British Journal uf Haemafolugy, 1979, 42, I 3 7-1 45.
Thrombin Receptors of Human Platelets: Thrombin Binding and Antithrombin Properties of Glycoprotein I P. GANCULY A N D N. L. GOULD
Laboratory of Hematology, St Jude Children’s Research Hospital, Memphis, Tennessee (Received20June 1978; acceptedforpublication 16 August 1978) SUMMARY. Washed human platelets were solubilized and the proteins were separated by preparative gel electrophoresis in the presence of sodium dodecyl sulphate. The gel was cut into slices and the effect of the eluted proteins on the clotting of fibrinogen by thrombin was evaluated. The isolate from only one gel slice strongly inhibited the clotting of fibrinogen. The prolongation of the clotting time was dependent on the concentration of the protein and reached a plateau around 5 p g . Gel electrophoresis of this isolate showed a prominent glycoprotein with an apparent M,= 150000. Gel filtration studies with [1251]thrombinshowed that the protein isolate bound a significant amount of thrombin which could be displaced with unlabelled thrombin. Another preparation from the same gel or purified y-globulin did not bind thrombin or prolong the clotting time of fibrinogen. Glycoprotein I was isolated from human platelets by affinity chromatography on lectin-Sepharose columns. The isolated glycoprotein prolonged the clotting of fibrinogen and bound [ 1251]thrombin which could be displaced by unlabelled thrombin. It is proposed that the high affinity receptor of thrombin on human platelets is glycoprotein I. In addition, the antithrombin activity of intact platelets is due to binding of thrombin to this glycoprotein. Human platelets treated with thrombin undergo a series of morphological and biochemical changes. The stimulated platelets rapidly secrete calcium, serotonin, adenine nucleotides and other substancesand then slowly aggregate. Thrombin causes these changes in human platelets a t concentrations less than I nM (0.I U/ml) suggesting that thrombin might be an important mediator of platelet function in haemostasis and thrombosis (for review, see Marcus, 1969). Recently, it has been shown that the initial step in the interaction of thrombin with platelets is the binding of the enzyme to specific structures on the platelet surface (Ganguly, 1974; Tollefsen et al, 1974;Martin et all 1975; Ganguly & Sonnichsen, 1976; Mohammed et all 1976). Evidence has also been presented which indicates that binding of thrombin is an important step in the physiological function of platelets (Tollefsen et a l , 1974; Ganguly & Sonnichsen, 1976; Workman et all 1977). To elucidate further the mechanism of action of thrombin on platelets, it is essential to determine the identity of the receptor of thrombin. In this paper we present direct evidence that a glycoprotein, designated as glycoprotein I (GP-I), binds thrombin Correspondence: Dr P. Ganguly, Laboratory of Hematology, St Jude Children’s Research Hospital, P.O. Box 318, Memphis, Tennessee 38101, U.S.A.
0007-1048/79/0500-o13 7$02.00
01979 Blackwell Scientific Publications I37
138
P . Ganguly and N. L. Could
specifically and reversibly. We propose that GP-I is the high affinity receptor of thrombin and the antithrombin activity in human platelets is due to binding of thrombin to this surface component. METHODS
Platelets. Blood was collected from volunteers of both sexes in plastic bags o r syringes utilizing 3.8% sodium citrate as the anticoagulant (9:1 v/v). The red cells were removed by centrifugation and the platelets were washed as described previously (Ganguly & Sonnichsen, 1976).
Electrophorrrir. The washed platelets were solubilized in 3 Yo sodium dodecyl sulphate (SDS)-I% P-mercaptoethanol and heated in a boiling water bath for 5 min. Preparative electrophoresis was carried out in 7.5% acrylamide gels in the presence of 0.1 YOSDS in a 32 mm x 6 0 mm cylindrical glass apparatus (Kontes, Vineland, N.J.) a t 3 0 mA for 40 h. After electrophoresis, a longitudinal slice was cut from the gel and stained to locate the bands. The gel was cut into 5 mrn sections, ground in a blender in 0. I M Tris-o. I M NaC1, pH 7.5 and centrifuged a t 1 5 000 rpm for 30 min. The supernatants were concentrated by ultrafiltration through a XM-50 membrane and dialysed for 48 h against Tris buffer with several changes. The effect of these crude protein preparations was then tested on the clotting of fibrinogen by thrombin. Analysis ofthe different protein fractions were carried out in slab gels by the method of Laemmli ( I 970). Thrombin. Commercial bovine thrombin (Parke Davis, Detroit, Mich.) was purified by ion exchange chromatography and labelled with 1251 by the chloramine T method. Excess free label was removed by gel filtration. Details of these procedures have been reported (Tollefsen el al, 1974; Ganguly & Sonnichsen, 1976). The labelled enzyme was first used for binding studies with intact platelets. The thrombin preparations utilized in this study showed typical binding characteristics (Tollefsen et al, 1974; Ganguly 8: Sonnichsen, 1976). Fibrinogen clotting. Bovine fibrinogen was purchased from Kabi (Stockholm, Sweden) and was 90% clottablc. 100 pl of the test material was incubated with IOOpl of thrombin (2-4 u/ml) for 1 5 min a t room temperature. 200 pl of a 0.5% fibrinogen solution was then rapidly added and the clotting time measured in duplicate. Geljltration. Columns of Sephadex G-200 or Sepharose 613 were prepared according to tho instructions of the manufacturer (Pharmacia, Piscataway, N.J.). Two matched columns packed with Sepharose 6 U were used for the determination of thrombin binding. The test material was incubated with 50 mu of [ i251]thrombinfor 15 min a t room temperature and thcn applied to one column. Control studies with labelled thrombin alone or in combination with other proteins were carried out at the same time on the other column. Samples were eluted with 0. I M Tris-0.1 M NaC1, pH 7.5, containing o. I O h serum albumin. Labelling ofplatelets. This procedure, which is specific for glycoproteins and glycolipids, was carried out according to the general method of Gahmberg & Hakomori (1973). Washed platelets (2-3 x Io8/ml) were incubated with 10 u/ml of neuraminidase for 1 5 min followed by 5 u/ml of galactose oxidase for 60 min and finally I mCi of borotritide for 10 min. The platelets were collected by centrifugation and dissolved as described under electrophoresis. We have shown that in this method most of the radioactivity is incorporated in GP-I (Sidhu & Ganguly, 1978).
