Plasmin immunization preferentially induces potentially prothrombotic IgG anticardiolipin antibodies in MRLMpJ mice.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 60, No. 10, October 2009, pp 3108–3117 DOI 10.1002/art.24818 © 2009, American College of Rheumatology Plasmin Immunization Preferentially Induces Potentially Prothrombotic IgG Anticardiolipin Antibodies in MRL/MpJ Mice Kaleo Ede, Kwan-Ki Hwang, Chen-Ching Wu, Meifang Wu, Yao-Hsu Yang, Wei-Shiang Lin, Daniel Chien, Pei-Chih Chen, Betty P. Tsao, Deborah K. McCurdy, and Pojen P. Chen while 4 of 10 bound to ␤2GPI, 3 of 10 bound to thrombin, and 4 of 10 bound to the activated coagulation factor X (FXa). Functionally, 4 of the 10 IgG mAb inhibited plasmin activity, 1 of 10 hindered inactivation of thrombin by antithrombin III, and 2 of 10 inhibited inactivation of FXa by antithrombin III. Conclusion. Plasmin immunization leads to production of IgG antiplasmin, aCL, and anti-␤2GPI in MRL/MpJ mice, but leads to production of only IgG antiplasmin in BALB/cJ mice. IgG mAb generated from plasmin-immunized MRL/MpJ mice bind to various antigens and exhibit procoagulant activity in vitro. These results suggest that plasmin may drive potentially prothrombotic aCL in genetically susceptible individuals. Objective. To test the hypothesis, utilizing 2 experimental mouse models, that plasmin is an important autoantigen that drives the production of certain IgG anticardiolipin (aCL) antibodies in patients with the antiphospholipid syndrome. Methods. BALB/cJ and MRL/MpJ mice were immunized with Freund’s complete adjuvant in the presence or absence of human plasmin. The mouse sera were analyzed for production of IgG antiplasmin, IgG aCL, and IgG anti–␤2-glycoprotein I (anti-␤2GPI) antibodies. IgG monoclonal antibodies (mAb) were generated from the plasmin-immunized MRL/MpJ mice with high titers of aCL, and these 10 mAb were studied for their binding properties and functional activity in vitro. Results. Plasmin-immunized BALB/cJ mice produced high titers of IgG antiplasmin only, while plasmin-immunized MRL/MpJ mice produced high titers of IgG antiplasmin, IgG aCL, and IgG anti-␤2GPI. Both strains of mice immunized with the adjuvant alone did not develop IgG antiplasmin or IgG aCL. All 10 of the IgG mAb bound to human plasmin and cardiolipin, The antiphospholipid syndrome (APS) is characterized by clinical manifestations of vascular thrombosis and pregnancy loss associated with the presence of persistently and significantly increased titers of antiphospholipid antibodies (aPL) (1–6). The antigenic specificities of aPL have been the subject of a number of studies, and these studies have shown that aPL represent a heterogeneous group of immunologically and functionally distinct antibodies that recognize various phospholipids, phospholipid-binding plasma proteins, and phospholipid–protein complexes (1,3,7,8). These plasma proteins include ␤2-glycoprotein I (␤2GPI) and various factors involved in hemostasis, such as prothrombin, protein C, and protein S (7,8). Although aPL have been shown to promote thrombosis and miscarriage in animal studies, the etiology and pathogenic mechanisms remain unclear. To characterize pathogenic aPL in APS, we previously generated 7 monoclonal IgG anticardiolipin (aCL) antibodies from 2 patients with APS (9,10). Of Supported by NIH grant AR-42506. Dr. Ede is recipient of a fellowship grant from the Southern California Chapter of the Arthritis Foundation. Dr. Chen-Ching Wu’s work was supported by a Faculty Development award from Kaohsiung Medical University, Taiwan. Dr. Yang’s work was supported by a Faculty Development award from National Taiwan University Hospital, Taiwan. Dr. Lin’s work was supported by a Research Training grant from the Taiwanese government. Kaleo Ede, MD, Kwan-Ki Hwang, PhD, Chen-Ching Wu, MD, Meifang Wu, MD, Yao-Hsu Yang, MD, PhD, Wei-Shiang Lin, MD, Daniel Chien, BS, Pei-Chih Chen, MD, Betty P. Tsao, PhD, Deborah K. McCurdy, MD, Pojen P. Chen, PhD: University of California, Los Angeles. Address correspondence and reprint requests to Kaleo Ede, MD, University of California, Los Angeles, Division of Pediatric Allergy, Immunology and Rheumatology, 10833 Le Conte Avenue, MDCC 12-430, Los Angeles, CA 90095. E-mail: firstname.lastname@example.org. Submitted for publication June 20, 2008; accepted in revised form June 15, 2009. 3108 PLASMIN INDUCTION OF POTENTIALLY PROTHROMBOTIC IgG aCL IN MRL/MpJ MICE these monoclonal antibodies (mAb), 5 were prothrombotic in an in vivo pinch–induced thrombosis model in mice (11). Importantly, we found that 4 of these 5 aCL directly bind to the key enzymes involved in hemostasis, namely, thrombin, activated protein C, tissue-type plasminogen activator, and plasmin (12–15). These enzymes belong to the trypsin family and are homologous in their enzymatic domains (16–19). Interestingly, these enzyme-reactive aCL bind to plasmin with relative Kd values in the range of 10⫺7M (14), which are 30–100-fold higher than the affinities of known IgG aCL toward ␤2GPI, the major autoantigen in APS (20). These findings, in combination, suggest that plasmin may be an important autoantigen that drives the activities of certain IgG aCL in some patients with APS. Indeed, Chen et al, in a study in China, found that plasmin could induce IgG aCL in immunized BALB/cJ mice, and that