IgG antibodies that recognize epitope Gly40-Arg43 in domain I of β₂–glycoprotein I cause LAC, and their presence correlates strongly with thrombosis

The antiphospholipid syndrome (APS) is an autoimmune disease characterized by the presence of circulating antiphospholipid antibodies (aPLs), notably anticardiolipin antibodies (aCLs) and antibodies that prolong phospholipid coagulation tests (lupus anticoagulant [LAC]) in patients with thrombosis and/or well-defined pregnancy morbidity.^1-3  Initially it was thought that aPLs bind directly to negatively charged phospholipids. However, it is now generally accepted that the clinically relevant aPLs bind to proteins with affinity for phospholipids.^4-7  Probably the most important epitope of APS-related aPL resides on β₂–glycoprotein I (β₂GPI).⁴  Most authorities do agree that anti-β₂GPI antibodies of the immunoglobulin G (IgG) class are more related to clinical symptoms like thrombosis than IgM-class antibodies.

β₂GPI is a 50-kDa glycoprotein that is present in plasma at a concentration of approximately 200 μg/mL. The protein is a member of the complement control proteins (CCPs).⁸  β₂GPI consists of 5 CCP repeats in which the first 4 domains are regular repeats consisting of 60 amino acids. The fifth domain is different from the other 4 domains as it consists of 82 amino acids, 4 conserved hydrophobic amino acids (313-316), 3 disulfide bonds instead of 2, and a cluster of positively charged amino acids (282-287), which is responsible for binding of β₂GPI to phospholipids.^9-12 

To understand the pathophysiology of APS, it is necessary to know the domains involved in anti-β₂GPI antibody binding and to know whether pathologic antibodies are all directed toward the same epitope. There is substantial controversy on the specificity of anti-β₂GPI antibodies. Some groups claim that the epitope for binding of anti-β₂GPI antibodies is located on domain IV.^13,14  Other studies showed that binding of anti-β₂GPI antibodies to β₂GPI was abrogated when there was a clip at 316-317 in domain V, suggesting that domain V is involved.¹⁵  Arvieux et al¹⁶  found that anti-β₂GPI antibodies inhibited the binding of β₂GPI to cardiolipin, suggesting again that domain V harbors the epitope for binding of anti-β₂GPI antibodies. Blank et al¹⁷  found that anti-β₂GPI antibodies react with peptides that cover sequences present in domains I, II, III, and IV. Most of these groups used murine monoclonal antibodies or a restricted number of affinity-purified patient anti-β₂GPI antibodies for their studies. Interestingly, studies using only anti-β₂GPI antibodies purified from patients concluded that domain I is the major domain of β₂GPI involved in the binding of anti-β₂GPI antibodies.^18-20 

Recently it was suggested that the positively charged epitope R39-R43 on domain I is important in binding of anti-β₂GPI antibodies.²¹  Here we further investigate that suggestion by testing the specificity of anti-β₂GPI antibodies in patients with systemic lupus erythematosus, lupuslike disease, and primary antiphospholipid syndrome by using deletion mutants of β₂GPI, recombinant full-length β₂GPI, point-mutated recombinant full-length β₂GPI, and both hydrophilic and hydrophobic enzyme-linked immunosorbent assay (ELISA) plates. Findings were related to the presence or absence of a history of thrombosis and anti-β₂GPI–dependent LAC activity.

Included in this study were 198 patients with a diagnosis of systemic lupus erythematosus (SLE; n = 176), lupuslike disease (LLD; n = 16), and primary antiphospholipid syndrome (PAPS; n = 6) that were seen consecutively at the lupus clinic of the University Medical Center, Utrecht, the Netherlands. Patients with SLE meet at least 4 American College of Rheumatology (ACR) criteria for the classification of SLE, and patients with LLD meet at least 1 to 3 of these criteria. Patients with PAPS have aPLs and a history of thrombosis in absence of other signs for a systemic autoimmune disease. By chart review, the number of objectively verified thromboembolic events was recorded for each patient. Computed tomographic scanning or magnetic resonance imaging were used to diagnose a patient having thrombosis of intracerebral vessels. Typical electrocardiographic features and an elevated-fraction creatine kinase muscle and brain (MB) diagnosed myocardial infarction. Peripheral arterial thrombosis and thrombosis of the distal aorta was diagnosed by arteriography or thrombectomy at surgery. Retinal thrombosis was documented by fundoscopy and fluorescence angiography. Deep vein thrombosis was diagnosed by ultrasonography or venography, pulmonary embolism by radionuclide lung scanning, and portal vein thrombosis by angiography. The Institutional Review Board of the University Medical Center Utrecht approved this study, and informed consent was obtained from all patients.

