Toll-like receptors (TLRs) and the interleukin-1 receptor superfamily (IL-1Rs) are integral to both innate and adaptive immunity for host defence1,2,3. These receptors share a conserved cytoplasmic domain4,5, known as the TIR domain. A single-point mutation in the TIR domain of murine TLR4 (Pro712His, the Lpsd mutation) abolishes the host immune response to lipopolysaccharide (LPS)6, and mutation of the equivalent residue in TLR2, Pro681His, disrupts signal transduction in response to stimulation by yeast and Gram-positive bacteria7. Here we report the crystal structures of the TIR domains of human TLR1 and TLR2 and of the Pro681His mutant of TLR2. The structures have a large conserved surface patch that also contains the site of the Lpsd mutation. Mutagenesis and functional studies confirm that residues in this surface patch are crucial for receptor signalling. The Lpsd mutation does not disturb the structure of the TIR domain itself. Instead, structural and functional studies indicate that the conserved surface patch may mediate interactions with the downstream MyD88 adapter molecule7,8,9,10,11, and that the Lpsd mutation may abolish receptor signalling by disrupting this recruitment.
A principal function of TIR domains is thought to be to mediate homotypic protein–protein interactions in the signal transduction process2. To elucidate the molecular basis of TIR domain signalling, we have determined the crystal structures of the TIR domains of human TLR1 and TLR2 (Table 1). The structures contain a central five-stranded parallel β-sheet (βA–βE) that is surrounded by a total of five helices (αA–αE) on both sides (Fig. 1a). To facilitate sequence and structure comparisons of this large family of protein domains, we use a residue numbering system based on the secondary structure elements observed in the current structures, similar to that adopted for Src homology (SH)-2 domains12. In the numbering system, a residue in a β-strand or α-helix is numbered according to its position in that strand or helix. The loops are named by the letters of the secondary structure elements that they connect. For example, the BB loop connects strand βB and helix αB. A residue in a loop is numbered according to its position in that loop, or according to its position relative to the nearest strand or helix. For example, the residue immediately preceding strand βA can be numbered as βA-1.
The structures reveal that the core TIR domain starts from the conserved (F/Y)DA amino-acid motif and ends roughly eight residues carboxy-terminal to the conserved FW motif (Fig. 1d). The alanine residue of the (F/Y)DA motif is the first residue in strand βA (the βA1 residue). The amino and carboxy termini of this core TIR domain are located within 14 Å of each other, suggesting that the TIR domain may be considered as a cassette. Most of the conserved residues are located in the hydrophobic core of the structure. There are large insertions or deletions in several loop regions in different TIR domains. Consequently, the sizes of the core TIR domains vary considerably—between 135 and 160 residues among the sequences reported so far.
Although the TIR domains of TLR1 and TLR2 share 50% amino-acid sequence identity, there are large conformational differences between the two structures (for example, helices αB, αC′, and αD) ( Fig. 1b). Noting that the sequence conservation among TIR domains is generally in the 20–30% range, as well as the variation in the sizes of the domains, our studies suggest that TIR domains may have a significant amount of sequence and structural diversity. This diversity may be crucial for the specificity in the signal transduction process, by ensuring the formation of the proper signalling complex among the many TLRs and IL-1Rs.
Three types of TIR domain interactions are possible for receptor signalling in animals. The first interface (which we call the R face) would mediate the oligomerization of receptor TIR domains, facilitated by the ligand-induced association of the extracellular domains of the receptors. The second interface (the A face, possibly equivalent to the R face in the receptor) would mediate the oligomerization of the TIR domains of the downstream adapter molecule (MyD88), which may be facilitated by death domain interactions in this molecule. It has been shown that MyD88 exists as dimers in solution11. The third interface (the S face) would mediate the association between the receptor and adapter TIR domains, and the formation of this TIR domain complex is critical for receptor signalling7,8,9,10,11. In Drosophila , the gene CG2078 has the same domain organization as MyD88 and may therefore be a functional homologue of MyD88, suggesting that the TIR domain signalling complex may be conserved between insects and mammals.
