Anticardiolipin IgG subclasses association of ... - Wiley Online Library

ARTHRITIS & RHEUMATISM Vol 40, No 11. November 1997, pp 1998-2006 0 1997, Amcrlcdn Collcgc of Rheumatology

1998

ANTICARDIOLIPIN IgG SUBCLASSES Association of IgG2 with Arterial and/or Venous Thrombosis

LISA R. SAMMARITANO, SONIA NG, RACHEL SOBEL, SIU KONG LO, RONIT SIMANTOV, RICHARD FURlE, ALAN KAELL, ROY SILVERSTEIN, and JANE E. SALMON Objective. To determine whether the presence of anticardiolipin antibodies (aCL) of a specific IgG subclass is associated with clinical complications of the antiphospholipid antibody syndrome (APS) and whether polymorphisms of Fc receptors for IgG (FcyR) with differential binding preferences contribute to an increased risk of thrombotic complications. Methods. In 60 patients with IgG aCL, we assessed clinical complications of the APS, measured the level of antibody activity, and determined the IgG subclass distribution of aCL by a modified enzymelinked immunosorbent assay (ELISA) with murine antihuman IgG subclass monoclonal antibodies. Selective IgG subclass adsorption studies were performed to determine the relative contribution of specific IgG subclasses to overall aCL activity. Fcy receptor IIA (FcyRIIA) genotypes of aCL patients with thrombosis and of non-systemic lupus erythematosus controls were determined by polymerase chain reaction amplification of genomic DNA and allele-specific probes. Results. IgG2 aCL, detected in 75% of the patients, was the major subclass of aCL. Selective adsorption studies demonstrated that IgG2, in contrast to IgGl, was the predominant subclass responsible for aCL reactivity. IgG2 aCL was the only subclass associated with clinical complications, specifically, arterial Supported in part by NIH grants P50-AR-42588 and AR-38889. Lisa R. Sammaritano, MD, Sonia Ng, BA, Rachel Sobcl, BA, Jane E. Salmon, MD: Hospital for Special Surgery, and New York Hospital, Cornell Medical Center, New York, New York; Siu Kong Lo, PhD, Ronit Simantov, MD, Roy Silverstein, MD: New York Hospital, Cornell Medical Center, New York, New York; Richard Furie, MD: North Shore University Hospital, Manhasset, New York; Alan Kaell, MD: State University of New York at Stony Brook. Address reprint requests to Jane E. Salmon, MD, Hospital for Special Surgery, 535 East 70th Street, New York, NY 10021. Submitted for publication December 27, 1996; accepted in revised form June 26. 1997.

and/or venous thrombosis (P < 0.04). The presence of FcyRIIA-H131, a receptor expressed on platelets, monocytes, and endothelial cells and the only human FcyR which efficiently recognizes IgG2, was associated with thrombosis in aCL patients. Among 45 high-titer (>40 GPL [IgG phospholipid] units) aCL patients with thrombosis, 40% were homozygous for FcyRIIA-H131, compared with 25% of disease-free controls (P = 0.042). Conclusion. While all 4 IgG subclasses are found in autoimmune aCL, only the presence of IgG2 is significantly associated with thrombotic complications. Reactivity in aCL ELlSA is largely due to the presence of IgG2 in high-titer patients. The presence of IgG2 aCL, particularly in association with FcyRIIA-H131, may be a useful clinical predictor of increased thrombotic risk in patients with autoimmune IgG aCL. Allelic variants of FcyRIIA with distinct capacities to interact with IgG subclasses provide a mechanism for genetic susceptibility to an autoantibody-induced prothrombotic state.

Patients with anticardiolipin antibodies (aCL) have a high incidence of arterial and venous thrombosis, recurrent fetal loss, and thrombocytopenia, although not all patients with aCL develop these complications. Predictors of such events in this syndrome are poorly defined. The presence of IgG aCL isotype and a high level of IgG aCL activity are clearly associated with increased risk for clinical complications. In contrast, phospholipid specificity, antibody avidity, and light chain predominance do not predict clinical events (1). It is likely that the development of clinical complications in patients with aCL is the result of multiple risk factors relating to characteristics of the autoantibody and the genetic background of the host. Potential mechanisms of aCL-induced thrombosis have been the subject of much investigation. Our recent work demonstrated endothelial activation in the

