By Antithrombin Interfere with the Inactivation of Thrombin The

Home , Cardiolipin

Identification of Anti-Thrombin Antibodies in the Antiphospholipid Syndrome That Interfere with the Inactivation of Thrombin by Antithrombin This information is current as of September 28, 2021. Kwan-Ki Hwang, Jennifer M. Grossman, Sudha Visvanathan, Reginald U. Chukwuocha, Virgil L. Woods, Jr., Dzung T. Le, Bevra H. Hahn and Pojen P. Chen J Immunol 2001; 167:7192-7198; ;

doi: 10.4049/jimmunol.167.12.7192 Downloaded from http://www.jimmunol.org/content/167/12/7192

References This article cites 41 articles, 14 of which you can access for free at: http://www.jimmunol.org/content/167/12/7192.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 28, 2021 • Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2001 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Identiﬁcation of Anti-Thrombin Antibodies in the Antiphospholipid Syndrome That Interfere with the Inactivation of Thrombin by Antithrombin1

Kwan-Ki Hwang,2* Jennifer M. Grossman,* Sudha Visvanathan,* Reginald U. Chukwuocha,* Virgil L. Woods, Jr.,† Dzung T. Le,‡ Bevra H. Hahn,* and Pojen P. Chen*

The combined presence of anti-phospholipid (PL) Ab, including lupus anticoagulants (LAC) and/or anticardiolipin Ab (aCL), and thrombosis is recognized as the antiphospholipid syndrome (APS). LAC are detected as an inhibitory effect on PL-restricted in ␤ vitro blood coagulation tests, and are comprised mainly of Ab against 2 glycoprotein I and prothrombin (PT). Recently, anti-PT Ab (aPT) were found to be associated with thrombosis by some investigators, although this is not confirmed by others. Considering that aPT are heterogeneous in patients and that PT is converted into thrombin, we hypothesize that certain aPT in patients may Downloaded from bind to thrombin, and that some of such anti-thrombin Ab may interfere with thrombin-antithrombin (AT) interaction and thus reduce the AT inactivation of thrombin. To test this hypothesis, we searched for anti-thrombin Ab in APS patients and then studied those found for their effects on the AT inactivation of thrombin. The results revealed that most, but not all, aPT-positive patient plasma samples contained anti-thrombin Ab. To study the functional significance of these Ab, we identified six patient- derived mAb that bound to both PT and thrombin. Of these mAb, three could reduce the AT inactivation of thrombin, whereas others had minimal effect. These findings indicate that some aPT in patients react with thrombin, and that some of such anti- http://www.jimmunol.org/ thrombin Ab could inhibit feedback regulation of thrombin. Because the latter anti-thrombin Ab are likely to promote clotting, it will be important to develop specific assays for such Ab and study their roles in thrombosis in APS patients. The Journal of Immunology, 2001, 167: 7192Ð7198.

