M. N. Abid Hussein, Department of Clinical Chemistry, B-1-235, Academic Medical Center, PO Box 22660, 1100 DD Amsterdam, the Netherlands. Tel.: +31 20 566 3946; fax: +31 20 609 1222; e-mail: firstname.lastname@example.org
Summary. Background: Endothelial activation and dysfunction are associated with several diseases. However, hardly any specific markers are available. Microparticles (MP) from endothelial cells (EC; EMP) were reported in patient groups and healthy individuals. The antibodies used to detect EMP, however, were mainly directed against antigens without EC specificity. Objectives: We evaluated the antigens on EC and EMP to establish proper markers for EMP detection. Methods: EMP were isolated from supernatants of resting and interleukin (IL)-1α activated human umbilical vein EC (HUVEC; n = 3; 0–72 h), stained with annexin V and monoclonal antibodies, and analyzed by flow cytometry. Human platelet-MP (PMP), the main MP population in plasma, were prepared in vitro. EMP and PMP were studied in plasma from systemic lupus erythematosus (SLE) patients (n = 11) and healthy individuals (n = 10). Results: Platelet–endothelial cell adhesion molecule-1 (PECAM-1), αν and β3 were constitutively exposed on HUVEC, but (almost) absent on EMP (<15% positive for αν and β3), or only exposed on a subpopulation (PECAM-1; 30–60%). Activated HUVEC (>80%) and (subpopulations of) EMP exposed E-selectin and tissue factor. PMP strongly exposed PECAM-1, β3, and glycoprotein (GP)Ib (CD42b), but not αν or E-selectin. GPIb and P-selectin (CD62P) were absent on EMP. Plasma samples contained 0.5% MP staining for E-selectin and/or αν. Plasma from one SLE patient contained E-selectin exposing MP (21%), but little αν-positive MP. Conclusions: EC release EMP in vitro. The antigenic phenotype of EMP released from resting and IL-1α-stimulated EC differs among each other as well as from resting and stimulated EC, respectively. E-selectin exposed on IL-1α-stimulated EC is a valid marker for EMP detection ex vivo to establish endothelial cell activation.
In physiological conditions, endothelial cells (EC) play an important role in homeostasis of the blood. This homeostasis is lost during pathological conditions, at least in part by increased exposure of procoagulant and proadhesive antigens on their surface. For example, only activated EC expose tissue factor (TF), the initiator of coagulation in vivo, and E-selectin, which facilitates reversible adhesion of white blood cells as part of the inflammatory response [1–4]. Endothelial dysfunction is associated with several disease states such as pre-eclampsia, thrombotic thrombocytopenic purpura (TTP), diabetes, systemic lupus erythematosus (SLE), lupus anticoagulant, atherosclerosis, inflammation, hypertension, and coronary artery disease [5–10]. At present, there are only a few markers for the detection of endothelial activation and/or dysfunction ex vivo such as von Willebrand factor (VWF) and soluble (s)E-selectin [11–13].
In vitro, activated EC show surface blebbing and the subsequent shedding of small vesicles (microparticles; MP) [14–18]. Recent studies report the presence of endothelial cell-derived microparticles (EMP) in peripheral blood from patients with lupus anticoagulant, TTP, acute coronary syndromes, and even in blood of healthy individuals [14,15,19]. However, in most studies the identification of EMP was based on the presence of surface antigens that are not exclusively exposed on EC, such as the platelet–endothelial cell adhesion molecule-1 (PECAM-1; CD31) or the vitronectin receptor (ανβ3; αν, CD51; β3, CD61). In addition, a recent study showed that MP from human erythrocytes significantly differed in their antigenic composition from their corresponding parent cells . Thus, a comprehensive characterization of the antigenic composition of EMP is required to identify accurately such vesicles in mixed populations of MP of various cellular origins as present in, for example, the venous blood of healthy individuals and patients. The aim of the present study was to compare the antigenic phenotype of EC and EMP under resting and activation conditions to establish reliable markers to quantify EMP.
