Structure Article

A Platform of C-type Lectin-like Receptor CLEC-2 for Binding O-Glycosylated Podoplanin and Nonglycosylated Rhodocytin

Masamichi Nagae,1 Kana Morita-Matsumoto,1 Masaki Kato,1 Mika Kato Kaneko,2 Yukinari Kato,2 and Yoshiki Yamaguchi1,* 1Structural Glycobiology Team, Systems Glycobiology Research Group, RIKEN-Max Planck Joint Research Center, RIKEN Global Research Cluster, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan 2Department of Regional Innovation, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan *Correspondence: [email protected] http://dx.doi.org/10.1016/j.str.2014.09.009

SUMMARY (Colonna et al., 2000), and then recognized as the receptor for rhodocytin, the snake venom toxin. Rhodocytin, produced by Podoplanin is a transmembrane O-glycoprotein that the Malayan pit viper (Calloselasma rhodostoma) is a nongly- binds to C-type lectin-like receptor 2 (CLEC-2). cosylated heterodimeric composed of one a and one b The O-glycan-dependent interaction seems to play subunit (Shin and Morita, 1998)(Figure 1) and induces platelet crucial roles in various biological processes, such aggregation (Suzuki-Inoue et al., 2006). Subsequently, CLEC-2 as platelet aggregation. Rhodocytin, a snake venom, was found to be the physiological receptor for the O-glyco- also binds to CLEC-2 and aggregates platelets in a sylated protein podoplanin (also known as aggrus, OTS8, T1a, E11 antigen, gp38, and PA2.26) (Suzuki-Inoue et al., 2007). glycan-independent manner. To elucidate the struc- Podoplanin is a type I sialomucin-like glycoprotein expressed tural basis of the glycan-dependent and independent in the cardinal vein and lymphatic endothelial cells (Figure 1). interactions, we performed comparative crystallo- Podoplanin-deficient mice have abnormal lymphatic vessels graphic studies of podoplanin and rhodocytin in because of blood-lymphatic misconnections (Schacht et al., complex with CLEC-2. Both podoplanin and rhodo- 2003). Importantly, the same abnormality occurs in mice lacking cytin bind to the noncanonical ‘‘side’’ face of CLEC- glycosyltransferase T-synthase (C1galt1), an enzyme critical for 2. There is a common interaction mode between the expression of podoplanin, suggesting that O-glycosylation consecutive acidic residues on the ligands and the is required for podoplanin function (Fu et al., 2008). Furthermore, same arginine residues on CLEC-2. Other inter- conditional knockout of CLEC-2 in platelets impairs blood/ actions are ligand-specific. Carboxyl groups from lymphatic vessel separation (Osada et al., 2012). Activation of the sialic acid residue on podoplanin and from the CLEC-2 by binding O-glycosylated podoplanin is, therefore, essential for blood/lymphatic separation. C terminus of the rhodocytin a subunit interact Accumulating evidence suggests that many other important differently at this ‘‘second’’ binding site on CLEC-2. biological pathways depend on the CLEC-2-podoplanin inter- The unique and versatile binding modes open a action. CLEC-2 deficiency in dendritic cells impairs their entry way to understand the functional consequences of into the lymphatics and their migration to lymph nodes (Acton CLEC-2-ligand interactions. et al., 2012). Podoplanin expressed on fibroblastic reticular cells activates CLEC-2 on extravasated platelets in the perive- nular space of high endothelial venules (HEVs), thus maintain- INTRODUCTION ing HEV integrity during immune responses (Herzog et al., 2013). The importance of podoplanin has been highlighted by In vertebrates, the growth of blood vessels (angiogenesis) and its frequent upregulation in tumors. Podoplanin is overex- lymphatic vessels (lymphangiogenesis) involves the coordi- pressed in and murine tumors in various tissues (Kato nation of many biological processes such as cell proliferation, et al., 2003, 2005; Kato and Kaneko, 2014) and enhances migration, and cell-cell communication (Adams and Alitalo, hematogenous metastasis by inducing platelet aggregation 2007). The mechanism of angiogenesis has received consider- through activation of CLEC-2 (Kato et al., 2008). In this context, able attention over the past few decades. However, the pro- an anti-human podoplanin antibody, NZ-1 (Kaneko et al., 2012; cesses regulating lymphangiogenesis have only been explored Kato et al., 2006, 2010), which inhibits podoplanin-induced very recently (Tammela and Alitalo, 2010). CLEC-2 (CLEC1B) is platelet activation, abrogates tumor metastasis by inhibiting a type II transmembrane, C-type lectin-like receptor (Figure 1), the interaction of podoplanin and CLEC-2 (Figure 1A). CLEC- and its activation in platelets blocks lymphaticovenous connec- 2 features in viral infections. CLEC-2 on platelets is an attach- tions, thereby helping to separate blood and lymphatic vascular ment factor for HIV type 1 (HIV-1) and mediates HIV-1 capture systems. CLEC-2 was first identified as a cell surface protein through the C-type lectin receptor DC-SIGN (Chaipan et al., expressed in the liver and blood cells, mostly of myeloid origin 2006, 2010).

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1711 Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 1. Molecular Interactions of CLEC-2 and Podoplanin (A) Interaction of CLEC-2 and O-glycosylated podoplanin (physiologically a trans cell-cell interaction). Snake toxin rhodocytin and the anti-podoplanin antibody NZ-1 block the interaction. hemITAM, hemi-immunoreceptor tyrosine-based activation motif. (B) Domain architecture of human podoplanin, CLEC-2, and rhodocytin. The amino acid sequence of three consecutive PLAG domains (PLAG1–PLAG3) is shown, and O-glycosylated Thr52 is indicated with asterisk. The N- and C-terminal sequences of the rhodocytin a subunit, which is crucial for CLEC-2 binding, are indicated. SP, signal peptide; TM, transmembrane domain; CT, cytoplasmic tail.

