Immunization of naive mice with β2-glycoprotein I (β2GPI) leads to the generation of pathogenic anticardiolipin antibodies associated with clinical manifestations of the antiphospholipid syndrome (APS). The aim of this study was to determine whether immunization of naive mice with human β2GPI, which shares homology with mouse β2GPI molecules, breaks tolerance to murine β2GPI and leads to the generation of anti-mouse β2GPI.
Twenty-four female BALB/c mice were immunized in the footpads with 10 μg of human β2GPI. Twelve age- and sex-matched BALB/c mice were immunized in the same manner with Freund's complete adjuvant and served as controls. The reactivity of whole sera, polyclonal IgG, and affinity-purified anti-β2GPI IgG antibodies against human, bovine, and mouse β2GPI was evaluated by enzyme-linked immunosorbent assay.
High titers of anti-human β2GPI IgG antibodies were detected 1 month after immunization. Progressively increasing titers against murine and bovine β2GPI were recorded 1–4 months after injection.
Immunization of mice with human β2GPI resulted in the generation of antibodies reacting with human, bovine, and murine β2GPI. The loss of tolerance to mouse β2GPI is attributable to the high interspecies homology of β2GPI. These results may point to molecular mimicry as a possible cause of APS.
Antiphospholipid syndrome (APS) is a clinical entity manifested by arterial and venous thromboses, recurrent miscarriages, and the presence of antiphospholipid antibodies (aPL) (1). These autoantibodies originally were described as being reactive with negatively charged phospholipids such as cardiolipin. However, it is now clear that most aPL recognize proteins that are able to bind negatively charged phospholipids, such as β2-glycoprotein I (β2GPI), particularly if they are bound to phospholipids or to activated plastic surface (1, 2).
Anti-β2GPI antibodies (anti-β2GPI) have a major role in the pathogenesis of APS (3, 4). In fact, it has been suggested that they might interfere with hemostatic reactions by reacting with β2GPI bound to coagulation factors and may affect cell function in different target tissues through the β2GPI present on cell membranes (4). We previously demonstrated that affinity-purified anti-β2GPI from APS patient sera bind β2GPI in the absence of any phospholipids, recognize soluble β2GPI, and appear to be mainly monoreactive autoantibodies (5). Human polyclonal anti-β2GPI are generally directed to epitopes preserved along the species evolution and cross-react with heterologous β2GPI from various species (5).
The mechanism by which the immune system reacts against self β2GPI is not clear. Molecular mimicry between β2GPI and exogenous antigens is a possible mechanism for anti-β2GPI autoimmunity (6–9). In fact, some studies have demonstrated a significant homology between β2GPI and microbial or viral proteins (6, 7, 9). Moreover, peptides that share such a structural similarity were shown to induce production of anti-β2GPI antibodies in experimental animals (6, 7). Most important, an experimental APS model was induced following passive transfer of anti-β2GPI antibodies that were generated by immunization with Haemophilus influenzae or Neisseria gonorrhoeae peptides (6). Conversely, immunization with human β2GPI whole molecule was associated not only with generation of antibodies binding β2GPI and β2GPI-cardiolipin complex in the anticardiolipin antibody (aCL) enzyme-linked immunosorbent assay (ELISA) (3, 10–14), but also with induction of experimental APS in naive animals (3, 12, 13). The aim of the present study was to examine whether immunization of mice with human β2GPI, which shares homology with mouse β2GPI molecules, breaks tolerance to murine β2GPI and leads to generation of anti-mouse β2GPI.
MATERIALS AND METHODS
Thirty-six female BALB/c mice were included in the study, 24 of which were immunized intradermally in the hind footpads with 10 μg of human β2GPI in Freund's complete adjuvant (CFA). Booster injections were given 3 weeks later. The control group included 12 age- and sex-matched BALB/c mice that were immunized with CFA only.
β2GPI purification and purification of IgG from mouse sera. Human β2GPI was purified from normal human sera according to a previously described method (15–17). Bovine and murine β2GPI were also purified as previously described (5). Pooled mouse sera, diluted 1:3 (volume/volume) in phosphate buffered saline (PBS), were subjected to chromatography on protein G-Sepharose (HiTrap Protein G; Pharmacia Biotech, Uppsala, Sweden). The elution buffer was 0.5N acetic acid.
β2GPI immobilization. Human β2GPI was immobilized on HiTrap N-hydroxysuccinimide (NHS)-activated Sepharose (Pharmacia Biotech), according to the manufacturer's advice. The protein-to-gel ratio was 5 mg of β2GPI to 1 ml of gel; the coupling efficiency was ∼90%.