␤ oagulation abnormalities, including thrombosis and re- plexes. The involved plasma proteins include plasma protein 2 gly- ␤ current fetal loss, have emerged as important clinical coprotein-1 ( 2GPI), prothrombin (PT), protein C, and protein S (9– complications in systemic lupus erythematosus (SLE3; 14). To date, the Ab against ␤ GPI and its complexes with cardiolipin C 2 Refs. 1–3). Patients with SLE and acquired coagulation abnormal- probably account for most of the positive findings on tests for aCL in by guest on September 28, 2021 ␤ ities often have anti-phospholipid (PL) Ab (aPL). These include APS (15, 16), whereas Ab against PT and 2GPI are responsible for lupus anticoagulants (LAC), as detected by their abilities to pro- the majority of the LAC activity (11, 17–19). long certain in vitro PL-restricted blood clotting tests, and anticar- Recently, increasing attention has been paid to anti-PT Ab (aPT) diolipin Ab (aCL; Refs. 2 and 4–8). Because LAC is neutralized and their roles in thrombosis in APS patients (18, 20–29). The prev- by addition of excess PL, it was suggested that the LAC Ab might alence of aPT in patients varies among different studies, ranging from interact with PL and thus interfere with blood coagulation on the 30 to 60% in APS patients when tested by ELISA using immobilized limited PL surface in the in vitro test. Therefore, LAC and aCL are human PT on activated polyvinyl chloride plates (20, 23, 25). How- generally referred to as aPL, and the association of thrombosis and ever, aPT were found to be associated with thrombosis (21, 25, 28), fetal loss with LAC and aCL is recognized as antiphospholipid although this is not confirmed by other investigators (23). These con- syndrome (APS; Refs. 2 and 8). flicting data may reflect the heterogeneity of aPT present in individual Accumulated studies show that aPL represent a heterogeneous patient sera and different sets of these autoantibodies in clinically di- group of immunologically and functionally distinct Ab that recognize verse patient populations in different studies. various PL, PL-binding plasma proteins, and/or PL-protein com- To understand the functional and pathogenic property of aPT, Rao and coworkers (11, 18) affinity purified IgG aPT and found *Division of Rheumatology, Department of Medicine, University of California, Los that the purified Ab bound to immobilized phosphatidylserine in † ‡ Angeles, CA 90095; and Departments of Medicine and Pathology, University of 2ϩ California at San Diego, La Jolla, CA 92093 the presence of Ca and PT. These results suggested that IgG aPT cross-linked PT molecules, and thus increased the valence Received for publication April 20, 2001. Accepted for publication October 12, 2001. of interactions between PT and phosphatidylserine. Subse- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance quently, those investigators showed that IgG purified from a with 18 U.S.C. Section 1734 solely to indicate this fact. LAC-positive plasma sample (designated LAC IgG; from a 1 This work was supported by a research grant from the Southern California Chapter patient with hypoprothrombinemia) enhanced the binding of PT of the Arthritis Foundation and by Grants AI32243 and AR42506 from the National Institutes of Health. to HUVEC and increased conversion of PT to thrombin on the surface of HUVEC (22). 2 Address correspondence and reprint requests to Dr. Kwan-Ki Hwang, Medicine/ Rheumatology, 167022, University of California, Los Angeles, CA 90095-1670. E- Thrombin is a key effector enzyme in the coagulation cascade. mail address: [email protected] It converts fibrinogen to fibrin, leading to the formation of fibrin 3 Abbreviations used in this paper: SLE, systemic lupus erythematosus; aCL, anti- clots. It also feedback amplifies the cascade by activating factors V cardiolipin Ab; PL, phospholipid; aPL, anti-PL Ab; APS, antiphospholipid syndrome; ␤ ␤ and VIII, which in turn, enhance conversion of PT to thrombin PT, prothrombin; aPT, anti-PT Ab; AT, antithrombin; AU, abstract unit; 2GPI, 2 glycoprotein-1; EC, endothelial cell; LAC, lupus anticoagulant; PT, prothrombin. (30). Therefore, once thrombin is generated in vivo, it is tightly