1999

ASSOCIATION OF IgG2 aCL WITH THROMBOSIS

presence of IgG aCL from patients with autoimmune disease (2), but the specific properties of the autoantibody required for this effect are undefined. A characteristic of aCL which may be important to the induction of a prothrombotic phenotype is the IgG subclass distribution. The nature of the IgG subclass determines the capacity to activate the classical complement pathway and to bind receptors for IgG. Several studies suggest that autoimmune disease-associated aCL have an IgG2 subclass predominance, whereas infection-induced aCL, which are not associated with thrombosis, are primarily composed of IgGl and IgG3 subclasses (3-5). However, the presence of specific subclasses has not been correlated with the development of particular clinical events. The qualitative nature of the autoimmune response in the context of effector cells may dictate pathogenetic potential. Receptors for the Fc portion of IgG (FcyR) provide the crucial link between humoral immunity and IgG-triggered effector functions. Human FcyR are diverse in structure and function. There are 3 families of FcyR (Fcy, FcyRII, and FcyR111) which vary in their affinity for IgG and the cell types where they are expressed (for review, see refs. 6 and 7). The IgG preferences have been described for all FcyR, with the most striking difference being the binding of IgG2 to 1 allele of FcyRIIA, the only FcyR capable of efficiently binding this subclass (8,9). The 2 alleles of FcyRIIA, which are codominantly expressed, differ substantially in their ability to ligate human IgG2 as a consequence of a single amino acid substitution at position 131 in the second extracellular domain (arginine R131 or histidine H131). FcyRIIA-H131 is the only receptor which binds IgG2 and, therefore, is the only receptor on monocytes (which express both FcyRI and FcyRIIA) that has the potential to generate procoagulant activity as a consequence of interaction with IgG2 complexes. As the only FcyR expressed on platelets and endothelial cells, FcyRIIA has the capacity to initiate a prothrombotic phenotype, but this potential is not realized unless the appropriate IgG subclass and FcyRIIA allele are present (10,ll). Thus, the existence of allelic variants of FcyRIIA with distinct capacities to interact with IgG subclasses provides a mechanism for genetic susceptibility to autoantibody-induced thrombosis. Given their differential capacity to trigger effector functions, the IgG subclass distribution of aCL are therefore of particular interest. We found that IgG2 aCL were significantly associated with a history of the presence of complications. Among aCL patients with thrombosis, there was also a significant increase in the frequency of homozygosity for FcyRIIA-H131, the variant

Table 1. Clinical characteristics of 60 patients with IgG aCL*

No. (9%) of patients 30 (SO) 30 (SO) 53 (88) 7 (12) 40 (66) 19 (33)

Primary aCL Secondary aCL (SLE + aCL) Females Males Patients with any complication Patients with > 1 complication Patients with specific complications Thrombosis Arterial Venous Arterial and venous Miscarriage, no. (5%) of females Thrombocytopcnia GPL value, mean ? SD Patients with 2 4 0 GPL units Patients with 280 GPL mils 'I aCL = anticardiolipin antibodies; SLE tosus; GPL = IgG phospholipid units.

27 (45) 18 (30) 13 (22) 4 (7) I 6 (30) 15 (25) 60 i- 22 46 (77) 19 (32) =

systcrnic lupus erytherna-

which recognizes IgG2. Confirming the contribution of IgG2 to aCL binding, removal of this subclass from aCL-positive sera greatly reduced overall aCL activity in enzyme-linked immunosorbent assay (ELISA). The identification of a specific IgG aCL subclass and an FcyRIIA that are associated with clinical complications has implications for disease pathogenesis and therapy. PATIENTS AND METHODS Subjects. Sixty patients with autoimmune IgG aCL were identified from the Hospital for Special Surgery Autoimmune Registry and Repository. All patients had IgG aCL detected on at Icast 2 occasions a minimum of 6 weeks apart. Patients with only IgM and/or IgA aCL but without IgG aCL were not included. Thirty patients fulfilled the American College of Rheumatology criteria for systemic lupus erythematosus (SLE) (12), and the remaining 30 patients had isolated or primary aCL. Clinical characteristics of the population are summarized in Table 1. Anticardiolipin antibody-related complications were defined as spontaneous venous or arterial thrombosis, including transient ischemic attacks; 2 or more unexplained first trimester miscarriages or I or more unexplained second or third trimester fetal losses; and thrombocytopenia with platelet counts 40 IgG phospholipid (GPL) units and a history of arterial or venous thrombosis. Sixteen patients were identified from the Hospital for Special Surgery Autoimmune Registry and Repository, and 29 patients were recruited from other centers in New York.

