Micromolar Ca2+ Concentrations Are Essential for Mg2+-dependent Binding of Collagen by the Integrin α2β1 in Human Platelets*

Integrin receptor α2β1 requires micromolar Ca2+ to bind to collagen and to the peptide GPC(GPP)5GFOGER(GPP)5GPC (denoted GFOGER-GPP, where O represents hydroxyproline), which contains the minimum recognition sequence for the collagen-binding α2I-domain (Knight, C. G., Morton, L. F., Peachey, A. R., Tuckwell, D. S., Farndale, R. W., and Barnes, M. J. (2000) J. Biol. Chem. 275, 35–40). Platelet adhesion to these ligands is completely dependent on α2β1 in the presence of 2 mmMg2+. However, we show here that this interaction was abolished in the presence of 25 μm EGTA. Adhesion of Glanzmann's thrombasthenic platelets, which lack the fibrinogen receptor αIIbβ3, was also inhibited by micromolar EGTA. Mg2+-dependent adhesion of platelets was restored by the addition of 10 μmCa2+, but millimolar Ca2+ was inhibitory. Binding of isolated α2β1 to GFOGER-GPP was 70% inhibited by 50 μm EGTA but, as with intact platelets, was fully restored by the addition of micromolar Ca2+. 2 mm Ca2+ did not inhibit binding of isolated α2β1 to collagen or to GFOGER-GPP. Binding of recombinant α2 I-domain was not inhibited by EGTA, nor did millimolar Ca2+ inhibit binding. Our data suggest that high affinity Ca2+ binding to α2β1, outside the I-domain, is essential for adhesion to collagen. This is the first demonstration of a Ca2+ requirement in α2β1function.

The platelet integrin ␣ 2 ␤ 1 is a collagen receptor that plays an important role in hemostasis. Injury to the endothelium of a blood vessel results in the exposure of collagen fibers to circulating platelets, resulting in their adhesion and activation, leading to platelet aggregation and clot formation (10,11). Integrin ␣ 2 ␤ 1 is essential for the recognition of collagen by platelets under flow conditions (10,12) and platelets lacking functional integrin ␣ 2 ␤ 1 do not respond to stimulation by collagen (13), resulting in bleeding disorders. The sequence GFOGER, 2 within a triple-helical structure, was recently identified as the minimum binding motif within collagen I for ␣ 1 and ␣ 2 I-domains (14) and has been co-crystallized with the ␣ 2 I-domain, verifying its interaction with the ␣ 2 MIDAS (15). The availability of this peptide allows the properties of ␣ 2 ␤ 1 to be resolved from those of other, non-integrin, platelet receptors for collagen (11,16).
Against this background, we examined the effects of the Ca 2ϩ chelator, EGTA, on adhesion of platelets, of ␣ 2 ␤ 1 , and of recombinant ␣ 2 I-domain to immobilized type I collagen and to the peptide GFOGER-GPP. We demonstrate a requirement for micromolar Ca 2ϩ in ␣ 2 ␤ 1 -mediated platelet adhesion and the binding of isolated ␣ 2 ␤ 1 to these substrates. Both the stimulatory and inhibitory Ca 2ϩ -binding sites appear to lie outside the I-domain.