regulated by antithrombin (AT) that binds to thrombin in the pres- min, generation of p-nitroaniline was monitored by measuring OD at 405 ence of heparin-like glycosaminoglycans on the endothelial cell nm. The activity of thrombin was determined as the rate of hydrolysis of (EC) surface and inactivates the enzyme irreversibly (30–32). Con- S-2238 from the linear range of absorbance at 405 nm with time. The effects of monoclonal anti-thrombin Ab on the AT inactivation of sidering that thrombin is derived from the zymogen PT, it is conceiv- thrombin were studied in a functional assay for the thrombin activity in the able that some aPT may bind to thrombin at a site where thrombin presence of AT and heparin, according to Bock et al. (35) with minor interacts with AT, and therefore inhibit AT inactivation of thrombin. modifications. In particular, human AT (Enzyme Research Laboratories) In this study, we report the detection of Ab against thrombin in APS was used at a concentration that was at least 10-fold higher than that of human ␣-thrombin, and experiments were conducted in 50 mM HEPES, patients and the inhibitory effects of three patient-derived IgG mono- 125 mM NaCl, 1 mM EDTA, and 0.1% polyethylene glycol 8000, pH 7.4, clonal anti-thrombin Ab on the AT inactivation of thrombin. These at 25°C in microtiter plates. The assay was initiated by incubating 25 ␮lof findings define a novel anti-thrombin autoantibody in APS and they thrombin (80 nM) separately with 25 ␮l of a test mAb (128 ␮g/ml), normal show that such Ab may interfere with negative feedback regulation of human IgG, or the isotype control monoclonal IgG3 in duplicate for1hat ␮ thrombin in circulation, and thus contribute to thrombosis. room temperature. Then, to each reaction mixture was added 50 lofAT (400 nM or the indicated concentrations) in the buffer containing heparin, resulting in a final heparin concentration of 0.1 USP unit/ml (U/ml) or the Materials and Methods indicated concentrations. Subsequently, 200 ␮l of the chromogenic sub- Patients and healthy controls strate S-2238 was added, and OD at 405 nm was measured at 1 min unless stated otherwise. The final concentrations of thrombin, AT, and IgG were Plasma samples were obtained from 13 patients and 5 normal controls at 6.7 nM, 67 nM, and 10.7 ␮g/ml, respectively. The percentage of thrombin University of California, Medical Center (Los Angeles, CA) and Univer- inactivation by AT was calculated as (1 Ϫ (the residual thrombin activity sity of California Medical Center (San Diego, CA). All 13 APS patients with AT)/(the initial thrombin activity without AT)) ϫ 100%. satisfy the Sapporo classification criteria for definite APS (8). The diag- For the comparative analysis of AT inactivation of human ␣-thrombin Downloaded from nosis of APS was confirmed by a medical record review. and ␥-thrombin (Hematologic Technologies), the experiments were done Of the 13 patients, there were 8 primary APS (62%) and 5 secondary similarly as the above, except that either one of the two above thrombin APS (38%). Three with secondary APS had SLE, one had SLE and variants was used in each functional assay of AT inactivation of thrombin. Sjogren’s syndrome, and the fifth had undifferentiated connective tissue disease. All were positive for LAC and 10 were positive for aCL. For Statistical analysis clinical manifestations, all had thrombosis (5 arterial and 10 venous), and two of the eight females had two or more unexplained fetal losses. The The mean AU plus 3 SD of the five normal controls was used as the cutoff, http://www.jimmunol.org/ ethnicity and gender for primary APS were six Caucasians and two His- and the plasma samples with AU values consistently higher than the cutoff panics, and six females and two males. All of the secondary APS were in two separate experiments were considered positive. Differences in the Caucasians with a gender ratio of two females/three males. The average test Ab-induced inhibition of thrombin activity or the thrombin inactivation age (in years) at diagnosis for primary APS was 43.3 (range of 17–63) and by AT were analyzed using paired ANOVA followed by the Bonferroni that for secondary APS was 26.8 (range of 12–41). multiple comparison test. Values of p Ͻ 0.05 were considered significant. Five healthy donors were recruited as normal controls and were designated N1 to N5. Their ethnicity was two Asians and three Caucasians, and Results gender was three females and two males. Their average age at the time of donation of their sample was 33.6 years (range of 21–53). Detection of anti-thrombin Ab in some APS patients To test the hypothesis that some aPT in patients may bind to thrombin,

