLR2–TLR6 hetrodimerization.25) Therefore, the peptide moiety of the lipopeptide is important for the heterodimerization of TLR2 with TLR1 or TLR6 by mediating interactions with the two TLRs, thereby providing specificity. Recently, the crystal structure of TLR2 in complex with a potent antagonist, staphylococcal superantigen-like protein 3 (SSL3) secreted from Staphylococcus aureus, has been reported; SSL3 binds near the lipopeptide-binding site of TLR2 and inhibits both lipopeptide binding to TLR2 and heterodimerization with TLR1 or TLR626) (Fig. 3c).
TLR3 recognizes dsRNA from viruses.10) TLR3 homodimerizes with dsRNA to form a 2 : 1 complex27,28); this has been confirmed by the crystal structure of mouse TLR3 in complex with 46-basepair (bp) dsRNA27) (Fig. 2c). The crystal structure reveals that interactions between TLR3 and dsRNA are mostly electrostatic. The two positively charged surfaces of TLR3 make contact with the negatively charged phosphate backbone of dsRNA (binding sites 1 and 2) (Fig. 4). TLR3–dsRNA involves no base-specific interactions. Therefore, TLR3 does not exhibit a sequence preference for dsRNA. TLR3–dsRNA interactions are pH dependent. TLR3 can interact with dsRNA only under acidic conditions. Several His resides with neutral pKa values involved in the interaction can account for the dependence on pH (Fig. 4). Although dsRNA must be at least 40–50 bp to bridge two TLR3 molecules and form a minimal 2 : 1 signaling complex, longer dsRNA could bind to multiple TLR3 molecules.28) It was proposed that the lateral clustering of multiple TLR3 molecules at one dsRNA is required for the efficient signal transduction of TLR3.29)
TLR3 residues involved in interaction with ssRNA in the TLR3/ssRNA complex (PDB 3CIY)27) are shown in stick representations. ssRNA-binding sites of TLR3 are indicated by dashed circles.
TLR4 in cooperation with its accessary protein MD-2 recognizes LPS, a component of the outer membrane of Gram-negative bacteria.11–14) Owing to its therapeutic potential for lethal endotoxin-induced shock, TLR4–MD-2 is one of the most extensively studied TLRs. LPS consists of an O-specific chain, a core oligosaccharide, and lipid A. The lipid A moiety is composed of a D-glucosamine β(1→6) disaccharide 1,4′-diphosphate backbone to which several acyl chains are covalently linked and is responsible for the immunostimulatory activity of LPS.30) The number of acyl chains is closely related to the immunostimulatory potency of LPS.31–33) For example, Escherichia coli LPS typically has six acyl chains and acts as a potent agonist against both human and mouse cells expressing TLR4–MD-2, while tetraacylated LPS (lipid IVa) acts as a weak agonist in mouse cells, but as an antagonist in human cells.34,35) MD-2, a soluble protein of about 160 amino acid resides, forms a stable 1 : 1 complex with the extracellular domain of TLR4, which functions collectively as an LPS receptor.36,37) The crystal structure of TLR4–MD-2–LPS (E. coli) reveals the 2 : 2 : 2 architecture of the complex, where LPS-bound MD-2 mediates the dimerization of two TLR4 molecules in the m-shaped activated configuration38) (Fig. 2d). The large hydrophobic pocket formed between the two β-sheets of MD-2 provides a major binding site for hydrophobic acyl chains of LPS (Fig. 5a). In the TLR4–MD-2–LPS complex, five of the six acyl chains of LPS are completely buried in the pocket. One acyl chain is partially exposed to the solvent on the shallow side of the MD-2 pocket, and this region contributes to hydrophobic interactions with TLR4* (where the asterisk identifies an alternative protomer in the dimer throughout this review) for dimerization38) (Fig. 5a). The species-specific difference in TLR4–MD-2 activation by tetraacylated lipid IVa can be explained by the recognition pattern of acyl chains of lipid IVa in the MD-2 pocket39); all acyl chains are completely buried in the cavity in human TLR4–MD-2, while one acyl chain is partially exposed to the solvent in mouse TLR4–MD-2 and can act as a hydrophobic patch on the MD-2 surface for interactions with TLR4* (Fig. 5b). The different binding mode of lipid IVa can be ascribed to the difference in the pocket shape and charge distribution of MD-2. In addition, the difference in the charge distribution on the TLR4 surface between human and mouse TLR4 would contribute to the species-specific TLR4–MD-2 activation by lipid IVa. Many reports described endogenous and exogenous molecules other than LPS that potentially activate TLR4–MD-2.40) Recently, the crystal structure of mouse TLR4–MD-2 in complex with the synthetic agonist Neoseptin-3, with no similarity to LPS, has been determined41) (Fig. 5c), where two Neoseptin-3 molecules occupy the MD-2 pocket in a completely different manner to LPS but induce the same activated form of the TLR4–MD-2 dimer, providing structural evidence that therapeutic intervention targeting the activation process of TLR4–MD-2 is possible.