2000

White patients were selected for this study because FcyRIIA allele distribution varies significantly among different ethnic groups (13,14). Twenty-one patients had primary antiphospholipid antibody syndrome (PAPS), and 24 patients had SLE and aCL. There were 39 women and 6 men. The mean -t S D GPL value was 68 i 18 units. Venous thrombosis was documented in 6 7 4 and arterial thrombosis in 42%. Controls (n = 103) were American white subjects with no history of systemic autoimmune disease or thrombosis. ELISA for aCL. The cardiolipin solid-phase assay was performed as previously described (15). Briefly, 100-pI samples of a 1:50 dilution of patient serum in phosphate buffered saline containing 10% adult bovine serum were incubated at 4°C in a microtiter plate coated with 1.2 pgiwell of cardiolipin (Sigma, St. Louis, MO) in ethanol. Plates were blocked with 2%1 bovine serum albumin (BSA), washed, and adherent antibody was identified by t h e addition of alkaline phosphatase-conjugated goat anti-human IgG. G P L units (1 GPL unit = activity of 1 pg of affinity-purified TgG aCL antibody) were calculated using international standards supplied by the Rayne Institute (London, UK). This assay satisfies the International Standardization Workshop criteria (16). Patients with GPL units >5 SD above the mean of 20 normal sera (>1S GPL units) were identified as positive. Values 2 4 0 GPL units were considered to be high levels of aCL activity. IgG subclass analysis. IgG subclasses were identified by ELISA as described elsewhere ( 3 ) , using optimal dilutions of sera based on the concentration of each IgG subclass (150 for IgGl and IgG2, 1:lO for lgG3, 1:5 for lgG4). Mouse monoclonal antibodies to human IgG subclasses were obtained from ICN Biomedical (Costa Mesa, CA) and used at working dilutions, which gave a similar value for equal quantities of purified human IgG of each subclass. A positive result was defined as a value >5 SD above the mean of 20 sera from normal donors. IgG subclass-specific adsorption. Agarose beads were coupled to murine anti-human IgGl antibody, murine antihuman IgG2 antibody, or bovine serum albumin with the AminoLink kit (Pierce, Rockford, IL) according to the manufacturer’s recommended protocol. Sera with G P L units 280 at a 1 5 0 dilution were incubated with an excess of each preparation of protein-coupled heads for 1 hour at room temperature, with frequent vortexing. After centrifugation, supernatants were removed and assayed with the standard aCL ELISA. Adhesion assay. Endothelial cell activation was measured utilizing a monocyte adhesion assay that has been previously described (2). Briefly, human umbilical vein endothelial cells (HUVEC) cultured on Terasaki plates first were incubated with aCL-containing sera for 4 hours at 37°C before the addition of MM6, a human monocyte cell line that is heterozygous, expressing FcyRIIA-H131 and FcyRIIA-R 131 (Salmon J: unpublished observations). Adhesion was allowed to proceed for 15 minutes at 3TC, after which nonadherent cells were removed by washing. Adherent cells on HUVEC were fixed with 1 % paraformaldehyde and counted on an inverted microscope using a 10 X 10-mm grid. The results of 4 rcplicates were averaged. FcyRIIA alleles. D N A was isolated from peripheral blood (Puregcne kit; Gentra Systems, Minneapolis, MN). For genotyping, we utilized polymerase chain reaction (PCR)

SAMMARITANO ET AL

amplification of FcyRIIA EC2 from genomic DNA and hybridization of DNA dot-blots with H131 and R131 allelespecific oligonucleotide (ASO) probes as previously described (13,17). Briefly, D N A was selectively amplified with FcyRIIA gene-specific primer pairs (distinguishable from FcyRIIB and C gene by differences in intron 4). Dot-blots of PCR products were hybridized with digoxigenin-labeled A S 0 probes. Sequences of A S 0 differ by 1 nucleotide (494A and 494G) and permit the characterization of H131 and R131 alleles, respectively. Because allele frequencies vary among ethnic groups, analysis was confined to white subjects: 45 patients with high levels of aCL and a history of thrombosis, and 104 controls. Statistical analysis. Groups of patients and GPL levels were compared using t-tests, chi-square, Fisher’s exact, or nonparametric tests of association as appropriate. Multivariate analysis was performed using those independent variables which indicated statistical significance in correlation with the dependent variable. The specific analysis is indicated in the Results section. Observed P values are quoted wherever possible ro that the reader can judge the level of significance.