Podoplanin contains the well conserved three tandem repeats CLEC-2-podoplanin complex (Figure 2B). The two complexes, of the platelet aggregation-stimulating (PLAG) domain in the referred to as complex A and complex B, include Pro101– extracellular region (Figure 1B) (Kaneko et al., 2006). In , Ala220 of CLEC-2 and Ala46–Gly54 and Gly45–Gly54 of the po- only the third PLAG domain, PLAG3, has an attached O-glycan, doplanin peptide, respectively. Structural superposition of the at Thr52 (Kaneko et al., 2007). Its structure is primarily a disialyl two complexes in the asymmetric unit revealed that the root- core 1 (NeuAca2-3Galb1-3(NeuAca2-6)GalNAca1-O-Thr) (Fig- mean-square deviation (rmsd) of 120 corresponding Ca atoms ure 2A) (Kaneko et al., 2007). Because sialylation is essential of CLEC-2 is only 0.4 A˚ (Figure S1A available online). In both com- for podoplanin-induced platelet aggregation, CLEC-2 must plexes, the PLAG2 (Glu38–Gly45) domain and a2-3 linked sialic specifically recognize the sialylated PLAG3 domain. This seems acid are exposed to solvent, and the corresponding electron den- at odds with the fact that rhodocytin interacts with CLEC-2 in the sities are missing (Figure 2C). Two podoplanin peptides in the absence of carbohydrate. The structural basis for these specific asymmetric unit were disulfide-linked with C terminus cysteine interactions is still unknown. residues (Figure 2B). Structural superposition of ligand-bound Here we report the crystal structures of the CLEC-2 lectin CLEC-2 with a ligand-free form (Watson et al., 2007) showed domain in complex with podoplanin O-glycosylated peptide only 0.7 A˚ rmsd for 120 corresponding Ca atoms (Figure S1B), or nonglycosylated rhodocytin. The ligands bind at a locus indicating no apparent structural change on binding. distinct from the canonical binding site, and both use a two- site interaction mode. The C-type lectin fold displays an unex- CLEC-2 Utilizes Its Noncanonical Side Face for Binding pected binding versatility, and the details of this binding mode, to Sialylated O-Glycan and the Adjoining Polypeptide described here, may help in the rational design of anti-platelet of Podoplanin aggregation lead compounds. In complexes A and B, the podoplanin-CLEC-2 interaction modes are almost identical (Table S1), and only that of complex RESULTS A is described hereafter. The C terminus of the PLAG3 domain points in a different direction than the N terminus of CLEC-2 (Fig- The Structure of CLEC-2 Complexed with the ure 2D), an orientation befitting trans cell-cell interaction via the O-Glycosylated Podoplanin PLAG Domain two (Figure 1A). The PLAG3 domain and attached We determined the crystal structure, at 1.9 A˚ resolution, of O-glycan fit into a noncanonical pocket formed by a1 and a2 a complex of the CLEC-2 extracellular lectin domain and a helices and b1 and b10 strands of the CLEC-2 lectin domain, a podoplanin glycopeptide encompassing the PLAG2 and PLAG3 position completely separate from the usual sugar-binding site domains (Glu38–Gly54) with a disialyl core 1 O-glycan (NeuAca2- (b3 and b4 strands) of typical C-type lectins (Figure 2D). Bound 3Galb1-3[NeuAca2-6]GalNAca1-O-Thr) at Thr52 (Figure 2A, PLAG3 peptide has no particular secondary structure. Interac- Table 1). The asymmetric unit contains two copies of the tion with CLEC-2 is mainly observed at Glu47 and Asp48 in the

1712 Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 2. Crystal Structure of CLEC-2 Complexed with the O-Glycosylated Podoplanin PLAG Domain (A) The O-glycosylated podoplanin peptide (Glu38–Gly54) used in this study (top panel). An artificial cysteine residue is introduced at the C terminus. The gray parts are the amino acid and carbohydrate residues with missing electron density in the crystal structure. The chemical structure of O-glycan (disialyl core 1) attached on Thr52 is shown in the bottom panel. The a2-6-linked sialic acid residue, which extensively interacts with CLEC-2, is highlighted in yellow. (B) Two CLEC-2-podoplanin complexes (A and B) in the asymmetric unit. Two CLEC-2 lectin domains bind to disulfide-linked podoplanin peptides. CLEC-2 and the podoplanin peptide are shown as ribbon and wire models, and O-glycan attached at Thr52 is shown as a rod model. (C) The observed electron density of podoplanin. The omit map of the ligand contoured at the 2.5 s level is shown as cyan mesh. The ligand is shown as a rod model. The peptide and carbohydrate moieties of podoplanin are shown in yellow and white, respectively. (D) The overall structure of the CLEC-2-podoplanin complex. The secondary structure element numbering of the CLEC-2 lectin domain is based on a previous report (Zelensky and Gready, 2005). The directions toward membranes are indicated with black arrows.

PLAG3 domain and the a2-6 linked sialic acid residue (Figure 3A; tains core 1 O-glycan with a2–6 linked sialic acid linked through Table S1). The side chains of the two acidic residues make elec- an N-acetyl-D-galactosamine (GalNAc) residue. The sequence trostatic contact with three arginine residues (Arg107, Arg152, motif is conserved among podoplanins of various species and Arg157) of CLEC-2 (Figure 3B). The Asp48 side chain is (Figure 4A) (Kaneko et al., 2006). further connected by hydrogen bonds with the Thr153 side chain and the backbone of His154 of CLEC-2. The a2–6-linked sialic The a2-6-Linked Sialic Acid Residue Is Essential acid residue interacts extensively with CLEC-2 (Figure 3C). The for CLEC-2 Binding carboxyl group of the sialic acid bonds with the side chain The contact surface area of the a2-6-linked sialic acid residue and backbone atoms of Arg118. The hydroxyl groups OH8 and (247 A˚ 2) is almost equal to those of the two consecutive acidic OH9 make hydrogen bonds with the side chains of His119 and residues (Glu47 and Asp48, 247 A˚ 2), suggesting that these Tyr129, respectively, and the N-acetyl oxygen atom interacts two regions contribute significantly to complex formation. To with the side chain of Asn105. Furthermore, the hydrophobic examine the contribution of the O-glycan part for binding face of the sialic acid is surrounded by the aromatic rings of CLEC-2, we performed a solution nuclear magnetic resonance Trp106 and Phe117. The main chain atoms of Asp49-Val50- (NMR) analysis using glycosylated and nonglycosylated podo- Val51, but not the side chain atoms, hydrogen bond with the planin peptides (Figure 3D). The addition of nonglycosylated side chain of Arg118 of CLEC-2 (Figure 3B). The mode of inter- PLAG3 peptide to CLEC-2 solution did not significantly change action suggests that CLEC-2 has a recognition motif in the the 1H-NMR spectrum of CLEC-2. In contrast, the addition of the form ‘‘EDXXXT/S,’’ where X is any amino acid and T (or S) con- podoplanin peptide with a disialyl core 1 O-glycan significantly

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1713 Structure Versatile Ligand Recognition Mechanism of CLEC-2

