Agonist-activated (cid:97) v (cid:98) 3 on Platelets and Lymphocytes Binds to the Matrix Protein Osteopontin*

The phosphorylated acidic glycoprotein osteopontin is present in the extracellular matrix of atherosclerotic plaques and the wall of injured but not normal arteries. To determine if osteopontin could serve as a substrate for platelet adhesion, we measured the adherence of resting and agonist-stimulated human platelets to immobilized recombinant human osteopontin. Agonist-stimulated but not resting platelets bound to osteopontin by a process that was mediated primarily by (cid:97) v (cid:98) 3. (cid:97) v (cid:98) 3-mediated adherence occurred at physiologic concentrations of calcium and was inhibited by an (cid:97) v (cid:98) 3-selective cyclic peptide. Assays using phorbol myristate acetate-stimulated transfected B lymphocytes expressing both (cid:97) v (cid:98) 3 and (cid:97) IIb (cid:98) 3 confirmed that activated (cid:97) v (cid:98) 3 not activated (cid:97) IIb (cid:98) 3 was responsible for the cellular adherence we measured. These studies indicate that (cid:97) v (cid:98) 3 can reside on the cell surface in an inactive state and can be converted to a ligand binding conformation by cellular agonists. Moreover, they suggest that platelet adherence to osteopontin mediated by activated (cid:97) v (cid:98)


The phosphorylated acidic glycoprotein osteopontin is present in the extracellular matrix of atherosclerotic plaques and the wall of injured but not normal arteries.
To determine if osteopontin could serve as a substrate for platelet adhesion, we measured the adherence of resting and agonist-stimulated human platelets to immobilized recombinant human osteopontin. Agoniststimulated but not resting platelets bound to osteopontin by a process that was mediated primarily by ␣v␤3. ␣v␤3-mediated adherence occurred at physiologic concentrations of calcium and was inhibited by an ␣v␤3selective cyclic peptide. Assays using phorbol myristate acetate-stimulated transfected B lymphocytes expressing both ␣v␤3 and ␣IIb␤3 confirmed that activated ␣v␤3 not activated ␣IIb␤3 was responsible for the cellular adherence we measured. These studies indicate that ␣v␤3 can reside on the cell surface in an inactive state and can be converted to a ligand binding conformation by cellular agonists. Moreover, they suggest that platelet adherence to osteopontin mediated by activated ␣v␤3 could play a role in anchoring platelets to disrupted atherosclerotic plaques and the walls of injured arteries. By inhibiting ␣v␤3 function, it may be possible to inhibit platelet-mediated vascular occlusion with a minimal effect on primary hemostasis.
The final event in a variety of cardiovascular diseases is often arterial occlusion by a platelet-rich thrombus. Members of the integrin superfamily of adhesion molecules play a key role in this pathologic process by anchoring platelets to the exposed subendothelium of damaged arteries and by mediating platelet aggregation. The integrin, ␣IIb␤3, mediates platelet aggregation when platelet stimulation converts it from a resting to a ligand-binding conformation (1). Platelets contain a second ␤3 integrin, ␣v␤3, but whether ␣v␤3 plays a role in platelet function is not known. However, a monoclonal antibody that binds to both ␣IIb␤3 and ␣v␤3 has been shown in clinical trials to have additional efficacy relative to compounds that bind only to ␣IIb␤3 by preventing the reocclusion that often occurs months after PTCA 1 (2)(3)(4).
Formation of an occlusive thrombus or a normal hemostatic platelet plug is initiated when platelets adhere to newly exposed components of the subendothelial extracellular matrix of diseased or damaged blood vessels. The matrix components assumed to function as substrates for platelet adherence include collagen, fibronectin, and von Willebrand's factor because platelets contain receptors for each of these proteins and adhere to these proteins in vitro (2). Nevertheless, the substrates that actually mediate platelet adherence to disrupted atherosclerotic plaques are not known. Osteopontin is an acidic phosphorylated glycoprotein secreted by a number of cells including osteocytes, osteoclasts, macrophages, and smooth muscle cells (5)(6)(7). Although not present in the walls of normal arteries, osteopontin is widely distributed throughout the matrix of calcified plaques in arteries involved by atherosclerosis (8 -10). Studies in vitro suggest that osteopontin may be involved in the formation of the neointima characteristic of the atherosclerotic process by serving as a substrate for ␣v-integrin-mediated smooth muscle and endothelial cell migration (8,11). Because it is likely that osteopontin is exposed to circulating blood by the plaque disruption that precedes acute coronary artery occlusion and results from PTCA, we examined the possibility that osteopontin could serve as an adhesive substrate for platelets. We found that activated but not resting platelets adhere to osteopontin and that their adherence is mediated by an activated conformation of ␣v␤3.

