Biochemical and Biophysical Research Communications 299 (2002)

Transcription

1 Biochemical and Biophysical Research Communications 299 (2002) BBRC An extensively associated dimer in the structure of the C713S mutant of the TIR domain of human TLR2 Xiao Tao, Yingwu Xu, 1 Ye Zheng, Amer A. Beg, and Liang Tong * Department of Biological Sciences, Columbia University, New York, NY 10027, USA Received 7 October 2002 Abstract The Toll/interleukin-1 receptor (TIR) domains are conserved modules in the intracellular regions of the Toll-like receptors (TLRs) and interleukin-1 receptors (IL-1Rs). The domains are crucial for the signal transduction by these receptors, through homotypic interactions among the receptor and the downstream adapter TIR domains. Previous studies showed that the BB loop in the structure of the TIR domain forms a prominent conserved feature on the surface and is important for receptor signaling. Here we report the crystal structure of the C713S mutant of the TIR domain of human TLR2. An extensively associated dimer is observed in the crystal structure and mutations of several residues in this dimer interface abolished the function of the receptor. Moreover, the structure shows that the BB loop can adopt different conformations, which are required for the formation of this dimer. This asymmetric dimer might represent the TLR2:TLRx heterodimer in the function of this receptor. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Toll-like receptors; Receptor signaling; TIR domain; Protein structure and function; Mutagenesis; Innate immunity Innate immunity represents the sole mechanism of host defense against microbial infections in invertebrates [1]. In mammals and other vertebrates, innate immunity provides the first line of host defense against these infections and the innate immune response is crucial for the stimulation of adaptive immunity [2]. Recent studies have shown that Toll-like receptors (TLRs) have crucial roles in the vertebrate innate and adaptive immune responses [3 5]. The intracellular region of TLRs contains a conserved domain of about 150amino acid residues, known as the TIR (Toll/interleukin-1 receptor) domain, as it also shares sequence homology with the intracellular region of members of the interleukin-1 receptor (IL-1R) superfamily [6,7]. Upon activation, the TLRs and IL- 1Rs interact with a cytoplasmic adapter molecule, MyD88, which also contains a TIR domain [3 5]. Therefore, it is believed that homotypic interactions * Corresponding author. Fax: address: tong@como.bio.columbia.edu (L. Tong). 1 Present address: Cancer Biology, Dana-Farber Cancer Institute, Boston, MA 02115, USA. between receptor and adapter TIR domains may mediate the signal transduction by these receptors. The crucial importance of the TIR domain for receptor function is demonstrated by the Lps d mutant of murine TLR4. This single-point mutation, P712H, in the TIR domain of the receptor abolished its response to bacterial LPS [8]. This proline residue is conserved among all mammalian TLRs except for TLR3. Mutation of the equivalent residue in TLR2, P681H, can abrogate signal transduction in response to stimulation by yeast and Gram-positive bacteria [9 12]. We recently reported the crystal structures of the TIR domains of human TLR1 and TLR2 [13]. The TIR domain contains a central fully parallel five-stranded b- sheet (ba through be), five helices (aa through ae), and the connecting loops (Fig. 1A). Structural and sequence analyses show that the BB loop contains several highly conserved residues and corresponds to a prominent feature on the surface of the domain. Mutations of several residues in this surface patch can abolish the function of the receptors [13]. Most importantly, the Lps d mutation is located at the tip of the BB loop. Crystal structure of the P681H mutant of human TLR2 is essentially identical to that of the wild-type, suggest X/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S X(02 )