EXPERIMENTAL PROCEDURES
Materials-Human platelets were from fresh whole blood, provided by the National Blood Service, Long Road, Cambridge, UK. Platelets lacking ␣ IIb ␤ 3 were prepared from whole blood, kindly provided by Dr. M. Makris (Royal Hallamshire Hospital, Sheffield, UK), from two type I Glanzmann's patients. Monomeric type I collagen for use in solid phase adhesion assays was purified from bovine skin, following limited pepsin digestion, as described previously (26,27). The anti-(human integrin ␣ 2 subunit) monoclonal antibody 6F1 (28)  UK). Chemicals were from Sigma-Aldrich (Poole, Dorset, UK) unless otherwise stated. Recombinant ␣ 2 I-domain, as a glutathione S-transferase (GST) fusion protein, was produced (5,29) and used in solid phase binding assays as described previously (5, 29 -31). Peptide GPC(GPP) 5 GFOGER(GPP) 5 GPC (denoted GFOGER-GPP) and collagen-related peptide (CRP; GCO(GPO) 10 GCOG) were synthesized as described previously (14,30,31). The central GFOGER sequence is the minimum recognition sequence for the ␣ 2 I-domain (14) and the flanking GPP sequences stabilize the triple-helical conformation, which is essential for recognition (14,26). Integrin ␣ 2 ␤ 1 was purified from solubilized membranes of human platelets by affinity chromatography on collagen-Sepharose (32,33), biotinylated using an Amersham Pharmacia Biotech ECL biotinylation module, according to the manufacturer's instructions. Purity of the preparation was assessed by separation on SDS-polyacrylamide gel electrophoresis, followed by staining with Gelcode blue stain reagent (Pierce and Warriner, Chester, UK) and image analysis with Leica Q500 (34).
Static Platelet Adhesion Assay-96-well plates (Immulon 2, Dynex Technologies, Ashford, Middlesex, UK) were coated with 100 l per well of monomeric type I collagen or peptides GFOGER-GPP or CRP at 10 g/ml in 0.01 M acetic acid for 1 h at 20°C. Platelet-rich plasma was prepared from fresh whole blood after 2 spins for 1 min at 1200 ϫ g. 10% (v/v) of ACD buffer (39 mM citric acid, 75 mM tri-sodium citrate⅐2H 2 O, 135 mM D-glucose, pH 4.5) and prostaglandin E 1 (100 ng/ml final concentration) were added, and the platelet-rich plasma was spun for 6 min at 700 ϫ g. The platelet pellet was resuspended in 6 ml of buffer (5.5 mM D-glucose, 128 mM NaCl, 4.26 mM Na 2 HPO 4 ⅐2H 2 O, 7.46 mM NaH 2 PO 4 ⅐2H 2 O, 4.77 mM tri-sodium citrate⅐2H 2 O, 2.35 mM citric acid, 0.35% bovine serum albumin (BSA), pH 6.5). Prostaglandin E 1 was added as before, and the platelets were spun for 6 min at 700 ϫ g. Platelets were resuspended to 2 ϫ 10 8 platelets/ml in adhesion buffer (0.05 M Tris-HCl, 0.14 M NaCl, 0.1% BSA, pH 7.4) and treated as appropriate with MgCl 2 , CaCl 2 , or EGTA and allowed to rest for 15 min at room temperature. Ligand-coated wells were blocked by incubation with 200 l of blocking buffer (0.05 M Tris-HCl, 0.14 M NaCl, 5% BSA, pH 7.4) for 30 min. The wells were washed three times with 200 l of adhesion buffer, then 50 l of platelet suspension (10 7 platelets) was added to each well and left for 1 h. The wells were emptied and washed three times with 200 l of adhesion buffer to remove non-adherent platelets. Adherent platelets were lysed by incubation for 1 h with 150 l per well of lysis buffer (0.07 M tri-sodium citrate, 0.3 M citric acid, 0.1% Triton X-100 (v/v), 5 mM p-nitrophenyl phosphate). The reaction was terminated by the addition of 100 l of 2 M NaOH to each well. Adhesion was measured colorimetrically as the absorbance of the pnitrophenol product at 405 nm in a Maxline Emax microplate reader (Molecular Devices Ltd., Crawley, UK). Values were corrected for background by subtraction of readings from BSA-coated wells. In agreement with others (35), the relationship between platelet number and A 405 was linear up to 3.0 ( Fig. 1) in our conditions, and in a typical experiment, adhesion to collagen I resulted in A 405 ϳ1 Ϯ 0.3, which corresponds to adhesion of ϳ20% of the cells applied. For clarity, absorbance values were scaled so that platelet adhesion to collagen I in the presence of 2 mM Mg 2ϩ alone resulted in A 405 ϭ 1, as we have done previously (14).
