Crystal structure of signal regulatory protein gamma (SIRPγ) in complex with an antibody Fab fragment
- Joanne E Nettleship†1, 2,
- Jingshan Ren†1,
- David J Scott3, 4,
- Nahid Rahman1, 2,
- Deborah Hatherley5,
- Yuguang Zhao1,
- David I Stuart1, 6,
- A Neil Barclay5 and
- Raymond J Owens1, 2Email author
© Nettleship et al.; licensee BioMed Central Ltd. 2013
Received: 6 December 2012
Accepted: 24 June 2013
Published: 4 July 2013
Signal Regulatory Protein γ (SIRPγ) is a member of a closely related family of three cell surface receptors implicated in modulating immune/inflammatory responses. SIRPγ is expressed on T lymphocytes where it appears to be involved in the integrin-independent adhesion of lymphocytes to antigen-presenting cells. Here we describe the first full length structure of the extracellular region of human SIRPγ.
We obtained crystals of SIRPγ by making a complex of the protein with the Fab fragment of the anti-SIRP antibody, OX117, which also binds to SIRPα and SIRPβ. We show that the epitope for FabOX117 is formed at the interface of the first and second domains of SIRPγ and comprises residues which are conserved between all three SIRPs. The FabOX117 binding site is distinct from the region in domain 1 which interacts with CD47, the physiological ligand for both SIRPγ and SIRPα but not SIRPβ. Comparison of the three domain structures of SIRPγ and SIRPα showed that these receptors can adopt different overall conformations due to the flexibility of the linker between the first two domains. SIRPγ in complex with FabOX117 forms a dimer in the crystal. Binding to the Fab fixes the position of domain 1 relative to domains 2/3 exposing a surface which favours formation of a homotypic dimer. However, the interaction appears to be relatively weak since only monomers of SIRPγ were observed in sedimentation velocity analytical ultracentrifugation of the protein alone. Studies of complex formation by equilibrium ultracentrifugation showed that only a 1:1 complex of SIRPγ: FabOX117 was formed with a dissociation constant in the low micromolar range (K d = 1.2 +/− 0.3 μM).
The three-domain extracellular regions of SIRPs are structurally conserved but show conformational flexibility in the disposition of the amino terminal ligand-binding Ig domain relative to the two membrane proximal Ig domains. Binding of a cross-reactive anti-SIRP Fab fragment to SIRPγ stabilises a conformation that favours SIRP dimer formation in the crystal structure, though this interaction does not appear sufficiently stable to be observed in solution.
Members of the signal regulatory protein family (SIRP) play important roles in the regulation of the immune response in man . The family comprises three type I transmembrane glycoproteins (α, β, γ) each with an extracellular region consisting of three Ig-like domains followed by a single transmembrane sequence and a cytoplasmic domain which varies in length between the three SIRPs. SIRPs are classified as “paired receptors” since they show the following characteristics: (1) they are encoded by different genes arranged in a gene cluster, (2) they share significant sequence homology in their extracellular domains and (3) they comprise both activating and inhibitory members. Thus SIRPα delivers an inhibitory signal via immunoreceptor tyrosine-based inhibition motifs (ITIMs) located in the cytoplasmic domain of the protein as well as interacting with the ligand CD47 . SIRPβ delivers an activating signal through association with DAP12, a transmembrane adaptor protein with an immunoreceptor tyrosine-based activation motif (ITAM), but does not bind to CD47 . By contrast, SIRPγ appears to have no signaling function but does bind to CD47, though this interaction is ten times weaker than that of SIRPα . Unlike the other SIRP proteins, SIRPγ is expressed by T cells where it interacts with CD47 on the surface of the cell, resulting in increased cell-cell adhesion in an integrin-independent manner . Therefore, it is thought that SIRPγ may be involved in T cell responses as an accessory protein . Note, SIRPγ was originally called SIRPβ2 but this term is no longer used .
There has been considerable interest in the structure of the SIRP family with regard to the subtle differences in ligand binding specificity [6, 7] and a putative relationship to primitive antigen receptors . Crystal structures have been determined for the N terminal domains of SIRPα [9, 10], SIRPβ and SIRPγ , a full extracellular region of SIRPα  and a complex of the N-terminal domain of SIRPα and CD47 .