RESULTS IgG subclasses of aCL. In 60 patients with aCL (mean t SD 60 ? 22 GPL units), we determined the presence of the 4 IgG subclasses (IgG1, IgG2, IgG3, and IgG4) that contribute to aCL activity. As shown in Table 2, IgG2 was the most common subclass (75% positive), whereas IgGl was the least common subclass (33% positive). IgG3 and IgG4 were present in 55% and 60%, respectively. This pattern of IgG2 dominance was evident in both PAPS and SLE (81% and 6996, respectively). The frequency of IgG2 positivity was 91% in the high-level aCL patients (240 GPL units; n = 46), suggesting an association between IgG2 and aCL level (Table 2), which was corroborated by regression analysis of IgG2 subclass level (by ELISA) and aCL level (r = 0.7, P < 0.001). When patients’ sera were stratified according to the presence or absence of each of the 4 IgG subclasses, the highest level of aCL activity was found in the IgG2-positive group (mean ? SD 68 i 16 GPL units). In contrast, IgGZnegative patients had 2-fold less aCL activity (34 i 15 GPL units; P < 0.001, by t-test). The differences in total aCL activity between patients positive and negative for IgG1, IgG3, or IgG4 were of much smaller magnitude (ratio mean GPL units for subclasspositive and subclass-negative were IgGl 1.4 [P < 0.0011, IgG3 1.3 [P < 0.011, IgG4 1.3 [P < 0.021). Among the 4 IgG subclasses, statistical analysis revealed the highest association between IgG2 and total aCL activity (P < 0.0001, by Mann-Whitney U test). In multivariate analysis, the aCL level correlated with the presence of


2001

Table 2. IgG subclass distribution in aCL-positive patients”

IgGl IgG2 IgC3 IgG4

PAPS (n = 30) No. (96)

SLE + APS (n = 30) No. (5%)

240 GPI, unit5 (n = 46) No. (55)

Total (n = 60) NO. (%,)

9 (30) 25 (81) 17 (56) 21 (65)

11 (35) 20 (69) 15 (52) 15 (55)

18 (39) 41 (91) 30 (65) 31 (67)

20 (33) 45 (75) 33 (55) 36 (60)

* Values are the percentage of patients positive for each IgG subclass. The anticardiolipin antibody (aCL) IgG subclass distribution was determined by enzyme-linked immunosorbent assay, as described in Patients and Metbods. Serum dilutions used in the assays for IgCl and IgG2 were 150; those for IgG3 and IgG4 were 1:10 and 1:5, respectively. PAPS = primary antiphospholipid syndrome; SZE = systcmic lupus erythematosus; GPL units = IgG phospholipid units.

IgG1, IgG2, IgG3, and IgG4, and all of the subclasses correlated highly with each other. Association of aCL IgG subclasses with clinical complications. We compared the distribution of the 4 IgG subclasses in patients who had complications with those who did not have clinical complications. IgG2 was associated with venous and/or arterial thrombosis (P < 0.04, by Fisher’s exact test, 2-tailed) (Figure l),but not with thrombocytopenia or fetal loss. In patients with IgG2 aCL, there was an increased risk of arterial and/or venous thrombosis (53%), whereas in patients without IgG2 aCL, 80% were free of thrombosis. In contrast,

there were no differences in the prevalence of IgG1, IgG3, and IgG4 aCL in patients with or without thrombosis, thrombocytopenia, or fetal loss (Figure 1). The presence of IgM aCL was found not to be associated with any of these clinical complications. Multivariate analysis of aCL level (continuous variable), aCL IgG subclass (categorical variable), and clinical complications was confounded by the strong correlations among the independent variables. The level of aCL activity was the most consistent predictor of clinical events (thrombosis or any complication) and was the only variable with a statistically significant associa-

*

+

+ lgGl

+

t

lgG3

lgG2

lgG4

aCL IgG Subclass

Figure 1. Association of the presence or absence of each IgG subclass with arterial and/or venous thrombosis in 60 patients. Open bars represent percentage of patients positive for IgC subclass; solid bars represent percentage of patients negative for IgG subclass. In patients with anticardiolipiri antibody (aCL) containing IgG2, there was an increased risk of thrombosis (P< 0.04, by Fisher’s exact test, 2-tailcd).