Integrin ␣ 2 ␤ 1 and ␣ 2 I-domain Binding Assay-The assays for ␣ 2 ␤ 1 (31, 32) and ␣ 2 I-domain (5, 29) adhesion have been described previously. Briefly, 96-well plates (Immulon 2) were coated and blocked as above. After three washes with 200 l of adhesion buffer, 100 l of either biotinylated ␣ 2 ␤ 1 (1 g/ml) or recombinant GST-␣ 2 I-domain fusion protein (5 g/ml) (both in adhesion buffer containing 2 mM MgCl 2 and other ions as required), was applied to the wells and incubated for 2 h at room temperature. Wells were then washed as above and incubated for 30 min with 100 l of either 0.67 g/ml streptavidinhorseradish peroxidase (Pierce and Warriner), for ␣ 2 ␤ 1 detection, or 22 g/ml rabbit anti-GST (Ig-peroxidase conjugate), for ␣ 2 I-domain detection. The wells were washed four times, and bound ligand was detected using a 3,3Ј,5,5Ј-tetramethylbenzidine-peroxidase substrate system (Pierce and Warriner). Absorbance at 450 nm was measured using a Maxline Emax plate reader. Results were corrected for background as above and are scaled to an absorbance reading of 1 for collagen.
Calcium Concentration Calculations-These were performed using the program WinMAXC v2.05 (36) from Dr. Chris Patton at Stanford University.
Replication and Presentation of Data-Data were obtained from Glanzmann's platelets for two identical experiments using blood from different donors. All other experiments were performed on at least three separate occasions. Mean values Ϯ S.D. from single representative experiments are shown throughout, with each condition tested in triplicate. Where error bars are absent, they were too small to reproduce.

RESULTS AND DISCUSSION
Platelet Adhesion via Integrin ␣ 2 ␤ 1 Is Calcium-dependent-We first demonstrate the linearity of the assay used to measure platelet adhesion (Fig. 1) and are in agreement with Bellavite and coworkers (35) that the relationship between absorbance at 405 nm and platelet number is linear, in our conditions up to A 405 ϭ 3.0. Platelet adhesion to monomeric bovine type I collagen and to the peptide GFOGER-GPP requires the presence of Mg 2ϩ (14,24,25), although other divalent cations such as Co 2ϩ and Mn 2ϩ , but not Ca 2ϩ , can replace Mg 2ϩ in ␣ 2 ␤ 1 binding (24) and ␣ 2 I-domain binding. 3 Previous work (5,24,25) suggests that Ca 2ϩ inhibits Mg 2ϩ -dependent adhesion. However, we found that micromolar concentrations of EGTA, a 3 L. F. Morton (this laboratory), unpublished observations.

FIG. 1. Platelet number is directly proportional to A 405 nm.
Platelets were loaded into 96-well plates and lysed for 1 h, and the absorbance of the p-nitrophenol product was read at 405 nm, as described under "Experimental Procedures." Data are from a single experiment, representative of five identical experiments using platelets from different donors and expressed as the mean of triplicate readings Ϯ S.D. Where error bars are absent, they were too small to be reproduced.
FIG. 2. EGTA inhibits platelet adhesion to collagen and to GFOGER-GPP. Platelets were preincubated with EGTA for 20 min in the presence of 2 mM MgCl 2 and allowed to adhere to wells coated with monomeric bovine type I collagen or GFOGER-GPP as described under "Experimental Procedures." Data are from a single experiment, representative of three identical experiments using platelets from different donors and expressed as the mean of triplicate readings Ϯ S.D., scaled to an A 405 ϭ 1 for adhesion to collagen in the presence of 2 mM Mg 2ϩ alone. Where error bars are absent, they were too small to be reproduced. Ca 2ϩ chelator, blocked platelet adhesion to collagen and to GFOGER-GPP in the presence of 2 mM Mg 2ϩ (Fig. 2). Platelet adhesion to collagen-related peptide (CRP), which is mediated by the receptor glycoprotein VI in a cation-independent manner (37), was unaffected by EGTA (data not shown).
