- Top of page
- PATIENTS and METHODS
Endothelial injury is believed to be a key initiating event in the pathogenesis of thrombotic thrombocytopenic purpura (TTP), leading to platelet activation and formation of platelet-rich thrombi in microvasculature. However, the nature of endothelial injury in TTP is poorly defined and clinical assays to rapidly and reliably monitor endothelial damage are not readily available. Using flow cytometry, we measured endothelial microparticles (EMPs) generated from cultured renal and brain microvascular endothelial cells (MVECs) during activation and apoptosis, and evaluated the effect of TTP plasma on them. EMPs were measured using positivity for monoclonal antibodies (mAbs) CD31 and CD51, and their procoagulant activity was assessed using a Russell viper venom assay. Both cell lines generated procoagulant EMPs when cultured with inducers of activation (tumour necrosis factor alpha; TNF-α) or apoptosis (mitomycin C). TTP plasma induced a five- to sixfold increase of EMP generation and a two- to threefold increase of procoagulant activity in cultured brain and renal MVECs. TTP plasma induced a threefold and 13-fold increase of intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) expression, respectively, on renal MVECs. Procoagulant activity tended to parallel EMP numbers. The effect of TTP plasma on cell viability was similar to that of TNF-α, implying that it induced activation rather than apoptosis. Control plasma and idiopathic thrombocytopenic purpura (ITP) plasma had little effect. In the clinical study, EMP assay of blood from acute TTP patients showed levels markedly elevated compared with normal controls, but values returned to normal in remission. In conclusion, TTP plasma activated and induced injury to MVECs in culture, judged by production of EMP and expression of activation markers. Released procoagulant EMP may play a role in the pathogenesis of TTP. Assay of EMP may be a useful marker of disease activity and endothelial injury in TTP and possibly other thrombotic disorders.
In thrombotic thrombocytopenic purpura (TTP), formation of platelet-rich microthrombi in microvasculature leads to neurological dysfunction, renal failure and microangiopathic haemolytic anaemia (Ridolphi & Bell, 1981). What promotes platelet activation and formation of platelet-rich thrombi is not entirely clear, but it is believed that insult to the endothelium is an initiating event. It has been suggested that unusually large multimers of von Willebrand factor (VWF) released from disturbed endothelium promote platelet adhesion and aggregation in TTP (Moake, 1982; Moake & McPherson, 1989; Moake et al, 1995). Alternatively, Lian et al (1979, 1984) reported a platelet aggregating factor released during acute TTP that appeared to be inhibited by normal immunoglobulins.
In the normal haemostatic state, microvascular endothelium is thromboresistant, but upon activation or insult it loses its anticoagulant properties and becomes procoagulant (Rodgers, 1988). Loss of thrombomodulin, tissue factor pathway inhibitor and heparan sulphate has been demonstrated in apoptotic endothelium (Bombeli et al, 1997). Procoagulant apoptotic bodies have also been demonstrated in cultures from human umbilical vein endothelial cells (HUVECs) undergoing apoptosis (Casciola-Rosen et al, 1996; Bombeli et al, 1997). It has been proposed that endothelial cells may undergo apoptosis in TTP, based on the finding that TTP plasma induced apoptosis in cultured brain- or kidney-derived microvascular endothelial cells (MVECs) but not in large-vessel ECs (Laurence et al, 1996; Laurence & Mitra, 1997; Mitra et al, 1997).
HUVECs release microparticles (EMPs) within minutes of exposure to complement C5b-9 (Hamilton et al, 1990). More recently, EMPs from HUVECs exposed to tumour necrosis factor alpha (TNF-α) for 24–48 h were found to express tissue factor (TF), in addition to constitutive antigens CD31 and CD51, and exhibited functional procoagulant activity, as defined by an assay using factor VII (FVII)-deficient plasma (Combes et al, 1999). Elevated EMPs were also detected in patients with lupus anticoagulant (LA).
The present study is an extension of our long interest in platelet microparticles (PMPs) (Jy et al, 1992; Horstman, 1999) and platelet activation in TTP (Ahn et al, 1996; Schultz et al, 1998; Valant et al, 1998). Endothelial damage may underlie the platelet activation reported in those studies. In this study, we examined the effect of TTP plasma on cultured renal and brain MVECs with emphasis on the EMPs produced and their procoagulant activity. In addition, we measured and monitored EMP in the plasma of two patients with active TTP.
