JOURNAL OF BONE AND MINERAL RESEARCH Volume 6, Number 10, 1991 Mary Ann Liebert, Inc., Publishers

Osteonectin Is an a-Granule Component Involved with Thrombospondin in Platelet Aggregation PHILIPPE CLEZARDIN,’.’ LUC MALAVAL,‘ MARIE-CHRISTINE MOREL,3 JOSETTE GUICHARD,4 THOMAS LECOMPTE,’ MARIE-CHRISTINE TRZECIAK,’ MARC DECHAVANNE,, JANINE BRETON-GORIUS,4 PIERRE D. DELMAS,’ and CECILE KAPLAN3

ABSTRACT We previously showed that thrombospondin, a major a-granule glycoprotein of human platelets, forms a specific complex with osteonectin, a phosphoglycoprotein originally described in bone that is also present in human platelets. The storage organelles and the function of osteonectin in platelets are still unknown. In this study, using electron microscopy in combination with immunogold staining, osteonectin was located within the major storage organelle for platelet-secreted proteins, the a-granules. Furthermore, osteonectin was qualitatively and quantitatively assessed by studying normal platelets and the platelets from a patient with gray platelet syndrome. Gray platelet syndrome is a rare congenital bleeding disorder characterized by a selective deficiency in morphologically recognizable platelet a-granules and in the a-granule secretory proteins. Binding of an iodinated antiosteonectin monoclonal antibody to gray platelet proteins transferred to nitrocellulose from SDS-polyacrylamide gels showed no band corresponding to osteonectin compared to control platelets. Using a polyclonal antiosteonectin antibody-based radioimmunoassay, gray platelets contained 0.2 f 0.03 ng osteonectin per lo6 platelets, which is only 20% of the normal platelet content of osteonectin (0.93 f 0.16 ng per lo6 platelets). Study of the localization of osteonectin to the surface of human platelets demonstrated that a radioiodinated antiosteonectin polyclonal antibody bound specifically to thrombinstimulated platelets but not to resting platelets. Binding was concentration-dependent, saturable (1710 f 453 binding sites per platelet, Kd = 1 pM), and inhibited by an excess of cold antiosteonectin polyclonal antibody. N o binding was observed on the surface of thrombin-stimulated gray platelets. To gain further insights into the role of osteonectin released from activated platelets, the effect of an antiosteonectin polyclonal antibody was tested on the aggregation of washed platelets. F(ab), fragments from the antiosteonectin polyclonal antibody inhibited in a dose-dependent manner the aggregation of collagen-stimulated, washed human platelets without affecting collagen-induced platelet serotonin release. To characterize the mechanism through which antiosteonectin F(ab’), fragments inhibit platelet aggregation, the expression of endogenous thrombospondin (TSP) on the surface of thrombin-activated platelets was studied using ‘251-labeled anti-TSP monoclonal antibody P10. The endogenous surface expression of TSP to thrombin-stimulated platelets was significantly inhibited in the presence of antiosteonectin F(ab’), fragments (6286 f 2065 molecules of P10 per platelet) compared to 11,230 f 766 molecules of P10 per platelet in the presence of nonimmune F(ab), fragments. This inhibitory effect of antiosteonectin F(ab’), fragments on the surface expression of endogenous TSP was not mediated by interference with binding of monoclonal antibody P10 to TSP as judged by enzyme-linked immunosorbent assay. In summary, these results demonstrate that osteonectin is an a-granule component that, by binding on the surface of activated platelets, is involved with TSP in the secretion-dependent phase of the platelet aggregation process.

’INSERM U 234, Laboratoire de Biochimie des Proteines Osseuses, HBpital Edouard Herriot, Lyon, France. ’INSERM U 33 I , Laboratoire d’Hemobiologie, Faculte de Medecine Alexis Carrel, Lyon, France. ’INTS, Service d’lmmunologie Leuco-Plaquettaire, Paris, France. ‘INSERM U 91, HBpital Henri Mondor, Creteil, France. 5HHBtel-Dieu de Paris, Paris, France.

1059

1060

CLEZARDIN ET AL.

INTRODUCTION HROMBOSPONDINis a

high-molecular-mass (450 kD) glycoprotein present in platelet a-granules and is thus secreted when platelets are stimulated with thrombin.“) The secreted thrombospondin binds to the surface of activated platelet^(^.^) and is involved in platelet aggregation. (4-6) Platelet aggregation is a crucial step in normal hemostasis. Recent evidence has indicated that the sequence of events leading to platelet aggregation involves stimulation by an agonist followed by shape change, a-granule secretion, and the membrane localization of several a-granule proteins. Thrombospondin appears to be essential for the secretion-dependent phase of platelet aggregation, making platelet aggregation irreversible.(4) Thrombospondin is also synthesized by a wide range of cultured cells, including fibroblasts, monocytes and macrophages, keratinocytes, endothelial, smooth muscle, glial, squamous carcinoma, and melanoma cells (for review see ~ ~ ~osteosarcoma ) cells.(7) Refs. 5 and 6), o s t e ~ b l a s t s , (and Source-specific differences in fragmentation patterns between thrombospondin molecules were reported following exposure to proteolytic enzymes, suggesting thrombospondin polymorphism. ( 9 - 1 2 ) The exact physiologic functions of thrombospondin in these cells is unknown; however, there is growing evidence that it is involved in cell adhesion, ( 1 1 , 1 3 . 1 4 ) migration, ( l 5 . l 6 ) and proliferation. ( L7,18) Thrombospondin is composed of three equivalent disulfide-linked chains of 150-160 kD.I5)Each thrombospondin chain is made up of several protease-resistant domains, which bind specifically to heparin, thrombin, sulfated glycolipids, fibronectin, fibrinogen, collagen, laminin, histidine-rich glycoprotein, plasminogen, and plasminogen activators (for review see Refs. 5 and 6). We recently reported that thrombospondin also forms a specific calciumdependent complex with osteonectin,(19)a 32 kD phosphoglycoprotein originally described in but also found in endothelial cells,(z2) fibroblasts,(23) and platelets.(24LPlatelet and bone osteonectin are different both ~ t r u c t u r a l l y ( ~ ~and . ’ ~ ) i m m u n ~ l o g i c a l l y . ~Such ~~) structural and immunologic differences may lead to functional differences between platelet- and bone-derived osteonectin. In this respect, bone osteonectin promotes the deposition of calcium phosphate mineral onto type I collagen(20,21) and inhibits the growth of hydroxyapatite crystals in vitro.(”) Osteonectin also modulates endothelial cell and fibroblast behavior in vitro, particularly with respect to the spreading of these cells to plastic- or collagen-coated surfaces.(27)Its function and that of the storage organelles in platelets are unknown. Measurement of circulating levels of osteonectin in serum of normal subjects and of patients with thrombocytopenia shows a logarithmic positive correlation with platelet count, suggesting that some trapping of platelet osteonectin occurred either in the clot or to the surface of activated platelets. (I8) These findings, taken together with the fact that thrombospondin and osteonectin are present as a complex when released from thrombin-activated platelets,(19) prompted us to examine whether osteonectin is an a-granule component that, by binding to the surface of activated platelets, takes part in the platelet aggregation process.

