Reduction of non-endothelial cell contamination of microvascular endothelial cell seeded grafts decreases thrombogenicity and intimal hyperplasia.

INTRODUCTION: fat derived microvascular endothelial cells (MVEC) seeded on prosthetic vascular grafts, improve patency in animals. Results in humans were disappointing, due to thrombogenicity and progressive intimal hyperplasia. Also in animals intimal hyperplasia was found. We postulate that contaminating cells present in the transplant are involved in the intimal hyperplasia. We developed a method to further purify human MVEC from 40-90%. Here we tested the effects of enrichment upon thrombogenicity and seeding-related intimal hyperplasia. METHODS: liposuction fat was enzymatically digested and centrifuged. To enrich MVEC, contaminating macrophages and fibroblasts were removed with dynabeads coated with macrophage- and fibroblast-specific antibodies. Thrombogenicity was assessed by measuring tissue factor and thrombomodulin activity, presence of endothelial nitric oxide synthase and via perfusion of the cells with whole blood. To investigate seeding-related intimal hyperplasia, PTFE grafts were seeded with the cells and cultured for 3 weeks. RESULTS: tissue factor activity of purified cells was reduced compared to nonpurified cells. Purified cells showed thrombomodulin activity and eNOS expression. Fragment 1+2 and Fibrinopeptide A generation after perfusion of purified cells were significantly lower than after perfusion of nonpurified cells, and only nonpurified cells were covered with platelets and fibrin. Prostheses seeded with nonpurified cells showed an EC monolayer above a multilayer of myofibroblasts, prostheses seeded with purified cells only showed a single EC monolayer. Mixing experiments with human umbilical cord EC (HUVEC) and fibroblasts showed that when more than 25% HUVEC were present a confluent EC layer was formed. When the amount of fibroblasts was 25% or less, no development of a subendothelial multilayer of myofibroblasts was found within 3 weeks. CONCLUSION: reduction of non-endothelial cell contamination of microvascular endothelial cell seeded grafts decreases thrombogenicity and might prevent seeding-related intimal hyperplasia.

Contaminants from the transplant contribute to intimal hyperplasia associated with microvascular endothelial cell seeding.

OBJECTIVES: seeding prosthetic grafts with fat-derived microvascular endothelial cells (MVEC) results not only in a non-thrombogenic EC layer, but also in intimal hyperplasia. Here we investigated incidence, composition, progression, and cause of this intimal hyperplasia. DESIGN: EPTFE grafts with MVEC were implanted as carotid interpositions in six dogs with 1 month, and in three dogs with 4, 8 and 12 months follow-up. Grafts seeded without cells, implanted in the contralateral carotid, served as a control. In another three dogs labelled cells were seeded to investigate the contribution of the seeded cells (2-3 weeks). MATERIALS AND METHODS: MVEC were isolated from the falciform ligament. Cells were pressure seeded on ePTFE grafts. Labelling was performed using retroviral gene transduction. The grafts were analysed with immunohistochemical techniques. RESULTS: after 1 month, all patent non-seeded grafts (5/6) showed fibrin and platelet deposition, and all patent seeded grafts (5/6) were covered with a confluent endothelial monolayer on top of a multilayer of myofibroblasts, elastin and collagen. After long term follow-up, all non-seeded grafts were occluded, all patent seeded grafts (4 and 12 months) were covered with an EC-layer with intimal hyperplasia underneath. The thickness of the intima did not progress after 1 month. Transduced cells were found in the endothelial monolayer, hyperplastic intima and luminal part of the prosthesis. CONCLUSIONS: MVEC seeding in dogs results in intimal hyperplasia in all patent grafts, which contains myofibroblasts. Contaminants from the transplant contribute to this intimal hyperplasia.

The 33-kDa platelet alpha-granule membrane protein (GMP-33) is an N-terminal proteolytic fragment of thrombospondin.

GMP-33 is a platelet membrane associated protein that is recognised by RUU-SP 1.77, an antibody raised against activated platelets. GMP-33 is predominantly associated with the membrane of platelet alpha-granules and it is translocated to the plasma membrane upon platelet activation (Metzelaar et al. Blood 1992; 79: 372-9). In this study we have isolated the protein by immunoaffinity chromatography. The N-terminus was sequenced and was identical to the N-terminal sequence of human thrombospondin. The protein was N-glycosylated and bound to heparin as would be expected of the N-terminal part of thrombospondin. RUU-SP 1.77 reacted only with reduced thrombospondin. Plasmin and trypsin digestion of thrombospondin yielded fragments of approximately the same size as GMP 33 that reacted with RUU-SP 1.77 after reduction. No evidence for alternative splicing was found. We postulate that GMP 33 is an N-terminal proteolytic fragment of thrombospondin that is membrane associated.

