T cell lines characterize events in the pathogenesis of the Wiskott-Aldrich syndrome.

The Wiskott-Aldrich syndrome (WAS) is a severe immunodeficiency and platelet deficiency disease arising from an X-linked defect. The disease is correctable by transplantation of hematopoietic stem cells, but the product of the defective gene is unidentified and the number of defects in patient blood cells is large. The current hurdle is the need to identify the early pathogenic event(s) that are the cause of other defects. As a step toward this goal, we have generated and examined a panel of interleukin 2-dependent allospecific T cell lines from peripheral lymphocytes of seven WAS patients and five normal individuals. WAS cell lines, like normal lines, undergo vigorous proliferation when challenged with specific allostimulant or with phorbol myristate acetate and ionomycin. Both normal and WAS T cell lines express cell surface molecules CD2, CD3, T cell receptor-alpha/beta, human histocompatibility leukocyte antigen class I, CD45 and CD11a, and varying ratios of CD4 and CD8, and are negative for natural killer cell and monocyte surface molecules. WAS T cell lines express CD43 (sialophorin/leukosialin) with molecular weight and in an amount comparable with normal T cell lines. WAS T cell lines thus do not express defects in CD43 (decreased amount, abnormal molecular weight), previously documented in WAS circulating lymphocytes. On the other hand, as detected by scanning electron microscopy, WAS cell lines exhibit severe morphological abnormalities, including decreased size and density of the microvillus surface projections. The morphological abnormalities of WAS T cell lines are similar to, or more extensive than, those previously reported for WAS peripheral lymphocytes, indicating that the generation of morphological (cytoarchitectural) defects is an early pathogenic event in this disease. The findings suggest that the gene that is defective in the WAS encodes a protein that normally functions to maintain or regulate the cytoskeletal structure of blood cells.

Characterization of a human lymphocyte surface sialoglycoprotein that is defective in Wiskott-Aldrich syndrome.

gpL115 is a lymphocyte surface component that is deficient in patients with the X-chromosome-linked immune deficiency Wiskott-Aldrich syndrome (6). The glycoprotein nature of gpL115 is demonstrated through labeling in carbohydrate moieties by [3H]NaBH4 and its synthesis by lymphocytes through labeling with [35S]methionine. Native gpL115 adheres to wheat germ lectin-Sepharose and sialidase-treated gpL115 does not adhere, indicating that native gpL115 adheres via clusters of sialic acid residues. When tested on peanut lectin, which shows specificity for the disaccharide Gal beta 1-3GalNAc, gpL115 is nonadherent and sialidase-treated gpL115 is adherent, indicating the presence of the sequence sialic acid-Gal beta 1-3GalNAc, which is characteristic for O-linked (mucin-type, acidic-type) carbohydrates. A surface glycoprotein with all the above characteristics was found on the lymphoblastoid cell line CEM. CEM cells were used as immunogen to generate the monoclonal antibody L10, an IgG1, which binds native and sialidase-treated gpL115 . Sialidase-treatment of gpL115 significantly alters its physical properties, reducing its electrophoretic mobility and changing its behavior on isoelectrofocusing. Cumulatively, these findings indicate that gpL115 , like glycophorin of erythrocytes and GPIb of platelets, is a sialoglyco protein with significant quantities of O-linked carbohydrate. On treatment with limiting sialidase concentrations, gpL115 of normal lymphocytes is transformed into a series of partially desialylated species of decreasing electrophoretic mobility. This finding resembles the situation with lymphocytes of some Wiskott-Aldrich syndrome patients. Lymphocytes of eight Wiskott-Aldrich syndrome patients were found to be deficient in 125I-labeled gpL115 . Lymphocytes from three of these patients displayed an abnormal 125I-component of apparent mol wt 135,000.

Properdin factor D: effects on thrombin-induced platelet aggregation.

