The structural basis for receptor recognition of human interleukin-18

Abstract

Interleukin (IL)-18 is a proinflammatory cytokine that belongs to the IL-1 family and plays an important role in inflammation. The uncontrolled release of this cytokine is associated with severe chronic inflammatory disease. IL-18 forms a signalling complex with the IL-18 receptor α (Rα) and β (Rβ) chains at the plasma membrane, which induces multiple inflammatory cytokines. Here, we present a crystal structure of human IL-18 bound to the two receptor extracellular domains. Generally, the receptors' recognition mode for IL-18 is similar to IL-1β; however, certain notable differences were observed. The architecture of the IL-18 receptor second domain (D2) is unique among the other IL-1R family members, which presumably distinguishes them from the IL-1 receptors that exhibit a more promiscuous ligand recognition mode. The structures and associated biochemical and cellular data should aid in developing novel drugs to neutralize IL-18 activity.

ARTICLE

Received 10 Jul 2014 | Accepted 20 Sep 2014 | Published 15 Dec 2014

Naotaka Tsutsumi1,*, Takeshi Kimura2,*, Kyohei Arita3, Mariko Ariyoshi1,4, Hidenori Ohnishi2, Takahiro Yamamoto2, Xiaobing Zuo5, Katsumi Maenaka6, Enoch Y. Park7, Naomi Kondo2,8, Masahiro Shirakawa1,9, Hidehito Tochio10 & Zenichiro Kato2,11

Interleukin (IL)-18 is a proinammatory cytokine that belongs to the IL-1 family and plays an important role in inammation. The uncontrolled release of this cytokine is associated with severe chronic inammatory disease. IL-18 forms a signalling complex with the IL-18 receptor

a (Ra) and b (Rb) chains at the plasma membrane, which induces multiple inammatory cytokines. Here, we present a crystal structure of human IL-18 bound to the two receptor extracellular domains. Generally, the receptors recognition mode for IL-18 is similar to IL-1b;

however, certain notable differences were observed. The architecture of the IL-18 receptor second domain (D2) is unique among the other IL-1R family members, which presumably distinguishes them from the IL-1 receptors that exhibit a more promiscuous ligand recognition mode. The structures and associated biochemical and cellular data should aid in developing novel drugs to neutralize IL-18 activity.

1 Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, Katsura, Nishikyo-ku, Kyoto 615-8510, Japan. 2 Department of Pediatrics, Graduate School of Medicine, Gifu University, Yanagido 1-1, Gifu 501-1194, Japan. 3 Graduate School of Nanobioscience, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama Kanagawa 230-0045, Japan. 4 Institute for Integrated Cell-Material Sciences, Kyoto University, Kyoto 606-8501, Japan. 5 X-Ray Science Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, USA. 6 Laboratory of Biomolecular Science and Center for Research and Education on Drug Discovery, Faculty of Pharmaceutical Sciences, Hokkaido University, , Kita-12, Nishi-6, Kita-ki, Sapporo 060-0812, Japan. 7 Research Institute of Green Science and Technology, Department of Bioscience, Graduate school of Science and Technology, Shizuoka University, 836 Ohya Suruga-ku, Shizuoka 422-8529, Japan. 8 Heisei College of Health Sciences, 180 Kurono, Gifu 501-1131, Japan. 9 Core Research of Evolution Science (CREST), Japan Sciences and Technology Agency, Tokyo 102-0076, Japan. 10 Department of Biophysics, Graduate School of Science, Kyoto University, Kitashirakawa-oiwake, Sakyo-ku, Kyoto 606-8502, Japan. 11 Biomedical Informatics, Medical Information Sciences Division, The United Graduate School of Drug Discovery and Medical Information Sciences, Gifu University, Gifu 501-1194, Japan. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to H.O. (email: mailto:[email protected]

Web End [email protected] ) or to H.T. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms6340 OPEN

The structural basis for receptor recognition of human interleukin-18

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

Interleukin (IL)-18 belongs to the IL-1 superfamily and was rst discovered as an interferon gamma (IFN-g)-inducing factor in sera from mice with hepatitis stimulated with

Propionibacterium acnes and lipopolysaccharide1. This proinammatory cytokine is secreted by various types of cells and strongly augments IFN-g production in type-1 helper T (Th1)

cells and natural killer (NK) cells following activation of NK-cell cytotoxicity; thus, it plays a critical role in inammation and the host defense against microbes. In addition to IL-1b2,3,

IL-18 is synthesized as a biologically inactive precursor (proIL-18) on activation of a certain class of receptors, such as Toll-like receptors and proinammatory cytokine receptors, and then stored in the cytosol. Once it matures via caspase-1 (ref. 4), which is regulated by a large protein complex referred to as the inammasome5, IL-18 is extracellularly secreted and binds IL-18 receptor a (Ra) as well as IL-18 receptor b (Rb) at the immunocyte plasma membrane in a stepwise manner. IL-18/ IL-18Ra/IL-18Rb ternary complex formation juxtaposes the intracellular Toll-Interleukin-1 receptor domains of IL-18Ra and IL-18Rb, to which the adaptor molecule myeloid differentiation factor 88 (MyD88) is recruited presumably with the aid of TRAM6. MyD88 further interacts with IL-1 receptor associating kinase (IRAK) 4 and IRAK1/2 to form the large molecular assembly referred to as Myddosome, which subsequently activates IKK via TRAF6. Finally, the signal activates the NF-kB and mitogen-activated protein kinase pathways7, which upregulate the expression of various inammatory cytokines.

Of the IL-1 family cytokines, IL-18 and IL-1b have garnered much attention because they are causal cytokines that lead to severe chronic inammatory syndrome. IL-1b is associated with immunological disorders, such as autoinammatory syndromes8,9. The central pathogenic feature of autoinammatory syndromes is excess production of mature IL-1b derived from abnormal inammasome activation due to certain gene mutations. IL-1b-related autoinammatory diseases are treated through neutralizing IL-1b by anti-IL-1b (canakinumab and gevokizumab), engineered soluble receptors (rilonacept) or the receptor antagonist IL-1Ra (anakinra), which is remarkably effective; thus, these treatments are currently in clinical use10. Similar to IL-1b, IL-18 overproduction likely leads to severe autoimmune, autoinammatory, allergic, neurological and metabolic disease, which might be associated with IL-18 or IL-18 receptor genetic polymorphisms1114. Two recent papers have revealed that constitutive activation of the inammasome caused by single point mutations in NLRC4 is associated with a novel autoinammatory disorder, and the patient with NLRC4-mediated macrophage activation syndrome showed ultra-high circulation levels of IL-18 even after IL-1 blockade15,16. Consistent with these observations, therapeutic approaches that block IL-18 activity have been effective in inammatory disease models17,18. Therefore, developing drugs that impede binding between IL-18 and the receptors is clinically important. Generally, the atomic structures of targeted proteins and their complexes play vital roles in drug design. Thus far, despite the reported structures for free IL-18 and its related complexes1922, a structure for the genuine complex between IL-18 and its receptors has not yet been determined.

Previously, we reported a solution structure for IL-18 and identied the functional residues for which mutation markedly decreased its binding afnity for IL-18Ra19. The results suggest that the binary complex between IL-18 and IL-18Ra exhibits an essentially identical binding mode to the complex between IL-1b and its receptors (IL-1RI or IL-1RII). However, the binding mode for IL-18Rb, which is the IL-18 co-receptor, to IL-18/IL-18Ra remained ambiguous. Recent structural studies on the ternary complex between IL-1b and its receptors ectodomains23,24

demonstrate that IL-1RAcP, which is the commonly used co-receptor for IL-1a, IL-1b, IL-33 and IL-36s, adopted a left

binding mode. In this mode, IL-1RAcP binds the IL-1b/IL-1RI or IL-1b/IL-1RII binary complexes from the left side as seen from the concave IL-1b recognition surface of IL-1RI or IL-1RII. Furthermore, the other IL-1 superfamily molecule, IL-33/ST2/IL-1RAcP, was also suggested to adopt the left binding mode based on the model structure from the small angle X-ray scattering (SAXS) proles25. Thus, left binding seems common in complexes that employ IL-1RAcP. In contrast to other IL-1 family cytokines, IL-18 is unique due to its pair of specialized receptors (IL-18Ra and IL-18Rb); hence, the recognition details are not sufciently understood based only on homology to the IL-1b and IL-33 system.

