Dr Isobel Ford Department of Medicine and Therapeutics, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, Scotland.
The clinical manifestations of the antiphospholipid syndrome (APLS) include arterial and venous thrombosis, thrombocytopenia and fetal loss, but the pathogenic mechanisms remain unclear. It has been hypothesized that platelet activation by autoantibody may be a pathogenic mechanism. We studied IgG binding, microparticle (mp) formation and P-selectin expression by flow cytometry in normal platelets after incubation in serum from 11 patients with antiphospholipid antibodies and that from 10 normal healthy subjects. Levels of platelet-associated IgG were significantly higher after incubation in patient sera (mean 17.2, range 2.0–75.0%) compared with normal sera (mean 2.0, range 1.2–3.7%, P < 0.05). Incubation of normal platelets in serum led to increased microparticle formation (P < 0.01) and P-selectin expression (P < 0.05), compared with unstimulated platelets. There was no significant difference, however, between microparticle formation nor P-Selectin expression induced by patient serum (mp 3.0 (1.6–5.0)%; P-selectin 8.0 (4.0–16.6)%) versus normal serum (mp 3.2 (2.1–4.5)%; P-selectin 10.1 (4.0–15.6); median (range)). Pre-activation of platelets with subthreshold ADP concentrations or thrombin receptor activator peptide resulted in a small increase in microparticle formation, but there was still no significant difference between the effects of patient and control sera. Despite the presence of platelet membrane binding IgG in serum from 5/11 patients with antiphospholipid antibodies, there was no evidence for associated enhanced platelet-activating ability. This study supports antiplatelet reactivity in antiphospholipid syndrome, but not a direct platelet-activating role for platelet-directed autoantibodies.
The clinical manifestations of the antiphospholipid syndrome (APLS) include arterial and venous thrombosis, thrombocytopenia and fetal loss, but the pathogenic mechanisms remain unclear. Recognition of platelet membrane components by ‘antiphospholipid’ autoantibodies, leading to activation of platelets, might provide a possible explanation, and an ability of antiphospholipid antibodies to activate platelets has been claimed ( Lin & Wang, 1992). Platelet activation by autoimmune mechanisms is known to underlie the arterial and venous thrombotic events in the syndrome of heparin-induced thrombocytopenia (HIT) with thrombosis, and it has been hypothesized that HIT and APLS share common mechanisms of cellular activation ( Arnout, 1996). The targets of autoantibodies in the primary antiphospholipid syndrome are largely unknown but appear to be mainly proteins with high affinity for negatively-charged phospholipids. Autoantibodies directed against epitopes on the platelet membrane are also commonly found in the serum of patients with antiphospholipid syndrome ( Galli et al, 1996 ).
Exposure of procoagulant phospholipid from the inner leaflet of the platelet membrane and budding of microvesicles or microparticles may follow in vitro activation of platelets by strong agonists such as thrombin plus collagen, as well as by murine monoclonal antibodies ( Nomura et al, 1992 ), complement complex C5b-9 ( Sims et al, 1988 ) and calcium ionophore ( Dachary-Prigent et al, 1995 ). There is evidence that platelet microparticles and activated platelets are present in the circulation in certain thrombotic disease states ( Abrams et al, 1990 ; Michelson, 1996), as well as in patients with antiphospholipid antibodies ( Galli et al, 1993b ). Some recent studies, using flow cytometry to detect the expression of platelet activation markers, have indicated that patients with the antiphospholipid syndrome have circulating activated platelets ( Fanelli et al, 1997 ; Galli et al, 1993b ; Joseph et al, 1997 ), although this has not been a consistent finding ( Out et al, 1991 ). It is unclear, however, whether ‘antiphospholipid’ antibodies are capable of initiating platelet activation, or if other components and conditions are required. The relationship between platelet activation and thrombotic disease or fetal loss in the antiphospholipid syndrome also remains uncertain.
