Endothelial microparticles induce formation of platelet aggregates via a von Willebrand factor/ristocetin dependent pathway, rendering them resistant to dissociation


Wenche Jy, 1600 NW 10th Ave. (R-36A), University of Miami School of Medicine, Miami, FL 33136, USA.
Tel.: +1 305 243 6617; fax: +1 305 243 5957; e-mail: wjy@med.miami.edu


Summary.  Endothelial microparticles (EMP) released from activated or apoptotic endothelial cells (EC) are emerging as useful markers for detection of EC dysfunction. Our recent observation that EMP carry von Willebrand factor (vWf) led us to investigate their interaction with platelets. EMP were incubated with normal washed platelets in the presence or absence of ristocetin, then platelet aggregates were measured by flow cytometry. In the absence of ristocetin, negligible EMP conjugated with platelets (< 5%) but in the presence of ristocetin (1 mg mL−1), EMP induced up to 95% of platelets to aggregate. EMP-platelet interaction was 80% blocked by anti-CD42b, or by 0.1 μm filtration to remove EMP. Platelet aggregates induced by normal plasma or high molecular weight vWf (Humate-P) dissociated 50% within 15–25 min following 1 : 20 dilution. In contrast, aggregates formed with EMP persisted two- to threefold longer with the same treatment, indicating greater stability. A similar degree of prolongation of dissociation was observed using plasma from thrombotic thrombocytopenic purpura (TTP) patients compared with normal plasma. Addition of EMP to plasma from severe von Willebrand disease restored his ristocetin-induced platelet aggregation. Multimer analysis of vWf on EMP showed unusually large vWf (ULvWf). In summary, EMP carries ULvWf multimers, promote platelet aggregates, and increase the stability of the aggregates thus formed.


Release of cell membrane-derived microparticles (MP) during cell activation has been demonstrated in platelets, erythrocytes, leukocytes and endothelial cells (ECs) [1,2]. Most MP expose procoagulant anionic phospholipids such as phosphatidylserine (PS) [3–6], giving rise to procoagulant activity known as platelet factor 3 (PF3) [7], believed to be a major functional activity of MP. More recently, tissue factor (TF) has been identified on leukocyte MP (LMP) [8–10], endothelial MP (EMP) [10–13], and platelet MP (PMP) [9,14], further suggesting important roles of MP in hemostasis and thrombosis.

There is also evidence for a role of MP in inflammation. We demonstrated that PMP bind to specific subsets of leukocytes, inducing their activation and aggregation [15]. This observation was subsequently extended by Barry et al. [16,17] who reported that PMP induced up-regulation of ICAM-1 on human umbilical vein endothelial cells (HUVEC) and promoted adhesion of monocytes to HUVEC. Mesri et al. showed that LMP, too, can induce endothelial activation including expression of TF [18]. Thus, MP may be regarded as diffusible extracellular messengers, mediating cell interaction and activation [19].

Platelets do not adhere to healthy endothelium owing to the anti-adhesive and anti-thrombotic properties of normal EC [20–23], whereas even non-activated platelets adhere avidly to activated or injured endothelium via several adhesion molecules which become expressed on stimulated EC [24–27], particularly under conditions of shearing stress [27,28]. It has been shown that fibrinogen can serve as a cross-linking bridge between activated GP IIb/IIIa on platelets and ICAM-1 on activated EC [29]. P-selectin on activated endothelium has been shown to participate in the rolling of platelets on EC [30]. Andre et al. [31] demonstrated that secretion of von Willebrand factor (vWf) from endothelium in response to inflammatory stimuli leads to increased local recruitment of platelets to the endothelial surface. However, it is not yet clear if EMP can interact with platelets.

We recently showed that vWf was associated with subspecies of EMP [32]. The present study was undertaken to investigate possible functional significance of EMP-bound vWf in their interaction with platelets. We here demonstrate that platelet-EMP interaction is mediated through vWf. We compare the stability of the platelet aggregates formed with EMP to those formed by normal plasma, Humate-P, and thrombotic thrombocytopenic purpura (TTP) plasma. In addition, we exhibit the effect of vWf-positive EMP on the ristocetin-induced aggregability of plasma from patients with von Willebrand disease (vWD). Finally, we compare and correlate the multimer size of EMP-bound vWf with free soluble vWf from normal plasma, TTP plasma, and Humate-P as a function of the stability of the resulting platelet aggregates.