2 X. Tao et al. / Biochemical and Biophysical Research Communications 299 (2002) Fig. 1. Structures of the TIR domain of human TLR2. (A) Schematic drawing of the structure of wild-type TIR domain of human TLR2. The b- strands are shown as arrowed ribbons in cyan, a-helices in yellow, and connecting loops in purple. Residues are missing in the atomic model and are indicated in grey. (B) Structure of molecule A of the C713S mutant. Molecules C and E are similar to this molecule. (C) Structure of molecule B of the C713S mutant. Molecule D is similar to this molecule. (D) Overlay of the Ca atoms of the structures of the wild-type TIR domain (in yellow) and molecules A (green) and B (cyan) of the C713S mutant. Note the large conformational changes for the BB loop in molecule B. (A C) are produced with ribbons [21] and (D) is produced with Grasp [22]. ing that the Lps d mutation may abrogate receptor function by disrupting the interactions with other molecules (possibly TIR domains) in the signaling complex [13]. It is expected that TIR domains may function as homo- or hetero-oligomers, by mediating interactions between the receptors, as well as the downstream adapter molecules [3 5]. In the crystals of the TIR domain of human TLR1, a tetramer of the molecule, obeying 222 symmetry, was observed. However, this tetramer contains two disulfide bridges, between residues Cys707 of two neighboring monomers [13]. Moreover, this disulfide bridge is buried at the center of the interface between the two monomers. Therefore, it is likely that the observed tetramer may be stabilized mostly by this disulfide linkage. In the structure of the TIR domain of TLR2, the equivalent residue, Cys713, is modified by cacodylate, which might also have affected the natural oligomerization behavior of the protein [13]. Therefore, to remove the potential for covalent modification at this Cys residue, we have produced the C713S mutant of the TIR domain of TLR2 and obtained its crystal structure at 3.2 A resolution. An extensively associated, asymmetric dimer is observed in the crystal and mutations of several residues in this dimer interface abolished receptor signaling. Moreover, the structure shows that the BB loop can adopt a few different conformations, which may be important for its function in signal transduction. Materials and methods Cloning, protein expression, and purification. The C713S mutant of the TIR domain was prepared with the QuikChange mutagenesis kit (Stratagene) and sequenced to confirm the incorporation of the mutation. The mutant protein was expressed and purified using the same protocol as that for the wild-type protein [13]. Crystallization and data collection. Crystals of the C713S mutant were obtained at 21 C by the hanging drop vapor diffusion method. The reservoir solution contained 100 mm MES (ph 6.5) or Hepes (ph 7.5), M ðnh 4 Þ 2 HPO 4, and 5 mm DTT. The protein was at mg/ml concentration, in a buffer containing 20 mm MOPS

3 218 X. Tao et al. / Biochemical and Biophysical Research Communications 299 (2002) (ph 7.0), 150 mm NaCl, and 2 mm DTT. The crystals were transferred in a few steps to the cryoprotection buffer (100 mm MES, ph 6.5, 1.6 M ðnh 4 Þ 2 HPO 4, and 25% glycerol) and flash-frozen in liquid propane. X-ray diffraction data to 3.2 A resolution were collected at Beamline 32-ID (ComCAT) of the Advanced Photon Source (APS). The diffraction images were recorded on a Mar CCD detector and processed with the HKL package [14]. The crystal belongs to space group P6 5 22, with cell parameters of a ¼ b ¼ 112:3 A and c ¼ 362:8 A. The data processing statistics are summarized in Table 1. Structure determination and refinement. Five copies of the TIR domain molecules were located by the combined molecular replacement protocol as implemented in the program COMO [15,16], using the structure of the wild-type TIR domain as the search model. The details of the structure determination have been described previously [16]. After fivefold non-crystallographic symmetry (NCS) averaging with the program DM [17], the atomic model was rebuilt against the electron density map with the program O [18]. The structure refinement was carried out with the program CNS [19] and NCS restraints were applied throughout the refinement. The atomic model was further rebuilt against the (omit) 2F o F c electron density maps. The refinement statistics are summarized in Table 1. Functional studies with TLR2 mutants. For expression, wild-type murine TLR2 was cloned into the pires vector (Clontech), as reported earlier [20]. A series of single-site mutants of murine TLR2, selected based on the structural information, were produced with the QuikChange mutagenesis kit (Stratagene) and sequenced to confirm the presence of the