SAMMARTTANO ET AL

2002

Table 3. Distribution of FcyRIIA alleles in aCL-positive patients with thrombosis and in disease-free controls*

Genotype, no. (%) of group H131iH131 R13l i H l 31 R131iR131 Allelic frequency, % H131 R131

aCL patients with thrombosis (n = 45)

Control subjects (n = 103)

18 (40) 18 (40) 9 (20)

25 (24) 53 (51) 25 (24)

60 40

50 50

* Odds ratio for the risk of thrombosis in patients with anticardiolipin antibodies (aCL) homozygous for FcyRII-H131/H131 compared with H131iR131 and R131iR131 was 2.08 (YS% confidence interval 0.9924.4; P = 0.042, by Fisher’s exact test, 1-tailed).

tion in the multivariate analysis, predicting 65% of the cases in the thrombosis analysis. Like the aCL level, when the continuous variable of IgG2 level was used and compared with the presence of other subclasses, it predicted 84% of the positive thrombosis cases. That the variables are highly correlated and that continuous variables are more sensitive to the analysis limit the utility of multivariate analysis for these data. The distribution of FcyRIIA alleles is altered in aCL patients with thrombosis. Given the suggestion of a role of IgG2 aCL, we determined the distribution of FcyRIIA allelic variants in a group of IgG aCL patients with thrombosis. We performed these studies based on the rationale that FcyRIIA-H131 is the only human FcyR which efficiently recognizes IgG2, and might therefore be associated with thrombotic risk. In a cross-sectional multicenter study, we identified 45 white patients with high-titer IgG aCL and thrombotic complications (42% with arterial thrombosis and 67% with venous thrombosis) and 103 disease-free white control subjects. We performed allele-specific PCR amplification of genomic DNA to determine FcyRIIA genotypes in these subjects. In patients with thrombosis, the frequency of the FcyRIIA-HI 31 allele was 0.6, and the frequency of the FcyRIIA-R131 allele was 0.4. In contrast, the frequencies of FcyRIIA-H131 and -R131 were both 0.5 in control subjects. The FcyRIIA-H131 gene product provides optimal IgG2 binding only in the homozygous state (H131iH131); non-H13 1 homozygotes (H13 1/R131 and R13 1/R13 1) have potentially decreased effector function when triggered by IgG2 (17). Therefore, we compared the frequency of H131/H131 with that of R131/H131 and R131/R131 in patients and controls. As shown in Table 3,40% of the 45 patients with thrombosis were homozy-

gous for FcyRIIA-H131. In contrast, 24% of the 103 controls were homozygous for FcyRIIA-H131. The odds ratio for thrombosis in Hl31iH131 as compared with H131iR131 and R131/R131 is 2.08 (95% confidence interval 0.992-4.4; P = 0.042 by Fisher’s exact test, 1-tailed). Contribution of IgG2 to aCL activity. Given the clinical observation that aCL is associated with thrombosis and our present finding that IgG2 is the most prevalent subclass of aCL, we performed the following adsorption studies. Using antibody-coupled agarose beads, we selectively removed IgGl or IgG2 from sera. In 5 patients with high levels of aCL that contained lgGl and IgG2 subclasses, removal of IgGl did not significantly alter the total aCL activity in our ELISA system (Figure 2). In contrast, IgG2 removal resulted in an 80% reduction in the total aCL activity (P < 0.001). BSAcoupled agarose beads failed to alter aCL activity. Since the selective removal of IgG2 abolished most of the total aCL activity, it appears that IgG3 and IgG4 play minor roles in total aCL activity. We further evaluated 1 IgG2-depleted serum for its capacity to stimulate cultured vascular endothelial cells, utilizing a monocyte adhesion assay as an indicator of endothelial activation (2). The MM6 human monocyte cell line is heterozygous and expresses FcyRIIAH131 and FcyRIIA-R131 (Salmon J: unpublished observations), and therefore recognized IgG1, IgG2, and IgG3. Monocyte binding to endothelial cells treated with aCL-positive sera was increased 4-fold compared with control serum (745 versus 170 adherent cells/mm2, respectively). Neither BSA-agarose bead treatment (602 adherent cells/mm2) nor IgGl removal (704 adherent cells/mm2) significantly altered the ability of the aCLpositive serum to enhance monocyte adhesion. In contrast, depletion of lgG2 from this serum markedly inhibited monocyte adhesion (406 adherent cells/mm’). Taken together, our rewlts demonstrate that IgG2 contributes to aCL activity and, as such, also has functional activity for endothelial cell activation. This is not likely to be a unique property of IgG2, but rather, provides evidence that aCL activity of sera is related to the capacity to generate a proadhesive phenotype of endothelial cells. DISCUSSION