TLR5 is the only TLR that can be activated by a protein ligand, bacterial flagellin.15,16,42) Flagellin is a main component of bacterial flagella and consists of the D0, D1, D2, and D3 domains43) (Fig. 6a). The conserved surface of the D1 domain of flagellin mediates host inflammatory responses via TLR5.44) The crystal structure of the truncated form of zebrafish TLR5 containing LRR1–14 as a variable lymphocyte receptor hybrid protein in complex with the D1/D2/D3 fragment of Salmonella flagellin (FliC) reveals the overall architecture of the 2 : 2 complex45) (Fig. 2e). The D1 domain of FliC consists of three α-helices, one β-hairpin, and a helix bundle on one side that interacts with the ascending lateral side of the LRR N-terminus (LRRNT) to LRR10 of one TLR5 protomer, with a large contact area of ca. 1320 Å2 (primary interface) (Fig. 6b). The other side of the D1 surface interacts with the second TLR5 protomer at LRR12 and LRR13, with a relatively small contact area of ca. 260 Å2, in addition to the TLR5–TLR5′ interaction at LRR9, LRR12, and LRR13, with a contact area of ca. 260 Å2, contributing to TLR5 dimerization (dimerization interface) (Fig. 6b). The contact surface area between TLR5 and FliC is the largest among the activated forms of TLR–ligand complexes, although the crystallized construct of TLR5 lacks approximately the C-terminal half of the extracellular domain. It should be noted that the 2 : 2 complex was only observed in the crystal structure using these truncated proteins; thus, additional interactions between the full-length extracellular domain of TLR5 and flagellin occur (Fig. 6a).
(a) Entire dimer model of the TLR5/FliC complex based on the structure of the truncated form of TLR5 in complex with the D1–D3 domains of FliC (PDB 3V47).45) Missing regions in the crystal structure are indicated by dashed lines. (b) TLR5 residues involved in interaction with FliC are shown as stick representations. Primary and dimerization interfaces are indicated by dashed circles.
TLR7, TLR8, and TLR9 form the TLR7 subfamily of TLRs, which recognize ssRNA or DNA.23) TLR7 and TLR8 are considered receptors for viral or bacterial ssRNA.17–19) TLR7 and TLR8 are also activated by small chemical ligands, such as imidazoquinoline compounds and nucleotide analogues.18,46,47) TLR9 can be activated by DNA containing the CpG motif.20) Members of the TLR7 subfamily have 26 LRR units and possess a predicted loop region between LRR14 and LRR15 with 40–50 amino acid residues (referred to as the Z-loop)3,48) (Fig. 7a). Accumulating evidence indicates that the proteolytic processing of the Z-loop is essential for the activation of this family of TLRs.49–54) After processing, whether the resultant N-terminal half can be separated from the C-terminal half has been controversial. The C-terminal half alone of TLR9 was reported to be functional in some studies,51,54) while the results of other studies suggested that both fragments are important for the activation of TLR9.55,56) The first crystal structure determined for a TLR7 subfamily member was that of TLR8.48) Although the purified TLR8 protein was cleaved at the Z-loop, the N- and C-terminal fragments remained associated during purification steps, suggesting a tight interaction. In accordance with this observation, the crystal structure of TLR8 showed the ring structure of the LRR, in which the N- and C-terminal halves associate with each other. LRR14 and LRR15 are interrupted by the Z-loop, but form a continuous β-sheet at the concave surface, and the N- and C-terminal regions interact directly. Moreover, the ordered region of the Z-loop in the C-terminal fragment makes extensive contact with the concave surface of the N-terminal fragment. The unliganded and small chemical ligand-bound forms of TLR8 are dimeric in both the crystal and solution structures (Figs. 7b, c). The ligand binds to two equivalent positions in the dimerization interface, and ligand binding induces a structural rearrangement of the dimer, in which the two C-terminal regions are in close proximity48) (Figs. 7b, c). The binding site consists of the N-terminal fragment of one TLR8 protomer and the C-terminal fragment of the other TLR8* protomer (Fig. 7d), strongly suggesting that both fragments are required for ligand recognition after the proteolytic processing of TLR8.
