IgG subclass distribution of antibodies against b2-GP1 and cardiolipin

Rheumatology 2001;40:1026–1032

IgG subclass distribution of antibodies against b2-GP1 and cardiolipin in patients with systemic lupus erythematosus and primary antiphospholipid syndrome, and their clinical associations M. Samarkos, K. A. Davies1, C. Gordon2, M. J. Walport1 and S. Loizou1 5th Department of Internal Medicine, Evangelismos Hospital, Athens, Greece, 1 Rheumatology Section, Division of Medicine, Hammersmith Hospital, Imperial College School of Medicine, London and 2Department of Rheumatology, Division of Immunity and Infection, University of Birmingham, Birmingham, UK Abstract Objectives. To determine the immunoglobulin G (IgG) subclass distribution of anticardiolipin (aCL) and anti-b2-glycoprotein 1 (b2-GP1) antibodies (ab2-GP1), and to examine possible associations between the different ab2-GP1 and aCL subclasses and the main clinical manifestations of the antiphospholipid syndrome (APS). Methods. We studied 130 patients with systemic lupus erythematosus and 35 patients with primary APS. We used enzyme-linked immunosorbent assays to measure IgG aCL and ab2-GP1 and to determine the IgG subclass distribution of these two autoantibodies. Results. When the number of patients positive for each subclass was examined, IgG3 and IgG2 aCL were more frequent (63.5 and 54.1% of patients were positive for the two subclasses, respectively), while for ab2-GP1 IgG2 was the most prevalent subclass (81.8% of patients were positive). IgG2 aCL was significantly associated with arterial thrombosis (P = 0.023) and fetal loss (P = 0.013), and IgG3 aCL was significantly associated with arterial thrombosis (P = 0.0003) and fetal loss (P = 0.045). IgG2 ab2-GP1 was associated with venous thrombosis (P = 0.012) and IgG3 ab2-GP1 was associated with venous thrombosis (P = 0.036) and fetal loss (P = 0.024). Conclusions. The IgG2 predominance of ab2-GP1 suggests that the antibody response against b2-GP1 may be T-cell-independent. As IgG2 and IgG3 differ in their effector functions, their association with the same clinical manifestations (i.e. thrombosis and fetal loss) suggests that more than one mechanism may be involved in the pathogenesis of thrombosis and fetal loss in APS. KEY WORDS: Systemic lupus erythematosus, Anticardiolipinuantiphospholipid antibodies, IgG subclasses.

Antiphospholipid antibodies (aPL) are a heterogeneous group of autoantibodies and are often found to be elevated in the sera of patients suffering from a wide range of clinical conditions, mainly autoimmune and infectious diseases w1–3x. In autoimmune diseases, aPL are associated with a spectrum of clinical manifestations, such as arterial and venous thrombosis, recurrent Submitted 4 July 2000; revised version accepted 27 March 2001. Correspondence to: M. Samarkos, 5th Department of Internal Medicine, Evangelismos Hospital, 45– 47 Ipsilantou Street, Athens 106 76, Greece.

fetal loss and thrombocytopenia. The antiphospholipid syndrome (APS) is characterized by the combination of increased aPL titres with at least one of these manifestations. Until relatively recently, aPL have been detected by a range of assays, each detecting a specific subgroup of aPL, e.g. anticardiolipin antibody (aCL)-specific enzyme-linked immunosorbent assays (ELISAs) and lupus anticoagulant assays w4 –6x, and it was thought that aPL were directed specifically against negatively charged phospholipids w1x. However, recent data support the view that, in autoimmune disease patients,