Materials and methods
Reagents and assays
Medium M199, penicillin, streptomycin, and l-glutamine were obtained from Gibco BRL, Life Technologies (Paisley, UK). IgG1–FITC and IgG1–PE (clone X40), CD31–PE (clone WM-59, IgG1), CD34–PE (clone My10, IgG1), and CD61–PE (clone VI-PL2, IgG1) were obtained from Becton Dickinson (San Jose, CA, USA). CD42b–PE (clone CLB-MB45, IgG1), CD42b–FITC (clone CLB-MB45, IgG1), fetal calf serum (heat inactivated during 30 min at 56 °C; FCSi), normal mouse serum, and human serum albumin (HSA) were obtained from the Central Laboratory of the Netherlands Red Cross Blood Transfusion Service (CLB; Amsterdam, the Netherlands). CD61–FITC (clone Y2/51, IgG1) was from Dako A/S (Glostrup, Denmark). CD51–FITC (clone AMF7, IgG1), CD62P–FITC (clone CLB-Thromb/6, IgG1), CD54–PE (clone 84H10, IgG1) and CD62P–PE (clone CLB-Thromb/6, IgG1) were from Immunotech (Marseille, France). CD62E–FITC (clone 1.2B6, IgG1) was obtained from Serotec Ltd. (Kidlington, UK), CD62E–PE (clone HAE-1f, IgG1) from Ancell (Lausen, Switzerland), CD106–FITC (clone 1.G11B1, IgG1) from Calbiochem (La Jolla, CA, USA), CD141–PE (clone B-A35, IgG1) from Diaclone (Besançon, France), antitissue factor–FITC (4508CJ, IgG1) from American Diagnostics Inc. (Greenwich, CT, USA), and CD144–FITC (clone BMS158F1, IgG1) from MedSystems Diagnostics GmbH (Vienna, Austria).
Recombinant human interleukin (IL)-1α, human recombinant basic fibroblast growth factor (bFGF), and epidermal growth factor (EGF) were from Gibco BRL (Gaithersburg, MD, USA). Annexin V–APC was from Caltag Labs (Burlingame, CA, USA), collagenase (type 1A) from Sigma (St Louis, MO, USA), EDTA from Merck (Darmstadt, Germany), heparin (400 U mL−1) from Bufa BV (Uitgeest, the Netherlands), calcium ionophore A23187 from Calbiochem (Darmstadt, Germany), and trypsin from Difco Labs (Detroit, MI, USA). Human serum was provided by the Blood Bank Center (Leiden University Medical Center) and was heat inactivated during 30 min at 56 °C (HuSi). Tissue culture flasks were from Greiner Labortechnik (Frickenhausen, Germany) and gelatin from Difco Labs (Sparks, MD, USA).
Isolation and culture of human umbilical vein endothelial cells
Human umbilical vein endothelial cells (HUVEC) were collected from human umbilical cord veins by minor modifications of previously described protocols [21,22]. Briefly, umbilical cords were filled with 1 mg mL−1 collagenase in M199 and subsequently incubated in PBS (154 mmol L−1 NaCl, 1.4 mmol L−1 phosphate; pH 7.5) for 20 min at 37 °C. Detached cells were collected by perfusion with medium M199 supplemented with 10% HuSi. The cell suspension was centrifuged for 10 min at 180 g and 20 °C. Cells were resuspended in M199 (37 °C) supplemented with 10% HuSi, 2 mmol L−1 L-glutamine, 1 mg mL−1 penicillin, 0.1 mg mL−1 streptomycin, 0.5 µg mL−1 fungizone, 10 ng mL−1 EGF, 20 ng mL−1 bFGF, and 5 U mL−1 heparin. HUVEC were cultured in 25-cm2 tissue culture flasks coated with 0.75% gelatin (passage 0). Upon confluence, the HUVEC were transferred to 75-cm2 tissue culture flasks coated with 0.75% gelatin (passage 1). Cells were detached by trypsinization (0.05% w/v trypsin and 2.7 mm EDTA in PBS, pH 7.4), and transferred twice more to 75-cm2 tissue culture flasks coated with 0.75% gelatin (passage 3).