In this paper we present the X-ray crystal structure of the complete extracellular portion of SIRPγ co-crystallized with the Fab fragment of the OX117 monoclonal antibody, which recognises SIRPα, SIRPβ and SIRPγ . The Fab fragments of antibodies have been used extensively as co-crystallization chaperones [13, 14] and binding to Fab fragments of OX117 facilitated the crystallization of SIRPγ in this study. The eptiope recognised by OX117 involved residues in both domains 1 and 2 (d1 and d2) of SIRPγ and was distinct from the CD47 binding site on d1. Interestingly, SIRPγ formed a dimer in the crystal structure through an interface between d1 and d2.
Results and discussion
The Fab binding site
From the structure it can be seen that the antibody Fab fragment binds to both d1 and d2 of SIRPγ (Figure 1A). The light chain variable region (Vl) appears to be inserted between the two domains contacting residues in both d1 and d2, whereas the heavy chain variable domain (Vh) only interacts with d1. Comparison of bound and unbound FabOX117 (PDB id 3DIF) shows only small changes in conformation upon binding (Figure 1B). Alignment using PDBeFold  of the Fd region (Vh and CH1 domains) gives a rmsd of 0.6 Å for the Cα backbone for 208 out of 220 residues. Regions of the Fd fragment that were disordered in the unbound Fab structure, namely the Vh loop consisting of residues G133 to S140 and the C-terminal residues 219 to 220 become ordered on binding to the antigen. For the light chain the rmsd for the Cα backbone is 0.7 Å for 213 out of 214 residues. The most significant conformational change in FabOX117 upon binding to SIRPγ is the re-positioning of the side-chain of R103 in the third Complementarity Determining Region (CDR3) of Vh (Figure 1B). R103 forms a network of hydrogen bonds with SIRPγ d1 involving Q8 and E10 in the loop connecting β-strands A1 and A2 and G109 in the G1/G2 loop (Figure 1B). Additional hydrogen bond interactions between FabOX117 and SIRPγ are observed between Y92 in Vl CDR3 and K11 and L12 in the A1/A2 loop of SIRPγ and between S26 of Vl CDR1 and D149 in d2 of SIRPγ. All these residues are conserved between SIRPα, β and γ indicating that OX117 would bind in a similar way to all three SIRPs, consistent with the cross-reactivity data previously published for this antibody .
The structure of the isolated d1 domain of SIRPγ has been solved previously (PDB id 2JJW ). Overlaying this structure onto the full length SIRPγ, showed that binding of the Fab fragment to d1 did not change the conformation of the domain. Thus alignment of the two d1 structures gave a rmsd for the Cα backbone of 0.5 Å for 109/116 amino acids, with the loop comprising G97 to E100 becoming ordered in the complex. Overall, binding of the FabOX117 to SIRPγ appears to result in very little conformational change in either the antibody or antigen.
The crystal structure of d1 of SIRPα in complex with CD47 has been solved by Hatherley et al. . When the d1 of the FabOX117: SIRPγ crystal structure is aligned with the SIRPα(d1): CD47 structure (PDB id 2JJS), the locations where Fab and CD47 bind to SIRPγ do not overlap (Figure 1C). The main contact residues for CD47 are in three loops connecting strands B2-C, D-E and F-G1. This suggests that SIRPγ could bind both FabOX117 and CD47 at the same time.
Comparison of SIRPγ with SIRPα
Analysis of the SIRPγ: FabOX117 complex by analytical ultracentrifugation
M calc c
S calc c
To assess the strength of the SIRPγ: FabOX117 interaction, sedimentation equilibrium centrifugation experiments were carried out on 2:1, 1:1 and 1:2 stoichiometric mixtures of SIRPγ: FabOX117. Data were extracted using SEDFIT v14.1  and processed with SEDPHAT . The results did not fit a single species model indicating that multiple forms were present in solution (Figure 4B). However, the data fitted well to an A + B ↔ AB model. From this model the derived dissociation constant was determined to be 1.2 (± 0.3) μM, indicating only a moderately strong antibody: antigen interaction. Such a value would also explain the behaviour of the sedimentation velocity profiles seen in Figure 4A, since at the loading concentration used, there would be a small amount of dissociation into individual components giving rise to skewed peaks in the c(s) plot (Figure 4A).
