Portal de Periódicos da CAPES

Specificities, properties, and clinical significance of antiprothrombin antibodies

2003; Wiley; Volume: 48; Issue: 4 Linguagem: Inglês

10.1002/art.10831

1529-0131

Olga Amengual, Tatsuya Atsumi, Takao Koike,

Systemic Lupus Erythematosus Research

Antibodies against phospholipid-binding proteins, such as β2-glycoprotein I (β2GPI) and prothrombin, are associated with the clinical manifestations of the antiphospholipid syndrome. In the last decade, there has been increasing interest in antiprothrombin antibodies (APTs). The immunologic and functional properties of these antibodies have been investigated, and they have been found to vary widely, depending mainly on their affinity for human prothrombin. Recently, antibodies against phosphatidylserine–prothrombin complex (APS/PT) have been detected, and these antibodies, rather than antibodies against prothrombin alone (APT-A), are closely associated with antiphospholipid syndrome and lupus anticoagulant (LAC). Despite increasing knowledge about their mechanisms of action, the clinical significance of APTs has not yet been established, and the question of whether antibodies to particular phospholipid-binding plasma proteins or phospholipid–protein complexes lead to different clinical presentations and/or thrombogenic mechanisms needs to be addressed. Antiphospholipid antibodies are immunoglobulins associated with a variety of clinical phenomena, including arterial and venous thrombosis, pregnancy morbidity, neurologic disorders, and thrombocytopenia. The term "antiphospholipid syndrome" is used to link these clinical manifestations to the persistence of aPL, and it is recognized as one of the most common causes of acquired thrombophilia (1, 2). It has been shown that aPL, despite their name, are not directed against anionic phospholipids, as had previously been thought, but are a part of a large family of autoantibodies against phospholipid-binding plasma proteins or phospholipid–protein complexes (3, 4). The most common and best characterized antigenic targets of these antibodies are β2GPI (5-7) and prothrombin (8). Other phospholipid-binding proteins, including high and low molecular weight kininogens (9), annexin V (10), and protein C and protein S (11, 12), may be important targets as well. Since most of these antigens are involved in blood coagulation, some aPL may hamper the regulation of the coagulation system, providing an explanation of the high incidence of thrombotic events in patients with antiphospholipid syndrome. β2GPI is a phospholipid-binding protein that has been studied extensively. β2GPI plays an important role in the binding of aPL to phospholipids. In 1990, three groups of investigators, Galli et al (5), McNeil et al (6), and Matsuura et al (7), independently reported that anticardiolipin antibodies (aCL), which were associated with the antiphospholipid syndrome, were not directed against CL alone but required a plasma protein (β2GPI) as a cofactor to bind CL in enzyme-linked immunosorbent assay (ELISA) plates. β2GPI bears the epitopes for aCL binding that are exposed when β2GPI binds to negatively charged phospholipids or irradiated plastic plates (13). Several studies have highlighted the significance of using anti-β2GPI antibodies as an alternative ELISA method that has higher specificity for the diagnosis of antiphospholipid syndrome than does the conventional aCL ELISA (14-16). Prothrombin, another major phospholipid-binding protein, was first reported to be a possible cofactor for LAC in 1959 (17). In subsequent years, prothrombin has attracted considerable interest, and several groups have dealt with APTs in an attempt to clarify their immunologic characteristics and the role they may play in thrombosis in patients with antiphospholipid syndrome. It has been shown that APTs detected by ELISA represent a heterogeneous family of antibodies that includes APT-A and APS/PT. This review summarizes recent information on the prothrombin molecule and on the specificities and properties of APTs. Prothrombin (factor II) is a vitamin K–dependent glycoprotein synthesized in the liver and present at a concentration of ∼100 μg/ml in normal plasma. Its gene spans 21 kbp on chromosome 11 (18). Mature human prothrombin that circulates in blood consists of a single-chain glycoprotein of 579 amino acid residues with a molecular weight of 72 kd, including 3 carbohydrate chains and 10 γ-carboxyglutamic acid residues (19). During its biosynthesis in the liver, prothrombin undergoes γ-carboxylation. These γ-carboxyglutamic residues, known as the GLA domain, are located on fragment 1 of the prothrombin molecule. The GLA domain is essential for the