Stimulation of platelets with a number of agonists leads to secretion from the alpha granules. P-selectin is translocated from the alpha-granule membrane to the platelet surface where it can be detected by fluorescently-conjugated antibodies. Flow cytometry provides a sensitive technique for the detection of platelet activation. The simultaneous labelling of platelets with an antibody to a platelet marker and an antibody to P-selectin conjugated with a different fluorophore, allows more accurate measurement of the proportion of activated platelets in a mixed population of cells. Platelet microparticles, generated during platelet activation, can also be detected by flow cytometry. They vary in diameter from 0.1 to 0.2 μm, compared with 2–4 μm for the average intact platelet, but carry the platelet glycoproteins GPIb and GPIIb–IIIa on their surface. Platelet microparticles can be discriminated from platelets by whole blood flow cytometry on the basis of their lower forward light scatter (FSC). The technique avoids the likelihood of centrifugation procedures to cause activation or disruption of platelets, or to lead to a bias towards the selection of particles of certain sizes or densities. Some overlap between platelet and microparticle size is probably unavoidable, however, and the smallest particles may fall below the limits of detection of the instrument. The use of flow cytometry has the advantage of permitting the discrimination of platelet-derived microparticles from those originating from other cells, since the former can be labelled with fluorescently-conjugated platelet markers such as anti-platelet GPIIIa.
In the present study we have measured platelet- recognizing IgG in serum from patients with antiphospholipid antibodies. We have examined the ability of patient serum to induce activation of normal unstimulated platelets in vitro, and of platelets pre-stimulated with ADP or thrombin receptor activator peptide, by using sensitive and specific flow cytometry methods for platelet microparticle formation and P-selectin expression. We have compared these with results obtained using serum from normal healthy controls.
Patients and controls
The study group comprised 11 patients (seven female) with antiphospholipid antibodies. Seven were persistently positive for lupus anticoagulant when tested using the DRVVT, KCT and KCCT; nine had elevated titres of IgG anti-cardiolipin antibodies (>2 standard deviations above the normal mean). Six patients had a history of thrombotic episodes and four of recurrent miscarriage. Eight of the patients were diagnosed as having primary antiphospholipid syndrome; one had systemic lupus erythematosus (SLE) and two had no evidence of previous thrombotic events or fetal loss despite elevated titres of anticardiolipin and positive lupus anticoagulant. Serum from a patient with heparin-induced thrombocytopenia with thrombosis was also employed as a positive control. Normal platelets and serum were obtained from 10 normal, healthy laboratory staff of comparable age (five female).
Preparation of washed normal platelets
Blood was collected by clean venepuncture from the antecubital vein of a normal healthy volunteer with a 19-gauge butterfly needle on the day of the assay. No subject had taken any anti-platelet drugs for the previous 7 d. 10 ml of blood was collected into citrate anticoagulant in 5 ml plastic tubes (1 part trisodium citrate 3.8 g 100 ml; 9 parts blood) and centrifuged at 200 g for 6 min to prepare platelet-rich plasma (PRP). Four aliquots of 0.5 ml PRP were washed with equal volumes of filtered Tris-EDTA buffer (Tris-HCl buffer pH 7.4 containing 10 m M EDTA) by centrifugation at 1000 g for 6 min. Platelets were resuspended in 0.5 ml of filtered HEPES-Mg buffer (10 m M HEPES, 145 m M NaCl, 5 m M KCl, 1 m M MgSO4, pH 7.4) and pooled. Platelets were counted on a Thrombocounter (Coulter Electronics Ltd, Luton) and the number of platelets in suspension was adjusted to 30 × 109/l by dilution in HEPES-Mg buffer. All buffers were sterile-filtered on disposable 0.2 μm filters (Gelman, U.S.A.), and stored at 4°C for no more than 7 d.
Preparation of serum
Venous blood was collected into glass tubes without anticoagulant and allowed to clot at room temperature. The blood was left for a further 60 min to allow retraction of the clot and then centrifuged at 3000 g for 15 min. Serum was dispensed into aliquots and stored at −70°C.
FITC-conjugated antibody to platelet glycoprotein IIIa (CD41b) purchased from Dako Ltd, Denmark, was used to identify platelets and platelet microparticles. Phycoerythrin (PE)-conjugated antibody to P-selectin (CD62P) was purchased from Coulter Electronics Ltd, Luton (Immunotech Antibodies) and FITC-conjugated-F(ab′)2 fragment of anti-human IgG from Dako Ltd. FITC- and PE-conjugated mouse Immunoglobulin G from Becton Dickinson, Oxford, were used as negative controls.