mutations. A pires vector containing GFP was used as the negative control for receptor activation assays, as well as an independent indicator for transfection efficiency. Human embryonic kidney (HEK) 293 cells were cultured in DulbeccoÕs modified EagleÕs medium (CellGro) supplemented with glutamine (2 mm), penicillin (100 U/ml), and fetal bovine serum (10%). The TLR2-IRES or the GFP-IRES expression construct, together with an NF-jB and a TK-promoter driven luciferase expression construct, was transfected into the HEK293 cells with Fugene6 (Roche Molecular Biochemicals). After 24 h, the cells were treated with peptidoglycan (Fluka) at 5 lg/ml concentration. Cell extracts were prepared 6 8 h later and tested for luciferase activity using the Dual- Luciferase Reporter Assay System (Promega). The transfection efficiency was assessed by the GFP fluorescence and the TK-promoter driven luciferase (Renilla luciferase) activity, an internal control that monitors the baseline response. The NF-jB-dependent luciferase (firefly luciferase) activity was normalized to the activity of the internal control to minimize experimental variability due to differences in transfection efficiency or cell viability. The expression levels of the various mutants were checked by Western blots using a TLR2 antibody (ebioscience). Table 1 Summary of crystallographic information Maximum resolution ( A) 3.2 Number of observations 99,505 R merge (%) a 8.7 (39.9) Resolution range for refinement Number of reflections 19,710 Completeness (%) 85 R factor b (%) 24.9 (30.8) Free R factor (%) 31.6 (36.8) rms deviation in bond lengths ( A) rms deviation in bond angles ( ) 1.4 R merge ¼ P P h i ji hi hi h ij= P P h i I hi. Numbers in parentheses are for the highest resolution shell. R ¼ P h jf h o h cj= h F h o. Results and discussion Overall structure and conformational variability of the BB loop The crystal structure of the C713S mutant of the TIR domain of human TLR2 has been determined at 3.2 A resolution. The statistics for the structure refinement are summarized in Table 1. The atomic structure has been deposited at the Protein Data Bank (Accession No. 1077). There are five copies of the TIR domain molecule in the crystallographic asymmetric unit and they will be called molecules A through E here. In all these molecules, the first 13 residues of the protein ( ) and a portion of the CD loop (residues ) are missing in the atomic model due to disorder (Figs. 1B and C), as also observed in the wild-type structure (Fig. 1A) [13]. Based on their conformations, the five molecules can be divided into two groups. The first group contains molecules A, C, and E (Fig. 1B). The root-meansquare (rms) distance between equivalent Ca atoms of any pair of these molecules is about 0.65 A. In the second group, the rms distance between equivalent Ca atoms of molecules B and D is 0.5 A (Fig. 1C). In contrast, there are large conformational differences between the two groups of molecules and wild-type (Fig. 1D). These differences are due to inherent flexibility of the TIR domain rather than to the C713S mutation, as the mutation caused only minor conformational changes in its vicinity. For molecules A, C, and E, in the first group, differences are seen for residues (BB loop and the beginning of the ab helix) and (ac 0 helix and the CD loop). In addition, there are smaller conformational differences for residues , in the DD loop. The largest conformational differences to the wildtype are seen for molecules B and D, in the second group, for residues (BB loop and the beginning of the ab helix), (ac 0 helix), and (DD loop and ad helix) (Fig. 1D). The conformational difference in the BB loop for molecules in the second group is especially noteworthy (Figs. 1C and D). In the wild-type protein and molecules of the first group, the BB loop protrudes away from the rest of the TIR domain and forms a prominent feature on the surface [13]. In molecules B and D, the BB loop adopts a conformation where it is located closer to the rest of the TIR domain, but remains a highly visible feature on the surface of the molecule. The BB loop in this new conformation is important for mediating interactions in a dimer of molecules A and B (see below). Similarly, residues in the CD loop are ordered in molecule A, but disordered in molecule B, because it is located in the AB dimer interface.