Anti-β2GPI activity absorption. Sera from immunized mice, diluted 1:3 in PBS (v/v), were subjected to chromatography on human β2GPI-NHS column. The protein concentration of both the original and the eluted samples was determined and made equal. Sera from the control mice underwent a comparable process.
Anti-β2GPI IgG affinity purification. Mouse IgG preparations from protein G-Sepharose were subjected to chromatography on human β2GPI-NHS column. Anti-β2GPI IgG were recovered by acid elution with 0.1M glycine-HCl, pH 2.8, 0.5M NaCl. The protein content of the eluted fractions was evaluated by bicinchoninic acid protein assay (Pierce Europe, Amsterdam, The Netherlands).
Anti-β2GPI ELISA. The presence of anti-β2GPI in the sera of the mice was determined as previously described (18). Both polyclonal and affinity-purified anti-β2GPI IgG fractions were tested for anti-β2GPI activity according to the same method, with the following modifications: 1) coating of polystyrene-irradiated plates (Combiplate Enhanced Binding; ThermoLabsystems, Helsinki, Finland) was performed with 10 μg/ml of β2GPI; 2) the blocking agent was 0.6% gelatin (Carlo Erba, Milan, Italy) in 0.15M PBS, pH 7.2; and 3) the IgG dilutions were performed in PBS containing 0.3% gelatin, 0.05% Tween 20. Sera and purified IgG from the mice treated with CFA only were used as control.
Modified anticardiolipin ELISA. Anticardiolipin activity in purified IgG preparations was examined according to the method reported previously by our group (5). High-binding polystyrene plates (Greiner Labortechnik, Frickenhausen, Germany) were coated with cardiolipin, the wells were blocked by 0.6% gelatin in PBS, and the IgG samples were diluted in 0.3% gelatin-PBS. In some experiments, the assay was performed in the presence of purified human, bovine, or murine β2GPI (20 μg /ml) in blocking and diluting buffers, or in the presence of 10% fetal calf serum (FCS) in blocking and diluting buffers. Purified IgG preparations from mice treated with CFA only were used as control.
Induction of anti-β2GPI activity. Immunization of mice with human β2GPI was associated with the generation of anti-human β2GPI antibodies, which were already detectable at high titers 1 month after immunization. Increasing titers of antibodies to the bovine and murine molecules were detectable during the observation period (1–4 months), although always at levels lower than those of anti-human β2GPI antibodies (Figure 1). Four months after immunization, significant amounts of both anti-human and anti-murine β2GPI were detectable in mouse sera, as shown by the titration curves. In fact, the binding of immune sera to murine β2GPI was significantly higher than that of sera from control mice, up to a dilution of ≥1:2,000. No reactivity was shown in the sera of control mice (Figure 2).According to these findings, the IgG fractions from protein G-Sepharose showed detectable anti-β2GPI activity up to a concentration as low as 1 μg/ml (Figure 3).
Characterization of murine anti-β2GPI antibodies. To verify whether anti-human and anti-murine β2GPI antibodies were directed to shared epitopes, sera of immunized mice underwent absorption on immobilized human β2GPI (β2GPI-NHS column). Figure 4 shows that, after absorption, anti-human β2GPI activity significantly decreased, while binding to murine β2GPI was undetectable in comparison with that of control sera. Conversely, the affinity-purified IgG antibodies eluted from the human β2GPI-NHS column were reactive with both human and murine β2GPI, confirming that the antibody activity was directed to shared epitopes (Figure 5).
Binding to cardiolipin. The IgG fractions from sera of human β2GPI-immunized mice were tested under various conditions for their binding to cardiolipin. They did not bind to cardiolipin in the presence of gelatin only; however, strong binding to cardiolipin occurred in the presence of FCS and human, bovine, and murine β2GPI (Figure 6). Together, these findings strongly suggest that the binding of the purified IgG to cardiolipin was β2GPI-dependent.
Both in vitro and in vivo experimental models are highly suggestive of aPL having a direct pathogenic role in APS rather than simply being serologic markers for the disorder (4, 19). Several studies have demonstrated that the binding of pathogenic aPL to phospholipids is closely related to antibody reactivity against the β2GPI molecule (16, 20, 21). Not only have studies shown that anti-β2GPI is a significant clinical indicator of APS (18, 22, 23), but it has also been suggested that anti-β2GPI has a direct pathogenic role in APS by interfering with hemostatic reactions and with different cell types involved in both the coagulation cascade and placental implantation (4). Moreover, experimental APS has been induced in mice following passive transfer of anti-β2GPI and by immunizing mice with exogenous (human or bovine) β2GPI, resulting in the generation of pathogenic antibodies against β2GPI and cardiolipin (3, 9–13, 18). However, because these results were derived from experiments performed using the classic aCL ELISA (which implies use of FCS), it is difficult to state whether the pathogenic antibodies are those directed against phospholipids or against phospholipid-binding proteins.