1026

ß 2001 British Society for Rheumatology

IgG subclasses of aCL and ab2-GP1 antibodies

the aPL detected by solid-phase assays do not bind to phospholipids alone but bind either to a complex of phospholipids and plasma proteins, such as b2-glycoprotein 1 (b2-GP1) and prothrombin, or bind directly to these proteins alone w6–8x. This has been shown to be the case in systemic lupus erythematosus (SLE) and primary antiphospholipid syndrome (PAPS) patients, in whom antibodies against b2-GP1 (ab2-GP1) have been detected by the use of b2-GP1 as the ligand in solid-phase ELISAs w9x. The immunoglobulin (Ig) G subclass distribution of autoantibodies may give insight into their mode of action, as IgG subclasses differ in their ability to activate complement and in their properties of binding to Fcc receptors w10, 11x. Furthermore, IgG subclass responses are greatly dependent on their target antigen: protein antigens generally elicit a T-cell-dependent IgG1 and IgG3 response, whereas carbohydrate antigens induce a T-cell-independent IgG2 response w12–14x. Hitherto, a relatively large number of studies have examined the prevalence of increased levels of antibodies against b2-GP1 (ab2-GP1) in patients with SLE anduor the APS w15–18x, but only one study has examined the IgG subclass distribution of ab2-GP1 w19x. In contrast, several studies on IgG aCL subclasses have been published, with conflicting results w20–23x. Interestingly, the most recent of these studies has reported an association between IgG2 aCL and the occurrence of thrombosis w23x. The present study is the first in which the IgG subclass distributions of ab2-GP1 and aCL have been determined simultaneously in the sera of SLE and PAPS patients. We also attempted to clarify whether there were any differences in subclass distribution in different patient subgroups (e.g. in patients with high vs low autoantibody levels). Additionally, we examined possible associations between the four ab2-GP1 and aCL subclasses, and any of the main clinical manifestations of the APS.

Patients and methods Patients We studied retrospectively 130 patients (120 female, 10 male; age 19–83 yr, median 43 yr) with SLE from the Department of Rheumatology, Division of Immunity and Infection, University of Birmingham, and 35 patients with PAPS (29 female, 6 male; age 20–71 yr, median 39) from the Rheumatology Section, Imperial College School of Medicine at Hammersmith Hospital. All SLE patients fulfilled the American College of Rheumatology criteria w24x and all PAPS patients had elevated levels of IgG anduor IgM anticardiolipin antibodies or a positive lupus anticoagulant test, and at least one of the following clinical manifestations of the antiphospholipid syndrome: arterial thrombosis (peripheral arterial thrombosis, thrombotic stroke), venous thrombosis (deep vein thrombosis or pulmonary embolism), at least two fetal losses, and thrombocytopenia

1027

(platelet count less than 120 000uml). The clinical data were obtained from medical records. Serum samples were stored at 2708C. The control group consisted of 88 healthy blood donors. At the time when blood was drawn, all patients had given informed consent for the use of their blood sample for research. Concentrations of IgG and IgM aCL and ab2-GP1 were measured in all patients and controls. We also determined the IgG subclass distribution of aCL and ab2-GP1 in all patients who tested positive for the IgG isotype for each of these two autoantibodies. aCL ELISA aCL were measured using the method of Loizou et al. with minor modifications w25x. Briefly, microtitre plates (MP01, Life Sciences International, Basingstoke, UK) were coated with 2 mguwell of cardiolipin (Sigma Biosciences, Poole, Dorset, UK) in ethanol and were left to dry at 48C. Subsequently the plates were blocked overnight with 5% adult bovine serum (ABS) in phosphate-buffered saline (PBS). Serum samples diluted 1 : 100 in 5% ABSuPBS were added and incubated for 2 h at room temperature. Alkaline phosphatase (AP)-conjugated goat anti-human IgG c-chain-specific or goat anti-human IgM m-chainspecific (Sigma) antibodies were added for the determination of the respective isotypes. After the addition of p-nitrophenyl phosphate, the optical density at 405 nm was measured with an automated ELISA reader (MultiScan MCCu340; Titertek Life Sciences International). The levels of both aCL isotypes were calculated in arbitrary ELISA units (AEU) from an eight-point standard curve. Serum samples with low and high antibody levels were included as internal controls in all cases. b2-GP1 purification b2-GP1 was purified from normal human plasma using a modification of the method of Polz et al. w26x. The initial precipitation steps were followed by affinity chromatography on a heparin–Sepharose column (HiTrap Heparin; Pharmacia LKB Biotechnology, Uppsala, Sweden). A second affinity chromatography step was performed on a protein G column (HiTrap Protein G; Pharmacia). The final product was delipidated by mixing with butanol. The antigenic properties and purity of the final product were confirmed by double radial immunodiffusion, by ELISA against a rabbit anti-human b2-GP1 antiserum (Dako, Glostrup, Denmark), and by sodium dodecyl sulphate polyacrylamide gel electrophoresis, which gave a single band at approximately 50 kDa (data not shown). ab2-GP1 ELISA Microtitre plates (Immulon 2; Dynex Technologies, Billingshurst, UK) were coated overnight at 48C with 0.25 mguwell of our b2-GP1 preparation in 0.2 M boratebuffered saline (BBS) pH 8.4, while an appropriate number of wells were coated with 0.2 M BBS alone for estimation of non-specific binding. Plates were subsequently blocked with 0.5% BSA, 0.4% Tween-20uPBS for 2 h at room temperature. Serum samples (50 ml)