- Top of page
- PATIENTS and METHODS
Representative examples of EMP assay using flow cytometry are shown in Fig 1. EMP generation from RMVECs exposed to normal control plasma (top) vs. plasma from a TTP patient (bottom) are illustrated in Fig 1A (left). The EMP population appears in region 4 at the lower right of each histogram; the counts are listed as ‘EMP #’ for these examples. An increase in EMP was seen in RMVECs exposed to TTP plasma compared with control: the EMP count in the supernatant increased from 146 to 9434 after exposure to TTP plasma, as measured by CD31. CD51 EMP increased only twofold, from 1019 to 2011.
Figure 1. (A) EMPs released from MVECs following exposure to normal plasma (top) or TTP plasma (bottom). The particles counted as EMPs appear in the lower right rectangle (region 4) of each histogram. The population counted in this example is listed as ‘EMP #’. Renal MVECs exposed to TTP plasma show increased EMPs (lower) compared with controls (upper). (B) EMPs in peripheral blood of a normal control (top) and a TTP patient (bottom). The figure shows the EMP pattern of normal control plasma (top) and TTP plasma (bottom). Notice that the CD31+ EMPs distinguish more sharply between the TTP plasma and the normal control than the CD51 marker.
Download figure to PowerPoint
Figure 1B shows examples of EMP counts from the PPP of a normal control (top) and a TTP patient (bottom). A large elevation in CD31 EMP number was seen in the plasma of the TTP patient (count 5260) compared with control plasma (count 543). The elevation of CD51 EMPs was only modest (731 for the TTP patient in this example vs. 549 for the control). The dense CD31 population in region 2 of the normal control were platelet microparticles (PMPs), i.e. are CD42+ and were not counted as EMPs. The PMPs were greatly reduced in the patient samples, presumably owing to severe thrombocytopenia.
Baseline studies of the effect of TNF-α and mitomycin C on cultured renal and brain MVECs
To establish the response of our MVECs to well-characterized agents, we used a known inducer of activation, TNF-α (Nawroth & Stern, 1986), and of apoptosis, mitomycin C (Hebert et al, 1998). Generation of EMPs was investigated in both renal and brain MVECs. Both cell lines released EMPs similar to previous work in HUVECs (Combes et al, 1999). EMPs released by mitomycin C were more abundant than those released by TNF-α in our experiments, as seen in following figures.
Figure 2A shows EMP release following TNF-α and mitomycin C incubation, as quantified by CD31 and CD51 labelling. Both agents induced EMP generation from renal and brain MVECs, but CD31-labelled EMPs were more prominent than CD51-labelled EMPs in both cell lines. The procoagulant activity of the same EMPs assayed by the RVV is shown in Fig 2B. TNF-α and mitomycin C both induced procoagulant EMPs from both cell lines, but mitomycin C generated greater activity. When EMPs were removed by filtration (0·1 μm), procoagulant activity was largely abolished, which demonstrated that the activity resides on EMPs. Renal MVECs consistently produced greater procoagulant activity than brain MVECs, again in line with EMP numbers by flow cytometry (Fig 2A).
Figure 2. A. EMPs liberated from cultured MVECs in response to TNF and mitomycin C at 48 h. Values shown are the mean plus standard error of four experiments. Axis scale ‘EMP Counted’ are raw data, i.e. to obtain EMPs per ml of MVEC supernatant, multiply by 1178 (see Patients and methods). It can be noted that (i) CD31(+) EMPs were three- to 10-fold more numerous than CD51(+), except in controls; (ii) mitomycin C was ≈twofold more potent than TNF in eliciting EMP; and (iii) renal MVECs were ≈twofold more sensitive than brain MVECs. B. Procoagulant activities of the EMPs. After removing the supernatants from the MVECs, a portion of each was tested by flow cytometry (Fig 2A) and a portion (usually 15 μl) was tested by RVV procoagulant assay as described in the Patients and methods section. The complete experiment shown was repeated three times. It was consistently observed that (i) renal cells released greater activity than brain cells with both agonists, (ii) mitomycin C gave greater activity than TNF, (iii) filtration of the mitomycin C-treated MVEC supernatant through 0·1 μm reduced the activity to less than the supernatant of the normal control-treated MVEC supernatantnt. All of these observations were significant at P < 0·05 or better by paired t-test. C. Procoagulant properties of EMP measured in the presence/absence of expgenous phospholipid (PL). Upper curve: EMPs shorten clotting time in the presence of saturating PL, indicating that EMPs accelerate intrinsic or extrinsic pathways of clotting. The procedure followed was the same as in the Patients and methods except that 50 μl of buffer replaced 50 μl of RVV, and PL (10·8 μg/ml) was added. Lower curve: in the absence of PL, EMPs shorten the clotting time by RVV, indicating that EMPs provide PL surface to accelerate clotting, as also shown in Fig. 2B.