Platelet Thrombin Receptors
I39
A f i n i t y chromatography. This procedure was carried out as described (Kahane et al, 1976; Nachman et al, I977b). RESULTS In this study we have used preparative gel electrophoresis for the separation of platelet proteins. This method requires small amounts of platelets and provides high resolution. The effect of the protein eluates from the gel slices on the clotting of fibrinogen by thrombin is shown in Fig I . The clotting of fibrinogen was strongly inhibited by the eluate from a single gel slice. Similar results were obtained from five different experiments and shows the specificity of the I
2
I
I
4
6
I
I
8
1
1
0
SLICE NUMBER FIG I . Effect of human platelet proteins on the clotting of fibrinogen ( 5 mg/ml) by thrombin (4 u/ml). Platelet proteins were separated by preparative gel electrophoresis in the presence of SDS. The gel was cut into 5 m m sections and the protein was eluted out by grinding the gels. The eluates were centrifuged at 1 5 ooo rpm for 3 0 min, concentrated by ultrafiltration and dialysed extensively against Tris-saline, pH 7.2. 100 p1 of thrombin was incubated with 100p1of each eluate for 1 5 min. 200 pl of fibrinogen was then added and the clotting time was determined in duplicate. The inset shows the protein staining pattern of a longitudinal slice of the preparative gel.
inhibitory reaction. The protein concentration (A280",,J of the eluate from the fifth gel slice was lower than those from slices 6-1 I . Thus, the prolongation in the fibrinogen clotting time is not due to a higher protein concentration in the eluate from this particular slice. The inhibitory property remained unchanged after extensive dialysis. The gel filtration pattern of the isolate from the fifth slice through a column of Sephadex G-zoo showed two peaks (Fig 2). The first peak which eluted near the void volume inhibited clotting of fibrinogen. These results show that the inhibitory activity was associated with a macromolecular component of platelets. The fractions of the first peak were pooled and centrifuged at 1 5 000 rpm for 30 min. Assuming an extinction coefficient of 10, the protein concentration was estimated to be 200 &ml. This
P . Canguly and N.L. Could I
I
1
1
I
I
1
I
I
Fraction number FIG2. Gel filtration pattern of the platelet protein through a Sephadex G-200 column (2.5 cm x 37 cm). Whole platelets were solubilized and the constituent proteins separated by preparative gel electrophoresis. 4 ml of protein isolated from the fifth gel slice which showed maximum inhibition of fibrinogen clotting was applied to the column and eluted with 0.1 M Tris-o.1 M NaCI, pH 7.5. Fraction volume 2.0 ml, flow rate 10 ml/h.
value is likely to be higher than the true value because of aggregation of the protein and consequent loss of light due to scattering (see later). The variation in the clotting time of fibrinogen by thrombin with different amounts of the platelet protein is shown in Fig 3 . The clotting time increased progressively reaching a maximum around 5 ,ug of protein. With further addition of the protein, there was a slow decrease in the degree of inhibition of clotting suggesting that the protein may be a restricted
- 60 u
-w k$
- 50
I-
W
Z 40 II-
s 30 0
1
1
2
I
I
I
I
J
4 6 8 1 0 PROTEIN ADDED ( p g )
FIG 3. Effect of varying amounts of platelet protein after gel filtration on the clotting time of fibrinogen by thrombin. Proteins of solubilized whole platelets were separated by electrophoresis and the eluate from the fifth gel slice was fractionated on a Sephadex G-zoo column. The fractions of the first peak were pooled and the effect of this protein o n the clotting time of fibrinogen was determined as in Fig I . 0 , Different amounts of protein; 0,equal amounts of elution buffer control.
Platelet Thrombin Receptors 141 inhibitor. As little as 4 pg of the protein doubled the clotting time of fibrinogen showing a highly specific reaction. These results show that the isolated platelet protein is an antithrombin. The platelet protein isolate was analysed by SDS gel electrophoresis and stained with PAS reagent. Besides some material near the origin, a single band, corresponding to an apparent mol wt of 150000, was observed. Staining for protein revealed two additional faint bands. This material did not react with antiserum to whole serum, fibrinogen or fibrin stabilizing factor. The elution pattern of [1251]thrombinthrough the Sepharose 6B column showed a single peak around fraction 70 (Fig 4a). Some of the thrombin applied binds to the gel. This binding of thrombin is reduced by including albumin in the elution buffer. In this experiment about 80% of the counts applied were recovered. The elution pattern of the protein isolate incubated with [ l251]thrombin showed three peaks of radioactivity (Fig 4b). The material eluting near fraction 70 (peak 111) was due to free [1251]thrombin.Based on the area of the peaks, approximately 70% of the thrombin added was bound and 30% remained free. O f the total thrombin bound, 40% and 60% were associated with peak I and peak I1 respectively. When the same protein preparation was analysed after 24 h at 4%, the amount ofradioactivity associated with the first peak increased considerably. The total amount of thrombin bound to both peaks I and 11 remained essentially the same as in Fig 4(b). However, the amount of radioactivity associated with peak I increased from 