Incubation of platelets with serum and preparation for flow cytometry
Diluted platelet suspension (35 μl containing approximately 106 platelets) was pipetted into polypropylene tubes (Eppendorf tubes, Sarstedt Ltd, Beaumont Leys). Serum from patients and normal controls was thawed at 37°C for 2 min and 10 μl of serum or filtered PBS was added to the normal platelets. As a positive control for microparticle formation, CaCl2 (1 m M final concentration) and calcium ionophore (1 μM final concentration, A23187, Sigma-Aldrich Company Ltd, Poole, Dorset) were added to an aliquot of platelet suspension. As a positive control for P-selectin secretion and expression in each assay, an aliquot of platelets was incubated with phorbol myristate acetate (PMA) (1 μM final concentration). After incubation at room temperature for 1 h, fluorescently-conjugated antibodies were added to the appropriate tubes and incubated for 30 min at room temperature in the dark. 1 ml of cold, filtered PBS pH 7.4 was then added and the tubes were stored at 4°C until analysed by flow cytometry. Analysis was performed within 2 h of the end of incubation, although readings were found to be stable for at least 6 h. Each serum was tested with platelets from at least three normal donors on different occasions and the mean of these three measurements was used. In some experiments platelets were pre-incubated with 1 μM ADP (adenosine diphosphate from equine muscle, Sigma-Aldrich Company Ltd, Poole, Dorset) for 5 min prior to incubation with serum.
PRP was prepared by centrifugation of EDTA-anticoagulated blood, and washed three times with sterile-filtered PBS containing 0.2% bovine serum albumin (globulin-free, Sigma) and 0.01 M EDTA. Washed platelets were resuspended and the platelet count adjusted by dilution in PBS/BSA/EDTA to give 2 × 108 platelets per ml. 300 μl of this suspension were pipetted into polypropylene tubes for each test, and platelets washed ×1 with 1 ml of PBS/BSA/EDTA at 1800 g for 5 min. After blotting, platelet pellets were resuspended in 300 μl of either control serum, patient serum or buffer and incubated at 37°C for 30 min. Serum from a patient with known high titres of HLA antibodies was used as a positive control. After two washing steps, platelets were incubated with 100 μl of FITC-conjugated F(ab′)2 fragments of anti-human IgG (Dako) pre-diluted 1/20, for 30 min at room temperature in the dark. Platelets were then washed twice as above and, after the final wash, were resuspended in 1 ml of buffer and analysed on the flow cytometer within 2 h. In experiments designed to investigate the effects of prestimulation of platelets on IgG binding, agonists (ADP at a final concentration of 1 μM, or thrombin receptor activator peptide (TRAP 14-mer peptide) at 25 μM plus peptidase inhibitor, Amastatin, at 100 μM), were added to the washed platelet pellet, and incubated for 10 min at 37°C before the addition of serum. The agonists were purchased from Sigma-Aldrich Company Ltd, Poole, Dorset.
Platelets and microparticles were analysed using the Coulter Epics-XL flow cytometer (Coulter Electronics Ltd). Platelets were identified initially by log forward-scatter and side-scatter characteristics. Antibody to GPIIIa was used to positively identify platelets and microparticles. Quadrant markers were set on a histogram of log forward scatter versus log FL1 fluorescence on a sample of unstimulated platelets so that 3.0% of positively-stained (for GPIIIa) particles fell below the minimum forward scatter boundary, i.e. smaller than smallest resting platelets in the sample (Fig 1a). Platelet microparticles were defined in subsequent samples as the percentage of positively stained particles below the minimum size boundary. In samples stimulated with calcium ionophore, 74.2–90.5% of GPIIIa-positive particles fell within quadrant 4 (Fig 1b). Colour compensation was set to minimize cross-contamination of red and green fluorescence in dual-labelled samples.
Results were expressed as means or medians with ranges. The Wilcoxon two-sample rank sum test was used for comparisons between groups. The paired t-test was used for comparisons of matched groups.
