SEARCH

SEARCH BY CITATION

Keywords:

  • collagen;
  • collagen-related peptide;
  • glycoprotein VI;
  • site-directed mutagenesis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Summary. Background: By site-directed mutagenesis of recombinant receptor fragments, we have previously identified residue lysine59 of the platelet collagen receptor glycoprotein VI (GPVI) as being critical for its interaction with the synthetic ligand collagen-related peptide (CRP) and the inhibitory phage antibody 10B12. Lysine59 is proposed to lie on the apical surface of the receptor near the linker joining the two immunoglobulin (Ig)-like extracellular domains. Recently, others have postulated the involvement of a portion of the first domain distant from the interdomain hinge as being involved in an extended collagen-binding site. Aim and Methods: To extend our knowledge of the primary collagen-binding site of GPVI, a number of neighboring residues on the apical surface of recombinant soluble GPVI were mutated to alanine and binding of these mutants, as well as the lysine59 mutant, to fibrillar collagen was measured. Results: Binding of recombinant GPVI to collagen, like CRP, was dramatically reduced by the mutation of residue lysine59 to glutamate. Remarkably, the mutation of residues arginine60 in domain one and arginine166 in domain two, individually to alanine, which had no significant affect on CRP binding, reduced binding of recombinant GPVI to collagen. Mutation of the residue lysine41 to alanine dramatically increased binding to both CRP and collagen. This mutation abolished 10B12 binding, confirming its position in the epitope of our inhibitory phage antibody. Conclusions: Residues lysine59, arginine60, and arginine166, from both Ig-like domains of GPVI, are critical for collagen binding by the receptor. This provides additional evidence for a basic patch on the apical surface of the receptor as the primary collagen-binding site of GPVI.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The activation of platelets following the vessel wall injury is a primary event in normal hemostasis, as well as in atherothrombotic diseases such as stroke and myocardial infarction [1]. When a blood vessel is damaged, components of the sub-endothelial matrix are exposed to the circulation, and interact directly or indirectly with receptors on the platelet membrane to initiate and localize the platelet response. Recognition of exposed sub-endothelial collagen by the receptor GPVI on platelets is a critical early step for platelet activation and subsequent thrombus formation. This interaction strengthens platelet adhesion through the activation of integrins α2β1 and αIIbβ3 [2], and induces aggregation, degranulation and procoagulant activity of the platelets in the area of the damaged blood vessel [3]. Localization of the collagen-binding site of GPVI is useful for developing effective platelet antagonists, as well as for understanding the interaction of GPVI with collagen.

Glycoprotein VI has previously been shown to recognize glycine–proline–hydroxyproline (GPO) repeat motifs in the collagen backbone [4]. A synthetic triple-helical collagen-related peptide, CRP, composed of 10 GPO repeats is a potent and specific GPVI agonist. GPVI has two extracellular C2-type immunoglobulin (Ig)-like domains (D1 and D2). Using the tandem ectodomains (D1D2) of human and mouse GPVI expressed in insect cells, we previously showed that both had affinity for CRP, but human D1D2 bound more strongly than did mouse D1D2 [5]. This was attributed to differences in the amino acid composition of their collagen-binding site. We prepared a three-dimensional model of human D1D2 based on the crystal structures of the killer Ig-like receptors [5], and used this model to identify residues in human D1D2, which differed from the murine counterpart. By site-directed mutagenesis of human D1D2, the residue K59 (E59 in mouse D1D2) was shown to be critical for the interaction of human D1D2 with CRP and the inhibitory single-chain antibody fragment (scFv) 10B12 [5]. Together, these results provided converging evidence that the primary collagen-binding site of GPVI lies in a basic patch on the apical surface containing the residue K59.

The existence of a collagen-binding site on human GPVI distinct from the CRP-binding site has recently been proposed by Lecut et al. [6]. Their study showed that three monoclonal antibodies directed against the extracellular portion of human GPVI displayed selective inhibitory properties on the interaction of GPVI with CRP and collagen selectively. They used phage display to identify short linear peptides, which bound to a monoclonal antibody shown to inhibit the interaction of GPVI with both CRP and collagen. A consensus motif was identified, containing three amino acids, which may lie in the epitope of the inhibitory antibody. By site-directed mutagenesis of recombinant GPVI, they implicated the residue V34 and to a lesser extent L36 as being critical for the interaction of GPVI with collagen and CRP. By homology modeling, these residues were proposed to be part of a contra-apical loop of domain one.