HUVEC stimulation and flow cytometric analysis
Upon confluence at passage 3, HUVEC were kept for 3–4 days in a resting state before stimulation with IL-1α (5 ng mL−1). IL-1α was prepared as stock solution (10 µg mL−1) in M199 and added (5 µL) to 10 mL culture medium. After incubation for various time intervals, culture supernatants were collected for MP analysis and the cells were harvested by trypsinization. After 4 min, trypsin was neutralized by 1% FCSi in PBS pH 7.4. The obtained HUVEC suspension was washed twice by centrifugation for 10 min at 180 × g and 4 °C, and resuspended in PBS/FCSi. HUVEC were then kept on melting ice for 15 min, centrifuged for 10 min at 180 × g and 4 °C and resuspended in PBS/FCSi. MoAbs (5 µL) were added to 45 µL cell suspension. For HUVEC staining, the final MoAb concentrations used were 0.5 µg mL−1 for IgG1–FITC and IgG1–PE, 0.06 µg mL−1 for CD31–PE, 0.5 µg mL−1 for CD34–PE, 20 µg mL−1 for CD51–FITC, 0.03 µg mL−1 for CD54–PE, 2 µg mL−1 for CD61–FITC, 1 µg mL−1 for CD62E–FITC and anti-TF–FITC, and 10 µg mL−1 for CD144–FITC. Dilutions of CD106–FITC and CD141–PE were 1 : 50 (v/v) and 1 : 10 (v/v), respectively. The cell suspension was incubated with MoAbs in the dark for 30 min and 4 °C. After incubation, the HUVEC were washed by addition of 1 mL PBS/FCSi, centrifuged for 10 min at 180 × g and 4 °C, and resuspended in 300 µL PBS/FCSi (on melting ice). In each sample, 5000 cells were analyzed in a FACScan flow cytometer with CellQuest software (Becton Dickinson) .
Isolation of MP
At various time intervals, culture supernatants were harvested and centrifuged for 10 min at 180 × g and 20 °C to remove whole cells. Subsequently, aliquots of the cell-free culture supernatant (250 µL each) were snap-frozen in liquid nitrogen and stored at − 80 °C. Alternatively, plasma samples from citrate-anticoagulated venous blood of SLE patients and healthy controls (with their informed consent) were collected and handled as described previously . Aliquots (250 µL each) were snap-frozen in liquid nitrogen and stored at − 80 °C. Before use, all samples were kept on melting ice to allow thawing for 1 h. After thawing, samples were centrifuged for 30 min at 17 570 g and 20 °C. Then, 225 µL of (MP-free) supernatant were removed. The remaining 25 µL (MP-enriched) suspension was diluted with 225 µL PBS containing 10.9 mmol L−1 trisodium citrate. MP were resuspended and again centrifuged for 30 min at 17 570 g and 20 °C. Again, 225 µL of supernatant were removed and MP were resuspended in the remaining 25 µL. For flow cytometry detection of platelet-MP (PMP) and EMP from SLE patients and controls, this MP suspension (25 µL) was diluted 4-fold with PBS/citrate (75 µL; pH 7.4).
Preparation of PMP in vitro
Venous blood (8.4 mL) from three healthy controls was collected (with their informed consent) into 1.6 mL 3.2% acid citrate dextrose solution (ACD; 85 mmol L−1 trisodium citrate, 11 mmol L−1 glucose, 7 mmol L−1 citric acid; pH 4.4). Blood was centrifuged for 15 min at 180 × g and room temperature, and platelet-rich plasma was collected. Platelet-rich plasma was centrifuged for 20 min at 1000 × g and room temperature. The supernatant was removed and the platelet pellet was gently resuspended in 10 mL Tyrode buffer (136.9 mm NaCl, 11.9 mm NaHCO3, 5.6 mm glucose, 1.0 mm MgCl2, 2.7 mm KCl and 0.36 mm NaH2PO4; pH 6.5) containing HSA (0.25% w/v) and EDTA (2.0 mm). This platelet suspension was again centrifuged for 20 min at 1000 × g and room temperature, and the supernatant was removed. The platelet pellet was resuspended in 0.5 mL Tyrode buffer (pH 7.4) containing 2 mm CaCl2 instead of EDTA. After further diluting the platelet suspension with 3.5 mL of Tyrode buffer (pH 7.4), platelets were removed, counted, and adjusted to approximately 2.0 × 105 µL−1. Subsequently, platelets were activated by addition of calcium ionophore A23187 (2.5 µm final concentration) at 37 °C (non-stirring conditions). After 20 min, EDTA (5 mm) was added, platelets were removed by centrifugation for 20 min at 1000 × g and room temperature. The supernatant, containing PMP, was used for flow cytometric analysis.