Mutations on fibrinogen (gamma 316-322) are associated with reduction in platelet adhesion under flow conditions.

In this paper we report on studies of platelet adhesion to several fibrinogen gamma chain variants under physiological flow conditions. Reduced platelet adhesion was found to patient dysfibrinogen Vlissingen and its recombinant form (deletion of gamma 319-320). Furthermore, substitutions of the amino acids 318, 320, or both in the recombinant fibrinogen gamma chain showed a strong decrease in platelet adhesion under flow conditions in our perfusion system. Antibodies raised against peptides covering these sequences inhibited platelet adhesion completely, which suggested that the gamma 316-322 sequence could be involved in platelet adhesion in flowing blood.

The structure of leech anti-platelet protein, an inhibitor of haemostasis.

Leech anti-platelet protein (LAPP) from the leech Haementeria officinalis is a collagen-binding protein that inhibits the collagen-mediated adhesion of blood platelets. The crystal structure of recombinant LAPP has been determined using single isomorphous replacement with anomalous scattering combined with solvent flattening and threefold molecular averaging. The model of LAPP has been refined to 2.2 A resolution (R factor 21.5%; free R factor 24.0%). LAPP contains an 89-residue C-terminal domain consisting of a central six-stranded antiparallel beta-sheet flanked on one side by an alpha-helix and on the other side by two extended loops with little secondary structure. A 36-residue N-terminal region is not visible in the electron-density map. This region is rich in glycine and lacks hydrophobic residues. It probably does not have a compact globular fold, but instead has an extended conformation and is flexible. The crystal packing suggests that LAPP may form tightly interacting dimers. The fold of the C-terminal domain of LAPP closely resembles that of the N-domain of hepatocyte growth factor (HGF), which classifies LAPP as a PAN domain. However, no significant sequence homology exists between LAPP and other PAN domains. Common structural features between LAPP and the HGF N-domain include two disulfide bonds that link the alpha-helix to the central region of the protein and five residues with a conserved hydrophobic nature that are located in the core of the domain. These conserved structural features may be an important determinant of the PAN-domain type of fold.

Excessive urinary albumin levels are associated with future cardiovascular mortality in postmenopausal women.

BACKGROUND: Microalbuminuria is an early predictor of cardiovascular morbidity and mortality, in both diabetic patients and hypertensive patients. Little is known about the relation of microalbuminuria to cardiovascular disease in women of the general population. METHODS AND RESULTS: We have studied the relation of urinary albumin levels to cardiovascular mortality in a cohort study of 12 239 postmenopausal women living in Utrecht, the Netherlands. The initial age was between 52 and 67 years. Women were followed on vital status between 1976 and 1995 (168 513 women-years). Albumin was determined in the urine of 561 cases and 557 controls. Data were analyzed by using a nested case-control design. The cardiovascular mortality rate (95% CI) for women who were in the highest quintile of urinary albumin levels was 13.2/1000 years (8.1 to 20.9) compared with 2.6/1000 years (2.3 to 3.1) in women without detectable urinary albumin. The age-adjusted rate ratio (95% CI) between these groups was 4.4 (2.6 to 7.6). CONCLUSIONS: This is the first large cohort study that confirms a predictive role of urinary albumin for the risk of future cardiovascular mortality independent of hypertension and diabetes. Our findings support the hypothesis that microalbuminuria is a reflection of vascular damage and a marker of early arterial disease in women from the general population.

Absence of fibrinogen in afibrinogenemia results in large but loosely packed thrombi under flow conditions.

We studied the role of fibrinogen in platelet thrombus formation under flow on adhesive proteins using afibrinogenemic blood (LMWH anticoagulated) in a perfusion system. Perfusions with afibrinogenemic blood showed strong increased surface coverage and thrombus volume that normalized upon addition of fibrinogen. Similar studies using citrate anticoagulated blood showed that this was due to fibrinogen and not fibrin. Morphological analysis showed that afibrinogenemic thrombi were loosely packed and consisted mainly of dendritic platelets that contacted one another through filopodia. However, in the presence of fibrinogen, platelets formed lamellipodia and spread out on top of one another. Studies with radiolabeled platelets showed similar numbers of platelets in both conditions demonstrating that the difference is one of packing and the larger size is due to absence of lamellipodia formation and spreading. The found increased thrombus size and loosely packed platelets might help to understand thrombotic complications sometimes seen in afibrinogenemia patients.