Here, we performed X-ray crystallography using human IL-18 and its complexes with the receptors extracellular domains. The structures demonstrate that the co-receptor (IL-18Rb) binding mode is generally identical to IL-1b; however, substantial differences were observed in the subdomain orientations and interaction details throughout the complex. Intriguingly, the second domain (D2) of the two IL-18 receptors lacked one b-strand, d2, which is conserved among other IL-1-related receptors, and was previously shown to contribute to the inter-receptor interaction. In addition, N-linked glycans played a role in bridging the two receptors, which was observed in the signalling IL-1b receptor complex but was absent in its decoy complex. We further show that other IL-18Ra N-linked glycans proximal to IL-18 in the complexes contributed to the binding afnity. With the associated biochemical and cell biological data, the structures comprehensively clarify the IL-18 receptor recognition mode, which will facilitate rational drug development to neutralize IL-18 activity, the uncontrolled release of which has been shown to cause severe chronic inammatory diseases.

Results

Structural comparison between the IL-18 and IL-1b complexes.

We determined the crystal structures of IL-18 (Fig. 1a), the IL-18/ IL-18Ra binary complex (Fig. 1b) and the IL-18/IL-18Ra/IL-18Rb signalling ternary complex (Fig. 1c,d) at the resolutions2.33, 3.10 and 3.10 , respectively. The crystallographic statistics are provided in Table 1.

IL-18Ra curls around IL-18, and IL-18Rb contacts the lateral portion of the IL-18/IL-18Ra binary complex in a similar manner as the IL-1b/IL-1RI(RII)/IL-1RAcP complex23,24. The IL-18 structure essentially does not change on complex formation, maintaining the b-trefoil fold that comprises 12 b-strands (b1-b12) and 2 a-helices (a1-a2) (Supplementary Fig. 1a), as previously reported19. The IL-18Ra ectodomain folds into three immunoglobulin (Ig)-like domains, which are referred to as D1, D2 and D3, in the same manner as the IL-1 receptors23,24,26. Each domain comprises a two-layer sandwich of six to nine b-strands and contains at least one intra-domain disulde bond (Supplementary Fig. 1b). Within IL-18Ra, D1 extensively contacts D2, whereas D3 is distant and is connected by the long D2-D3 linker (Fig. 1d middle), which implies that D1 and D2 behave as a single module, similar to IL-1-related primary receptors. In the Ig superfamily, including the IL-1 receptor family (Fig. 2a), the core cysteine residues on the b and f strands are highly conserved (Fig. 2b). However, for IL-18Ra-D1, the f strand cysteine is replaced with phenylalanine (Fig. 2b), which yields two unexpected surface disulde bonds (Fig. 2c). In addition, the D2 domain lacks one b-strand (d2 in Fig. 2b,d) that is structurally conserved among most IL-1 receptor family members, IL-1RI, IL-1RII and ST2 as well as IL-1RAcP.

On ternary complex formation, the IL-18/IL-18Ra binary complex structure essentially does not change; the root mean

2 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

+IL-18R

+IL-18R

90

IL-18

D1

D2

D2

D1

90

IL-18 90

D3

D3

IL-18R

IL-18R

IL-18

IL-18R

IL-18R

Figure 1 | Overall structures for IL-18 and its extracellular complexes. (ac) Schematic ow diagram of the stepwise complex formation for IL-18/ IL-18Ra/IL-18Rb. Crystal structures of (a) IL-18, (b) IL-18/IL-18Ra and (c) IL-18/IL-18Ra/IL-18Rb are shown as a ribbon (IL-18, blue) or surface (IL-18Ra, palegreen; IL-18Rb, wheat) representation, respectively. (d) Ribbon diagrams of the ternary complex structure for IL-18 (blue), IL-18Ra (green) and

IL-18Rb (orange) from three perspectives.

square deviations (RMSD) for the backbone Ca atoms are0.410.90 compared with the asymmetric unit (ASU) molecules. IL-18Rb is also composed of three Ig-like domains, similar to IL-18Ra except for the aforementioned disulde bonds (Supplementary Fig. 1c); however, the spatial arrangement of these domains is markedly distinct from IL-18Ra in the ternary complex (Fig. 1d). D2 and D3 of IL-18Rb are close to each other and directly associated with IL-18/IL-18Ra, while D1 does not contribute to ternary complex formation. In fact, a part of the IL-18Rb-D1 electron density is ambiguous presumably because the region was loosely packed in the crystal (Supplementary Fig. 2). The D1 isolation of the co-receptor was also observed previously in the IL-1b/IL-1RI(RII)/IL-1RAcP complex, where

IL-1RAcP-D1 is outstretched and does not participate in the molecular interface. Despite of the similarity in the binding mode, relative orientation of IL-18Rb-D2 D3 to IL-18/IL-18Ra is

different from that of IL-1RAcP-D2 D3 to IL-1b/IL-1RI(RII)

(Fig. 3ac). This difference in the orientations is attributable to the interaction manner at the interface between IL-18Rb-D2 and

IL-18/IL-18Ra (Fig. 3b,d). Owing to the aforementioned absence of the d2 strands in D2s (Fig. 2b,d), IL-18Ra-D2 and IL-18Rb-D2 supply only two strands (b2 and e2) and loops at the interface, respectively, while for IL-1RI(RII) and IL-1RAcP-D2, three strands (b2, e2 and d2) and loops interact (Fig. 1d and Fig. 3ac). Furthermore, IL-18Rb-D2 supplies not only electro-static side chains but also aromatic rings that interact with IL-18, whereas IL-1RAcP-D2 utilizes electrostatic and aliphatic side chains for ligand binding. The b4-b5 loop of IL-18 and IL-18Rb-

D3 do not interact, which may also contribute to the orientation difference (Fig. 3d). A more marked dissimilarity is observed for the IL-18Rb-D1 orientation relative to -D2, which differs from

IL-1RAcP in the complex (Fig. 3e). As a result, despite the same left binding mode, the spatial arrangements of the subdomains in the IL-18/IL-18Ra/IL-18Rb ternary complex differ somewhat from IL-1b/IL-1RI/IL-1RAcP with Ca atom RMSD values of6.646.73 .

The binding interface between IL-18 and IL-18Ra. IL-18Ra recognizes IL-18 through a large interaction surface area at

B1,850 2. Two IL-18 sites, Sites I and II, contact IL-18Ra; Site I is located on a side of the core barrel of the b-trefoil structure, and Site II is at the top of the b-barrel (Fig. 4a).

Our previous NMR study showed that the residues 34 to 42 are exible despite partial a-helix formation in this region (residues 3841)19. In addition, two isomeric forms that originate from a cis or trans peptide bond between Ala42IL-18-Pro43IL-18 in the loop are equally populated in the solution structure. The exible nature of the segment produces a variety of architectures, as demonstrated in several crystallographic reports, including a loop with a trans isomer22, an a-helix with a cis isomer21 or, primarily, unobservable exible loop with a trans isomer20,22 (Fig. 4ac). In both the binary and ternary complexes, the segment between residues 35 and 40 adopts an a-helix structure with a cis

Ala42IL-18-Pro43IL-18 bond, which is stabilized by hydrogen bonds and electrostatic interactions with IL-18Ra (Fig. 4c).

Surrounding Site I, a-helix I mediated the interaction (Fig. 4c and Fig. 5a), wherein the Arg25Ra side chain is buried within the b3-a1 acidic groove of IL-18. Two disulde bonds, Cys22Ra-

Cys41Ra and Cys43Ra-Cys81Ra, bridge the b1-c1 loop to the N-terminal loop and d1-e1 turn, respectively (Figs 2c and 5a), which likely reinforce the proximal loop structure, to which the long b10-b11 hairpin of IL-18 is tted. This unique feature

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

Table 1 | X-ray crystallographic statistics of IL-18 and its extracellular complexes.