1028

M. Samarkos et al.

diluted 1 : 100 in 0.5% BSA, 0.4% Tween-20 in PBS were added and incubated at room temperature for 1 h. Alkaline phosphatase-conjugated goat F(ab9)2 fragments of anti-human c-chain-specific IgG diluted 1 : 3000 or m-chain-specific IgM diluted 1 : 1000 (Sigma) were added for the determination of the different isotypes. All washing steps were performed using 0.075% Tween-20uPBS. The final step was the same as for the aCL ELISA. Non-specific binding from non-antigencoated wells was subtracted from all samples, and the levels of both ab2GP1 isotypes were calculated in AEU from a standard curve of eight points. Internal controls were included on each plate. aCL and ab2-GP1 IgG subclasses In preliminary experiments we established the equipotency of the four anti-human IgG subclass-specific monoclonal antibodies w10x. Four microtitre plates were each coated with a different purified human myeloma IgG subclass (IgG1k to IgG4k; all from Binding Site, Birmingham, UK) at concentrations of 0.025–0.8 mguml. After blocking, chequerboard dilutions (1 : 250–1 : 100 000) of each of the respective murine anti-human subclass-specific monoclonal antibodies were added to each plate (anti-IgG1 IUISuWHO clone 8cu6–39, anti-IgG2 clone HP6014, anti-IgG3 clone HP6050 and anti-IgG4 clone HP 6025; all from Sigma) followed by an AP-conjugated rabbit anti-mouse IgG. The optimum dilution for each subclass-specific antibody was the dilution which gave a similar optical density for the same concentration for each of the different myeloma subclasses (Fig. 1). These were 1 : 500 for anti-IgG1, 1 : 6000 for anti-IgG2, 1 : 500 for anti-IgG3 and 1 : 50 000 for anti-IgG4. The microtitre plates used, antigen coating and blocking steps for the IgG subclass ELISAs were the same as those used for the total IgG aCL and ab2-GP1 ELISAs, except that all serum samples to be tested were diluted 1 : 50 and added to each of four plates used for

the determination of each of the four subclasses. This was followed by addition of the appropriate concentration (as determined above) for each of the subclassspecific antibodies, to one of each of the four plates. After 2 h of incubation at room temperature, AP-conjugated rabbit anti-mouse IgG (Sigma) was added, followed by addition of the chromogenic substrate. The colour reaction on all four plates was stopped simultaneously. Two wells on each plate were coated with 0.5 mguwell human IgG subclass calibrator (Binding Site), which served as an interplate control. After subtraction of non-specific binding, the net absorbances for the four IgG subclasses were summed (total absorbance), and for each subclass we calculated the absorbance as the percentage of the total absorbance of the antigen-specific IgG. To measure rheumatoid factor (RF), we used a semiquantitative latex agglutination kit (Rapitex RF; Behringwerke, Marburg, Germany). Normal ranges Patient sera were considered positive if the IgG and IgM aCL and ab2-GP1 levels were more than four standard deviations (SD) above the mean level of 88 healthy blood donors. Patient sera were considered positive for one subclass if the percentage levels of this subclass exceeded the upper limit of the accepted normal range (Fig. 2) for each of the respective subclasses in normal total serum w10, 11x. Statistical analysis The results were analysed with the statistical package Prism v2.0 (GraphPad Software, San Diego, CA, USA). We used the Mann–Whitney test to compare means and Spearman’s rank sum test for correlations between numerical variables. To compare proportions and detect associations between nominal variables, we used Pearson’s two-tailed x2 test or Fisher’s exact test as appropriate. A result was considered statistically significant when P < 0.05.