Blood samples were taken at an arbitrary visit of the patients to the outpatient department and were collected by venipuncture using plastic tubes containing 3.8% trisodium citrate (0.129 M) as the anticoagulant (9:1; vol/vol). To obtain platelet-poor plasma, the samples were centrifuged twice at 2000g for 10 minutes and subsequently stored at –50°C until use.

All patient samples were measured with 2 LAC assays: an activated partial thromboplastin time (APTT)–like assay (PTT-LA) and a dilute Russell's Viper Venom time (dRVVT). The APTT (PTT-LA; Diagnostica Stago, Gennevilliers, France) was performed as follows: 50 μL patient plasma was diluted 1:1 with normal pool plasma of 40 healthy volunteers and incubated with 50 μL APTT reagent (Diagnostica Stago). Coagulation was initiated by the addition of 50 μL CaCl₂ (25 mM). As a control, samples were tested in an APTT with actin-FS (Dade-Behring, Marburg, Germany) as activator, a LAC-insensitive assay.²²  Patients were considered positive when the ratioAPTT-LA/APTT-FS was greater than 1.20. The dRVVT was performed according to the instructions of the manufacturer (Gradipore, North Ryde, Australia) and considered positive when LAC screen/LAC confirm was greater than 1.20. A patient was considered LAC positive if 1 of the 2 LAC assays was positive.

Anticardiolipin antibodies were measured in an ELISA as described before.²³  Nine IgG/IgM calibrators were used to report aCL levels as GPL or MPL units. Levels above 10 GPL or GPM units were considered positive.

Anti–β₂GPI IgG antibody ELISA. IgG-class antibodies against β₂GPI were measured in an ELISA.²⁴  In short, plasma-purified β₂GPI (10 μg/mL in Tris-based solution [TBS]: 50 mM Tris in 100 mM NaCl) was coated onto in a high-binding ELISA plate (9102; Costar, New York, NY) by incubation for 1 hour. Then the plates were washed with 0.1% Tween in TBS (washing solution) and incubated with 4% bovine serum albumin (BSA) in TBS for 1 hour (blocking buffer) to prevent aspecific binding. The plates were washed again, followed by incubation with patient plasma (1:50 diluted in blocking buffer) for 1 hour. Then the plates were washed and incubated with a goat antihuman alkaline phosphatase–labeled antibody (diluted 1:1000; Biosource, Camarillo, CA) followed by staining with para-nitrophenyl phosphatase (PnPP; 0.4 mg/mL). A sample was considered positive if the optical density (OD) was greater than 3SD above the mean OD obtained with plasma from 40 healthy volunteers.

Mutants of β₂GPI. By a generous gift of Dr Iverson of La Jolla Pharmaceutical Company, we obtained recombinant full-length β₂GPI (DI-V); 8 deletion mutants of β₂GPI (comprising domain I [DI]; domains I and II [DI-II]; domains I, II, and III [DI-III]; domains I, II, III, and IV [DI-IV]; domains II, III, IV, and V [DII-V]; domains III, IV, and V [III-V]; domains IV and V [DIV-V]; and domain V [DV]); and 3 recombinant β₂GPI molecules with different point mutations in domain I (aspartate-alanine mutation at position 8 [D8A], glycine-glutamate mutation at position 40 [G40E], and an arginine-glycine mutation at position 43 [R43G]).²¹ 