IL-18 IL-18/IL-18Ra IL-18/IL-18Ra/IL-18Rb

Data collection

Beamline BL38B1 (SPring-8) BL44XU (SPring-8) BL17A (Photon Factory) Wavelength 1.0000 0.9000 0.9800Space group P21 P21212 P212121

Cell dimensionsa, b, c () 68.15, 79.51, 73.46 135.49, 174.81, 183.40 72.56, 111.56, 134.57 a, b, g () 90.00, 100.97, 90.00 90.00, 90.00, 90.00 90.00, 90.00, 90.00

Resolution () 45.02.33 (2.462.33) 43.93.10 (3.273.10) 50.03.10 (3.213.10) Rmerge 4.9 (39.6) 10.0 (58.2) 11.5 (68.3)

I/sI 20.5 (3.6) 14.9 (2.5) 19.8 (2.9) Completeness (%) 97.8 (97.1) 99.6 (100) 99.9 (99.3)

Redundancy 3.8 (3.8) 3.7 (3.8) 9.8 (9.2)

Renement

Resolution () 34.822.33 (2.412.33) 42.003.10 (3.183.10) 42.003.10 (3.293.10) No. reections 32,157 (2,861) 79,130 (5,821) 17,368 (1,825) Rwork/Rfree 21.8/27.1 (27.6/29.8) 19.1/22.2 (22.7/26.3) 18.8/23.2 (20.9/29.7)

No. atoms

Protein 5,022 20,533 5,663 Glycan 1,225 335 Others 711 289 32 B-factors

Protein 46.13 78.31 74.38 Glycan 121.43 139.74 Others 60.94 109.02 36.55 R.m.s. deviations

Bond lengths () 0.010 0.009 0.009 Bond angles () 1.38 1.11 1.17

Ramachandran plot 95.2, 4.8 94.4, 5.0, 0.5 93.7, 5.4, 0.8

Values for Ramachandran plots are presented as favoured, allowed, outlier examined by RAMPAGE. Data collection statistics were summarized from the report35.

implies that the loop structure may be loosened under certain reductive conditions, which affects the afnity for the ligand. The Ser42Ra amide proton supplies a hydrogen bond for the Asp132IL-18 side chain, which also exhibited a backbone backbone hydrogen bond with Cys22Ra. Around Site II (Fig. 5b), the IL-18Ra acidic surface, which is composed of the

Glu253Ra and Glu263Ra carboxylates as well as Trp249Ra backbone oxygen, captures the Lys53IL-18 e-amino group.

IL-18Rb recognition by IL-18/IL-18Ra. The IL-18/IL-18Ra/IL-18Rb signalling ternary complex is formed by IL-18Rb binding with the lateral portion of the binary complex (Fig. 1). The IL-18Rb-D2 convex surface with the key Tyr212Rb aromatic residue ts into the concave surface jointly formed by IL-18 and IL-18Ra-D2 (Fig. 5c and Supplementary Fig. 3a) with a 613 2 buried surface area, which is shallower than that in the IL-1b complexes. The concave surface area is divided into Site III on IL-18 (354 2, Fig. 5c) and part of IL-18Ra-D2 (259 2,

Supplementary Fig. 3a). Note that IL-18 Site III is revised in this work19. Site III of IL-18 comprises the prominent b8-b9 hairpin and b11-a2 loop (Fig. 5c), where the aromatic ring of His109IL-18 forms p-p stacking with Tyr212Rb at a 3.4 distance, which is surrounded and stabilized by multiple hydrogen bonds. IL-18Ra-D2 is composed of an antiparallel b-sheet formed by b2 and e2 (Supplementary Fig. 3a) but lacks the conserved d2 strand compared with the corresponding b-sheets of IL-1RI and ST2.

Although the structure of IL-36R has not been determined yet, the multiple sequence alignment suggests that the b2/e2/d2 sheet is conserved among the primary receptors (Fig. 1b and Supplementary Fig. 1b). In addition to these interactions, IL-18Rb-D3 extensively contacts IL-18Ra-D3 with the buried area B550 2, to which both electrostatic (Supplementary

Fig. 3b) and hydrophobic interactions (Supplementary Fig. 3c) contribute. These inter-receptor interfaces in the IL-1b complex are designated as Site IV23, in which the IL-1b b4-b5 loop interacts with IL-1R1-D3 and IL-1RAcP-D3 to at least partly establish the ligands agonism/antagonism. The same would be true for the IL-36 system based on the crystal structure of IL-36 (ref. 27). In contrast, the corresponding IL-18 loop does not contact IL-18Rb-D3 and only interacts with IL-18Ra-D3 (Fig. 3d).

N-linked IL-18Ra glycans and its interactions. In the IL-18/IL-18Ra and IL-18/IL-18Ra/IL-18Rb crystal structures, seven N-linked glycosyl chains were identied in IL-18Ra at Asn91, 102, 150, 197, 203, 236 and 297 (Fig. 6a). These carbohydrates are high-mannose glycans (Fig. 6b) because the receptor proteins were prepared using the Sf9 expression system. The glycan on Asn197Ra, which are located in the D2 domain, forms moderate intramolecular interactions with Arg114Ra and His117Ra at the D1-D2 loop (Fig. 6c), seemingly contributing to the subdomains spatial arrangement. The core-NAG (N-acetyl-D-glucosamine) directly linked to Asn297Ra branches into the second-NAG and

L-fucose (FUC), which is referred to as the core-FUC (Fig. 6bd). Interestingly, the core-FUC and second-NAG on Asn297Ra are proximal to the b4-b5 loop of IL-18 (o4 ); hence, they likely interact through electrostatic and hydrophobic interactions with the ligand (Fig. 6d) and partly contribute the unique D3:D3 interaction.

Notably, the NAG-NAG extended from the IL-18Ra-D3 Asn236Ra and points to the receptor C-terminus in the binary complex; however, in the ternary complex, the NAG-NAG chain changes direction and points to IL-18Rb-D3 within the distance possibly to form electrostatic interaction with Val257Rb and Asp259Rb (Fig. 6e). A similar inter-receptor interaction via an

4 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

IL-37

IL-18

IL-36s

IL-33 IL-1 IL-1

IL-18R ST2 IL-1RI IL-1RII

Plasma membrane

IL-18R IL-1RAcP

IL-36R

NF-B activation via the MyD88-dependent pathway

f1

IL-18R ST2 IL-36R IL-1RII IL-1RI

91 80 88 101

89

104

95 103 116 104

f1

e2

IL-18R ST2 IL-36R IL-1RII IL-1RI

168 159 172 183 174

176 162 185 198 187

d2 e2

e2

201

IL-18R 187

IL-1RAcP

209 200

d2 e2

IL-18R-D2 IL-1RI-D2

IL-18R-D1 IL-1RI-D1

IL-18R-D2 IL-1RAcP-D2

Figure 2 | Structural features of IL-18 and its receptors. (a) Promiscuous interactions between the IL-1 family agonists and IL-37 with the receptor molecules. Except IL-18, all agonists employ IL-1RAcP as a co-receptor, while IL-18 uses IL-18Rb. (b) Certain regions of the receptor multiple sequence alignments manifest the unique qualities of IL-1R family IL-18Rs. The full set of sequence alignments is in Supplementary Fig. 1. (c) Structural comparison of D1 and (d) D2 of the receptors from the signalling complex. The green circles in (c) show the disulde bond positions.

N-linked oligosaccharide was previously observed between IL-1RI-D3 and IL-1RAcP-D3 in the IL-1b signalling complex, but it is intriguingly absent in the IL-1b/IL-1RII/IL-1RAcP non-signalling decoy complex, which suggests that the N-linked oligosaccharides bridging the two receptors are important for IL-1 family signal transduction.

Characterization of IL-18/IL-18Ra/IL-18Rb in solution. A previous SPR analysis showed that IL-18Rb does not bind free IL-18, but it does bind a preformed IL-18/IL-18Ra binary complex19. To better understand the formation of this complex under more physiological conditions, we performed titration experiments using solution NMR spectroscopy. The 1H15N correlation spectrum for [2H,15N]-IL-18 did not change on adding a small excess of IL-18Rb, indicating that the molecules do not interact. However, marked spectral changes were observed when IL-18 was titrated with IL-18Ra (Supplementary Fig. 4a,b).