calcium-dependent binding of phospholipid to prothrombin, which in time, is necessary for the conversion of prothrombin to biologically active α-thrombin. The GLA domain is followed by a kringle domain containing 2 kringle structures and a carboxyl-terminal serine protease. Kringle is a structure with 3 characteristic intradisulfide bonds that has been identified in a number of proteins, including prothrombin (20). Kringles of each protein have distinct functions in many cell types and interact with substrate cofactor receptors (21). In the prothrombin molecule, the kringle domain is involved in the binding of thrombin to fibrin (18). Prothrombin is physiologically activated by the prothrombinase complex (activated factor X, factor V, calcium, and phospholipids). Once negatively charged phospholipids bind prothrombin, prothrombinase complex converts prothrombin to thrombin, which triggers fibrinogen polymerization into fibrin (22). In addition, thrombin binds thrombomodulin on the surface of endothelial cells and activates protein C, which then exerts its anticoagulant activity by digesting factor V and depriving the prothrombinase complex of its most important cofactor. Because of this negative-feedback pathway, prothrombin/thrombin behaves as an "indirect" anticoagulant. Prothrombin activation catalysis by prothrombinase complex constitutes an ordered sequential reaction that proceeds via the cleavage at Arg322–Ile323 and results in the liberation of meizothrombin, a catalytically active intermediate composed of fragment 1+2 and A and B chains of thrombin linked by a disulfide bond. Further cleavage at Arg273–Thr274 results in the formation of α-thrombin (23) (Figure 1). Factor Xa alone catalyzes the activation of human prothrombin by proteolytic cleavage at Arg273–Thr274, which results in the liberation of fragment 1+2 and prethrombin 2. Prethrombin 2 is composed of the A and B chains of thrombin with the bond at Arg322–Ile323 intact. Further proteolysis at this bond yields α-thrombin (Figure 1). One of the most potent enzymes known, α-thrombin not only converts fibrinogen to fibrin, but also acts on factors V, VIII, and XIII, protein C, platelets, and endothelial cells (24). Conversion of human prothrombin to α-thrombin. Left, Activated factor Xa catalyzes the activation of human prothrombin by proteolytic cleavage at Arg273–Thr274, resulting in the liberation of fragment 1+2 and prethrombin 2. Prethrombin 2 is composed of the A and B chains of thrombin with the disulfide bond (S–S) at Arg322–Ile323 intact. Further proteolysis at this bond yields α-thrombin. Right, Prothrombinase complex activates human prothrombin in an ordered sequential reaction via cleavage at Arg322–Ile323 and results in the liberation of meizothrombin, followed by cleavage at Arg273–Thr274, which leads to the formation of α-thrombin. The involvement of prothrombin as a cofactor for a circulating anticoagulant was first considered in 1959 by Loeliger (17). The addition of normal plasma to that of a patient with LAC increased the degree of coagulation inhibition. A low plasma level of prothrombin was also found, suggesting that the cofactor associated with the expression of LAC activity was most likely prothrombin. In 1960, Rapaport et al (25) described the case of a child with LAC who underwent recurrent episodes of bleeding. Further investigations showed a severe prothrombin deficiency, a prolonged prothrombin time, and a prolonged activated partial thromboplastin time. During the 1980s, more research was performed to clarify the hypoprothrombinemia in patients with LAC, and in 1983, Bajaj et al (26) were the first to ascertain the presence of prothrombin-binding antibodies in patients with LAC and severe hypoprothrombinemia. They postulated that hypoprothrombinemia results from the rapid clearance of prothrombin–APT complexes from the circulation. In 1984, Edson et al (27) demonstrated prothrombin–APT complexes by counterimmunoelectrophoresis (CIE) in the plasma of patients with LAC but without severe hypoprothrombinemia. These findings were confirmed and extended by Fleck et al (28), who found that 74% of the population with LAC under study had prothrombin–APT complexes by CIE and showed that APT had LAC activity. Subsequently, circulating prothrombin–APT complexes were also observed in patients with LAC and with normal prothrombin levels. In 1991, Bevers et al (8) highlighted the importance of APT in causing LAC activity when they studied 16 patients with both aCL and LAC. After incubation with CL-containing liposomes, the LAC activity remained in the supernatant from 11 patients. These 11 samples demonstrated LAC activity in a phospholipid-bound prothrombin–dependent