Table 1. Data Collection and Refinement Statistics rhodocytin possess C-type lectin folds and form a domain- swapped dimer, as observed in typical snake venom C-type CLEC-2-Podoplanin CLEC-2-Rhodocytin Crystal Complex Complex lectin domains (Zelensky and Gready, 2005). In the crystal, the rhodocytin a subunit, b subunit, and CLEC-2 lie along a straight Data Collection Statistics line. Structural superposition of the four complexes reveals that Space group P2 C2 1 the rhodocytin molecules bend slightly at the hinge region of the Cell constant a = 54.4, b = 51.2, a = 130.9, b = 117.1, domain-swapped dimer (Figure S2). Rhodocytin binds to the ˚ ˚ c = 54.1 A, b = 112.8 c = 152.2 A, b = 115.8 same face of CLEC-2 as podoplanin, except that rhodocytin Beamline AR-NW12A BL-5A only interacts through the N- and C-terminal regions of its a sub- Wavelength (A˚ ) 1.0000 1.0000 unit (Figure 5B; Table S2). The contact surface of rhodocytin is Resolution (A˚ )a 100–1.90 (1.93–1.90) 100–3.00 (3.05–3.00) 337–446 A˚ 2 (Table S2). Therefore, it is almost comparable with Completeness (%)a 100 (100) 99.7 (99.9) that of podoplanin. The N-terminal loop region of the rhodocytin a a subunit contains three acidic residues, Glu3, Asp4, and Asp6, Rsym (%) 7.2 (49.5) 9.9 (45.8) Redundancy a 4.1 (4.1) 3.6 (3.6) and these three residues are located close to four basic residues of CLEC-2, Arg107, Arg118, Arg152, and Arg157. Intriguingly, a 24.3 (3.3) 16.9 (2.9) the carboxyl group of C-terminal Tyr136 in the rhodocytin a Refinement Statistics subunit also resides close to Arg118 of CLEC-2. The additional ˚ Resolution (A) 50.13–1.91 37.56–2.98 backbone interactions seem to stabilize complex formation (Fig- R (%) 21.9 28.4 ure 5C; Table S2). The electrostatic interaction may be important

Rfree (%) 25.9 32.5 for complex formation between the negatively charged rhodocy- Rmsd from Ideal Values tin and the positively charged CLEC-2. Bond length (A˚ ) 0.009 0.009 Podoplanin and Rhodocytin Interact with Common Bond angle ( ) 1.386 1.619 Arginine Residues on CLEC-2 Average B factor (A˚ 2) 28.0 60.9 To confirm the contributions of the four arginine residues to rho- Ramachandran Plot docytin binding, we performed surface plasmon resonance Favored (%) 97.6 92.7 assays using rhodocytin and a set of CLEC-2 lectin domain mu- Allowed (%) 2.4 7.1 tants. Rhodocytin was immobilized on a sensor chip, and control Outlier (%) 0.0 0.2 CLEC-2 or its mutants were injected as analytes. The maximum a Values in parentheses are the highest-resolution shells. resonance units (RUmax) were plotted against a series of concen- trations of analytes (Figure 5D). From a steady-state analysis, the

KD of control CLEC-2 was estimated to be 3 mM, which is com- parable with that of a previous report (1 mM; Watson et al., 2007). broadened the signals originating from CLEC-2, clearly indicating All arginine mutations in CLEC-2 greatly reduce the affinity for a glycosylation-dependent CLEC-2 interaction. The broadening rhodocytin. Evidently, these four arginine residues are essential of the CLEC-2 signals can be interpreted as formation of the for binding rhodocytin and podoplanin. 2+2 dimer observed in the crystal. The solution NMR analysis supports the conclusion that the a2-6 sialic acid residue is DISCUSSION essential for the interaction. Generally, C-type lectins bind to both carbohydrate and protein Arginine Residues on CLEC-2 Are Essential ligands through the ‘‘top’’ face (b3 and b4 strands) of the lectin for Podoplanin Binding (Drickamer, 1999; Zelensky and Gready, 2005). The carbohy- To evaluate the contribution of each arginine residue of CLEC-2 drate moiety binds via a calcium ion coordinated at this site. to podoplanin binding, we constructed four CLEC-2 lectin A typical example is the complex between P-selectin and sialyl domain mutants (R107A, R118A, R152A, and R157A) fused to LewisX-containing PSGL-1 (Figure 6B; Somers et al., 2000). Fc proteins (Figure S8) and performed a flow cytometric assay The interaction mode of CLEC-2 as elucidated in our study is using podoplanin-expressing Chinese hamster ovary (CHO) cells quite different (Figure 6A). CLEC-2 lacks the critical calcium- (Figure 3E). All four alanine mutants failed to bind podoplanin- binding site for carbohydrate recognition (Figure 4C) and binds expressing cells, indicating that the arginine residues are critical. the O-glycosylated peptide ligand on the side face (a1, a2, b1, and b10) with no calcium assistance (Figure 3C) as well as nongly- CLEC-2 Utilizes the Same Side Face for Binding cosylated rhodocytin (Figure 5B). The ligand-binding site on to Rhodocytin CLEC-2 is characterized by a clustering of arginine residues. To compare the interaction mode of podoplanin with that of the An electron surface potential map shows a large positive patch nonglycosylated ligand rhodocytin, we further solved the crystal around the ligand binding site (Figure S3). structure of the CLEC-2 lectin domain in complex with rhodo- Structural comparison of CLEC-2-podoplanin and CLEC-2- cytin at a 3.0 A˚ resolution. The asymmetric unit of the crystal rhodocytin complexes reveals no apparent difference in CLEC- contains four CLEC-2-rhodocytin complexes. In each complex, 2 structures (Figure 7A). Intriguingly, the N-terminal region of one CLEC-2 lectin domain interacts with one rhodocytin the rhodocytin a subunit contains a ‘‘Glu-Asp’’ sequence, and molecule via the a´ subunit (Figure 5A). The a and b subunits of the spatial position of this region relative to CLEC-2 coincides

1714 Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 3. Dual Recognition of CLEC-2 for Podoplanin (A) The binding interface of the podoplanin peptide (wire model) and the CLEC-2 lectin domain (surface model). Two acidic residues (Glu47 and Asp48) and carbohydrate moieties are shown as rod models and in yellow. The critical amino acid residues of CLEC-2 are mapped and labeled. (B) The interaction between CLEC-2 and the podoplanin peptide. Amino acid residues involved in the interactions are shown as rod models. Electrostatic interactions and hydrogen bonds are represented as red dashed lines. (C) Recognition of the a2-6-linked sialic acid residue-intermolecular hydrogen bonds and hydrophobic contacts. (D) CLEC-2 binds O-glycosylated podoplanin peptides, but not unglycosylated peptides, in solution. Shown are the 1H-NMR spectra (methyl region) of 100 mM CLEC-2 lectin domain in the absence of ligand (top panel), in the presence of 200 mM nonglycosylated PLAG3 domain (center panel), and 200 mM glycosylated PLAG2-3 domain (bottom panel). (E) Flow cytometry analysis using Fc-fused CLEC-2 lectin domains and podoplanin-expressing CHO cells. well with that of the podoplanin PLAG3 domain (left panel in Fig- The human PLAG2 domain lacks the Glu-Asp doublet (corre- ure 7B). Both complexes exhibit a two-locus interaction, and sponding to Glu47–Asp48 in the PLAG3 domain) and the both use the Glu-Asp signature (Figure 7C). The negatively O-glycosylation site, and the domain does not contribute to charged patch of the Glu-Asp doublet accepts the positively CLEC-2 binding. The PLAG1 domain does have the doublet. charged patch formed by Arg107, Arg152, and Arg157. The However, there is no attached O-glycan (Kaneko et al., 2007), second binding sites are slightly apart, although the carboxyl suggesting no role in binding CLEC-2. Intriguingly, threonine- groups of both the sialic acid on podoplanin and of the to-alanine conversions in the PLAG1 domains of the mouse, C terminus of rhodocytin interact with Arg118 on CLEC-2. In rat, hamster, and dog abolish platelet aggregation (Kato et al., comparison with the rhodocytin complex, the sialic acid in 2003), indicating that species-specific O-glycosylation on the the podoplanin complex is positioned on the opposite side of first or last PLAG domain maintains a podoplanin-CLEC-2 inter- the side chain of Arg118 in CLEC-2 (right panel in Figure 7B). action (Figure 4A). Our data suggest that the sialyl-Tn structure Therefore, CLEC-2 possesses three ligand binding sites. One (NeuAca2-6GalNAca1-O-Thr) in the PLAG3 domain is sufficient is common to both podoplanin and rhodocytin (site 1), but the for CLEC-2 binding. Consistent with this, a previous report other sites (sites 2 and 3) are ligand-specific. Each ligand uses showed that human podoplanin with a sialyl-Tn structure its own second binding site (site 3 for podoplanin and site 2 for in the PLAG3 domain induces platelet aggregation (Amano rhodocytin) (Figure 7C). et al., 2008). The anti-podoplanin antibody NZ-1 reacts with