Methods and materials


Human cultured EC of microvascular renal or brain origin, and of coronary artery origin, were obtained from Cell Systems (Kirkland, WA, USA). FITC-labeled anti-CD62E (clone 1.2B6, Cat. no. F-0674) was obtained from Sigma (St Louis, MO, USA). FITC-labeled anti-CD42b (clone SZ2, Cat. no. IM0409) and anti-CD41 (clone P2, Cat. no. IM0649) were obtained from Beckman-Coulter (Fullerton, CA, USA). HRP-conjugated anti-vWf (Cat. no. AHP 062-P) was purchased from Serotec Inc. (Raleigh, NC, USA). Gel electrophoresis reagents and instruments were obtained from Bio-Rad (Richmond, CA, USA). Ristocetin (Cat. no. 396) was purchased from Chrono-Log. Humate-P, a therapeutic agent containing concentrated Factor VIII and high-multimer vWf, was obtained from Aventis-Behring (Marburg, Germany). Other chemicals were purchased from Sigma.

Preparation of EMP and platelets

Human EC were activated with tumor necrosis factor (TNF)-α (10 ng mL−1) for 24 h to induce EMP generation [13]. The culture supernatants were then centrifuged at 15 000 g for 30 min to sediment EMP, which were then washed three times with PBS buffer and re-suspended in PBS to one of 10 original volume. Concentration of EMP was measured by flow cytometry using FITC-labeled anti-CD62E as described [33].

Washed platelets were prepared by centrifuging platelet-rich plasma (PRP) at 600 g for 10 min in the presence of 10 mm EGTA and 1 μm PGE1. The pellets were washed twice with phosphate-buffered saline (PBS), then suspended in PBS at 1 × 108 mL−1.

Assay of EMP-platelet interaction and ristocetin-dependent aggregation by flow cytometry

Endothelial microparticles at 5–100 × 106 mL−1 final concentration were incubated with normal washed platelets at 1 × 107 mL−1 final, in the presence or absence of ristocetin (1 mg mL−1) for 10 min with gentle orbital shaking (100 rpm). Binding of EMP to platelets was assessed by co-expression of EMP marker CD62E with platelet marker CD41 in flow cytometry. In most experiments, plasma at 1–15% was used as a source of vWF, indicated; and Humate-P at 0.02 to 0.4 U mL−1, as indicated. Platelet aggregation was measured flow cytometrically [34] by counting the number of free platelets (< 5 μm) shifted to a bit-map representing platelet aggregates (> 5 μm). The flow rate of the Coulter XL flow cytometer (Beckman Coulter Inc., Fullerton, CA, USA) was at medium setting and discriminator was forward scatter (FS), level 3. Platelet counts were calibrated with standard beads of known concentrations. A decrease in singular free platelets (accompanied by increase in number of platelet aggregates) was observed when ristocetin was present in the plasma. At maximal plasma or Humate-P with ristocetin (1 mg mL−1), very few platelets, < 5%, remains singular (free). Reduction of number of singular platelets with/without ristocetin was taken to indicate degree of platelet aggregation, rather than counting the number of micro-aggregates, because the latter is ambiguous owing to heterogeneous aggregate size and sticking to the flow chamber and tubing.

Dissociation of ristocetin-induced platelet aggregates

After 20 min of platelet aggregate formation at room temperature, samples were diluted with PBS (1 : 20 volume ratio) to initiate time-dependent dissociation of aggregates. Release of free platelets after dilution was monitored at intervals to determine the time-course of dissociation by flow cytometry.