Anti–β₂GPI IgG ELISA with β₂GPI deletion mutants. To determine the IgG reactivity against β₂GPI deletion mutants we used both 96-well hydrophilic high-binding ELISA plates (9102; Costar) and hydrophobic ELISA plates (2595; Costar). The plates were coated with deletion mutants in a concentration (10 μg/mL) that allows the use of the maximum binding capacity of the wells (1 hour at 37°C). The plates were washed 4 times with TBS/0.1% Tween and subsequently blocked with a 4% BSA/TBS solution. Patient plasma (diluted 1:50 in blocking solution) was added to the wells (50 μL/well) and incubated for 1 hour at 37°C. The plates were washed 4 times (0.1% Tween/TBS) and incubated with alkalic-phosphatase–labeled goat antihuman IgG antibodies (Abs; diluted 1:1000; Biosource) for 1 hour at 37°C. Staining was performed by using PnPP (Sigma, St Louis, MO) at a concentration of 0.6 mg/mL diluted in diethynolamine (DEA) buffer. Adding 2.4 M/L NaOH to the wells after 10 minutes at room temperature stopped the reaction. Absorption was measured at 405 nm. Extinctions obtained with normal pooled plasma were subtracted from those with patient plasma.

Inhibition experiments with point-mutated β₂GPI coupled to Sepharose beads. Recombinant β₂GPI and 3 different point-mutated forms of β₂GPI were covalently coupled to chelating Sepharose beads by the following procedure. Chelating Sepharose beads (Pharmacia, Uppsala, Sweden) were centrifuged for 15 seconds at 10 000g followed by removal of ethanol and washing of the beads with 4 volumes 50 mM EDTA (ethylenediaminetetraacetic acid)/H₂O. All washing procedures were done at room temperature by spinning down the beads at 57g. Subsequently beads were washed with 4 volumes H₂O, 4 volumes 50 mM CoCl₂/HO₂, 4 volumes H₂O, and 4 volumes PBS, followed by incubation for 15 minutes with point-mutated β₂GPI or full-length recombinant β₂GPI (2 volumes, 20 μM/mL) at room temperature during rotation. Then the beads were washed (with 6 volumes PBS) and incubated with 2 volumes 0.03% H₂O₂/PBS for 2 hours at room temperature during rotation, followed by washing with 4 volumes PBS, 1 volume 50 mM EDTA, 1 volume PBS, 1 volume 0.5 M imidazole/PBS, and 1 volume PBS. Blocking of aspecific binding places was done by incubation of beads with 4% BSA/PBS for 1 hour at room temperature during rotation.

Purified patient IgG was incubated for 1 hour at room temperature during rotation with Sepharose beads alone and Sepharose beads to which recombinant β₂GPI or mutated β₂GPI–Sepharose beads was coupled. Then the mixture was centrifuged for 1 minute at 57g. The supernatant was added to ELISA plates (Costar; high absorbing plates) coated with full-length recombinant β₂GPI and incubated for 1 hour at 37°C. After washing with 0.1% Tween-20/TBS five times, the plates were incubated with 50 μL/well alkalic-phosphatase–labeled goat antihuman IgG diluted 1:1000 in 4% BSA/PBS for 1 hour at 37°C. Then plates were washed 5 times with 0.1% Tween/TBS, and color development was performed by adding 100 μL/well PNPP at a concentration of 6 mg/10 mL DEA buffer. After 20 minutes, addition of 50 μL/well 2.4 M/L NaOH stopped the coloring reaction. Absorption was measured at 405 nm. From extinctions with purified patient IgGs, extinctions obtained with normal pool IgG were subtracted.

To discriminate between a β₂GPI-dependent LAC and a β₂GPI-independent LAC we made use of an aPTT-based clotting test (PTT-LA; Diagnostica Stago), as described before.²⁵  In short, 25 μL PTT-LA reagent, 50 μL patient plasma that was mixed 1:1 with normal pool plasma (of 40 healthy volunteers), and 25 μL of different concentrations (0, 25, 50, 100, 200 μM) of cardiolipin vesicles diluted in TBS were mixed and added to a KC-10 microcoagulometer (Amelung, Lemgo, Germany). The cardiolipin vesicles were made as described before.²⁶  After 3 minutes of incubation at 37°C, coagulation was initiated by the addition of 50 μL CaCl₂ and clotting time was measured. A LAC was considered β₂GPI dependent when the ratio of coagulation times of patient plasma and normal pool plasma was decreased by at least 0.05 with addition of cardiolipin vesicles at a concentration of 25 μM cardiolipin vesicles.²²  An example of a patient with a β₂GPI-dependent LAC and a patient with a β₂GPI-independent LAC is given in Table 1.