Download figure to PowerPoint
Figure 2C shows the clotting activity of a cell supernatant in the absence of RVV but in the presence of ample phospholipid (10·8 μg/ml). In this example experiment, the clotting time with heparin-free medium alone was 440 s and with the standard volume (15 μl) of MVECs, the supernatant fell to 220 s. This contrasts with the normal RVV assay (lower curve) designed to measure platelet factor 3 (PF3)-like procoagulant activity. The purpose of this experiment was to see if other pathways (contact- or TF-initiated) contributed to a significant shortening of the clot time on top of the PF3 activity. As the times in the upper curve were long compared with the lower curve, this activity contributed only a small fraction of the measured PF3 activity at the standard volume of 15 μl.
To assess the effect of mitomycin C, TNF-α and serum deprivation on BMVECs and RMVECs, all cultures were assessed for viability by trypan blue dye exclusion and by TUNEL assay to detect DNA fragmentation. As expected, mitomycin C induced apoptosis, as judged by TUNEL-positive results (65% positive in BMVECs and 70% in RMVECs). Viability for BMVECs was 5% and 3% for RMVECs. Stimulation by TNF-α gave no significant change in cell count or viability, as detected by trypan blue dye exclusion (mean viability 94% for RMVECs and 93% for BMVECs), nor was DNA fragmentation observed in the TUNEL assay (2% for RMVECs and 1% for BMVECs). Conversely, deprivation of serum/growth factor led to findings of apoptosis similar to mitomycin C: 50% positive in BMVECs and 65% in RMVECs by TUNEL assay, and viabilty 12% for BMVECs and 8% for RMVECs. This indicates the importance of culture conditions when assessing putative inducers of cell death/apoptotsis.
Effect of ITP, TTP and control plasma on cultured renal and brain MVECs
Filtered plasma from normal controls and patients with ITP or TTP was added at 20% concentration to confluent cultured RMVECs or BMVECs and, after 48 h, supernatants were collected and prepared for flow cytometry and procoagulant assay as described in Patients and methods. As shown in Fig 3A, renal EMPs measured by CD31 expression were similar between controls (1·19 × 106/ml) and ITP (0·84 × 106/ml), but CD51 expression was significantly lower in controls (1·20 × 106/ml) than in ITP (2·01 × 106/ml). Alternatively, RMVECs incubated with TTP plasma released six- to eightfold more CD31 EMPs than normal or ITP plasma. Release of EMPs measured by CD51 was less dramatic, being 1·8- to threefold compared with normal or ITP plasma. Also shown in Fig 3A is the response of BMVECs, which also responded strongly to TTP plasma, showing an increase in CD31 EMPs of ≈sixfold relative to normal controls and a ≈twofold rise in CD51 EMPs.
Figure 3. EMPs released from MVECs following exposure to TTP plasma. A. EMPs per ml released from renal and brain MVECs after incubation with plasma of normal controls and in patients with ITP and TTP, measured by both markers (CD31, CD51). Errors bars are standard error. B. The corresponding procoagulant activity by RVV assay. Error bars are standard deviation.
Download figure to PowerPoint
Cell viability and apoptosis detection in TTP plasma-exposed MVECs
Renal and brain MVECs were cultured to confluence and then incubated for 48 h in 20% ITP, TTP or control plasma. Cells were then detached with trypsin/EDTA and viability assessed by trypan blue dye exclusion. Viability in all cultures incubated with all plasmas was 95% to 100%. Parallel studies to determine induction of apoptosis by TUNEL assay were also performed. Incubation with plasma of ITP, TTP or normal subjects failed to induce apoptosis in either BMVEC or RMVEC cultures. The percentage of cells positive for TUNEL was 0% to 2% in both renal and brain cultures exposed to all plasmas.