To determine if the basic patch on the apical surface of human GPVI identified in our previous study is involved in the interaction of human D1D2 with fibrillar collagen as well as CRP, four of the mutants used in the initial study, as well as two additional mutants designed as a result of that study, were expressed in insect cells. The effect of these mutations on binding to CRP, collagen, and the inhibitory scFv 10B12 was measured in the ligand-binding assay.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Site-directed mutagenesis and expression of the D1D2 mutants

Human D1D2 (residues Q1-T185 of mature human GPVI) and mutants were expressed in the Drosophila expression system (DESTM; Invitrogen, Paisley, UK) as fusion proteins with a c-terminal calmodulin (CaM) tag, as described previously [5]. CaM-tagged proteins were purified on a W-7 resin affinity column (Jennings et al., University of Cambridge, UK; in prep.). The purity was verified by SDS–PAGE and the concentration measured using a BCA assay kit (Perbio, Chester, UK).

Point mutations were introduced into the human D1D2 fragment in the vector pCR2.1 following the QuickchangeTM protocol (Stratagene, La Jolla, CA, USA). Primers for the K59E, R60A, F120A and S164A mutations have been described previously [5]. Primers for the K41A and R166A mutations were K41Aback 5′-CCTGTACCGCCTGGAGGCGCTGAGTTCCAGCAGGTACC-3′, K41Afor 5′-GGTACCTGCTGGAACTCAGCGCCTCCAGGCGGTACAGG-3′, R166Aback 5′-GCTTCGCTAGCGCGGATCCATACCTGTGGTCGGC-3′ and R166Afor 5′-GCCGACCACAGGTATGGATCCGCGCTAGCGAAGC-3′. The mutated fragments were cloned in expression vectors, as described above.

Synthesis of peptides and conjugates

Collagen-related peptide and GPP10 were synthesized as previously described [7]. Horm collagen (fibrillar type I from equine tendon) was purchased from Nycomed (Munich, Germany). N9A peptide (CAAARWKKAFIAVSAANRFKKIS) [8] was synthesized as described [9]. This peptide binds CaM in the presence of Ca2+ ions, and is conjugated to BSA by a standard method [10] to give BSA-N9A. In this form, it is used to immobilize D1D2 molecules in ELISA. It is conjugated to peroxidase to give HRP-N9A [5], the form used to detect D1D2 binding to its ligands. The anti-GPVI scFvs 10B12 and 1C3 were selected and characterized as described previously from human V gene phage display libraries [5]. The scFvs, which have a hexahistidine tag, were purified using a HiTrap Chelating column (Amersham Biosciences, Little Chalfont, UK) and desalted into 150 mm NaCl, 10 mm HEPES, pH 7.3 (HBS) before use.

Ligand-binding assay

Binding of wild-type D1D2 and mutants to collagen-related peptides and fibrillar collagen was measured in the ligand-binding assay as described previously [5]. Briefly, wells of 96-well MaxisorpTM microplates (Life Technologies, Paisley, UK) were coated with 100 μL of ligand (fibrillar collagen, CRP, GPP10) at 10 μg mL−1 in 0.01 m acetic acid overnight at 4 °C. After washing with assay buffer (TBS/1 mm CaCl2/0.1% BSA), non-specific binding sites on the plastic surface were blocked with TBS/5% BSA for 2 h at 37 °C. After blocking, the wells were washed three times with 200 μL of assay buffer. 100 μL of recombinant protein was then added and incubated at room temperature for 2 h. After four washes with assay buffer, 100 μL of HRP-N9A (0.16 μg mL−1) was added for 1 h. Finally, after six washes, 100 μL of HRP substrate (SureBlue TMB substrate kit; KPL) was added. After 5 min, 50 μL of 0.5 m H2SO4 was added to stop the reaction and the absorbance measured at 450 nm on an ELISA plate reader (Dynex Technologies, Middlesex, UK). Values for non-specific binding to blocked wells (typically <0.06 AU) were subtracted from other readings before analysis. Data were analyzed and graphs were produced in PRISM (GraphPad, San Diego, CA, USA).