manner. Later, Oosting et al (11) showed that LAC inhibited endothelial cell–mediated prothrombinase activity and that the IgG fraction containing LAC activity bound to phospholipid–prothrombin complex. It was shown that most LACs depend on the presence of phospholipid-bound prothrombin and phospholipid-bound β2GPI, and the anticoagulant properties of APTs have been studied by several groups. Permpikul et al (29) purified IgG fractions from 10 patients with LAC, showing that LAC activity was due to APTs in at least 9 of the samples. Some investigators (30, 31) reported the existence of two types of circulating APTs which can be distinguished on the basis of their effects in coagulation assays: 1) functional, which cause LAC activity, and 2) nonfunctional, which do not contribute to the LAC activity. The differences in effects are probably caused by differences in epitope specificities between the different APTs (32). The first techniques used for screening APTs were double diffusion, CIE (26-28, 33), and assays based on the impairment of prothrombin activation by APTs (11, 34). In 1995, Arvieux et al (35) showed that APTs could be detected by a standard ELISA. This ELISA used prothrombin as antigen coated onto irradiated plates (APT-A ELISA). The behavior of APTs in ELISA closely resembles that of anti-β2GPI. In fact, APTs cannot be detected when prothrombin is immobilized on nonirradiated plates (35), but binding is observed if prothrombin is immobilized on a suitable anionic surface, adsorbed on gamma-irradiated plates, or exposed to immobilized anionic phospholipids. In 1996, Matsuda et al (36) demonstrated that APS/PT could be detected in LAC-positive patients. Later, Galli et al (30) reported that the ELISA system with PS-bound prothrombin (APS/PT ELISA) was more efficient in demonstrating the presence of APT than the system using prothrombin alone coated on high-binding plates (APT-A ELISA), suggesting the existence of two types of APTs. Our group (37) recently reported that the detection of APS/PT strongly correlated with the presence of LAC. Most investigators have used only the APT-A ELISA, presuming that both systems of detection (APT-A and APS/PT ELISAs) detect the same antibodies. However, we showed that there was no correlation of APS/PT and APT-A titers in patients with antiphospholipid syndrome or other autoimmune diseases (Figure 2). Human monoclonal APS/PT, derived from a patient with antiphospholipid syndrome, did not bind to prothrombin immobilized on oxidized plates (38). Therefore, the APS/PT assay detects, at least in part, a population of antibodies different from APT-A, and these antibodies have significant clinical relevance because of their correlations with clinical features of antiphospholipid syndrome and with the presence of LAC. Table 1 summarizes the historical background of research on APTs. Relationship between IgG titers of antiprothrombin antibodies against prothrombin alone (aPT-A) and antibodies against phosphatidylserine–prothrombin complex (aPS/PT). The titers of IgG aPT-A and IgG aPS/PT in patients with antiphospholipid syndrome or other autoimmune diseases were determined by enzyme-linked immunosorbent assay. The dashed line indicates the cutoff for positivity. Many patients with only 1 positive determination were found. U = units. The mechanisms by which APTs cause LAC activity are not yet clearly understood. Pierangeli et al (39) have suggested that APTs cause prolongation of in vitro clotting by inhibiting the conversion of prothrombin into thrombin. However, it seems unlikely that APTs prolong the clotting times by hampering the activation of prothrombin through binding near the activation sites in the molecule, since such a mechanism would not explain the neutralizing effects of high phospholipid concentrations. Recently, Simmelink et al (40) showed that addition of affinity-purified APTs from LAC-positive plasma to normal plasma induced LAC activity, and that LAC activity was neutralized upon increasing the phospholipid concentration. They also showed that complexes of prothrombin and APT with LAC activity inhibited both prothrombinase and tenase complex. Thus, in the procoagulant pathway, APTs might increase the affinity of prothrombin for negatively charged phospholipids, thereby competing with clotting factors for the available catalytic phospholipid surface, a mechanism similar to that of anti-β2GPI. The model, based on increased affinity of protein–antibody complexes (β2GPI or prothrombin) for negatively charged phospholipids, can explain why LAC activity caused by both anti-β2GPI and APT can be neutralized by the addition of extra phospholipids. Furthermore, in their