T

In this study, using electron microscopy in combination with immunogold staining, osteonectin was located within the major storage organelle for platelet-secreted proteins, the a-granules. Furthermore, osteonectin was qualitatively and quantitatively assessed by studying normal platelets and the platelets from a patient with the gray platelet syndrome. Gray platelet syndrome is a rare congenital bleeding disorder characterized by a selective deficiency in morphologically recognizable platelet a - g r a n u l e ~ ( *and ~ ) in the a-granule secretory proteins. (3031) Using western blot and radioimmunoassay techniques, platelets of a patient fulfilling the previously described criteria for the gray platelet syndrome possess a severe deficiency of osteonectin, confirming that a-granules are a storage organelle for this protein. Study of the localization of osteonectin on the surface of human platelets demonstrates that a polyclonal antiosteonectin antibody binds specifically to thrombin-activated normal platelets but not to thrombin-activated gray platelets. Such a polyclonal antibody inhibits both collagen-induced platelet aggregation and expression of endogenous thrombospondin on the surface of activated platelets, suggesting that osteonectin is involved with thrombospondin in the secretion-dependent phase of the platelet aggregation process.

MATERIALS AND METHODS Materials Unless otherwise stated chemicals were of analytic grade from Merck. Sodium dodecyl sulfate (SDS) was from BDH Chemicals, Ltd. Acrylamide, N’,N’-methylenebisacrylamide, N, N, N’, N’-tetramethylenediamine, ammonium persulfate, nitrocellulose membranes. Tween 20, and the silver-staining kit were from Bio-Rad. Carrier-free l Z 5 I Na and [14C]serotoninwere from Amersham. Iodo-Beads were from Pierce. PD-10 columns (Sephadex G-25), protein A-Sepharose CL4B, and molecular mass standards were from Pharmacia Fine Chemicals AB. X-Omat S films for indirect autoradiography were obtained from Kodak. Coomassie brilliant blue R-250, fibronectin, heparin, apyrase grade I11 (1.3 mg/ml diluted in 0.9% NaCI), and bovine serum albumin (Cohn, fraction V) were from Sigma. Adenosine 5’-diphosphate (ADP) was from Boehringer, collagen was from Hormon-Chemie, bovine thrombin (64U h g ) was from Hoffmann-La Roche, and D-phenylalanyl-L-prolyl-L-arginine chloromethylketone (PPACK) was from Calbiochem. Fibrinogen was purchased from Imco (Sweden). Platelet concentrates were kindly provided by Dr. F. Robert (Centre de Transfusion Sanguine, Beynost), and platelet thrombospondin purified by anion-exchange chromatography as previously described.(3z)Antiosteonectin mouse monoclonal antibodies ON3 and ON6 were produced by CIS BioIndustries (France). The characterization of these antibodies has been previously described.

Case report The patient with the gray platelet syndrome was a 40year-old female originally described by Kaplan et

1061

OSTEONECTIN IN PLATELETS She had been complaining of hypermenorrhae and frequent epistaxis since childhood. Severe bleeding occurred at her deliveries, and one child died at birth. She was then referred for evaluation and was found to have large gray agranular platelets on May-Grunwald-Giemsa-stained blood smears. Electron microscopy on thin sections of human blood platelets from this patient showed a selective deficiency in morphologically recognizable a-granules (Fig. I). Her Ivy bleeding time was prolonged ( > 20 minutes), and her platelet count decreased (70 x lo9 liter). Platelet aggregation occurred subnormally with ADP, thrombin, and collagen. Analysis of patient platelet proteins by periodic acid-Schiff (PAS) staining, in combination with one-dimensional SDS-polyacrylamide gel electrophoresis (SDS-PAGE), revealed the absence of intracellular fibrinogen and thrombospondin. Such results were confirmed by crossed i m m u n o e l e c t r ~ p h o r e s i s ~and ~ ~ ~by) twodimensional SDS-PAGE followed by Coomassie blue and/or silver staining (results not shown).

centrifuged and fixed in a mixture of 2% (vol/vol) paraformaldehyde and 0.5% (vol/vol) glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, for 90 minutes at 22°C. After incubation, fixed platelets were washed three times in phosphate buffer, p H 7.2, and embedded in glycol methacrylate (GMA) as previously described.134)Ultrathin sections were collected on nickel grids coated with Formvar. Immunogold staining was performed on thin GMA sections as detailed Briefly, thin sections were first incubated for 2 h at room temperature with 4 pg/ml of antiosteonectin mouse monoclonal antibody ON3. Alternatively, a mouse monoclonal antibody (NP57) directed against neutrophil elastase or normal mouse serum was used as negative control. After washing, thin GMA sections were then incubated with goat antimouse IgG coupled with 15 nm colloidal gold particles (GAMGl5, Janssen Pharmaceutica, Beerse, Belgium). Incubation was for 1 h at room temperature. Thin sections were finally stained with lead citrate and examined on CM 10 Phillips electron microscope.

Electron microscopy

Radioiodination of monoclonal antibodies

Blood samples were taken from healthy volunteers by venipuncture on sodium EDTA. Platelet-rich plasma was

The characterization of antiosteonectin mouse monoclonal antibodies ON3 and ON6'251and of antithrombospon-

FIG. 1. Electron microscopy on thin sections of human blood platelets. (A) Platelet from a normal donor showing the open canalicular system (OCS) and numerous a-granules (arrows) (original magnification x 20,000). (B) Platelet from a patient with the gray platelet syndrome. Note the selective deficiency in morphologically recognizable a-granules (original magnification x 20,000).

CLEZARDIN ET AL.

1062

din mouse monoclonal antibody P10(36)was previously described. Each monoclonal antibody (100 pg) was radioiodinated with 1 mCi of carrier-free I z 5 I Na for 5 minutes in the presence of lodo-Beads according to the manufacturer’s directions. Such a procedure was also used to radioiodinate antiosteonectin rabbit polyclonal antibody 2.2. Following iodination, free I z 5 I was removed from the sample by gel filtration through Sephadex (3-25 (PDlO column). The specific activity of each antibody ranged between 500 and 1000 cpm/ng. Radioiodinated antibodies were stored at -80°C.

Preparation of washed platelets The study was conducted according to the principles embodied in the Declaration of Helsinki. Blood (6 vol) from informed and consenting normal volunteers and from the patient with the gray platelet syndrome was drawn into acid/citrate/dextrose anticoagulant (1 vol). The blood was centrifuged for 20 minutes at 160 x g to obtain plateletrich plasma. Platelets were isolated from the platelet-rich plasma by centrifugation at 1200 x g for 15 minutes and the platelet pellet washed three times by centrifugation at 1200 x g, following the technique of Mustard et Briefly, platelets were washed once in Tyrode’s buffer (140 mM NaCI, 3 mM KCI, 12 mM NaHCO,, 4 mM NaH,PO,, 1 mM MgCI,, 2 mM CaCI,, and 5.5 mM glucose, pH 7.35) containing 0.35% (mass per vol) bovine serum albumin, 5 mM Hepes, apyrase (10 pl/ml), and heparin (50 U/ml). After centrifugation, the platelet pellet was washed in Tyrode’s buffer without heparin and platelets were finally resuspended at 2 x lo8 per ml in Tyrode’s buffer containing 2 pl/ml of apyrase.