Methylenetetrahydrofolate reductase 677 C/T genotype and cardiovascular disease mortality in postmenopausal women.

Methylenetetrahydrofolate reductase (MTHFR) is involved in the reduction of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate. A 677 C/T single nucleotide polymorphism localized in the MTHFR gene is associated with both thermolability and reduced activity of the enzyme and is associated with increased homocysteine levels. The authors investigated the relation between the MTHFR 677 C/T polymorphism and risk of cardiovascular disease mortality in a cohort study of 12,239 women initially aged 52--67 years with a maximum follow-up time of 18 years (1976--1995; 153,732 woman-years of follow-up). The cardiovascular disease mortality rate was highest among women with the MTHFR 677 CC wild-type genotype and lowest among MTHFR 677 TT homozygotes. In comparison with women with the 677 CC wild-type genotype, age-adjusted rate ratios were 0.7 (95% confidence interval: 0.5, 0.9) for 677 CT heterozygotes and 0.6 (95% confidence interval: 0.4, 1.0) for 677 TT homozygotes. The possibility that this relation is a chance finding must be considered, because the relation is weak and of borderline significance. However, it provides an important argument against the view that increased levels of homocysteine directly raise cardiovascular disease risk.

Identification of the collagen-binding site of the von Willebrand factor A3-domain.

Von Willebrand factor (vWF) is a multimeric glycoprotein that mediates platelet adhesion and thrombus formation at sites of vascular injury. vWF functions as a molecular bridge between collagen and platelet receptor glycoprotein Ib. The major collagen-binding site of vWF is contained within the A3 domain, but its precise location is unknown. To localize the collagen-binding site, we determined the crystal structure of A3 in complex with an Fab fragment of antibody RU5 that inhibits collagen binding. The structure shows that RU5 recognizes a nonlinear epitope consisting of residues 962-966, 981-997, and 1022-1026. Alanine mutants were constructed of residues Arg(963), Glu(987), His(990), Arg(1016), and His(1023), located in or close to the epitope. Mutants were expressed as fully processed multimeric vWF. Mutation of His(1023) abolished collagen binding, whereas mutation of Arg(963) and Arg(1016) reduced collagen binding by 25-35%. These residues are part of loops alpha3beta4 and alpha1beta2 and alpha-helix 3, respectively, and lie near the bottom face of the domain. His(1023) and flanking residues display multiple conformations in available A3-crystal structures, suggesting that binding of A3 to collagen involves an induced-fit mechanism. The collagen-binding site of A3 is located distant from the top face of the domain where collagen-binding sites are found in homologous integrin I domains.

Sequence alignment between vWF and peptides inhibiting the vWF-collagen interaction does not result in the identification of a collagen-binding site in vWF.

We previously found that two peptides (N- and Q-peptide) selected by phage display for binding to an anti-vWF antibody, were able to inhibit vWF-binding to collagen (1). The sequence of those peptides could be aligned with the sequence in vWF at position 1129-1136 just outside the A3-domain. As the peptides represent an epitope or mimotope of vWF for binding to collagen we next wanted to study whether the alignment resulted in the identification of a new collagen binding site in vWF. We mutated the 1129-1136 VWTLPDQC sequence in vWF to VATAPAAC. Expressing this mutant vWF (7.8-vWF) in a fur-BHK cell line resulted in well processed 7.8-vWF containing a normal distribution of molecular weight multimers. However, binding studies of this mutant vWF to rat tail, human and calf skin collagens type 1, to human collagen types III and VI, revealed no decrease in vWF-binding to any of these collagens. Thus, although the N- and Q-peptides did inhibit the vWF-collagen interaction, the resulting alignment with the vWF sequence did not identify a collagen binding site, pointing out that alignments (although with a high percentage of identity) do not always result in identification of binding epitopes. However, suprisingly removal of the A3-domain or changing the vWF sequence at position 1129-1136 resulted in an increase of vWF-binding to human collagen type V1 and to rat tail collagen type 1, implying that these changes result in a different conformation of vWF with an increased binding to these collagens as a consequence.

Homozygosity for 807 T polymorphism in alpha(2) subunit of platelet alpha(2)beta(1) is associated with increased risk of cardiovascular mortality in high-risk women.