(a) Monomer structure of TLR8 with an uncleaved Z-loop (PDB 5HDH),59) and dimer structures of (b) unliganded form of TLR8 (PDB 3W3G)48) and (c) R848-bound form of TLR8 (PDB 3W3N).48) (d) Ligand recognition at the first sites of TLR7 and TLR8. Guanosine in the TLR7/ssRNA/guanosine complex (top left, PDB 5GMF),60) uridine in the TLR8/ssRNA/uridine complex (top right, PDB 4R07),62) R848 in the TLR7/R848 complex (bottom left, PDB 5GMH),60) and R848 in the TLR8/R848 complex (bottom right, PDB 3W3N).48) (e) ssRNA recognition at the second sites of TLR7 (top)60) and TLR8 (bottom),62) viewed from the bottom side of the dimer structures.
Surprisingly, the co-crystal structure of TLR8 and ssRNA reveals two distinct ligand-binding sites in TLR8, one for uridine mononucleoside and the other for short ssRNA57) (Fig. 2g). The first binding site corresponds to the chemical ligand-binding site and the second binding site is located at the concave surface of TLR8, which is outside the dimerization interface (Fig. 7e). Thus, ssRNA binding to TLR8 itself would not induce the activated form of TLR8. Biophysical analysis of the interaction between TLR8 and several mononucleosides revealed a TLR8 preference for uridine. Although only uridine exhibits an affinity for TLR8, its affinity (50 µM) is far weaker than that for small chemical ligands (0.2 µM). To compensate for the low affinity for uridine at the first site, TLR8 utilizes the second binding site for ssRNA for the synergistic activation of TLR8. The binding affinity for uridine was markedly enhanced (1 µM) in the presence of ssRNA to a level comparable to those for small chemical ligands. Ligand binding to the first site of TLR8, e.g., the binding of uridine and small chemical ligands, triggers the structural rearrangement of TLR8 to the activated form; this explains why both ssRNA and small chemical ligands, despite differences in size and chemical properties, can activate the same receptor. According to structural analysis, the 20-mer ssRNA is degraded to uridine and small fragments of ssRNA, which bind to the first and second sites, respectively (Fig. 2g). Similar degradation would occur in cells with TLR8 activation. This clearly explains why ssRNA lacking uridine cannot activate TLR8.58)
The significance of Z-loop cleavage for TLR8 activation was clearly demonstrated in a structural and biochemical study of TLR8 with an uncleaved Z-loop59) (Fig. 7a). TLR8 with an uncleaved Z-loop is monomeric, irrespective of the presence of agonistic ligands such as ssRNA or chemical ligands. The uncleaved Z-loop is located on the surface at which TLR8 dimerizes (Fig. 7a). Therefore, the uncleaved Z-loop sterically occludes the dimerization partner to enter the appropriate position for dimerization. In line with this structural observation, ssRNA still binds to TLR8. The chemical ligand fails to bind to TLR8 with an uncleaved Z-loop because ligand binding to the first site of TLR8 is coupled with the dimerization of TLR8.