40% to 57%. With further storage, the sample became visibly turbid. Thus, it appears reasonable to suggest that the first peak is due to aggregated protein while the second peak, eluting in a position similar to that of y-globulin (m.w. 160 ooo), corresponds to monomeric protein. The fractions (48-60) ofthe second peak (Fig 4b) were pooled and concentrated by ultrafiltration through a XM-so membrane. An aliquot of this material analysed through the Sepharose 6B column showed one peak at the original position. The concentrated material was then incubated with excess unlabelled thrombin and analysed through the column. A substantial amount of the bound [ 1251]thrombin eluted around the position of the free enzyme indicating displacement by the unlabelled thrombin (Fig 4c). As determined from the area of the peaks, about 65% of the thrombin was in the free form. When the platelet protein was incubated with [1251]thrombinin the presence of excess unlabelled thrombin, a reduction in the binding of [ 1251]thrombinsimilar to that shown in Fig 4(c) was observed. To determine further the specificity of thrombin-binding to GP-I, the contents of the fourth and sixth gel slices were recovered. These materials did not prolong the clotting time of fibrinogen by thrombin. In gel filtration studies, these preparations bound about 15% of the [1251]thrombinadded which could not be displaced with excess unlabelled thrombin. Further, y-globulin with a molecular weight similar to the thrombin binding protein neither inhibited fibrinogen clotting nor bound thrombin. The results presented above show that a protein fraction of human platelets which we isolated by gel electrophoresis contains a significant amount of antithrombin activity. This same protien fraction also binds thrombin specifically and reversibly. The major constituent of this protein isolate was a glycoprotein of apparent mol wt I 5 5 000. These observations suggest that GP-I in human platelets may be the receptor of thrombin. T o determine this point, GP-I was isolated by affinity chromatography through columns of wheat germ agglutinin linked to Sepharose 4B (Fig 5 ) . The bound material eluted with N-acetyl-wglucosamine showed a
P. Ganguly and N.L. Could
142 18
-
16
-
12
-
N
0 x
k
0
84 -
Fraction number
Fraction number I
20
30
40
50 60 70 Fraction number
80
90
FIG4.Gel filtration studies on the binding of [ 1251]thrombinto the platelet protein isolate. 50 p1 of thrombin (so mu) containing approximately 20 ooo cpm was incubated at room temperature for 15 min with I ml of the protein solution (zoo pg, c I .3 nmol). The sample was applied to a Sepharose 6B column (1.2 cm x 20 cm) and eluted with 0.1 M Tris-o.1 M NaC1, pH 7.5, containing 0.1% albumin. The flow rate was maintained at 6 ml/h with a pump and fractions of 0.6 ml each was collected in plastic cups. (a) Thrombin+buffer control, (b) thrombin+protein, (c) fractions (48-60) of pattern (b) were pooled and concentrated by ultrafiltration. This material containing 3000 cpm was incubated with 10 u of unlabelled thrombin for 1 5 min and then applied to the column. In this pattern both peaks eluted slightly earlier presumably due to sinking of the gel.
prominent peak of radioactivity corresponding to a glycoprotein of M,= 150ooo (GP-I). A minor component of M,=220 ooo was also detected (Nachman et al, 1977; Sidhu & Ganguly, 1978). The eluted material, after lyophilization and reconstitution, was then tested for its ability to bind thrombin. In addition to the peak due to free [1251]thrombin,a single peak of radioactivity in a position similar to that observed with the protein isolate from the preparative gel was noted (Fig 6a). The binding of thrombin to this material could be reversed with excess unlabelled thrombin (Fig 6b). Further, the glycoprotein isolate inhibited the clotting of fibrinogen by thrombin. DISCUSSION The interaction of thrombin with human platelets is complex. The initial step in this interaction is binding of thrombin to specific structures on the platelet surface. Thrombin binds to about 500 high affinity sites and a considerably larger number of low affinity sites. The
Platelet Thrombin Receptors
rt
I
1
I
1
10
3 P X
10
20
30
Fraction number FIG5. Isolation of glycoprotein I from solubilized tritium-labelled platelets. z mg of platelet protein was applied to a 0 . 5 cm x 6.0 cm column of wheat germ agglutinin covalently linked to Sepharose 4B. The column was washed with 0.25 M NaCI, 0.1%)sodium dodecyl sulphate and 0.05 M phosphate buffer, pH 7.5, until the cpm of the fractions dropped to the baseline level. The bound material was eluted with 0 . 1 M N-acetyl-D-glucosamine (arrow). T h e fractions, z ml each, o f this peak were pooled, dialysed and lyophilized.
I
36-
I
I
I
I
I
1
( 0 )
30 -
7 0
k
I
1
I
I
1
1
I
I
I
I
8
24 -
? 6 0
18-
X
12 06 -
2
-
n
0I
1
I
30
40
50 60 70 Fraction number
1
1
80
90
FIG6. Gel filtration studies on the binding of [1251]thrombin to platelet glycoprotein I isolated by affinity chromatography. (a) so pl of thrombin ( 5 0 mu) wasincubated with 350pl of isolated GP-I (c 70 pg) for 1 5 minat room temperature. The sample was analysed by gel filtration under conditions as described in Fig 4. (b) Fractions (sods) of Fig 6(a) were pooled and concentrated by ultrafiltration through an XM-SOmembrane. An aliquot of this material was then incubated with I u of unlabelled thrombin and analysed as in Fig 4.
