The interaction of thrombospondin with platelet glycoprotein GPIIb-IIIa.

The interaction of human platelet thrombospondin (TSP) with human platelet glycoproteins GPIIb-IIIa was studied using a solid-phase binding assay. Polystyrene test tubes were coated with TSP, and 125I-labeled GPIIb-IIIa was added, allowed to bind, and the bound radioactivity was measured. After 90 min, the binding became time independent, and in most experiments, more than 10% of the exogenously added radioactivity was bound to the tube. Analysis of the bound radioactivity by polyacrylamide gel electrophoresis and autoradiography indicated that it was from labeled GPIIb-IIIa. Several lines of evidence indicate that the binding of GPIIb-IIIa to TSP was specific. (a) TSP immobilized on plastic or Sepharose bound 3-10-fold more GPIIb-IIIa than immobilized bovine serum albumin. (b) Addition of unlabeled excess GPIIb-IIIa reversed the binding of 125I-labeled GPIIb-IIIa to immobilized TSP. (c) Addition of EDTA inhibited the binding of GPIIb-IIIa to TSP by more than 90%, whereas addition of 1 mM CaCl2 and 1 mM MgCl2 potentiated the binding by more than 100%. (d) Monoclonal antibodies against TSP and GPIIb-IIIa inhibited the binding by 30-70% as compared with control and polyclonal anti-fibrinogen anti-serum. (e) A plot of GPIIb-IIIa bound versus GPIIb-IIIa added was best described as a rectangular hyperbola by regression analysis with half-saturation at 60 ng/ml GPIIb-IIIa. Similar results were obtained when labeled TSP was added to tubes coated with GPIIb-IIIa. These results show that TSP and GPIIb-IIIa can specifically interact in vitro and suggest that GPIIb-IIIa may function as a platelet TSP receptor during platelet aggregation.

table Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. (1). TSP is a major platelet-secreted protein accounting for more than 25% of the total protein secreted by activated platelets (2). Much information is known about the physical and chemical properties of TSP. For example, TSP, like fibronectin, contains distinct structural domains that, when cleaved from the intact molecule by proteolytic enzymes, still maintain their ligand-binding properties. These TSP fragments have been shown to interact with heparin (3), fibrinogen (2), and type V collagen ( 4 ) . The structure of the intact molecule as revealed by electron microscopy consists of several globular regions connected by three strands that resemble a South American bola, a weapon made of long cord with heavy balls at the end, used for roping cattle.
Platelet aggregation is an important step in the hemostatic process. As a consequence of vessel injury, platelets become activated, aggregate, and, in concert with the plasma-clotting system, seal the damaged vessel. The importance of platelets in normal hemostasis can be appreciated when patients who develop thrombocytopenia run the risk of severe hemorrhage. Therefore, an understanding of how platelets and their associated proteins function in blood coagulation is critical to the development of antithrombotic therapies.
The role of thrombospondin in the mechanism of platelet aggregation and blood coagulation is poorly understood. It is widely believed that TSP functions to cross-link platelet aggregates irreversibly during the secretion phase of platelet aggregation. This conclusion is based on several experimental observations. ( a ) At least 80% of the total TSP secreted by platelets in response to thrombin, a potent aggregating agent, binds the platelet surface ( 5 ) . ( b ) TSP binds fibrinogen (2).
One interpretation of the studies described above suggests that during platelet aggregation, TSP and fibrinogen share the same platelet receptor. To test this hypothesis, we developed an in vitro binding assay to measure the interaction of TSP with the platelet fibrinogen receptor, GPIIb-III,. Our results indicate that TSP specifically and saturably binds the GPIIb-III, complex.

EXPERIMENTAL PROCEDURES
Materials-All reagents, unless specified otherwise, were purchased from Sigma. Reagents for sodium dodecyl sulfate-polyacrylamide gel electrophoresis were obtained from Bio-Rad. Iodobeads were purchased from Pierce Chemical Co. NalZ5l was purchased from DU Pont-New England Nuclear. The monoclonal antibodies AzAs and SSA, to the GPIIb-III. complex were gifts of Dr. Joel Bennett (University of Pennsylvania) and were stored in a 10.9 mg/ml stock solution in phosphate-buffered saline. Monoclonal anti-TSP was characterized previously (7). AzAs is directed against the GPIIb-III, complex (lo), and SSAG is GPIII, specific (11). TSP was purified by fibrinogen-Sepharose chromatography from the released proteins of ionophore-stimulated platelets essentially as described previously (2). Platelet glycoproteins GPIIh and 111, were purified by a three-step procedure consisting of Lens culinaris affinity chromatography, anion-exchange chromatography, and anti-GPIIb-III. (A&) monoclonal antibody affinity chromatography as described previously (12). Preparations of GPIIh-III. contained no GPIV as indicated by the absence of cross-reactivity with anti-GPIV antibody, kindly provided by Dr. Tandon of the American Red Cross, Rockville, MD.
Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis was performed as described previously (13). Gels were stained, dried onto paper, and autoradiograms were prepared from the dried gels using intensifying screens (Du Pont Cronex Lightning Plus screens mounted in Spectroline cassettes, Reliance X-Ray Inc., Oreland, PA). Kodak X-Omat-AR film was used and developed according to the instructions provided with the film. Films were exposed overnight at Protein Assays-Protein concentrations were measured by the method of Bradford (14) using bovine serum albumin as standard.
Labeling of Proteins-TSP and GPIIb-111. were labeled with [1251] iodine using Iodobeads. Briefly, 100 p1 of a 1 mg/ml solution of either TSP or GPIIh-III, dissolved in 10 mM Tris-HCI, pH 7.4, containing 0.1% Tween 20, 1 mM CaC12, 1 mM MgC12, 149 mM NaCl (labeling buffer), was treated with 1 mCi of NaIz5I for 5 min in the presence of one Iodobead. Unreacted iodide was separated from protein by means of a PD-10 column (Pharmacia LKB Biotechnology Inc.) equilibrated in labeling buffer. Labeled TSP was further purified by heparinagarose chromatography. TSP was bound to heparin-agarose equilibrated in labeling buffer and eluted from heparin-agarose with labeling buffer containing 0.55 M NaC1. Proteins were typically labeled to a specific activity of 4000 cpm/ng.
Binding Assay-The inside surfaces of 3-ml polystyrene test tubes (Elkay) were coated with either GPIIb-III., bovine serum albumin (BSA), fibrinogen, or TSP. Tubes were coated overnight at 4 "C by the addition of 2 pg of protein dissolved in 20 mM Tris-HCI, pH 7.4, containing 50 mM NaCI, 1 mM CaC12, and 1 mM MgCL (coating buffer). Tubes were washed three times with 200 p1 of coating buffer containing 0.05% Tween 20 (binding buffer). Uncoated sites in the tube were blocked at room temperature for 60 min by the addition of 300 p1 of a 1 mg/ml BSA solution prepared in binding buffer. After blocking, tubes were washed three more times with 200 ~1 of binding buffer and were ready to use in the bninding assay. Tubes bound 0.3-1 Wg of protein as estimated from radiolabeled tracer studies. Binding assays were carried out for 2 h at room temperature in 200 p1 of binding buffer containing either radiolabeled GPIIb-III., TSP, or BSA. For all studies except those in which the concentration of the ligand in the fluid phase was varied, the amount of the ligand in the fluid phase was adjusted to equal the amount of receptor protein immobilized on the tube. The specific activity of ligand in a typical binding assay was usually lo6 cpm/pg of protein. The amount of bound ligand was counted after extensive washing of the tubes with binding buffer. In antibody-blocking experiments, ligand solutions were treated for 30 min with antibody prior to the start of the binding assay.
Affinity Chromatography-Approximately 2 mg of TSP or ovalbumin was coupled to 300 pl of CNBr-activated Sepharose according to the instructions provided by the manufacturer. The washed proteincoupled gels were packed into small 5-ml disposable columns (Isolab Inc.). Two ml of '*'I-GPIIb-III. (4 X lo6 cpm) in binding buffer was slowly passed over each column. Columns were washed with 25 ml of binding buffer and eluted with 2 ml of binding buffer containing 0.25 M NaCI. 0.5-ml fractions were collected, and 30 p1 of each fraction was analyzed on SDS-polyacrylamide gels. 9.6% of the input radioactivity bound the column.

RESULTS
Binding of GPIIb-III, to Immobilized TSP-In 1 h, approximately 1.5 ng of GPIIb-111, bound to 1 pg of immobilized TSP as calculated from the specific activity of the '251-GPIIb-III, added to the incubation mixture, whereas less than 0.10 ng bound in the presence of unlabeled excess GPIIb-111, during the same period of time (Fig. 1). The binding of GPIIb-111, to TSP became time independent after 40 min and was partially reversible with excess unlabeled GPIIb-111, added 30 min after initiation of binding (Fig. 1). These results indicate that unlabeled excess GPIIb-111, can compete for binding of radiolabel, but unlabeled GPIIb-111, can only partially reverse binding of radiolabel once saturation has occurred. Additionally, these results indicate that binding of GPIIb-111, to TSP is partially reversible after time-independent binding has been achieved and that binding constants calculated from concentration dependences cannot be regarded as representing thermodynamically valid equilibrium constants.