The amino-acid residues that were perturbed during the titration

are at the molecular interface of the binary complex, which indicates that the binding mode in solution is identical to the crystal structure (Supplementary Fig. 4b, bottom). Cross-saturation NMR experiments28 further conrmed this result with more precision (Fig. 7a, forest and Supplementary Fig. 4c). Next, IL-18Rb was added to the preformed [2H,15N]-IL-18/IL-18Ra binary complex. The 1H15N correlation spectrum was then changed due to ternary complex formation. Although signicant changes were not observed in the region including the previously dened Site III (Supplementary Fig. 4d), the cross-peaks that disappeared or shifted on ternary complex formation (Supplementary Fig. 4e) are mostly at the interface with IL-18Rb, which is consistent with the ternary complex crystal structure (Fig. 7a, orange). Furthermore, we performed SAXS to analyse the architecture of the ternary complex (Supplementary Fig. 4fh); these data were used to construct a low-resolution dummy atom model. The crystal structure of the complex reasonably ts the SAXS-derived model envelope (Fig. 7b). Together, the crystal

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

IL-18R/IL-1RI

IL-18R /IL-1RAcP

D2 axis

IL-18/IL-1

D2 axi

axi

axi

axi

axi

axi

axi

i

axi

axi

axi

xi

axi

axi

axi

xi

axi

axi

axi

axi

axi

axi

IL-18R/IL-1RI

ax

axisssssssssssssssssssss

IL-18R/IL-1RAcP IL-18

/IL-1

180

D3 axis

D2 axis

IL-18R/IL-1RII

D3 axis

IL-18/IL-1

IL-18R /IL-1RAcP

180

D1 D2 D2 D1

D3 D3

IL-18R

90

IL-1RAcP

Figure 3 | Structural comparison of the IL-1 family cytokine ternary complexes. (a) Structural alignment of IL-18/IL-18Ra/IL-18Rb (blue) and IL-1b/IL-1RI/ IL-1RAcP (4DEP, red) using the binary complex region. The backbone Ca RMSD between IL-18/IL-18Ra and IL-1b/IL-1RI is 4.354.36 . (b) The orientation of D2 and D3 from IL-18Rb and IL-1RAcP in the complexes. The dotted lines show the approximate orientation of the longitudinal D2 and D3 axes.

A close-up of the binary complex interface, which recognizes the co-receptor protein D2 domains, is also shown. (c) A comparison of IL-18/IL-18Ra/ IL-18Rb (blue) and IL-1b/IL-1RII/IL-1RAcP (3O4O, green). The backbone Ca RMSD was 4.39 surrounding the binary complex portion. (d) A close-up of the b4-b5 loops of the ligands in IL-18/IL-18Ra/IL-18Rb (left) and IL-1b/IL-1RII/IL-1RAcP (right). (e) Superimposition of IL-18Rb and IL-1RAcP(from 3O4O). The IL-18Rb-D1 orientation relative to D2 is the opposite of IL-1RAcP.

6 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

IL-18 in complex with IL-18R

IL-18 in complex with yldvIL-18BP

-helix I

Site I

Site II

IL18

D2-D3 linker

IL-18R

D3

Free IL-18 (NMR)

Ser127

Asp35

Val125

Val125

Ser127

Val125

Val125

Gln124

Thr34

Thr34

Gln124

Gln124

D2

Arg123

Arg123

Asp35

Asp37 Ser36

Ser36

Asp37

Asp37

IL18

D1 D2

Arg25

Arg25

Thr29

120

Met33 Arg25

Arg25

Met33

Asp37

Ala42

D1

Ala42

Asp40

Asp40

Asn41 His27

Asn41 His27

His27

His27

Gln124

Thr29

2.5~3.0 []3.1~4.0 []

Pro43

Pro43

Figure 4 | IL-18 structural perturbations on binding IL-18Ra or other proteins. (a) Superimposition of the structures of IL-18 in free and various complex forms: crystal structure of IL-18 with CHPAS (red); the solution structure (1J0S, raspberry); the crystal structure of IL-18 in complex with

Ectromelia virus IL-18BP (3F62, pink); the crystal structure of IL-18 complexed with murine reference antibody 125-2H Fab (2VXT, salmon); the crystal structure for IL-18 in complex with YLDV 14L IL-18BP (4EEE, brown); the crystal structure for IL-18 in complex with a DVD-Ig molecule (4HJJ, gray); the crystal structure of IL-18 in complex with IL-18Ra (blue). (b) The zoom window of the dashed line box shown in 2a. (c) The stabilized structure and IL-18 a-helix I interactions. The structure of IL-18 a-helix I stabilized by interactions with IL-18Ra.

structures solved in this study are consistent with the data collected in the solution state, which underscores the physiological signicance of the crystal structures.

Effects of mutation on the signaling and binding afnities. To conrm the importance of the IL-18 Site III interactions in signal transduction, we performed cell-based assays using IL-18Rb mutants (Fig. 8a). IL-18Rb-WT and its mutants were transiently expressed in HEK293 cells with a background of stably expressed IL-18Ra; the NF-kB activity was measured using the luciferase reporter system with or without an IL-18 stimulus. A complete loss of function was observed for the IL-18Rb-E210A-Y212A

Y214A triple mutants, while IL-18Rb-E210A, -Y212A and -K313A exhibited approximately half the activity compared with -WT. Remarkably, only a minor decrease in activity was observed when the D1 region of IL-18Rb was deleted (D1565 and D15

146, Fig. 8b), likely because IL-18Rb-D1 is distal to the other parts of the ternary complex and seemingly does not affect binding. However, a substantial decrease in activity was observed when the D1-D2 loop (D15153) was deleted, and the activity was fully abolished in the D1/D2-decient experiments (D15176 and D15243). Therefore, IL-18Rb-D2 D3 is sufcient, but D1 is

not essential for signalling.

To identify the amino-acid residues that are critical for ternary complex formation, we performed a binding study using surface plasmon resonance (SPR) analysis (Supplementary Figs 5 and 6 and Supplementary Table 1). The prominent binding residues are summarized in Table 2 and Supplementary Fig. 7. First, IL-18Rb was immobilized on the sensor chip, and binary complexes that formed between one of IL-18 mutants and IL-18Ra were examined for binding. When the binary complex contained

either of three IL-18 mutants, G108A, H109A or K112A, it lost afnity for IL-18Rb, even though these mutants maintained full binding activity for IL-18Ra. Thus, the mutated residues are only important for IL-18Rb binding. Next, one of IL-18Rb mutants was immobilized on the sensor chip, and the IL-18/IL-18Ra binary complex was examined for binding. Our data show that IL-18Rb-Y212A did not bind the binary complex, which is consistent with the structure, wherein substantial p-p stacking were observed between Tyr212Rb and His109IL-18 (Fig. 5c). IL-18Rb-K313A and -E210A exhibited 7- and 20-fold lower afnity for the IL-18/IL-18Ra complex relative to the wild-type receptor, respectively. These data are also consistent with the NF-kB luciferase reporter assay and the structure, wherein the mutated residues extensively interact with His109IL-18, Asp110IL-18, Lys112IL-18 and Phe135Ra (Fig. 5c and Supplementary Fig. 3a).

To determine the contribution of the IL-18Ra N-linked oligosaccharides to the receptor complex formation, we mutated two IL-18Ra asparagine residues to glutamine (Table 2). The afnity of IL-18 for IL-18Ra decreased to one-third when Asn297Ra was mutated to Gln compared with the wild type, which indicates that the sugar chain is important for the recognition of ligand, wherein the Asn297Ra core-FUC and second-NAG are proximal to IL-18 (Fig. 6d). However, we did not observe a different afnity on mutating Asn236Ra.