I44
P. Canguly and N.L. Cotrid
binding of thrombin to the high affinity sites closely parallels the release of serotonin from platelets (Tollefsen et al, 1974; Ganguly & Sonnichsen, 1976). Further, the selective inhibition of high affinity binding by suitable modification of the thrombin molecule leads to a significant loss in the capability of thrombin to stimulate platelets (Workman et al, 1977). Platelets from patients with defective platelet function also show a lower capacity to bind thrombin (Ganguly et al, 1978). These observations suggest that binding of thrombin to the high affinity sites represents a physiologically significant and essential step in the activation of platelets. W e have shown previously that binding of thrombin to platelets is exquisitely sensitive to the action of trypsin (Ganguly, i977a). It is known that GP-I is the most trypsin-sensitive component on the platelet surface. Further, platelets from Bernard-Soulier patients which have reduced GP-I on their membranes also show a lower capacity to bind thrombin. In contrast, platelets from thrombasthenic patients which havc a normal amount of GP-I but arc defective in other glycoproteins bind thrombin similar to normal platelets (Ganguly , 1977b; White et al, 1978;Jamieson & Okumura, 1978). These results strongly implicate GP-I as the high affinity receptor of thrombin. Okumura & Jamieson (1976) demonstrated that glycocalicin selectively and competitively inhibited thrombin-induced aggregation of platelets. Thus, there is considerable evidence in the literature which suggests that GP-I on the surface of human platelets may be the receptor of thrombin. However, these data are at best indirect. In this study, we have isolated a protein fraction from human platelets by preparative gel electrophoresis and gel filtration. The major component in this preparation was a glycoprotein with molecular weight and stability properties similar to those reported for GP-I (Okumura et al, 1976). In gel filtration studies, this preparation a t nmol protein concentrations showed the capacity to bind thrombin. Glycoprotein I, isolated by affinity chromatography, showed a similar thrombin binding capability. This will suggest that the binding of thrombin to GP-I is a specific event and GP-I may be the high affinity receptor of thrombin in human platelets. This association has the characteristics of a receptor-ligand interaction in that it is specific and reversible. Glycoprotein I appears to be unstable and forms aggregates on storage. Another intriguing possibility is that interaction with thrombin induces polymerization of the receptor. This might lead to patching or capping of the receptors on the platelet surface. W e have shown that GP-I is not hydrolysed by physiological concentrations ( IOO mu) of thrombin (unpublished observation). While these data directly demonstrate binding of thrombin to GP-I the physiological significance of this association in the stimulation of platelets still remains to be established (Nachman et al, 1977a). Studies in several laboratories have shown that platelets contain a significant amount of antithrombin activity (Watanabe et al, 1977). This antithrombin activity of whole platelets has properties different from the known plasma inhibitors (Watanabe et al, 1977). T h e results of this study clearly show that the antithrombin property of intact platelets is due to binding of thrombin to platelet surface GP-I. The platelet antithrombin activity may be physiologically important in removing small amounts of circulating thrombin generated in vivo. ACKNOWLEDGMENTS
We are grateful to Drs A. Mauer, A. Granoff, D. Kingsbury, C. Jackson and K. Subbarao for reviewing the manuscript. This work was supported by grants HL 16720, CA 21765 from NIH and by ALSAC.
Platelet Thrombin Receptors
I45
REFERENCES
GAHMBERG, C.G. & HAKOMORI, S. (1973) External labeling of cell surface galactose and galactosamine in glycolipid and glycoprotein of human erythrocytes. Journal of Biological Chemistry, 248, 4311-4317.
GANGULY, P. (1974) Binding of thrombin to human platelets. Nature, 247. 306-307. GANGULY, P. (1977a) The perturbation of thrombin binding to human platelets by trypsin and neuraminidase. Biothimica et Biophysica Acta, 498, 21-28. GANCULY, P. (1977b) Binding of thrombin to functionally defective platelets: a hypothesis on the nature of the thrombin receptor. British Journal of Haemafology, 37, 47-5 I . GANGULY, P. & SONNICHSEN, W.J. (1976) Binding of thrombin to human platelets and its possible significance. Britishjournal of Haematology, 34,291-301, GANGULY, P., SUTHERLAND, S.B. & BRADFORD, H.R. (1978)Defective binding of thrombin to platelets in myeloid leukaemia. British Journal of Haematology, 39, 5 9 9 4 0 s .
G.A. & OKUMURA, T. (1978) Reduced thrombin binding and aggregation in BernardSoulier platelets. Journal ofclinical Invesfigation, 61, 861-864. KAHANE,I., FURTHMAYR, H. & MARCHESI, V.T. (1976) Isolation of membrane glycoproteins by affinity chromatography in the presence of detergents. Biochimica et Biophysica Ada, 426, 464-476. LAEMMLI,U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T,. Nature, 227, 680-68s. MARCUS, A.J. (1969) Platelet function. N e w Englattd jourual $Medicine, 280, 1213-1220, 1278-1284 MARTIN, B.M., FEINMAN,R.D. & DETWILER, T.C. (1975) Platelet stimulation by thrombin and other proteases. Biochemistry, 14, 1308-13 14. MOHAMMED,S.F., WHITWORTH,C., CHUANG, H.Y.K., LUNDBLAD,R.L. 81 MASON,R.G. (1976) Multiple active forms of thrombin: binding to
JAMIESON,
platelets and effects on platelet function. Proceedings ofthe Nafional Academy of Sciences ofthe United States of America, 73, 1660-1663. NACHMAN,R.L.,JAFFE, E.A.& WEKSLER, B.B. (1977a) Immunoinhibition of ristocetin-induced platelet aggregation. Journal of Clinical Investigation, 59, 143-1 48.
NACHMAN, R.L., TARASOV, E., WEKSLER, B.B. & FERRIS,B. (i977b) Wheat germ agglutinin chromatography of human platelet membrane glycoproteins. Thrombosis Research, 12, 91. OKUMURA, T. &JAMIESON, G.A. (1976) Platelet glycocalicin: a single receptor for platelet aggregation induced by thrombin and ristocetin. Thrombosis Research, 8, 701-706. OKUMURA, T . , LOMBART, C. & JAMIESON, G.A. (1976) Platelet glycocalicin. 11. Purification and characterization. Journal of Biological Chemistry, 251, 5950-5955.
SIDHU, P. & GANGULY, P. (1978)Interactionofthrombin with mammalian platelets: role of surface glycoproteins. (Abstract). Federation Roceedings, 37, 406. TOLLEFSEN, D.M., FEAGLER, J.R. & MAJERUS, P.W. (1974) The binding of thrombin to the surface of human platelets. Journal ofBiological Chemistry, 249, 2646-2651.
WATANABE, K . , C H A O , F.C. & TULLIS, J.L. (1977) Platelet antithrombins: role of thrombin binding and the release of platelet fibrinogen. British Journal ofHaemafology, 35, 123-133. W H ~ T G.C., E , WORKMAN, E.F., JR & LUNDBLAD, R.L. (1978) Thrombin binding to thrombasthenic platelets. Journal of Laboratory and Clinical Medicine, 91, 76-82.
WORKMAN, E.F., WHITE,G.C. & LUNDBLAD, R.L. (1977)Structure-function relationships in the interaction of a-thrombin with blood platelets. Chemical modification of the macromolecular substrate binding site of a-thrombin. journal of Biological Chemistry, 252, 7118-7123.