Capture ELISA

The effect of the single amino acid mutations in human D1D2 on the binding of anti-GPVI scFvs 10B12 and 1C3 was investigated by ELISA. Fusion proteins were captured via the CaM tag to BSA-N9A in the presence of Ca2+ and binding detected as described previously [5]. In brief, wells of 96-well plates were coated with 50 μL of BSA-N9A at 5 μg mL−1 in 0.05 m borate pH 8.3 overnight at 4 °C. After washing once with 200 μL assay buffer (TBS/1 mm CaCl2/0.1% Tween 20), wells were blocked with 200 μL assay buffer for 2 h at 37 °C. After washing twice with assay buffer, 50 μL of CaM fusion protein at 5 μg mL−1 was added and incubated for 30 min at room temperature. This allows high affinity binding of the CaM-tagged protein to the N9A-coated surface. Wells were washed four times with assay buffer and a titration of scFv added and incubated for 2 h at room temperature. Wells were washed four times with assay buffer and 100 μL of HRP-labeled mouse monoclonal anti-c-myc (clone 9E10; Roche, Lewes, UK) added at a 1:5000 dilution in assay buffer for 30 min at room temperature. After washing six times with assay buffer, 100 μL of TMB was added and the reaction was stopped and read as described earlier.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Residues in the apical surface of recombinant human GPVI are critical for binding to immobilized fibrillar collagen

Binding of wild-type human D1D2 and the mutants to immobilized fibrillar collagen was measured by the ligand-binding assay. As with wild-type, binding of each mutant to collagen was dose-dependent and specific relative to the blocked plate, approaching saturation at 50 μg mL−1 (Fig. 1A). We have previously shown that the K59E mutation reduced binding to CRP, whereas R60A was innocuous [5]. Here, we show that R60A and R166A mutations, as well as K59E, all reduced binding of GPVI to collagen in this study, whereas the mutations F120A and S164A had a relatively minor effect. Binding to the synthetic ligand CRP is unaltered by the R166A mutation (Fig. 1B), whereas K41A increases the binding of GPVI to both CRP and collagen (Fig. 1A,B). In each case, specificity for the hydroxyproline-containing peptide, CRP, over the control peptide, and GPP10 is retained (Fig. 1B).

image

Figure 1. Analysis of binding of recombinant human GPVI mutants to immobilized fibrillar collagen and CRP. Fibrillar collagen (A) or CRP and GPP10 (B) were immobilized on maxisorp plates and incubated with increasing concentrations of wild-type human D1D2 (closed square) and the single amino acid mutants produced specifically for this study, K41A (closed triangle) and R166A (closed circle) (shown in both A and B), or those produced for our previous study [5], S164A (open triangle), K59E (open diamond), R60A (closed diamond), and F120A (open circle) (only shown in A). In B, the binding of human D1D2, K41A and R166A to the control peptide GPP10 is shown using the corresponding open symbol to that used for CRP binding. These were then incubated for 2 h at room temperature. Binding of CaM-tagged D1D2 was detected with N9A-HRP. Non-specific binding to the blocked plate has been subtracted from each data point, and values are the mean ± SD of at least three determinations. On each plate binding of at least one mutant was compared with binding of wild-type D1D2 to collagen, and the above graphs are composites of two to four experiments. Binding of wild-type D1D2 to collagen at 50 μg mL−1 was assigned 100% binding, and values for other points are relative to that value.

Download figure to PowerPoint

Epitope mapping of the scFvs 10B12 and 1C3 confirms that the collagen-binding surface of GPVI is identified by the residues K41 and K59

Two scFvs which bind with similar affinities to separate epitopes on GPVI have been produced in our laboratory. Clone 10B12 directly and dose-dependently inhibits collagen-induced platelet aggregation and abolishes binding of human D1D2 to immobilized collagen [5]. The other clone, 1C3, does not. In our previous study, binding of both scFvs to the mutants R60A, F120A, and S164A was similar to wild-type. The mutant K59E, however, though fully recognized by 1C3, showed reduced binding to 10B12. The capture ELISA was here employed to determine the effect of new mutations on scFv binding.

The binding of 1C3 to K41A and R166A was identical to its binding to wild-type D1D2 (Fig. 2A). 1C3 does not bind to either D1 or D2 alone, implying that its epitope is formed from a combination of the two [5]. Therefore, recognition of the mutants by 1C3 provides additional evidence that the mutants are correctly folded. 10B12 bound to R166A in a manner identical to its binding to the wild-type D1D2, but binding to K41A was reduced (Fig. 2B). Therefore, the 10B12 epitope on GPVI is defined by residues K41 and K59, residues which also affect binding to fibrillar collagen.