one of the mAb generated from these mice, IgG1 aCL, displayed lupus anticoagulant activity and induced fetal loss when injected into pregnant mice (21). However, the titers and kinetics of the plasmin-induced IgG aCL were not given; the IgG aCL values were only expressed as the fold change (in SD) above the mean value for control mice. Furthermore, although 2 of the mAb inhibited plasmin activity, the effects of the mAb on other cross-reacting target proteases (such as thrombin) were not explored. To address these issues, we immunized BALB/cJ mice with human plasmin, which resulted in only transient and very low titers of IgG aCL. Therefore, in addition to BALB/cJ mice, we also immunized MRL/ MpJ mice with plasmin and analyzed the immune sera for IgG antiplasmin antibodies and IgG aCL. The MRL/ MpJ strain was chosen because mild immunologic defects (i.e., the presence of low-titer anti–double-stranded DNA autoantibodies and low levels of glomerulonephritis) have been observed in older mice (⬎1 year of age) in this strain, and MRL/MpJ mice are the parent and control strain for the well-studied spontaneous lupus model in MRL/lpr mice. The results showed that immunized MRL/MpJ mice, as compared with control BALB/cJ mice, produced high titers of both IgG antiplasmin antibodies and IgG aCL. Moreover, the immunized MRL/MpJ mice also produced high titers of IgG anti-␤2GPI antibodies. Furthermore, when mAb were generated from the sera of the MRL/MpJ mice with high titers of IgG antiplasmin and IgG aCL, these mAb were found to bind human plasmin, cardiolipin, ␤2GPI, thrombin, and activated coagulation factor X (FXa), to varying degrees. Importantly, some of the mAb inhibited plasmin activity and also hindered the inactivation of 3109 thrombin and FXa by antithrombin III (AT). Thus, these findings show that plasmin may serve as a driving autoantigen for certain prothrombotic IgG aCL that have functional significance in vitro. MATERIALS AND METHODS Immunization of mice. All mouse studies were performed following protocols in accordance with our institutional guidelines. Six-to-seven-week-old female BALB/cJ and MRL/ MpJ mice (The Jackson Laboratory, Bar Harbor, ME) were immunized subcutaneously with either 30 g of human Lysplasmin (Haematologic Technologies, Essex Junction, VT) in Freund’s complete adjuvant (CFA; Sigma, St. Louis, MO) or CFA alone. The mice received booster immunizations subcutaneously with the same amount of plasmin in Freund’s incomplete adjuvant (Sigma) at month 1 and month 3 after the first immunization. Serum samples were obtained from the mice before immunization on day 0 and monthly thereafter. Analysis of sera for IgG antibodies against human plasmin, cardiolipin, and ␤2GPI. IgG antiplasmin antibodies were analyzed as described previously (14). Briefly, highbinding plates were coated with 5 g/ml of human plasmin in phosphate buffered saline (PBS) (pH 7.4). After incubating the plates overnight at 4°C, the cultures were blocked with 0.25% gelatin/PBS. Serum samples at 1:100 dilution in 0.1% gelatin/ PBS were distributed to separate wells and incubated for 1.5 hours at room temperature (RT). Bound IgG antibodies were then measured in the sera, and those samples considered positive for IgG antiplasmin were analyzed further at a series of 1:10 dilutions, up to a 1:105 dilution, to determine the antibody titers. IgG aCL were analyzed in a similar manner, using a previously described method (22), except that the microtiter plates were coated with cardiolipin at 50 g/ml in ethanol and blocked with 10% bovine serum in PBS. Finally, IgG anti-␤2GPI antibodies were also analyzed in a similar manner, as described previously (9), except that high-binding plates were coated with ␤2GPI at 5 g/ml in PBS. In all of these experiments, a murine serum sample with high reactivity for each of the respective antigens (antiplasmin, aCL, or anti-␤2GPI) was used repeatedly in all of the enzyme-linked immunosorbent assay (ELISA) experiments as the reference standard to determine antibody titers or reference units (RU). Generation of monoclonal IgG antiplasmin/IgG aCL from the immunized mice. Mice with high titers of both IgG antiplasmin antibodies and IgG aCL received a booster immunization peritoneally with 30 g of plasmin 3–4 days before fusion. The spleen cells were fused with the P3X63-AG8.653 mouse myeloma cell line (American Type Culture Collection, Manassas, VA). The culture supernatants were then screened for IgG antiplasmin and IgG aCL, and the hybridomas from the double-positive wells were subcloned twice at 1 cell/well. Of note, all of the IgG aCL–positive wells were also positive for IgG antiplasmin. The mAb were affinity-purified using a HiTrap Protein G–HP column (GE Lifesciences, Piscataway, NJ) and used for all studies. All mAb were then isotyped using the Mouse-Typer Isotyping ELISA panel (Bio-Rad, Hercules, CA), and all were determined to be of the IgG1 subtype (results not shown). 