Results

FIG. 1. Binding curves of IgG subclass-specific monoclonal antibodies, each at its optimum dilution, to respective myeloma proteins coated on the microtitre plate.

IgG and IgM aCL and ab2-GP1 Our normal range for IgG aCL was less than 14 AEU and for IgM aCL less than 10 AEU. The numbers of patients positive for each isotype in the SLE and PAPS groups are shown in Table 1. Our normal range for IgG ab2-GP1 was less than 12.5 AEU and for IgM ab2-GP1 less than 9.5 AEU. The numbers of positive patients for each isotype in the two patient groups (SLE and PAPS) are shown in Table 1. When patients positive for only IgG or IgM aCL were examined, it was found that levels of IgG ab2-GP1 were higher in PAPS than in SLE patients (54.1 vs 9.6 AEU; Mann–Whitney test, P < 0.0001); in the same subgroup of patients (positive for only IgG or IgM aCL), more PAPS than SLE patients were positive for IgG ab2-GP1 (15u34 vs 10u57; x2 = 7.54, P = 0.006).

IgG subclasses of aCL and ab2-GP1 antibodies

1029

(P < 0.0001). When IgG ab2-GP1 and IgG aCL levels were compared in SLE patients who were positive for only IgG ab2-GP1, a strong correlation was also seen (P = 0.0042). Only nine of the 40 (22.5%) IgG or IgM ab2-GP1-positive patients were negative for both IgG and IgM aCL. IgG aCL subclasses The percentage levels for each of the IgG aCL subclasses in all 74 IgG aCL-positive patients wmean value and 95% confidence intervals (CI)x were as follows: IgG1, 40.1% (33.6–46.7%); IgG2, 32.8% (26.6–39%); IgG3, 23.7% (18.6–28.8%); and IgG4, 3.3% (1.3–5.2%), as shown in Fig. 2a. Although IgG1 was the predominant subclass in terms of mean percentage levels, when we considered positivity for each subclass, 47 out of 74 (63.5%) patients were positive for IgG3 and 40 out of 74 (54.1%) patients were positive for IgG2 (Table 2). We categorized all our IgG aCL-positive patients as ‘weakly positive’ when the level was lower than 40 AEU and ‘moderately to highly positive’ when the level was higher than 40 AEU. Moderately and highly positive IgG aCL patients had significantly higher mean IgG1 levels (45.8 vs 30.9%, P = 0.03) and lower IgG3 levels (18.8 vs 30.4%, P = 0.006) than weakly positive IgG aCL patients.

FIG. 2. Subclass distribution of IgG aCL (a) and ab2-GP1 (b) in the total study population. Total IgG is the sum of the absorbance of all the subclasses. For each subclass, the ratio of its absorbance to the total absorbance is shown as a percentage. Columns represent mean and bars represent 95% CI. TABLE 1. Frequencies of IgG anduor IgM aCL- and ab2-GP1-positive patients

Autoantibody

SLE (n = 130)

IgG aCL IgM aCL IgG or IgM aCL IgG ab2-GP1 IgM ab2-GP1 IgG or IgM ab2-GP1

42 33 57 17 8 24

(32.0%) (25.4%) (43.8%) (13.1%) (6.1%) (18.7%)

Pa IgG1 > IgG3 > IgG4. When the number of patients positive for each subclass was examined, IgG2 and IgG3 aCL were more frequently elevated, whereas IgG2 was the most prevalent subclass for ab2-GP1. The relative concentrations of IgG subclasses for aCL were different in patients who were also positive for IgG ab2-GP1 in comparison with IgG ab2-GP1-negative patients. Patients who were positive for IgG ab2-GP1 had significantly higher levels of IgG2 aCL than patients who were negative for IgG ab2-GP1 (47.4 vs 26.7%, P = 0.003). We also found that patients

M. Samarkos et al.

1030

TABLE 2. Numbers of patients positive for each IgG subclass and autoantibody Positive for IgG subtype IgG1 IgG2 IgG3 IgG4

TABLE 3. IgG subclass distribution in relation to aCL and ab2-GP1 profile IgG aCL+

IgG aCL-positive (n = 74) 11 40 47 10

(14.9%) (54.1%) (63.5%) (13.5%)

Pa n.s.