ELISA for Ab against PT and thrombin by guest on September 28, 2021 we developed an ELISA for anti-thrombin Ab and used the assay to The ELISA for aPT was done as described previously (33). Briefly, high- analyze plasma samples from 13 APS patients (designated P1 to P13) binding ELISA plates (Costar, Cambridge, MA) were coated with 10 and 5 healthy normal controls (designated N1 to N5); multiple plasma ␮g/ml of human PT (Enzyme Research Laboratories, South Bend, IN) in TBS (0.05 M Tris-HCl and 0.15 M NaCl, pH 7.5). After incubating over- samples from 3 patients were available and were all analyzed. All night at 4°C, plates were blocked with TBS containing 0.3% gelatin. Then samples were analyzed at the 1/100 dilution. The results showed that test plasma samples (1/100 dilution) or mAb (1.0 ␮g/ml) in TBS/0.1% using mean plus 3 SD of the normal controls as the cutoff, anti-throm- gelatin were distributed to wells in duplicate and were incubated for 1.5 h bin Ab were detected in 10 of 13 patients, and IgG anti-thrombin Ab at room temperature. The IS6 monoclonal IgG aPT was used as a positive control and as a reference aPT in all assays (33). After washing with TBS, were detected in 3 of 13 patients (Fig. 1, A and B). Patient P1 had the bound human IgG and total Ig (denoted as Igs) were detected, respectively, highest titer of anti-thrombin Ab, but had no detectable IgG anti- with HRP-conjugated goat anti-human IgG (␥-chain specific; BioSource thrombin Ab. In contrast, patient P4 had the highest titer of IgG anti- International, Camarillo, CA), anti-human Igs (all isotypes; Jackson Im- thrombin Ab, but only a medium titer of anti-thrombin Ab. Of the munoResearch Laboratories, West Grove, PA), and peroxidase substrate three patients with multiple plasma samples, their anti-thrombin Ab tetramethylbenzidine (Kirkegaard & Perry Laboratories, Gaithersburg, MD). The ELISA for anti-thrombin Ab was done similarly except that titers fluctuated over time, which paralleled the previously reported plates were coated with 10 ␮g/ml of human ␣-thrombin (Hematologic fluctuation of aCL titers in APS patients. Specifically, only one of Technologies, Essex Junction, VT). Of note, the IS6 monoclonal aPT was three samples from patient P1 had anti-thrombin Ab, and only one of found to react with thrombin during the initial study of anti-thrombin Ab three samples from patient P6 had IgG anti-thrombin Ab. (see Results) and thus was used as a reference Ab in all subsequent anti- thrombin Ab ELISA. The results were expressed in abstract units (AU) To determine the relationship between aPT and anti-thrombin with 1 AU equivalent to the OD of the IS6 mAb at 2 ␮g/ml. Ab, we analyzed the same samples for aPT. As can be seen in Fig. A competitive inhibition assay was used to study the binding properties of 1, C and D, aPT were found in 10 of 13 patients and IgG aPT was selected mAb to PT and thrombin. Briefly, each mAb (1.5 ␮g/ml) was prein- found in 5 of 13 patients. Except for the P2c plasma sample, there cubated for 1.5 h with various concentrations of either PT or thrombin. Then, was a good quantitative correlation between aPT and IgG aPT the mixture was distributed to the PT- or thrombin-coated wells in duplicate. After incubation, bound IgG was measured. The inhibition data of each mAb titers; noticeably, patient P5 had abundant aPT and IgG aPT. Of