3110 Analyses of mAb for reactivity with plasmin, cardiolipin, and ␤2GPI, and for cross-reactivity with thrombin and FXa. Purified mAb were analyzed for reactivity against human plasmin, cardiolipin, and ␤2GPI by ELISA in a manner similar to that described above, except that polyclonal mouse IgG and isotype mouse IgG1 (Chemicon, Temecula, CA) were used as negative controls, and all IgG were used at a concentration of 10 g/ml. The mAb were also tested for thrombin reactivity as described previously (12). Briefly, high-binding ELISA plates were coated with 5 g/ml of human ␣-thrombin (Haematologic Technologies) in Tris buffered saline (TBS; 0.05M Tris HCl and 0.15M NaCl, pH 7.5). After incubating the plates overnight at 4°C, the cultures were blocked with TBS containing 0.3% gelatin. The ELISA to test cross-reactivity with human FXa was conducted similarly, except that the plates were coated with 5 g/ml of human FXa (Haematologic Technologies). Effects of mAb on plasmin activity. The effects of the mAb on plasmin activity were studied using the chromogenic substrate S-2251 (Sigma) as described previously (21). Briefly, plasmin (20 mM) in 50 l of HEPES buffer was incubated with 50 l of mAb (200 g/ml) for 1 hour at RT in microtiter wells. S-2251 at a concentration of 100 l (250 g/ml final concentration) was then added, and the plate was incubated in a humidified environment for 1 hour. The optical density (OD) value at 405 nm was measured. Effects of mAb on the inactivation of thrombin by AT. The effects of the thrombin-reactive mAb on the inactivation of thrombin by AT were studied in a functional assay for the thrombin activity in the presence of AT and heparin, as described previously (12). Briefly, 25 l of thrombin was incubated with 25 l of either test mAb, normal polyclonal mouse IgG, or the isotype control monoclonal mouse IgG1, in duplicate for 1 hour at RT. Fifty microliters of AT (Enzyme Research Laboratories, South Bend, IN) was then added to each reaction mixture in buffer containing heparin, and finally, 200 l of the chromogenic substrate S-2238 (150 M; Chromogenix, Molndal, Sweden) was added, and after ⬃1 minute of incubation, the OD at 405 nm was measured. The percentage of thrombin inactivation by AT was calculated as (1 ⫺ [residual thrombin activity with AT]/[initial thrombin activity without AT]) ⫻ 100. Effects of mAb on the inactivation of FXa by AT. A functional assay for the effects of the FXa-reactive mAb on FXa inactivation by AT was done using a previously described method (23). The assay used was similar to the thrombin functional assay described above, except that 25 l of FXa was incubated separately with 25 l of test mAb or control IgG for 1 hour at RT. Fifty microliters of AT was then added in the buffer containing heparin, and subsequently, 50 l of the chromogenic substrate S-2765 (660 M; DiaPharma, West Chester, OH) was added, and the OD at 405 nm was measured over time. The percentage of FXa inactivation by AT was calculated as (1 ⫺ [residual FXa activity with AT]/[initial FXa activity without AT]) ⫻ 100. Statistical analysis. Differences in the titers of IgG antiplasmin antibodies and IgG aCL between the 2 immunization groups (the control group and plasmin-immunized group) were analyzed with the Kruskal-Wallis test followed by Dunn’s multiple comparison test. Of note, when a data point in a group was outside of the mean ⫾ 3 SD of the remaining data points of the same group, it was considered an outlier and was EDE ET AL excluded from all analyses. Differences in reactivity to ␤2GPI (OD values) in the sera between the 2 groups were analyzed with the unpaired t-test. Relationships between 2 of the variables (aCL titers versus anti-␤2GPI antibody titers) were assessed by Pearson’s correlation analyses using logtransformed values of aCL and anti-␤2GPI titers (both expressed in RU ⫻ 103). Differences in the plasmin activity and percentages of either thrombin or FXa inactivation by AT between the mAb-treated and untreated control groups were determined by one-way analysis of variance followed by Dunnett’s multiple comparison test. RESULTS Induction by plasmin immunization of mainly IgG antiplasmin antibodies in BALB/cJ mice, but both IgG antiplasmin and IgG aCL in MRL/MpJ mice. When BALB/cJ and MRL/MpJ mice were immunized with human plasmin, both strains produced high titers of IgG antiplasmin antibodies (Figures 1A and B). The IgG antiplasmin antibody titers peaked at month 2 and remained high up to month 5 in both strains of mice. Importantly, immunized MRL/MpJ mice also displayed significantly raised titers of IgG aCL. From month 2 to month 5, the titers of IgG aCL in the plasmin-immunized mice were significantly higher than those in the corresponding control mice (Figure 1D). Of note, MRL/MpJ mouse sera were also analyzed for the presence of aCL of both the IgA and IgM isotypes, but no significant differences in the levels of IgA or IgM aCL were found in comparison with control mice (results not shown). Immunized BALB/cJ mice also developed titers of IgG aCL that were significantly higher (P ⬍ 0.05) than those in the control mice at month 2, although the significance of this difference was not sustained throughout the remainder of the experiment (Figure 1C). However, the levels of IgG aCL in the BALB/cJ mice were much lower than in their MRL/MpJ counterparts (Figure 1C). The failure of induction of aCL in plasmin-immunized BALB/cJ mice might imply that plasmin can drive aCL production only in autoimmuneprone individuals. The immunization experiments were repeated separately on 2 occasions for the BALB/cJ mice and 3 times for the MRL/MpJ mice. Analysis of the sera was completed on all samples, and the results shown in Figure 1 are representative