When platelets from normal donors were preincubated with an RGD mimetic, GR144053F, at the previously established maximal level for blockade of ␣ IIb ␤ 3 , 4 adhesion to either monomeric collagen I or to GFOGER-GPP was inhibited (Fig. 3). One explanation for the involvement of ␣ IIb ␤ 3 might be that platelet microaggregates form on the initial layer of adherent platelets. Alternatively, the adhesion might involve indirect binding of platelets to substrate in an ␣ IIb ␤ 3 -dependent manner, as has been proposed previously (26,28). We and others (14,26,28) have shown that this component of adhesion is secondary to the initial ␣ 2 ␤ 1 -dependent adhesive process, being entirely inhibited by 6F1 (Fig. 3).
If the observed effects of EGTA were solely due to inhibition of ␣ IIb ␤ 3 , the degree of inhibition of adhesion induced by GR144053F would be the same as for EGTA alone and no additional inhibition of adhesion would occur in the presence of both of these substances. However, the left-hand side of Fig. 3 demonstrates that this is not the case; with normal platelets, 2 M GR144053F resulted in ϳ50% inhibition of adhesion to collagen I and ϳ70% inhibition of adhesion to GFOGER-GPP, whereas 25 M EGTA reduced adhesion to about a quarter of these values. The effect of EGTA was greater than that of GR144053F for adhesion to both collagen and to GFOGER-GPP (p Ͻ 0.001, analysis of variance). Therefore, blockade of normal platelet adhesion by EGTA cannot be attributed solely to inhibition of ␣ IIb ␤ 3 but must include inhibition of binding through ␣ 2 ␤ 1 . We do not understand why GR144053F differentially inhibits platelet adhesion to collagen I and GFOGER-GPP.
Preincubation with a monoclonal antibody specific for the integrin ␣ 2 subunit, 6F1, completely abrogated platelet adhesion to either collagen or to GFOGER-GPP (Fig. 3), thus demonstrating the absolute requirement of this receptor for adhesion to these ligands. In control experiments (Fig. 3), 6F1 did not block adhesion of platelets to CRP (37). These observations suggest that the inhibition of normal platelet adhesion by EGTA in excess of that caused by GR144053F is due to direct action on ␣ 2 ␤ 1 .
To confirm the independence of the effect of EGTA from ␣ IIb ␤ 3 , we examined adhesion using platelets lacking the ␣ IIb ␤ 3 receptor, 5 from two Type I Glanzmann's patients (Fig. 3, righthand side). Adhesion of these platelets to collagen I and to the ␣ 2 ␤ 1 -specific GFOGER-GPP was 80% inhibited by micromolar EGTA (Fig. 3)   validating the specificity of GR144053F for ␣ IIb ␤ 3 . The data in Fig. 3 show that the EGTA-induced inhibition of platelet adhesion occurs in the absence of ␣ IIb ␤ 3 , supporting the concept that the inhibition occurs at the level of ␣ 2 ␤ 1 .