Flow cytometric analysis
MP samples were analyzed in a FACSCalibur flow cytometer (Becton Dickinson). Forward scatter (FSC) and side scatter (SSC) were set at logarithmic gain and MP were identified as described previously [24,25]. MP were identified on FSC, SSC, and binding of a MoAb. More than 80% of the events identified using these criteria also stained for annexin V (data not shown). Fluorescence thresholds for MoAb were set in terms of binding of isotype-matched control antibodies (all IgG1). Fluorescence was measured in the FL-1 channel (FITC), FL-2 channel (PE) and FL-4 channel (APC). MP (5 µL) were diluted with 35 µL PBS containing 2.5 mmol L−1 CaCl2 (pH 7.4) and 5 µL of (1–500 diluted in PBS) normal mouse serum. After incubation for 15 min at room temperature, 5 µL annexin V–APC (0.66 µg mL−1 final concentration) and 5 µL MoAb or isotype-matched control antibody (IgG1) were added. For MP analysis, antibody concentrations used were 0.5 µg mL−1 for IgG1–FITC and IgG1–PE, 0.06 µg mL−1 for CD31–PE, 0.5 µg mL−1 for CD34–PE, 1 µg mL−1 for CD42b–FITC and CD42b–PE, 10 µg mL−1 for CD51–FITC, 0.03 µg mL−1 for CD54–PE, 1 µg mL−1 for CD61–FITC, 1.6 µg mL−1 for CD62E–PE, 2.5 µg mL−1 for CD62P–FITC, 0.0625 µg mL−1 for CD62P–PE, 0.5 µg mL−1 for anti-TF–FITC and CD144–FITC. Dilution of CD61–PE was 1 : 100 (v/v), CD106–FITC and CD141–PE were both diluted 1 : 50 (v/v). The mixture of MP, normal mouse serum and MoAbs was then incubated for 15 min in the dark at room temperature. To remove the excess of free MoAb, 200 µL PBS/calcium buffer were added and the suspension was centrifuged for 30 min at 17 570 × g at 20 °C. Finally, 200 µL of supernatant were removed, and MP were resuspended with 300 µL PBS/calcium. All samples were analyzed for 2 min.
Patients and healthy controls
In the present study, 11 SLE patients (all women) were included, all of whom fulfilled the revised criteria of the American College of Rheumatology for the diagnosis of SLE . Their age was 42 years (median; range 23–64). The SLE Disease Activity Index (SLEDAI)  was 9 (median; range 0–22). As controls, 10 age-matched women were included. The study fulfilled the guidelines of the Medical Ethical Committee of the Slotervaart Hospital.
Data were analyzed with Prism (3.0) for Windows. For direct comparison of the binding of MoAbs to HUVEC and EMP, paired t-tests were used. To compare the differences in MoAb binding to plasma samples from SLE patients and healthy volunteers, the Mann–Whitney U-test was used. Two-tailed significance levels (P) are presented. Differences were considered statistically significant at P < 0.05.
Antigen exposure on resting and IL-1α-activated HUVEC
HUVEC were incubated for various time intervals up to 72 h with and without IL-1α (5 ng mL−1). Figure 1 shows representative dot plots of the surface antigen exposure of PECAM-1 and E-selectin. Both resting (Fig. 1A) and activated (Fig. 1B) HUVEC exposed PECAM-1. In contrast, E-selectin was exposed only on the activated HUVEC (Fig. 1D vs. C). The overall data are summarized in Fig. 2 and in Table 1. As is evident from Fig. 2, PECAM-1 (CD31), αν (CD51), and β3 (CD61) are exposed on all HUVEC independent of their activation status (Fig. 2B,D,F). About 20% of the resting HUVEC exposed TF (Fig. 2J), whereas E-selectin was not exposed (Fig. 2H). Three hours after addition of IL-1α (the first measuring period), HUVEC exposed TF and E-selectin. The exposure of both antigens was transient and gradually diminished after 12 h. Because the maximal antigen exposure of inducible antigens on the HUVEC occurred 12–24 h after addition of IL-1α, we summarized the overall data for all studied antigens in Table 1 at that activation period. From Table 1, it is apparent that the antigens PECAM-1, αν, β3, GP105-120 (CD34), vascular endothelial (VE-) cadherin (CD144), and to a lesser extent intercellular adhesion molecule (ICAM; CD54), vascular cell adhesion molecule-1 (VCAM-1; CD106) and thrombomodulin (CD141), were all exposed on most of the HUVEC regardless of their activation status, although the surface exposure of ICAM and VCAM-1 increased upon cell activation. For other antigens studied, i.e. E-selectin and TF, the surface exposure was inducible upon activation of the HUVEC.