First, β₂GPI-deficient plasma was obtained by adding normal pool plasma (mixture of 40 healthy volunteers) to a Sepharose column coupled with 21B2 (a monoclonal murine anti-β₂GPI antibody, generous gift of Prof J. Arnout). After this procedure, β₂GPI could not be demonstrated anymore by chromatography. The β₂GPI-deficient plasma was reconstituted with either recombinant β₂GPI or mutated β₂GPI (R43G). Of this mixture, 75 μL was added to the KC-10 (Amelung) and incubated at 37°C. After 3 minutes of incubation, a dRVVT solution (Gradipore, Frenchs Forrest, Australia) was added and the coagulation time was measured.

Chi-square statistics was used to compare the prevalence of thrombosis with serologic findings. Odds ratios and 95% confidence intervals were calculated by binary logistic regression. Student t test or Mann-Whitney test was used to calculate differences between 2 groups. For these calculations we used SPSS (Chicago, IL). P values less than .05 were considered significant.

The median age of the 198 patients (180 female) was 33 years. In 60 (30%) of 198 patients there was a history of objectively verified thrombosis. Thirty-two patients had a history of arterial thrombosis (stroke, n = 19; myocardial infarction, n = 5; peripheral artery, n = 7; distal aorta, n = 2; retinal artery, n = 3) and 38 patients had a history of venous thrombosis (deep venous thrombosis, n = 30; pulmonary embolism, n = 19; thrombophlebitis, n = 7; portal vein thrombosis, n = 1). Anticardiolipin antibodies were present in 112 (57%) of 198 patients, LAC in 63 (32%) of 198 patients, and anti–β₂GPI IgG antibodies in 52 (26%) of 198 patients (Table 2). Table 3 shows the interassay variability as an indication for the quality of the assays used in this study.

To investigate the domain specificity of the anti–β₂GPI IgG antibodies that was found in 52 patients, separate wells of hydrophilic and hydrophobic ELISA plates were coated with 8 domain-deleted mutants of β₂GP1 and full-length recombinant β₂GP1. It appeared that the 52 samples with anti–β₂GPI IgG antibodies could be divided into 2 types (designated A and B) based on the reactivity that these had (Figure 1). On a hydrophilic high-binding plate the samples of type A (n = 30) only recognized full-length recombinant β₂GP1, whereas those of type B (n = 22) recognized full-length recombinant β₂GP1, domains I and V, and also domains I and V extended with additional domains. When a hydrophobic plate was used, samples of type A recognized not only full-length recombinant β₂GP1 (albeit the OD was lower) but also domain I. On a hydrophobic plate the reactivity of samples from type B was similar to that found with hydrophilic ELISA plates, albeit the ODs were lower. A typical example of the reactivity of samples of types A and B is presented in Figure 1.

We hypothesized that patients of group A recognize a positive epitope on domain I that is shielded off from recognition when coated to hydrophilic plates. To demonstrate the involvement of this positively charged epitope on domain I in recognition by type A anti-β₂GPI antibodies, we performed inhibition experiments. We coated Sepharose beads with native full-length recombinant β₂GPI or 1 of 3 point mutations (D8A, G40E, or R43G) of full-length recombinant β₂GPI, blocked the beads with BSA, and incubated them with IgG fractions isolated from plasma from 3 patients from type A. The supernatants were added to a hydrophilic plate that was coated with full-length recombinant β₂GPI. We found that in all 3 samples with type A anti-β₂GPI antibodies, the anti-β₂GPI activity could be fully absorbed by full-length recombinant β₂GPI and with D8A and only partially with G40E or R43G (Figure 2).