Factor D, when preincubated with platelet suspensions, at concentrations as low as 1.2 micrograms/ml, inhibited thrombin-induced platelet aggregation. No inhibition of collagen or arachidonic acid-induced platelet aggregation was found. Inhibition occurred, but to a lesser extent, when thrombin and factor D were added to platelets at the same time. No inhibition occurred when factor D was added after thrombin. Thrombin was able to overcome inhibition by factor D by increasing its concentration. Diisopropyl-phosphorofluoridate-inactivated factor D also inhibited thrombin-induced platelet aggregation so that enzymatic activity of factor D was not required for inhibition. Factor D absorbed with hirudin coupled to Sepharose 6B showed no decrease in inhibitory capacity. 125I-Factor D bound to platelets in a manner suggesting an equilibrium reaction similar to thrombin. At low factor D input, binding was linear, whereas at higher input, binding began to approach saturation. Binding of 125I-labeled thrombin to platelets was inhibited by factor D. Analysis of these data show that factor D does not alter the total number of thrombin molecules which bind to the platelet surface at saturation. However, the dissociation constant for thrombin is altered from 2.78 to 6.90 nM in the presence of factor D (20 micrograms/ml). Factor D is thus a competitive inhibitor of thrombin binding, although the affinity of factor D for the platelet thrombin receptor is much less than that of thrombin. These phenomena occur at physiologic concentrations of factor D. Therefore, factor D may function in vivo as an inhibitor of platelet aggregation.

Moesin, the major ERM protein of lymphocytes and platelets, differs from ezrin in its insensitivity to calpain.

The ERM proteins, ezrin, radixin and moesin, provide regulated linkage of the cytoskeleton with the plasma membrane, particularly in cell surface projections. Ezrin and moesin were found co-expressed, and radixin was not detected, in human blood lymphocytes, monocytes and neutrophils. Moesin is the quantitatively dominant ERM protein in these cells and the only one in platelets. Because Ca signaling pathways involving calpain cleavages are important in blood cells, we examined ERM protein sensitivity to this protease. A striking difference was discovered: sensitivity of ezrin and resistance of moesin (and radixin) to calpain. In intact stimulated lymphocytes, ezrin was cleaved, while moesin was not, strongly suggesting that differential sensitivity to calpain contributes to specialized functions of these proteins.

Effect of sodium chloride on limulus amebocyte lysate. Inhibition of endotoxin activation of procoagulase.

Concentrations of sodium chloride up to 3 M increase the time necessary for the clot formation from Limulus amebocyte lysate (LAL) induced with endotoxin. Sodium chloride at a concentration of 4 M prevents clot formation by either precipitation or denaturation of procoagulase. The time necessary for the activation of procoagulase by endotoxin is increased by a change in the sodium chloride concentration from 0.15 M to 0.588 M. No effect on the proteolytic phase or the polymerization phase of the clotting reaction is detected by the increase in sodium chloride concentration from 0.15 M to 0.588 M. The authors conclude that increased sodium chloride concentrations may aid the isolation of procoagulase.

Ionophore-induced disassembly of blood platelet microtubules: effect of cyclic AMP and indomethacin.

Cytoplasmic calcium levels are believed to be important in blood platelet activation. Upon activation, the discrete marginal microtubule band, which maintains the discoid shape of non-activated platelets, becomes disrupted. Present studies demonstrate that the extent of assembly of the marginal microtubule band is related to cytoplasmic calcium levels. The divalent cationophore, A23187, causes platelet aggregation, secretion, and contraction by promoting calcium transport from intraplatelet storage sites into the cytoplasm. A23187 caused disassembly of platelet microtubules. Quantitation of electron micrographs revealed that numbers of microtubules were reduced by approximately 80% after A23187 treatment. Secondly, assembled microtubules in homogenates of platelets, in which microtubules were stabilized prior to homogenization, were decreased in favor of free tubulin in A23187-treated platelets. Thirdly, A23187 increased 14C-colchicine binding by intact platelets; this also indicated a shift in the microtubule subunit equilibrium to favor free, colchicine-binding tubulin subunits. In control experiments, A23187 did not affect the stability of platelet tubulin, the colchicine binding reaction, or the total tubulin content of platelets. Stimulation of colchicine binding depended on A23187 concentration (0.05-0.5 microM) and did not require extracellular calcium. A23187-stimulation of colchicine binding was blocked by dibutyryl cyclic AMP (0.80 mM) and/or 3-isobutyl-1-methylxanthine (50 microM) and by indomethacin (10 microM). Cyclic AMP or indomethacin also interferes with A23187-induced platelet activation, but indomethacin is not likely to completely inhibit the perturbation of intraplatelet calcium gradients by A23187. It is suggested that A23187-induced microtubule disassembly may be an indirect effect of calcium on microtubules.