4 X. Tao et al. / Biochemical and Biophysical Research Communications 299 (2002) An extensively associated, asymmetric dimer of the TIR domain Molecules A, B, C, and D form a tetramer obeying non-crystallographic twofold symmetry, which relates the AB and the CD asymmetric dimers (Fig. 2A). This tetramer is located near a crystallographic twofold axis, such that an octamer of the TIR domain, obeying 222 symmetry, is generated (not shown). The A, B, C, and D monomers are arranged in a side-by-side fashion in the tetramer, such that their N-termini are all on the same face of the oligomer (Fig. 2A). The fifth molecule (E) in the asymmetric unit does not belong to this complex and has only small contacts with the other four monomers. The most significant interactions in this octamer occur in the AB and equivalently the CD asymmetric dimers (Fig. 2B), where about 600 A 2 of the surface area of each monomer is buried. Therefore, this dimer interface contributes about 4800 A 2 or roughly half of the total surface area burials (8700 A 2 ) in the octamer. The two molecules are related by a rotation of about 103. Their interface is formed by the contact of the ab, ac 0, ad helices, and the CD, DD loops of molecule A with the ab helix and the BB loop of molecule B (Fig. 2B). The conformational difference of the BB loop in molecule B is important for the formation of this interface, with most of the loop buried at the interface. Interestingly, the side chain of Pro681 (the BB7 residue) is exposed to the solvent (Fig. 2B). To facilitate comparison among TIR domains, residues in the BB loop are numbered according to their positions in the loop [13], as shown in Fig. 3. On the other hand, the BB loop of molecule A is not involved in the formation of the AB dimer (Fig. 2A). Near the center of this interface, Leu717 in molecule A is in van der Waals contact with residues Tyr641, Cys673, and Phe679 (BB5) in molecule B (Fig. 2B), and Fig. 2. An oligomer of the TIR domain. (A) The tetramer of the TIR domain in the crystals of the C713S mutant. The N-termini of the four monomers are on the same face. The disulfide links are indicated. The twofold symmetry axis of the tetramer is indicated with a blue oval. (B) Detailed interactions in the interface of the AB dimer. Molecule A is shown in green and molecule B is shown in cyan. The BB7 residue in molecule B, Pro681, is shown in red, but it does not contribute to the interface. Produced with ribbons [21].

5 220 X. Tao et al. / Biochemical and Biophysical Research Communications 299 (2002) Fig. 3. Alignment of representative human TIR domain sequences. The secondary structure elements (S.S.) are labeled. Residues in the BB loop are numbered according to their positions in the loop. 90 A 2 of its surface area is buried by the formation of this dimer. The Arg748 side chain is in an ion-pair interaction with the highly conserved acidic residue at the BB4 position of the BB loop (Figs. 3 and 2B) and 60 A 2 of its surface area is buried at this interface. To our surprise, despite efforts to minimize Cys modification, we observed the formation of disulfide links between neighboring molecules of the tetramer in the crystal (Fig. 2A). In the AB dimer, the disulfide bond is formed between Cys640of molecule A and Cys750of molecule B (Fig. 2B). This disulfide is located at the very edge of the interface and contributes only about 20% of the buried surface area. It is unlikely that the AB dimer interface is stabilized solely by the S S link. This is supported by the observation that identical residues (Cys640, Cys750) are involved in the S-S bond between molecules A and D, but the AD interface is much less extensive (Fig. 2A). Most importantly, the relevance of the AB interface is supported by our functional studies (see below). In contrast to the AB and equivalently the CD dimer interface, interactions between these two dimers appear to be much weaker. The two S S bonds between the two dimers make large contributions to this interface. Therefore, it is probably unlikely that this tetramer, and consequently the entire octamer, can be stable in the absence of the disulfide bonds. Consequently, the tetramer and octamer interfaces will not be analyzed further here. Residues at