vWf multimer analysis

The method of Raines et al. [35] was employed with minor modifications, as follows. Cooling during electrophoresis was accomplished by resting the horizontal gel electrophoresis apparatus on an aluminum block immersed in ice-water slurry and the buffer chambers were also on ice. Several agarose gel concentrations were tested and 0.8% was found to be optimal for showing a wide range of multimer sizes. Western blotting was according to Raines et al. except that the anti-vWf was preconjugated with HRP (Serotec Inc., Raleigh, NC, USA; Cat. no. AHP 062-P) and was used at 1 : 500 dilution (50 μL in 25 mL). The proteins in the gel were transferred (blotted) to polyvinylidene fluoride (PVDF) membrane by capillary diffusion aided by layers of paper towels on top of PVDF membrane overnight with PBS as transfer buffer. The PVDF membrane was then blocked with 0.5% casein solution, then stained by the method of Nakane [36] using dye 4-chloro-1-napthol (4CN; Sigma, Cat. no. C-8890) prepared fresh by dissolving 30 mg in 5 mL of ethanol, then bringing to volume 100 mL by addition of 50 mm Tris buffer pH 7.6 containing 0.03% H2O2 (1 mL of 3% H2O2 in 100 mL).

Clinical studies

The protocol was approved by the Institutional Review Board, and informed consent was obtained from all patient subjects. Citrated blood was obtained from four TTP and four Type I vWD patients. The four TTP patients all presented with the classic triad of TTP; severe thrombocytopenia (platelet count < 2 × 107 mL−1), microangiopathic hemolytic anemia, and mental dysfunction. The four vWD patients were characterized by low vWf total antigen (0.30–0.45 U mL−1) and ristocetin cofactor activity (0.16–0.45 U mL−1).

Statistical analysis

For comparing three or more groups, one-way anova was use to determine the P-values. If P < 0.05, then two-tailed Student's t-test was used to analyze the significance (P < 0.05) of difference between the means of two groups. In cases where the data failed the normality test, then the Mann–Whitney rank sum test was used. All data analyses were performed using Windows-based program, Statmost (Dataxiom Software Inc., Los Angeles, CA, USA).


Platelet aggregation induced by EMP

The presence of EMP at 4 × 107 mL−1 final induced strong platelet aggregation which was dependent on ristocetin. As seen in Fig. 1, the degree of ristocetin-dependent platelet aggregation caused by EMP was similar to that caused by 8% normal plasma. In the absence of ristocetin, negligible platelet aggregates were formed with either EMP, Humate-P, or normal plasma. Both EMP-induced and plasma-induced platelet aggregation was inhibited by anti-CD42b blocking mAb. These results demonstrate that EMP induced platelet aggregation, which was vWf-dependent.

Figure 1.

Endothelial microparticles (EMP)-induced platelet aggregate formation in the presence or absence of ristocetin. All samples were 50 μL final volume and contained washed normal platelets at 1 × 107 mL−1 final concentration, to which was added the indicated agents. Plasma, 8%; ristocetin, 1 mg mL−1; EMP of microvascular renal origin, 4 × 107 mL−1; blocking antibody CD42b, 20 μg mL−1. All experiments were repeated five times (n = 5). Error bars indicate ±SD. *P < 0.01 between ‘plasma + ristocetin’ and ‘plasma + ristocetin + anti-CD42b.’**P < 0.03 between ‘EMP + ristocetin’ and ‘EMP + ristocetin + anti-CD42b’.

The dose-response curves of platelet aggregation induced by EMP, normal plasma, and Humate-P are shown in Fig. 2. Notice that the shapes of the curves for all three agents are similar. As both the platelets and the EMP were prewashed and essentially plasma-free, these results demonstrate that EMP-bound vWf can substitute for soluble vWf in plasma or Humate-P in inducing full platelet aggregation with ristocetin. The data indicate that 50% aggregation occurs with 3.5% plasma, equivalent to 1 × 107 mL−1 of EMP, and to 0.075 U mL−1 of Humate-P.

Figure 2.

Dose-response curves of platelet aggregate formation induced by plasma, Humate-P, or EMP. Procedure was as detailed in Methods and materials. Analysis of these results showed that the ED50's were: control plasma, 3.5%; Humate-P, 0.075 U mL−1; EMP, at 1 × 107 mL−1. Error bars show ±SD; n = 5 experiments.

To further confirm the existence of MP-bound vWf, we tested the effect of filtration through 0.1 μm filter, which is known to retain the majority of EMP ≥ 0.1 μm. As shown in Fig. 3, this filtration largely abolished EMP-induced platelet aggregation but had no significant effect on normal plasma- or Humate-P-induced platelet aggregation.

Figure 3.