Identification of platelets and microparticles
In unstimulated samples of washed platelets, at least 82% of particles counted within the electronic gate were labelled with anti-glycoprotein IIIa antibody (91.8 (82.9–97.9%); median (range)) and were therefore identifiable as platelets or platelet-derived microparticles. The percentage of particles expressing GPIIIa was similar before and after stimulation with in vitro agonists such as ADP (data not shown). It was possible to stain unstimulated platelets with antibody to glycoprotein Ib (GPIb), but GPIb expression was found to decrease after activation, as previously reported ( Nurden et al, 1995 ), and so GPIIIa was used as a platelet marker throughout the present studies.
Following incubation in normal or patient serum, the percentage of particles in the sample which expressed GPIIIa was noted to be slightly reduced (P < 0.05 for normal serum, but not significant for patient serum; 1 Table I). This may have been due to the presence of non-platelet particles of similar size in some serum samples, rather than to a reduction in GPIIIa expression on the incubated platelets.
Table 1. Table I. Platelet microparticles and P-selectin expression following incubation in normal or APLS serum.
Isolated platelets from normal subjects were incubated with buffer, normal serum or serum from patients with antiphospholipid antibodies, for 1 h at room temperature, then labelled with anti-GPIIIa/FITC or anti-P-selectin/PE and assayed by flow cytometry. Each serum was tested with platelets from three normal donors, and the mean of these three experiments was used. Results are presented as median (range); *P < 0.05 **P < 0.01 versus unstimulated platelets; incubations with normal sera versus patient sera: not significant.
Platelet-recognizing immunoglobulin in APLS sera
The percentage of platelets binding immunoglobulin G was significantly higher after incubation of normal washed platelets in patient sera (mean 17.2, range 2.0–75.0%) compared with normal sera (mean 2.0, range 1.2–3.7%, P < 0.05). Sera from 5/11 antiphospholipid patients in the study contained platelet-binding antibody (Fig 2). Serum from a patient with multispecific anti-HLA antibodies was included in each experiment as a positive control; platelet-associated IgG binding with this serum ranged from 85.8% to 98.1%.
In order to examine the effect of pre-activation of platelets on binding of platelet-recognizing IgG, normal platelets were incubated at 37°C for 10 min with TRAP 14-mer peptide (25 μM), in the presence of the peptidase inhibitor, Amastatin (100 μM), before incubation with serum from five of the patients and two controls. Incubation with TRAP induced platelet activation, as measured by microparticle formation or P-selectin expression, but the effect of TRAP peptide on binding of IgG from patients' serum to normal platelets was not consistent, producing decreased binding in some cases. The median change in IgG binding was +2.5% (not significant by Wilcoxon signed rank test).
Table 2. Table II. Effect of pre-stimulation of normal platelets with ADP or TRAP on microparticle formation induced by serum from normal subjects or patients with antiphospholipid antibodies.
* TRAP was tested with only three normal and three patient sera.
Effect of exogenous calcium ions
Microparticle formation is known to be influenced by the calcium ion concentration ( Pasquet et al, 1996 ). In order to examine the effect of calcium ions on microparticle formation in the presence of serum, platelets from seven normal subjects were tested for microparticle formation with patient and normal sera, with and without added calcium chloride (1 m M). There was no significant difference between percentage microparticle formation in unstimulated platelets and in those incubated with calcium chloride alone. Adding calcium chloride to serum did result in a highly significant increase in microparticle formation compared with that induced by serum alone ( 3 Table III). In the presence of calcium chloride there was, however, no significant difference between microparticle formation by patient sera or by normal sera.
Table 3. Table III. The effect of 1 m M calcium chloride on microparticle formation by washed platelets, with and without serum.
Microparticles are expressed as a percentage of total GPIIIa-positive particles. Results are medians with ranges. Differences between groups were analysed by the Wilcoxon two-sample rank sum test.