were used to calculate its relative Kd toward PT and thrombin (34). the three patients with multiple plasma samples, their aPT and IgG aPT titers ﬂuctuated over time, similar to the above ﬂuctuation in Functional assay for thrombin activity and the AT inactivation anti-thrombin Ab titers. of thrombin When the presence of aPT and anti-thrombin Ab in patient sam- The effects of thrombin-reactive mAb on thrombin activity were studied by ples were compared, 11 of 13 aPT-positive patient plasma samples mixing 25 ␮l of human ␣-thrombin (80 nM) separately with 25 ␮l of a test had anti-thrombin Ab, and three of the seven IgG aPT-positive mAb (128 ␮g/ml), normal human IgG, or an isotype control monoclonal IgG3 for1hatroom temperature. Then, to each reaction mixture was samples had IgG anti-thrombin Ab. These data suggest that many added 200 ␮l of the thrombin chromogenic substrate S-2238 (H-D-Phe- aPT (including the IgG isotype) may also react with thrombin, or Pip-Arg-p-nitroanilide, 150 ␮M; Chromogenix, Molndal, Sweden). After 1 that two different Ab for each Ag often occur together, similar to 7194 ANTI-THROMBIN Ab IN THE APS Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 1. Presence of aPT and anti-thrombin Ab in some APS patients. A, Total anti-thrombin Ab; B, IgG anti-thrombin Ab; C, Total aPT; and D, IgG aPT. Plasma samples from 13 APS patients (designated P1 to P13; the designations followed by alphabets denote multiple samples from the respective patient) and 5 normal controls (designated N1 to N5) were analyzed at the 1/100 dilution. The IS6 monoclonal aPT was found to react with thrombin (see latter) and thus was used as a reference Ab in all assays. The OD of all samples were standardized against IS6 (2 ␮g/ml) and was expressed in AU. The mean AU and the range are given (n ϭ 2); the dashed lines represent the cutoff, which is mean AU plus 3 SD of the five normal controls; and an asterisk denotes positive samples. autoantibodies against Sjogren syndrome Ag A and B in SLE (1). In CL15, and IS3 were even better than that of the IS6 monoclonal IgG contrast, sample P1b had high titers of anti-thrombin Ab, but no aPT, aPT (33) at the same concentration. indicating that some anti-thrombin Ab in APS patients do not react Subsequently, all monoclonal aPT were analyzed for binding to with PT and that anti-thrombin Ab are heterogeneous in patients. thrombin. As can be seen in Fig. 2B, all six aPT (CL1, CL15, CL24, IS3, IS4, and IS6) bind to thrombin. Interestingly, their relative bind- Identification of monoclonal aPT and anti-thrombin Ab ing pattern to thrombin is similar to their binding to PT. Because anti-thrombin Ab are heterogeneous in patients, it would be important to obtain monoclonal anti-thrombin Ab and use such mAb to study the functional significance of anti-thrombin Ab in APS. Ac- Effects of anti-thrombin mAb on thrombin activity and thrombin cordingly, we searched for patient-derived monoclonal aPT and anti- inactivation by AT thrombin Ab. Previously, we found that the IS6 monoclonal aPT To study the effects of anti-thrombin mAb on thrombin activity, cross-reacts with cardiolipin (33), and that the P11 monoclonal aPT thrombin was incubated separately with test mAb for 1 h, and then the Ј F(ab )2 (isolated by panning a phage display Ab library on PT) reacts thrombin chromogenic substrate activity was assessed. As can be seen ␤ ϫ Ϫ6 with both PT and 2GPI, with relative Kd values of 3.2 10 M for in Fig. 3, CL15 and IS3 slightly reduced thrombin activity, whereas ϫ Ϫ6 ␤ PT vs 1.6 10 M for 2GPI (36). These data raised the possibility the other four anti-thrombin mAb did not affect thrombin activity. ␤ that some of our monoclonal aCL/anti- 2GPI Ab might reciprocally Thereafter, we studied thrombin-reactive mAb for their abilities cross-react with PT. Therefore, we screened seven patient-derived to interfere with AT inactivation of thrombin in a functional assay, monoclonal IgG aCL (37) for binding to PT; IS1 and IS2 are IgG1, which contained 0.1 U/ml of heparin and used AT at a concentra- and the other mAb are IgG3. The results showed that five of seven tion that was at least 10-fold higher than that of thrombin. The mAb reacted with PT (Fig. 2A). Moreover, the aPT activities of CL1, addition of heparin is to approximate the in vivo inactivation of The Journal of Immunology 7195