TLR7 and TLR8 are functionally related receptors and both recognize ssRNA as well as small chemical ligands17–19,23); TLR7 is expected to be activated in a manner similar to TLR8. Recently, the crystal structures of TLR7 in complex with agonistic ligands have been reported60) (Fig. 2f), revealing similarities and dissimilarities in the ligand recognition mechanisms of TLR7 and TLR8. Unlike the recombinant extracellular domain of TLR8, which forms a dimer irrespective of the presence of an agonistic ligand,48) that of TLR7 exists as a monomer in the absence of the ligand and dimerizes in response to the agonistic ligands.60) Guanosine and its derivatives activate TLR7 in combination with ssRNA.61) Accordingly, the activated dimer form of TLR7 is induced by simultaneous binding to ssRNA and guanosine, similar to TLR8. In addition, chemical ligands alone, such as R848, can also induce TLR7 dimerization. The guanosine- and chemical ligand-binding site of TLR7 (first site) is the same as the uridine- and chemical ligand-binding site of TLR8, and the residues forming the first site are mostly conserved between TLR7 and TLR8 (Fig. 7d). This high conservation is why some chemical ligands activate both TLR7 and TLR8. On the other hand, TLR7 and TLR8 show distinct nucleoside preferences at the first site; TLR7 prefers guanosine and TLR8 prefers uridine, although their recognition modes are quite similar60,62) (Fig. 7d). The discrimination of nucleosides at the first site can be explained by subtle differences in the size and electrostatic potential of the ligand-binding pocket. In contrast to the conserved nucleoside- or chemical ligand-binding site, the ssRNA-binding site of TLR7 (second site) is spatially and structurally distinct from that of TLR8 (Fig. 7e). The second site of TLR7 is located at the concave surface of the N-terminal side of the LRR horseshoe structure (LRR1–5) and accommodates the UUU moiety of ssRNA,60) while that of TLR8 is located at the concave surface of the middle region of the LRR structure (LRR9–13) and binds to the UG moiety of ssRNA.62) The bound UUU in TLR7 is surrounded by the ordered structure of the Z-loop and the loop structures of LRR2 and LRR5; only the middle U position is specifically recognized via multiple interactions with TLR7 (Fig. 7e). The importance of the middle U position for recognition by TLR7 was demonstrated biochemically, and hence it was proposed that ssRNA containing the U residue at the internal position acts as a ligand for TLR7. The characteristic feature of the second site of TLR7 is that a disulfide bond (Cys98–Cys475) is formed between LRR2 and the Z-loop and is important for the formation of the second site (Fig. 7e), as demonstrated by mutational analyses.63) Similar to the enhanced activation of TLR8 by uridine in the presence of ssRNA,62) guanosine, when used alone, cannot activate or induce dimerization of TLR7 effectively, according to the results of an nuclear factor-kappaB (NF-κB) reporter assay and biochemical assay using recombinant proteins, but can induce the dimerization of TLR7 and activate TLR7 when used with ssRNA.60,61)
TLR9 is another member of the TLR7 subfamily and recognizes the DNA-containing CpG motif.20) The extracellular domain of TLR9 exists as a monomer in the absence of agonistic DNA, similar to TLR7.64) The ring-shaped monomer structure of TLR9 is very similar to those of TLR7 and TLR8.64) TLR9 with the uncleaved Z-loop still can bind CpG DNA but fails to dimerize, while TLR9 with the Z-loop artificially cleaved by V8-protease can dimerize, depending on the agonistic DNA,64) further supporting the inhibitory role of the Z-loop for dimerization, as in TLR8.59) The structure of the Z-loop of TLR9 after processing is mostly disordered and not involved in ligand binding,64) although those of TLR7 and TLR8 form an ordered structure at the concave surface of the LRR horseshoe structure and contribute to the formation of the second binding site for ssRNA binding60,62) (Fig. 7e). Overall, the agonistic CpG DNA-bound form of TLR9 forms a 2 : 2 complex with essentially the same configuration as that of the activated form of TLR7 and TLR848,60,62) (Fig. 2h). The bound CpG DNA, which contains a core CpG motif with the GAC GTT sequence, is sandwiched by two molecules of TLR9, thus acting as a mediator of TLR9 dimerization (Fig. 8a). The CpG dinucleotide is mainly recognized in the groove formed by the