Platelet granule secretion after incubation in serum
P-selectin expression was significantly increased by incubation of platelets in serum, when compared with unstimulated washed platelets ( 1 Table I). There was no significant difference in the number of platelets expressing P-selectin after incubation in normal sera (median 10.1 (range 4.0–15.6)%) compared with APLS patient sera (median 8.0 (range 4.0–16.6)%). Nor was there a difference between P-selectin expression induced by patient sera which contained platelet-binding antibody, compared with the other patient sera. Stimulation of platelets with TRAP before incubation with serum led to activation manifested by a marked increase in expression of P-selectin, but again there was no significant difference between the effects in patient or control sera; percentage of platelets expressing P-selectin after incubation with TRAP and normal sera (median 63.1 (range 62.3–65.9)%); after incubation with TRAP and antiphospholipid antibody-containing sera (median 63.3 (range 60.9–64.8)%).
There is convincing evidence for the occurrence of anti-platelet antibodies in patients with antiphospholipid syndrome. In one study ( Galli et al, 1994 ), 40% of patients with antiphospholipid syndrome had antibodies to GPIIb–IIIa, to GPIb/IX, or to both, and between 20% and 40% of APLS patients had thrombocytopenia ( Galli et al, 1996 ). Conversely, anticardiolipin antibodies are found in patients with autoimmune thrombocytopenia ( Galli et al, 1996 ; Harris et al, 1985 ) and cross-reactivity may exist between antiphospholipid antibody and GPIIIa ( Tokita et al, 1996 ) and other platelet membrane components ( Hasselaar et al, 1990 ). Antibodies to platelet glycoproteins were have been shown to be distinct from anticardiolipin antibodies, and their separate activities could be depleted by absorption ( Galli et al, 1994 ). That platelet activation mechanisms can be triggered by antibodies or immune complexes is also well known. Brandt et al (1996 ) demonstrated aggregation of platelets by HLA-related antibodies in the presence of serum. It has been known for some time ( Luscher & Pfüeller, 1978) that IgG-containing immune complexes can induce platelet activation and secretion, and that the activating complexes do not necessarily have to be directed against platelet antigens. Our data contrast with those of Lin & Wang (1992) who showed that purified rabbit anticardiolipin antibodies could activate human platelets in vitro, although it should be borne in mind that there are species differences and that aggregated rabbit Fc, for example, is much more potent than aggregated human Fc in inducing platelet release ( Luscher & Pfüeller, 1978). It is also unclear whether pre-activation of platelets by other agonists is necessary, but synergy was demonstrated between other agonists and the rabbit anticardiolipin antibodies in their activation of human platelets ( Lin & Wang, 1992).
In the present study, despite the presence of significant anti-platelet IgG in serum from around half of the APLS patients, there was no evidence found for associated enhanced platelet-activating ability. Microparticle formation and P-selectin expression in normal platelets were both significantly increased by incubation in sera from patients with primary APLS. There was no significant difference, however, between the degree of activation induced by patient sera compared with sera from normal controls. The addition of calcium chloride to serum produced a greater increase in microparticle formation, presumably due to opening of calcium ion channels and influx of calcium ions leading to enhancement of activation. It is known that calcium ions are necessary for platelet microparticle formation and for exposure of negatively-charged phospholipid on the platelet surface ( Zwaal & Schroit, 1997; Dachary-Prigent et al, 1995 ). Even after increasing the calcium ion concentration, however, there was no difference in the present study between the percentage of microparticles formed by normal or patient sera, so these results are unlikely to be due to an effect of anti-platelet or anti-phospholipid antibody. The reasons for the effects of normal serum on platelet activation are unclear, but the results confirm the importance of including suitable normal controls. The functional capacity of platelets used in the present study with respect to granule secretion and microparticle formation was confirmed in each set of experiments by incubation of platelets with phorbol myristate acetate (PMA) or calcium ionophore.