the interface of the AB dimer are important for receptor signaling To assess the functional relevance of this dimer, we have mutated several of the residues that make major contributions to this interface and determined the effects of the mutations on signaling by murine TLR2 [20]. The mutants have comparable levels of expression, as confirmed by Western blots. The mutants include F679D (BB5 residue), L717E (in ac 0 helix), and R748E and R748S (in DD loop). The F679D and L717E mutations changed TLR2 residues into their equivalents in IL-1R type I, whereas the R748S mutation changed it to the TLR1 sequence (Fig. 3). We also included the P681H mutant as a positive control, and the C713S mutant to assess the importance of the Cys713 residue, which makes a small contribution to the AB interface (Fig. 2B). As negative controls, we produced the R709K (in helix ac), V735I (bd), R742K (DD loop), and G769V (EE loop) mutants, changing murine TLR2 to human TLR2 sequence. These residues are located outside the AB interface and, with the exception of residue 735, are poorly conserved among the TIR domains (Fig. 3). The functional studies showed that the P681H Lps d mutant of TLR2 has essentially no signaling activity in response to peptidoglycan (PGN) (Fig. 4), consistent with earlier results [3 5]. Interestingly, the P681V mutant has little signaling activity either, even though a Val residue at the BB7 position is found in IL-1R type I (Fig. 3) and Drosophila Toll. The V! P mutation in Toll was tolerated by that receptor [13]. Most importantly, mutations at all three sites (679, 717, and 748) in the interface of the AB dimer also abolished receptor signaling, confirming the functional relevance of this interface (Fig. 4). On the other hand, mutations outside this dimer interface have little effects on the activation of the receptor (Fig. 4). The Cys713 residue may be located in a different interface in the TIR domain signaling complex, as our studies show that the C713S mutation also has deleterious effects on receptor function (Fig. 4). Nonetheless, this Fig. 4. Functional studies with mutants in the TIR domain of murine TLR2. The relative receptor activation in response to peptidoglycan (PGN) stimulation for the wild-type and various mutants of the receptor is plotted. The error bar on the wild-type data is based on multiple experiments, carried out over different days.

6 X. Tao et al. / Biochemical and Biophysical Research Communications 299 (2002) observation does not conflict with the functional importance of the surface patch identified in our study here. First of all, our structural analysis showed that the C713S mutation did not disrupt the structure of the TIR domain (see above). Second, the C713S residue makes only a small contribution to the AB dimer interface (Fig. 2B), therefore it is expected that the AB dimer can also be maintained in the wild-type TLR2 structure. At the same time, the side chain of the Cys713 residue is available for interactions with other TIR domain molecules in the complex (Fig. 2B). Further studies are needed to understand the molecular mechanism for the functional requirement of a Cys residue at this position in TLR2. This residue is conserved among the TLRs and MyD88, but not among Drosophila Toll homologs and the IL-1Rs (Fig. 3). TLR2 is believed to function as a heterodimer, possibly with TLR1 or TLR6, in the recognition of foreign pathogens [4,5]. Our observation of an asymmetric dimer of the TIR domain appears to be consistent with the biological observations. Residues in this dimer interface in molecule A (Leu717 and Arg748 for example) are unique to TLR2 (Fig. 3), whereas their contacts in molecule B (the BB loop) are more conserved among other TIR domains (Fig. 2B). It is possible that the asymmetric AB dimer observed here may reflect the natural heterodimeric TLR2:TLRx signaling complex, with molecule A corresponding to TLR2 and molecule B corresponding to TLR1 or TLR6. In this complex, the entire BB loop of molecule A and the site of the Lps d mutation (BB7) in molecule B are available for recruitment of downstream adapter molecules (Fig. 2A). Such a model is also consistent with earlier biochemical results, which showed that the Lps d mutation abolished