Western blot procedure Washed human platelets were solubilized in 2% (mass per vol) SDS under nonreducing or reducing conditions and samples (80 pg protein) electrophoresed on 5-20070 SDS-gradient polyacrylamide gels. The immunoblotting procedure, including the electrophoretic transfer of proteins from unstained polyacrylamide gels to nitrocellulose membranes, was as reported by Malaval et a1.(2s)Nitrocellulose membranes were saturated for 1 h at room temperature in a Tris buffer (50 mM Tris, 20 mM CaCI,, and 0.03% sodium azide, pH 8.0) containing 5% (mass per vol) bovine serum albumin and were then incubated for 12 h at room temperature with [~zsI]monoclonalantibody ON3 diluted in Tris buffer containing 1% (mass per vol) bovine serum albumin (325,000 cpm/ml, 5 ml). After washing in Tris buffer containing 0.05% (vol/vol) Tween 20, nitrocellulose membranes were dried and exposed for autoradiography as previously reported.(zS)

Radioirnrnunoassay for osteonectin The procedure was as described by Malaval et al.(3*)The radioimmunoassay used a specific rabbit antiserum raised against purified bovine bone osteonectin that cross-reacts with human osteonectin. Bovine bone osteonectin stan-

dard was calibrated on the basis of amino acid content. The standard curve showed linearity over a significant range comprised between 10 and 1000 ng/ml of bovine bone osteonectin. The mean intra- and interassay variations of samples were 10 and 20070, respectively.

Platelet aggregation studies Aggregation studies were performed with washed platelets labeled with [L4C]~er~tonin.(40) The aggregation pattern was recorded in a Chrono-Log dual-channel aggregometer at 37°C under continuous stirring. ADP (after addition of 100 pg/ml of purified fibrinogen) was used to induce aggregation at concentrations of 1, 2, and 5 pM. Thrombin was used at a concentration of 0.06 U/ml. Collagen was used at concentrations of 1 and 2 pg/ml. F(ab), fragments from nonimmune and antiosteonectin IgG antibodies were prepared by pepsin digestion, separated from the Fc fragments by affinity chromatography on a protein A-Sepharose column, and extensively dialyzed against phosphate-buffered saline, pH 7.2, to remove sodium azide before their use in aggregation studies. The effect of the antiosteonectin polyclonal antibody on platelet aggregation was measured by preincubation washed platelets with the indicated concentrations of F(ab’), fragments for 2 minutes at 37°C before the induction of aggregation.

Binding experiments The localization of osteonectin on the surface of stimulated platelets was assessed using radioiodinated antiosteonectin polyclonal antibody 2.2, which was shown previously to immunoprecipitate from platelet extracts a single protein with a molecular mass identical to that of platelet osteonectin. I 3 O ) In addition to immunoprecipitation experiments, antiosteonectin polyclonal antibody 2.2 did not show any cross-reactivity with fibrinogen, fibronectin, or thrombospondin using an enzyme-linked immunosorbent assay or with collagen using a solid-phase radioimmunoassay (results not shown). Platelets previously washed in Tyrode’s buffer (2 x lo8 per ml) were stimulated with thrombin (0.72 U/ml) at room temperature under nonstirring conditions. Thrombin was inactivated 10 minutes later by the addition of PPACK M). After 10 minutes of incubation, binding was initiated by the addition of [1z51]antio~te~nectin polyclonal antibody 2.2 to final concentrations ranging from 13 to 130 nM. The reaction was terminated 30 minutes later by centrifugation (12,000 x g for 3 minutes) of triplicate aliquots (50 pl each) through 0.45 ml of 20% sucrose dissolved in Tyrode’s buffer. The centrifuge tube tips were computed and the radioactivity associated with the pellets determined in a gamma counter. Nonspecific binding was determined in parallel tubes using a 100-fold excess of unlabelled antiosteonectin polyclonal antibody 2.2, and this value was subtracted from the total radioactivity in the platelet pellet to yield specifically bound counts per minute. For binding experiments designed to study the inhibitory effect of antiosteonectin polyclonal antibody 2.2 on the

OSTEONECTIN IN PLATELETS endogenous expression of platelet thrombospondin, antiosteonectin and nonimmune antibodies in the form of F(ab)', fragments (60pg/ml) were allowed to incubate with washed platelets (2 x lon per ml) for 5 minutes at room temperature before the addition of thrombin (0.72 U/ml). Platelet activation initiated with thrombin was performed at room temperature under nonstirring conditions. After neutralization of thrombin by PPACK M), binding was initiated by the addition of [1251]antithrombospondin monoclonal antibody P 10 to final concentrations ranging from 13 to 200 nM. The remaining steps were performed as described previously. Nonspecific binding was determined using a 100-fold excess of unlabeled antithrombospondin monoclonal antibody P10.

1063 unreactive. Gold particles were mostly localized at the periphery of the matrix, suggesting a possible association with the inner face of the a-granule membrane. Controls performed using nonimmune mouse serum or a mouse monoclonal antibody directed against neutrophil elastase did not show any labeling (results not shown).

Qualitative and quantitative assessment of osteonectin in normal and gray platelets