While the relative distribution of IgG subclasses in normal human sera is IgCl (65%) > IgG2 (20%) > IgG3 (10%) > IgG4 (5%) (lS), most autoantibodies have been shown to comprise mainly IgG1 and IgG3,

2003


100,

I

T

m

.-5 .Ee

60

*

20

I

T

I 0 1

Anti-lgG2

Anti-lgG1

BSA

Adsorption Figure 2. Effect of removal of specific IgG subclasses on anticardiolipin antibody ( C L ) activity. Sera containing aCL (280 GPL [IgG phospholipid] units) were treated with anti-IgG1- or anti-IgG2-coupled agarose beads to selectively adsorb out either of the 2 subclasses prior to determination of aCL activity. Bovine serum albumin-coupled beads served as the control treatment. Valucs are the mean t SD of sera from 5 different patients. Removal of IgG2 decreascd aCL levcls by 80% (anti-IgG2 versus BSA, P < 0.001, by paired t-test).

although patterns of distribution may vary. Antinuclear antibody (ANA) subclass ratios are dependent on the ANA pattern (19). Anti-double-stranded DNA (antidsDNA) have been well established as containing primarily IgGl and IgG3 (3,20). Anti-Ro, La, and U1 RNP are primarily IgG1, and anti-Sm BB’ and anti-ribosomal P protein antibodies contain equal amounts of IgGl and IgG2 (21). In contrast, antineutrophil cytoplasmic antibodies in patients with Wegener’s granulomatosis are skewed toward IgG3, particularly in active disease (22), and anti-Clq antibodies are predominantly IgG2 in SLE and hypocomplementemic urticaria1 vasculitis (23,24). Restriction of the subclass response within an autoantibody population seems to depend on the antigen targeted (25) and the disease. Patients with SLE may have altered subclass responses to a common antigen, such as tetanus toxoid (26), suggesting that global immune dysfunction may override the normal mechanism of IgG restriction. Herein, we demonstrate the relationship of IgG subclass distribution to complications of aCL. The

data suggest that the thrombogenic potential of aCL in SLE patients is related to an abundance of a specific IgG subclass, possibly because of an enhanced capacity to interact with specific Fcy receptors. The distribution of IgG subclasses of aCL in a total of 60 aCL-positive patients was determined to be IgGl33%, lgG2 75%, IgG3 55%, and IgG4 60%. There was a predominance of IgG2 over IgG1. Because the sensitivities of the ELISA for IgGl and IgC2 were identical, our assay allows a direct comparison of the frequencies of these 2 subclasses. However, because of lower concentrations of IgG3 and IgG4, 5- and 10-fold higher concentrations of aCL sera were used in our ELISA for IgG3 and IgG4. This accounts for the observed high frequency of IgG3 and IgG4. It should be noted that the skewing toward IgG2 is not reflected in the total serum IgG. In SLE patients, the serum IgG distribution is normal (mean ? SD ratio IgGl:IgG2 3.8 5 2.4, n = 43; Schachter L, Salmon J: unpublished observations). Even in the face of hypergammaglobu-