The cystoskeleton of unstimulated blood platelets: structure and composition of the isolated marginal microtubular band.

Detergent-insoluble, marginal microtubular band (MB) cytoskeletons were isolated from unstimulated blood platelets after pretreatment with glycerol or with Taxol. MB cytoskeletons retained the shape of intact platelets and behaved in suspension as coherent structural units. The major structural component was a continuous coil of long microtubule(s), often with granular/amorphous material present in the centre; few typical actin filaments were observed. The coiled microtubules often had an amorphous surface coating, but no discrete inter-microtubule bridges were seen. Tubulin and actin (identified by immunochemical staining) were major polypeptides. None of the minor (greater than 10) polypeptide components comigrated with high molecular weight microtubule-associated proteins in brain tubulin. A novel polypeptide, resolved by two-dimensional electrophoresis and designated IEF-51K, was present in MB cytoskeletons in amounts approximately equivalent to each of the tubulin polypeptides. Evidence suggests that IEF-51K is a distinct, previously undescribed component of the platelet cytoskeletal system.

Localizaton of fibrin converted by thrombin from released platelet fibrinogen.

Treatment of platelets (10(9) cells/ml) with thrombin (1 U/ml) resulted in rapid disappearance of fibrinogen from the system as measured by the tanned red cell hemagglutination inhibition immunoassay (TRCHII). Plasmin digestion of individual pellet and supernatant fractions that had been previously separated from thrombin-treated platelet suspensions by centrifugation resulted in recovery of TRCHII-detectable material in platelet pellets. To elucidate the specific association of fibrin to platelet membranes, control and thrombin-treated platelets were homogenized by a modified glycerol-loading and nitrogen decompression technique. Ultracentrifugation of homogenates through 27% sucrose cushions yielded three subcellular fractions: supernatant, small membrane vesicles, and a particulate fraction for controls; and supernatant membrane vesicles, and aggregated membrane "ghosts" for thrombin preparations. Ultrastructurally identifiable fibrin was noted only in the thrombin fraction containing membrane ghosts. Fibrinogen recovered from 3 thrombin fractions was markedly decreased (3% of the control). Plasmin digestion produced 23% and 46-fold increase in TRCHII-detectable material from 3 subcellular fractions of control and thrombin preparations, respectively. More than 97% of TRCHII material recovered from thrombin preparations was in the fraction containing aggregated membrane fractions. Results suggest that platelet plasma membranes function as surfaces for fibrin deposition.

Isolated cytoskeletons of human blood platelets: dark-field imaging of coiled and uncoiled microtubules.

Detergent extraction of human blood platelets pre-treated with Taxol to stabilize microtubules allows isolation of marginal band (MB) cytoskeletons. We studied MB cytoskeleton structure using dark-field light microscopy and negative stain electron microscopy (EM). Dark-field illumination clearly demonstrated the "hoop" shape of MB cytoskeletons in unfixed suspensions where the microtubule coils had a mean diameter of 2.87 microns (+/- 0.18 micron, SD). Microtubules were uncoiled by brief exposure to trypsin (2 ng/micrograms protein) or by NaCl (154-600 mM) but not by DNase I, which removed approximately 40% of total actin, but had no effect on dark-field images of microtubule coils. As microtubules uncoiled, a single fiber emerged from the hoop and gradually lengthened as the brightness of the hoop diminished; these fibers correspond to the single microtubules seen by EM. Polypeptides of coiled and uncoiled MB cytoskeletons were analyzed by SDS-PAGE. When microtubules became uncoiled, no changes in the major components (alpha- and beta-tubulin, IEF-51K, or actin) were found. However, a number (greater than 10) of minor polypeptides, each less than 5% of total cytoskeletal protein and with an Mr ranging from 80,000- greater than 260,000, were decreased in "uncoiled" MB cytoskeletons. These results implicate one or more of these minor polypeptides in maintenance of hoop integrity. Dark-field light microscopy thus provides an approach toward investigating the mechanism(s) involved in maintaining the microtubule coil of the platelet marginal band.

A novel microtubule protein in the marginal band of human blood platelets.