Synthesis of Recombinant Human
Osteopontin-Recombinant human osteopontin was synthesized as a histidine-tagged fusion protein using the pET system (Novagen). A cDNA for human osteopontin was inserted into the plasmid pET16b, and recombinant protein was synthesized as insoluble inclusion bodies in Escherichia coli BL21(DES)pLysS. Following lysis of the pelleted bacteria in 20 mM Tris-HCl buffer, pH 7.9, containing 0.5 M NaCl, 1 mg/ml lysozyme, and 0.1% Triton X-100, osteopontin was solubilized using 6 M guanidine HCl and isolated by metal chelate affinity chromatography on a Ni 2ϩ nitrilotriacetic acid resin (His⅐Bind Resin, Novagen). Recombinant osteopontin was eluted from the resin using 20 mM Tris-HCl buffer, pH 7.9, containing 0.5 M NaCl and 500 mM imidazole and renatured by dialysis against phosphate-buffered saline, pH 7.4. 0.1% SDS-7.5% polyacrylamide gel electrophoresis of the renatured protein revealed a single band with an apparent molecular weight of 58,000. The mass of the recombinant protein as determined by electrospray mass spectroscopy was 35,518, consistent with the calculated mass of the full-length osteopontin amino acid backbone (12) plus the polyhistidine tag and Factor Xa cleavage site contributed by pET16b.
Measurement of Platelet Adherence to Osteopontin and Fibrinogen-96-well flat bottom microtiter plates (Immulon 2, Dynatech) were coated with 5 g/ml recombinant osteopontin, purified human fibrinogen, or bovine serum albumin, each dissolved in 50 mM NaHCO 3 buffer, pH 8.0, containing 150 mM NaCl. Unoccupied protein binding sites on the wells were blocked with 5 mg/ml bovine serum albumin dissolved in the same buffer. Human platelets were isolated from blood anticoagulated with 0.1 volume of 3.8% sodium citrate by gel filtration using a 4 * This work was supported by National Institutes of Health Grants HL40387 (to J. S. B.) and HL51258 (to J. S. B.) and National Science Foundation Grant CHE-9634646 (to W. F. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed: Hematology-Oncology Division, Stellar-Chance Laboratories, Rm. 1005, 422 Curie Blvd., Philadelphia, PA 19104. Tel.: 215-662-4028; Fax: 215-662-7617; E-mail: bennetts@mail.med.upenn.edu. mM HEPES buffer, pH 7.4, containing 135 mM NaCl, 2.7 mM KCl, 5.6 mM glucose, 3.3 mM NaH 2 PO 4 , 0.35 mg/ml bovine serum albumin, and various concentrations of CaCl 2 or MgCl 2 according to the experiment (13). 100-l aliquots of the gel filtered platelet suspension containing ϳ2-5 ϫ 10 6 platelets were added to the protein-coated wells in the absence or the presence of a platelet agonist. Following an incubation for 30 min at 37°C without agitation, the plates were washed four times with the gel filtration buffer, and the number of adherent platelets was measured using the colorimetric assay reported by Bellavite et al. (14). Briefly, 150 l of a 0.1 M citrate buffer, pH 5.4, containing 5 mM p-nitrophenyl phosphate (Sigma) and 0.1% Triton X-100 was added to the wells after washing. After an incubation for 60 min at room temperature in the dark, color was developed by the addition of 100 l of 2 N NaOH and read in a microtiter plate reader at 405 nm.