investigation of the anticoagulant activity of the protein C system, Galli et al (41) demonstrated the dominant inhibitory effect on activated protein C activity of anti-β2GPI compared with the effect of APT using isolated total IgG fraction. APTs bind to prothrombin coated on gamma-irradiated (35) or activated polyvinyl chloride ELISA plates (APT-A) (12, 30) or exposed to immobilized PS (APS/PT) (30). Recently, it has been shown that APTs also recognize prothrombin when it is bound to hexagonal (II) phase phosphatidylethanolamine, a neutral phospholipid used as an excess of phospholipids in the clotting assay to confirm the presence of LAC (42). APTs may be directed against cryptic epitopes or neoepitopes (antigens) exposed when prothrombin binds to anionic phospholipids, and/or they may be low-affinity antibodies binding bivalently to immobilized prothrombin. However, experimental evidence has not clearly established whether (or which) APTs are antineoepitopes or low-affinity antibodies. Wu and Lentz (43) observed that human prothrombin undergoes a conformational change upon binding to PS-containing surfaces in the presence of calcium. On the other hand, Galli et al (30) demonstrated that prothrombin was recognized more efficiently when it was bound to PS-coated ELISA plates using calcium ions and that APTs could be of low affinity. They suggested that prothrombin complexed with PS could allow clustering and better orientation of the antigen, offering optimal conditions for antibody recognition. A high percentage of APTs have species specificity for the human protein (8), and a minority of them react with bovine prothrombin (44). The epitope(s) recognized by APTs remains to be defined. Rao et al (45), used purified IgG preparations from LAC-positive patients and demonstrated binding of APT to prethrombin 1 and fragment 1 as well as to the whole prothrombin molecule. However, none of the antibodies reacted with immobilized thrombin. These findings suggest that dominant epitopes are likely to be located near the phospholipid-binding site of the prothrombin molecule, although they may have heterogeneous distribution. Recently, Akimoto et al (46) investigated the epitope distribution of APTs using recombinant prothrombin fragments. They demonstrated that 61.5% of APTs were dominantly directed to fragment 1 of prothrombin and 38.4% to the prethrombin site (fragment 2 plus α-thrombin). The clinical significance of APTs has not yet been established (47). A number of clinical studies have investigated the clinical relevance of APTs detected by ELISA using prothrombin alone as antigen (APT-A ELISA), and these have been recently summarized by Galli (48). Some of these studies showed positive correlations between APT-A and some of the clinical features of antiphospholipid syndrome. Puurunen et al (49) reported the presence of APT-A in 34% of patients with systemic lupus erythematosus (SLE) and found positive correlations of both APT-A and anti-β2GPI with deep vein thrombosis in this population. Sorice et al (50) described a higher frequency of APT-A in patients with SLE and antiphospholipid syndrome than in those without antiphospholipid syndrome and demonstrated a significant correlation between the presence of these antibodies and the clinical features of antiphospholipid syndrome. Bertolaccini et al (51) reported positivity for APT-A in 58 of 207 patients with SLE and observed that 53% of patients with APT-A had a history of thrombotic events (odds ratio [OR] 2.5, 95% confidence interval [95% CI] 1.4–4.6). Furthermore, in a series of 177 patients with SLE or primary antiphospholipid syndrome, Muñoz-Rodriguez et al (52) found that thrombotic events were more prevalent in patients with APT-A (45% versus 28%; P = 0.02) and that these antibodies represented an independent risk factor for arterial thrombosis (OR 2.4, P = 0.04). Additionally, those investigators found an association between APT-A and thrombocytopenia in patients with primary antiphospholipid syndrome (OR 6.7, P = 0.007). Using multivariate logistic regression analysis, Nojima et al (53) demonstrated that the coexistence of IgG APT-A and LAC activity represented a risk factor for venous thromboembolism in patients with SLE (OR 19.1, 95% CI 4.7–77.1). Lakos et al (54), who evaluated a population of 70 patients with systemic autoimmune disorders, claimed high specificity of APT-A for the diagnosis of antiphospholipid syndrome. Conversely, Pengo et al (12) found no correlation between the presence of APT-A and thrombosis in 22 patients with aPL and a history of at least one thromboembolic event. Horbach et al (31), in a large population of patients with SLE, found that both IgG and IgM