A characterization is reported of the major cytoskeletal protein, called IEF (isoelectric focusing)-51K, of marginal band microtubule coils from human blood platelets (Kenney, D. M. and Linck, R. W. (1985) J. Cell Sci. 78, 1-22). IEF-51K is a unique biochemical species which is distinguishable from platelet and mammalian neuronal alpha-tubulin and beta-tubulin by 1) its faster mobility on discontinuous sodium dodecyl sulfate electrophoresis corresponding to an apparent Mr 51,000; 2) its more alkaline relative isoelectric point at pH 5.7 compared with that of alpha- and beta-tubulin at pH 5.3 and 5.5, respectively; 3) lack of coincidence in peptide maps prepared with chymotrypsin or Staphylococcus aureus V8 protease; and 4) lack of immunochemical cross-reactivity of polyclonal anti-IEF-51K with alpha- and beta-tubulin and of monoclonal anti-alpha-tubulin and anti-beta-tubulin with IEF-51K. In contrast to its chemical uniqueness, IEF-51K is tubulin-like in some of its properties. IEF-51K is localized in the marginal band of intact platelets by immunofluorescence; it undergoes cycles of microtubule disassembly/reassembly both in vitro and in vivo. Furthermore, IEF-51K was not extracted from isolated Taxol-stabilized marginal band microtubules by elevated NaCl concentrations (to 0.45 M), conditions that do not disrupt the polymeric structure of alpha- and beta-tubulin. These results indicate that IEF-51K together with alpha-tubulin and beta-tubulin are the major structural polypeptides of platelet marginal band microtubules. The unusual subunit composition of the platelet marginal band microtubule may be related to specialization(s) of microtubule structure and function in the marginal band coil of platelets.

Extended storage of single-donor apheresis platelets in CLX blood bags: effect of storage on platelet morphology, viability and in vitro function.

A CLX (Cutter Laboratories, Berkeley, Calif.) bag system was evaluated for storage of single-donor apheresis platelets collected with the Haemonetics V-50 blood processor. Concentrates (n = 21) containing 3.9-5.2 x 10(11) platelets in 292 (+/- 41.8) ml were stored in two 1-liter bags for 7 days at 22 degrees C. pH was well maintained, declining from an initial pH of 7.0 (+/- 0.04) to 6.92 (+/- 0.20) after 7 days. Platelet morphology, response to a hypotonic stimulus and aggregation induced by paired agonists (epinephrine and ADP, or collagen) were also well-preserved. Concentrates with a wide variation of platelet yields (2.0 greater than or equal to 5.2 x 10(11), n = 43) also maintained pH (6.96 +/- 0.26), morphology and aggregation when stored for 7 days. All platelet concentrates (n = 64) were sterile at collection. Single-donor apheresis platelets may be stored in this bag system for up to 7 days.

The surge technique: a method to increase purity of platelet concentrates obtained by centrifugal apheresis.

A surge technique has been developed to increase the purity of high-yield platelet concentrations prepared on a blood processor with the Latham bowl. The surge technique combines elutriation with centrifugal separation, utilizing plasma recirculated from the plasma/air bag back into the centrifuge bowl to elute platelets from the red cell mass. Platelet concentrates prepared by surge collection with six separation cycles (n = 22), contained an average of 3.9 +/- 1.4 X 10(11) platelets, with 0.15 +/- 0.11 X 10(9) leukocytes, and red cells below the level of detection. The surge technique reduces collection time by 4 minutes per cycle and eliminates the need for a secondary centrifugation; thus, 96 minutes donor processing time would permit eight separation cycles. Platelets collected by the surge technique exhibited unaltered morphology and capacity to take up radioactively labeled serotonin in vitro compared to pre-apheresis controls. Results from in vitro functional studies also indicate that the capacity of platelets collected with the surge technique to respond to various concentrations of adenosine diphosphate, collagen, and thrombin by aggregation and secretion of both serotonin (dense bodies) and beta-thromboglobulin (alpha-granules) was not significantly different (p less than or equal to 0.05) from that of pre-apheresis controls.

Platelet antithrombin defect in malignancy: platelet protein alterations.