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1715 Structure Versatile Ligand Recognition Mechanism of CLEC-2

A B

C

Figure 4. Amino Acid Sequence Alignment of Podoplanin, C-Type Lectin-like Snake Venom Proteins, and CLEC-2 (A) Sequence alignment of the podoplanin PLAG domains of human, mouse, rat, hamster and dog. The figure is modified from our previous report (Kaneko et al., 2006). (B) Sequence alignment of N-terminal parts of rhodocytin and related snake venom proteins. Three acidic residues that interact with CLEC-2 are indicated with arrows above the sequence. The figure is modified from the report by Shin and Morita (1998). (C) Amino acid sequence alignment of CLEC-2 lectin domains from human (UniProt ID code Q9P126), mouse (UniProt ID code Q9JL99), rat (GenBank ID code ADK94892), hamster (GenBank ID code ERE66437), dog (GenBank ID code XP_543823), human DC-SIGN (GenBank ID code AAK20997), and human P-selectin (GenBank ID code NP002996) were aligned by ClustalW. The secondary structure of human CLEC-2 is indicated. Amino acid residues that interact with po- doplanin are highlighted in black. EPN and WND motifs that are critical for calcium binding in DC-SIGN and P-selectin are indicated by asterisks. the nonglycosylated podoplanin peptide, and the point muta- where X is any amino acid and O-glycan is attached on T/S, tions of Met43 (PLAG2) and Asp49 (PLAG3) abolish binding to seems to be found ubiquitously in cellular proteins. However, NZ-1 (Figure 1A) (Fujii et al., 2014). It is likely that NZ-1 covers only 39 entries containing both the EDXXXT/S sequence and PLAG2 and PLAG3, especially at the two acidic residues O-glycan predicted at the T or S site are found in the UniProt (Glu47 and Asp48), and blocks access of CLEC-2. This supports database (Table S3). Furthermore, several proteins are either the idea that a two-locus interaction is essential for CLEC-2- core proteins of proteoglycans, such as glypican-2 and synde- podoplanin binding. A sequence comparison of CLEC-2 among can-1, or O-GlcNAclylated proteins, such as peroxisome pro- different mammalian species highlights the conservation of the liferator-activated receptor-g (PPARg). Because the sialylation four Arg residues Arg107, Arg118, Arg152, and Arg157 (human is essential for interaction, the endogenous binding partner protein numbering) responsible for binding podoplanin (Fig- seems almost limited to being podoplanin. Although we cannot ure 4B). Three are especially conserved, although Arg118 is exclude the possibility that a sialylation-independent ‘‘rhodocy- replaced with lysine in the dog (Figure 4C). Our mutagenesis tin-type’’ interaction exists endogenously, the short motif of study clearly shows that these electrostatic interactions between podoplanin must provide, at the very least, a highly selective podoplanin and CLEC-2 are essential for binding. landmark for CLEC-2. Recent evidence suggests that, because The simultaneous recognition of peptide and glycan moieties HIV-1-infected T cells do not express podoplanin, they possess of podoplanin by CLEC-2 must enhance affinity and form the an as yet unidentified CLEC-2 ligand (Chaipan et al., 2010). It has basis for the unique recognition of podoplanin among many also been proposed that CLEC-2 expressed on neutrophils may O-glycosylated proteins. The dissociation constant of CLEC-2 function as a pattern recognition receptor (Kerrigan et al., 2009). for podoplanin is 24 mM(Christou et al., 2008), which is suffi- Other possible endo- and exoligands of CLEC-2 may be identi- ciently strong and in line with other ligand-transmembrane re- fied using the recognition motif as a probe. Analogous with the ceptor interactions. The binding motif of podoplanin, EDXXXT/S, podoplanin-CLEC-2 interaction, a C-type lectin, P-selectin,

1716 Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 5. Crystal Structure of CLEC-2 in Complex with the Snake Venom Rhodocytin (A) Overall structure of the CLEC-2 rhodocytin complex. The a and b subunits of rhodocytin and CLEC-2 molecules are shown in blue, orange, and green, respectively. (B) The interface between rhodocytin and CLEC-2. Rhodocytin and CLEC-2 are shown as wire and surface models, respectively. The four arginine residuesof CLEC-2 are labeled and highlighted in blue. (C) The interaction between rhodocytin and CLEC-2. Complex 3 is shown as a representative structure. Hydrogen bond and electrostatic interaction networks are shown as red dotted lines. (D) Surface plasmon resonance assays. Rhodocytin was immobilized on a sensor chip, and wild-type CLEC-2 (black circle) and four mutants, R107A (green), R118A (blue), R152A (red), and R157A (magenta), were analyzed at a series of concentrations. binds O-glycan and sulfated tyrosine residues of P-selectin 2003; Mizuno et al., 2001). In contrast, rhodocytin has a different glycoprotein-1 (PSGL-1) (Somers et al., 2000). Additionally, sia- interaction mode with CLEC-2 (Figure S4). The area on CLEC-2 lylated O-glycan is essential in the recognition of the paired that contacts rhodocytin partially overlaps with that interacting immunoglobulin (Ig)-like type 2 receptor (PILR2) for endogenous with podoplanin, and mutations to the same four arginines mouse CD99 (Kuroki et al., 2014; Wang et al., 2008) and exoge- dramatically reduce binding of both protein ligands. This sug- nous herpes simplex virus glycoprotein B (gB) (Wang et al., gests that the toxin competes with podoplanin for binding to 2009). From these observations, it is tempting to speculate that CLEC-2. The identification of a ligand-binding site on CLEC-2 O-glycans often assume a specific role as initial recognition casts new light on its oligomerization and subsequent intracel- guides for target proteins, with subsequent involvement of adja- lular signaling. Analysis of the stoichiometry of CLEC-2 on cells cent amino acid residues. suggests that it is normally a dimer but forms larger complexes The N-terminal sequence of the rhodocytin a subunit is very upon rhodocytin binding (Watson et al., 2009). The crystal pack- similar to those of other C-type lectin-like venom proteins, but ing of the rhodocytin complex shows that rhodocytin forms a the rhodocytin a subunit alone possesses the Glu-Asp doublet multimer by associating laterally (Figure S5) and that the inter-