Measurement of Lymphocyte Adherence to Osteopontin and Fibrinogen-pREP vectors containing cDNAs for ␣IIb and ␤3 were introduced into 7.5 ϫ 10 6 GM1500 B lymphocytes by electroporation, and stable cotransfectants were selected by growth in media containing G418 and hygromycin as described previously (15). The phorbol myristate acetate (PMA)-stimulated adherence of transfected and untransfected lymphocytes to osteopontin and fibrinogen was measured as described previously (15). Briefly, 1.5 ϫ 10 5 B lymphocytes, metabolically labeled overnight with [ 35 S]methionine, were suspended in 100 l of 50 mM Tris-HCl buffer, pH 7.4, containing 150 mM NaCl, 0.5 mM CaCl 2 , 0.1% glucose, and 1% bovine serum albumin and added to the wells of microtiter plates coated with osteopontin or fibrinogen, either in the presence or the absence of 200 ng/ml PMA. Following an incubation at 37°C for 30 min without agitation, the plates were washed four times with the lymphocyte suspension buffer, and adherent cells were dissolved using 2% SDS. The SDS solutions were then counted for 35 S in a liquid scintillation counter.

RESULTS AND DISCUSSION
Platelet Adherence to Osteopontin and Fibrinogen-To determine if osteopontin could serve as a substrate for platelet adherence, we used a solid phase assay to measure the adherence of gel filtered human platelets to either purified recombinant human osteopontin or purified human fibrinogen (13), a known adhesive ligand for platelets (16). In the presence of Mg 2ϩ , there was substantial adherence of unstimulated platelets to fibrinogen but little adherence to osteopontin (Fig. 1A). Platelet stimulation with 10 M ADP resulted in a dramatic increase in the number of platelets adherent to osteopontin, as well as a smaller increase in the number of platelets adherent to fibrinogen. Inspection of the assay plates by light microscopy confirmed these results and revealed that ADP stimulation resulted in both platelet adherence and spreading on the fibrinogen and osteopontin-coated surfaces (Fig. 1B). Similar results were observed when the platelets were stimulated by 20 M epinephrine, 0.1 unit/ml thrombin, or 200 ng/ml PMA, and the presence of 25 M indomethacin had no effect on the adherence of ADP-stimulated platelets. The addition of EDTA prevented platelet adherence to either substrate. To verify that activated platelets adhere to native osteopontin, as well as to recombinant protein, assays were repeated using purified osteopontin isolated from human urine (uropontin). We found no difference in the ability of uropontin and recombinant osteopontin to support platelet adherence (data not shown). Thus, stimulated but not unstimulated human platelets are able to use immobilized osteopontin as an adhesive substrate.
Identification of the Receptor Responsible for Platelet Adherence to Osteopontin-To identify the receptor on activated platelets that mediates platelet adherence to osteopontin, we repeated the adherence assays in the presence of monoclonal antibodies against ␣IIb␤3 and against ␣v␤3, the only ␣v-containing integrin present in platelets (Fig. 2). Although agonistmediated platelet adherence is generally mediated by ␣IIb␤3 (1), platelet adherence to osteopontin was consistently inhibited by 84 -93% by the ␣v␤3-specific mAb LM609 (19) and 96 -99% by the ␤3-integrin-specific mAb 7E3 (20). In contrast, saturating concentrations of the ␣IIb␤3-selective mAbs A2A9 (17) and 10E5 (18) inhibited platelet adherence to osteopontin by only 30 -40%, perhaps because these antibodies also crossreact with ␣v␤3 to some extent (15), whereas an antibody specific for ␣5␤1 (mAb 16, Becton Dickinson) was not inhibitory FIG. 1. Adherence of gel filtered human platelets to immobilized osteopontin and fibrinogen. Platelet adherence to the wells of microtiter platelets coated with either recombinant human osteopontin or purified human fibrinogen was measured as described under "Experimental Procedures." A, quantitation of platelet adherence using a colorimetric based on measurement of platelet acid phosphatase activity (14). The data shown are the mean and standard error of measurements made in triplicate and are representative of 16 separate experiments. B, visualization of adherent platelets adherent by light microscopy (200ϫ).

FIG. 2. Identification of the receptor mediating platelet adherence to osteopontin.