Effects of 0.1 μm filtration on plasma-, Humate-P-, or EMP-induced platelet aggregate formation. Figure shows effect of filtration through 0.1 μm filter (Whatman ‘Anotop 10,’ Cat. no. 6809-1012) affixed to a 1 mL tuberculin syringe. Notice that filtration of soluble vWF, Humate-P, gave negligible losses, whereas filtration of EMP largely abolished platelet aggregation. Error bars show ±SD; n = 4; *P < 0.01 between ‘filtered EMP’ and ‘unfiltered EMP’.

EMP from endothelia of different origins

We also compared activities of EMP from three different sources of EC: microvascular renal and brain, and macrovascular coronary artery. All three EC were cultured under similar conditions and were stimulated with the same concentration of TNF-α for 24 h. EMP were collected and washed as described in Methods and Materials, counted by flow cytometry, and adjusted to equal concentrations. Table 1 shows the relative specific activities of the EMP from these three sources in inducing vWf-dependent platelet aggregation with ristocetin. It is seen that EMP derived from renal or brain microvascular EC were more potent than those from coronary artery EC. This is consistent with our previous finding, that renal or brain EMP contained higher percentage of vWf+ EMP [32] and with the fact that the clinical manifestations of abnormally active vWf are mainly related to microangiopathic thrombosis.

Table 1.  Comparison of vWf activity of EMP derived from different sources of human tissue cultures
 Subtypes of EMP
  1. *P < 0.01 as the ‘renal EMP’ or ‘brain EMP’ group compared with the ‘coronary EMP’ group.

  2. Endothelial microparticles (EMP) obtained from renal, brain and coronary artery endothelial cells (EC) as described in the Methods and Materials section were adjusted to equal concentrations prior to evaluate their proaggregatory activity in the presence of ristocetin. The table shows that EMP from different EC lines exhibited different activities in ristocetin-induced platelet aggregation, in the following order, renal > brain >> coronary EC. n = 4, mean ± SD.

% platelet aggregates induced by 2 × 107 mL−1 EMP68 ± 13%*54 ± 10%*29 ± 6%

Assessment of aggregate stability

In the course of pilot studies, we observed that when platelet aggregates induced by plasma plus ristocetin were diluted 20-fold with PBS buffer, the aggregate population gradually declined and the number of free platelet rose in a time-dependent manner. Figure 4 depicts the time course of dissociation of platelet aggregates induced by plasma, Humate-P, and EMP. After platelet aggregates were induced for 10 min, the mixtures were diluted with PBS (1 : 20) to promote dissociation. The time for 50% dissociation for plasma, Humate-P and EMP were about 15, 25 and 60 min respectively. These results demonstrate that platelet aggregates induced by EMP are more stable than those induced by plasma or Humate-P. We postulate that the greater stability of the aggregates formed by EMP may be because of (i) the presence of very large multimers of vWf on EMP and/or (ii) the presence of other adhesion molecules contributing to cross-linking between EMP and platelets.

Figure 4.

Time course of dissociation of platelet aggregates. Procedure was detailed in Methods and Materials section. Briefly, after 20 min incubation of 50 μL total volume of platelets with the indicated agents (control plasma, Humate-P, EMP), 20-fold volume of PBS was added at t = 0, then samples were run in flow cytometer at the specified times for measurement of dissociation of aggregates, i.e. increases of free platelets. The times to 50% dissociation for plasma, Humate-P, and EMP were respectively 15, 25, and 60 min. Error bars show mean ± SD; n = 4.

Effects of TTP plasma on ristocetin induced platelet aggregate formation

Because abnormal degree of vWf multimerization has been implicated in TTP [37,38], we investigated plasmas from four TTP patients in acute (A) and remission (R) phases, compared with normal pooled plasma. As shown in Table 2, TTP patients exhibited significantly increased ristocetin-induced platelet aggregation, in both acute and remission states.

Table 2.  Comparison of vWf activity of plasma from TTP in acute and remission phases with control
 Sources of plasma
  1. *P < 0.05 comparing thrombotic thrombocytopenic purpura (TTP) in acute phase or TTP in remission phase vs. the ‘control’ group.

  2. Platelet poor plasma (PPP) (4%) from four different TTP patients in acute (A) and remission (R) phases and pooled control plasma were incubated with platelets and ristocetin for 10 min, then the remaining free platelets were assayed by flow cytometry. The platelet aggregate formation by TTP plasma in acute or remission phase was compared with the control group, mean ± SD.