The interactions that take place at the R face are probably a main determinant of specificity in the receptor signalling process, in addition to those between the ligand and the receptor. This is supported by the observation that chimaeric receptor containing the extracellular domain of IL-1R type I (IL-1RI) and the intracellular domain of TLR1 cannot function in response to IL-1 (ref. 13), a possible explanation being that the TIR domains of the IL-1R accessory protein and TLR1 cannot form a proper complex. Therefore, it is likely that interactions at the R face may show a significant degree of diversity to allow for the specificity, consistent with the sequence and structural diversity that we have observed. Residues at this interface are probably not strictly conserved among the various receptor TIR domains.
In contrast, interactions at the S face should be primarily mediated by a highly conserved region among the TIR domains. Despite the identification of more than 20 TLRs and IL-1Rs in mammals, only one putative adapter molecule (MyD88) is currently known to contain a TIR domain7,8,9,10,11. Tollip has recently been identified as a receptor-proximal component of the IL-1RI pathway, but it does not contain a TIR domain14. In Drosophila, there are a total of nine Toll homologues, but there is only one putative MyD88 homologue (CG2078). For this large number of receptors to signal through a common adapter molecule, the receptors must present a conserved surface area (S face) for coupling to the downstream adapter.
To identify the surface feature of the TIR domains that may be important for the recruitment of MyD88 (the S face), we analysed the molecular surface of the TIR domain in terms of amino-acid sequence conservation. This analysis reveals a large conserved surface patch (Fig. 2a), which mostly contains the BB loop, with additional contributions from helix αA, strand βB and the aromatic side chain of the (F/Y)DA motif (the βA-2 residue). The BB loop extends away for the rest of the TIR domain, forming a protrusion on the surface of the structure (Fig. 1a). This loop contains three highly conserved residues, Arg BB3, Asp BB4 and Gly BB8, in the RDxΦ1Φ2G motif (where Φ represents a hydrophobic residue, and x any residue). The side chains of both hydrophobic residues are exposed in the structure (Fig. 2a). The Arg BB3 residue has ion-pair interactions with the Asp BB4 residue (asparagine in TLR1 and TLR6 only) and the strictly conserved glutamate residue near the end of helix αA (Glu αA13) (Fig. 2b). The conformational difference for the BB loop in TLR2 (Fig. 1b ) is probably due to the cacodylate modification of Cys βB4 during crystallization (see Methods), which is situated directly below the loop.
We performed functional studies to show that this conserved surface patch is crucial for signal transduction by the receptors. On the basis of the structural information, we mutated residues in this surface region and tested their functional effects using transfection assays with human TLR4 and Drosophila Toll10,15. These studies show that mutations at almost every position in this region lead to significantly reduced signalling activity of the receptors (Fig. 3a). Residues that were mutated include BB3, BB4, BB7 and BB8 in the BB loop, and residue αA13 (Fig. 3a ). In addition, it has been reported that mutation of the BB3 residue in IL-1RI (R to A)16 and the βA-2 residue of Drosophila Toll (F to I)5 also blocked receptor signalling. The ion-pair between Arg BB3 and Glu αA13 seems to stabilize the conformation of the BB loop. Notably, the double mutant, BB3 R to E and αA13 E to R, could not rescue the signalling activity (Fig. 3a). It is likely that this ion-pair is involved in additional interactions in the signalling complex, or that the double mutant did not fully restore the ion-pair interactions.
The importance of this surface patch for receptor signalling is also emphasized by the fact that the Lpsd mutation6 in murine TLR4, P712H, is at the Φ2 position in the BB loop (residue BB7). To understand the structural basis for the elimination of receptor signalling caused by this mutation6,7, we determined the crystal structure of the P681H mutant of the TLR2 TIR domain (Table 1). There are no significant structural differences between this mutant and the wild type (Fig. 1c). The mutation site is located at the tip of the BB loop, farthest from the rest of the TIR domain. The structures show that the proline residue does not have a special structural role. The lack of signalling by the Lpsd mutation is thus not due to disruption of the TIR domain structure itself, but rather to the disruption of a direct point of contact with other molecule(s), and specifically other TIR domains. Our observations are in contrast to those for the Fas death domain, where a naturally occurring mutation (V to N) eliminates signalling by destroying the native conformation of the domain17. The structural analysis is consistent with sequence and functional observations, as small hydrophobic residues (alanine, valine, isoleucine) are present at the BB7 position in several TIR domains (Fig. 1d). Drosophila Toll contains a valine residue at the BB7 position, and our studies showed that a V to P mutation has little effect on the function of the receptor (Fig. 3a). In contrast, a V to H mutation, equivalent to the Lpsd mutation of TLR4, as well as mutation to arginine or glutamate at the BB7 position, greatly reduced receptor function (Fig. 3a).