data from 1 experiment. The human Lys-plasmin used in these experiments was tested by Western blotting for contamination with ␤2GPI and was found to have ⬍0.25% ␤2GPI, which suggests that PLASMIN INDUCTION OF POTENTIALLY PROTHROMBOTIC IgG aCL IN MRL/MpJ MICE 3111 Figure 1. Effects of plasmin immunization on production of IgG antiplasmin antibodies (Ab) (A and B) and IgG anticardiolipin antibodies (aCL) (C and D) in BALB/cJ and MRL/MpJ mice. Mice were immunized (denoted by arrows) with either 30 g human plasmin in adjuvant (P) or adjuvant alone (control [C]) on day 0 and 1 month and 3 months later. Serum samples were obtained before immunization on day 0 and monthly thereafter and were analyzed at 1:102–1:105 dilutions for IgG antiplasmin and at a 1:102 dilution for IgG aCL. Results are expressed either in titers (for antiplasmin) or in relative units (RU) (for aCL). The experiments were repeated twice in BALB/cJ mice and 3 times in MRL/MpJ mice, and similar results were obtained in both strains of mice. Representative data from 1 experiment are shown. Horizontal bars indicate the geometric mean for each group. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01 versus controls treated with adjuvant alone. there was little to no contamination (results are available from the corresponding author upon request). The induced IgG aCL did not react with cardiolipin in the absence of bovine serum (results not shown), suggesting that these murine mAb have the characteristics of IgG aCL in APS. Taken together, these results showed that plasmin could drive IgG aCL production and suggested that genetic factors could play a significant role in the production of the plasmin-driven IgG aCL. Induction of IgG anti-␤2GPI antibodies in MRL/ MpJ mice, but not in BALB/cJ mice. ␤2GPI is recognized as the major autoantigen in APS, and the reactivity of antibodies against ␤2GPI and its complexes with cardiolipin may account for most of the positive findings on aCL activity in APS (24). Therefore, it is of interest to investigate whether the plasmin-induced aCL could also react with ␤2GPI in these murine models. To this end, the immune sera that displayed peak titers of IgG aCL (i.e., at the fifth month in MRL/MpJ mice) were analyzed for reactivity with ␤2GPI. As shown in Figure 2B, the immune MRL/MpJ mouse sera displayed significantly higher titers of IgG anti-␤2GPI antibodies as compared with the control MRL/MpJ mouse sera (P ⬍ 0.001). Moreover, the corresponding immune sera from BALB/cJ mice had no IgG anti-␤2GPI antibodies compared with the sera from control (nonimmunized) mice (Figure 2A). Thus, these results confirm that the plasmin-induced aCL in MRL/ MpJ mice mimic the characteristics of aCL in APS. Association of IgG aCL and IgG anti-␤2GPI antibodies with IgG antiplasmin antibodies. To further test our hypothesis that plasmin is a driving autoantigen for some IgG aCL, we studied the association of the production of IgG aCL with that of IgG antiplasmin antibodies in the immune sera of mice obtained at month 5 after the first immunization. As can be seen in Figure 2C, the levels of IgG aCL were highly correlated with the levels of antiplasmin antibodies (Pearson’s r ⫽ 0.64, P ⫽ 0.026). Moreover, we analyzed the relationship between 3112 EDE ET AL Figure 2. Serologic analyses of immune sera for IgG anti–␤2-glycoprotein I (anti-␤2GPI) antibodies (Ab) in BALB/cJ and MRL/MpJ mice (A and B) and analyses of association of anticardiolipin (aCL) and anti-␤2GPI antibodies with antiplasmin antibodies in immunized MRL/MpJ mice (C and D). Serum samples at month 5 after the first immunization were analyzed for IgG anti-␤2GPI antibodies at a 1:102 dilution (A and B). Murine IgG anti-␤2GPI antibodies at a 1:300 dilution were used as the reference serum, and results are expressed in relative units (RU). Horizontal bars indicate the geometric mean (n ⫽ 2 experiments). Associations of antiplasmin antibodies with aCL (C) and with anti-␤2GPI antibodies (D) were assessed by Pearson’s correlation coefficient analyses, using log-transformed values of antiplasmin antibodies (expressed as the optical density [OD] ⫻ 103), aCL (expressed as RU ⫻ 103), and anti-␤2GPI antibodies (expressed as RU ⫻ 103). For antiplasmin antibodies, mouse sera were analyzed at a 1:104 dilution in order to express the values in OD (as a continuous variable). ⴱⴱⴱ ⫽ P ⬍ 0.001. anti-␤2GPI antibody production and IgG antiplasmin antibody production. As is evident in Figure 2D, the levels of anti-␤2GPI antibodies were highly correlated with the levels of antiplasmin antibodies (Pearson’s r ⫽ 0.61, P ⫽ 0.048). Reactivity of some plasmin-induced murine monoclonal IgG aCL with human ␤2GPI. To further characterize the plasmin-induced IgG aCL, we initiated efforts to generate and analyze monoclonal IgG aCL from the immunized MRL/MpJ mice with high titers of IgG aCL. The hybridomas were screened for IgG antiplasmin antibodies and IgG aCL, and only the doublepositive hybridomas were utilized, resulting in the generation of 10 test mAb. These mAb clones were designated M2-11, M4-0, M9-6, M9-12, M9-14, M9-15, M9-16, M9-17, M9-43, and M20-48 (Figures 3A and B). As noted in the preceding experiments, the plasmin-induced aCL were observed to bind to ␤2GPI and were thus shown to mimic the major characteristic of aCL in APS. To explore this finding further, we