SDS-solubilized platelet proteins from a control and a patient with the gray platelet syndrome were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and incubated with [12'I]antiosteonectin monoclonal antibody ON3. An autoradiograph of the immunoblot performed with control platelets Enzyme-linked immunosorbent assay (ELISA) showed a band of immunoreactivity at a molecular mass of The procedure was similar to that described previ- 45 kD under reducing conditions (Fig. 3A, lane 1) and of o ~ s I y . Both ( ~ ~ proteins ~ (thrombospondin and osteonectin) 39 kD under nonreducing conditions (Fig. 3A, lane 2). The and antibodies (P10, 2.2) used in this study were diluted to apparent molecular mass of this band under both reduced concentrations previously found to be optimal for this and nonreduced conditions is consistent with that of hu~ , ~ ~using ) Laemmli SDSELISA assay.(1p1Purified platelet thrombospondin (di- man platelet ~ s t e o n e c t i n ( ~when luted to a concentration of 0.35 pg/ml in bicarbonate buf- polyacrylamide gels. In sharp contrast, no band correfer, pH 9.6) was first incubated in siliconized tubes with sponding to platelet osteonectin was seen with gray platepurified bovine bone osteonectin (diluted to a concentra- lets under reducing and nonreducing conditions (Fig. 3B, tion of 10 pg/ml in bicarbonate buffer) for a 1 h incuba- lanes 1 and 2). In addition to the osteonectin band immution at room temperature. Thrombospondin and osteonec- nodetected in control platelets, there was a faint band with tin were then added either separately or together to 96-mi- a molecular mass of 52 kD under both reduced (Fig. 3A, crotiter plates with flat-bottomed wells (100 1 1 per well). lane 1) and nonreduced conditions (Fig. 3A, lane 2), which After an overnight incubation at 4"C, the wells were was also detected in gray platelets (Fig. 3B, lanes 1 and 2). washed three times with Tris-Tween buffer (15 mM Tris The presence of an excess of cold antiosteonectin monoand 2 mM CaCI,, pH 7.4, containing 0.05% by volume clonal antibody ON3 with the tracer abolished all the imTween 20; 250 pl per well). Purified antithrombospondin munoreactive bands (results not shown). The total osteonectin content of normal and gray platemonoclonal antibody P10,136'diluted to 1:500,000 in TrisTween buffer, was first incubated with antiosteonectin lets was quantitated by radioimmunoassay using washed, polyclonal antibody 2.213n)diluted to 1:8000 in Tris-Tween buffer. After a 5 minute incubation at room temperature, wells were emptied and the P10/2.2 antibody mixture was then added to thrombospondin-, osteonectin, and thrombospondin-osteonectin-coated wells. After a 1 h incubation, the washing procedure was repeated and a goat antimouse IgG or a goat antirabbit IgG conjugated with horseradish peroxidase (diluted to 1 :3000 in Tris-Tween buffer) was added to each well (100 pl per well) for an additional 1 h incubation at 37°C. The wells were emptied, the washing procedure repeated, and 100 pl substrate o-phenylenediamine (0.4 mg/ml in 0.1 M citrate buffer, p H 5, containing 1.3 mM hydrogen peroxide) added to each well. The enzyme reaction was allowed to proceed for 5 minutes in the dark at room temperature. The reaction was then stopped with 100 pl per well 2 M sulfuric acid. The absorbance was read at 492 nm using a microplate photometer (Bio-Tek Instruments). FIG. 2. Localization of osteonectin on thin section of human blood platelets. Fixed platelets were incubated with antiosteonectin monoclonal antibody ON3 and bound RESULTS ON3 visualized using a goat antimouse IgG coupled with 15 nm colloidal gold particles. Gold particles are excluIm m unogold labeling sively localized on a-granules near the inner face of the aThe labeling occurs within the platelet a-granules (Fig. granule membrane (arrows). (Original magnification x 2). Plasma membranes and other storage organelles were 27,600.)

CLEZARDIN ET AL.

1064

unactivated platelets solubilized with 2% (mass per vol) SDS. Experiments conducted with platelets from seven different normal donors indicated that the platelet osteonectin content was ranged from 0.7 to 1.1 ng per lob platelets (16,000-23,000 molecules per platelet) (Table 1). Gray platelets had a markedly reduced level of osteonectin (20% of normal; Table 1).

A

B

nn

r

Osteonectin+ 20.1 1

Binding of an antiosteonectin polyclonal antibody to thrombin-stimulated normal and gray platelets From a general point of view, experiments designed to study the localization of proteins to the platelet surface are usually performed with platelets stimulated with thrombin at room temperature under nonstirring conditions. (2-4,41,42' Under such experimental conditions, secretion of platelet a-granule proteins occurs without subsequent aggregation of stimulated platelet^.(^-^.^^^^^) On the other hand, stimulation of platelets with thrombin at 37°C under stirring conditions induces the secretion of a-granule proteins and irreversible platelet aggregation as observed in aggregometer studies (for a general review see Ref. 39). The localization of osteonectin to the surface of stimulated platelets was assessed using radioiodinated antiosteonectin polyclonal antibody 2.2, which has been shown previously to immunoprecipitate from platelet extracts a single protein with a molecular mass identical to that of platelet o s t e ~ n e c t i n . (In ~ ~addition ) to immunoprecipitation experiments, antiosteonectin polyclonal antibody 2.2 did not show any cross-reactivity with fibrinogen, fibronectin, or thrombospondin using an enzyme-linked immunosorbent assay or with collagen using a solid-phase radioimmunoassay (results not shown). Although resting normal platelets did not bind [llsl]antiosteonectin polyclonal antibody 2.2 (results not shown), thrombin stimulation of normal platelets induced specific binding of radioiodinated polyclonal antibody 2.2 on the surface of these cells (Fig. 4). Quantitation by Scatchard analysis of the binding data obtained with five different donors indicated that there are 1710 f 453 antibody binding sites per platelet (Fig. 4 inset and Table 2). Using an antiosteonectin mouse monoclonal antibody (ON6)(2s)instead of a polyclonal antibody, similar results were obtained regarding the binding of radioiodinated ON6 to thrombin-stimulated platelets compared to binding of radioiodinated polyclonal antibody 2.2 (results not shown). Assuming a stoichiometric relationship of antibody to antigen of 1:1, at least 1700 molecules of osteonectin may be expressed on the surface of an activated platelet. The Kd of the antibody was estimated as 1 pM. Because of the low binding of antiosteonectin polyclonal antibody 2.2 to thrombin-stimulated platelets, parallel binding experiments were performed with [125]antithrombospondinmonoclonal antibody P 10 using thrombin-stimulated platelets from the same donors. The results obtained indicated that 12,014 f 1873 P10 molecules were bound per platelet (Table 2). This is in agreement with previous v a l ~ e s , ( ~ "and - ~ ~these ) results

r

14.6

1 2

tn

1 2

FIG. 3. Qualitative assessment of osteonectin in normal and gray platelets. SDS-solubilized platelet proteins (80 pg) from a control subject and a patient with the gray platelet syndrome were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and incubated with [12SI]antio~te~nectin monoclonal antibody ON3. An autoradiograph of the immunoblot performed with control platelets (A) showed an immunoreactive band at a molecular mass of 45 kD under reducing conditions (lane 1) and of 39 kD under nonreducing conditions (lane 2). The apparent molecular mass of this band under both reduced and nonreduced conditions is consistent with that of human platelet osteonectin.12s~2b1 In contrast, no band corresponding to platelet osteonectin was seen with gray platelets (B) under reducing (lane 1) or nonreducing conditions (lane 2).

TABLE1. RADIOIMMUNOASSAY FOR OSTEONECTIN IN PLATELETS OF A PATIENT WITH THE GRAYPLATELET SYNDROME (GPS) COMPARED TO NORMAL PLATELETSa

Osteonectin content Donor Control

Mean GPS

(ng per lobplatelets)

LM BF FD FT AT J-FK PK f

0.8 0.7 0.8 0.9 1.1 1.1 1.1 0.93 f 0.16

SD A-MK

0.2

f

0.03b

aThe total osteonectin content of normal and gray platelets was quantitated by radioimmun~assay'~~~ using washed, unactivated platelets solubilized with 2% (mass/vol) SDs. The radioimmunoassay used a specific rabbit antiserum raised against purified bovine bone osteonectin that cross-reacts with human platelet osteonectin. bThree determinations were made for osteonectin concentration in GPS platelets.