(Figure 4B) (Shin and Morita, 1998). The doublet may be the action mode is different from that of the (ab)2 heterotetramer, key for specific binding to CLEC-2. Crystallographic analyses as reported in unliganded structures (Hooley et al., 2008; Watson of other snake venom toxins (EMS16-Integrin a2-I domain, Bitis- et al., 2008). In contrast, CLEC-2 molecules weakly dimerize in cetin-von Willebrand Factor A1 domain, and Factor-X binding the crystal at a site close to the a3-b3 loop region, which is on protein-Factor-X Gla domain) in complex with their target pro- a side opposite to the ligand binding site (Figure S6). This inter- teins reveal that heterodimeric toxins bind to their target through action may reflect a dimerization mode upon binding to a multi- cavities formed by two subunits (Horii et al., 2004; Maita et al., valent ligand. The human CLEC-2 (CLEC1B) is encoded in

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1717 Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 6. The Side Face of CLEC-2 Is Utilized as a Platform for Ligand Binding (A–D) CLEC-2 in complex with podoplanin (yellow) or rhodocytin (blue) (A), P-selectin in complex with PSGL-1 (PDB ID code 1G1S) (B), the NKG2D dimer in complex with ULBP3 (PDB ID code 1KCG) (C), and the KACL dimer in complex with NKp65 (PDB ID code 4IOP) (D) are shown.

into Escherichia coli strain Rosetta2 (DE3) (Nova- gen, Merck Millipore). The hexahistidine-tagged CLEC-2 protein was found in inclusion bodies. The inclusion bodies were solubilized in 50 mM 2-(N-morpholino)ethanesulfonic acid (pH 6.5), 10 mM EDTA, 2 mM DTT, and 6 M guanidine hy- drochloride at a protein concentration of 5 mg/ml. Solubilized protein was diluted into 2 l of buffer containing 150 mM Tris-HCl (pH 8.0), 1 M L-argi- nine, 2 mM EDTA, 10 mM reduced glutathione, and 0.5 mM oxidized glutathione. The mixture was equilibrated at 4C for 16 hr with slow stirring and then concentrated to a volume of 250 ml by the QuixStand Benchtop system (GE Healthcare). The concentrated protein solution was extensively dialyzed against 2 l of buffer containing 20 mM a natural killer gene complex (NKC) on 12, which Tris-HCl (pH 8.0) and 100 mM NaCl at 4C. The dialyzed protein was then contains a group of C-type lectin receptor , including purified by nickel-nitrilotriacetic acid chromatography. After removal of the NKG2 (A, C, D, E, and F), Ly49A-W, keratinocyte-associated hexahistidine tag by TEV protease, the protein was further purified with size C-type lectin (KACL, also known as CLEC2A), and CD94. Several exclusion column chromatography. Mutant CLEC-2 proteins (C99S/R107A, C99S/R118A, C99S/R152A, and C99S/R157A) were expressed and purified C-type lectin receptors of NKC form homo- or heterodimers via in the same manner. The purity of each protein was confirmed by SDS- their C-type lectin domains (Radaev and Sun, 2003; Yokoyama PAGE (Figure S7A, upper panel) and the proper folding of each mutant by and Plougastel, 2003). Importantly, the ligand binding interface thermal shift assay (Figure S7B). of CLEC-2 we have elucidated partially overlaps the dimerization 0 interface of NKG2s, KACL, and Ly49s (a2 and b1 , Figures 6C Purification of Rhodocytin from Calloselasma rhodostoma Venom and 6D). The putative dimerization interface of CLEC-2 contrasts Lyophilized powder of pure venom from Calloselasma rhodostoma was pur- markedly with those of NKG2D and Ly49. The ligand interface chased from Latoxan. Rhodocytin was purified from the lyophilized powder and crystal packing of CLEC-2 suggest its dimerization mode. as reported previously (Shin and Morita, 1998). In brief, the lyophilized powder The structural basis for the specific interaction between was gently dissolved in a buffer containing 20 mM Tris-HCl (pH 8.0) and 100 mM NaCl. The dissolved solution was separated by centrifugation, and CLEC-2 and O-glycosylated podoplanin can now be under- the supernatant was subjected to gel filtration and anion exchange column stood. Receptor specificity is critically dependent on a two-locus chromatography. The purity of rhodocytin was checked by SDS-PAGE interaction involving both the polypeptide and the carbohydrate (Figure S7A, lower panel). of podoplanin. It explains how a common O-glycan modification, together with a particular adjoining peptide motif, can be Crystallization of CLEC-2 Lectin Domains in Complex with recognized as being unique. Such bifunctional recognition O-Glycosylated Podoplanin Peptide or Rhodocytin could become a fundamental principle for specificity in recep- Human podoplanin glycopeptide, including both the PLAG2 and PLAG3 do- tor-O-glycan ligand interactions. Furthermore, our elucidation mains (Glu38–Gly54) and the disialyl core 1 O-glycan on Thr52, has been of a glycan-independent binding mode as seen in the CLEC-2- synthesized and reported previously (Kato et al., 2008). An additional cysteine residue (Cys55) was introduced at the C terminus of the peptide. For complex rhodocytin interaction may provide a basis for the rational design formation, purified CLEC-2 protein was mixed with glycosylated peptide with of noncarbohydrate-based drugs that inhibit platelet-tumor a 2-fold molar excess of glycopeptide and concentrated up to 57 mg/ml. interaction (Kato et al., 2008) and CLEC-2-dependent HIV-1 For the rhodocytin complex, purified rhodocytin was mixed with equal molar spread in infected individuals (Chaipan et al., 2006). amounts of CLEC-2 protein. The concentration was 45 mg/ml. All crystals were obtained by the sitting drop vapor diffusion method using 0.8 ml drops containing a 50:50 (v/v) mix of protein and reservoir solution at EXPERIMENTAL PROCEDURES 20C. The crystallization conditions were determined by screening using Index (Hampton Research). Crystals of the CLEC-2-podoplanin peptide complex Expression and Purification of the CLEC-2 Lectin Domain were grown in a reservoir containing 0.1 M Tris-HCl (pH 8.5) and 25% (w/v) The extracellular lectin domain of CLEC-2 (C99S mutant) from residues 96–221 polyethylene glycol (PEG) 3,350 (Index #45), and those of the CLEC-2-rhodo- was cloned, expressed, and purified as reported previously, with slight cytin complex were grown in a reservoir containing 0.1 M HEPES (pH 7.6), modifications (Watson and O’Callaghan, 2005). In brief, the CLEC-2 lectin 0.2 M L-proline, and 10% (w/v) PEG 3,350. domain gene incorporated into the pCold-TEV vector was modified to include Before X-ray diffraction experiments, crystals were soaked in reservoir a tobacco etch virus (TEV) protease cleavage sequence between the solution containing 20% (v/v) PEG400 and flash-cooled in liquid nitrogen. hexahistidine tag and CLEC-2. The expression plasmid was introduced X-ray diffraction data sets for the crystals were collected at the synchrotron