Table 1. Antigen exposure on human umbilical vein endothelial cells (HUVEC) and microparticles from endothelial cells (EMP) in the absence or presence of interleukin (IL)-1α
Next, the antigen exposure on EMP, obtained from resting and activated HUVEC, was analyzed 12 h after addition of IL-1α. Figure 3 shows that a subpopulation of EMP from resting and stimulated HUVEC exposed PECAM-1 (Fig. 3A,B). Approximately 20–30% of the EMP, released from resting HUVEC, exposed PECAM-1. Upon activation, the percentage of PECAM-1-exposing EMP increased to almost 60% (Fig. 3B). Figure 3C,D show the exposure of E-selectin on EMP from resting (Fig. 3C) and activated (Fig. 3D) HUVEC. Whereas EMP from resting HUVEC hardly stained for E-selectin (Fig. 3C), EMP from HUVEC strongly stained for this antigen upon cell activation.
Figure 4 summarizes the exposure of PECAM-1, αν, β3, E-selectin and TF on EMP from resting and IL-1α-activated HUVEC. The antigens that were constitutively exposed on the HUVEC (PECAM-1, αν and β3), were exposed only on a subpopulation of the EMP (PECAM-1, αν) or even absent (β3). The antigens that were inducible on the HUVEC, i.e. E-selectin and TF, were virtually absent on EMP derived from resting HUVEC, but as also shown in Fig. 3, the EMP from activated HUVEC strongly exposed E-selectin up to 72 h. On these EMP, TF was present but barely detectable by flow cytometry. Table 1 presents the percentages of EMP that exposed the indicated antigens. The antigenic phenotype of the EMP differed remarkably from the HUVEC and depended on the activation status of the parent cells. Not only αν and β3, but also GP105-120, VCAM-1, CD141 and CD144 were not or hardly detectable on the EMP, regardless of the activation status of the HUVEC. Interestingly, not only E-selectin but also ICAM was exposed on the EMP, albeit to a lesser extent. As with E-selectin, this marker occurred only on the EMP from activated HUVEC.
Comparison of the antigenic profile of PMP prepared in vitro with EMP
Subsequently, we compared the antigenic profile of EMP and PMP prepared in vitro (Fig. 5). Both EMP and PMP exposed PECAM-1 (Fig. 5A and B, respectively). In contrast to the EMP, PMP strongly stained for GPIb (CD42b), GPIIIa (CD61) and P-selectin (CD62P) (Fig. 5C,G,K vs. D,H,L, respectively). αν (CD51) was nearly absent on EMP (Fig. 5E) and PMP (Fig. 5F). The only marker that positively and selectively identified EMP was E-selectin (Fig. 5I), which was absent on PMP (Fig. 5J).
Detection of EMP and PMP in plasma samples of SLE patients and healthy individuals
Based on our current observations, we reinvestigated the presence of EMP and PMP in plasma samples from patients with SLE and healthy individuals. In order to detect EMP and PMP in these samples, MP were isolated and stained with combinations of MoAbs directed against αν and E-selectin, and GPIb and GPIIIa, respectively. αν had previously been used to quantify MP . As shown in Table 2, most by far of the cell-derived MP in plasma samples studied from SLE patients and controls (median 66%) strongly stained for GPIb (CD42b), GPIIIa (CD61), or a combination of these two MoAbs. In contrast, hardly any MP stained for either αν, E-selectin, or a combination of these two MoAbs (Fig. 6B,D,F). Also, no ICAM-positive events were detected in these samples (data not shown). Interestingly, an E-selectin-positive subpopulation of MP occurred in plasma from the SLE patient who had the highest SLEDAI (22). Of the E-selectin-positive EMP, only 12% double-labeled for αν (Fig. 6A,C,E). These findings indicate that MP of endothelial origin indeed occur in vivo.
Table 2. Percentages of microparticles from endothelial cells (EMP) and platelet microparticles (PMP) in plasma samples from systemic lupus erythematosus (SLE) patients (n = 11) and healthy controls (n = 10)
The present finding that the antigenic phenotype of EMP differs considerably from the HUVEC suggests that a sorting of membrane proteins occurs during membrane vesiculation. Interestingly, a recent study showed that calcium ionophore-activated erythrocytes release microparticles that antigenically differ from their parent cells . The selective sorting of membrane proteins into MP is likely to be a general phenomenon, which is not cell type specific.