We conclude that the 2:2 SIRP: FabOX117 complex observed in the crystal structure is not formed in solution under the conditions of the AUC experiments. It is important to note that the highest protein concentration of the SIRPγ: FabOX117 mixture used in the AUC experiments was 1 mg/ml (corresponding to 13.7 μM) compared to 16.8 mg/ml (230 μM) for crystallization. Therefore it is possible that the 2:2 complex predominates at high concentrations of the SIRPγ FabOX117 mixture. However we calculate that this would require that the dissociation constant of the 2:2 complex is </= 10 μM. If this was the case then it is surprising that no 2:2 complex was observed at the concentration of 13.7 μM used in the sedimentation velocity experiments. The results of the AUC analyses imply that self-association of SIRP γ only occurs, if at all, at very high protein concentrations and therefore is unlikely to be physiologically relevant.
A cross-reactive anti-SIRP monoclonal antibody (OX117) has proved useful in crystallizing human SIRPγ and subsequently obtaining the structure of the antibody: antigen complex. This has enabled the cross-reactive epitope to be mapped to the first and second domains of SIRPγ distal from the CD47 ligand binding site on domain 1. The structure of the FabOX117: SIRPγ complex has also revealed the potential for SIRPs to form head-to-head dimers through an interaction between the domains 1 and 2 on adjacent molecules in the crystal. However this interaction was not observed in solution, so its physiological significance is questionable. Comparison of the three domain structures of SIRPγ and SIRPα showed that these receptors can adopt different overall conformations due to the flexibility of the linker between the first two domains. Given the sequence similarity between all members of the SIRP family it seems likely that this is a property shared by all of the receptors.
SIRPγ and the Fab fragment of the anti-SIRP monoclonal antibody, OX117  (FabOX117), were expressed as recombinant proteins in mammalian cells. FabOX117 was produced by co-transfection of HEK 293 T cells (available from the American Type Culture Collection as CRL-11268) with vectors encoding the heavy and light chain genes using a CompacT SelecT robotic system (The Automation Partnership, Royston, UK) . The Fab fragment was purified from the cell culture media by nickel affinity chromatography followed by size exclusion chromatography . The extracellular region of SIRPγ was expressed using the SIRPα leader sequence residues 1–30 (accession number CAA71403) followed by residues 29–347 of SIRPγ (accession number NP_061026) with a C-terminal tag of STRHHHHHH using the pEE14 vector in CHO Lec126.96.36.199 cells  as previously described . N-glycosylation in these cells is arrested at the high mannose state enabling de-glycosylation by treatment with endoglycosidase H/F1. The SIRPγ was purified from the cell culture media using nickel affinity chromatography followed by size exclusion chromatography . The FabOX117 and SIRPγ proteins were mixed in a 1:1 molar ratio and incubated overnight at 4°C. The FabOX117: SIRPγ complex was treated with Endo F1 to remove the N-linked glycans then purified by size exclusion chromatography.
Crystallization and structure solution
The FabOX117: SIRPγ (16.8 mg/ml) complex was crystallized from 0.2 M magnesium acetate tetrahydrate, 20% w/v polyethylene glycol 3350 (Hampton PEG/Ion Screen #25). Data were collected to 2.5 Å resolution at Diamond Light Source beamline I03 from a single crystal. Diffraction images each of 1.0° oscillation and 0.5 s exposure were recorded on a ADSC Quantum 315 CCD detector at a X-ray wavelength of 0.9763 Å. The crystal was soaked in a cryoprotectant solution containing 20% (v/v) glycerol and 80% (v/v) crystallization reservoir solution for about 10 s before being plunged into liquid nitrogen and maintained at 100 K under a stream of nitrogen gas during data collection. Data were indexed, integrated and scaled using HKL2000/SCALEPACK . The crystal belongs to a space group of P 21212 with unit cell dimensions of a = 140.4 Å, b = 174.2 Å and c = 81.7 Å. The solvent content is 54% by assuming 2 complexes in one crystallographic asymmetric unit.