OSTEONECTIN IN PLATELETS

1065

indicate that the activated state of platelets was normal. To nal antibody 2.2 correlates with the presence of osteonecevaluate the specificity of the [12Sl]antio~te~nectin polyclo- tin to the surface of thrombin-stimulated platelets. The nal antibody binding to normal thrombin-stimulated plate- value obtained with antithrombospondin monoclonal antilets, binding experiments were performed with gray plate- body P10 (1,856 molecules per platelet) is in agreement lets, which contain only 20% (i.e., 0.2 ng per lo6 platelets) with the binding of antithrombospondin monoclonal antiof the normal platelet osteonectin content as judged by ra- body 5Gll to thrombin-activated platelets of two unre~ - patients with the gray platelet syndrome.(42) dioimmunoassay (Table 1). No binding of [ L 2 S I ] a n t i o ~ t elated nectin polyclonal antibody 2.2 was observed on the surface of resting and thrombin-stimulated gray platelets, but a Effect of antiosteonectin F(ab)I2fragments slight binding was observed with [lZSI]antithrombospondinon platelet aggregation monoclonal antibody PI0 (Table 2). These results thereFrom a general point of view, platelets can be stimulated polyclofore indicate that binding of [12sI]antioste~ne~tin by a wide variety of agonists, including ADP, collagen, and thrombin, and respond by changing shape from a diskoid to a spherical form with extended pseudopodia, by sticking to one another ( a g g r e g a t i ~ n ) . ' ~As ~ ' delineated by aggregometer studies, platelet aggregation can be viewed as occurring in two phases. The primary phase of aggregation reflects the initial response of the stimulated platelet associated with fibrinogen receptor induction and fibrinogen binding. However, this primary phase of platelet aggregation is reversible; for example, ADP stimulation of washed platelets in the presence of fibrinogen resulted in platelet aggregation and deaggregation because of the lack of platelet Thus, the binding of fibrinogen to the platelet surface is a necessary but insufficient condition for the platelet aggregation process. O n the other hand, the reversible primary phase of aggregation of washed platelets is followed by the secondary phase of aggregation when stronger agonists, such as collagen and thrombin, are used. This secondary phase of aggregation involves the secretion and the participation of a-granule proteins (throm10 50 100 bospondin, fibronectin, and von Willebrand factor) in 2.2 Input Concentration [nM] membrane-mediated events, making platelet aggregation FIG. 4. Binding of osteonectin to thrombin-stimulated irreversible (for a general review see Ref. 39). F(ab), fragments from antiosteonectin polyclonal antinormal platelets. The localization of osteonectin on the surface of thrombin-stimulated platelets was assessed using body 2.2 had no effect on the aggregation of washed plate[12SI]antio~te~nectin polyclonal antibody 2.2. The experi- lets induced by ADP (2 pM; reversible aggregation) and mental procedure was as described in Table 2. thrombin (0.06 U/ml; irreversible aggregation; Table 3).

U

TABLE 2. BINDINGOF ANTIOSTEONECTIN POLYCLONAL ANTIBODY 2.2 AND ANTITHROMBOSPONDIN MONOCLONAL ANTIBODY P 10 TO THROMBIN-STIMULATED NORMAL AND GRAY PLATELETS~

Normal platelets Antibody-bound (molecules or platelet)

A

B

c

D

E

Mean

2.2 PI0

1,742 11,444

950 10,439

2,100 10,840

1,750 12,200

2.008 15,148

12,014

* SD 1,710 * 453 +

1873

Gray platelets, A-MK N Db 1856

aBinding experiments were performed with platelets isolated from five different volunteers (A-E) and one patient (A-MK) with the gray platelet syndrome. The expression of osteonectin and thrombospondin to the surface of activated platelets was studied in parallel for each donor. Washed platelets (2 x lo8 per ml) were stimulated with thrombin (0.72 U/ml) at room temperature under nonstirring conditions. Thrombin was inactivated 10 minutes later by the addition of PPACK M). Binding was then initiated by the addition of ['2sI]antiosteonectin polyclonal antibody 2.2 or ["sI]antithrombospondin monoclonal antibody PI0 to final concentrations ranging from 13 to 200 nM. After 30 minutes of incubation at room temperature, aliquots were layered over 20% sucrose and the platelets were sedimented. The radioactivity associated with the platelet pellets was quantitated in a gamma counter. Nonspecific binding was determined using a 100-fold excess of unlabeled antibody, and this value was subtracted from the total radioactivity to yield the specific binding of 'l51-antibodies to platelet pellets. bNot detected.

CLEZARDIN ET AL.

1066

On the other hand, antiosteonectin F(ab)z fragments (80 pg/ml) significantly inhibited the collagen-induced aggregation of washed platelets (2 x lo8 per ml; irreversible aggregation) but control studies using nonimmune F(ab), fragments did not show any inhibition at similar concentrations (Fig. 5). Decreasing the concentration of antiosteonectin F(ab‘), fragments from 80 to 40 pg/ml caused a decreased inhibition (Table 3), indicating that the inhibitory effect was dose dependent. Inhibition of aggregation by antiosteonectin F(ab), fragments occurred without af-

fecting the collagen-induced platelet serotonin release (Fig. 5), indicating that the inhibitory effect did not result from

a direct inhibition of platelet secretion. However, the inhibitory effect of antiosteonectin F(ab), fragments was related to the secretion-dependent phase of the platelet aggregation process since these F(ab’), fragments (80 pg/ml) no longer had any effect when the platelet release reaction (and the subsequent extent of platelet aggregation) was increased using either a higher dose of collagen (2 pg/ml) or a higher concentration of platelets (5 x 10” per ml; Table

TABLE3. EFFECTOF ANTIOSTEONECTIN F(AB’),FRAGMENTS ON COLLAGEN-, ADP-, THROMBIN-INDUCED AGGREGATION OF WASHED PLATELETS~ Platelet concentration ~

Agonist

AND

A n tibody concentration (dml)

Platelet aggregation (%)

0

45 35 5 55 52 32 33 55 55 62 58

~~

2 x lo8 per ml

Collagen (1 pg/ml

40 80

Collagen (2 pg/ml)

0 80

ADP (2 pM)

0 80

Thrombin (0.06 U/ml) 5 x lo* per ml

Collagen (1 pg/ml)

0 80 0 80

apreparation of washed platelets in Tyrode’s buffer was performed according to the technique of Mustard et al.lJ7)The effect of the antiosteonectin polyclonal antibody on platelet aggregation was measured by preincubating washed platelets with the indicated concentrations of F(ab’), fragments for 2 minutes at 37°C before the induction of aggregation. Control experiments were performed with nonimmune F(ab’), fragments. The aggregation pattern was recorded in an aggregometer at 37°C under continuous stirring. Tyrode’s buffer was used to calibrate light transmission at 100%. The platelet aggregation response (%) was measured as the peak height of the aggregation curve. Values obtained are representative of experiments conducted with five different donors.

Fab’2

Collagen

I

L

Ant I -0steonect In F b b ’ ?

(2511

FIG. 5. Effect of F(ab), fragments from antiosteonectin polyclonal antibody 2.2 on collagen-induced aggregation of washed platelets. Washed platelets (2 x lo8 per ml) in Tyrode’s buffer were incubated with antiosteonectin or nonimmune F(ab‘), fragments (80 pg/ml) for 2 minutes at 37°C before the induction of aggregation by collagen (1 pg/mI). The aggregation pattern was recorded in an aggregometer at 37°C under continuous stirring. Secretion of [“C]serotonin is indicated in parentheses.