When we tested the affinity for β₂GPI in an anti–β₂GPI IgG-class ELISA in the presence of 1 M NaCl, 26 of 30 patients from group A recognized an epitope in the presence of 1 M NaCl (data not shown), indicating the recognition of a specific epitope rather than charge. Only 8 of 22 of the IgG of group B recognized β₂GPI in the presence of 1 M NaCl, indicating that the binding of those antibodies was of lower affinity (data not shown).

Out of the 30 samples from type A, 28 (93%) were LAC positive. In 23 of 28 samples the LAC activity was dependent on β₂GPI (Table 4). To give support to the association between anti-β₂GPI reactivity and β₂GPI-dependent LAC, we added anti–β₂GPI IgG isolated from plasma of 3 patients with type A reactivity to β₂GPI-depleted plasma that was reconstituted with either recombinant full-length β₂GPI or point-mutated β₂GPI (R43G). We observed an increase in coagulation time from 104.0 ± 3.4 (mean ± SEM, n = 4) seconds to 175.5 ± 32.4 (mean ± SEM, n = 4) seconds when plasma was reconstituted with 150 μg/mL recombinant β₂GPI and 100 μg/mL type A anti–β₂GPI IgG antibodies. However, little prolongation (104.0 ± 3.4 → 122.8 ± 17.3, mean ± SEM, n = 4) was seen when plasma was reconstituted with 150 μg/mL R43G and 100 μg/mL type A anti–β₂GPI IgG antibodies. This experiment is representative of 4 experiments performed in the same way.

A history of thrombosis was present in 32 of 52 patients with IgG anti-β₂GPI antibodies, in 25 (83%) of 30 patients IgG anti-β₂GPI antibodies with type A reactivity, and in 7 (32%) of 22 patients with type B reactivity (Table 4). The odds ratios for thrombosis were 6.7 (95% confidence interval [95% CI]: 13.5-3.4) for IgG anti-β₂GPI antibodies, 18.9 (95% CI: 53.2-6.8) for type A reactivity, and 1.1 (95% CI: 2.8-0.4) for type B reactivity.

To enable differentiation between patients with IgG anti-β₂GPI antibodies of type A and type B reactivity, one can coat domain I of β₂GPI onto a hydrophilic and a hydrophobic ELISA plate. Plasma from patients with type A IgG anti-β₂GPI antibodies do not recognize domain I on a hydrophilic plate but recognize domain I on a hydrophobic plate, whereas samples from patients with type B reactivity recognize domain I on both plates. A ratio of greater than 2 between the OD measured with the hydrophobic plate and the OD with the hydrophilic plate discriminates between type A and type B IgG anti-β₂GPI antibodies. This ratio is an indication for the relative amount of IgG anti-β₂GPI antibodies that recognize the positive charge on domain I. We determined a coefficient of variation of 3.7% (interassay variability) as an indication for the quality of this assay (Table 3). Figure 3 shows that within the group of patients with IgG anti-β₂GPI antibodies with type A reactivity, the ratio in those with a thrombotic history is significantly higher than in patients without thrombosis (P = .014). Within the group of patients with IgG anti-β₂GPI antibodies of type B reactivity, the ratios are low and there is no difference in ratio between patients with or without thrombosis.

The identification of the domains of β₂GPI that are involved in binding of anti-β₂GPI antibodies has been the subject of many studies.^13-21  The cumulative results show that each domain of β₂GPI has been indicated as the location where anti-β₂GPI antibodies may bind. A recent paper strongly suggests that for binding of anti-β₂GPI antibodies purified from patient plasma, a positively charged epitope, spanning amino acids 40-43 on domain I of β₂GPI, is very important. We tested this suggestion by studying the reactivity of 52 samples (52 patients) with anti–β₂GPI IgG antibodies that were found within a group of 198 patients with SLE, LLD, or PAPS. The major conclusions from this study are as follows: (1) not all anti–β₂GPI IgG antibodies have the same epitope specificity; (2) a substantial number of antibodies found in patients recognize an epitope around G40-R43 on domain I; (3) these latter antibodies seem to be the pathologic ones because their presence was highly correlated with a history of thrombosis; (4) the anti–β₂GPI IgG antibodies against epitope G40-R43 in domain I are antibodies that cause LAC activity; and (5) anti–β₂GPI IgG antibodies that do not recognize G40-R43 in domain I are not related to a history of thrombosis. For anti-β₂GPI antibodies of the IgM class we did not find an increase in their association with thrombosis by using the anti–domain I ELISA. Based on these observations we have developed a simple assay to detect a subset of anti–β₂GPI IgG antibodies that highly correlates with thrombosis. Although the proportion of patients with primary APS was very small in our population, previous studies found similar clinical and laboratory features for primary and secondary APS.²⁷  It seems likely that this also accounts for this anti–domain I ELISA.

