[ Nature ref: 2006-04-04152A; Ward_SupplMethods2.doc ]
McKern et al. Supplementary Methods
Protein and peptide preparation. The IR-A ectodomain construct was generated as described previously1 and converted to IRDb by oligonucleotide-directed in vitro mutagenesis using the USBT7 GenTM kit. In IRDb, residues 731-735 (VTVA) near the start of the b-chain are replaced with the less hydrophobic sequence AGNN and the nucleotides coding for the 19 residues, 735-753 (VPTVAAFPNTSSTSVPTSP), are removed. IRDb thus lacks the two N-linked glycosylation sites at 730 and 743, all six IR O-glycosylation sites at T732, T737, S745, S746, T747 and T7512 and K917, the last residue of the ectodomain, which was removed in constructing the termination codon. For mammalian cell expression IRDb cDNA was inserted into the pEE14 vector and the resulting plasmid used to transfect Lec8 mutant CHO cells as previously described3. IRDb protein was produced in spinner flasks and purified by affinity chromatography and gel-filtration chromatography as previously described1. Final purification was by ion-exchange chromatography on Bio Q (BioRad, USA) using shallow salt gradients3, which enabled the separation into three fractions corresponding to well-defined peaks.
Fab fragments for Mabs 83-7 and 83-144 were prepared by digesting Mabs 83.7 (isotype IgG1) and 83.14 (isotype IgG2a) with dithiothreitol-activated papain (Sigma 2xcrystallized suspension from Papaya latex ) at 37oC. The digestion was stopped by adding iodoacetamide (IAA), and the reaction mixture passed down a Prosep vA (Millipore) column. Unbound 83-14 Fab and 83.7 F(ab′) 2 were isolated by Superdex 200 (Pharmacia) gel filtration chromatography. The 83-7 F(ab′) 2 was subsequently reduced with mercaptoethylamine and alkylated with IAA, followed by further gel filtration and anion-exchange chromatography on Mono S (Pharmacia, Sweden) for Fab 83-7 or Bio S (BioRad, USA) for Fab 83-14.
The S519N20 peptide was supplied by AusPep (Australia), oxidized by overnight stirring in 1% NH4HCO3, purified by RP-HPLC and characterized by mass spectrometry and amino acid analysis.
Cloning of Mab variable region cDNA. Complementary DNA corresponding to the heavy- and light-chain variable regions of Mabs 83-7 and 83-144 was prepared as described5 and cloned using the Zero Blunt TOPO PCR Cloning Kit (Invitrogen). A minimum of three individual clones for each variable region was sequenced.
Crystallization. A complex of IRDb was prepared from the first fraction of the ion-exchange chromatography with Fab 83-7, Fab 83-14 and S519N20 at a molar ratio of 3:8:8:10 respectively in 10 mM HEPES (pH 7.5), 0.02% sodium azide and 10% D-trehalose. After at least 4 hours at 4°C, the complex was passed over a Superdex 200 column to remove excess reagents, and concentrated to 3 mg/ml in 10mM HEPES (pH 7.5). Initial crystallization conditions were found by vapour diffusion using an in-house screen of 808 conditions. Subsequent refinement of these conditions led to crystals being obtained via vapour diffusion using 1 ml of the above protein solution and 1 ml of 0.24 M ammonium tartrate and 15% PEG 3350 well solution at either 4°C or 20°C. A di-μ-iodobis(ethylenediamine) diplatinum (II) nitrate (PIP) derivatized crystal was generated by soaking a crystal, grown in the above conditions in the well solution plus 2 mM PIP.
X-ray data collection. A native data set and a PIP derivative data set were collected at the IMCA-CAT 17-ID undulator beamline at the Advanced Photon Source (Argonne, USA), each from single crystals which diffracted to 4.5Å and 5.5 Å respectively. A further native data set was collected from a single crystal up to 3.8Å resolution at the BL5A wiggler beamline at the Photon Factory (Tsukuba, Japan). Data processing was conducted with the HKL suite6 and with D*trek7. Details of collection and processing are presented in Supplementary Table S1. The space group was identified as C2221 with unit cell dimensions a=123.0 Å, b=319.7 Å, c=204.9 Å.