Under time-independent binding conditions, the amount of GPIIb-111, bound to immobilized TSP uersus GPIIb-111, added shows saturation at approximately 1.0 pg of GPIIb-111. added (Fig. 2). The divalent cation requirements for the binding of GPIIb-III, to immobilized TSP were investigated. The binding required both M e and Ca2+ because EDTA, which complexes both ions, totally abolished binding, whereas EGTA, which preferentially complexes Ca'+, only partially prevented binding (Fig. 3). Furthermore, the data in Fig. 3 show that binding can occur without the addition of divalent cations (bar graph designated no Ca/Mg), suggesting that protein-bound Ca2+ can support substantial binding.
To establish that the bound radioactivity represented GPIIb-111, and not some labeled impurity, tubes were eluted with SDS and the eluted material analyzed by SDS-polyacrylamide gel electrophoresis followed by autoradiography (Fig.   4). Labeled bands at 155,000 and 97,000 daltons under nonreducing conditions (lanes 2 and 3 in Fig. 4) and 135,000 and 110,000 under reducing conditions (lunes 4 and 5 in Fig. 4), corresponding to GP& and GPIII., respectively, were observed in the labeled GPIIb-111, preparation added to the tubes as well as in the material eluted from the TSP-coated tubes. These results indicate that our assay measures bound GPIIb-111,.
Binding of TSP to Immobilized GPII~-GPIII,-In 2 h, approximately 1.2 ng of TSP bound to immobilized GPIIb-III, calculated from the specific activity of the "'1-TSP added to the incubation mixture, whereas less than 0.4 ng bound im- to immobilized TSP. TSP-coated tubes were incubated with 3 pg of '2sII-GPIIh-III. in binding buffer containing the various additions designated in the figure. The tubes were then washed and counted, and the amount of protein bound was calculated from the specific activity of the '2'I-GPIIb-III. added. The data points are the mean of triplicate determinations, and the bars represent the standard deviation. mobilized BSA during the same period of time (Fig. 5). The binding of TSP to GPIIb-111. became time independent after 2 h, indicating that the interaction had gone to completion (Fig. 5). The SDS-gel profile of the "'1-TSP used in these studies is shown in lane 1 of Fig. 4. These results indicate that fluid-phase TSP is capable of binding GPIIb-111. immobilized on a surface.

Specificity of Binding of GPZZb-ZZZa to Immobilized TSP-
To establish further the specificity of the interaction of GPIIb-111. with TSP, incubation mixtures containing GPIIb-111. were treated with either control IgG, fibrinogen, the peptides Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) or Gly-Arg-Gly-Glu-Ser-Pro (GRGESP) or antibodies against TSP, GPIIb-III., or fibrinogen and tested in the binding assay (Fig. 6). Only antibodies to GPIIb-III., SSAe, and A2A9 and a monoclonal antibody against TSP significantly inhibited the binding of GPIII,-III. to immobilized TSP. These results indicate that GPIIh-111. interacts specifically with TSP immobilized on our solid-phase assay tubes. As an additional control to rule out nonspecific interactions on the plastic surface of our solid-phase assay tubes, TSP as well as BSA were coupled to Sepharose, and the derivatized gels were examined for GPIIb-III.-binding activity. As can be seen in Fig. 7, only TSP-Sepharose bound '2sI-GPIIt,-III., which was eluted with NaCl. These results further support the conclusion that GPIIb-111, specifically interacts with TSP. in platelet aggregation (18, 19). Of great interest is the identification of the cellular TSP receptor. Ash and co-workers (20) have presented evidence that an 88,000-dalton membrane glycosprotein present in platelets, endothelial cells, and a variety of human tumor cells is the membrane-binding site of TSP. This protein was identified as glycoprotein IV (GPIV) by the use of specific anti-GPIV antisera (21). Although TSP was shown to bind to monocytes and mediate platelet-monocyte adhesion, no functional role for GPIV in these adhesive interactions was described (22).