X-ray data collection and refinement statistics
Unit cell (Å)
a = 140.4, b = 174.2, c = 81.7
Resolution range (Å)
30.0 – 2.50 (2.59-2.50)
Resolution range (Å)
30.0 – 2.50
No. of atoms (protein/other atoms)
Rms bond length deviation (Å)
Rms bond angle deviation (°)
Mean B-factor (protein/other atoms[Å2])
Residues in preferred regions (%)
Residues in allowed regions (%)
Residues in disallowed regions (%)
Analytical ultracentrifuge experiments were performed on a Beckman-Coulter Proteome Lab XL-I running version 5.8 of the data collection software. Data was obtained using both absorbance and interference optics. Sedimentation velocity data was obtained at 40 000 rpm in 2 channel meniscus-matching centrepiece cells (SpinAnalytical, NH, USA), while sedimentation equilibrium data was obtained at 15 000, 20 000, 28 000 and 36 000 rpm in 2 channel centrepieces (Beckman, USA).
Data were scanned every 2 hours until equilibrium had been reached, as determined by calculated RMSDs between successive scans using SEDFIT v14.1 and analysed using SEDPHAT. Loading concentrations were 1.0, 0.2, 0.1 mg/ml for Sirpγ and FabOX117 alone. These corresponded to molar loading concentrations of 20, 4 and 2 μM for Sirpγ and 23, 4.2 and 2.1 μM for FabOX117. For the mixtures, these were performed at a total concentration of 1 mg/ml at molar ratios of 1:2, 1:1 and 2:1, respectively. These corresponded to molar loading concentrations of 6.3/13.7 μM, 10/10 μM and 13.7/6.3 μM. The buffer used throughout was 20 mM Tris–HCl pH 7.5 200 mM NaCl. The solvent density and viscosity were calculated to be 1.0070 and 1.002, respectively using SEDNTERP . The molecular weight and partial specific volumes for FabOX117 was calculated using SEDNTERP to be 48 380 Da and 0.7281 g/ml, respectively. The corresponding values for SIRPγ were calculated on the basis of an additional 1.8 kDa of glycosylation to the protein sequence and found to be 52 800 Da and 0.7321 g/ml, respectively. All data were obtained at 20°C.
Data analysis of analytical ultracentrifuge data
Sedimentation velocity data were analysed using SEDFIT v14.1, and sedimentation coefficients determined from the weight averaged integration of the peaks using the integration functions contained in the software. Corrected sedimentation coefficients (S20, w values) were noted for each loading concentration and used to extrapolate back to infinite dilution to obtain values for for each species. Sedimentation equilibrium data were excised using SEDFIT v14.1 and then exported into SEDPHAT. Data were then fitted to a heterogeneous association A + B ⇔ AB, where A = FabOX117 (light and heavy chain) and B = SIRPγ monomer. Errors in the determined dissociation constant were calculated within SEDPHAT using a Monte Carlo routine, and errors were quoted at the 95% confidence level.
The OPPF-UK is funded by the Medical Research Council and the Biotechnology and Biological Research Council. DIS is supported by the MRC, ANB and DH are funded by the MRC (grant ref; G9826026) and DJS is supported by the Science and Technology Facilities Council. The Wellcome Trust Centre for Human Genetics is supported by the Wellcome Trust (grant no.075491/Z/04). We thank the staff of beamline I03 at Diamond Light Source for help with data collection.