Phasing, model building and refinement. The initial structure solution was by molecular replacement using PHASER8 with native data set 1 (Supplementary Table S1). This process succeeded to locate in turn the Fab domains, the L1-CR fragment and the L2 domain. The Fab search model was that of the Hy-HEL5 antibody (PDB entry 1BQL) set at 5° hinge angle increments, whilst that of L1-CR and the L2 domain were derived from the earlier structure of the L1-CR-L2 fragment9. These results suggested that the solvent content of the unit cell was likely high (ca 75%) with the asymmetric unit containing only one IRDb monomer and its attached pair of Fabs. Phase improvement then followed using both RESOLVE10 and BUSTER-TNT11 with the resultant electron density map revealing the likely location of three fibronectin type-3 like domains. At this stage it was clear that the CDRs of one Fab were positioned so as to interact with the CR region, identifying the Fab as 83-7, whilst the other was positioned to interact with one of the fibronectin domains, identifying it as 83-14 and the fibronectin domain as FnIII-1 (see REF 12). Interestingly, the majority of the inter-molecular contacts within the crystal (i.e. between one Fab-complexed IR ectodomain dimer and its neighbour) are Fab-to-Fab. The Fabs thus obviate the need for the heavily-glycosylated IR ectodomain to be involved in crystal contacts and arguably have played a key role in the success of crystallization. The only exception is the involvement of the FnIII-3 domain of the receptor in a crystal contact with the constant domain of the 83-7 Fab attached to a neighbouring receptor. No connected electron density was visible for the ID nor for the constant domain of Fab 83-14. At this stage native data set 2 became available.
The molecular replacement solution was then verified afresh using a combination of single isomorphous replacement phases derived from the PIP-derivative data set (Supplementary Table S1) processed with SHARP13 and subsequent solvent-flattening using DM and SOLOMON. Models of the FnIII domains were then built directly into the density using known FnIII domains 7, 8 and 9 of the fibronectin structure (PDB code 1FNF) to guide the building (see Fig. S7a). Crystallographic refinement followed using iterative cycles of BUSTER-TNT11 and/or REFMAC514 and manual model building using XtalView/Xfit15. Within BUSTER-TNT scattering from the missing atoms was modelled with a low-resolution homographic exponential distribution and maximum entropy density completion was used at the end of each round of refinement to recover the density for missing parts of the structure. During this process, the sequences of Fabs 83-7 and 83-14 became available (Supplementary Figs S5 and S6) and were included in the model building. The 518-529 region of the large CC' loop of FnIII-1 was modelled into somewhat broken and diffuse electron density with the C524-C524 disulphide bond included between the 2-fold related a-chains of the homodimer.
No convincing density appeared for residues 656 to 754, these making up the major part of the ID except for the significant continuous and extended electron density that lay across the ligand-binding face of L1 domain (Supplementary Figs S4 and S8) Whilst we initially interpreted this segment as corresponding to the 20-residue peptide S519N20, subsequent analysis of crystals grown in the presence of alternative binding peptides16 showed that this density remained unchanged (data not shown), suggesting that it may not arise from S519N20 bound at that location. It may represent instead the CT peptide, which is known to be in juxtaposition with L1, since the adjacent insulin B-chain residues F24 and F25 can be chemically cross-linked to the L1 and CT regions respectively17. Electron density consistent with N-linked glycan (see Fig. S7a) was detected at 10 of the 16 potential N-linked sites within the modeled fragment. The final model comprises residues 4 to 655 and 755 to 909 of IRDb, 439 residues from Fab 83-7 and 429 residues from Fab 83-14. No sugar residues are included in the model. Final refinement statistics are shown in Supplementary Table S1.
