Supplemental Methods Protein expression and purification The isolated collagen-binding domain of hlair-1, amino acid 22-122, was cloned into pet3xa using introduced BamHI and NotI sites at the 5 and 3 end respectively. Protein was expressed in Escherichia coli Origami (DE3) using LB medium supplemented with 0.5% (v/v) glycerol, 0.05% (w/v) D- glucose, 0.02% (w/v) lactose, 5 mm MgSO 4 and 50 mm K-phosphate buffer ph 6.9 at 20 C. The expressed hlair collagen-binding domain has an N-terminal hexa-histidine tag followed by a TEVprotease cleavage site (ENLYPQGS) and contains three additional alanine residues at the C-terminus. After proteolysis of the tag a glycine and serine residue remain at the N-terminus of the protein. hlair1-cbd was purified by affinity chromatography using Nickel-Sepharose (GE Healthcare, Belgium), followed by anion exchange chromatography on a 1 ml MonoQ column with 25 mm Tris-Cl, ph 8.2 as buffer and a 0-300 mm NaCl gradient to elute the protein. After anion exchange the protein was cleaved using hexa-histidine tagged TEV-protease followed by removal of the protease by washing through a Nickel-Sepharose column. Finally, the hlair1-cbd protein was purified through a Superdex75 size-exclusion column with 25 mm Tris-Cl ph 8.2 as the running buffer. The purified protein was concentrated to 20-30 mg/ml. 15 N-labeled hlair1-cbd was produced by over-expression in Escherichia coli via auto-induction at 20 C in M9-medium containing 1g/L 15 N-ammoniumchloride supplemented with 0.5% (v/v) glycerol, 0.05% (w/v) D-glucose, 0.02% (w/v) lactose, 5 mm MgSO 4 and 50 mm potassium phosphate buffer ph 6.9 enriched with 10% (v/v) Silantes E. coli OD2 15 N-labelled medium (Buchem BV, Netherlands). Purifcation of 15 N-labeled hlair1-cbd was done as for the unlabelled protein except that the buffer was exchanged for 10 mm potassium phosphate, ph 7.1 during the size exclusion step. Docking of hlair1-cbd and peptide III-30 using Haddock2.0 3D models of the hlair1-cbd/collagen III-30 complex were generated using the Haddock webserver (http://haddock.chem.uu.nl/haddock/haddock.php). Starting models for the docking were molecule A of the hlair1-cbd crystals structure and theoretical model of peptide III-30. The central 27 amino acids (collagen III-sequence) of the peptide III-30 model were constructed from the main chain of a collagen III peptide bound to SPARC (PDB code 2V53) to which side chains were added using CNS. 1 This model was flanked on either side with 3 GPP-triplets taken from the crystal structure of (GPP) 10 (PDB code 1K6F). Docking was performed using an ensemble of 20 collagen models that had their side chain conformations randomized by simulated annealing in CNS. Haddock uses a knowledge based approach in which the available experimental data is used in conjunction with structural information to drive the docking. In this procedure all residues that are thought to be involved in the interaction are designated active residues and surface accessible residues that flank interacting/active residues and therefore might be involved in the interaction are designated passive residues. This information is used in the form of distance restraints between all atoms of active residues on one protein and all active and passive residues on the interacting molecule. We defined active residues as all surface accessible residues in LAIR-1 that were indentified as part of the collagen binding surface in the NMR titration experiments plus those residues that were shown to affect collagen binding in the mutagenesis experiments, i.e. R59, E61, R62, E63, R65, T67, Y68, N69, R85 R100, W109, E111, Q112 and Y115. As no experimental data on interacting residues from collagen is available, all surface accessible residues of the collagen model (all residues at the X and X positions of the G-X-X collagen repeat), except for those of the 9 N-terminal or C-terminal residues were designated passive residues. These terminal residues were excluded to prevent the docking being driven towards the termini of the triple-helical peptide that are highly charged. To preserve the collagen tertiary structure we imposed additional restraints on
distances between the terminal regions of the three collagen chains. A HADDOCK run consisted of 3 docking steps; in the first step 5000 complexes were generated by randomization of orientations and rigid-body energy minimization. In the 2 nd step, the 500 best solutions were further refined using semiflexible simulated annealing and finally the top 200 solutions were refined in explicit solvent. Solutions were clustered using a ligand interface rmsd cut off of 7.5 Å and the clusters were sorted based on their HADDOCK score. REFERENCES (1) Brunger AT. Version 1.2 of the Crystallography and NMR system. Nat Protoc. 2007;2:2728-2733. (2) Shiroishi M, Kuroki K, Rasubala L et al. Structural basis for recognition of the nonclassical MHC molecule HLA-G by the leukocyte Ig-like receptor B2 (LILRB2/LIR2/ILT4/CD85d). Proc Natl Acad Sci U S A. 2006;103:16412-16417. (3) Willcox BE, Thomas LM, Bjorkman PJ. Crystal structure of HLA-A2 bound to LIR-1, a host and viral major histocompatibility complex receptor. Nat Immunol. 2003;4:913-919. (4) Boyington JC, Motyka SA, Schuck P, Brooks AG, Sun PD. Crystal structure of an NK cell immunoglobulin-like receptor in complex with its class I MHC ligand. Nature. 2000;405:537-543. (5) Herr AB, Ballister ER, Bjorkman PJ. Insights into IgA-mediated immune responses from the crystal structures of human FcalphaRI and its complex with IgA1-Fc. Nature. 2003;423:614-620. (6) Krissinel E, Henrick K. Inference of macromolecular assemblies from crystalline state. J Mol Biol. 2007;372:774-797.