By coating 8 domain-deleted mutants of β₂GPI and full-length recombinant β₂GPI on both hydrophilic and hydrophobic ELISA plates we discriminated 2 different reactivity patterns (designated types A and B) for the anti-β₂GPI antibodies. Samples with type A reactivity (present in 58% of anti-β₂GPI antibody–positive samples) recognize full-length β₂GPI but not deletion mutants when hydrophilic ELISA plates are used and a strong reactivity for domain I (and extensions of this domain) with use of hydrophobic ELISA plates (Figure 1). Samples designated type B do not seem to have a preferred binding site on β₂GPI. With hydrophilic ELISA plates these antibodies react both with full-length β₂GPI, domain I and V, and extensions of these. When hydrophobic ELISA plates are used, the pattern of reactivity against β₂GPI and domain-deleted mutants is similar, albeit the ODs are much lower.

We hypothesized that our observation of the reactivity of type A antibodies is dependent on the type of ELISA plate used and could indicate that type A anti-β₂GPI antibodies are directed against the G40-R43 epitope in domain I because it is conceivable that interactions between the positive charge of this epitope and the negative charge of a hydrophilic ELISA plate may hamper interaction of antibodies with that epitope.²¹  We showed that the reactivity of type A anti-β₂GPI antibodies against β₂GPI can be fully absorbed by full-length recombinant β₂GPI and full-length recombinant β₂GPI mutated at position 8 (D8A). When we absorbed type A anti-β₂GPI antibodies with full-length recombinant β₂GPI that had point mutations at 1 amino acid within the epitope G40-R43, namely R40E or R43G, 35% to 40% of the reactivity against β₂GPI remained (Figure 2). This strongly suggests that these amino acids are part of the epitope to which type A anti-β₂GPI antibodies are directed. We previously reported that β₂GPI-dependent LAC activity is a strong risk factor for thrombosis.²²  We now demonstrated that 23 (92%) of 25 samples with β₂GPI-dependent LAC activity contained type A anti-β₂GPI antibodies and that presence of type A antibodies is a strong risk factor for thrombosis (Table 4). The finding that reconstitution of β₂GPI-depleted plasma with full-length β₂GPI and type A anti-β₂GPI antibodies induces more prolongation of the clotting time than reconstitution with R43G and type A anti-β₂GPI antibodies suggested that type A anti-β₂GPI antibodies cause LAC activity. As a β₂GPI-dependent LAC activity correlates very strongly with thrombosis (odds ratio, 42.3), it also supports the concept that pathologic antibodies are directed against the epitope G40-R43.

Collectively, type B antibodies directed to β₂GPI are not correlated with a history of thrombotic complications. Whether this population contains clinical relevant subpopulations of antibodies needs further studies.

In conclusion, our study shows that anti-β₂GPI antibodies react with different epitopes on β₂GPI. However, a substantial number of patient samples with anti-β₂GPI antibodies (in our series 58%) recognize an epitope on domain I, including positions 40 and 43 (type A antibodies). The strong correlation between presence of such type A antibodies with histories of thrombosis and presence of β₂GPI-dependent LAC activity strongly suggests that type A antibodies constitute a pathologic subset. Our observations form the basis for a simple assay (using domain I coated onto hydrophilic and hydrophobic ELISA plates) for the detection of type A anti-β₂GPI antibodies.