% platelet aggregates induced by 4% plasma45 ± 6% (n = 8)79 ± 16%* (n = 4)70 ± 12%* (n = 4)

The platelet aggregates produced by TTP plasma were also markedly more stable than with normal plasma. As shown in Fig. 5, the platelet aggregates with TTP plasma were much more resistant to dissociation after 1 : 20 dilution than normal plasma, and this was seen in both acute (exacerbation) and remission phases. Filtration of the TTP plasmas through 0.1 μm to deplete MP of size ≥0.1 μm reduced stability of aggregates as seen by reduced t1/2. Although filtration through 0.1 μm filter removed significant vWf/ristocetin activity from TTP plasma (approximately 30%), it is possible that this over-estimates MP-bound vWF (if significant non-specific retention of soluble unusually large vWf (ULvWf) occurs), or under-estimates (if significant vWf-EMP are < 0.1 μm and therefore pass through). The time course of dissociation with TTP plasma in exacerbation was similar to that of EMP (Fig. 4). These results indicate that EMP-bound vWf in TTP plasma may contribute to stabilizing platelet aggregates.

Figure 5.

Dissociation rates of platelet aggregates formed in TTP plasma. Platelet-poor plasma at 4% v/v final concentration from four different TTP patients and normal plasma was added to washed platelets in presence of ristocetin, without 0.1 μm filtration (solid black symbols) or with filtration (open symbols). Then degree of dissociation was measured at intervals as in the previous figure. Notice that filtration of TTP plasma resulted in 20–30% loss of ristocetin co-factor activity. For normal control plasmas, each data point is the mean of six different donors; for TTP plasmas, each data point is the mean of four different patients.

Restoration of platelet aggregation to vWD plasma by EMP

As shown in Fig. 6, plasma from four vWD patients was tested for vWf/ristocetin-dependent platelet aggregating activity. vWD plasma alone showed very weak activity, 6%. This activity increased significantly upon addition of low-dose Humate-P (0.1 U mL−1, final concentration), to 48%. Addition of low-dose EMP alone (5 × 106 mL−1, final concentration) to the vWD patient plasma also partially restored the ristocetin cofactor activity, to 34%. However, the combination of low-dose EMP with low-dose Humate-P appeared to be synergistic in restoring ristocetin cofactor activity to vWD plasma, giving 78% aggregation.

Figure 6.

Addition of EMP to vWD plasma restored ristocetin cofactor activity. As in Fig. 1, all samples were 50 μL final and contained washed normal platelets at 1 × 107 mL−1, to which was added 4% of the indicated plasmas, or EMP at 5 × 106 mL−1 final, or Humate-P at 0.1 U mL−1 final, in the various combinations shown. Error bars show mean ± SD; n = 4.

Multimer analysis of EMP-bound vWf

The above findings suggest that the observed effects of EMP on ristocetin-induced platelet aggregation may be because of the presence of ULvWf on EMP. To test this conjecture, we performed multimer analysis on vWf-EMP compared with vWF from normal plasma, TTP plasma, vWD plasma, Humate P, and platelet microparticles (PMP). As shown in Fig. 7, multimer analysis confirms the conjecture that EMP-bound vWf (lane 4) has the largest multimers, followed by Humate-P (lane 5), TTP (lane 2), PMP (lane 6), normal plasma (lane3), and vWD plasma (lane 1). Notice that vWf multimers from a vWD type I patient contain the fewest bands (lane 1). Notice also that the vWf multimer bands from the EMP sample are not clearly separated. We repeated this experiment many times under different conditions, always with similar results (Fig. 7, lane 4). In contrast, vWf multimers from PMP were always clearly separated (lane 6). This peculiarity may reflect the distinctive functional properties of EMP-bound vWF. The reason for this behavior is not fully understood, but we speculate that membrane-bound vWf multimers may be tightly bound to other membrane-associated proteins or phospholipids, and therefore cannot be fully dissociated by sodium dodecyl sulfate, resulting in more diffuse bands.

Figure 7.