Table S1. Data collection and refinement statistics hlair1-cbd Data collection Space group P3 2 Cell dimensions a, b, c (Å) 69.4, 69.4, 54.3 α, β, γ ( ) 90, 90, 120 Resolution (Å)* 34.69-1.80 (1.90-1.80) R merge 0.067 (0.208) I / σi 20.2 (5.2) Completeness (%) 96.3 (78.4) Redundancy 5.5 (3.9) Refinement No. reflections 26079 Twin law h, -h-k, -l Twin Fraction 0.294 R work / R free 0.1225 / 0.1575 No. atoms 2843 Protein 2389 Water/Other 443/11 Average B/ Wilson B (Å 2 ) 23.2/17.46 R.m.s. deviations Bond lengths (Å) 0.009 Bond angles ( ) 1.279 Ramachandran Plot (%) Core 93.4 Allowed 6.6 Generously allowed 0.0 Disallowed 0.0
* Values between brackets refer to the highest resolution shell of data Table S2. Root mean square differences between hlair1-cbd and D1 domains of other LRC receptors Receptor PDB code Alignment length Cα rmsd (Å) Sequence identity (residues) D1-domains GPVI 2GI7 85 1.09 0.355 LIR-1 1G0X 90 1.74 0.352 LIR-2 2GW5 83 1.14 0.336 LILRA5 2D3V 91 1.36 0.394 KIR2DL2 1EFX 93 1.34 0.252 FcαRI 1OVZ 83 2.14 0.256
Table S3. Results of docking of hlair1-cbd and peptide III-30 using Haddock 2.0 Cluster a Haddock score (kcal mol -1 ) b N c RMSD-E min (Å) d E VDW (kcal mol -1 ) e E elec (kcal mol -1 ) f E desolv (kcal mol -1 ) g E AIR (kcal mol -1 ) h BSA (Å 2 ) i orientation ( ) j 1-82.5 ± 0.5 39 16.0 ± 1.0-44.1 ± 2.5-323.6 ± 32.0 13.7 ± 9.3 125.7 ± 17.5 1278 ± 59 60 2-81.4 ± 9.2 4 0.6 ± 0.3-46.2 ± 3.7-285.5 ± 16.3 7.0 ± 2.5 149.3 ± 3.4 1414 ± 42 60 3-77.3 ± 6.5 16 12.6 ± 0.1-37.4 ± 3.8-411.6 ± 40.8 34.0 ± 3.0 83.4 ± 13.0 1449 ± 43-30 4-69.4 ± 1.6 11 4.7 ± 0.7-41.9 ± 4.3-206.1 ± 20.9 2.3 ± 4.7 115.1 ± 4.7 1142 ± 24 50 5-69.3 ± 8.2 5 15.4 ± 0.5-41.5 ± 5.9-282.4 ± 62.3 14.3 ± 5.0 143.8 ± 14.3 1294 ± 123-155 6-66.9 ± 7.4 5 18.3 ± 0.1-38.2 ± 1.9-252.3 ± 33.3 8.9 ± 1.5 128.6 ± 6.6 1263 ± 10-145 7-60.6 ± 1.0 14 16.6 ± 0.2-45.1 ± 1.8-182.5 ± 17.0 8.0 ± 4.7 130.0 ± 1.3 1253 ± 63 120 8-59.1 ± 1.3 8 21.3 ± 0.2-43.9 ± 4.2-160.2 ± 41.7 2.9 ± 7.1 138.9 ± 14.2 1333 ± 33-155 9-55.8 ± 5.2 7 7.5 ± 0.2-44.6 ± 2.0-150.7 ± 20.0 9.7 ± 1.3 92.5 ± 3.6 1341 ± 63 30 10-53.7 ± 3.2 14 8.7 ± 0.2-42.9 ± 3.4-149.4 ± 27.1 6.9 ± 1.7 121.9 ± 2.6 1216 ± 86 100 a The final 200 solutions were clustered using an interface ligand RMSD cut off of 7.5 Å. Values represent the average of the four best scoring solutions in each cluster. b The HADDOCK score was calculated as the sum of: E vdw + 0.2 E elec + E desolv + 0.1 E AIR. c Number of structures in a given cluster d Overall backbone RMSD from the overall lowest energy structure e Intermolecular VanderWaals energy f Intermolecular electrostatic energy g Desolvation energy h HADDOCK ambiguous interaction restraint energy i Buried surface area j Estimated angle between the collagen peptide and β-strands F, C and C of LAIR-1
Figure S1. Ligand binding surfaces of LRC-family members Surface drawing of the D1D2 domains of LIR-1, KIR2DL2 and FcαR showing in red surface residues that in the ligand-complex crystal structures were shown to be in contact with the ligand. The D1 domain is in the same orientation as LAIR-1 in Fig. 2B, ribbon drawing. In the leukocyte inhibitory receptors LIR-1 and LIR-2, the interaction site of MHC I-β 2 M is located to the D1D2 hinge region, whereas the HLA-binding site is located in D1 and involves residues from β-strand C, and the 3 10 -helix that replaces strand C in these proteins. 2,3 In the KIR receptors, the interaction with MHC class I molecules is localised to the D1D2 hinge region. 4 The FcαRI receptor interacts with IgA through residues in the BC and FG loops and the tip of strand D at the apex of the D1 domain. 5