DiscussionIt is well-known that the IL-1 family system has only two co-receptors, IL-1RAcP and IL-18Rb, despite its seven agonists (IL-1a, IL-1b, IL-18, IL-33, IL-36a, IL-36b and IL-36g) and four

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

IL-18R D1

IL-18

d1

f1

e1

b1 Ser42

Cys81

Cys81

Cys43

Ser42

Cys41

Cys41

Asp37 Arg25

Asp37

Arg25

Ser24

Ser24

1

Cys22 Cys43

Cys22

Glu31 Asp132

Asp132

Glu31

Asp32

Met33

Met33

Arg13

Asp32

11 10

Ser127

Asp17

Ser127

Asp17

Lys128

Lys128

Arg13

IL-18R

Site I

IL-18R

D2 D1

D2

D1

D3

D2

IL-18

D3

IL-18R D1

D3

D3

D2

Site III

11

Gly145

Glu210

Gly145

Tyr214

Tyr214

2

Arg147

Tyr212

Arg147

Tyr212

Asp213

Asp213

Val211

Met150

Met150

9

Lys112

Lys112

Val211

Glu210

His109

His109

8

Gly108

Gly108

Lys313

Lys313

Asp110

Asp110

Lys93

Lys93

D1

Met60

Met60 Leu5

Lys53

IL-18

IL-18R

Glu254

Glu254

Leu5

Lys53

Glu263

Val246

Val246

Ala301

Ala301

D2

Glu253

Glu253

Tyr248

Trp249

D3

Glu263

Tyr248

Site II

Trp249

IL-18R

Figure 5 | The key residues involved in forming the ternary signalling complex. (a) The interactions in the IL-18 receptor binding Site I. In additionto the area surrounding the stable a-helix I, Asp132IL-18 at the tip of the b10-b11 hairpin exhibits ionic interactions with the structure, which are formed by the two unique disulde bonds in IL-18Ra. (b) Interactions surrounding the IL-18 Site II. The key residue of Lys53IL-18 is recognized by the IL-18Ra acidic surface through multiple hydrogen bonds. (c) The interface between Site III of IL-18 and IL-18Rb. In addition to His109IL-18/Tyr212Rb stacking, multiple hydrogen bonds were observed: the His109IL-18 Ne-H to the Val211Rb backbone carbonyl oxygen as well as the Lys112IL-18 Nz-H to the Glu210Rb side chain and Tyr212Rb OZ.

primary receptors (IL-1RI, IL-18Ra, ST2 and IL-36R)29. Thus, certain receptors must be used by multiple ligands (Fig. 2a). In fact, IL-1RAcP is commonly involved in signals with six ligands other than IL-18, which is recognized by its specic receptors, IL-18Ra and IL-18Rb. Accordingly, the IL-18 binding mode remained unclear even after several crystal structures of the IL-1b and receptor complexes became available. The IL-18 complex structures determined in this work exhibited essentially the same binding mode as IL-1b2325. This feature was conrmed under more physiological conditions by using two solution techniques, NMR and SAXS. On the basis of these results, it is highly likely that the left binding mode is the common mode in the IL-1 family; however, the IL-36/IL-36R/IL-1RAcP ternary complex structure has not been determined. Although the general binding mode was conserved, substantial differences were observed at the ligand-receptor or receptor-receptor interfaces in the IL-18 and IL-1b ternary complexes, which was expected based on the signicant sequence deviations (Supplementary Fig. 1). The pronounced effects from such deviations include the unique compositions of secondary structure elements in IL-18Ra-D1 and IL-18Rs-D2, such as the lack of a b-strand that is typically conserved among Ig-like domains. This feature has led to the unique subdomain orientation of IL-18Rb, and may partly explain why IL-1RAcP exhibits the promiscuous ligand recognition mode, although IL-18Rb does not. On the basis of the presumable existence of d2 in IL-36R-D2 and the IL-36 (ref. 27) structure, it is likely that the IL-36 system also adopts the left binding mode with similar orientation to the IL-1b system.

Furthermore, we found a potential common trait among the IL-1 family, which is that the N-linked glycans presumably bridge two receptors in the ternary complex. Intriguingly, this feature was only observed in two functional ternary complexes (IL-1b/IL-1RI/IL-1RAcP and IL-18/IL-18Ra/IL-18Rb) but was absent in the decoy ternary complex (IL-1b/IL-1RII/IL-1RAcP) and may contribute the binding afnity. Nevertheless, our SPR analyses showed that the difference in IL-18Ra-binding afnity for IL-18Rb was only trivial even without glycosylation (N236QRa in Table 2). This discrepancy may be due to the distinct sugar modication patterns between insects and mammals. N-glycans are mostly pauci-mannose (Fig. 6b, not greater than three MAN) oligosaccharides in silkworm30, while mammals contain more varied outer sugar chains; certain chains are longer with more complicatedly mixtures (that is, MAN, NAG, galactose and sialic acid). Thus, larger interaction surfaces on N236Ra sugar chains in humans are expected, which could strengthen the afnity. In contrast, one N-glycan chain likely plays a unique role in the IL-18 system, as the IL-18Ra Asn297Ra sugar chain moderately interacts with IL-18 in both the binary and ternary complexes (Fig. 6d). In fact, IL-18Ra without the sugar chain (N297Q) exhibited a threefold lower binding afnity (Table 2). To our best knowledge, such sugarligand interactions have not been reported for other IL-1 family members. For the IL-1b complexes, any N-glycan chains on IL-1RI(RII) appear too distal to the ligand for direct contact.

IL-1 family activity in vivo are modulated by the counteractions of natural inhibitors, such as receptor antagonists (IL-1Ra, IL-36Ra and IL-38), soluble receptors (sIL-1RI, sIL-18R and

8 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

High-mannose (for structural analysis)

1,61,2 1,6 1,6

-linkage between C1 and N

1,3

Asn91

1,2

1,2

1,6

1,3

1,2

1,3

1,4 1,4

Asn102 Asn203

Pauci-mannose (for SPR)

1,6

Asn NAG FUC MAN

1,4 1,4

NAG-(FUC)-NAG

IL-18

IL-18R-D2

NAG-NAG-MAN-(MAN)2

Gln56

IL-18R-D3

His190

His190

Pro57

Pro57

Gln56

His117

His117

Arg114

Arg114

Loop(D1-D2)

NAG-(FUC)-NAG-MAN

Val257

Val257

Asp259

Asp259

+IL-18R

Binary complex (NAG-NAG-MAN) Ternary complex (NAG-NAG)

Figure 6 | The sugar contacts that maintain IL-18/IL-18Ra/IL-18Rb. (a) The N-linked glycans identied from the structure. (b) The N-linked glycosylation in this study. (c) Intramolecular interactions between the carbohydrates on Asn150Ra and His190Ra as well as on Asn197Ra and the D1-D2 loop.

(d) Interactions between the N-linked carbohydrates on Asn297Ra and IL-18. Core-FUC moderately interacts with IL-18 at an B4 distance. (e) Interactions between the N-linked carbohydrates on Asn236Ra and IL-18Rb.

sST2) and IL-18BP. These proteins tightly bind their target cytokines or receptors, which impedes formation of ternary complexes that initiate intracellular signal transduction. This balancing feature facilitates secure control of inammatory responses in vivo. Our structures not only demonstrate the receptors molecular recognition mode for IL-18, but they also explain the IL-18 inhibitors mechanism of action.

Superimposition of the IL-18/IL-18Ra and IL-18/IL-18BPs complex structures20,22 shows that the IL-18BP binding site on IL-18 precisely corresponds with the IL-18Ra-D3 binding site.

Thus, IL-18BP clearly inhibits IL-18Ra binding through steric hindrance (Supplementary Fig. 8a). Recently, IL-37 (IL-1F7b) was proposed to function as an anti-inammatory cytokine31. IL-37 weakly bound IL-18Ra with 50 times lower afnity than IL-18

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

Table 2 | Functional epitopes of IL-18 and IL-18 receptors.

Kd (10 8 M)

Ligand: IL-18Rb WT

Analyte: IL-18 mutant/IL-18Ra WT complex

IL-18 WT(control) 1.50.1 IL-18 G108A No binding IL-18 H109A No binding IL-18 D110A 5.60.8 IL-18 K112A No binding IL-18 G145A 2.80.1 IL-18 R147A 5.70.3 IL-18 M150A 6.00.5

Ligand: IL-18Rb mutantAnalyte: IL-18 WT/IL-18Ra WT complex

IL-18Rb WT (control) 1.50.1 IL-18Rb L167A 3.30.4

IL-18Rb E210A 32.70.9 IL-18Rb Y212A No binding

IL-18Rb Y214A 5.00.5 IL-18Rb K313A 10.90.9

Ligand: IL-18Rb WTAnalyte: IL-18WT/IL-18Ra mutant complex

IL-18Ra WT (control) 1.50.1 IL-18Ra N236Q 1.70.3

Ligand: IL-18Ra mutant Analyte: IL-18 WT

IL-18Ra WT (control) 6.90.2 IL-18Ra N297Q 25.82.6

Ligand: the interactant captured on the sensor surface. Analyte: the interactant in free solution. WT: wild type. Values are the meanss.d. of the results derived from three independent experiments.