APT-A were more frequent in patients with a history of venous thrombosis, but the correlation was not significant when examined by multivariate analysis. Horbach et al showed that LAC was the strongest risk factor for thrombosis and that neither APT-A nor anti-β2GPI gave additional information concerning a thrombotic risk in SLE. Similar data were obtained by D'Angelo et al (55) using samples from 110 inpatients referred for evaluation of their aPL status, 32 of whom had a history of thrombotic events. D'Angelo et al found that only the presence of LAC, but not aCL, anti-β2GPI, or APT-A, was significantly associated with a history of previous thrombosis (OR 15.8, 95% CI 4.3–59.6, P < 0.001). Forastiero et al (56), in a group of 233 patients with LAC and/or aCL, showed that APT-A were related to venous thrombosis; however, the multivariate analysis demonstrated that anti-β2GPI, but not APT-A, were the only independent risk factor for venous thrombosis in those patients. Galli et al (30) detected APT-A in 58% of patients with aPL, but the overall prevalence of these antibodies was similar in patients with and without a history of thrombosis (95% versus 86%, respectively). Guerin et al (57) described the presence of APT-A in a variety of disorders, including antiphospholipid syndrome, SLE, and conditions associated with a high rate of false-positive results in conventional aPL tests. Those investigators observed lack of specificity of APT-A for antiphospholipid syndrome and failed to demonstrate an association between the presence of these antibodies and thrombosis; however, they found a correlation between APT-A and antiphospholipid syndrome, suggesting that these antibodies might be associated more with other clinical manifestations, such as fetal loss or thrombocytopenia. Swadzba et al (58) did not find a correlation between IgG APT-A and thrombosis in SLE, although they found a high prevalence of the IgM isotype in patients with lupus-like disease and a history of thrombotic events. Additionally, in a longitudinal study of 7 patients with SLE and thrombotic events, Inanç et al (59) demonstrated that APT-A were far less predictive of thrombosis than were anti-β2GPI. In a study of patients without autoimmune disease, Vaarala et al (60) found that high levels of APT-A conferred high risk of myocardial infarction or cardiac death in middle-aged men, and Palosuo et al (61) found a close relationship between high APT-A levels and deep venous thrombosis and pulmonary embolism in middle-aged men. However, Eschwege et al (62) showed no correlation between APT-A and thrombotic events in a large population of unselected patients with a history of venous thrombosis. In that study, LAC positivity was the only aPL test result that was strongly associated with the severity of thrombosis. Table 2 summarizes the clinical associations of APT-A in autoimmune disorders. Taken together, the controversial nature of the available data regarding APT-A, the lack of a well-standardized assay, and the fact that the majority of the studies are retrospective make it difficult to make definite conclusions regarding the clinical significance of these antibodies. More "cross-sectional" and/or prospective clinical studies are warranted to establish the clinical relevance of APT-A. We have reported the detection of APS/PT and have shown that the results obtained depend closely on the assay performed (37). The presence of APS/PT, but not APT-A, correlated significantly with the clinical manifestations of antiphospholipid syndrome (OR 8.29, 95% CI 3.03–22.71, P < 0.0001 versus OR 1.89, 95% CI 0.71–5.06, P not significant). Furthermore, APS/PT were more closely associated with LAC than were APT-A. Thus, APS/PT are a marker for antiphospholipid syndrome, and their detection may help to confirm the presence of LAC. The pathophysiologic mechanisms of APTs are not completely known. However, there is increasing evidence that they play a role in the hypercoagulable state of antiphospholipid syndrome. The antigens are present in plasma or on cell surfaces exposed to plasma, and they are therefore accessible to circulating antibodies. Some effects on endothelial cells have been proposed: 1) APTs inhibit thrombin-mediated endothelial cell prostacyclin release and hamper protein C activation (3); 2) APTs could recognize the prothrombin–anionic phospholipid complex on the endothelial cell surface, thus activating endothelial cells and inducing procoagulant substances via prothrombin (63); or 3) APTs could increase the affinity of prothrombin for negatively charged phospholipids (45). Field et al (64) extended the last of these observations and reported that a murine monoclonal APT