As a positive control for apoptosis, both RMVECs and BMVECs treated with mitomycin C gave clear positivity in the TUNEL assay. To address the possibility of early stage apoptosis during long-term exposure to TTP plasma, viability and cell counts were assessed after 72 h incubation. Measures of viability remained similar to those at 48 h (95–100% for both cell lines treated with control or TTP plasma) and the number of cells in cultures treated with TTP plasma (157 650 ± 4000 for RMVECs and 140 000 ± 3500 for BMVECs) was only slightly decreased compared with those exposed to normal plasma (170 000 ± 5500 for RMVECs and 165 000 ± 4500 for BMVECs). To further corroborate our finding of endothelial activation by TTP plasma rather than apoptosis, renal MVECs incubated with TTP plasma were assessed for ICAM-1 and VCAM-1, both of which are regarded as activation markers (Augustin et al, 1994; Haraldsen et al, 1996). There was a ≈threefold increase of ICAM-1 expression in MVECs exposed to TTP plasma compared with MVECs incubated with normal plasma for 24 h (3·44 ± 0·02 and 1·56 ± 0·04 respectively). Positivity for VCAM-1 was more dramatic, increasing almost 13-fold in TTP-treated (19·02 ± 1·04) MVECs compared with controls (1·49 ± 0·05).
The procoagulant assay by RVV is summarized in Fig 3B. The TTP plasma induced greater coagulant activity than ITP or control plasma. The coagulant activity from BMVECs was significantly less than from RMVECs (P < 0·001). The same trend was seen in EMP counts by CD31 (Fig 3A), however, results in Fig 3A were almost, but not quite, significant.
Detection of EMPs in controls and TTP patients
In normal controls (n = 45), the mean level of CD31+ EMPs was 734 ± 227 per μl. Both acute TTP patients had significantly elevated EMPs compared with controls. During the acute phase, CD31+ EMPs reached 5260 per μl in case 1 and 1583 per μl in case 2. In remission, the levels had fallen to normal (752 and 674 per μl respectively).
Sequential studies of TTP case 1 were conducted through the entire clinical course, with results shown in Fig 4. This patient presented with mental confusion, thrombocytopenia (platelets 11 × 109/l), microangiopathic haemolytic anaemia (Hb; 8 g/dl), high lactate dehydrogenase (LDH; 2000 U/l) and numerous fragmented red blood cells (RBCs) in blood smears. CD31 EMPs were markedly elevated at ∼5000/μl. After two exchange plasmapheresis and daily FFP infusion, the patient became alert, orientated and platelets rose to > 100 × 109/l, while LDH fell. The EMP count also fell towards normal (< 1000/μl). On d 8, the patient had a line sepsis-associated relapse in spite of continuing infusion of FFP and exchange plasmapheresis. The platelet count again fell and LDH rose, as did the EMP count. Subsequently, when the patient entered remission, her EMP levels remained low; see Fig 4.
Figure 4. EMPs in the blood of a TTP patient over the course of acute phase, as related to other clinical measures. A. Platelet count (×109/l). B. Lactic dehydrogenase (LDH), in units/l. C. EMPs, numbers per μl of blood.
Download figure to PowerPoint
- Top of page
- PATIENTS and METHODS
The results obtained in this study demonstrate that (i) TTP plasma stimulates MVECs of renal and brain origin, inducing release of procoagulant EMPs, and (ii) a marked elevation of EMPs occurs in the acute phase of TTP and normalizes in remission. These findings lend further support to the hypothesis that endothelial damage is a key initiating event in TTP. Whether EMPs themselves contribute to the pathogenesis of TTP remains to be elucidated. It was noticed that CD31 was a more consistent marker of EMPs than CD51, in both in vitro and in vivo studies of normal controls and patients.
Combes et al (1999) focused on the morphological, immunological and functional characteristics of EMPs derived from TNF-α-treated HUVECs, observing that they are procoagulant, expressing tissue factor (TF) as well as certain adhesins, such as E-selectin, ICAM-1, αvβ3 and platelet endothelial cell adhesion molecule-1 (PECAM-1). Additionally, in vivo they detected EMPs at low levels in healthy individuals and at higher levels in patients with lupus anticoagulant.
Combes et al (1999) recommended use of CD31+/CD51+ as markers for EMPs. However, we found that only ≈20% of EMPs obtained from endothelial culture were positive for both these markers. Therefore, we measured EMPs using these markers separately. As platelets also express CD31, we used additional monoclonal antibodies (mAbs) specific for platelets and anti-CD42, and excluded particles co-expressing CD31 and CD42 to exclude platelet microparticles.