We previously described the binding characteristics of anti-β2GPI (5). Human anti-β2GPI derived from APS patient sera were found to specifically recognize epitopes located on β2GPI in the absence of any phospholipids. Binding to β2GPI was inhibited by the addition of soluble β2GPI but not by the addition of cardiolipin in free solution. Furthermore, affinity-purified human anti-β2GPI reacted with β2GPI from different species, including bovine, murine, and guinea pig β2GPI.
In the present study, immunization of mice with human β2GPI resulted in the generation of antibodies reacting with human, bovine, and murine β2GPI. This autoimmune response is likely the result of the high interspecies homology of β2GPI molecules (24, 25). Consistent with this hypothesis, no anti-murine β2GPI reactivity was detectable after absorption of sera from the human β2GPI-immunized mice on human β2GPI-NHS columns. Affinity-purified mouse anti-human β2GPI also reacted with murine and bovine β2GPI. Taken together, these data suggest that after immunization with human β2GPI, the immune response induces secretion of antibodies directed to epitopes shared by human and murine β2GPI, suggesting a loss of immune tolerance to murine β2GPI. It is of interest that anti-murine β2GPI antibodies were not detectable until 2 months after immunization, and that titers did not become significantly high until 4 months after immunization, indicating involvement of mechanisms different from those responsible for the early immunization response against the heterologous molecule. In the current study, the time course of the appearance of anti-murine β2GPI was consistent with our previous data showing that experimental APS becomes evident 4 months after immunizing mice with human β2GPI (2, 3, 11, 12, 18).
Investigators have previously reported that immunization with β2GPI induces production of aCL (9–13). We demonstrated that IgG fractions from sera of immunized mice did not bind cardiolipin when gelatin only was used as blocking agent, but that they reacted strongly with cardiolipin when murine, bovine, and human β2GPI as well as FCS were used as blocking agents. Our results suggest that the previously reported aCL activity following immunization of naive mice with human β2GPI was actually attributable to antibodies directed to β2GPI epitopes rather than to cardiolipin itself. In fact, positive binding to cardiolipin-coated plates was shown by using the standard aCL ELISA, in which FCS contained enough bovine β2GPI to allow detection of anti-β2GPI antibodies (26).
In summary, the current data suggest that in experimental APS models, immunization of mice with human β2GPI leads to a state of autoimmunity manifested by generation of pathogenic anti-mouse β2GPI antibodies as the result of molecular mimicry between human and mouse β2GPI epitopes. The primary antigen for those autoantibodies appears to be the β2GPI molecule rather than cardiolipin. Consistent with this observation is the possibility that exposure to foreign antigens that share molecular homology with self β2GPI molecules may be associated with development of APS in humans and animals. Anti-β2GPI and aPL with binding properties similar to those found in APS patient sera were generated in mice following immunization with synthetic peptides and hexapeptides that shared structural similarity between β2GPI and cytomegalovirus (7) or between β2GPI and various bacterial antigens (6), respectively. Recently, we induced experimental APS by passive transfer of anti-β2GPI derived from mice immunized with the H influenzae or N gonorrhoeae TLRVYK hexapeptide (6).
In humans, aCL and other aPL develop following various infections (Table 1). It has been suggested that infection-associated aPL are neither pathogenic nor β2GPI-dependent (27). However, several reports indicated that thromboembolic episodes do occur in patients with infection, in association with high titers of aPL (28–31). In addition, cases of catastrophic APS were reported following various bacterial infections, including pneumonia, gram-negative septicemia, and Salmonella typhi (32, 33), suggesting a pathogenic role for infection in the induction of autoimmunity and APS.
Table 1. Associations between infection and the antiphospholipid syndrome reported in the literature*
LAC = lupus anticoagulant; aCL = anticardiolipin antibodies; HIV = human immunodeficiency virus; DIC = disseminated intravascular coagulation.
Besides the possibility of the presence of alternative thrombophilic mechanisms in these conditions (34–36), it is reasonable to hypothesize that the autoimmune response in APS might develop following exposure to a common infective agent, such as a foreign antigen that shares homology with β2GPI molecule. Alternatively, β2GPI bound to negatively charged phospholipids on the membranes of bacteria could be taken up by phagocytes during the infectious process and presented to β2GPI-reactive T cells in a manner that elicits a specific autoimmune response. However, because the majority of aPL that appear following infection are not pathogenic, it should be pointed out that the clinical features of APS may develop only in patients with the “proper” genetic, hormonal, and immunologic status.