Adherence of ADP-stimulated platelets to osteopontin was measured as described in the legend to Fig. 1. The inhibitory effect of saturating concentrations of the following monoclonal antibodies was then compared with that of 5 mM RGDS: A2A9 (50 g/ml), 10E5 (20 g/ml), 7E3 (20 g/ml), and LM609 (30 g/ml). The data shown are the means and standard errors of triplicate determinations and were normalized to 100% binding in the absence of inhibitors.
(data not shown). Inhibition by the tetrapeptide RGDS was nearly complete, consistent with platelet adherence to osteopontin being an integrin-mediated process. Conversely, there was nearly complete inhibition of platelet adherence to fibrinogen by A2A9, 10E5, and 7E3 and only minimal inhibition by LM609 (data not shown). Thus, these data indicate that whereas platelet adherence to fibrinogen is mediated by ␣IIb␤3, the receptor primarily mediating platelet adherence to osteopontin is ␣v␤3. The data also indicate that the ability of platelet ␣v␤3 to recognize osteopontin requires platelet stimulation.
There are 50-fold (21) to 500-fold (22) fewer copies of ␣v␤3 compared with ␣IIb␤3 on the platelet surface. Moreover, ␣v␤3 is generally considered to reside on the surface of most cells in a constitutively active state (23). To verify that agonist-stimulated platelet adherence to osteopontin is mediated by ␣v␤3 rather than ␣IIb␤3, we used a B lymphocyte model of platelet integrin function. B lymphocytes constitutively express ␣v␤3 but express ␣IIb␤3 after transfection (15,24). Following exposure to PMA, only the transfected cells expressing ␣IIb␤3 bind soluble fibrinogen (24) or adhere to immobilized fibrinogen (15). We found little adherence of transfected lymphocytes to either osteopontin or fibrinogen in the absence of PMA stimulation (Fig. 3). Following PMA stimulation, lymphocyte adherence to osteopontin and fibrinogen increased by 13.6-and 8.1fold, respectively. Nevertheless, whereas adherence to fibrinogen was inhibited by the mAb A2A9, indicating it was mediated by ␣IIb␤3, adherence to osteopontin was inhibited by the mAb LM609, indicating it was mediated by ␣v␤3. Next, we measured the adherence of the parental line GM1500 that expresses ␣v␤3 but not ␣IIb␤3 to both substrates (Fig. 3). As anticipated, the parental cells did not adhere to fibrinogen; however, their adherence to osteopontin was identical to that of the transfected cells. These data indicate that ␣v␤3 on lymphocytes, like ␣v␤3 on platelets, resides on the cell surface in an inactive state and confirm that agonist-generated intracellular signals can induce ␣v␤3 binding to osteopontin.
Because the ␣v␤3 on activated platelets and lymphocytes interacts with osteopontin, whereas ␣IIb␤3 on these cells interacts with fibrinogen, the ␣ subunit of ␤3 integrins regulates their ligand binding specificity. The heavy chains of ␣v and ␣IIb exhibit 38% overall homology, but homology increases to 57% when their amino-terminal halves containing their putative calcium binding domains are compared (25). Recent studies of ␣IIb and ␣v concluded that the amino-terminal one-third of each protein is involved in ligand recognition (26), a conclusion consistent with peptide cross-linking studies implicating two sites in ␣v encompassing amino acids 139 -349 (27) and a site in ␣IIb that contains amino acids 294 -314 (28). A major difference between the proximal end of the putative ␣v ligand binding domain and the corresponding region of ␣IIb is the presence of a stretch of 10 additional amino acids in ␣IIb (Gly 148 -Glu 157 ) (25), perhaps accounting for the different ligand preference of the integrins containing these ␣ subunits.