To understand further the molecular mechanism for the disruption of signalling by the Lpsd mutation and other changes of this conserved surface patch, we performed protein-binding assays between purified recombinant glutathione S-transferase (GST) fusion TIR domains and in vitro translated TIR domains. The binding assays show that there is a significant interaction between the TIR domains of MyD88 and TLR2, and also that the P681H mutant of TLR2 almost completely abolished this interaction (Fig. 3b ). At the same time, mutations in this surface patch do not seem to affect the oligomerization of the receptor (interactions at the R face), as shown by the binding assays with the G to V mutant at the BB8 position of Drosophila Toll (Fig. 3c). The experiment also shows that the previously documented interaction with the Pelle kinase was not affected by the mutation (Fig. 3c)18. Our biochemical data are supported by the observation that the intracellular domain of TLR4 containing the Lpsd mutation is a potent inhibitor of tumour-necrosis factor production in response to LPS stimulation as signalled by endogenous (full-length) TLR4 (ref. 19). Overall, the experimental data provide strong evidence that the R face is not disturbed by the Lpsd mutation. Both structural and functional studies therefore suggest that the conserved surface patch corresponds to the S face and that the Lpsd mutation abolishes receptor signalling by disrupting the recruitment of MyD88.
The backbone topology of the TIR domain is similar to that of the chemotaxis regulator CheY, as was proposed earlier20, as well as many other proteins. The structural similarity between the TIR domain and CheY is however limited to the strands of the β-sheet (see Supplementary Information). In addition, TIR domains do not contain a cation-binding site equivalent to that in CheY (ref. 21), and therefore have a different mechanism of action from the response regulators. Nonetheless, it is interesting to note that a domain of similar architecture is used for signal transduction in response to exogenous stimulus in bacteria and eukaryotes.
We performed gel-filtration and dynamic light-scattering experiments to characterize the oligomerization state of isolated TIR domains in solution. These studies indicate that the inherent affinity for self-association of the TIR domains is rather low (dissociation constant (Kd) in the millimolar range; data not shown). At high concentrations, dimers and tetramers of TIR domains are observed in solution. This low affinity is consistent with the observation that signal transduction through TIR domains requires receptor oligomerization, induced either by ligand binding or by overexpression7,8,9,10,11. It is likely that avidity may be important in the assembly of the TIR domain signalling complex. Our structural and functional information provide a molecular basis for understanding the mechanism and complexity of signal transduction by the TIR domains.
Protein production and crystallization
Details on the expression, purification and crystallization of the TIR domains of human TLR1 and TLR2 will be presented elsewhere. Briefly, the TIR domain of human TLR1 (residues 625–786, about 20 residues from the transmembrane region) was overexpressed in Escherichia coli and purified with nickel-agarose, cation exchange and gel-filtration chromatography. We purified the TIR domain of TLR2 (residues 626–784; about 16 residues from the transmembrane region) with cation exchange and gel-filtration chromatography. The P681H mutant of this domain was produced with the QuickChange mutagenesis kit (Stratagene) and sequenced to confirm the presence of the mutation. We obtained crystals of the TIR domain of TLR1 at 21 °C by the hanging-drop vapour-diffusion method. The reservoir solution contained 100 mM Tris (pH 8.0), 1.2 M NaH2PO4/K2HPO4, 5 mM dithiothreitol (DTT) and 20% glycerol. The crystals belong to space group P6422. The TIR domain of TLR2 was crystallized at 4 °C. The reservoir solution contained 100 mM cacodylate (pH 6.8), 10% PEG 8,000, 20% (v/v) DMSO, 200 mM MgCl2 and 5 mM DTT. The crystals belong to space group P6222.
We collected X-ray diffraction data using rotating anode and synchrotron radiation sources (32-ID beamline (ComCAT) at APS, X4A beamline at NSLS and A-1 beamline at CHESS). The diffraction images were processed and scaled with the HKL package22. For TLR1, phases were obtained from the selenomethionyl multiwavelength anomalous diffraction (MAD) method23 and the multiple isomorphous replacement (MIR) method24. The atomic model was built with the program O (ref. 25). We carried out the structure refinement with the program CNS26. There is a disulphide bond between Cys 707 and its twofold symmetry mate, which was confirmed by non-reducing SDS–PAGE and MALDI-TOF analysis on the crystals. A comparison with the structure of the TIR domain of TLR2 shows that the disulphide bond did not introduce significant conformational changes in the TIR domain of TLR1.