1718 Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved Structure Versatile Ligand Recognition Mechanism of CLEC-2

Figure 7. Two-Locus Interaction of CLEC-2 toward Podoplanin and Rhodocytin (A) Superposition of two CLEC-2 molecules in two complex structures. The CLEC-2 podoplanin complex is shown in green and yellow, and the CLEC-2 rhodocytin complex is shown in cyan and blue. CLEC-2 molecules are shown as ribbon models. (B) Structural comparison of ligand binding sites. The Glu-Asp doublet and the second site are shown in the left and right panels, respectively. The carboxyl groups at the second sites (Sia-4 in podoplanin, Asp6 and C-terminal Tyr136 in rhodocytin) are indicated by red arrows in the right panel. The four arginine residues in CLEC-2 are shown as stick and semitransparent sphere models. (C) Schematic of the two-locus interaction in the CLEC-2-ligand complexes.

Surface Plasmon Resonance Assay A surface plasmon resonance assay was per- formed with BIACORE 2000 (GE Healthcare). Rhodocytin was directly captured on a sensor chip (CM5, GE Healthcare) by amine coupling. radiation source at BL-5A and AR-NW12A in the Photon Factory. All data A series of CLEC-2 lectin domain mutants were used as analytes at a flow sets were processed and scaled using the HKL2000 program suite (Otwi- rate of 20 ml/min. After each run, bound CLEC-2 protein was completely nowski and Minor, 1997). Phase determination was performed by the molec- stripped off from the ligand by regeneration of the surface with 3 M magne- ular replacement method with ligand-free CLEC-2 ( [PDB] sium chloride. We calculated the dissociation constants of monomeric

ID code 2C6U; Watson et al., 2007) and rhodocytin (PDB ID code 2VRP; CLEC-2 lectin domain using a 1:1 steady-state binding model: Req =C3

Watson et al., 2008) using MOLREP (Vagin and Teplyakov, 2010). Further Rmax /(KD + C), where C is the analyte CLEC-2 concentrations, Rmax is model building was performed manually using the program COOT (Emsley the maximum binding response, and KD is the equilibrium dissociation and Cowtan, 2004). Refinement was conducted using REFMAC5 (Murshu- constant. dov et al., 1997). Of note, the data of the rhodocytin complex were diag- nosed as twinned by CTRUNCATE of the CCP4 program suite (Collaborative Database Analysis Computational Project, Number 4, 1994). Therefore, refinement was per- A comprehensive sequence motif search was performed against all formed using REFMAC5 with the twin refinement option. The stereochemical deposited entries (544,996 entries) of the UniProtKB/SWISS-PROT data- quality of the final models was assessed by MolProbity (Lovell et al., 2003). base as of April 2014. The potential O-glycosylation site was assigned The MolProbity scores of podoplanin and rhodocytin complexes are 99th based on the annotation (FT column) of each entry. During the course of and 96th percentiles, respectively. Data collection and refinement statistics the analysis, we realized that annotated O-glycosylation sites in UniProt are summarized in Table 1. All figures were prepared with PyMOL (DeLano seem to include glycosaminoglycan (GAG)-linked Ser in addition to Scientific). Ser/Thr-linked mucin-type O-glycans. Because no suitable method is avail- able to select the mucin-type O-glycosylation sites, GAG-linked sites are Solution NMR Analysis of the CLEC-2-Podoplanin Interaction included in Table S3. 1H-NMR data were obtained at a frequency of 500 MHz using a DRX-500 spec- trometer (BrukerBioSpin) equipped with a triple-resonance inverse cryogenic probe. The CLEC-2 lectin domain (0.1 mM) was dissolved in 20 mM Tris-HCl ACCESSION NUMBERS

(pH 8.0), 100 mM NaCl, and 10% D2O/90% H2O, and the nonglycosylated PLAG3 domain or glycosylated PLAG2-3 domain used in the crystallographic The atomic coordinates and structure factors CLEC-2-podoplanin complex study was added to the CLEC-2 solution. The probe temperature was set to (PDB ID code 3WSR) and CLEC-2-rhodocytin complex (PDB ID code 25C, and the water signal was suppressed using a watergate pulse sequence. 3WWK) have been deposited in the Protein Data Bank (http://wwpdb.org/). Data processing was performed by XWIN-NMR (BrukerBiospin), and NMR spectra were displayed using XWIN-PLOT (BrukerBiospin). SUPPLEMENTAL INFORMATION Flow Cytometric Analysis Supplemental Information includes eight figures and three tables and can be The extracellular region of CLEC-2 (DN57) fused to the human Fc fragment was found with this article online at http://dx.doi.org/10.1016/j.str.2014.09.009. expressed using CHO cells. Four CLEC-2-Fc mutants (R107A, R118A, R152A, and R157A) were constructed in addition to wild-type CLEC-2. The expression of each mutant in culture supernatant was confirmed by western blotting using AUTHOR CONTRIBUTIONS anti-human IgG (Figure S7). Flow cytometric analysis was performed using the CLEC-2-Fc chimera proteins and CHO cells expressing human podoplanin. M.N., Y.K., and Y.Y. designed the research; K.M.M. and M.K.K. prepared the We initially tried to detect the binding between the monomeric CLEC-2 lectin samples; M.N. performed the crystallographic analysis; Y.Y. performed the domain and podoplanin in the flow cytometric assay but were unsuccessful. NMR analysis; Y.K. performed the flow cytometry analysis; M.N. performed We therefore used Fc-fused CLEC-2 to enhance the apparent affinity toward the SPR analysis; M.K performed the database analysis; and M.N., Y.K., and podoplanin on CHO cells. Y.Y. wrote the manuscript.