analyzed the test mAb for reactivity with ␤2GPI. As shown in Figure 3C, 4 (40%) of the 10 mAb (i.e., M2-11, M4-0, M9-6, and M9-12) bound to ␤2GPI, providing further support that a subset of plasmin-induced aCL mimic the main characteristic of aCL in APS. Reactivity of some plasmin-induced murine monoclonal IgG aCL with the serine protease thrombin and FXa. To test our hypothesis that some of the plasmin-driven aCL cross-react with certain serine proteases that share homologous enzymatic domains with plasmin, we analyzed our mAb for cross-reactivity with thrombin and FXa. Figure 3D demonstrates that several of the mAb bound to thrombin, including M2-11, M4-0, M9-12, and M9-43. In addition, as shown in Figure 3E, some of the mAb also bound to FXa, namely, M2-11, M4-0, and M9-43. Taken together, these results indicate that a significant subset of the plasmin-induced aCL cross-react with other serine proteases. PLASMIN INDUCTION OF POTENTIALLY PROTHROMBOTIC IgG aCL IN MRL/MpJ MICE 3113 Figure 3. Binding of plasmin-induced murine monoclonal anticardiolipin antibodies (aCL) to different antigens, including human plasmin (A), cardiolipin (B), ␤2-glycoprotein I (␤2GPI) (C), thrombin (D), and activated coagulation factor X (FXa) (E). Splenocytes from plasmin-immunized MRL/MpJ mice with high titers of IgG aCL were used to generate murine hybridoma cells that were positive for both IgG antiplasmin antibodies and IgG aCL. The resulting 10 murine monoclonal antibodies (mAb) were analyzed at a concentration of 10 g/ml for binding to the various antigens. A polyclonal mouse IgG antibody (mIgG) was included as a control on each plate. In addition, buffer alone was used as the untreated control. Bars show the mean and SEM optical density (OD) (n ⫽ 2 experiments). Murine mAb inhibition of the amidolytic activity of plasmin. To assess the pathologic significance of the plasmin-driven aCL, we first studied the effects of the mAb on the amidolytic activity of plasmin, using the chromogenic substrate S-2251. The results showed that 4 mAb (M2-11, M9-16, M9-17, and M20-48) significantly inhibited the amidolytic activity of plasmin, with the extent of inhibition between 18% and 39% (P ⬍ 0.01 versus untreated controls) (Figures 4A and B). These results indicate that the plasmin-driven aCL may promote thrombosis by directly interfering with the fibrinolytic function of plasmin, similar to the previously described effects of patient-derived CL15 monoclonal IgG aCL (14). Murine mAb inhibition of the inactivation of thrombin by AT. To further characterize the thrombinreactive mAb, we investigated the ability of these mAb to interfere with AT inactivation of thrombin, using a functional assay containing 0.1 units/ml of heparin. The addition of heparin is needed to approximate the in vivo inactivation of thrombin by AT, since AT often binds to heparin-like glycosaminoglycans (including heparan sulfates on the endothelial cell surface) (25–27). Under these conditions, AT inactivated 98% of thrombin activity, and the degrees of thrombin inactivation by AT were not changed by the presence of either polyclonal mouse IgG or a monoclonal mouse IgG1 isotype control (Figure 5A). In contrast, the mAb M2-11 reduced the degree of thrombin inactivation to 29% (SEM 1.4 [n ⫽ 2 experiments]; P ⬍ 0.01 versus untreated controls), whereas the remaining 2 mAb (M4-0 and M9-43) did not significantly affect AT inactivation of thrombin. Because M2-11 could reduce the thrombin inactivation from 98% in buffer alone to 29% in cultures with the M2-11 mAb, the resultant increase in thrombin activity over time could present a significant procoagulant effect, since the residual thrombin may continue to convert fibrinogen into fibrin at a constant rate. To visualize this rapid cumulative effect over time, overall 3114 EDE ET AL interfere with the FXa inactivation by AT. The final concentrations of FXa, AT, and IgG were 1.25 nM, 12.5 nM, and 33.3 g/ml, respectively. Under these conditions, AT inactivated 92% of FXa activity (in buffer alone), and the degrees of FXa inactivation by AT were not significantly changed by the presence of either Figure 4. Effects of the 10 murine IgG aCL mAb on plasmin amidolytic activity, as determined using the chromogenic substrate S-2251. A, Plasmin amidolytic activity was assessed at the 60-minute time point using substrate S-2251 in the absence or presence of test mAb, polyclonal mouse IgG control, or murine IgG1 monoclonal isotype control (mIgG1) (all at 100 g/ml); buffer alone was used as the untreated control. Bars show the mean and SEM OD (n ⫽ 2 experiments). B, The mAb-mediated inhibition of plasmin activity was calculated using the data obtained in the activity assay. Bars show the mean and SEM percentage of inhibition relative to that with buffer alone (as the untreated control). ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01 versus untreated control. See Figure 3 for other definitions. conversion of a thrombin substrate in the absence or presence of test mAb, polyclonal mouse IgG, or an isotype control mAb was measured over a period of 5 minutes. As shown in Figure 5B, accumulated substrate conversion in the presence of M2-11 increased dramatically over that in the absence of M2-11 in the span of 5 minutes. Murine mAb inhibition of the inactivation of FXa by AT. In a similar manner, we used a functional assay to study the FXa–cross-reactive mAb for their ability to Figure 5. Inhibition of thrombin inactivation by antithrombin III (AT) with M2-11, one of the murine IgG aCL mAb. A, Thrombin was preincubated with the test mAb, polyclonal mouse IgG, or monoclonal mouse IgG1 isotype control (mIgG1), and AT was then added to the reaction mixtures (in the presence of heparin), which was followed immediately by the addition of the thrombin chromogenic substrate S-2238. The OD after 1 minute was measured. Bars show the mean and SEM percentage of thrombin inactivation by AT (n ⫽ 2 experiments). ⴱⴱ ⫽ P ⬍ 0.01 versus buffer alone (as untreated control). B, The effect of M2-11 on accumulated thrombin activity in the presence of AT and heparin was further assessed, in a manner similar to that described in A, except that generation of p-nitroaniline was monitored continuously for 5 minutes. Representative results (expressed as the OD) from 1 of 2 experiments are shown. See Figure 3 for other definitions. PLASMIN INDUCTION OF POTENTIALLY PROTHROMBOTIC IgG aCL IN MRL/MpJ MICE 3115 to a mean ⫾ SEM 3 ⫾ 0.7% and 38 ⫾ 4.2%, respectively (each n ⫽ 2 experiments; P ⬍ 0.01 versus untreated controls). As noted, these experiments were done using IgG at a concentration of 33.3 g/ml (222 nM) and FXa at a concentration of 1.25 nM, resulting in an IgG:FXa molar ratio of 178:1. Assuming that the plasma concentration of total IgG is 10 mg/ml and that IgG anti-FXa antibodies in a patient with APS account for 1% of the possible total IgG, the concentration of antibodies used in this assay was ⬃30% of the possible total IgG anti-FXa antibodies. Therefore, we studied the concentrationdependent effects of anti-FXa antibodies on AT inactivation of FXa. The mAb, polyclonal mouse IgG, and mouse monoclonal IgG1 isotype were analyzed in a graded series of 2-fold lower concentrations (from 3.13 g/ml up to 50 g/ml). The results showed that at a lower concentration of 12.5 g/ml (resulting in an IgG: FXa molar ratio of 67:1), M2-11 and M4-0 reduced the degree of FXa inactivation to 26% and 74%, respectively (Figure 6B). Of note, increasing the concentration of M2-11 almost completely inhibited the anticoagulant function of AT on FXa (Figure 6B). DISCUSSION Figure 6. Reduction of antithrombin III (AT) inactivation of activated FXa with M2-11 and M4-0, 2 of the IgG aCL mAb. A, FXa was preincubated separately with the test mAb (each at a concentration of 33.3 g/ml), polyclonal mouse IgG, or monoclonal mouse IgG1 isotype control (mIgG1), and AT was then added to each reaction mixture (in the presence of heparin), which was followed immediately by the addition of the FXa chromogenic substrate S-2765. Bars show the mean and SEM percentage of FXa inactivation by AT (n ⫽ 2 experiments). ⴱⴱ ⫽ P ⬍ 0.01 versus buffer alone (as the untreated control). B, The concentration dependence of the inhibition of AT inactivation of FXa by the test mAb was assessed. The experiment was performed in a manner similar to that described in A except that a series of graded concentrations of mAb (ranging from 0 to 50 g/ml) was used. Bars show the mean ⫾ SEM percentage of inactivation of FXa by AT in the presence of heparin (n ⫽ 2 experiments). See Figure 3 for other definitions. polyclonal mouse IgG control or monoclonal mouse IgG1 isotype control (Figure 6A). In contrast, the mAb M2-11 and M4-0 reduced the degree of FXa inactivation Previously, we showed that plasmin could induce both IgG antiplasmin antibodies and limited IgG aCL in BALB/cJ mice, and that one monoclonal aCL had lupus anticoagulant activity and induced fetal loss in mice (21). To characterize further the plasmin-driven IgG aCL, we used plasmin to immunize BALB/cJ mice and MRL/ MpJ mice. We found that plasmin immunization in MRL/MpJ mice induced IgG aCL and IgG antiplasmin antibodies, whereas only persistent IgG antiplasmin antibodies were induced in BALB/cJ mice (Figure 1). Moreover, the antibodies produced in the plasminimmunized MRL/MpJ mice reacted with human ␤2GPI (Figure 2B), and the IgG antiplasmin antibody levels were highly correlated with those of IgG aCL and IgG anti-␤2GPI (Figures 2C and D), demonstrating that the plasmin-driven aCL displayed a major characteristic of the diagnostic aCL in APS patients. Of note, to rule out the possibility that the observed anti-␤2GPI activity might be due to ␤2GPI contamination in plasmin, we used Western blotting analysis with 40 g plasmin; the results (available from the corresponding author upon request) indicated that there was no detectable ␤2GPI contamination. Taken together, these results suggest that aCL are tightly regulated in normal mice with a normal 3116 immune system but are more ready to escape immune regulation and persist in autoimmune-prone mice that probably have deficient/defective immune regulation. Moreover, these findings imply that plasmin could drive aCL production more readily in certain genetically susceptible individuals who are prone to autoimmune disorders. Subsequently, we generated 10 monoclonal IgG aCL from the plasmin-immunized MRL/MpJ mice. Importantly, 4 (40%) of these monoclonal IgG aCL reacted with ␤2GPI (Figure 3C), and some of the monoclonal IgG aCL cross-reacted with other human serine proteases, namely, thrombin and FXa (Figures 3D and E). Of note, the sera from the sixth blood