180

180

Figure 7 | The solution-state IL-18/IL-18Ra/IL-18Rb binding mode. (a) The results of cross-saturation experiments with [2H15N]-IL-18/IL-18Ra (forest) and the chemical shift change of [2H15N]-IL-18/IL-18Ra on adding

IL-18Rb (orange) are coloured on the crystal structure of IL-18 in complex with IL-18Ra and IL-18Rb. (b) Superimposition of the IL-18/IL-18Ra/IL-18Rb crystal structure and the low-resolution SAXS envelope.

a

IL-18

0 20 40 60 80 100

0 20 40 60 80 100

(R.L.U.)

(R.L.U.)

mock wild S169A-T170A

L167A S169A

T170A Y214A E210A Y212A K313A E210A-Y212A-Y214A

mock

wild

(15-65)

(15-146)

(15-153)

(15-176)

(15-243)

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

(Fig. 2a), and did not activate signal transduction32. Consistent with this notion, the essential IL-18 amino acids for the IL-18Rb interactions, as well as for IL-18Ra, determined in this study are not conserved in IL-37 (Supplementary Fig. 1a), which suggests that IL-37 is incapable of forming the ternary complex that enables signal transduction. In addition to the natural inhibitors, several anti-IL-18 antibodies have been developed for therapeutic purposes21,33. The structures herein clearly explain how these antibodies impede binding between the ligand and receptors (Supplementary Fig. 8b).

The molecular interface differences herein are important because they may provide numerous opportunities for designing molecules that specically stabilize or interrupt IL-18 ternary complex formation without affecting other IL-1 family members. Unlike molecules that target more downstream signalling pathways, these molecules can be used to treat ligand-specic diseases and avoid side effects due to interference with common intracellular signalling pathways. The structural information collected in this work will facilitate development of an IL-18 modulator in multiple ways. For example, recombinant derivatives of IL-18, which bind to IL-18Ra more strongly relative to wild type of IL-18, can be logically designed based on the structures34. The structures will especially benet small molecule design; small molecules are generally more cost effective once developed, and thus preferable to proteinous compounds. The data obtained in this study provide an atomic framework for molecular interfaces between IL-18 and receptors as well as between the two receptors, which serve as promising drug target sites, and hence will aid in development of effective IL-18 inhibitors.

*

*

*

*

b

IL-18

*

*

*

Figure 8 | The effect of mutations on proteinprotein interfaces for signal transduction. Cell-based NF-kB activity assay for IL-18Rb with(a) an alanine substitution and (b) N-terminal deletion. Each column indicates the relative luciferase activity unit (R.L.U.), for the stimulated cells( ) and the non-stimulated cells( ). The data are the means.d. for

triplicate experiments. Asterisks inside the graphs indicate the signicance from comparing mutant versus wild-type IL-18 Rb, which was determined using a one-way ANOVA with Bonferronis multiple comparison test;

*Po0.05. This experiment is representative of three independent experiments.

10 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

In summary, the structures of IL-18 and its receptor have advanced our precise understanding of molecular recognition and suggest a single architectural paradigm for signalling complexes in the IL-1 family cytokines. Furthermore, with biochemical and cellular data, the structures reveal detailed interaction properties at the molecular interfaces, presenting an atomic framework that will aid in rational drug development for IL-18-related diseases.

Methods

Construction of expression vectors. The coding region of the extracellular domains of human IL-18Ra (NM_003855, residues 20329) and IL-18Rb (NM_003853, residues 15356) were cloned into the pFastBac1 vector (Invitrogen, Carlsbad, CA, USA). Full-length IL-18Rb was also cloned into the pcDNA3.1

vector (Invitrogen). The pGL4.32[luc2P/NF-kB-RE/Hygro] vector was used as an NF-kB luciferase reporter, and the pGL4.70[hRluc] vector was used as an internal control Renilla luciferase reporter; both were purchased from Promega (Fitchburg, WI, USA). For the crystallographic studies, the signal-peptide sequence for Sf9 insect cells, an 8 His tag and an HRV 3C protease cleavage site were placed

immediately upstream of the mature sequence35. The same constructs with a C-terminal 6 His tag and without the HRV 3C site were also prepared for

solution structure analysis and SPR experiments. Mutations were introduced into

the pGEX4T-1[IL-18]19, pFastBac1[IL-18Rs] and pcDNA3.1 [IL-18Rb] vectors

using an inverse PCR-based site-directed mutagenesis method.

Protein expression and purication. Mature IL-18 (residues 1157) and its single amino-acid mutants were prepared as previously reported19. In brief, human IL-18 was expressed as a glutathione S-transferase (GST) fusion protein in the Escherichia coli strain BL21(DE3) (Novagen, Madison, WI, USA). GST-tagged IL-18 was afnity puried followed by GST digestion with Factor Xa and further puried using gel ltration column chromatography. Expression of IL-18Rs using Sf9 insect cell system and their detailed purication procedures were also described35. The extracellular human IL-18Ra or IL-18Rb domains were each separately secreted from Sf9 insect cells (Invitrogen) for structural analyses. The supernatant was puried through three chromatography steps, including ion exchange chromatography, Ni-NTA afnity chromatography and gel ltration chromatography with or without an HRV 3C treatment. To obtain the IL-18/IL-18Ra binary and IL-18/IL-18Ra/IL-18Rb ternary complexes, IL-18, IL-18Ra and

IL-18Rb were mixed at equimolar ratios and puried through gel ltration chromatography. Both the wild type and the mutants of the IL-18Rs for SPR analysis were expressed using a silkworm system36,37. The donor plasmids of the pFastBac1 vectors containing the IL-18 receptor gene were transformed into Escherichia coli BmDH10Bac. Then, 1 mg of BmNPV bacmid DNA and 1 ml of

Cellfectin reagent (Invitrogen) suspended in Grace insect cell medium were injected into the ventral side of B. mori silkworm larvae. After 6 days, haemolymph was recovered from the larvae, and sodium thiosulfate (nal 0.5%) and EDTA were immediately added. The IL-18 receptor proteins from the silkwormwere puried using the same protocols as the purication from the Sf9 insect cell system.

Cell culture. HEK293 cells (Japanese Collection of Research Bioresources, Osaka, Japan) were cultured in Dulbeccos modied Eagles medium (high glucose-containing D-MEM, Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (Sigma-Aldrich, Missouri, USA), penicillin (100 unit ml 1) and streptomycin (100 mg ml 1). All cells were incubated at 37 C in a humidied atmosphere of 5% CO2.

Luciferase reporter gene assay. HEK293 cells were transfected with an empty pcDNA3.1 vector or pcDNA3.1 [IL-18Rb] (wild-type or mutants),

pGL4.32[luc2P/NF-kB-RE/Hygro] and pGL4.70[hRluc] vectors using Lipofectamine 2000 reagent according to the manufacturers instructions. These transfectants were stimulated with recombinant human IL-18 (10 ng ml 1) for 6 h.

The luciferase reporter gene activities were analysed using a Dual-Luciferase Reporter Assay System (Promega). The statistical signicance of the differences was determined using one-way ANOVA with Bonferronis multiple comparison test. The statistical signicance was assigned to be Po0.05.

Surface plasmon resonance analysis. The real-time binding afnities between IL-18 and IL-18Ra and between IL-18Rb and IL-18/IL-18Ra were analysed using a BIAcore 3000 surface plasmon resonance spectrometer (GE Healthcare, Little Chalfont, UK) at 25 C with a Ni-NTA sensor chip. The Kd (dissociation constant)

estimated using the Ni-NTA sensor chip was an average of one order of magnitude lower than estimates obtained using an Anti-His-tag Ab covalently linked to a CM5 sensor chip. However, we used the Ni-NTA sensor chip because it did not decrease the binding capacity after repeated measurement cycles, which is desirable when comparing the Kd of many mutants.