- Barclay AN, Brown MH: The SIRP family of receptors and immune regulation. Nat Rev Immunol 2006, 6(6):457–464. 10.1038/nri1859View ArticlePubMedGoogle Scholar
- Barclay AN: Signal regulatory protein alpha (SIRPalpha)/CD47 interaction and function. Curr Opin Immunol 2009, 21(1):47–52. Epub 2009 Feb 2014 10.1016/j.coi.2009.01.008PubMed CentralView ArticlePubMedGoogle Scholar
- Brooke G, Holbrook JD, Brown MH, Barclay AN: Human lymphocytes interact directly with CD47 through a novel member of the signal regulatory protein (SIRP) family. J Immunol 2004, 173(4):2562–2570.View ArticlePubMedGoogle Scholar
- Piccio L, Vermi W, Boles KS, Fuchs A, Strader CA, Facchetti F, Cella M, Colonna M: Adhesion of human T cells to antigen-presenting cells through SIRPbeta2-CD47 interaction costimulates T-cell proliferation. Blood 2005, 105(6):2421–2427. 10.1182/blood-2004-07-2823View ArticlePubMedGoogle Scholar
- Van den Berg TK, Van Beek EM, Buhring HJ, Colonna M, Hamaguchi M, Howard CJ, Kasuga M, Liu Y, Matozaki T, Neel BG, et al.: A nomenclature for signal regulatory protein family members. J Immunol 2005, 175(12):7788–7789.View ArticlePubMedGoogle Scholar
- Lee WY, Weber DA, Laur O, Severson EA, McCall I, Jen RP, Chin AC, Wu T, Gernert KM, Parkos CA: Novel structural determinants on SIRP alpha that mediate binding to CD47. J Immunol 2007, 179(11):7741–7750.View ArticlePubMedGoogle Scholar
- Liu Y, Tong Q, Zhou Y, Lee HW, Yang JJ, Buhring HJ, Chen YT, Ha B, Chen CX, Yang Y, et al.: Functional elements on SIRPalpha IgV domain mediate cell surface binding to CD47. J Mol Biol 2007, 365(3):680–693. 10.1016/j.jmb.2006.09.079PubMed CentralView ArticlePubMedGoogle Scholar
- Van den Berg TK, Yoder JA, Litman GW: On the origins of adaptive immunity: innate immune receptors join the tale. Trends Immunol 2004, 25(1):11–16. 10.1016/j.it.2003.11.006View ArticlePubMedGoogle Scholar
- Hatherley D, Harlos K, Dunlop DC, Stuart DI, Barclay AN: The structure of the macrophage signal regulatory protein alpha (SIRPalpha) inhibitory receptor reveals a binding face reminiscent of that used by T cell receptors. J Biol Chem 2007, 282(19):14567–14575. 10.1074/jbc.M611511200View ArticlePubMedGoogle Scholar
- Nakaishi A, Hirose M, Yoshimura M, Oneyama C, Saito K, Kuki N, Matsuda M, Honma N, Ohnishi H, Matozaki T, et al.: Structural insight into the specific interaction between murine SHPS-1/SIRP alpha and its ligand CD47. J Mol Biol 2008, 375(3):650–660. 10.1016/j.jmb.2007.10.085View ArticlePubMedGoogle Scholar
- Hatherley D, Graham SC, Turner J, Harlos K, Stuart DI, Barclay AN: Paired receptor specificity explained by structures of signal regulatory proteins alone and complexed with CD47. Mol Cell 2008, 31(2):266–277. 10.1016/j.molcel.2008.05.026View ArticlePubMedGoogle Scholar
- Hatherley D, Graham SC, Harlos K, Stuart DI, Barclay AN: Structure of signal-regulatory protein alpha: a link to antigen receptor evolution. J Biol Chem 2009, 284(39):26613–26619. 10.1074/jbc.M109.017566PubMed CentralView ArticlePubMedGoogle Scholar
- Kovari LC, Momany C, Rossmann MG: The use of antibody fragments for crystallization and structure determinations. Structure 1995, 3(12):1291–1293. 10.1016/S0969-2126(01)00266-0View ArticlePubMedGoogle Scholar
- Rader C: Overview on concepts and applications of Fab antibody fragments. Curr Protoc Protein Sci 2009, 6(6):6–9.Google Scholar