1067

OSTEONECTIN IN PLATELETS 3). In addition, antiosteonectin F(ab), fragments did not have any effect on the ADP-induced aggregation of washed platelets (Table 3), in which the platelet release reaction does not occur.(43)

Nonimmune F(ab’),

Effect of antiosteonectin F(ab’), fragments on the binding of endogenous thrombospondin to thrombin-activated platelets Since osteonectin and thrombospondin are present as a complex when secreted from thrombin-stimulated platelets‘LP) and osteonectin, like t h r o m b o ~ p o n d i n , ‘4 ~ ~~ 2 ) , binds specifically on the surface of thrombin-activated platelets, we determined whether the inhibitory effect of antiosteonectin F(ab), fragments on platelet aggregation was mediated by interference with platelet thrombospondin binding. F(ab), fragments from antiosteonectin polyclonal antibody 2.2 significantly inhibited the specific binding of radioiodinated antithrombospondin monoclonal antibody P 10 to thrombin-stimulated platelets but, using nonimmune F(ab), fragments, no effect was observed at similar concentrations (Fig. 6). Scatchard plot analysis of binding experiments performed with platelets from four different healthy. donors indicated that, in the presence of nonimmune f(ab), fragments, thrombin stimulation induced the binding of 11,230 f 766 [lzsI]P1]molecules per platelet, with an apparent dissociation constant Kd of 0.45 X lo-’ M (Table 4). This is in agreement with previously published r e s ~ l t s . ( ~On . ~ *the ’ other hand, despite a Kd of 0.35 f lo-’ M, only 6286 f 2065 [1z51)P10 molecules per platelet were bound in the presence of antiosteonectin F(ab), fragments (Table 4). Using an ELISA, binding of monoclonal antibody P10 to thrombospondin-osteonectin-coated wells was assessed in the presence of antiosteonectin polyclonal antibody 2.2 (Fig. 7). Monoclonal antibody P10, in the presence of antibody 2.2, bound to the same extent of thrombospondin(Fig. 7A) and thrombospondin-osteonectin-coated wells (Fig. 7B). Furthermore, in the presence of antibody P10, binding of antiosteonectin polyclonal antibody 2.2 to thrombospondin-osteonectin- (Fig. 7C) and osteonectincoated wells was similar (Fig. 7D). Such results indicate therefore that anti-osteonectin antibody 2.2 did not interfere by steric hindrance with the binding of P10 to solidphase adsorbed and platelet-bound thrombospondin.

DISCUSSION In this study, using electron microscopy in combination with immunogold staining, we have demonstrated that osteonectin is located within the major storage organelle for platelet-secreted proteins, the a-granules. In addition, we have shown by both western blot and radioimmunoassay techniques that platelets of a patient fulfilling the previously described criteria for the gray platelet synd r ~ m e ( * ~possess - ~ ~ ~a ~severe ~ ) deficiency of osteonectin (20% of normal) together with an apparent absence of thrombospondin and fibrinogen. The absence of thrombospondin and fibrinogen, as well as the markedly reduced

P 50

100

150

200 PI0 Input C once n t r a t ion (nM)

FIG. 6. Effect of F(ab), fragments from antiosteonectin polyclonal antibody 2.2 on expression of endogenous thrombospondin to thrombin-stimulated normal platelets. The expression of endogenous thrombospondin on the surface of thrombin-stimulated platelets was assessed using [lzSI]antithrombospondinmonoclonal antibody P10. The experimental procedure was as described in Table 4.

level of other a-granule proteins, including fibronectin, platelet factor 4, P-thromboglobulin, von Willebrand factor, and platelet-derived growth factor, has already been documented in other patients with the gray platelet syndrome.(29-31.42) The severe deficiency of osteonectin in gray platelets therefore confirms that a-granules are a storage organelle for osteonectin. Using monoclonal antibovine bone osteonectin antibodies together with bovine bone osteonectin standards, Stenner et al. ( 2 4 ) recently developed a radioimmunoassay to measure the concentration of osteonectin in serum, plasma, and platelet extracts. They reported a mean value of 9.5 ng per lo6 platelets, which is 10 times higher than what we observed in this study using a polyclonal antibovine bone osteonectin antibody-based radioimmunoassay.(38) Because human platelet and bone osteonectin are ~ t r u ~ t u r a l and l y ~immunologically ~~~~~~ different,(25)discrepancies in platelet osteonectin values that have been reported by Stenner et aLtZ4)and by us (this study) may reflect species- and/or sources-specific differences in the cross-reactivity of the antibovine bone osteonectin antibody used. Using such a radioimmunoassay, measurement of circulating levels of osteonectin in serum of normal subjects and of patients with thrombocytopenia shows a logarithmic positive correlation with platelet count, suggesting that

CLEZARDIN ET AL.

1068

TABLE4. QUANTITATIVE EFFECTOF ANTIOSTEONECTIN F(AB')z ON EXPRESSION OF ENDOGENOUS THROMBOSPONDIN ON FRAGMENTS THE SURFACE OF THROMBIN-STIMULATED PLATELETS~

Donor A A B B C C D D

Antibodyb NI Anti-ON NI Anti-ON NI Anti-ON NI Anti-ON

Dissociation constant (x 10- M)

Molecules bound per platelet

0.4 0.6

11,444 8,059 10,439 3,980

0.7 0.2 0.2 0.2 0.5

0.4

10.840 5,105

12,200 8,OoO

aThe effect of nonimmune or antiosteonectin F(ab'), fragments on the expression of endogenous thrombospondin to the surface of activated platelets was studied in parallel for each donor. Nonimmune or antiosteonectin F(ab'), fragments, at a final concentration of 60 pg/ml, were allowed to incubate with washed platelets (2 x lo8 per ml) for 5 minutes before the addition of thrombin (0.72 U/ml). After neutralization of thrombin by PPACK (10.' M), ["'l]antithrombospondin monoclonal antibody PI0 was added in increasing amounts to thrombin-stimulated platelets until saturation was reached. The remaining steps were performed as described in Table 2. The mean platelet binding of ["'I]monoclonal antibody PI0 in the presence of antiosteonectin and nonimmune F(ab), fragments was 6286 f 2065 and 11,230 f 766 molecules per platelet, respectively. Significant differences in [1251]monoclonalantibody PI0 binding were observed between samples when analyzed by Student's paired f-test (P < 0.01). The dissociation constant Kd was not significantly affected (0.35 f 0.19 x lo-' versus 0.45 f 0.2 x lo-' M). bNI, the binding of ["Sl]antithrombospondin monoclonal antibody P I 0 in the presence of nonimmune F(ab), fragments; Anti-ON, binding was performed in the presence of antiosteonectin F(ab), fragments.