image

Figure 2. Binding of the anti-GPVI scFvs 1C3 and 10B12 to the recombinant human GPVI mutants. Wild-type human D1D2 (square), K41A (inverted triangle), and R166A (circle) were captured via the CaM tag to a maxisorp plate coated with BSA-N9A. This ensured equal loading of antigen in the same orientation. Increasing concentrations of 1C3 (A) or 10B12 (B) were then added to each well and incubated for 1 h at room temperature. Bound scFvs were detected via their Myc-tag with 9E10-HRP. Values are the mean ± SD of three determinations. Binding of wild-type D1D2 to 1C3 and 10B12 at 10 and 20 μg mL−1, respectively, was assigned 100% binding, and values for other points are relative to that value. These graphs are representative of two experiments.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Elucidation of the mechanism of interaction of the receptor GPVI with the platelet agonist fibrillar collagen is an essential step toward understanding the complex process of platelet activation in the initial stages of hemostasis and thrombosis. The structure of human GPVI, alone or in complex with a ligand, is unknown. To date, two studies have utilized site-directed mutagenesis of recombinant forms of the extracellular portion of the receptor to identify residues, which may play a role in the GPVI–collagen interaction. The first study, carried out in this laboratory, identified residue K59, in the apical surface of GPVI, to be critical for the interaction of GPVI with the peptide CRP and the scFv 10B12 [5]. 10B12, which contains a relatively acidic antigen-binding site, has been shown to inhibit the interaction of GPVI with both CRP and collagen, allowing us to propose that a basic patch on the apical surface of GPVI contains the primary collagen-binding site.

A more recent study from Lecut et al. [6] implicated the residues V34 and L36 in the interaction of GPVI with CRP and collagen. This group employed peptide phage display to identify residues binding to monoclonal antibody 9O12 that block the GPVI–collagen interaction. The consensus peptide sequence of the 9O12 binders matched a segment of the GPVI sequence found in a loop of domain one, distal to the apical surface and containing residues V34 and L36. When mutated to alanine, these residues affected GPVI binding to collagen, CRP and 9O12, and the authors concluded that this region of the receptor is critical for the direct interaction of GPVI with these ligands.

To examine if the apical surface of human GPVI is involved in the GPVI–collagen interaction, and to investigate the extent of this interaction by identifying additional residues critical for binding to collagen, a number of single amino acid mutants of recombinant human D1D2 were created using site-directed mutagenesis, based on the results of our previous study [5]. All mutants were correctly folded as judged by the specific binding to CRP (Fig. 1B) and recognition by the anti-GPVI scFvs 10B12 and 1C3 (Fig. 2). The effect of each of these mutations on binding of human D1D2 to fibrillar collagen was measured using the ligand-binding assay (Fig. 1A).

Mutation of the residues K41, R60, and R166 to alanine, and K59 to glutamate, all affected binding of human D1D2 to collagen (Fig. 1A). Notably, the mutant K41A demonstrated increased binding of human D1D2 to collagen, while the others showed reduced binding. K41A also showed increased binding to CRP, though binding of R60A and R166A was similar to wild-type (Fig. 1B) [5]. This suggests that though the CRP- and collagen-binding sites of human GPVI may overlap, the binding sites are chemically different.

We have thus demonstrated that a number of basic charged residues are involved in GPVI–collagen interaction, all of which form a cluster on the apical surface (Fig. 3), arising from two loops in domain one (C-C’ for K41, E-F for K59 and R60) and one in domain two (F-G for R166). Binding of the inhibitory scFv 10B12 to human D1D2 was inhibited by the mutations K59E [5] and K41A (Fig. 2B), providing additional evidence that the epitope of this scFv contains the apical loops of D1.

image

Figure 3. Model of recombinant human D1D2 showing residues implicated in binding to collagen. A ribbon diagram displaying the side chains of residues identified in this study involved in the interaction of GPVI with collagen (K41, K59, R60, R117, R166 on the apical surface) and in the Lecut study (V34 and L36 in domain one) [6]. D1 is on the right and D2 on the left.

Download figure to PowerPoint

Prolyl hydroxylation is essential for binding of GPVI to collagen, as shown in studies of platelet binding to recombinant unhydroxylated collagen from transgenic plants [11]. The repeating GPO triplets present in CRP represent the simplest motif containing that structure, and CRP also requires the presence of hydroxyproline to be bound by GPVI [7]. For some reason, the replacement of lysine41 by alanine permits a stronger interaction with these essential hydroxyproline-containing motifs. The cavity generated would presumably reduce constraints on the positioning of the remaining side chains in the basic cluster, but may also positively contribute to a hydrophobic surface. It is interesting then that the loss of arginine side chains at positions 60 and 166 do not affect CRP binding, yet reduce collagen binding. This implies that the latter residues are unnecessary for interaction with hydroxyproline, and may support the interaction with collagen by interacting with another part of the collagen structure, or conceivably, another constituent of the collagen preparation.