Although EGTA chelates Ca 2ϩ , its inhibitory effect could be due to the removal of other ions such as Zn 2ϩ or Co 2ϩ , possibly present at trace levels in the medium, which might be essential for ligand binding. However, addition of micromolar Ca 2ϩ to EGTA-inhibited platelets restored adhesive function, confirming that Ca 2ϩ is sufficient to support adhesion in the presence of Mg 2ϩ (Fig. 4). However, platelet adhesion to collagen I and to GFOGER-GPP was inhibited in the presence of millimolar Ca 2ϩ , in agreement with the work of others (24,25). Over several different experiments, rescue of platelet adhesion occurred in the estimated free Ca 2ϩ concentration ranges (36): 58 -323 nM (collagen I) and 20 -323 nM (GFOGER-GPP), where 50% maximal adhesion occurred at a free Ca 2ϩ concentration of 88 Ϯ 28 or 110 Ϯ 57 nM for collagen I and GFOGER-GPP, respectively (the latter values are given as mean Ϯ S.E. of five determinations).
It is not surprising that the amount of Ca 2ϩ needed to restore EGTA-inhibited platelet adhesion varied between experiments. The platelets obtained from individual donors are likely to vary both in expression of ␣ 2 ␤ 1 and in sensitivity to activation, so that they may secrete their granule load of Ca 2ϩ to different extents. However, it is important to note that Ca 2ϩ concentrations of around 100 -200 nM can restore EGTA-inhibited, Mg 2ϩdependent platelet adhesion via the ␣ 2 ␤ 1 receptor. This suggests that Ca 2ϩ binds at high affinity site(s) on the platelet surface. The location of these sites may be either on ␣ 2 ␤ 1 itself or on other surface proteins that interact with ␣ 2 ␤ 1 and modify its binding affinity.
Ligand Binding to Isolated ␣ 2 ␤ 1 Shows Partial Ca 2ϩ Dependence-The purity of the ␣ 2 ␤ 1 preparation used in these assays was judged to be ϳ90% by densitometric analysis of the polyacrylamide gel shown in Fig. 5. As with platelets, binding of purified ␣ 2 ␤ 1 to collagen and to GFOGER-containing peptides is completely abolished by 6F1 (14,30) and is also inhibited by increasing concentration of EGTA (Fig. 6A). At 2 mM EGTA, there is very little further inhibition of adhesion (data not shown). Eleven repeat experiments found that in the presence of 50 M EGTA, ␣ 2 ␤ 1 binding to GFOGER-GPP is reduced to 30 Ϯ 3% but adhesion to collagen is only reduced to 66 Ϯ 4% (mean Ϯ S.E.), perhaps because ␣ 2 ␤ 1 binds to collagen sequences other than GFOGER in a Ca 2ϩ -independent manner. It is possible that, when removed from its proper membrane context, the unconstrained integrin displays novel collagen binding activity in regions other than its I-domain. It has also been speculated that integrin ␤ subunits may contain I-domain-like elements (7,38), and it is possible that ␤ 1 adheres to sites other than GFOGER within collagen. However, use of monoclonal antibodies directed against the ␤ 1 subunit did not result in significant blockade of binding (data not shown). Others have found that recombinant constructs, including ␣ 2 sequence up to the end of the first EF-hand as well as the I-domain show enhanced capacity to bind collagen (4), although the mechanism is unclear. These regions of ␣ 2 may either increase the affinity of the MIDAS or bind to collagen at a site distinct from GFOGER. However, it is clear that adhesion of isolated ␣ 2 ␤ 1 to the ␣ 2 I-domain-specific peptide, GFOGER-GPP, is highly sensitive to EGTA, suggesting that Ca 2ϩ has a role in affinity regulation of the I-domain.
Binding of Recombinant ␣ 2 I-domain Is Ca 2ϩ -independent-By marked contrast, binding of recombinant GST-␣ 2 I-domain fusion protein to collagen and to GFOGER-GPP was completely unaffected by EGTA, even up to 2 mM levels (Fig. 6B). In addition, up to 10 mM Ca 2ϩ had no significant effect on adhesion of the I-domain to these ligands. The insen-FIG. 5. ␣ 2 ␤ 1 preparation is at least 90% pure. 9 g of ␣ 2 ␤ 1 preparation was separated by SDS-polyacrylamide gel electrophoresis on a 6% gel.