We determined the initial structure of the TIR domain of TLR2 by the combined molecular replacement protocol as implemented in the Replace package27, using the structure of TIR domain of TLR1 as the model. The cysteine side chains had been modified by the cacodylate buffer, and the anomalous diffraction contained contributions from both Se and As atoms. The acentric reflections were phased with the anomalous diffraction data, and the centric reflections were phased with the molecular replacement solution. See Supplementary Information for more details on the structure determination.
There is only one molecule of the TIR domain in the asymmetric unit for crystals of both TLR1 and TLR2. This gives rise to a solvent content of about 80% and a unit cell volume to mass ratio of 5.5 Å3 per dalton for these crystals.
Functional studies with human TLR4 and Drosophila Toll
A truncated version of human TLR4, containing residues 558–825 and lacking the leucine-rich-repeat domain, was used for the mammalian functional studies as it has a higher constitutive activity. The mutants were made by site-directed polymerase chain reaction (PCR) mutagenesis and cloned into the pFLAG CMV1 vector. The receptor activity was measured by its ability to activate an NF-κB-dependent luciferase reporter after transient transfections of 2 µg of DNA into 293T cells10. The T110b mutant of Toll, which contains a C to Y mutation in its extracellular domain5, was used for the functional studies in Drosophila cells as it has a higher constitutive activity15. The mutants were made by site-directed mutagenesis with the QuickChange kit (Stratagene). We assayed signalling activity by co-transfection of 2 µg of Toll and 0.4 µg of dorsal expression vectors and a dorsal-dependent chloramphenicol acetyl transferase (CAT) reporter into Drosophila Schneider cells15. In both systems, all mutants were sequenced to confirm the presence of the respective mutations, and the assays were repeated several times to ensure reproducibility. Equivalent expression levels of the wild-type and mutant proteins were verified by western blotting.
Protein binding assays
The TIR domain of MyD88 was purified as a GST fusion protein, immobilized with glutathione-agarose and incubated with [35S]methionine-labelled wild type and P681H mutant of the TIR domain of TLR2 that were obtained by in vitro translation (Promega TNT system). For Drosophila Toll, the purified intracellular domain (GST–Toll IC) and GST–Pelle (K240R mutant) were incubated with [35S]methionine-labelled intracellular domains of wild-type Toll and the G to V mutant at the BB8 position18. After washing, the bound proteins were eluted and separated by SDS–PAGE. Gels were fixed and exposed to X-ray film.
Hoffmann, J. A., Kafatos, F. C., Janeway, C. A. Jr & Ezekowitz, R. A. B. Phylogenetic perspectives in innate immunity. Science 284, 1313–1318 (1999).
Kopp, E. B. & Medzhitov, R. The Toll-receptor family and control of innate immunity. Curr. Opin. Immunol. 11, 13–18 (1999).
Anderson, K. V. Toll signaling pathways in the innate immune response. Curr. Opin. Immunol. 12, 13–19 ( 2000).
Gay, N. J. & Keith, F. J. Drosophila Toll and IL-1 receptor. Nature 351, 355– 356 (1991).
Schneider, D. S., Hudson, K. L., Lin, T.-Y. & Anderson, K. V. Dominant and recessive mutations define functional domains of Toll, a transmembrane protein required for dorsal-ventral polarity in the Drosophila embryo. Genes Dev. 5, 797–807 ( 1991).
Poltorak, A. et al. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282, 2085 –2088 (1998).
Underhill, D. M. et al. The Toll-like receptor 2 is recruited to macrophage phagosomes and discriminates between pathogens. Nature 401, 811–815 (1999).
Muzio, M., Ji, J., Feng, P. & Dixit, V. M. IRAK (Pelle) family member IRAK-2 and MyD88 as proximal mediators of IL-1 signaling. Science 278, 1612–1615 ( 1997).
Wesche, H., Henzel, W. J., Shilinglaw, W., Li, S. & Cao, Z. MyD88, an adapter that recruits IRAK to the IL-1 receptor complex. Immunity 7, 837–847 (1997).