- Nettleship JE, Ren J, Rahman N, Berrow NS, Hatherley D, Barclay AN, Owens RJ: A pipeline for the production of antibody fragments for structural studies using transient expression in HEK 293T cells. Protein Expr Purif 2008, 62(1):83–89. 10.1016/j.pep.2008.06.017View ArticlePubMedGoogle Scholar
- Krissinel E, Henrick K: Secondary-structure matching (SSM), a new tool for fast protein structure alignment in three dimensions. Acta Crystallogr D: Biol Crystallogr 2004, 60(Pt 12 Pt 1):2256–2268.View ArticleGoogle Scholar
- Krissinel E: On the relationship between sequence and structure similarities in proteomics. Bioinformatics 2007, 23(6):717–723. 10.1093/bioinformatics/btm006View ArticlePubMedGoogle Scholar
- Lo Conte L, Chothia C, Janin J: The atomic structure of protein-protein recognition sites. J Mol Biol 1999, 285(5):2177–2198. 10.1006/jmbi.1998.2439View ArticlePubMedGoogle Scholar
- Brookes E, Demeler B, Rosano C, Rocco M: The implementation of SOMO (SOlution MOdeller) in the UltraScan analytical ultracentrifugation data analysis suite: enhanced capabilities allow the reliable hydrodynamic modeling of virtually any kind of biomacromolecule. Eur Biophys J 2010, 39: 423–435. 10.1007/s00249-009-0418-0PubMed CentralView ArticlePubMedGoogle Scholar
- Cann J: Interacting Macromolecules. Volume 12. New York, London: Academic; 1970.Google Scholar
- Nichol LaW DJ: Migration of Interacting Systems. Oxford: Oxford University Press; 1972.Google Scholar
- Schuck P: Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys J 2000, 78: 1606–1619. 10.1016/S0006-3495(00)76713-0PubMed CentralView ArticlePubMedGoogle Scholar
- Vistica J, Dam J, Balbo A, Yikilmaz E, Mariuzza RA, Rouault TA, Schuck P: Sedimentation equilibrium analysis of protein interactions with global implicit mass conservation constraints and systematic noise decomposition. Anal Biochem 2004, 326: 234–256. 10.1016/j.ab.2003.12.014View ArticlePubMedGoogle Scholar
- Zhao Y, Bishop B, Clay JE, Lu W, Jones M, Daenke S, Siebold C, Stuart DI, Jones EY, Aricescu AR: Automation of large scale transient protein expression in mammalian cells. J Struct Biol 2011, 175(2):209–215. 10.1016/j.jsb.2011.04.017PubMed CentralView ArticlePubMedGoogle Scholar
- Stanley P: Glycosylation mutants of animal cells. Annu Rev Genet 1984, 18: 525–552. 10.1146/annurev.ge.18.120184.002521View ArticlePubMedGoogle Scholar
- Nettleship JE, Rahman-Huq N, Owens RJ: The production of glycoproteins by transient expression in mammalian cells. Methods Mol Biol 2009, 498: 245–263. 10.1007/978-1-59745-196-3_16View ArticlePubMedGoogle Scholar
- Otwinowski ZMW: Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol 1997, 276: 307–326.View ArticleGoogle Scholar
- Vagin A A, Teplyakov A: MOLREP: an automated program for molecular replacement. J Appl Cryst 1997, 30: 1022–1025. 10.1107/S0021889897006766View ArticleGoogle Scholar
- Murshudov GN, Vagin AA, Dodson EJ: Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D: Biol Crystallogr 1997, 53(Pt 3):240–255.View ArticleGoogle Scholar
- Emsley P, Cowtan K: Coot: model-building tools for molecular graphics. Acta Crystallogr 2004, 60(Pt 12 Pt 1):2126–2132.Google Scholar
- Stuart DI, Levine M, Muirhead H, Stammers DK: Crystal structure of cat muscle pyruvate kinase at a resolution of 2.6 A. J Mol Biol 1979, 134(1):109–142. 10.1016/0022-2836(79)90416-9View ArticlePubMedGoogle Scholar
- Laue TM, Shah BD, Ridgeway TM, Pelletier SL: Analytical Ultracentrifugation in Biochemistry and Polymer Science. In Royal Society of Chemistry Edited by: Harding S, Rowe A. 1992, 90–125.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.