FIG. 6. EGTA inhibits ␣ 2 ␤ 1 adhesion to collagen and to GFOGER-GPP, but adhesion of ␣ 2 I-domain to these ligands is insensitive to EGTA or Ca 2ϩ . A, biotinylated ␣ 2 ␤ 1 was preincubated with EGTA for 15 min, before measuring adhesion. B, recombinant GST-␣ 2 I-domain fusion protein was preincubated with EGTA or CaCl 2 for 15 min before measuring adhesion. All adhesions were performed in the presence of 2 mM MgCl 2 . Data are from a single experiment, representative of three identical experiments and expressed as the mean of triplicate readings Ϯ S.D., scaled to an A 450 ϭ 1 for adhesion to collagen in the presence of 2 mM Mg 2ϩ alone. Where error bars are absent, they were too small to be reproduced. sitivity to EGTA suggests that the activatory Ca 2ϩ binding site is not found within the I-domain and must lie elsewhere within ␣ 2 ␤ 1 . Also, the well-documented inhibitory effect of Ca 2ϩ on ␣ 2 ␤ 1 -mediated adhesion (24,25) cannot be due to direct competition for the Mg 2ϩ ion bound at the MIDAS site within the I-domain, which is in agreement with others (4).
Ca 2ϩ Restores EGTA-inhibited ␣ 2 ␤ 1 Binding-The EGTA-induced inhibition of isolated ␣ 2 ␤ 1 adhesion can be restored by the addition of Ca 2ϩ (Fig. 7). A 60% reduction in ␣ 2 ␤ 1 binding to GFOGER-GPP was observed in the presence of 50 M EGTA, which could subsequently be restored by addition of Ca 2ϩ . A similar pattern was observed for binding to collagen. Recovery of adhesion corresponds to estimated free Ca 2ϩ concentrations in the range of 2-50 M (36). The rescue of adhesion observed here is similar to that observed with whole platelets, but a striking difference is that 2 mM Ca 2ϩ does not significantly inhibit adhesion of integrin ␣ 2 ␤ 1 to its ligands. In fact, the presence of 10 mM Ca 2ϩ results in only partial reduction of adhesion to either collagen I or to GFOGER-GPP (Fig. 7), suggesting that the inhibitory Ca 2ϩ -binding site is either not present or non-functional in the isolated integrin. Alternatively, millimolar Ca 2ϩ might bind other cell surface proteins, which then regulate ␣ 2 ␤ 1 ligand-binding affinity.
In conclusion, we have demonstrated an essential requirement for micromolar concentrations of Ca 2ϩ in Mg 2ϩ -dependent ␣ 2 ␤ 1 -mediated platelet adhesion. Adhesion of purified ␣ 2 ␤ 1 to GFOGER-GPP was significantly Ca 2ϩ -dependent, whereas, adhesion to collagen had a large Ca 2ϩ -independent component. The stimulatory Ca 2ϩ binding site(s) is not situated within the I-domain and, in agreement with others, Ca 2ϩ -mediated inhibition of ␣ 2 ␤ 1 adhesion does not act by competition with the Mg 2ϩ ion at the MIDAS (4). These findings contribute to the understanding of the role of cations in the regulation of ␣ 2 ␤ 1 function. FIG. 7. The EGTA-induced inhibition of ␣ 2 ␤ 1 adhesion to collagen or to GFOGER-GPP can be restored by addition of Ca 2؉ . Biotinylated ␣ 2 ␤ 1 was preincubated for 15 min with the concentrations of CaCl 2 shown and EGTA in the presence of 2 mM MgCl 2 , before application to the ligand-coated wells. Data are from a single experiment, representative of three identical experiments and expressed as the mean of triplicate readings Ϯ S.D., scaled to an A 450 ϭ 1 for adhesion to collagen in the presence of 2 mM Mg 2ϩ alone. Where error bars are absent, they were too small to be reproduced.