some trapping of platelet osteonectin occurred either in the clot or on the surface of activated platelets.(18' These observations prompted us to examine whether osteonectin was expressed on the surface of activated platelets. We showed that a radiolabeled antiosteonectin polyclonal antibody binds specifically to thrombin-activated platelets but not to resting platelets. Binding is concentration dependent, saturable, and inhibited with an excess of cold antiosteonectin polyclonal antibody. Assuming a stoichiometric relationship of antibody to antigen of 1 :1, at least 1700 molecules of osteonectin may be bound on the surface of an activated platelet, which is approximately 10% of the total platelet osteonectin content. Kelm and Mann{l6)recently published similar observations regarding the surface localization of osteonectin to thrombin-activated platelets and to resting platelets. Using an antibovine bone osteonectin monoclonal antibody, they reported that there are 2200 molecules of osteonectin exposed on an activated platelet as opposed to resting platelets, where osteonectin is not surface bound.(16)Although the expression of osteonectin on the surface of normal activated platelets is low, the absence of binding of an antiosteonectin polyclonal antibody on the surface of activated gray platelets, which contain only 20% of the normal osteonectin content, further validates the binding experiments obtained with normal platelets. To gain further insights into the role of osteonectin re-

leased from activated platelets, the effect of antiosteonectin F(ab), fragments was tested on aggregation of washed platelets. Antiosteonectin F(ab), fragments significantly inhibit collagen-induced platelet aggregation in a dose-dependent manner, but nonimmune F(ab), fragments have no effect at similar concentrations. The inhibitory effect of antiosteonectin F(ab), fragments on collagen-induced platelet aggregation is not mediated by interference with collagen and does not result from a direct inhibition of platelet secretion. In addition, antiosteonectin F(ab), fragments do not have any effect on ADP-induced aggregation of washed platelets, in which the platelet release reaction does not occur.(43'Taken together, these results strongly suggest that the inhibitory effect of antiosteonectin F(ab), fragments depends upon the secretion-dependent phase of the platelet aggregation process. To characterize the mechanism through which antiosteonectin F(ab), fragments inhibit platelet aggregation, the expression of endogenous thrombospondin on the surface of thrombin-activated platelets was studied using radiolabeled antithrombospondin monoclonal antibody P10. In the present of nonimmune F(ab'), fragments, the apparent K a f monoclonal antibody PI0 is 0.45 x lo-' M, with approximately 11,OoO binding sites per platelet. This is in agreement with previous published v a l ~ e s . ' ~ . 'On ~ ) the other hand, despite a similar Kd (0.35 x 10'' M), only 6OOO binding sites per platelet are present in the presence of antiosteonectin

1069

OSTEONECTIN IN PLATELETS P10

2.2

sults presented in this study, it is conceivable that osteonectin and thrombospondin share similar roles in the platelet aggregation stabilization process. That thrombospondin and osteonectin already share similar properties in modulating the attachment and spreading of cultured cells to extracellular matrix p r o t e i n ~ ( ~further ~ ' ~ ~ ~suggests ~1 that the thrombospondin-osteonectin complex plays an important role in cell-cell and cell-extracellular matrix interactions.

c.

.-

100

:

T

SO

60

7 X

40

ACKNOWLEDGMENTS

F,

9

20 0

A

B

C

D

FIG. 7. Binding of antithrombospondin monoclonal antibody P10 to the thrombospondin-osteonectin complex in the presence of antiosteonectin polyclonal antibody 2.2. Thrombospondin (0.35 pg/ml) and osteonectin (10 pg/ml) in bicarbonate coating buffer were added either separately or together to plastic wells overnight at 4°C. Antithrombospondin monoclonal antibody PI0 (diluted to 1:500,000 in Tris-Tween buffer) was first incubated with antiosteonectin polyclonal antibody 2.2 (diluted to 1:8OOO in TrisTween buffer). After a 5 minute incubation at room temperature, wells were emptied and washed and the P10/2.2 antibody mixture added to thrombospondin- (A), thrombospondin-osteonectin- (B and C), and osteonectin-coated wells (D). After a 1 h incubation at 37"C, the washing procedure was repeated and a goat antimouse IgG (lanes A and B) or a goat antirabbit IgG (lanes C and D) conjugated with horseradish peroxidase was added to each well for an additional 1 h incubation at 37°C. The remaining steps were carried out as described in Materials and Methods.

This study was supported in part by a grant from the Fondation de France, by a grant from the FCdCration Nationale des Centres de Lutte Contre le Cancer, and by Grant 6840 from the Association pour la Recherche contre le Cancer. A report of this study was presented at the Twelfth Annual Meeting of the American Society for Bone and Mineral Research, Atlanta, August 28-31, 1990, and published in abstract form (J Bone Miner Res 55164, 1990). Clezardin was a recipient of a Young Investigator Award for this study.

REFERENCES 1 . Baenziger NL, Brodie GN, Majerus PW 1972 Isolation and

properties of a thrombin-sensitive protein of human platelets. J Biol Chem 247:2723-2731. 2. Wolff R, Plow EF, Ginsberg MH 1986 Interaction of thrombospondin with resting and stimulated human platelets. J Biol Chem 261:6840-6846. 3. Aiken ML, Ginsberg MH, Plow EF 1987 Divalent cation-dependent and independent surface expression of thrombospondin on thrombin-stimulated human platelets. Blood 69: 58-64.

F(ab), fragments. This inhibitory effect of antiosteonectin F(ab), fragments on the surface expression of endogenous thrombospondin is not mediated by interference with binding of monoclonal antibody PI0 to thrombospondin as judged by ELISA. Because thrombospondin and osteonectin are present as a complex when released from thrombinstimulated platelet^,"^) it is therefore conceivable that the binding of endogenous thrombospondin on the surface of activated platelets is in part mediated by osteonectin. In this respect, the localization of osteonectin to the surface of activated platelets could be partially masked by thrombospondin. Steric considerations could therefore limit the availability of osteonectin epitopes and explain why only 1700 molecules of antiosteonectin antibody bound per platelet. In summary, we provided evidences that osteonectin is an a-granule component that, by binding o n the surface of activated platelets, is involved with thrombospondin in the secretion-dependent phase of the platelet aggregation process. It has been proposed that thrombospondin, by reinforcing the strength of interplatelet interactions, could convert the reversible aggregation into an irreversible state.(4)Based on this as well as on the re-

4. Leung LLK 1984 Role of thrombospondin in platelet aggregation. J Clin Invest 74:1764-1772. 5 . Lawler J 1986 The structural and functional properties of thrombospondin. Blood 67:1197-1209. 6 . Silverstein RL, Leung LLK, Nachman RL 1986 Thrombospondin: A versatile multifunctional glycoprotein. Arteriosclerosis 6:245-253. 7. Clezardin P, Jouishomme H, Chavassieux P, Marie PJ 1989 Thrombospondin is synthesized and secreted by human osteoblasts and osteosarcoma cells. A model to study the different effects of thrombospondin in cell adhesion. Eur J Biochem 181:721-726. 8. Gehron-Robey P, Young MR, Fisher LW, McLain TD 1989 Thrombospondin is an osteoblast-derived components of mineralized extracellular matrix. J Cell Biol 108:719-727. 9 . Clezardin P, Hunter NR, Lawler J , Pratt DA, McGregor JL, Pepper DS, Dawes J 1986 Structural and immunological comparison of human thrombospondins isolated from platelets and from culture supernatants of endothelial cells and fibroblasts. Evidence for a thrombospondin polymorphism. Eur J Biochem 159569-579. 10. Dardik R, Lahav J 1987 The structure of endothelial cell thrombospondin. Characterization of the heparin-binding domains. Eur J Biochem 168:347-355. 1 1 . Clezardin P, Bourdillon MC, Hunter NR, McGregor JL 1988 Cell attachment and fibrinogen binding properties of platelet and endothelial cell thrombospondin are not affected by structural differences in the 70 and 18 kDa protease-resistant