C-terminal 6 His tagged IL-18Ra was immobilized approximately

200-resonance units (RU) on an NTA sensor chip. Then, various concentrations of IL-18 in HBS-P (10 mM HEPES-Na, pH7.4, 150 mM NaCl, 0.01% (v/v) surfactant P-20) buffer were injected over the sensor surface as an analyte at a ow rate of 30 ml min 1 for 180 s. After association, it was allowed to run for another 360 s for dissociation. In the same way, C-terminal 6 His tagged IL-18Rb was immobilized

approximately 100 RU on the sensor chip, and un-tagged IL-18Ra that was saturated with 500 nM IL-18 in HBS-P buffer was injected. The sensor surface was regenerated with 350 mM EDTA. For the mutational analysis, mutants of IL-18, IL-18Ra and IL-18Rb were used instead of wild type, as shown in

Supplementary Table 1. The sensor chip was analysed using BIAevaluation software (GE Healthcare). Analyses with the same concentration series were performed in triplicate.

Crystal structure determination. Step by step crystallization method and preliminary crystallographic analysis were done as described35. X-ray diffraction data were collected at 100 K on the BL38 (IL-18) or BL44XU (IL-18/IL-18Ra) beamlines at SPring-8 (Harima, Japan) and on the BL17A beamline (IL-18/IL-18Ra/IL-18Rb) at Photon Factory (Tsukuba, Japan). For IL-18 and IL-18/IL-18Ra, the diffraction data were processed using the XDS38 and SCALA39,40 software. The initial phases were determined by molecular replacement using Phaser41 with the crystal structure of IL-18 (ref. 20) (PDB code: 3F62) as the search model. The initial phases of the binary complex were improved by NCS averaging using RESOLVE42.

The model was further manually built using COOT43 and rened using BUSTER44 with autoNCS45. In the IL-18 and IL-18/IL-18Ra crystals, four and six copies of

IL-18 and the binary complexes were observed in each ASU, respectively. The structures of these copies in the ASUs are essentially the same, so we referred to chain A for IL-18 and chain A/B for IL-18/IL-18Ra, if not otherwise indicated. The structure of IL-18 included 13 molecules of CHAPS. The diffraction data of IL-18/ IL-18Ra/IL-18Rb were processed using HKL2000 software46. To determine the initial phase for IL-18/IL-18Ra/IL-18Rb, a rened structure of the binary complex was used as the search model for molecular replacement. The model was further manually built using COOT and rened using BUSTER. Only one copy of the ternary complex was in the IL-18/IL-18Ra/IL-18Rb crystal ASU. Ramachandran diagrams were examined using RAMPAGE47. Structural gures were prepared using PyMol (Schrdinger, LLC). The secondary structures were assigned using the DSSP software48.

NMR spectroscopy. All NMR spectra were measured at 308 K on a Bruker Avance II 700 MHz spectrometer equipped with cryogenic probes using TROSY-type pulse sequences. The samples for NMR measurements were in 20 mM potassium phosphate (pH 6.0), 50 mM KCl in H2O/D2O (90%/10%). Spectra were processed using NMRPipe49 and analysed using Sparky analysis software50.

Chemical shift assignments for IL-18 bound to IL-18Ra are based on a HN(CO)CA/HNCA data set and conrmed by HNCACB and NOESY spectra with a mixing time of 200 ms. The IL-18Ra binding surface of IL-18 in solution is dened by the cross-saturation28 intensity ratio of I2000ms/I0ms. The IL-18Rb binding site of IL-18 is inferred from the titration experiment, by shifted or eliminated peaks, which can be assigned for all but the IL-18Ra binding surface.

Structural gures were prepared using PyMol (Schrdinger, LLC).

Small-angle X-ray scattering. SAXS data of the IL-18/IL-18Ra/IL-18Rb ternary complex were collected at the beamline 12ID-B of the Advanced Photon Source at Argonne National Laboratory (Argonne, IL, USA). The sample concentrations were 1, 3 and 5 mg 1 ml 1. The samples were run at 12 keV radiation energy, with a sample-to-detector distance of 2 m. The scattered X-rays were measured using a Pilatus 2 M detector. A ow cell was used to reduce radiation damage. Thirty images were taken for each blank and each sample.

After background subtraction, the data were superimposed using Primus51 (Supplementary Fig. 4f). S is the momentum transfer equal to 4psin(y/2)/l, where y and l are the scattering angle and X-ray wavelength, respectively. Rg is the radius of gyration, which was determined using the Guinier approximation of the data in the low s region (sRgo1.3), the linearity of which also served as an initial assessment of data quality (Supplementary Fig. 4g). The maximum particle dimension, Dmax, and the distance distribution function, P(r), were calculated using auto GNOM52 (Supplementary Fig. 4h). The low-resolution envelopes of the ternary complex were produced using DAMMIN53 by directly tting the reciprocal space scattering prole. Fifteen DAMMIN models were generated and then aligned and averaged using DAMAVER and DAMFILT54. Structural gures were prepared using Chimera55.

References

1. Okamura, H. et al. Cloning of a new cytokine that induces IFN-gamma production by T cells. Nature 378, 8891 (1995).

2. Kuida, K. et al. Altered cytokine export and apoptosis in mice decient in interleukin-1 beta converting enzyme. Science 267, 20002003 (1995).3. Li, P. et al. Mice decient in IL-1 beta-converting enzyme are defective in production of mature IL-1 beta and resistant to endotoxic shock. Cell 80, 401411 (1995).

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340

4. Ghayur, T. et al. Caspase-1 processes IFN-gamma-inducing factor and regulates LPS-induced IFN-gamma production. Nature 386, 619623 (1997).

5. Rathinam, V. a. K., Vanaja, S. K. & Fitzgerald, K. a. Regulation of inammasome signaling. Nat. Immunol. 13, 3332 (2012).

6. Ohnishi, H. et al. TRAM is involved in IL-18 signaling and functions as a sorting adaptor for MyD88. PLoS ONE 7, e38423 (2012).

7. Adachi, O. et al. Targeted disruption of the MyD88 gene results in loss of IL-1-and IL-18-mediated function. Immunity 9, 143150 (1998).

8. Hoffman, H. M., Mueller, J. L., Broide, D. H., Wanderer, A. A. & Kolodner, R.D. Mutation of a new gene encoding a putative pyrin-like protein causes familial cold autoinammatory syndrome and Muckle-Wells syndrome. Nat. Genet. 29, 301305 (2001).9. Park, H., Bourla, A. B., Kastner, D. L., Colbert, R. a. & Siegel, R. M. Lighting the res within: the cell biology of autoinammatory diseases. Nat. Rev. Immunol. 12, 570580 (2012).

10. Dinarello, C. a., Simon, A. & van der Meer, J. W. M. Treating inammation by blocking interleukin-1 in a broad spectrum of diseases. Nat. Rev. Drug Discov. 11, 633652 (2012).

11. Alboni, S., Cervia, D., Sugama, S. & Conti, B. Interleukin 18 in the CNS.J. Neuroinamm. 7, 9 (2010).12. Mellins, E. D., Macaubas, C. & Grom, A. A. Pathogenesis of systemic juvenile idiopathic arthritis: some answers, more questions. Nat. Rev. Rheumatol. 7, 416426 (2011).

13. Hirota, T. et al. Genome-wide association study identies eight new susceptibility loci for atopic dermatitis in the Japanese population. Nat. Genet. 44, 12221226 (2012).

14. Dinarello, C. A., Novick, D., Kim, S. & Kaplanski, G. Interleukin-18 and IL-18 binding protein. Front. Immunol. 4, 289 (2013).

15. Romberg, N. et al. Mutation of NLRC4 causes a syndrome of enterocolitis and autoinammation. Nat. Genet. 46, 11351139 (2014).

16. Canna, S. W. et al. An activating NLRC4 inammasome mutation causes autoinammation with recurrent macrophage activation syndrome. Nat. Genet. 46, 11401146 (2014).

17. Terada, M. et al. Contribution of IL-18 to atopic-dermatitis-like skin inammation induced by Staphylococcus aureus product in mice. Proc. Natl Acad. Sci. USA 103, 88168821 (2006).

18. Brydges, S. D. et al. Divergence of IL-1, IL-18, and cell death in NLRP3 inammasomopathies. J. Clin. Invest. 123, 46954705 (2013).

19. Kato, Z. et al. The structure and binding mode of interleukin-18. Nat. Struct. Biol. 10, 966971 (2003).

20. Krumm, B., Meng, X., Li, Y., Xiang, Y. & Deng, J. Structural basis for antagonism of human interleukin 18 by poxvirus interleukin 18-binding protein. Proc. Natl Acad. Sci. USA 105, 2071120715 (2008).