withdrawal (i.e., 2 months after the third plasmin immunization) were analyzed for reactivity with thrombin and FXa, and these sera were found to contain antithrombin activity but not anti-FXa activity (P ⫽ 0.02 and P ⫽ 0.16, respectively, versus control mice immunized with adjuvant only) (results not shown). Finally, the pathologic significance of the plasmin-induced IgG aCL was assessed by studying the effects of the mAb on the function and regulation of the reactive target proteases. The results showed that 4 of the mAb (M2-11, M9-16, M9-17, and M20-48) could directly inhibit the amidolytic activity of plasmin, with the extent of inhibition ranging from 18% to 39% (Figure 4). Interestingly, the strength of binding of the mAb to plasmin (Figure 3A) did not directly correlate with the profiles of plasmin inhibition (Figure 4), which is likely a reflection of the fact that the mAb bind to different regions on plasmin, and that only the mAb that bind to or near the active site of plasmin may inhibit the activity of plasmin. In addition, the thrombin-reactive monoclonal IgG aCL M2-11 was also found to reduce the inactivation of thrombin by AT down to 29% (Figure 5), and 2 mAb, M2-11 and M4-0, reduced the inactivation of FXa by AT to 3% and 38%, respectively (Figure 6). These latter data suggest that ⬃20% of the plasmin-driven IgG aCL may interfere with feedback regulation of FXa, resulting in unchecked activation of FXa and a procoagulant state. Taken together, these data indicate that the plasmin-driven IgG aCL may promote thrombosis from 2 ends, via the unregulated conversion of fibrinogen to fibrin and the reduced rate of fibrin resolution. As reported in previous studies, 5 of 7 monoclonal IgG aCL derived from 2 patients with APS reacted with plasmin (9,10,14). Furthermore, these 5 plasminreactive aCL were observed to bind to plasmin with relative Kd values in the range of 10⫺7 (14), which are EDE ET AL 30–100-fold higher than the affinities of known IgG aCL toward ␤2GPI, the major autoantigen in APS (20). Taken together with the present findings, these data strongly suggest that plasmin is an important autoantigen that drives the production of certain prothrombotic IgG aCL in APS patients. In this context, it is interesting to note that many bacteria use human plasmin to dissolve the surrounding fibrin clots that function as a host defense mechanism to prevent spreading of bacteria via the circulation (28). For example, group A Streptococcus (GAS) uses its plasminogen receptors to recruit and activate human plasminogen to plasmin on the bacterial surface (28,29). Consequently, plasmin is presented to the host together with streptococci and thus may induce the immune response to plasmin, according to the “danger signal” hypothesis (30,31). In addition to GAS, several other common pathogens, including Staphylococcus aureus and Yersinia pestis, use the human plasminogen system for survival (28). Although the majority of aCL generated in the postinfection period are transient in nature (32), it is possible that repeated exposure to common pathogenic bacteria may allow the development of antiplasmin antibodies in genetically susceptible individuals, which may, in turn, lead to the development of aCL, anti-␤2GPI antibodies, and APS. Future studies are warranted to test the above-mentioned hypothesis. Alternatively, we recently reported that some aPL in APS patients recognize conformational epitopes shared by ␤2GPI and the homologous enzymatic domains of several serine proteases, including plasmin (33). Specifically, 2 patient-derived IgG anti-␤2GPI mAb bound to thrombin and plasmin, and 1 antithrombin mAb reacted with ␤2GPI. In addition, the binding of a cross-reactive mAb to ␤2GPI was inhibited by ␣-thrombin, which contains only the catalytic domain of thrombin. Taken together with the results of the present study, these findings suggest that plasmin may induce and/or drive the production of aCL and anti-␤2GPI antibodies in genetically susceptible individuals via epitope spreading. Thus, the present study shows that plasmin immunization induces significantly and persistently raised titers of disease-relevant IgG aCL in MRL/MpJ mice, but not in BALB/cJ mice, suggesting that plasmin-driven aCL production is under genetic control. Therefore, plasmin is more likely to serve as a driving autoantigen in certain genetically susceptible individuals. ACKNOWLEDGMENT We thank Dr. Brian Skaggs for teaching Dr. M. Wu and providing reagents for the Western blot analyses. PLASMIN INDUCTION OF POTENTIALLY PROTHROMBOTIC IgG aCL IN MRL/MpJ MICE AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ede had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. Ede, Hwang, Tsao, P. P. Chen. Acquisition of data. Ede, Hwang, C.-C. Wu, M. Wu, Yang, Lin, Chien, P.-C. Chen, McCurdy, P. P. Chen. Analysis and interpretation of data. Ede, Hwang, C.-C. Wu, M. Wu, Lin, P.-C. Chen, Tsao, P. P. Chen. REFERENCES 1. Harris EN, Pierangeli SS. Antiphospholipid antibodies and the antiphospholipid syndrome. Springer Semin Immunopathol 1994; 16:223–45. 2. Lockshin MD, Sammaritano LR, Schwartzman S. Validation of the Sapporo criteria for antiphospholipid syndrome. Arthritis Rheum 2000;43:440–3. 3. Levine J, Branch D, Rauch J. Medical progress: the antiphospholipid syndrome. N Engl J Med 2002;346:752–63. 