Thrombin Receptors of Human Platelets: Thrombin Binding and Antithrombin Properties of Glycoprotein I P. GANCULY A N D N. L. GOULD
Laboratory of Hematology, St Jude Children’s Research Hospital, Memphis, Tennessee (Received20June 1978; acceptedforpublication 16 August 1978) SUMMARY. Washed human platelets were solubilized and the proteins were separated by preparative gel electrophoresis in the presence of sodium dodecyl sulphate. The gel was cut into slices and the effect of the eluted proteins on the clotting of fibrinogen by thrombin was evaluated. The isolate from only one gel slice strongly inhibited the clotting of fibrinogen. The prolongation of the clotting time was dependent on the concentration of the protein and reached a plateau around 5 p g . Gel electrophoresis of this isolate showed a prominent glycoprotein with an apparent M,= 150000. Gel filtration studies with [1251]thrombinshowed that the protein isolate bound a significant amount of thrombin which could be displaced with unlabelled thrombin. Another preparation from the same gel or purified y-globulin did not bind thrombin or prolong the clotting time of fibrinogen. Glycoprotein I was isolated from human platelets by affinity chromatography on lectin-Sepharose columns. The isolated glycoprotein prolonged the clotting of fibrinogen and bound [ 1251]thrombin which could be displaced by unlabelled thrombin. It is proposed that the high affinity receptor of thrombin on human platelets is glycoprotein I. In addition, the antithrombin activity of intact platelets is due to binding of thrombin to this glycoprotein. Human platelets treated with thrombin undergo a series of morphological and biochemical changes. The stimulated platelets rapidly secrete calcium, serotonin, adenine nucleotides and other substancesand then slowly aggregate. Thrombin causes these changes in human platelets a t concentrations less than I nM (0.I U/ml) suggesting that thrombin might be an important mediator of platelet function in haemostasis and thrombosis (for review, see Marcus, 1969). Recently, it has been shown that the initial step in the interaction of thrombin with platelets is the binding of the enzyme to specific structures on the platelet surface (Ganguly, 1974; Tollefsen et al, 1974;Martin et all 1975; Ganguly & Sonnichsen, 1976; Mohammed et all 1976). Evidence has also been presented which indicates that binding of thrombin is an important step in the physiological function of platelets (Tollefsen et a l , 1974; Ganguly & Sonnichsen, 1976; Workman et all 1977). To elucidate further the mechanism of action of thrombin on platelets, it is essential to determine the identity of the receptor of thrombin. In this paper we present direct evidence that a glycoprotein, designated as glycoprotein I (GP-I), binds thrombin Correspondence: Dr P. Ganguly, Laboratory of Hematology, St Jude Children’s Research Hospital, P.O. Box 318, Memphis, Tennessee 38101, U.S.A.
0007-1048/79/0500-o13 7$02.00
01979 Blackwell Scientific Publications I37
138
P . Ganguly and N. L. Could
specifically and reversibly. We propose that GP-I is the high affinity receptor of thrombin and the antithrombin activity in human platelets is due to binding of thrombin to this surface component. METHODS
Platelets. Blood was collected from volunteers of both sexes in plastic bags o r syringes utilizing 3.8% sodium citrate as the anticoagulant (9:1 v/v). The red cells were removed by centrifugation and the platelets were washed as described previously (Ganguly & Sonnichsen, 1976).
Electrophorrrir. The washed platelets were solubilized in 3 Yo sodium dodecyl sulphate (SDS)-I% P-mercaptoethanol and heated in a boiling water bath for 5 min. Preparative electrophoresis was carried out in 7.5% acrylamide gels in the presence of 0.1 YOSDS in a 32 mm x 6 0 mm cylindrical glass apparatus (Kontes, Vineland, N.J.) a t 3 0 mA for 40 h. After electrophoresis, a longitudinal slice was cut from the gel and stained to locate the bands. The gel was cut into 5 mrn sections, ground in a blender in 0. I M Tris-o. I M NaC1, pH 7.5 and centrifuged a t 1 5 000 rpm for 30 min. The supernatants were concentrated by ultrafiltration through a XM-50 membrane and dialysed for 48 h against Tris buffer with several changes. The effect of these crude protein preparations was then tested on the clotting of fibrinogen by thrombin. Analysis ofthe different protein fractions were carried out in slab gels by the method of Laemmli ( I 970). Thrombin. Commercial bovine thrombin (Parke Davis, Detroit, Mich.) was purified by ion exchange chromatography and labelled with 1251 by the chloramine T method. Excess free label was removed by gel filtration. Details of these procedures have been reported (Tollefsen el al, 1974; Ganguly & Sonnichsen, 1976). The labelled enzyme was first used for binding studies with intact platelets. The thrombin preparations utilized in this study showed typical binding characteristics (Tollefsen et al, 1974; Ganguly 8: Sonnichsen, 1976). Fibrinogen clotting. Bovine fibrinogen was purchased from Kabi (Stockholm, Sweden) and was 90% clottablc. 100 pl of the test material was incubated with IOOpl of thrombin (2-4 u/ml) for 1 5 min a t room temperature. 200 pl of a 0.5% fibrinogen solution was then rapidly added and the clotting time measured in duplicate. Geljltration. Columns of Sephadex G-200 or Sepharose 613 were prepared according to tho instructions of the manufacturer (Pharmacia, Piscataway, N.J.). Two matched columns packed with Sepharose 6 U were used for the determination of thrombin binding. The test material was incubated with 50 mu of [ i251]thrombinfor 15 min a t room temperature and thcn applied to one column. Control studies with labelled thrombin alone or in combination with other proteins were carried out at the same time on the other column. Samples were eluted with 0. I M Tris-0.1 M NaC1, pH 7.5, containing o. I O h serum albumin. Labelling ofplatelets. This procedure, which is specific for glycoproteins and glycolipids, was carried out according to the general method of Gahmberg & Hakomori (1973). Washed platelets (2-3 x Io8/ml) were incubated with 10 u/ml of neuraminidase for 1 5 min followed by 5 u/ml of galactose oxidase for 60 min and finally I mCi of borotritide for 10 min. The platelets were collected by centrifugation and dissolved as described under electrophoresis. We have shown that in this method most of the radioactivity is incorporated in GP-I (Sidhu & Ganguly, 1978).