Multimer analysis of vWf from different sources. As described in Methods and materials section, samples were applied to 0.8% agarose gel electrophoresis followed by Western blotting. Lane 1: plasma from a type I vWD patient (5 μL); lane 2, plasma from a TTP patient in acute phase (5 μL); lane 3, plasma from a normal control (5 μL); lane 4, renal EMP (2 × 107 counts); lane 5, Humate-P (0.01 U); lane 6, platelet microparticles (PMP) (2 × 107 counts).


Since the initial report on elevated EMP in lupus anticoagulant patients [9], EMP levels have been shown to reflect many thrombotic and inflammatory conditions, as recently reviewed [2]. Jimenez et al. [33] demonstrated that the surface antigens of EMP are distinctive depending on the type of EC injury, as in apoptosis vs. activation: in activation, the number of CD62E+ or CD54+ EMP is much higher than that of CD31+ EMP and vice versa in apoptosis. As referenced above, EMP have been shown to exhibit procoagulant activity and to carry TF. Sabattier et al. [39] has shown that EMP bind to monocytes to induce TF expression, and we have shown that EMP interaction with monocytes is predominantly to the CD54+ subset of EMP and results in enhanced transmigration of monocytes through endothelial junctions [40]. However, although EMP assay is emerging as a marker of endothelial dysfunction in many disorders, their functional roles have not been extensively investigated.

The present study demonstrates that vWf+ EMP bind to platelets to induce formation of platelet aggregates, chiefly via a vWf-dependent pathway, and that platelet aggregates induced by EMP are more stable (more resistant to dissociation) than those induced by free soluble vWf. This appears to be explained by the presence of membrane-bound ULvWf. However, EMP express other adhesins which could further contribute to the stability of EMP-platelet complexes, such as ICAM-1, VCAM-1, E-selectin, P-selectin, PECAM-1, or vitronectin receptor [2]. Some of these may have counter-receptors on the platelet side [25,41], potentially stabilizing the EMP-platelet complexes and their aggregates, in concert with vWf on EMP.

Our results also showed that in four TTP patients, their ristocetin-induced vWf-dependent platelet aggregate formation remained elevated even 1–2 weeks of remission. The dissociation rates of the platelet aggregates induced by TTP plasma were also significantly prolonged, even in remission, compared with normal plasma or Humate-P. Multimer analysis showed that Humate-P had larger multimers than found in TTP plasma, raising the question of whether multimer size alone is the critical factor in TTP pathology. It is possible that TTP plasma contains additional active substances such as vWF+ EMP, which further enhance the stability of platelet aggregates in TTP. This suggests that our understanding of abnormally active vWf in TTP is incomplete, and that vWF+ EMP may be an important factor now overlooked.

Some limitations of this study must be pointed out. The ristocetin-based assay, which is independent of calcium, is not physiological, and therefore these results may not strictly apply in vivo. Increasing calcium concentrations seem to increase the effect of vWf plus ristocetin. It is possible that PBS may lower the free calcium levels of citrated plasma. This may result in decreased response of platelet aggregate formation induced by plasma plus ristocetin. However, this effect is likely to be minimal at low level of calcium (50 μm) (W. Jy, unpublished data). On the other hand, the ristocetin cofactor activity has been used for many years to evaluate vWF activity in vWD patients, confirming its physiological relevance. Another limitation is that the concentration of EMP used in these experiments was two- to fourfold higher than usually found in the circulation of TTP patients [13]. However, local concentrations at sites of endothelial injury may be expected to be much higher than in the general circulation. Furthermore, we observed synergy between EMP and free vWf, as low-dose of either had only weak effects but the combination gave very pronounced effects, see Fig. 6. These points support the physiological relevance of this study.

We found that measuring platelet aggregate formation and dissociation by flow cytometry was simple, economical, and reproducible. This suggests potential clinical applications, and we are working to refine these methods for this purpose.

We also demonstrated that addition of EMP to plasma from Type I vWD patients can restore the deficient response to ristocetin (Fig. 6). We also showed that mixing EMP with Humate-P synergistically enhanced the platelet aggregating action of Humate-P. These data indicate that EMP may play an important role in hemostasis. We postulate that EMP released during vascular injury may contribute to arrest of bleeding by rapid interaction with platelets via membrane-associated vWf multimers and adhesins to stabilize platelet aggregates in the microenvironment.


This project was supported by a grant from the Wallace H Coulter Foundation.