Role of Divalent Cations in Platelet Adherence to Osteopontin-␣v␤3-mediated cell adhesion to osteopontin occurs in the presence of Mg 2ϩ , although Ca 2ϩ can support osteopontin binding to purified ␣v␤3 (29). To verify that Ca 2ϩ can support platelet adherence to osteopontin, we measured the adherence of ADP-stimulated platelets suspended in buffer containing either Ca 2ϩ or Mg 2ϩ (Fig. 4). In the presence of EDTA, there was no adherence of ADP-stimulated platelets to osteopontin. However, both Ca 2ϩ and Mg 2ϩ supported platelet adherence to osteopontin in a concentration-dependent manner. At cation concentrations up to 1 mM, there was little difference in the ability of Ca 2ϩ and Mg 2ϩ to support platelet adherence; although at higher concentrations, the ability of Ca 2ϩ to support adherence declined relative to that of Mg 2ϩ . Nevertheless, at a physiologically relevant concentration of 1 mM, Ca 2ϩ supported platelet adherence to osteopontin nearly as well as Mg 2ϩ . Moreover, adherence in either Ca 2ϩ -or Mg 2ϩ -containing buffer was inhibited by LM609, indicating that it was mediated by ␣v␤3 regardless of the divalent cation present (data not shown).
The divalent cation dependence of platelet ␣v␤3 deviates significantly from that of ␣v␤3 in certain cell lines where Ca 2ϩ does not support cell adherence to osteopontin and can even be inhibitory (29). It has been reported that the ␣v␤1-mediated adherence of 293 cells to osteopontin occurs in media containing Ca 2ϩ but only when ␣v␤1 is exposed to an activating monoclonal antibody, suggesting that the affinity state of ␣v␤1 determines its ability to interact with Ca 2ϩ (30). We found that only ␣v␤3 on activated platelets mediates adherence to os- teopontin in the presence of Ca 2ϩ , suggesting that the affinity of this integrin also determines its ability to interact with Ca 2ϩ . However, in contrast to ␣v␤1 and to ␣v␤3 expressed by cells other than platelets, platelet ␣v␤3 is not constitutively active in the presence of Mg 2ϩ .
Selective Inhibition of Platelet Adherence to Osteopontin by a Cyclic Peptide Selective for ␣v␤3-The ability of ␣v␤3 to mediate platelet adherence to osteopontin at physiologic Ca 2ϩ concentrations suggests that RGD-based peptides with selectivity for ␣v␤3 over ␣IIb␤3 could be of clinical utility. To test this possibility in vitro, we compared the ability of XJ735 (cyclo(Ala-Arg-Gly-Asp-Mamb), where Mamb is meta-aminomethyl benzoic acid), a cyclic RGD-based peptide selective for ␣v␤3 (31), to inhibit ADP-stimulated platelet adherence to osteopontin and fibrinogen (Fig. 5). XJ735 abolished platelet adherence to osteopontin with an IC 50 of ϳ6 M, whereas its effect on adherence to fibrinogen was incomplete with an IC 50 that was Ͼ1 mM. Thus, peptides selective for the ␣v␤3 integrin can discriminate between osteopontin and fibrinogen and accordingly could inhibit platelet adhesion to the wall of injured arteries without impairing the ␣IIb␤3-mediated platelet aggregation responsible for primary hemostasis.
Our results have a number of important implications. First, The affinity state of ␣v␤3 on platelets and lymphocytes, like that of ␣IIb␤3, is regulated by cellular agonists. Moreover, because agonist-generated signals interact with the ␤3 cytoplasmic tail to up-regulate ␣IIb␤3 function (24), it is reasonable to postulate that a similar mechanism is involved in up-regulating the function of ␣v␤3. Second, because ␣v␤3, but not ␣IIb␤3, binds to osteopontin, our results establish that the ␣ subunit regulates the selectivity of ␤3 integrins for natural ligands. Third, because osteopontin is a major constituent of atherosclerotic plaques (8 -10), is absent from the endothelium of normal arteries (8 -10), and is strongly up-regulated in areas of endothelial damage (6,8,32), it may be possible to impair the formation of platelet thrombi in arteries by preventing the interaction of platelet ␣v␤3 with osteopontin. A potential advantage of this approach is that it may be less prone to impair hemostasis than current therapies. Indeed, peptides that bind to ␣v␤3 have been shown to be effective in animal models for restenosis (33), and we have shown that XJ735 has significant, ␣IIb␤3-independent, in vivo antithrombotic efficacy but does not increase bleeding times in dogs and swine. 2 Thus, ␣v␤3mediated platelet adherence to osteopontin could be involved in the pathogenesis of acute arterial occlusion, and inhibitors of this integrin may be useful pharmaceutical agents for treating arterial thrombotic disorders.