BACKGROUND: Platelet adhesion to collagen is the initial step in both hemostasis and thrombosis; this adhesion is mediated by alpha(2)beta(1) on the surface of platelet membranes. An 807 C to T single nucleotide exchange polymorphism close to the gene coding for the alpha(2) subunit of alpha(2)beta(1) is associated with the density of alpha(2)beta(1) on the platelet membrane. METHODS AND RESULTS: We studied the relation of the alpha(2)beta(1) 807 C/T genotype to cardiovascular mortality in a prospective cohort study of 12 239 women who were invited for the breast cancer screening program of Utrecht, the Netherlands. The initial age was between 52 and 67 years. Women were followed on vital status between 1976 and 1995 (168 513 women-years). Data were analyzed by using a nested case-control design. The alpha(2)beta(1) 807 C/T genotype was not associated with cardiovascular mortality in the total population: the rate ratio for cardiovascular mortality in 807 TT homozygotes compared with 807 CC wild types was 1.2 (95% CI 0.8 to 1.7). However, the alpha(2)beta(1) 807 T polymorphism was associated with an increased risk of cardiovascular mortality in women who smoked or in women who had indications of compromised endothelium, such as diabetes and microalbuminuria. In those who were exposed to >/=2 of these factors, the risk ratio (95% CI) between alpha(2)beta(1) 807 TT homozygotes and 807 CC wild types was 14.1 (5.0 to 39.9). CONCLUSIONS: alpha(2)beta(1) 807 TT homozygosity, coding for increased alpha(2)beta(1) density on the platelet membrane, is associated with an increased risk of cardiovascular mortality in those women with indications of compromised endothelium.

Protein A is the von Willebrand factor binding protein on Staphylococcus aureus.

Endovascular infection is a highly critical complication of invasive Staphylococcus aureus disease. For colonization, staphylococci must first adhere to adhesive endovascular foci. Von Willebrand factor (vWF) is a large, multimeric glycoprotein mediating platelet adhesion at sites of endothelial damage. Earlier it was demonstrated that vWF binds to and promotes the surface adhesion of S. aureus, prompting this effort to identify the vWF adhesin. In Western ligand assays of S. aureus lysates, staphylococcal protein A (SPA) was recognized by purified vWF. Surface plasmon resonance demonstrated the binding of soluble vWF to immobilized recombinant protein A with a K(d) of 1.49 x 10(-8) mol/L. Using flow cytometry, the binding of fluorescein isothiocyanate-labeled vWF to S. aureus was found to be saturable and inhibitable by unlabeled vWF, antiprotein-A antibodies, or IgG. Isogenic Deltaspa::Tc(r) mutants were constructed by the insertion of a tetracycline resistance cassette into spa using allelic replacement, and it exhibited decreased binding of soluble vWF and decreased adhesion to vWF-adsorbed surfaces. The interaction was restored on complementation of the mutants with spa-containing plasmid pSPA7235. In conclusion, protein A confers interaction of S. aureus with soluble and immobilized vWF in a newly discovered function characterizing protein A as a novel member of the staphylococcal surface protein adhesin superfamily and suggesting its potential role in the pathogenesis of endovascular staphylococcal disease.

Platelet adhesion to collagen in healthy volunteers is influenced by variation of both alpha(2)beta(1) density and von Willebrand factor.

Platelet thrombus formation on collagen is initiated by platelet GPIb interaction with von Willebrand factor (vWF) bound to collagen, followed by firm attachment of the platelet to collagen by the integrin alpha(2)beta(1). Platelet and plasma vWF levels and alpha(2)beta(1) density on the platelet surface are highly variable among normal subjects; however, little is known about the consequences of this variability on platelet adhesion to collagen. A population of 32 normal subjects was studied to evaluate the relation between genetic and phenotypic variations of alpha(2)beta(1) density on the platelet surface, plasma vWF levels, platelet vWF levels, and adenosine diphosphate and adenosine triphosphate concentrations on the one hand and platelet adhesion to collagen under flow on the other hand. Platelet adhesion to collagen types I and III under flow was correlated with plasma levels of vWF (r(2) = 0.45 and 0.42, respectively) and alpha(2)beta(1) density on the platelet surface (r(2) = 0.35 and 0.17, not significant). Platelet adhesion to collagen type IV under flow was significantly correlated with platelet vWF levels (r(2) = 0.34) and alpha(2)beta(1) density on the platelet surface (r(2) = 0.42). Platelet adhesion to collagen types I and III depends on both plasma levels of vWF and alpha(2)beta(1) density on the platelet surface, whereas platelet adhesion to collagen type IV is mediated by both platelet vWF levels and alpha(2)beta(1) density on the platelet surface. (Blood. 2000;96:1433-1437)

Platelet thrombus formation on collagen at high shear rates is mediated by von Willebrand factor-glycoprotein Ib interaction and inhibited by von Willebrand factor-glycoprotein IIb/IIIa interaction.