Figure S2. Collagen binding properties of the recombinant hlair-1 ectodomain Binding of recombinant hlair-1 to Elisa plates coated with collagen I ( ), collagen III ( ) or BSA ( ). The apparent dissociation constant was determined by fitting the equation OD = OD background + OD max * [LAIR- 1] / (K D,app + [LAIR-1]). This gives half maximal binding at 2.07 ± 0.22 μg/ml and 10.6 ± 1.0 μg/ml protein for collagen I and collagen III, respectively. Since the molecular weight of the tagged LAIR-1 protein is 14.1 kd this corresponds to apparent K D values of 147 ± 16 nm and 750 ± 71 nm, respectively. Curve fitting was done using the program SigmaPlot 11.
Figure S3. An unlikely dimer of LAIR-1 Cartoon drawing of LAIR-1 molecules A and B in the asymmetric unit. Both molecules are rainbowcoloured from N-terminus (blue) to C-terminus (red). The location of the two-fold rotational axis is indicated by the crossed circle. The C-termini of molecules A and B point in opposite directions along the long axis of the molecules. We analyzed the significance of the dimer using the PISA web server 6 (http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html). On the basis of the relatively small size of the dimer interface (439 Å 2 ) and the nature of the interaction surface, PISA suggests that the dimer observed in the crystal is not likely to be of physiological relevance.
Figure S4. Binding of K562 cells transfected with hlair-1a mutants to peptide III-30 (A) Summary flow cytometry analyses showing binding of the triple helical peptide III-30 to parent K562 cells (-) and K562 cells expressing wt LAIR-1a or mutants as indicated. (B) Adhesion of parent K562 cells (-) and K562 cells expressing wt LAIR-1a or mutants as indicated to plate bound peptide III-30. Data represent mean + SD of at least three independent experiments.
Figure S5. Sequence conservation within the LAIR and LRC-families Top: alignment of hlair1 with LAIR1 and LAIR2 from several species. Sequence identities vary from 31.7% with mouse LAIR-1 to 96.1% with chimpanzee LAIR-1. Bottom: alignment with the D1 domains of a selection of human LRC-members. Residues that are 100% conserved are indicated by grey bars. The alignment was produced by 3DCoffee/Espresso which includes 3D structural information (http://www.tcoffee.org/). Arrows indicate key residues in the collagen-binding site.
Figure S6. Sequence conservation within the GPVI family Multiple sequence alignment of the D1D2 domains of GPVI molecules from several species. Residues that are 100% conserved residues are indicated by grey bars. Residues that are also conserved in the LAIR-family are indicated by red bars and key residues R59, E61 and W109 from the LAIR-1 collagen-binding site are indicated by arrows. The alignment was produced by 3DCoffee/Espresso which includes 3D structural information (http://www.tcoffee.org/).