2004

linemia, the ratio of IgG subclasses in SLE patients is reported to be similar to that of disease-free individuals (27). Selective adsorption studies using an anti-IgG2 antibody to remove IgG2 resulted in a loss of 77% aCL activity, whereas an anti-IgG1 antibody had minimal effect. These data, therefore, confirm an important contribution of IgG2 to the total aCL activity. Our observation that the IgG2 level is highly correlated with the aCL level underscores the relative importance of this subclass. The decrease in aCL-induced endothelial cell activation by selective removal of IgG2 in the patient serum studied raises the possibility that IgG2 also contributes to this effect of aCL. Our observation that IgG2 is an important component of aCL extends data previously reported, which described the presence of this subclass (3,4,28-30). Gharavi et a1 reported the presence of all 4 IgG subclasses in aCL; complications were noted in 5 of 6 patients with IgG2 and/or IgG4. Anti-dsDNA in the same group of patients were IgGl and IgG3 (3). Loizou et a1 found a general elevation of IgGl and IgG3 in SLE aCL sera, with 25% IgG2, but they identified a prominent elevation of IgG2 in the subset with the highest GPL levels (28), a group at increased risk of thrombosis. The presence of IgG2 in aCL may be related to antigenic targets of the response. Levy et a1 reported that infection-induced (nonthrombogenic) aCL were predominantly IgGl and IgG3, whereas IgG2 and IgG4 predominate in the aCL of autoimmune patients (4). Arvieux and colleagues noted an interesting association between sera reactivity with P,-glycoprotein I (P,GPI) and IgG subclass distribution (5). While aCL sera reactive with P,GPI predominantly contained IgG2, aCL sera nonreactive with P,GPI predominantly contained IgG3 subclass. Further support for the role of IgG2 in the pathogenesis of thrombosis comes from the observation that patients who are anti-&GPI positive, and therefore more likely to be IgG2 aCL positive, have an increased frequency of thrombosis (31) as compared with patients who are anti-P,GPI negative. Using an in vitro adhesion assay, we observed that removal of IgG2 from aCL sera greatly attenuated endothelial activation, These data confirm the potential important role of IgG2 as an initiator of a hypercoagulable state. The presence of IgG2, but not other IgG subclasses, was associated with clinical complications in a series of bivariate, but not multivariate, analyses. Regarding complications, IgG2 positivity was associated with arterial and/or venous thrombosis. There was an absence of thrombotic complications in patients without IgG2. The lack of correlation of IgG2 with fetal loss or

SAMMARITANO ET AL

thrombocytopenia may have been a result of confounding variables. Of significance is the evidence that the pathology of aCL is largely one that is composed of a noninflammatory vasculopathy rather than a complementmediated vasculitis, a response that may be related to the fact that IgG2 (in contrast to IgGl and IgG3) is a weak activator of complement. The maternal presence of aCL does not confer IgG2 aCL to the fetus because of the low placental transfer of IgG2. Anticardiolipin antibody detected in cord blood is primarily limited to the IgGl and IgG3 subclasses (Sammaritano L: unpublished observations). The lack of IgG2 in cord blood may be the basis for the rare incidence of intrauterine fetal thrombosis. The mechanism of thrombosis in APS may have an FcyR-dependent component as well as an Fcindependent component. F(ab’), fragments of aCL have been shown to trigger monocyte procoagulant activity (32), indicating an Fc-independent response. Similarly, aCL have been shown to activate cultured vascular endothelial cells in the presence of FcyR blockade (2). With multiple mechanisms likely, unique features of the Fc portion of IgG2 may contribute to the selective pathogenic potential of this IgG subclass. Allelic variants of FcyR may provide a perspective within which to define the interplay between the qualitative and quantitative humoral immune response and host genotype. The H131 allele of FcyRIIA is the only reported receptor which has been shown to efficiently recognize IgG2 (6-9,17). The binding to FcyRIIA-H131 can initiate monocyte, platelet, and endothelial cell activation in the context of an IgG2 stimulus. Using allele-specific PCR, we observed that FcyRIIA-H131 is increased in frequency among aCL patients with thrombosis (thrombosis versus control 40% versus 24%). The number of patients without clinical complications of aCL was too small to allow comparison of genotypes of aCL patients with thrombosis versus aCL patients without complications. If our model is correct, the prediction would be enrichment of FcyRIIA-H131 homozygotes only in the group with complications. The current data suggest that FcyRIIAR131 homozygosity protects against an FcyRIIAtriggered hypercoagulable state in APS, whereas it has been shown to increase the risk for lupus nephritis, presumably due to inefficient immune complex clearance (17). The importance of immunoglobulin receptors as potential initiators of thrombosis is supported by 1) emerging evidence that aCL binding to platelets and neutrophils leads to activation via cellular FcyR (33,34), 2) the precedent of an FcyR-dependent mechanism of


thrombosis in heparin-induccd thrombocytopenia (35), and 3) the case reports of successful treatment of APS with high-dose intravenous gamma globulin (36,37). The presence of IgG2 aCL in association with FcyRIIAH131 may therefore be a useful clinical predictor of increased thrombotic risk in patients with autoimmune IgG aCL. Analysis of aCL IgG subclass, host FcyR alleles, and antigenic specificity, in large populations using multivariate analysis, may allow us to identify patients with high risk for thrombosis and provide novel targets for therapeutic intervention. This first glimpse of the relationship between IgG autoantibody subclasses and FcyR alleles of effector cells establishes a novel and testable model for future studies examining pathogenic mechanisms of autoantibodies. ACKNOWLEDGMENTS The authors would like to thank Azzudin E. Gharavi, MD, for thoughtful discussions, Margaret Peterson, PhD, for assistance with statistical analysis, and Ying Wang, MD, for technical assistance.