Platelet Thrombin Receptors
I39
A f i n i t y chromatography. This procedure was carried out as described (Kahane et al, 1976; Nachman et al, I977b). RESULTS In this study we have used preparative gel electrophoresis for the separation of platelet proteins. This method requires small amounts of platelets and provides high resolution. The effect of the protein eluates from the gel slices on the clotting of fibrinogen by thrombin is shown in Fig I . The clotting of fibrinogen was strongly inhibited by the eluate from a single gel slice. Similar results were obtained from five different experiments and shows the specificity of the I
2
I
I
4
6
I
I
8
1
1
0
SLICE NUMBER FIG I . Effect of human platelet proteins on the clotting of fibrinogen ( 5 mg/ml) by thrombin (4 u/ml). Platelet proteins were separated by preparative gel electrophoresis in the presence of SDS. The gel was cut into 5 m m sections and the protein was eluted out by grinding the gels. The eluates were centrifuged at 1 5 ooo rpm for 3 0 min, concentrated by ultrafiltration and dialysed extensively against Tris-saline, pH 7.2. 100 p1 of thrombin was incubated with 100p1of each eluate for 1 5 min. 200 pl of fibrinogen was then added and the clotting time was determined in duplicate. The inset shows the protein staining pattern of a longitudinal slice of the preparative gel.
inhibitory reaction. The protein concentration (A280",,J of the eluate from the fifth gel slice was lower than those from slices 6-1 I . Thus, the prolongation in the fibrinogen clotting time is not due to a higher protein concentration in the eluate from this particular slice. The inhibitory property remained unchanged after extensive dialysis. The gel filtration pattern of the isolate from the fifth slice through a column of Sephadex G-zoo showed two peaks (Fig 2). The first peak which eluted near the void volume inhibited clotting of fibrinogen. These results show that the inhibitory activity was associated with a macromolecular component of platelets. The fractions of the first peak were pooled and centrifuged at 1 5 000 rpm for 30 min. Assuming an extinction coefficient of 10, the protein concentration was estimated to be 200 &ml. This
P . Canguly and N.L. Could I
I
1
1
I
I
1
I
I
Fraction number FIG2. Gel filtration pattern of the platelet protein through a Sephadex G-200 column (2.5 cm x 37 cm). Whole platelets were solubilized and the constituent proteins separated by preparative gel electrophoresis. 4 ml of protein isolated from the fifth gel slice which showed maximum inhibition of fibrinogen clotting was applied to the column and eluted with 0.1 M Tris-o.1 M NaCI, pH 7.5. Fraction volume 2.0 ml, flow rate 10 ml/h.
value is likely to be higher than the true value because of aggregation of the protein and consequent loss of light due to scattering (see later). The variation in the clotting time of fibrinogen by thrombin with different amounts of the platelet protein is shown in Fig 3 . The clotting time increased progressively reaching a maximum around 5 ,ug of protein. With further addition of the protein, there was a slow decrease in the degree of inhibition of clotting suggesting that the protein may be a restricted
- 60 u
-w k$
- 50
I-
W
Z 40 II-
s 30 0
1
1
2
I
I
I
I
J
4 6 8 1 0 PROTEIN ADDED ( p g )
FIG 3. Effect of varying amounts of platelet protein after gel filtration on the clotting time of fibrinogen by thrombin. Proteins of solubilized whole platelets were separated by electrophoresis and the eluate from the fifth gel slice was fractionated on a Sephadex G-zoo column. The fractions of the first peak were pooled and the effect of this protein o n the clotting time of fibrinogen was determined as in Fig I . 0 , Different amounts of protein; 0,equal amounts of elution buffer control.
Platelet Thrombin Receptors 141 inhibitor. As little as 4 pg of the protein doubled the clotting time of fibrinogen showing a highly specific reaction. These results show that the isolated platelet protein is an antithrombin. The platelet protein isolate was analysed by SDS gel electrophoresis and stained with PAS reagent. Besides some material near the origin, a single band, corresponding to an apparent mol wt of 150000, was observed. Staining for protein revealed two additional faint bands. This material did not react with antiserum to whole serum, fibrinogen or fibrin stabilizing factor. The elution pattern of [1251]thrombinthrough the Sepharose 6B column showed a single peak around fraction 70 (Fig 4a). Some of the thrombin applied binds to the gel. This binding of thrombin is reduced by including albumin in the elution buffer. In this experiment about 80% of the counts applied were recovered. The elution pattern of the protein isolate incubated with [ l251]thrombin showed three peaks of radioactivity (Fig 4b). The material eluting near fraction 70 (peak 111) was due to free [1251]thrombin.Based on the area of the peaks, approximately 70% of the thrombin added was bound and 30% remained free. O f the total thrombin bound, 40% and 60% were associated with peak I and peak I1 respectively. When the same protein preparation was analysed after 24 h at 4%, the amount ofradioactivity associated with the first peak increased considerably. The total amount of thrombin bound to both peaks I and 11 remained essentially the same as in Fig 4(b). However, the amount of radioactivity associated with peak I increased from 