Therefore, in contrast to the conclusions of Lecut et al., it is clear from the results presented here that the collagen-binding surface of human GPVI contains at least four basic residues from both Ig-like domains, and is located on the apical surface joining those domains. The same surface is used for the interaction of GPVI with CRP and collagen, though the nature of the interaction is chemically different for each ligand. A crystal structure of the extracellular domains of GPVI in contact with a collagen fragment or synthetic collagen-like peptide is necessary to further understand the binding mechanism of this clinically relevant interaction.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

P. A. S. was supported by the British Heart Foundation, R. W. F. by the Medical Research Council, W. H. O. by the National Blood Service R&D.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Lusis AJ. Atherosclerosis. Nature 2000; 407: 23341.
  • 2
    Kuijpers MJ, Schulte V, Bergmeier W, Lindhout T, Brakebusch C, Offermanns S, Fassler R, Heemskerk JW, Nieswandt B. Complementary roles of glycoprotein VI and alpha2beta1 integrin in collagen-induced thrombus formation in flowing whole blood ex vivo. Faseb J 2003; 17: 6857.
  • 3
    Heemskerk JW, Siljander P, Vuist WM, Breikers G, Reutelingsperger CP, Barnes MJ, Knight CG, Lassila R, Farndale RW. Function of glycoprotein VI and integrin alpha2beta1 in the procoagulant response of single, collagen-adherent platelets. Thromb Haemost 1999; 81: 78292.
  • 4
    Kehrel B, Wierwille S, Clemetson KJ, Anders O, Steiner M, Knight CG, Farndale RW, Okuma M, Barnes MJ. Glycoprotein VI is a major collagen receptor for platelet activation: it recognizes the platelet-activating quaternary structure of collagen, whereas CD36, glycoprotein IIb/IIIa, and von Willebrand factor do not. Blood 1998; 91: 4919.
  • 5
    Smethurst PA, Joutsi-Korhonen L, O'Connor MN, Wilson E, Jennings NS, Garner SF, Zhang Y, Knight CG, Dafforn TR, Buckle A, MJ IJ, De Groot PG, Watkins NA, Farndale RW, Ouwehand WH. Identification of the primary collagen-binding surface on human glycoprotein VI by site-directed mutagenesis and by a blocking phage antibody. Blood 2004; 103: 90311.
  • 6
    Lecut C, Arocas V, Ulrichts H, Elbaz A, Villeval JL, Lacapere JJ, Deckmyn H, Jandrot-Perrus M. Identification of residues within human GPVI involved in the binding to collagen: evidence for the existence of distinct binding sites. J Biol Chem 2004; 279: 522939.
  • 7
    Knight CG, Morton LF, Onley DJ, Peachey AR, Ichinohe T, Okuma M, Farndale RW, Barnes MJ. Collagen-platelet interaction: Gly-Pro-Hyp is uniquely specific for platelet Gp VI and mediates platelet activation by collagen. Cardiovasc Res 1999; 41: 4507.
  • 8
    Montigiani S, Neri G, Neri P, Neri D. Alanine substitutions in calmodulin-binding peptides result in unexpected affinity enhancement. J Mol Biol 1996; 258: 613.
  • 9
    Li CQ, Garner SF, Davies J, Smethurst PA, Wardell MR, Ouwehand WH. Threonine-145/methionine-145 variants of baculovirus produced recombinant ligand binding domain of GPIbalpha express HPA-2 epitopes and show equal binding of von Willebrand factor. Blood 2000; 95: 20511.
  • 10
    Bernatowicz MS, Matsueda GR. Preparation of peptide-protein immunogens using N-succinimidyl bromoacetate as a heterobifunctional crosslinking reagent. Anal Biochem 1986; 155: 95102.
  • 11
    Perret S, Eble JA, Siljander PR, Merle C, Farndale RW, Theisen M, Ruggiero F. Prolyl hydroxylation of collagen type I is required for efficient binding to integrin alpha 1 beta 1 and platelet glycoprotein VI but not to alpha 2 beta 1. J Biol Chem 2003; 278: 298739.