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1719 Structure Versatile Ligand Recognition Mechanism of CLEC-2

ACKNOWLEDGMENTS Herzog, B.H., Fu, J., Wilson, S.J., Hess, P.R., Sen, A., McDaniel, J.M., Pan, Y., Sheng, M., Yago, T., Silasi-Mansat, R., et al. (2013). Podoplanin maintains high We are grateful to the staff of the beamlines at the Photon Factory (Tsukuba, endothelial venule integrity by interacting with platelet CLEC-2. Nature 502, Japan) for providing data collection facilities and support, and Yukishige Ito 105–109. (RIKEN) and Osamu Kanie (Tokai University) for use of the NMR spectrometer. Hooley, E., Papagrigoriou, E., Navdaev, A., Pandey, A.V., Clemetson, J.M., We also thank Noriko Tanaka (RIKEN) for secretarial assistance and Clemetson, K.J., and Emsley, J. (2008). The crystal structure of the Shinya Hanashima (RIKEN/Osaka University), Yuki Fujii, Yukiko Matsunaga, platelet activator aggretin reveals a novel (alphabeta)2 dimeric structure. Yu Kitago, and Junichi Takagi (Osaka University) for helpful discussions and Biochemistry 47, 7831–7837. technical assistance. This work was supported in part by Grant-in-aid for Horii, K., Okuda, D., Morita, T., and Mizuno, H. (2004). Crystal structure of Young Scientists (B) 2477011 (to M.N.), Grant for Scientific Research (C) EMS16 in complex with the integrin a2-I domain. J. Mol. Biol. 341, 519–527. 25460054 (to Y.Y.), Grant 25462242 (to Y.K.), and Grant 26440019 (to M.K.K.) from the Ministry of Education, Culture, Sports, Science, and Technol- Kaneko, M.K., Kato, Y., Kitano, T., and Osawa, M. (2006). Conservation of a ogy (MEXT) of Japan; by the Platform for Drug Discovery, Informatics, and platelet activating domain of Aggrus/podoplanin as a platelet aggregation- Structural Life Science from MEXT of Japan (to Y.K.); and by the Regional inducing factor. Gene 378, 52–57. Innovation Strategy Support Program from MEXT of Japan (to Y.K.). Kaneko, M.K., Kato, Y., Kameyama, A., Ito, H., Kuno, A., Hirabayashi, J., Kubota, T., Amano, K., Chiba, Y., Hasegawa, Y., et al. (2007). Functional glyco- Received: July 29, 2014 sylation of human podoplanin: glycan structure of platelet aggregation- Revised: September 13, 2014 inducing factor. FEBS Lett. 581, 331–336. Accepted: September 17, 2014 Kaneko, M.K., Kunita, A., Abe, S., Tsujimoto, Y., Fukayama, M., Goto, K., Published: November 6, 2014 Sawa, Y., Nishioka, Y., and Kato, Y. (2012). Chimeric anti-podoplanin antibody suppresses tumor metastasis through neutralization and antibody-dependent REFERENCES cellular cytotoxicity. Cancer Sci. 103, 1913–1919. Kato, Y., and Kaneko, M.K. (2014). A cancer-specific monoclonal antibody Acton, S.E., Astarita, J.L., Malhotra, D., Lukacs-Kornek, V., Franz, B., Hess, recognizes the aberrantly glycosylated podoplanin. Sci. Rep. 4, 5924. P.R., Jakus, Z., Kuligowski, M., Fletcher, A.L., Elpek, K.G., et al. (2012). Kato, Y., Fujita, N., Kunita, A., Sato, S., Kaneko, M., Osawa, M., and Tsuruo, T. Podoplanin-rich stromal networks induce dendritic cell motility via activation (2003). Molecular identification of Aggrus/T1a as a platelet aggregation- of the C-type lectin receptor CLEC-2. Immunity 37, 276–289. inducing factor expressed in colorectal tumors. J. Biol. Chem. 278, 51599– Adams, R.H., and Alitalo, K. (2007). Molecular regulation of angiogenesis and 51605. lymphangiogenesis. Nat. Rev. Mol. Cell Biol. 8, 464–478. Kato, Y., Kaneko, M., Sata, M., Fujita, N., Tsuruo, T., and Osawa, M. (2005). Amano, K., Chiba, Y., Kasahara, Y., Kato, Y., Kaneko, M.K., Kuno, A., Ito, H., Enhanced expression of Aggrus (T1alpha/podoplanin), a platelet-aggrega- Kobayashi, K., Hirabayashi, J., Jigami, Y., and Narimatsu, H. (2008). tion-inducing factor in lung squamous cell carcinoma. Tumour Biol. 26, Engineering of mucin-type human glycoproteins in yeast cells. Proc. Natl. 195–200. Acad. Sci. USA 105, 3232–3237. Kato, Y., Kaneko, M.K., Kuno, A., Uchiyama, N., Amano, K., Chiba, Y., Chaipan, C., Soilleux, E.J., Simpson, P., Hofmann, H., Gramberg, T., Marzi, A., Hasegawa, Y., Hirabayashi, J., Narimatsu, H., Mishima, K., and Osawa, M. Geier, M., Stewart, E.A., Eisemann, J., Steinkasserer, A., et al. (2006). DC- (2006). Inhibition of tumor cell-induced platelet aggregation using a novel SIGN and CLEC-2 mediate human immunodeficiency virus type 1 capture anti-podoplanin antibody reacting with its platelet-aggregation-stimulating by platelets. J. Virol. 80, 8951–8960. domain. Biochem. Biophys. Res. Commun. 349, 1301–1307. Chaipan, C., Steffen, I., Tsegaye, T.S., Bertram, S., Glowacka, I., Kato, Y., Kato, Y., Kaneko, M.K., Kunita, A., Ito, H., Kameyama, A., Ogasawara, S., Schmo¨ kel, J., Mu¨ nch, J., Simmons, G., Gerardy-Schahn, R., and Po¨ hlmann, Matsuura, N., Hasegawa, Y., Suzuki-Inoue, K., Inoue, O., et al. (2008). S. (2010). Incorporation of podoplanin into HIV released from HEK-293T cells, Molecular analysis of the pathophysiological binding of the platelet aggrega- but not PBMC, is required for efficient binding to the attachment factor tion-inducing factor podoplanin to the C-type lectin-like receptor CLEC-2. CLEC-2. Retrovirology 7, 47. Cancer Sci. 99, 54–61. Christou, C.M., Pearce, A.C., Watson, A.A., Mistry, A.R., Pollitt, A.Y., Fenton- Kato, Y., Vaidyanathan, G., Kaneko, M.K., Mishima, K., Srivastava, N., May, A.E., Johnson, L.A., Jackson, D.G., Watson, S.P., and O’Callaghan, C.A. Chandramohan, V., Pegram, C., Keir, S.T., Kuan, C.T., Bigner, D.D., and (2008). Renal cells activate the platelet receptor CLEC-2 through podoplanin. Zalutsky, M.R. (2010). Evaluation of anti-podoplanin rat monoclonal antibody Biochem. J. 411, 133–140. NZ-1 for targeting malignant gliomas. Nucl. Med. Biol. 37, 785–794. Kerrigan, A.M., Dennehy, K.M., Moura˜ o-Sa´ , D., Faro-Trindade, I., Willment, Collaborative Computational Project, Number 4 (1994). The CCP4 suite: pro- J.A., Taylor, P.R., Eble, J.A., Reis e Sousa, C., and Brown, G.D. (2009). grams for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50, CLEC-2 is a phagocytic activation receptor expressed on murine peripheral 760–763. blood neutrophils. J. Immunol. 182, 4150–4157. Colonna, M., Samaridis, J., and Angman, L. (2000). Molecular characterization Kuroki, K., Wang, J., Ose, T., Yamaguchi, M., Tabata, S., Maita, N., Nakamura, of two novel C-type lectin-like receptors, one of which is selectively expressed S., Kajikawa, M., Kogure, A., Satoh, T., et al. (2014). Structural basis for simul- in human dendritic cells. Eur. J. Immunol. 30, 697–704. taneous recognition of an O-glycan and its attached peptide of mucin family Drickamer, K. (1999). C-type lectin-like domains. Curr. Opin. Struct. Biol. 9, by immune receptor PILRa. Proc. Natl. Acad. Sci. USA 111, 8877–8882. 585–590. Lovell, S.C., Davis, I.W., Arendall, W.B., 3rd, de Bakker, P.I., Word, J.M., Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular Prisant, M.G., Richardson, J.S., and Richardson, D.C. (2003). Structure valida- graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132. tion by Calpha geometry: f,c and Cbeta deviation. Proteins 50, 437–450. Fu, J., Gerhardt, H., McDaniel, J.M., Xia, B., Liu, X., Ivanciu, L., Ny, A., Maita, N., Nishio, K., Nishimoto, E., Matsui, T., Shikamoto, Y., Morita, T., Hermans, K., Silasi-Mansat, R., McGee, S., et al. (2008). Endothelial cell Sadler, J.E., and Mizuno, H. (2003). Crystal structure of von Willebrand factor O-glycan deficiency causes blood/lymphatic misconnections and consequent A1 domain complexed with snake venom, bitiscetin: insight into glycoprotein fatty liver disease in mice. J. Clin. Invest. 118, 3725–3737. Ibalpha binding mechanism induced by snake venom proteins. J. Biol. Chem. Fujii, Y., Kaneko, M., Neyazaki, M., Nogi, T., Kato, Y., and Takagi, J. (2014). PA 278, 37777–37781. tag: a versatile protein tagging system using a super high affinity antibody Mizuno, H., Fujimoto, Z., Atoda, H., and Morita, T. (2001). Crystal structure of against a dodecapeptide derived from human podoplanin. Protein Expr. an anticoagulant protein in complex with the Gla domain of factor X. Proc. Natl. Purif. 95, 240–247. Acad. Sci. USA 98, 7230–7234.