We studied the role of von Willebrand Factor (vWF) in platelet thrombus formation in flowing blood by using a perfusion system and mutant forms of vWF lacking either interaction with glycoprotein Ib (GpIb) or with glycoprotein IIb/IIIa (alphaIIb-beta3). These mutants were added to the blood of patients with severe von Willebrand's disease (vWD) or to normal blood reconstituted with a human albumin solution instead of plasma. This blood was then perfused over collagen type III spray-coated on a glass surface and preincubated for 2 hours with 20 microg/mL plasma vWF. In this way, the adhesion step was mediated by the preincubated plasma vWF bound to collagen type III, whereas thrombus formation was mediated by mutant vWF added to the perfusate. Thrombus formation was absent at all 3 shear rates studied (300, 800, and 2600 s(-1)) when DeltaA1-vWF, lacking interaction with GpIb, was added to the perfusate, indicating the importance of GpIb-vWF interaction for thrombus formation. The interaction of vWF and GpIb is currently thought to be possible under physiological conditions in which the conformation of vWF has been changed by adsorption to a surface. Our results regarding the role of GpIb-vWF interaction in thrombus formation suggest that a second mechanism may operate by which a change may occur in GpIb on the surface of adhered platelets either by activation of the molecule or as a consequence of shear stress. Increased thrombus formation was observed when the Arg-Gly-Gly-Ser-vWF, which does not interact with alphaIIb-beta3, was added to vWD blood and perfused at 2600 s(-1). This increase was not observed in vWD blood at lower shear rates or after addition of Arg-Gly-Gly-Ser-vWF to reconstituted normal blood. Thrombus formation at a high shear rate was largest when either vWF or fibrinogen was present as a single ligand for alphaIIb-beta3 at a high shear rate. When both were present, thrombus formation was decreased. We postulate that thrombus formation is less efficient because of incomplete bridge formation when vWF and fibrinogen are both present as ligands for alphaIIb-beta3.

The half-life of infused factor VIII is shorter in hemophiliac patients with blood group O than in those with blood group A.

A considerable inter-individual variation in half-life of infused factor VIII is observed among patients with hemophilia A. The factors contributing to this wide range in factor VIII half-life are not known in detail. We analysed the pharmacokinetics of infused factor VIII in 32 patients with hemophilia A, comprising 20 brothers from 10 families, 3 and 4 brothers from 2 families, and 5 patients from 5 single families, respectively. Multiple linear regression analysis was used to assess the effect of several variables on factor VIII half-life. We found that the pre-infusion von Willebrand factor antigen levels (vWF:Ag) were positively correlated with factor VIII half-life (r = 0.52, p = 0.002), i.e., each variable was associated with about 27% of the variance of the other. In fraternal pairs, familial clustering was significant for ABO blood group (p < 0.001), but could not be detected for factor VIII half-lives or pre-infusion vWF:Ag levels. vWF:Ag level (p = 0.001) and ABO blood group (p = 0.003) significantly determined factor VIII half-life, whereas age, length, bodyweight, the presence or absence of a factor VIII gene inversion, and Rhesus phenotype did not. Patients with blood group O exhibited a statistically significant shorter factor VIII half-life than patients with blood group A (15.3 versus 19.7 h, respectively) (p = 0.003). Patients with blood group A and O differ in respect to the presence of anti-A antibodies in the latter. It is possible that these anti-A antibodies interact with endogenous vWF, thus affecting the half-life time of the factor VIII/vWF complex.

Plasminogen activator inhibitor 4G polymorphism is associated with decreased risk of cerebrovascular mortality in older women.