Figures and Supplementary Figures were generated using combinations of XtalView/Xfit15, MOLSCRIPT18, CONSCRIPT19, ALSCRIPT20 and Raster3D21.
Low-resolution projection images. Low-resolution projection images were calculated from the final atomic coordinates using routines within SPIDER22.
Supplementary Methods References
1. Tulloch, P. A. et al. Single-molecule imaging of human insulin receptor ectodomain and its Fab complexes. J. Struct. Biol. 125, 11-18 (1999).
2. Sparrow, L. G. et al. The location and characterisation of the O-linked glycans of the human insulin receptor.(unpublished data).
3. McKern, N. M. et al. Crystallization of the first three domains of the human insulin-like growth factor-1 receptor. Protein Sci. 6, 2663-2666 (1997).
4. Soos, M. A. et al. Monoclonal antibodies reacting with multiple epitopes on the human insulin receptor. Biochem. J. 235, 199-208 (1986).
5. Gilliland, L.K. et al. Rapid and reliable cloning of antibody variable regions and generation of recombinant single chain antibody fragments. Tissue Antigens 47, 1-20 (1996).
6. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307-326 (1997).
7. Pflugrath, J. W. The finer things in X-ray diffraction data collection. Acta Crystallogr. D Biol. Crystallogr. 55, 1718-1725 (1999).
8. McCoy, A.J., Grosse-Kunstleve, R.W., Storoni, L.C. & Read, R.J. Likelihood-enhanced fast translation functions. Acta Crystallogr. D Biol. Crystallogr. 61, 458-464 (2005).
9. Lou, M. et al. Crystal structure of the first three domains of the human insulin receptor reveals major differences from the IGF-1 receptor in the regions governing ligand specificity. Proc. Natl Acad. Sci. U S A. In press (2006).
10. Terwilliger, T.C. Maximum-likelihood density modification. Acta Crystallogr. D Biol. Crystallogr. 56, 965-972 (2000).
11. Blanc, E. et al. Refinement of severely incomplete structures with maximum likelihood in BUSTER-TNT. Acta Crystallogr. D Biol. Crystallogr. 60, 2210-2221 (2004).
12. Adams, T. E., Epa, V. C., Garrett, T. P. J. & Ward, C. W. Structure and function of the type 1 insulin-like growth factor receptor. Cell. Mol. Life Sci. 57, 1050-1093 (2000).
13. de la Fortelle, E. & Bricogne, G. Maximum-likelihood heavy-atom parameter refinement for multiple isomorphous replacement and multiwavelength anomalous diffraction methods. Methods Enzymol. 276, 472-494 (1997).
14. Murshudov, G.N., Vagin, A.A., and Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240-255 (1997).
15. McRee, D. E. XtalView/Xfit - a versatile program for manipulating atomic coordinates and electron density. J. Struct. Biol. 125, 156–165 (1999).
16. Schaffer, L. et al. Assembly of high-affinity insulin receptor agonists and antagonists from peptide building blocks. Proc. Natl Acad. Sci. USA. 100, 4435-4439 (2003).
17. Xu, B. et al. Diabetes-associated mutations in insulin: consecutive residues in the B chain contact distinct domains of the insulin receptor. Biochemistry 43, 8356-8372 (2004).
18. Kraulis, P. J. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946-950. (1991).
19. Lawrence, M. C. & Bourke, P. CONSCRIPT: a program for generating electron density isosurfaces for presentation in protein crystallography. J. Appl. Crystallogr. 33, 990-991. (2000).
20. Barton, G. J. ALSCRIPT: a tool to format multiple sequence alignments. Protein Eng. 6, 37-40. (1993).
21. Merritt, E. A. & Bacon, D. J. Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505-524 (1997).
22. Frank, J. et al., SPIDER and WEB: Processing and visualization of images in 3D electron microscopy and related fields. J. Struct. Biol. 116, 190-199 (1996).
1