4. De Groot PG, Derksen RH. Antiphospholipid antibodies: update on detection, pathophysiology, and treatment. Curr Opin Hematol 2004;11:165–9. 5. Mackworth-Young CG. Antiphospholipid syndrome: multiple mechanisms. Clin Exp Immunol 2004;136:393–401. 6. Miyakis S, Lockshin MD, Atsumi T, Branch DW, Brey RL, Cervera R, et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost 2006;4:295–306. 7. Fleck RA, Rapaport SI, Rao LV. Anti-prothrombin antibodies and the lupus anticoagulant. Blood 1988;72:512–9. 8. Oosting JD, Derksen RH, Bobbink IW, Hackeng TM, Bouma BN, de Groot PG. Antiphospholipid antibodies directed against a combination of phospholipids with prothrombin, protein C, or protein S: an explanation for their pathogenic mechanism? Blood 1993;81:2618–25. 9. Olee TW, Pierangeli SS, Handley HH, Le DT, Wei X, Lai CJ, et al. A monoclonal IgG anticardiolipin antibody from a patient with the antiphospholipid syndrome is thrombogenic in mice. Proc Natl Acad Sci U S A 1996;93:8606–11. 10. Zhu M, Olee T, Le DT, Roubey RA, Hahn BH, Woods VL Jr, et al. Characterization of IgG monoclonal anti-cardiolipin/ anti␤2GP1 antibodies from two patients with the antiphospholipid syndrome reveals three species of antibodies. Br J Haematol 1999;105:102–9. 11. Pierangeli SS, Liu XW, Espinola R, Olee T, Zhu M, Harris NE, et al. Functional analyses of patient-derived IgG monoclonal anticardiolipin antibodies using in vivo thrombosis and in vivo microcirculation models. Thromb Haemost 2000;84:388–95. 12. Hwang K, Grossman J, Visvanathan S, Chukwuocha R, Woods V, Le D, et al. Identification of anti-thrombin antibodies in the antiphospholipid syndrome that interfere with the inactivation of thrombin by antithrombin. J Immunol 2001;167:7192–8. 13. Hwang KK, Yang CD, Yan W, Grossman JM, Hahn BH, Chen PP. A thrombin–cross-reactive anticardiolipin antibody binds to and inhibits the anticoagulant function of activated protein C. Arthritis Rheum 2003;48:1622–30. 14. Yang CD, Hwang KK, Yan W, Gallagher K, FitzGerald J, 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 3117 Grossman JM, et al. Identification of anti-plasmin antibodies in the antiphospholipid syndrome that inhibit degradation of fibrin. J Immunol 2004;172:5765–73. Lu CS, Horizon AA, Hwang KK, FitzGerald J, Lin WS, Hahn BH, et al. Identification of polyclonal and monoclonal antibodies against tissue plasminogen activator in the antiphospholipid syndrome. Arthritis Rheum 2005;52:4018–27. St. Charles R, Matthews J, Zhang E, Tulinsky A. Bound structures of novel P3–P1⬘ beta-strand mimetic inhibitors of thrombin. J Medicinal Chem 1999;42:1376–83. Mather T, Oganessyan V, Hof P, Huber R, Foundling S, Esmon C, et al. The 2.8 Angstrom crystal structure of gla-domainless activated protein C. EMBO J 1996;15:6822–31. Wang X, Lin X, Loy JA, Tang J, Zhang XC. Crystal structure of the catalytic domain of human plasmin complexed with streptokinase. Science 1998;281:1662–5. Pennica D, Holmes WE, Kohr WJ, Harkins RN, Vehar GA, Ward CA, et al. Cloning and expression of human tissue-type plasminogen activator cDNA in E. coli. Nature 1983;301:214–21. Tincani A, Spatola L, Prati E, Allegri F, Ferremi P, Cattaneo R, et al. The anti-␤2-glycoprotein I activity in human anti-phospholipid syndrome sera is due to monoreactive low-affinity autoantibodies directed to epitopes located on native ␤2-glycoprotein I and preserved during species’ evolution. J Immunol 1996;157:5732–8. Chen XX, Gu YY, Li SJ, Qian J, Hwang KK, Chen PP, et al. Some plasmin-induced antibodies bind to cardiolipin, display lupus anticoagulant activity and induce fetal loss in mice. J Immunol 2007;178:5351–6. Harris EN, Pierangeli SS, Birch DJ. Anticardiolipin wet workshop report: Fifth International Symposium on Antiphospholipid Antibodies. Am J Clin Pathol 1994;101:616–24. Yang YH, Hwang KK, FitzGerald J, Grossman JM, Taylor M, Hahn BH, et al. Antibodies against the activated coagulation factor X (FXa) in the antiphospholipid syndrome that interfere with the FXa inactivation by antithrombin. J Immunol 2006;177: 8219–25. Roubey RA, Hoffman M. From antiphospholipid syndrome to antibody-mediated thrombosis. Lancet 1997;350:1491–3. Pearson JD. Endothelial cell function and thrombosis. Baillieres Best Prac Res Clin Haematol 1999;12:329–41. Dahlback B. Blood coagulation. Lancet 2000;355:1627–32. Esmon CT. Regulation of blood coagulation. Biochim Biophys Acta 2000;1477:349–60. Lahteenmaki K, Kuusela P, Korhonen TK. Bacterial plasminogen activators and receptors. FEMS Microbiol Rev 2001;25:531–52. Ringdahl U, Svensson M, Wistedt AC, Renne T, Kellner R, Muller-Esterl W, et al. Molecular co-operation between protein PAM and streptokinase for plasmin acquisition by Streptococcus pyogenes. J Biol Chem 1998;273:6424–30. Matzinger P. Tolerance, danger, and the extended family. Annu Rev Immunol 1994;12:991–1045. Gallucci S, Matzinger P. Danger signals: SOS to the immune system. Curr Opin Immunol 2001;13:114–9. Harel M, Aron-Maor A, Sherer Y, Blank M, Shoenfeld Y. The infectious etiology of the antiphospholipid syndrome: links between infection and autoimmunity. Immunobiology 2005:743–7. Lin WS, Chen PC, Yang CD, Cho E, Hahn BH, Grossman J, et al. Some antiphospholipid antibodies recognize conformational epitopes shared by ␤2-glycoprotein I and the homologous catalytic domains of several serine proteases. Arthritis Rheum 2007;56: 1638–47.