Sixty-eight patients with malignant disease were divided into two groups based on the results of the platelet antithrombin test (PAT). The normal group had a PAT clotting time ranging from 21.4 to 29.8 seconds, which was equivalent to 25% to 65% inactivation of the 2 U of thrombin added to the test system. The other group showed abnormal PAT clotting time, less than 21.4 seconds or less than 25% thrombin inactivation. The polypeptide composition of platelets from the two patient groups was analyzed by sodium dodecyl sulfate (SDS)-electrophoresis on 7.5% polyacrylamide gels. A polypeptide of 180,000 apparent mol wt was decreased or absent in both Coomassie blue- and Alcian blue-stained gels of the platelets from patients whose PAT was abnormal; this polypeptide comigrated with purified platelet thrombospondin. Tritium labeling of platelet surface glycoproteins by the periodate-borohydride method followed by two-dimensional electrophoresis was performed on platelets of seven patients with abnormal PAT. When they were compared with ten patients with normal PAT, a glycoprotein of 140,000 apparent mol wt with a pl of 4.5 to 5.2 was decreased in platelets of all seven patients with abnormal PAT. Nitrocellulose replicas of one-dimensional gels of platelets from 13 of 14 patients with abnormal PAT showed decreased reaction with an anti-human platelet glycocalicin antiserum. Platelets of these same patients also showed a decreased or absent platelet agglutination induced by ristocetin. Patients with normal PAT had a mean agglutination slope of 1.25 +/- 0.6 (n = 26) as compared with 0.37 +/- 0.34 (n = 26) for the abnormal PAT group (P less than .001). Results indicate that platelets from a subpopulation of tumor patients characterized by decreased platelet antithrombin activity have alterations in two platelet glycoproteins, identified as GPIb and thrombospondin.

Localization of the domain of actin-binding protein that binds to membrane glycoprotein Ib and actin in human platelets.

The Mr approximately 540,000 dimeric actin gelation protein, actin-binding protein (ABP), has previously been shown in human platelets to link actin to membrane glycoprotein Ib (GPIb) (Fox, J. E. B. (1985) J. Biol. Chem. 260, 11970-11977; Okita, J. R., Pidard, D., Newman, P. J., Montgomery, R. R., and Kunicki, T. J. (1985) J. Cell Biol. 100, 317-321). We have examined further the interaction between ABP and GPIb. Platelet extracts were depleted of ABP by precipitation with anti-ABP monoclonal antibodies (mAbs); in resulting precipitates, ABP monomer is complexed with GPIb in a 5:1 molar ratio. The ABP.GPIb complex is resistant to chaotropic solvents but dissociated by the ionic detergent, sodium dodecyl sulfate. Treatment of intact platelets with the ionophore A23187 activates a Ca2+-dependent protease which cleaves the Mr approximately 270,000 ABP subunit into three fragments of Mr 190,000, 100,000, and 90,000; the latter fragment is derived from the Mr 100,000 fragment. Anti-ABP mAbs coprecipitated GPIb with the Mr 100,000 and 90,000 fragments, but not with the Mr 190,000 fragment which contains the ABP self-association site. In the reciprocal experiment, anti-GPIb antibodies co-precipitated only the Mr 100,000 and 90,000 ABP fragments. Actin also co-precipitated with the Mr 100,000 and 90,000, but not with the Mr 190,000 ABP fragment. The anti-ABP mAb that precipitated the Mr 100,000-90,000 GPIb-binding ABP fragment recognizes a trypsin cleavage fragment of ABP that binds actin filaments in vitro. These findings establish that both the GPIb-binding site and actin-binding sites are in the same region of the ABP monomer. Because of the extended bipolar conformation of the ABP molecule, the data suggest that the GPIb.actin-binding region is located remote from the self-association, or dimerization, site of the ABP subunit.

Effect of platelet calpain on normal T-lymphocyte CD43: hypothesis of events in the Wiskott-Aldrich syndrome.