21. Argiriadi, M. A., Xiang, T., Wu, C., Ghayur, T. & Borhani, D. W. Unusual water-mediated antigenic recognition of the proinammatory cytokine interleukin-18. J. Biol. Chem. 284, 2447824489 (2009).

22. Krumm, B., Meng, X., Wang, Z., Xiang, Y. & Deng, J. A unique bivalent binding and inhibition mechanism by the yatapoxvirus interleukin 18 binding protein. PLoS Pathog. 8, e1002876 (2012).

23. Wang, D. et al. Structural insights into the assembly and activation of IL-1b with its receptors. Nat. Immunol. 11, 905911 (2010).

24. Thomas, C., Bazan, J. F. & Garcia, K. C. Structure of the activating IL-1 receptor signaling complex. Nat. Struct. Mol. Biol. 19, 455457 (2012).

25. Liu, X. et al. Structural insights into the interaction of IL-33 with its receptors. Proc. Natl. Acad. Sci. USA 110, 1491814923 (2013).

26. Vigers, G. P., Anderson, L. J., Caffes, P. & Brandhuber, B. J. Crystal structure of the type-I interleukin-1 receptor complexed with interleukin-1beta. Nature 386, 190194 (1997).

27. Gnther, S. & Sundberg, E. J. Molecular determinants of agonist and antagonist signaling through the IL-36 receptor. J. Immunol. 193, 921930 (2014).

28. Takahashi, H., Nakanishi, T., Kami, K., Arata, Y. & Shimada, I. A novel NMR method for determining the interfaces of large protein-protein complexes. Nat. Struct. Biol. 7, 220223 (2000).

29. Garlanda, C., Dinarello, C. a. & Mantovani, A. The interleukin-1 family: back to the future. Immunity 39, 10031018 (2013).

30. Park, E. Y. et al. Human IgG1 expression in silkworm larval hemolymph using BmNPV bacmids and its N-linked glycan structure. J. Biotechnol. 139, 108114 (2009).

31. Nold, M. F. et al. IL-37 is a fundamental inhibitor of innate immunity. Nat. Immunol. 11, 10141022 (2010).

32. Kumar, S. et al. Interleukin-1F7B (IL-1H4/IL-1F7) is processed by caspase-1 and mature IL-1F7B binds to the IL-18 receptor but does not induce IFN-gamma production. Cytokine 18, 6171 (2002).

33. Hamasaki, T. et al. Human anti-human IL-18 antibody recognizing the IL-18-binding site 3 with IL-18 signaling blocking activity. J. Biochem. 138, 433442 (2005).

34. Kim, S. H. et al. Site-specic mutations in the mature form of human IL-18 with enhanced biological activity and decreased neutralization by IL-18 binding protein. Proc. Natl Acad. Sci. USA 98, 33043309 (2001).

35. Kimura, T. et al. Purication, crystallization and preliminaryX-ray crystallographic analysis of human IL-18 and its extracellular complexes. Acta Crystallogr. Sect. F, Struct. Biol. Commun. 70, 13511356 (2014).

36. Motohashi, T., Shimojima, T., Fukagawa, T., Maenaka, K. & Park, E. Y. Efcient large-scale protein production of larvae and pupae of silkworm by Bombyx mori nuclear polyhedrosis virus bacmid system. Biochem. Biophys. Res. Commun. 326, 564569 (2005).

37. Kajikawa, M. et al. Silkworm Baculovirus Expression System for Molecular Medicine. J. Biotechnol. Biomater. S9, 15 (2012).

38. Kabsch, W. Xds.. Acta Crystallogr. D. Biol. Crystallogr. 66, 125132 (2010).39. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D. Biol. Crystallogr. 62, 7282 (2006).

40. Evans, P. R. An introduction to data reduction: space-group determination, scaling and intensity statistics. Acta Crystallogr. D. Biol. Crystallogr. 67, 282292 (2011).

41. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658674 (2007).

42. Terwilliger, T. C. Statistical density modication with non-crystallographic symmetry. Acta Crystallogr. D. Biol. Crystallogr. 58, 20822086 (2002).

43. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D. Biol. Crystallogr. 60, 21262132 (2004).

44. Bricogne, G. et al. BUSTER. 2.10.0 Ed., Global Phasing Ltd, UK (2011).45. Smart, O. S. et al. Exploiting structure similarity in renement: automated NCS and target-structure restraints in BUSTER. Acta Crystallogr. D. Biol. Crystallogr. 68, 368380 (2012).

46. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307326 (1997).

47. Lovell, S. C. et al. Structure validation by Calpha geometry: phi,psi and Cbeta deviation. Proteins 50, 437450 (2003).

48. Joosten, R. P. et al. A series of PDB related databases for everyday needs. Nucleic Acids Res. 39, D411D419 (2011).

49. Delaglio, F. et al. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277293 (1995).

50. Lee, W., Westler, W. M., Bahrami, A., Eghbalnia, H. R. & Markley, J. L. PINE-SPARKY: graphical interface for evaluating automated probabilistic peak assignments in protein NMR spectroscopy. Bioinformatics 25, 20852087 (2009).

51. Konarev, P. V., Volkov, V. V., Sokolova, A. V., Koch, M. H. J. & Svergun, D. I. PRIMUS: a Windows PC-based system for small-angle scattering data analysis.J. Appl. Crystallogr. 36, 12771282 (2003).52. Svergun, D. I. Determination of the regularization parameter in indirect-transform methods using perceptual criteria. J. Appl. Crystallogr. 25, 495503 (1992).

53. Svergun, D. I. Restoring low resolution structure of biological macromolecules from solution scattering using simulated annealing. Biophys. J. 76, 28792886 (1999).

54. Volkov, V. V. & Svergun, D. I. Uniqueness of ab initio shape determination in small-angle scattering. J. Appl. Crystallogr. 36, 860864 (2003).

55. Pettersen, E. F. et al. UCSF Chimera-a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 16051612 (2004).

Acknowledgements

X-ray data collection was supported by SPring-8 (Harima, Japan), Photon Factory (Tsukuba, Japan), Argonne National Laboratory (Illinois, USA, supported by DE-AC02-06CH11357) and Platform for Drug Discovery, Informatics, and Structural Life Science (Japan). We thank T. Tsunaka (Kyoto University) for the use of his beam time of BL44XU at SPring-8. We also thank T. Fukao, N. Kawamoto, K. Tsuji, M. Yamamoto,S. Ninomiya, T. Yano and K. Kasahara (Gifu University) for their scientic advice or technical help. Affymetrix, Inc. provided discontinued detergents to us. SAXS graphics were prepared using the UCSF Chimera package, which developed by the group at the University of California. (San Francisco, USA, supported by NIGMS P41-GM103311). This work was supported by JSPS KAKENHI Grant Number 22370038 to H.T., Grant-in-Aid for JSPS Fellows to N.T., Health and Labour Science Research Grants for Research on Intractable Diseases from the Ministry of Health, Labour and Welfare to H.O., Health and Labour Science Research Grants for Research from the Ministry of Health, Labour and Welfare to Z.K., Science Research Grant from Eishukai to Z.K, and MEXT KAKENHI Grant Number 21790979 to T.K.

Author contributions

H.T. and Z.K. designed the work; M.S and N.K. supervised the study; N.T. carried out the experimental design, experiments and structural analyses, supported by T.K., K.A., M.A. and H.T.; T.K. prepared the samples for the study, supported by H.O., N.T., K.M. and E.Y.P.; T.K. performed and analysed the SPR experiments; T.Y. performed the cell assays,

12 NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6340 ARTICLE

supported by H.O. and T.K.; X.Z. performed the SAXS measurements; and N.T., T.K., K.A., H.O., Z.K. and H.T. contributed to writing the manuscripts and preparing the gures.

Additional information

Accession codes: The atomic coordinates and structure factors for IL-18, IL-18/IL-18Ra and IL-18/IL-18Ra/IL-18Rb have been deposited in the RCSB Protein Data Bank under accession codes 3WO2, 3WO3 and 3WO4, respectively.

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Tsutsumi, N. et al. The structural basis for receptor recognition of human interleukin-18. Nat. Commun. 5:5340 doi: 10.1038/ncomms6340 (2014).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 5:5340 | DOI: 10.1038/ncomms6340 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

& 2014 Macmillan Publishers Limited. All rights reserved.