In vitro cultured endothelial cells not only release MP upon activation with IL-1α, but also upon activation with tumour necrosis factor (TNF)-α[14,15,17], TNF-α with cycloheximide or camptothecin , or lipopolysaccharide . The phenotype of the released EMP may be dependent on the agonist used to activate the parent cells. For instance, PMP expose the fibrinogen receptor GPIIb-IIIa (αIIbβ3) when released in vitro upon activation of platelets. This receptor is in its fibrinogen-binding conformation upon platelet activation by thrombin plus collagen, but not upon activation by the complement C5b−9 complex . Whether the antigenic composition of EMP is agonist dependent remains to be investigated. Because the stimuli involved in EMP release in vivo are unknown, the antigenic phenotype of EMP in vitro vs. in vivo may then differ as well.
In accordance with Combes et al. , we also found that resting HUVEC constitutively exposed antigens such as PECAM-1, αν and β3, and that EMP derived there from exposed PECAM-1. Thus far, EMP have been identified ex vivo, i.e. in plasma samples, using combinations of MoAbs directed against PECAM-1 plus αν, or PECAM-1 plus ανβ3[14,15,19]. In our present study, however, we found αν and β3 exposed only on a minor subpopulation (αν) or absent (β3) on EMP prepared in vitro. Our present findings confirm earlier reports on the occurrence of EMP in vivo. However, earlier studies in which PECAM-1 plus αν, or PECAM-1 plus ανβ3 were used to detect EMP ex vivo, may have overestimated their presence, since most of the MP ex vivo are of platelet origin which expose high levels of PECAM-1 and β3.
In agreement with Combes et al. , we also found no evidence for E-selectin-exposing EMP in plasma samples of SLE patients and healthy individuals. There was one exception, however. One patient, actually the one with the highest SLEDAI in our study, had a subpopulation of E-selectin-exposing MP, suggesting that in this particular patient the endothelium may have been more activated than in the other patients. However, this could not be confirmed because we measured the plasma concentration of VWF and found it in this patient to be not significantly higher than in other SLE patients (data not shown). We have no explanation yet for the low percentage PMP in the plasma of this SLE patient when compared with other SLE patients. Our present findings suggest that part of the soluble E-selectin, which is known to be elevated in plasma of patients with SLE, is MP associated. Whether this E-selectin originates from the parent cell during MP formation or resembles originally soluble E-selectin subsequently bound to the MP from other cells, is also open for discussion. Only some 12% of the E-selectin-positive MP exposed αν, which supports our in vitro data that this is not a proper marker for EMP detection ex vivo. Since we found E-selectin exposing MP in plasma from only one out of 11 SLE patients, we also analyzed plasma samples from severely Dengue virus infected patients. These patients are known to suffer from increased vascular permeability  and sera from such patients contain antibodies that directly trigger endothelial damage . We found that plasma from two of these three patients contained a subpopulation of 8% and 17% of E-selectin-exposing MP. About 90% of this subpopulation did not double stain for αν (data not shown). These data support our finding that αν may not be a proper marker to detect EMP.
Like E-selectin, TF was strongly inducible on the HUVEC. The exposure of TF peaked at 12 h and diminished afterwards. Despite the fact that only a small subpopulation of EMP exposing TF was observed, reconstitution of these EMP strongly generated TF- and factor VII(a)-mediated thrombin generation in plasma (data not shown). This indicates that functional TF must be exposed on these vesicles. Possibly, the antigenic density, i.e. the number of exposed TF molecules, is too low to be detected by flow cytometry. Of course, this may also hold true for all other surface markers in the present study.
We also studied the exposure of GPIb (CD42b) on HUVEC (data not shown). Two different MoAbs failed to detect exposure of this antigen. However, a third MoAb gave conflicting results. Therefore, we are uncertain about the exposure of GPIb on HUVEC.
The present study shows that (i) EMP are released from HUVEC, (ii) the antigenic surface composition of EMP differs from HUVEC, and (iii) the surface composition of EMP is highly dependent on the activation status of the parent cells. EMP and PMP share various important surface antigens, which implies that the measurement of EMP by flow cytometry should be performed carefully, since most by far of the MP found in plasma samples of patients and controls are of platelet origin. E-selectin can be used as a specific marker to detect EMP ex vivo but may underestimate their presence because only a subpopulation of the EMP stained positively for E-selectin and only upon activation of the parent cell. Whether ICAM-1 can also be used as a specific marker for EMP detection ex vivo remains to be established, since this adhesion receptor also occurs on lymphocytes and monocytes.