40% to 57%. With further storage, the sample became visibly turbid. Thus, it appears reasonable to suggest that the first peak is due to aggregated protein while the second peak, eluting in a position similar to that of y-globulin (m.w. 160 ooo), corresponds to monomeric protein. The fractions (48-60) ofthe second peak (Fig 4b) were pooled and concentrated by ultrafiltration through a XM-so membrane. An aliquot of this material analysed through the Sepharose 6B column showed one peak at the original position. The concentrated material was then incubated with excess unlabelled thrombin and analysed through the column. A substantial amount of the bound [ 1251]thrombin eluted around the position of the free enzyme indicating displacement by the unlabelled thrombin (Fig 4c). As determined from the area of the peaks, about 65% of the thrombin was in the free form. When the platelet protein was incubated with [1251]thrombinin the presence of excess unlabelled thrombin, a reduction in the binding of [ 1251]thrombinsimilar to that shown in Fig 4(c) was observed. To determine further the specificity of thrombin-binding to GP-I, the contents of the fourth and sixth gel slices were recovered. These materials did not prolong the clotting time of fibrinogen by thrombin. In gel filtration studies, these preparations bound about 15% of the [1251]thrombinadded which could not be displaced with excess unlabelled thrombin. Further, y-globulin with a molecular weight similar to the thrombin binding protein neither inhibited fibrinogen clotting nor bound thrombin. The results presented above show that a protein fraction of human platelets which we isolated by gel electrophoresis contains a significant amount of antithrombin activity. This same protien fraction also binds thrombin specifically and reversibly. The major constituent of this protein isolate was a glycoprotein of apparent mol wt I 5 5 000. These observations suggest that GP-I in human platelets may be the receptor of thrombin. T o determine this point, GP-I was isolated by affinity chromatography through columns of wheat germ agglutinin linked to Sepharose 4B (Fig 5 ) . The bound material eluted with N-acetyl-wglucosamine showed a
P. Ganguly and N.L. Could
142 18
-
16
-
12
-
N
0 x
k
0
84 -
Fraction number
Fraction number I
20
30
40
50 60 70 Fraction number
80
90
FIG4.Gel filtration studies on the binding of [ 1251]thrombinto the platelet protein isolate. 50 p1 of thrombin (so mu) containing approximately 20 ooo cpm was incubated at room temperature for 15 min with I ml of the protein solution (zoo pg, c I .3 nmol). The sample was applied to a Sepharose 6B column (1.2 cm x 20 cm) and eluted with 0.1 M Tris-o.1 M NaC1, pH 7.5, containing 0.1% albumin. The flow rate was maintained at 6 ml/h with a pump and fractions of 0.6 ml each was collected in plastic cups. (a) Thrombin+buffer control, (b) thrombin+protein, (c) fractions (48-60) of pattern (b) were pooled and concentrated by ultrafiltration. This material containing 3000 cpm was incubated with 10 u of unlabelled thrombin for 1 5 min and then applied to the column. In this pattern both peaks eluted slightly earlier presumably due to sinking of the gel.
prominent peak of radioactivity corresponding to a glycoprotein of M,= 150ooo (GP-I). A minor component of M,=220 ooo was also detected (Nachman et al, 1977; Sidhu & Ganguly, 1978). The eluted material, after lyophilization and reconstitution, was then tested for its ability to bind thrombin. In addition to the peak due to free [1251]thrombin,a single peak of radioactivity in a position similar to that observed with the protein isolate from the preparative gel was noted (Fig 6a). The binding of thrombin to this material could be reversed with excess unlabelled thrombin (Fig 6b). Further, the glycoprotein isolate inhibited the clotting of fibrinogen by thrombin. DISCUSSION The interaction of thrombin with human platelets is complex. The initial step in this interaction is binding of thrombin to specific structures on the platelet surface. Thrombin binds to about 500 high affinity sites and a considerably larger number of low affinity sites. The
Platelet Thrombin Receptors
rt
I
1
I
1
10
3 P X
10
20
30
Fraction number FIG5. Isolation of glycoprotein I from solubilized tritium-labelled platelets. z mg of platelet protein was applied to a 0 . 5 cm x 6.0 cm column of wheat germ agglutinin covalently linked to Sepharose 4B. The column was washed with 0.25 M NaCI, 0.1%)sodium dodecyl sulphate and 0.05 M phosphate buffer, pH 7.5, until the cpm of the fractions dropped to the baseline level. The bound material was eluted with 0 . 1 M N-acetyl-D-glucosamine (arrow). T h e fractions, z ml each, o f this peak were pooled, dialysed and lyophilized.
I
36-
I
I
I
I
I
1
( 0 )
30 -
7 0
k
I
1
I
I
1
1
I
I
I
I
8
24 -
? 6 0
18-
X
12 06 -
2
-
n
0I
1
I
30
40
50 60 70 Fraction number
1
1
80
90
FIG6. Gel filtration studies on the binding of [1251]thrombin to platelet glycoprotein I isolated by affinity chromatography. (a) so pl of thrombin ( 5 0 mu) wasincubated with 350pl of isolated GP-I (c 70 pg) for 1 5 minat room temperature. The sample was analysed by gel filtration under conditions as described in Fig 4. (b) Fractions (sods) of Fig 6(a) were pooled and concentrated by ultrafiltration through an XM-SOmembrane. An aliquot of this material was then incubated with I u of unlabelled thrombin and analysed as in Fig 4.