Our recent studies (7, 18) suggest that GPIIb-III., the cell surface heterodimeric glycoprotein complex that mediates platelet aggregation and other platelet adhesive functions by its interaction with fibrinogen, fibronectin, and von Willebrand factor (22), may mediate TSP-promoted platelet aggregation. We found that antibodies against GPIIb-111. inhibited the TSP potentiation of thrombin-stimulated platelet aggregation (7, 18). That GPIIb-111. may function as a platelet TSP receptor is also suggested from studies showing co-localization of TSP and GPIIb-111. on the surface of thrombin-stimulated platelets (9) and from a study showing that anti-GPIIb-111. inhibits binding of TSP to activated platelets (6). Clearly, however, other TSP receptors must also exist on the platelet surface since thrombasthenic platelets that are deficient in GPIIb-111. also bind TSP (23).
In order to determine whether GPIIb-III. can serve as a platelet TSP receptor, a simple in vitro binding assay utilizing purified TSP and GPIIb-III. was developed. Tubes were coated with either TSP or GPIIb-III., the corresponding radiolabeled GPIIb-111. or TSP was added, and the bound ligand was counted after extensive washing to remove unbound material. We found that either GPIIb-111, bound to immobilized TSP, or TSP bound to immobilized GPIIb-111. in a time-independent manner after less than 2 h of incubation. The binding was saturable and could be inhibited by either antibodies against GPIIb-111. or TSP. Binding was dependent on the presence of the divalent cations M e and Ca2', consistent with the results reported previously for the binding of TSP to activated platelets (24).
The stoichiometries in these solid-phase reactions do not reflect stoichiometries that might be expected in the fluid phase. In our study, approximately one molecule of GPIIb-111. bound 100 molecules of TSP. A similar stoichiometry was observed by Leung and Nachman (25) for the interaction of fibrinogen with TSP absorbed to plastic microtiter plates. As pointed out by Leung and Nachman (25), the poor stoichiometry could be due to randomly oriented TSP on the plastic giving rise to significant steric hindrance of binding. Thus, in our system, only one out of every 100 absorbed TSP molecules is capable of interacting with fluid-phase GPIIb-111..
Our results provide direct evidence for the interaction of GPIIb-111. with TSP. The monoclonal antibody specific to GPIIIa, SSA6, inhibited binding of GPIIb-111. to TSP by 60%, whereas anti-GPIIb-111, (&A9), specific for the heterodimeric complex, inhibited binding by 35%. These levels of inhibition were maximal in the presence of excess antibody. We interpret these results as suggesting that the TSP-binding epitope of GPIIb-111. is complex and may be located closer to the GPIII. portion of GPIIb-111.. We could also demonstrate the interaction of GPIIb-111. with TSP by enzyme-linked immunoassay.* In addition, our preparations of GPIIb-111. bound fibrinogen to the same extent as TSP (data not shown). We could also demonstrate binding of GPIIb-111. to TSP immobilized on Sepharose (Fig. 7), suggesting that our observed ligand binding was not due to an artifactual modification of the receptor proteins from adsorption on the tube surface.
Previous efforts to demonstrate an interaction between TSP and GPIIb-111. have yielded conflicting results. Leung and Nachman (25), using an enzyme-linked immunoassay, could demonstrate an interaction between GPIIb-111. and fibrinogen but not between GPIIb-111. and TSP. Similarly,

Thrombospondin Interaction with
Platelet Glycoprotein GPIIb-III, as well as a related integrin receptor present on human endothelial cells, smooth muscle cells, monocyte-like cells, normal rat kidney cells, and normal and thrombasthenic platelets. However, in contrast to the findings of Lawler and co-workers (28, 29), we could not demonstrate inhibition of binding of GPIIb-III, to TSP in our solid-state assay by Arg-Gly-Asp-containing peptides (Fig. 6). The reason for these discrepancies is presently unknown, although differences in the biological activities of either purified TSP or GPIIb-III, used in the various studies could account for some of the observed differences in binding. In conclusion, our studies suggest that GPIIb-III. may function as one of the receptors for TSP on the surface of activated pIatelets. During aggregation, TSP may not directly cross-link platelet aggregates but rather act synergistically with fibrinogen through GP&-111, on the platelet surface. TSP could either provide additional fibrinogen receptor sites on the platelet surface or increase the binding constant for fibrinogen by interaction with either fibrinogen or the GPIIb-III, complex. Finally, TSP could directly interact with GPIIb-III, and cross-link platelets through a fibrinogen-independent pathway of platelet aggregation, which has been suggested by two recent studies (30, 31). Further work is needed to differentiate between these possibilities.