REFERENCES 1. Sammaritano LR, Gharavi AE: Antiphospholipid antibody syndrome. Clin Lab Med 12:41-59, 1992 2. Simantov R, LaSala J, Lo SK, Gharavi AE, Sammaritano LR, Salmon JE, Silverstein R L Activation of cultured vascular endothelial cells by antiphospholipid antibodies. J Clin Invest 96:22112219, 1996 3. Gharavi AE, Harris EN, Lockshin MD, Hughes GRV, Elkon KB: IgG subclass and light chain distribution of anticardiolipin and anti-DNA antibodies in systemic lupus erythematosus. Ann Rheum Dis 47:286-290, 1988 4. Levy RA, Gharavi AE, Sammaritano LR, Habina L, Qamar T, Lockshin MD: Characteristics of IgG antiphospholipid antibodies in patients with systemic lupus erythematosus and syphilis. J Rheumat01 17:1036-1041, 1990 5. Arvieux J, Roussel B, Ponard D, Colomb MG: IgG2 subclass restriction of anti-0, glycoprotein I antibodies in autoimmune patients. Clin Exp Immunol 95310-315, 1994 6. Hulett MD, Hogarth PM: Molecular basis of Fc receptor function. Adv Immunol57:1-127, 1994 7. Kimberly RP, Salmon JE, Edberg JC: Receptors for immunoglobulin G: molecular diversity and implications for disease. Arthritis Rheum 38:306-314, 199.5 8. Salmon JE, Edberg JC, Brogle NL, Kimberly RP: Allelic polymorphism of human Fcy receptor IIa and Fcy receptor IIIb: independent mechanisms for differences in human phagocyte function. J Clin Invest 89:1274-1281, 1992 9. Parren PWHI, Warmerdam PAM, Boeije LCM, Arts J, Westerdaal NAC, Vlug A, Capel PJA, Aarden LA, van de Winkel JGJ: On the interaction of IgG subclasses with the low al'finity FcyRTlA (CD32) on human monocytes, neutrophils, and platelets: analysis of a functional polymorphism to human IgG2. J Clin In& 90:1537-1546. 1992 10. Horsewood P, Hayward CP, Warkentin TE, Kelton JG: Investiga-

2005

tion of the mechanisms of monoclonal antibody-induced platelet activation. Blood 78:1019-1026, 1991 11. Goger M, Sarmay G, Fiebigcr E, Wolff K, Petzelbauer P: Dermal microvascular endothelial cells express CD32 receptors in vivo and in vitro. J Immunol 156:1549-1556, 1996 12. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, Schaller JG, Tala1 N, Winchester RJ: The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 25:1271-1277, 1982 13. Osborne JM, Chacko GW, Brandt JT, Anderson CL: Ethnic variation in frequency of an allelic polymorphism of human FcyRIIA determined with allele specific oligonucleotide probes. J Immunol Methods 173:207-217, 1994 14. Abo T, Tilden AB, Balch CM, Kumagai K, Troup GM, Cooper MD: Ethnic difference in the lymphocyte proliferation rcsponse induced by a murine IgGl antibody, Leu-4, to the T3 molecule. J Exp Med 160:303-309, 1984 15. Gharavi AE, Harris EN, Asherson RA, Hughes GRV: Anticardiolipin antibodies: isotype distribution and phospholipid specificity. Ann Rheum Dis 46:1-6, 1987 16. Harris EN, Gharavi AE, Patel SP, Hughes GR: Evaluation of the antiphospholipid antibody test: report of an international workshop. Clin Exp Immunol68:215-222, 1986 17. Salmon JE, Millard SS. Schachter LA, Arnett FC, Ginzler EM, Gourley MF, Ramsey-Goldman R, Kimberly RP: FcyRIIA alleles are heritable risk factors for lupus nephritis in African Americans. J Clin Invest 97:1348-1354, 1996 18. Schur PH: IgG subclasses: a review. Ann Allergy 58:XY-100, 1987 19. Puritz EM, Yount WJ, Newell M, Utsinger PD: Immunoglobulin classes and IgG subclasses of human antinuclear antibodies: a correlation of complement-fixation and the nephritis of SLE. Clin Immunol Immunopathol2:98-113, 1973 20. Zouali M, Jefferis R, Eyquem A: IgG subclass distribution