It is known that heparin-induced antibodies will form complexes with heparin and platelet factor 4 which can activate platelets, inducing secretion from alpha granules and dense bodies ( Lee et al, 1996 ), and promoting expression of negatively charged phospholipid ( Tomer, 1997) and formation of procoagulant microparticles ( Warkentin et al, 1994 ). It has been proposed that mechanisms of cellular activation are common to HIT and APLS ( Arnout, 1996). In the present study we could find no evidence for the direct activation of normal platelets by serum which contained antiphospholipid antibodies. This is in contrast to the results with serum from an HIT patient, examined under identical conditions, which were consistent with the studies of Warkentin et al (1994 ) and Lee et al (1996 ). It may be that for activation of platelets in APLS, there is a need for an additional cofactor not present in serum. Some workers have shown a requirement for beta-2 glycoprotein I (β2GPI) in the binding of antiphospholipid antibodies to activated platelets ( Shi et al, 1993 ; Vasquez-Mellao et al, 1994 ), but this would be available in serum. The variable reactivity of platelets to HIT sera and heparin has recently been attributed to a polymorphism of the FcγRII-A receptor, although this remains to be proven ( Brandt et al, 1995 ; Burgess et al, 1995 ). We tested platelets from a total of 10 normal donors, some of whom have been shown on previous occasions to respond to heparin-induced antibodies, so it is unlikely that this or another polymorphism of platelets is required for a response to APLS serum.
The so-called ‘double-hit’ hypothesis which has been proposed for thrombosis associated with HIT or APLS ( Arnout, 1996) requires initial cell injury or activation to occur before the ‘anti-phospholipid’ can bind to its target and produce further cell injury. Low-level activation of platelets may be necessary for further stimulation by the antibodies in APLS serum. Arvieux et al (1993 ) reported that murine monoclonal antibodies to β2GPI could induce platelet aggregation and secretion only if platelets were pre-activated with ADP or adrenaline. The reaction required whole antibody and interaction with the platelet Fc receptor but did not require exogenous fibrinogen. Elsewhere, purified F(ab′)2 fragments of antiphospholipid antibodies were found to induce 5-HT release and thromboxane production by thrombin-stimulated platelets ( Martinuzzo et al, 1993 ). Purified IgG from patients with antiphospholipid syndrome, in contrast to syphilitic anticardiolipin antibodies, was capable of inducing P-selectin expression, but only if platelets were pre-activated by ADP or collagen ( Campbell et al, 1995 ). Other groups, however, have found that purified antibodies from antiphospholipid syndrome or SLE patients can lead to inhibition of in vitro platelet aggregation or adhesion ( Ostfeld et al, 1992 ) or of platelet prothrombinase activity ( Galli et al, 1993a ). This latter reaction seems to be dependent on binding of β2GPI. In the present study, preactivation of platelets led to an apparent small increase in IgG binding when incubated in serum, and to enhanced P-selectin expression, but these changes were found with normal control sera as well as with sera from patients with antiphospholipid antibodies and so cannot be attributed to the autoantibodies. The discrepancy in this regard may be attributable to the use of whole serum rather than the less physiological conditions of earlier studies in which purified components were used.
There are few studies of in vivo activation in primary APLS. Circulating microparticles were found in 50% of patients with antiphospholipid antibodies ( Galli et al, 1993b ), and were associated with a history of thrombotic events. More recently, surface expression of platelet lysosomal protein, CD63, but not P-selectin (CD62P), was found to be increased on platelets from patients with antiphospholipid syndrome ( Joseph et al, 1997 ). Another group ( Fanelli et al, 1997 ) found that CD62P, but not CD63 expression, on circulating platelets was significantly increased in primary antiphospholipid syndrome patients compared to normal healthy controls. In that study increased CD62P expression was found in patients with ‘low’ platelet counts (<150 × 109/l) compared with those with normal platelet counts, but there was no correlation with platelet-associated IgG ( Fanelli et al, 1997 ). Elsewhere, an imbalance between the synthesis of thromboxane and prostaglandin I2 has been reported in patients with lupus anticoagulant ( Lellouche et al, 1991 ). Other studies have failed to confirm that platelets circulate in an activated state in patients who display platelet binding anti-phospholipid antibodies ( Out et al, 1991 ).
The present study does not support a platelet-activating role for platelet-directed autoantibody in patients with primary anti-phospholipid syndrome. It seems unlikely that the autoantibodies present in patients with primary antiphospholipid syndrome are capable of initiating platelet activation, despite the presence of antibodies which are apparently able to bind to normal unstimulated platelets or pre-activated ones. An alternative explanation for the reported in vivo findings is that low-level activation of circulating platelets is secondary to other mechanisms, such as autoimmune-mediated damage to the vascular endothelium.