Medzhitov, R. et al. MyD88 is an adaptor protein in the hToll/IL-1 receptor family signaling pathways. Mol. Cell 2, 253– 258 (1998).
Burns, K. et al. MyD88, an adapter protein involved in interleukin-1 signaling. J. Biol. Chem. 273, 12203– 12209 (1998).
Eck, M. J., Schoelson, S. E. & Harrison, S. C. Recognition of a high-affinity phosphotyrosyl peptide by the Src homology-2 domain of p56lck. Nature 362, 87–91 (1993).
Mitcham, J. L. et al. T1/ST2 signaling establishes it as a member of an expanding interleukin-1 receptor family. J. Biol. Chem. 271, 5777–5783 (1996).
Burns, K. et al. Tollip, a new component of the IL-1RI pathway, links IRAK to the IL-1 receptor. Nature Cell Biol. 2, 346–351 (2000).
Norris, J. L. & Manley, J. L. Selective nuclear transport of the Drosophila morphogen dorsal can be established by a signaling pathway involving the transmembrane protein Toll and protein kinase A. Genes Dev. 6, 1654–1667 (1992).
Slack, J. L. et al. Identification of two major sites in the type I interleukin-1 receptor cytoplasmic region responsible for coupling to pro-inflammatory signaling pathways. J. Biol. Chem. 275, 4670– 4678 (2000).
Huang, B., Eberstadt, M., Olejniczak, E. T., Meadows, R. P. & Fesik, S. W. NMR structure and mutagenesis of the Fas (APO-1/CD95) death domain. Nature 384, 638–641 (1996).
Shen, B. & Manley, J. L. Phosphorylation modulates direct interactions between the Toll receptor, Pelle kinase and Tube. Development 125, 4719–4728 (1998).
Du, X., Poltorak, A., Silva, M. & Beutler, B. Analysis of Tlr4-mediated LPS signal transduction in macrophages by mutational modification of the receptor. Blood Cells Mol. Dis. 25, 328– 338 (1999).
Rock, F. L., Hardiman, G., Timans, J. C., Kastelein, R. A. & Bazan, J. F. A family of human receptors structurally related to Drosophila Toll. Proc. Natl Acad. Sci. USA 95, 588–593 (1998).
Stock, J. B., Stock, A. M. & Mottonen, J. M. Signal transduction in bacteria. Nature 344, 395–400 ( 1990).
Otwinowski, Z. & Minor, W. processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 ( 1997).
Hendrickson, W. A. Determination of macromolecular structures from anomalous diffraction of synchrotron radiation. Science 254, 51– 58 (1991).
Furey, W. & Swaminathan, S. PHASES-95: A program package for processing and analyzing diffraction data from macromolecules. Methods Enzymol. 277, 590–620 (1997).
Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119 (1991).
Brunger, A. et al. Crystallography & NMR System: A new software suite for macromolecular structure determination. Acta Crystallogr. D 54, 905–921 (1998).
Tong, L. Combined molecular replacement. Acta Crystallogr. A 52, 905–921 (1998).
Carson, M. Ribbon models of macromolecules. J. Mol. Graph. 5, 103–106 (1987).
Nicholls, A., Sharp, K. A. & Honig, B. Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins Struct. Funct. Genet. 11, 281–296 (1991).
We thank K. D'Amico and S. Wasserman for setting up the beamline at the Advanced Photon Source (APS) (supported by the US Department of Energy), R. Abramowitz and C. Ogata for setting up the beamline at the National Synchrotron Light Source (NSLS), and the MacCHESS staff for setting up the beamline at CHESS. We thank G. Bhargava, L. Duan and G. Xu for technical help; R. Khayat, G. Jogl, and Z. Yang for help with data collection at the synchrotron sources; H. Wu and W. Hendrickson for discussions; and Columbia University (L.T.) and an NIH grant (J.L.M.) for financial support.
About this article
Cite this article
Xu, Y., Tao, X., Shen, B. et al. Structural basis for signal transduction by the Toll/interleukin-1 receptor domains. Nature 408, 111–115 (2000). https://doi.org/10.1038/35040600
Nature Structural & Molecular Biology (2021)
Nature Structural & Molecular Biology (2021)
O-specific polysaccharide structure isolated from the LPS of the Antarctic bacterium Pseudomonas ANT_J38B
Carbohydrate Research (2020)
Nature Reviews Nephrology (2020)