I44
P. Canguly and N.L. Cotrid
binding of thrombin to the high affinity sites closely parallels the release of serotonin from platelets (Tollefsen et al, 1974; Ganguly & Sonnichsen, 1976). Further, the selective inhibition of high affinity binding by suitable modification of the thrombin molecule leads to a significant loss in the capability of thrombin to stimulate platelets (Workman et al, 1977). Platelets from patients with defective platelet function also show a lower capacity to bind thrombin (Ganguly et al, 1978). These observations suggest that binding of thrombin to the high affinity sites represents a physiologically significant and essential step in the activation of platelets. W e have shown previously that binding of thrombin to platelets is exquisitely sensitive to the action of trypsin (Ganguly, i977a). It is known that GP-I is the most trypsin-sensitive component on the platelet surface. Further, platelets from Bernard-Soulier patients which have reduced GP-I on their membranes also show a lower capacity to bind thrombin. In contrast, platelets from thrombasthenic patients which havc a normal amount of GP-I but arc defective in other glycoproteins bind thrombin similar to normal platelets (Ganguly , 1977b; White et al, 1978;Jamieson & Okumura, 1978). These results strongly implicate GP-I as the high affinity receptor of thrombin. Okumura & Jamieson (1976) demonstrated that glycocalicin selectively and competitively inhibited thrombin-induced aggregation of platelets. Thus, there is considerable evidence in the literature which suggests that GP-I on the surface of human platelets may be the receptor of thrombin. However, these data are at best indirect. In this study, we have isolated a protein fraction from human platelets by preparative gel electrophoresis and gel filtration. The major component in this preparation was a glycoprotein with molecular weight and stability properties similar to those reported for GP-I (Okumura et al, 1976). In gel filtration studies, this preparation a t nmol protein concentrations showed the capacity to bind thrombin. Glycoprotein I, isolated by affinity chromatography, showed a similar thrombin binding capability. This will suggest that the binding of thrombin to GP-I is a specific event and GP-I may be the high affinity receptor of thrombin in human platelets. This association has the characteristics of a receptor-ligand interaction in that it is specific and reversible. Glycoprotein I appears to be unstable and forms aggregates on storage. Another intriguing possibility is that interaction with thrombin induces polymerization of the receptor. This might lead to patching or capping of the receptors on the platelet surface. W e have shown that GP-I is not hydrolysed by physiological concentrations ( IOO mu) of thrombin (unpublished observation). While these data directly demonstrate binding of thrombin to GP-I the physiological significance of this association in the stimulation of platelets still remains to be established (Nachman et al, 1977a). Studies in several laboratories have shown that platelets contain a significant amount of antithrombin activity (Watanabe et al, 1977). This antithrombin activity of whole platelets has properties different from the known plasma inhibitors (Watanabe et al, 1977). T h e results of this study clearly show that the antithrombin property of intact platelets is due to binding of thrombin to platelet surface GP-I. The platelet antithrombin activity may be physiologically important in removing small amounts of circulating thrombin generated in vivo. ACKNOWLEDGMENTS
We are grateful to Drs A. Mauer, A. Granoff, D. Kingsbury, C. Jackson and K. Subbarao for reviewing the manuscript. This work was supported by grants HL 16720, CA 21765 from NIH and by ALSAC.
Platelet Thrombin Receptors
I45
REFERENCES
GAHMBERG, C.G. & HAKOMORI, S. (1973) External labeling of cell surface galactose and galactosamine in glycolipid and glycoprotein of human erythrocytes. Journal of Biological Chemistry, 248, 4311-4317.
GANGULY, P. (1974) Binding of thrombin to human platelets. Nature, 247. 306-307. GANGULY, P. (1977a) The perturbation of thrombin binding to human platelets by trypsin and neuraminidase. Biothimica et Biophysica Acta, 498, 21-28. GANCULY, P. (1977b) Binding of thrombin to functionally defective platelets: a hypothesis on the nature of the thrombin receptor. British Journal of Haemafology, 37, 47-5 I . GANGULY, P. & SONNICHSEN, W.J. (1976) Binding of thrombin to human platelets and its possible significance. Britishjournal of Haematology, 34,291-301, GANGULY, P., SUTHERLAND, S.B. & BRADFORD, H.R. (1978)Defective binding of thrombin to platelets in myeloid leukaemia. British Journal of Haematology, 39, 5 9 9 4 0 s .
G.A. & OKUMURA, T. (1978) Reduced thrombin binding and aggregation in BernardSoulier platelets. Journal ofclinical Invesfigation, 61, 861-864. KAHANE,I., FURTHMAYR, H. & MARCHESI, V.T. (1976) Isolation of membrane glycoproteins by affinity chromatography in the presence of detergents. Biochimica et Biophysica Ada, 426, 464-476. LAEMMLI,U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T,. Nature, 227, 680-68s. MARCUS, A.J. (1969) Platelet function. N e w Englattd jourual $Medicine, 280, 1213-1220, 1278-1284 MARTIN, B.M., FEINMAN,R.D. & DETWILER, T.C. (1975) Platelet stimulation by thrombin and other proteases. Biochemistry, 14, 1308-13 14. MOHAMMED,S.F., WHITWORTH,C., CHUANG, H.Y.K., LUNDBLAD,R.L. 81 MASON,R.G. (1976) Multiple active forms of thrombin: binding to
JAMIESON,
platelets and effects on platelet function. Proceedings ofthe Nafional Academy of Sciences ofthe United States of America, 73, 1660-1663. NACHMAN,R.L.,JAFFE, E.A.& WEKSLER, B.B. (1977a) Immunoinhibition of ristocetin-induced platelet aggregation. Journal of Clinical Investigation, 59, 143-1 48.
NACHMAN, R.L., TARASOV, E., WEKSLER, B.B. & FERRIS,B. (i977b) Wheat germ agglutinin chromatography of human platelet membrane glycoproteins. Thrombosis Research, 12, 91. OKUMURA, T. &JAMIESON, G.A. (1976) Platelet glycocalicin: a single receptor for platelet aggregation induced by thrombin and ristocetin. Thrombosis Research, 8, 701-706. OKUMURA, T . , LOMBART, C. & JAMIESON, G.A. (1976) Platelet glycocalicin. 11. Purification and characterization. Journal of Biological Chemistry, 251, 5950-5955.
SIDHU, P. & GANGULY, P. (1978)Interactionofthrombin with mammalian platelets: role of surface glycoproteins. (Abstract). Federation Roceedings, 37, 406. TOLLEFSEN, D.M., FEAGLER, J.R. & MAJERUS, P.W. (1974) The binding of thrombin to the surface of human platelets. Journal ofBiological Chemistry, 249, 2646-2651.
WATANABE, K . , C H A O , F.C. & TULLIS, J.L. (1977) Platelet antithrombins: role of thrombin binding and the release of platelet fibrinogen. British Journal ofHaemafology, 35, 123-133. W H ~ T G.C., E , WORKMAN, E.F., JR & LUNDBLAD, R.L. (1978) Thrombin binding to thrombasthenic platelets. Journal of Laboratory and Clinical Medicine, 91, 76-82.
WORKMAN, E.F., WHITE,G.C. & LUNDBLAD, R.L. (1977)Structure-function relationships in the interaction of a-thrombin with blood platelets. Chemical modification of the macromolecular substrate binding site of a-thrombin. journal of Biological Chemistry, 252, 7118-7123.