1070 domains. FEBS Lett 228:215-218. 12. Dardik R, Lahav J 1989 Multiple domains are involved in the interaction of endothelial cell thrombospondin with fibronectin. Eur J Biochem 185581-588. 13. Roberts DD, Sherwood JA, Ginsburg V 1987 Platelet thrombospondin mediates attachment and spreading of human melanoma cells. J Cell Biol 104:131-139. 14. Murphy-Ullrich JE, Hook M 1989 Thrombospondin modulates focal adhesions in endothelial cells. J Cell Biol 109: 1309-13 19. 15. Taraboletti G, Roberts DD, Liotta LA 1987 Thrombospondin-induced tumor cell migration: Haptotaxis and chemotaxis are mediated by different molecular domains. J Cell Biol 105:2409-2415. 16. Vischer P, Volker W, Schmidt A, Sinclair N 1988 Association of thrombospondin of endothelial cells with other matrix proteins and cell attachment sites and migration tracks. Eur J Cell Biol 47:36-46. 17. Majack RA, Goodman LV, Dixit VM 1988 Cell surface thrombospondin is functionally essential for vascular smooth muscle cell proliferation. J Cell Biol 106:415-422. 18. O'Shea KS, Dixit VM 1988 Unique distribution of the extracellular matrix component thrombospondin in the developing mouse embryo. J Cell Biol 107:2737-2748. 19. Clezardin P, Malaval L, Ehrensperger AS, Delmas PD, Dechavanne M, McGregor JL 1988 Complex formation of human thrombospondin with osteonectin. Eur J Biochem 175:275-284. 20. Termine JD, Kleinman HK, Whitson SW, Conn KM, McGarvey ML, Martin GR 1981 Osteonectin, a bone-specific protein linking mineral to collagen. Cell M:99-105. 21. Romberg RW, Werness PG, Lollar P, Riggs BL, Mann KG 1985 Isolation and characterization of native adult osteonectin. J Biol Chem 260:2728-2736. 22. Sage H, Johnson C, Bornstein P 1984 Characterization of a novel serum albumin-binding glycoprotein secreted by endothelial cells in culture. J Biol Chem 259:3993-4007. 23, Wasi S, Otsuka K, Yao KL, Tung PS, Aubin JE, Sodek J, Termine JD 1984 An osteonectin-like protein in porcine periodontal ligaments and its synthesis by periodontal ligament fibroblasts. Can J Biochem Cell Biol 62:470-478. 24. Stenner DD, Tracy RP, Riggs BL, Mann KG 1986 Human platelets contain and secrete osteonectin, a major protein of mineralized bone. Proc Natl Acad Sci USA 83:6892-6896. 25. Malaval L, Darbouret B, Preaudat C, Jolu JP, Delmas PD 1990 Monoclonal antibody distinguishes between bone and platelet osteonectin. J Bone Miner Res (in press). 26. Kelm RJ, Mann KG 1990 Human platelet osteonectin: Release, surface expression, and partial characterization. Blood 75: 1105-1 1 13. 27. Sage H, Vernon RB, Funk SE, Everitt EA, Angello J 1989 SPARC, a secreted protein associated with cellular proliferation, inhibits cell spreading in vitro and exhibits Ca"-dependent binding to the extracellular matrix. J Cell Biol 109:341356. 28. Malaval L, Ffrench M, Delmas PD 1990 Circulating levels of osteonectin in normal subjects and patients with thrombocytopenia. Bone Miner 9:129-136. 29. White JG 1979 Ultrastructural studies of the gray platelet syndrome. Am J Pathol 95445-462. 30. Gerrard JM, Phillips DR, Rao GHR, Plow EF, Walz DA, Harker A, White JG 1980 Biochemical studies of two patients with the gray platelet syndrome. Selective deficiency of platelet alpha granules. J Clin Invest 66:102-109.

CLEZARDIN ET AL. 31. Nurden AT, Kunicki TJ, Dupuis D, Soria C, Caen J P 1982

32.

33.

34.

35.

36.

Specific protein and glycoprotein deficiencies in platelets isolated from two patients with the ray platelet syndrome. Blood 59:709-718. Clezardin P, McGregor JL, Nanach M, Robert F, Dechavanne M, Clemetson KJ 1984 Isolation of thrombospondin released from thrombin-stimulated human platelets by fast protein liquid chromatography on an anion-exchange Mono Q column. J Chromatogr 2%:249-256. Kaplan C, Champeix P, Morel MC, Lecompte T, Ferrer-leCoeur F, Samama M 1988 Grey platelet syndrome with GPIIIa abnormality: A case report (abstract). XXII Congress of the International Society of Hematology, 278. Cramer E, Pryzwansky KB, Villeval JL, Testa U, BretonGorius J 1985 Ultrastructural localization of lactoferrin and myeloperoxidase in human neutrophils by immunogold. Blood 65~423-432. Cramer E, Meyer D, Le Menn R, Breton-Gorius J 1985 Excentric localization of von Willebrand factor in an internal structure of platelet a-granule resembling that of WeibelPalade bodies. Blood 66:710-713. Clezardin P, McGregor JL, Lyon M, Clemetson KJ. Huppert J 1986 Characterization of two murine monoclonal antibodies (PIO, P12) directed against different determinants on human blood platelet thrombospondin. Eur J Biochem 154:

95-204. 37. Mustard JF, Perry DW, Ardlie NG, Packham MA 1972

38.

39.

40.

41.

42.

43.

Preparation of suspensions of washed platelets from humans. Br J Haematol 22:193-204. Malaval L, Fournier B, Delmas PD 1987 Radioimmunoassay for osteonectin. Concentrations in bone, nonmineralized tissues, and blood. J Bone Miner Res 2:457-465. Nachman RL, Leung LLK, Polley MJ 1984 Molecular mechanisms of platelet adhesion and platelet aggregation. In: George JN, Nurden AT, Phillips DR (eds.) Platelet Membrane Glycoproteins. Plenum Press, New York, pp. 245-257. Valdorf-Hansen JF, Zucker MB 1971 Effect of temperature and inhibitors on serotonin-["C] release from human platelets. Am J Physiol 220:105-109. Boukerche H, McGregor JL 1988 Characterization of an anti-thrombospondin monoclonal antibody (P8) that inhibits human blood platelet functions. Normal binding of P8 to thrombin-activated Glanzmann thrombasthenic platelets. Eur J Biochem 171:383-392. Legrand C, Dubernard V, Kieffer N, Nurden AT 1988 Use of a monoclonal antibody to measure the surface expression of thrombospondin following platelet activation. Eur J Biochem 171:393-399. Mustard JF, Perry DW, Kinlough-Rathbone RL, Packham MA 1975 Factors responsible for ADP-induced release reaction of human platelets. Am J Physiol 228:1757-1765.

Address reprint requests to: Dr. P. Clezardin INSERM U 234 Laboratoire de Biochimie des Protdines Osseuses Hepita1 Edouard Herriot 3, Place d'Arsonval 69437 Lyon Cpdex 03, France Received for publication on October 19, 1990; in revised form on January 21, 1991; and accepted on March 15, 1991.