BACKGROUND: A common 4G allele of a 4G/5G polymorphism in the promoter region of the plasminogen activator inhibitor-1 (PAI-1) gene is associated with increased transcription of the PAI-1 protein, which may lead to decreased fibrinolysis. It has therefore been proposed as a candidate risk factor for myocardial infarction or stroke. METHODS AND RESULTS: We studied the relationship between PAI-1 4G/5G genotype and the risk of cardiovascular mortality in a prospective cohort study among 12 239 women initially aged between 52 and 67 years, with a maximum follow-up time of 18 years (153 732 follow-up years). PAI-1 4G/5G genotype was measured in DNA obtained from urine samples, which were collected at baseline, of 498 women who died of a cardiovascular disease and a random sample of 512 women from the same cohort who did not die of cardiovascular disease. The PAI-1 4G/5G genotype was not associated with risk of myocardial infarction or other cardiovascular mortality. However, PAI-1 4G4G homozygotes had a markedly reduced risk of cerebrovascular mortality compared with PAI-1 5G5G homozygotes: the relative risk was 0.4, with a 95% CI of 0.2 to 0.7, whereas the relative risk of cerebrovascular mortality in PAI-1 4G5G heterozygotes compared with PAI-1 5G5G homozygotes was 0.7, with a 95% CI of 0.4 to 1.1. CONCLUSIONS: These findings are suggestive of an important contribution of PAI-1 in cerebrovascular pathology, probably via pathways other than fibrinolysis. PAI-1 may protect against destabilization of the atherosclerotic plaque, or it may inhibit neurotoxicity of tissue plasminogen activator in the brain.

Binding of von Willebrand factor to collagen type III: role of specific amino acids in the collagen binding domain of vWF and effects of neighboring domains.

Binding of von Willebrand Factor (vWF) to sites of vascular injury is the first step of hemostasis. Collagen types I and III are important binding sites for vWF. We have previously determined the three-dimensional structure of the collagen binding A3 domain of vWF (Huizinga et al., Structure 1997; 5: 1147). We hypothesized that the top face of this domain might be the collagen-binding site. Based on this hypothesis, we made seven vWF mutants (D934A/S936A, V1040A/ V1042A, D1046A, D1066A, D1069A, D1069R, and R1074A). Collagen binding of these mutants was investigated in ELISA and with Surface Plasmon Resonance (BIAcore). In addition, we studied collagen binding of mutants lacking the A2 or D4 domains, which flank the A3 domain. In ELISA, all point mutants and deletion mutants bound to collagen in amounts similar to wild type (WT)-vWF. In the BIAcore we found that WT-vWF has an apparent KD for collagen of 1-7 nM on a subunit base. The apparent kinetic parameters of the point mutants and deletion mutants were not significantly different from WT-vWF, except for DA2-vWF, which had a lower KD. indicating that the A2 domain somehow modulates binding of vWF to collagen type III. Based on our results, we conclude that the amino acid residues mutated by us are not critically involved in the interaction between vWF and collagen type III, which suggests that the collagen binding site is not located on the top face of the A3 domain.

Activated platelets release two types of membrane vesicles: microvesicles by surface shedding and exosomes derived from exocytosis of multivesicular bodies and alpha-granules.

Platelet activation leads to secretion of granule contents and to the formation of microvesicles by shedding of membranes from the cell surface. Recently, we have described small internal vesicles in multivesicular bodies (MVBs) and alpha-granules, and suggested that these vesicles are secreted during platelet activation, analogous to the secretion of vesicles termed exosomes by other cell types. In the present study we report that two different types of membrane vesicles are released after stimulation of platelets with thrombin receptor agonist peptide SFLLRN (TRAP) or alpha-thrombin: microvesicles of 100 nm to 1 microm, and exosomes measuring 40 to 100 nm in diameter, similar in size as the internal vesicles in MVBs and alpha-granules. Microvesicles could be detected by flow cytometry but not the exosomes, probably because of the small size of the latter. Western blot analysis showed that isolated exosomes were selectively enriched in the tetraspan protein CD63. Whole-mount immuno-electron microscopy (IEM) confirmed this observation. Membrane proteins such as the integrin chains alpha(IIb)-beta(3) and beta(1), GPIbalpha, and P-selectin were predominantly present on the microvesicles. IEM of platelet aggregates showed CD63(+) internal vesicles in fusion profiles of MVBs, and in the extracellular space between platelet extensions. Annexin-V binding was mainly restricted to the microvesicles and to a low extent to exosomes. Binding of factor X and prothrombin was observed to the microvesicles but not to exosomes. These observations and the selective presence of CD63 suggest that released platelet exosomes may have an extracellular function other than the procoagulant activity, attributed to platelet microvesicles.