It has been reported (Laurence et al, 1996; Laurence & Mitra, 1997; Mitra et al, 1997, 1998) that exposure of MVECs to TTP plasma results in apoptosis. This was observed in brain, renal and dermal MVECs incubated with TTP plasma diluted from 1:100 to 1:1000 (1% to 0·1% v/v in culture medium). Fas upregulation and cross-linking and subsequent caspase-1 and caspase-3 induction were thought to be involved (Mitra et al, 1998). Wu et al (1999) also reported the effect of TTP plasma on the cell line ECV-304. Originally, these cells were believed to be of endothelial origin, however, more recently they have been classified as a derivative of the human bladder line T24.
In our culture conditions, we failed to detect apoptosis in renal or brain MVECs incubated with 20% TTP plasma. We attribute this disparity to the fact that our cells were grown in optimal health conditions, always maintained in the same basal nutrient medium and at no time were deprived of serum or growth factor. Our negative finding of apoptosis contradict the reports of Laurence et al (1996, 1997), Laurence & Mitra (1997) and Mitra et al (1997, 1998). The reason for this discrepancy is not clear but we suspect that differences in technique or culture media could account for this difference. Specifically, in their papers an ‘apoptosis culture medium’ was used. This apoptosis culture medium consisted of a change of basal nutrient medium from MCDB131 to M199 and deprivation of human serum and bovine brain extract. These culture conditions may have predisposed the cells in their study to apoptosis (Knedler & Ham, 1987).
Our data is consistent with clinical observation in TTP patients. They recover quickly with proper therapy and seldom exhibit residual neurological deficits or other organ failure, achieving complete remission. If extensive microvascular endothelial apoptosis was the culprit, such a prompt and complete recovery would be difficult to explain. In the work by Hamilton et al (1990), the endothelial cells remained viable, not apoptotic, even after exposure to complement C5b-9, and the same group proposed that a function of microparticle shedding is to rid the cell of noxious stimuli (Wiedmer & Sims, 1985).
A positive correlation between EMP concentration and thrombotic complications was reported in patients with LA (Combes et al, 1999). They used co-expression of CD31 and CD51 for assay of EMPs. They also demonstrated TF activity in TNF-α stimulated HUVECs, but required a large amount of EMPs to measure TF; i.e. per assay, they used supernatant from at least 6 × 105 cells (5–25 μl of 100 μl concentrated from 1·2 × 107 cells). In contrast, our RVV-based assay of procoagulant activity measures PF3 activity (Jy et al, 1995; Horstman, 1999) and responded strongly to supernatant from 200-fold fewer cells (15 μl of 1·0 ml supernatant of 2 × 105 cells). The relationship between TF and PF3 procoagulant activity has recently been addressed (Kunzelmann-Marche et al, 2000). Figure 2C is suggestive of TF activity in the EMPs.
Many efforts have been made to identify the factors in TTP plasma that induce EC injury. One of the candidates is autoantibodies against endothelium that fix complement to induce endothelial damage. The presence of anti-MVEC autoantibody in TTP plasma has been reported (Wright et al, 1999). We and others have reported the presence of anti-CD36 in TTP plasma (Tandon et al, 1994; Schultz et al, 1998). CD36 is expressed in both platelets and endothelium and is a good candidate for further investigation. Studies in other disease states, such as primary anti-phospholipid syndrome and systemic lupus erythematosus (SLE), indicate antibodies capable of activating endothelium with consequent prothrombotic activity (Oosting et al, 1992; Simantov et al, 1995). Unusually large mulitmers of VWF have been found to be markedly elevated in TTP. It was suggested that these large adhesive molecules are released from disturbed endothelium, prompting platelet adhesion and aggregation (Moake et al, 1982; Moake & McPherson, 1989; Moake, 1995). Recent studies have demonstrated that many patients with TTP lack the metalloprotease that cleaves unusually large VWF multimers to less adhesive monomers (Furlan et al, 1998; Tsai & Lian, 1998). Autoantibodies against these enzymes were found in many patients with TTP and lend further support to a role for autoimmunity in the pathogenesis of TTP (Furlan et al, 1998; Tsai & Lian, 1998).
In summary, the presence of procoagulant EMPs in TTP patients and in MVEC cultures exposed to TTP plasma supports the theory of endothelial cell injury as the immediate cause of TTP. EMPs may trigger or augment platelet microthrombi formation, playing a pivotal role in the pathogenesis of TTP. Quantification of EMPs may be a useful marker of endothelial injury in TTP. More detailed study of the coagulant and antigenic properties of EMPs in TTP may offer new insights into the pathogenesis of thrombotic microangiopathies.