1720 Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved Structure Versatile Ligand Recognition Mechanism of CLEC-2

Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macro- Tammela, T., and Alitalo, K. (2010). Lymphangiogenesis: Molecular mecha- molecular structures by the maximum-likelihood method. Acta Crystallogr. nisms and future promise. Cell 140, 460–476. D Biol. Crystallogr. 53, 240–255. Vagin, A., and Teplyakov, A. (2010). Molecular replacement with MOLREP. Osada, M., Inoue, O., Ding, G., Shirai, T., Ichise, H., Hirayama, K., Takano, K., Acta Crystallogr. D Biol. Crystallogr. 66, 22–25. Yatomi, Y., Hirashima, M., Fujii, H., et al. (2012). Platelet activation receptor CLEC-2 regulates blood/lymphatic vessel separation by inhibiting prolife- Wang, J., Shiratori, I., Satoh, T., Lanier, L.L., and Arase, H. (2008). An essential ration, migration, and tube formation of lymphatic endothelial cells. J. Biol. role of sialylated O-linked sugar chains in the recognition of mouse CD99 by Chem. 287, 22241–22252. paired Ig-like type 2 receptor (PILR). J. Immunol. 180, 1686–1693. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data Wang, J., Fan, Q., Satoh, T., Arii, J., Lanier, L.L., Spear, P.G., Kawaguchi, Y., collected in oscillation mode. Macromolecular Crystallography Pt. A 276, and Arase, H. (2009). Binding of herpes simplex virus glycoprotein B (gB) to 307–326. paired immunoglobulin-like type 2 receptor a depends on specific sialylated Radaev, S., and Sun, P.D. (2003). Structure and function of natural killer cell O-linked glycans on gB. J. Virol. 83, 13042–13045. surface receptors. Annu. Rev. Biophys. Biomol. Struct. 32, 93–114. Watson, A.A., and O’Callaghan, C.A. (2005). Crystallization and X-ray diffrac- Schacht, V., Ramirez, M.I., Hong, Y.K., Hirakawa, S., Feng, D., Harvey, N., tion analysis of human CLEC-2. Acta Crystallogr. Sect. F Struct. Biol. Cryst. Williams, M., Dvorak, A.M., Dvorak, H.F., Oliver, G., and Detmar, M. (2003). Commun. 61, 1094–1096. T1a/podoplanin deficiency disrupts normal lymphatic vasculature formation and causes lymphedema. EMBO J. 22, 3546–3556. Watson, A.A., Brown, J., Harlos, K., Eble, J.A., Walter, T.S., and O’Callaghan, C.A. (2007). The crystal structure and mutational binding analysis of the Shin, Y., and Morita, T. (1998). Rhodocytin, a functional novel platelet agonist extracellular domain of the platelet-activating receptor CLEC-2. J. Biol. belonging to the heterodimeric C-type lectin family, induces platelet aggrega- Chem. 282, 3165–3172. tion independently of glycoprotein Ib. Biochem. Biophys. Res. Commun. 245, 741–745. Watson, A.A., Eble, J.A., and O’Callaghan, C.A. (2008). Crystal structure of Somers, W.S., Tang, J., Shaw, G.D., and Camphausen, R.T. (2000). Insights rhodocytin, a ligand for the platelet-activating receptor CLEC-2. Protein Sci. into the molecular basis of leukocyte tethering and rolling revealed by struc- 17, 1611–1616. tures of P- and E-selectin bound to SLeX and PSGL-1. Cell 103, 467–479. Watson, A.A., Christou, C.M., James, J.R., Fenton-May, A.E., Moncayo, G.E., Suzuki-Inoue, K., Fuller, G.L., Garcı´a, A., Eble, J.A., Po¨ hlmann, S., Inoue, O., Mistry, A.R., Davis, S.J., Gilbert, R.J., Chakera, A., and O’Callaghan, C.A. Gartner, T.K., Hughan, S.C., Pearce, A.C., Laing, G.D., et al. (2006). A novel (2009). The platelet receptor CLEC-2 is active as a dimer. Biochemistry 48, Syk-dependent mechanism of platelet activation by the C-type lectin receptor 10988–10996. CLEC-2. Blood 107, 542–549. Yokoyama, W.M., and Plougastel, B.F. (2003). Immune functions encoded Suzuki-Inoue, K., Kato, Y., Inoue, O., Kaneko, M.K., Mishima, K., Yatomi, Y., by the natural killer gene complex. Nat. Rev. Immunol. 3, 304–316. Yamazaki, Y., Narimatsu, H., and Ozaki, Y. (2007). Involvement of the snake toxin receptor CLEC-2, in podoplanin-mediated platelet activation, by cancer Zelensky, A.N., and Gready, J.E. (2005). The C-type lectin-like domain super- cells. J. Biol. Chem. 282, 25993–26001. family. FEBS J. 272, 6179–6217.

Structure 22, 1711–1721, December 2, 2014 ª2014 Elsevier Ltd All rights reserved 1721