and 7 IgG LACs enhanced the binding of prothrombin to phospholipid vesicles (75:25 phosphatidylcholine:PS) in a concentration-dependent manner. They also demonstrated that the monoclonal antibody and 4 of 6 IgG LACs from patients with a history of thrombosis increased thrombin production by purified prothrombinase components in a flow system. These in vitro observations have been reported in vivo, showing that thrombin production is increased in patients with LAC (65) or aPL (66). The precise mechanisms by which IgG LACs enhanced interaction of prothrombin with phospholipid vesicles are not known, although Field et al showed that antibody bivalency was essential for this effect, as was the elevated microenvironmental concentration of prothrombin on the surfaces of the phospholipid vesicles (67). They demonstrated that this enhanced binding was due to a decrease in the "off rate" for prothrombin interaction with phospholipid vesicles in the presence of LAC. Some aPL bind to endothelial cells, suggesting that aPL may interact with endothelial cells and thus promote thrombosis (68, 69). APTs can bind to immobilized PS in the presence of calcium and prothrombin (28, 29), and a LAC IgG preparation can enhance the binding of prothrombin to endothelial cells and increase thrombin generation on these cells (45), suggesting that APT may concentrate prothrombin on cell surface phospholipids and thus lead to a hypercoagulable state. Zhao et al (70) generated and characterized 1 monoclonal IgG APT (from a patient with antiphospholipid syndrome) that showed LAC activity. This monoclonal IgG enhanced the binding of prothrombin to damaged endothelial cells and shortened plasma coagulation times in the presence of LAC. Therefore, a highly purified monoclonal APT can behave like LAC in endothelial cell–based coagulation assays and can paradoxically increase the rate of thrombosis in patients with antiphospholipid syndrome. It has been suggested that aPL alone are unlikely to be triggering agents for thrombosis in patients with antiphospholipid syndrome and that aPL might promote and sustain thrombosis initiated by other factors. APTs may promote thrombosis by facilitating prothrombin interactions with damaged blood vessel walls and by promoting thrombin generation in flow, leading to a hypercoagulable state and consequently to a thrombotic tendency (Figure 3). Proposed mechanism of action of antiprothrombin antibodies at the endothelial cell level. After an activation signal, anionic phospholipids (phosphatidylserine [PS]) are exposed on the endothelial cell. Prothrombin binds those phospholipids on the cell surface, and circulating antibodies against PS–prothrombin complexes (aPS/PT) recognize the epitope(s) on the prothrombin molecule and bind to them. Those antibodies (or immune complex) activate endothelial cells via specific pathways or via Fcγ receptors. This induces the release of procoagulant substances, which leads to coagulation activation, platelet aggregation, and thrombosis. PAI-1 = plasminogen activator inhibitor 1. APTs could bind platelets and produce thrombocytopenia. However, antibodies directed toward specific platelet glycoproteins have been detected in patients with aPL in a proportion similar to that reported for patients with idiopathic thrombocytopenic purpura (71), suggesting that the cause of thrombocytopenia in patients with aPL is similar to that of idiopathic thrombocytopenic purpura. So far, the pathogenic role of APTs in morbidity during pregnancy has not been clarified. Further investigations are needed to determine whether these antibodies play any role in the pathogenesis of these common features of the antiphospholipid syndrome. Finally, other congenital or acquired factors may contribute to the final thrombotic risk in patients with aPL. These factors include prothrombin gene mutation, factor V "Leiden," hyperhomocystinemia, elevated plasma levels of prothrombin, factor VII, and von Willebrand factor, and decreased protein C and protein S activities. APTs are frequently found in patients with aPL, but their immunologic characteristics and mechanisms of action are not completely understood. The evaluation of the clinical relevance of APTs is influenced by many variables and depends on the applied method of detection. APS/PT are more closely associated with antiphospholipid syndrome and LAC than are APT-A. The determination of APS/PT in clinical practice, in conjunction with other aPL, may improve the likelihood of recognizing antiphospholipid syndrome. Knowledge of the behavior of specific aPL would aid in defining specific thrombogenic pathways and improve the management of patients with antiphospholipid syndrome.