The Wiskott-Aldrich syndrome (WAS) is an inherited disease involving defects of platelets (small size, severe thrombocytopenia due to accelerated destruction) and T lymphocytes (progressive immunodeficiency, lymphopenia). The best-characterized molecular defect is the deficiency and, in some cases, abnormal forms of the T-lymphocyte surface mucin molecule CD43; deficiency of the platelet surface mucin GPIb was observed previously in two of four patients. Neither of these defects is primary, since CD43 and GPIb are encoded by autosomal genes and the disease is X-linked. This study uses cellular biological approaches to explore the possibility that destruction of structurally defective WAS platelets, mimicked experimentally by sonication of normal platelets, plays a role by releasing protease and generating other cellular defects. We show that a protease of normal platelets, identified as Ca(2+)-dependent neutral protease (calpain), which is known to cleave platelet GPIb, also specifically cleaves CD43 on the surface of neighboring desialylated T lymphocytes. The identification of the CD43 cleaving protease was based on its requirement for Ca2+ and inhibition by leupeptin, but not by diisopropylfluorophosphate (DFP). The approximate site of CD43 cleavage was identified by the use of a rabbit antibody. Sensitivity of GPIb to calpain is shown to be sialylation-independent and that of CD43 to be sialylation-dependent, and these findings are explained in terms of molecular structures. These and previous findings are incorporated into a putative mechanism, which explains most of the defects in the WAS. The mechanism suggests that the primary defective molecule in the WAS is unlikely to be a surface glycoprotein, but rather a cytoplasmic molecule with a function in cytoskeletal interactions and/or calcium ion regulation and calpain activation.

Colchicine uptake and binding by human platelets.

The uptake and binding of antimitotic alkaloid colchicine has been demonstrated in washed preparations of human pletelets. A silicone oil technique was adapted so that both uptake and binding of 14C-colchicine were examined in the same platelet preparations. The time dependence and amount of colchicine taken up and bound by different pletelet preparations during a 90 to 120 min incubation period were highly reproducible. Both colchicine uptake and binding by intact platelets, and colchicine binding by preparations of lysed platelets were specific and temperature dependent. Colchicine uptake was slowly reversible. Magnesium and GTP enhanced colchicine binding by lysed platelet preparations but calcium decreased binding. exposure of platelets to either cold (4 degrees C) or to thrombin, which disrupt platelet microtubules, produced significant increases in colchicine uptake and binding. The thrombin effect was maximal at 37 degrees C and resulted in a greater increase in uptake and binding than that produced by either cold treatment alone or, by cold treatment followed by incubation with thrombin at 37 degrees C. The amount of increase in uptake and binding produced by thrombin was independent of both thrombin (1--5 Units/10(9) platelets) and colchicine concentrations (1--50 X 10(-6) M). It is postulated that thrombin may initiate the formation, or make available, colchicine binding sites (microtubule subunits) within platelets.

Surface protein abnormalities in lymphocytes and platelets from patients with Wiskott-Aldrich syndrome.

Lymphocytes from three patients with the Wiskott-Aldrich syndrome were subjected to surface radioiodination and their 125I-labelled surface proteins were analysed by SDS-polyacrylamide gel electrophoresis and autoradiography. Autoradiographs demonstrated the absence in all three patients of an 125I-labelled protein, molecular weight of 115 000, that was present in normal individuals. In addition, one patient had an additional labelled protein, molecular weight 135 000, not found in normal individuals. The platelets from one patient were radioiodinated and the 125I-labelled membrane proteins were analysed. Glycoproteins Ia and Ib were reduced in amount and restricted in heterogeneity while glycoproteins IIb and IIIa were normal. These results suggest that the primary defect in the Wiskott-Aldrich syndrome may be abnormalities of (glyco)proteins normally present on the surface of lymphocytes and platelets.

The cytoskeletal linker protein moesin: decreased levels in Wiskott-Aldrich syndrome platelets and identification of a cleavage pathway in normal platelets.

The Wiskott-Aldrich syndrome (WAS) is a severe disease of platelets (small size, thrombocytopenia) and lymphocytes (immunodeficiency) arising from mutations of the X-chromosome gene WASP. Because of the prominent role of cytoskeletal abnormalities, particularly the paucity of surface microvilli, in the cellular pathology of this disease, blood cells from WAS patients were examined for moesin, a cytoskeletal linker protein that stabilizes cell surface microvilli, filopodia and lamellipodia. Comparison of patient and normal lymphocytes by immunofluorescence microscopy and immunoblotting showed normal levels and distribution of moesin in lymphocytes of WAS patients. In contrast, platelets from WAS patients stained only dimly for moesin relative to normal platelets. Quantitation by immunoblot revealed significantly decreased moesin levels in WAS patient platelets relative to normal platelets (63.5 +/- 4.9% of normal levels, n = 8, P < 0.0001). A novel reaction of normal platelets was discovered that may play a role in the depletion of moesin in patient platelets, namely the cleavage of moesin as a late event in platelet activation in response to certain platelet agonists.