N-terminal region of one TLR9 protomer (LRRNT, LRR1, and LRR2), in which the two bases of the CpG motif stick into the groove and form multiple specific interactions with the protein (Fig. 8a). Moreover, the region flanking the CpG dinucleotide also contributes to the interaction. The importance of the CpG dinucleotide and its flanking region sequence for interactions with TLR9 was confirmed by determining the binding affinities for oligonucleotides with various sequences. The interactions between TLR9 and CpG DNA are sequence and pH dependent. Importantly, the introduction of a methyl group at the 5′ cytosine base of the CpG dinucleotide reduces the binding affinity for TLR9, thus resulting in reduced immunostimulatory activity.64) The interaction between TLR9 and CpG DNA is strong under acidic conditions and weak under basic conditions, possibly due to the involvement of His residues, as suggested for TLR3.27) CpG DNA is also recognized by the other TLR9* protomer via the backbone phosphate group, contributing to the induction of the dimerization of TLR9 (Fig. 8a). Interestingly, although the overall dimeric structure of TLR9 is very similar to those of TLR7 and TLR8, the oligonucleotide-binding sites are completely different.60,62,64) The species specificity of TLR9 activation by certain types of CpG DNA was reported.65–70) This species specificity cannot be explained by the current structural study, but it may be related to another DNA-binding site on TLR9.71–73) Further structural studies will be needed to establish the basis for the species specificity of TLR9 activation. The structures of the inhibitory DNA-bound form of TLR9 were also determined64) (Fig. 8b). Inhibitory DNA binds to the concave surface of the N-terminal region of TLR9 by forming a compact stem-loop conformation. The binding site partially overlaps with that of CpG DNA and therefore acts as an antagonist in a competitive manner.
Unlike humans, mice have TLR13, which recognizes bacterial ssRNA.21,22) The crystal structure of TLR13 in complex with 13-mer ssRNA (ssRNA13) was determined74) (Fig. 2i). Like most other TLRs, the unliganded form of TLR13 exists as a monomer and exhibits dimerization in response to the ligand. The monomer structure of TLR13 with 27 LRR units exhibits an oval-shaped structure in which the N- and C-termini make direct contact. The overall dimer structure of the TLR13–ssRNA13 complex exhibits a typical m-shaped configuration. ssRNA13 in the complex forms a stem-loop structure, where one C-G base pair is observed. This stem-loop structure is completely different from that in the 23S rRNA with an extended conformation and is important for TLR13 binding. By forming a compact stem-loop structure, ssRNA13 binds to the positively charged concave surface of TLR13 at the junction between the N- and C-terminal regions. Most of the RNA bases flip out for interactions with TLR13. RNA-specific 2′-OH groups also contribute to interactions with TLR13 and to the maintenance of the stem-loop structure, establishing the structural basis for ssRNA recognition in a sequence- and stem-loop structure-dependent manner. Similar to TLR3 and TLR9,27,64) TLR13–ssRNA binding is pH dependent, with increased interactions under acidic conditions.
In the past 10 years, structural studies of TLRs have exhibited an explosive increase. The function of each TLR has been characterized based on interactions between each ligand and TLR at the atomic level. This structural information has not only clarified their detailed ligand recognition mechanisms but has also made it possible to make predictions regarding interactions with unknown ligands. Thus, these findings are expected to promote the development of therapeutic agents aimed at the control of TLR function.
This review of the author’s work was written by the author upon receiving the 2016 Pharmaceutical Society of Japan Award for Young Scientists.
I express my profound gratitude to Prof. Toshiyuki Shimizu and Prof. Yoshinori Satow for their directions, continuous support, and encouragement. I thank my collaborators whose names appear in the references. This work was supported by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan; the Takeda Science Foundation; the Mochida Memorial Foundation for Medical and Pharmaceutical Research; the Daiichi Sankyo Foundation of Life Science; and the Naito Foundation.
The author declares no conflict of interest.