the recruitment of the MyD88 molecule [13]. On the other hand, TLR4 is believed to function as a homodimer [4,5]. The signaling complex for the TLR4 TIR domain may be different from that observed here for TLR2 and a symmetrical dimer of TIR domains may be involved in the signal transduction by TLR4. Acknowledgments We thank Kevin DÕAmico and Steve Wassermann for access to the 32-ID beamline, Craig Ogata for access to the X4A beamline, Ming Li for helping to set up the TLR2 receptor assays, and John Sims for helpful discussions. This research was supported in part by a grant from the National Institutes of Health (AI49475 to L.T.). References [1] J.A. Hoffmann, F.C. Kafatos, C.A. Janeway Jr., R.A.B. Ezekowitz, Phylogenetic perspectives in innate immunity, Science 284 (1999) [2] R. Medzhitov, C.A. Janeway Jr., Innate immunity: the virtues of a nonclonal system of recognition, Cell 91 (1997) [3] A. Aderem, R.J. Ulevitch, Toll-like receptors in the induction of the innate immune response, Nature 406 (2000) [4] S. Akira, K. Takeda, T. Kaisho, Toll-like receptors: critical proteins linking innate and acquired immunity, Nat. Immunol. 2 (2001) [5] D.M. Underhill, A. Ozinsky, Toll-like receptors: key mediators of microbe detection, Curr. Opin. Immunol. 14 (2002) [6] N.J. Gay, F.J. Keith, Drosophila Toll and IL-1 receptor, Nature 351 (1991) [7] D.S. Schneider, K.L. Hudson, T.-Y. Lin, K.V. Anderson, Dominant and recessive mutations define functional domains of Toll, a transmembrane protein required for dorsal ventral polarity in the Drosophila embryo, Genes Dev. 5 (1991) [8] A. Poltorak, X. He, I. Smirnova, M.-Y. Liu, C.V. Huffel, X. Du, D. Birdwell, E. Alejos, M. Silva, C. Galanos, M. Freudenberg, P. Ricciardi-Castagnoli, B. Layton, B. Beutler, Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene, Science 282 (1998) [9] A.O. Aliprantis, R.-B. Yang, M.R. Mark, S. Suggett, B. Devaux, J.D. Radolf, G.R. Klimpel, P. Godowski, A. Zychlinsky, Cell activation and apoptosis by bacterial lipoproteins through Tolllike receptor-2, Science 285 (1999) [10] E. Lien, T.J. Sellati, A. Yoshimura, T.H. Flo, G. Rawadi, R.W. Finberg, J.D. Carroll, T. Espevik, R.R. Ingalls, J.D. Radolf, D.T. Golenbock, Toll-like receptor 2 functions as a pattern recognition receptor for diverse bacterial products, J. Biol. Chem. 274 (1999) [11] O. Takeuchi, K. Hoshino, T. Kawai, H. Sanjo, H. Takada, T. Ogawa, K. Takeda, S. Akira, Differential roles of TLR2 and TLR4 in recognition of gram-negative and gram-positive bacterial cell wall components, Immunity 11 (1999) [12] D.M. Underhill, A. Ozinsky, A.M. Hajjar, A. Stevens, C.B. Wilson, M. Bassetti, A. Aderem, The Toll-like receptor 2 is recruited to macrophage phagosomes and discriminates between pathogens, Nature 401 (1999) [13] Y. Xu, X. Tao, B. Shen, T. Horng, R. Medzhitov, J.L. Manley, L. Tong, Structural basis for signal transduction by the Toll/ interleukin-1 receptor domains, Nature 408 (2000) [14] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in oscillation mode, Methods Enzymol. 276 (1997) [15] L. Tong, Combined molecular replacement, Acta Cryst. A 52 (1996) [16] G. Jogl, X. Tao, Y. Xu, L. Tong, COMO: a program for combined molecular replacement, Acta Cryst. D 57 (2001) [17] CCP4, The CCP4 suite: programs for protein crystallography, Acta Cryst. D 50(1994) [18] T.A. Jones, J.Y. Zou, S.W. Cowan, M. Kjeldgaard, Improved methods for building protein models in electron density maps and the location of errors in these models, Acta Cryst. A 47 (1991) [19] A.T. Brunger, P.D. Adams, G.M. Clore, W.L. DeLano, P. Gros, R.W. Grosse-Kunstleve, J.-S. Jiang, J. Kuszewski, M. Nilges, N.S. Pannu, R.J. Read, L.M. Rice, T. Simonson, G.L. Warren, Crystallography & NMR system: a new software suite for macromolecular structure determination, Acta Cryst. D 54 (1998) [20] M. Li, D.F. Carpio, Y. Zheng, P. Bruzzo, V. Singh, F. Ouaaz, R.M. Medzhitov, A.A. Beg, An essential role of the NF-jB/Tolllike receptor pathway in induction of inflammatory and tissuerepair gene expression by necrotic cells, J. Immunol. 166 (2001) [21] M. Carson, Ribbon models of macromolecules, J. Mol. Graph. 5 (1987) [22] A. Nicholls, K.A. Sharp, B. Honig, Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons, Proteins 11 (1991)