FIGURE 3. CL15 and IS3 monoclonal anti-thrombin Ab inhibit (weak- ly) thrombin activity. Thrombin was preincubated with test mAb for 1 h, and then the thrombin chromogenic substrate S-2238 was added. After 1 min, generation of p-nitroaniline was measured in OD at 405 nm. The Downloaded from .p Ͻ 0.05 ,ء .(mean and the range are given (n ϭ 2

have a significant prothrombotic effect, as the residual thrombin continues to convert fibrinogen into fibrin at a constant rate. To

visualize this rapid cumulative effect over time, overall conversion http://www.jimmunol.org/ of a thrombin substrate in the absence or presence of test mAb or an isotype control mAb was measured over a period of 5 min. As can be seen in Fig. 4B, accumulated substrate conversion in the presence of CL24 increases dramatically over those in the absence of CL24 in 5 min. Because the plasma concentration of AT is 2 ␮M (38) and the above experiments were done with the final concentration of AT at 67 nM, we studied the effects of CL24 on thrombin inactivation in the presence of AT from 67 to 533 nM, due to the prohibitory cost of AT. by guest on September 28, 2021 The results showed that the CL24-mediated reduction in AT inactivation of thrombin remained constant over the tested range of AT concentrations (Fig. 4C). Subsequently, we studied the effects of FIGURE 2. Identification of monoclonal aPT and anti-thrombin Ab. A, CL24 and CL15 over a range of heparin concentrations that had been Some patient-derived monoclonal aCL bind to PT. B, All monoclonal aPT used by other investigators (39). The results show that CL24 and bind to thrombin. Microtiter wells were coated with PT or thrombin, and CL15 significantly reduced AT inactivation of thrombin in the pres- the test mAb, normal human IgG, or the monoclonal isotype controls (IgG1 ence of heparin concentrations from 0.025 to 0.2 U/ml (final concen- ␮ or IgG3) were analyzed at 1 g/ml. IS1 and IS2 are IgG1, and the other trations in the thrombin and AT mixture), and the Ab-mediated re- mAb are IgG3. Bound IgG were measured and expressed in OD, and the duction disappeared when heparin concentration reached 0.4 U/ml mean and SEM are given (n ϭ 3). (Fig. 4D). A maximal reduction in the thrombin inactivation by AT occurred at 0.05 U/ml of heparin, reducing the inactivation of throm- thrombin by AT, which often binds to heparin-like glycosamino- bin from 76% in the presence of the control IgG3 to 30% in the glycans (including heparan sulfates) on the EC surface (31). The presence of CL24 (Fig. 4D). In contrast, IS6 did not significantly final concentrations of thrombin, AT, and IgG were 6.7 nM, 67 affect the AT inactivation of thrombin. nM, and 10.7 ␮g/ml, respectively. The thrombin concentration was dictated by the need to determine the initial rate of thrombin ac- The binding properties of three chosen mAb to PT and thrombin tivity from the linear range of hydrolysis of the substrate (S-2238) Based on the above data, three representative mAb were chosen with time. The results were expressed as the percentage of throm- for analysis by competitive and cross inhibition. These were bin inactivation by AT. CL24 (which inhibits AT inactivation of thrombin), IS3 (which Under these conditions, AT inactivated 88% of thrombin activity inhibits thrombin per se), and CL15 (which inhibits both throm- (data not shown), and the degrees of thrombin inactivation by AT bin per se and the AT inactivation of thrombin). The results were not changed by the presence of either polyclonal human IgG or showed that soluble thrombin is more effective than PT in a monoclonal human IgG3 isotype control (Fig. 4A). In contrast, inhibiting all three mAb binding to either PT or thrombin (Fig. CL24 and CL15 reduced the degrees of thrombin inactivation to 61% 5). Importantly, PT could only inhibit binding of all tested mAb and 71%, respectively. CL1 showed a small but significant inhibitory to the immobilized PT but not thrombin. These results demon- effect on AT inactivation of thrombin, reducing thrombin inactivation strate that these three mAb are more specific for thrombin than to 78%. The remaining three other monoclonal anti-thrombin Ab did PT. Based on the thrombin inhibition data for binding to

not affect AT inactivation of thrombin. thrombin, the relative Kd values of these Ab to thrombin were Although CL24 reduced thrombin inactivation only from 88% calculated to be 7.5 ϫ 10Ϫ6, 1.7 ϫ 10Ϫ6, and 7.4 ϫ 10Ϫ6 M for to 61%, the resultant increase in thrombin activity over time could CL15, CL24, and IS3, respectively. 7196 ANTI-THROMBIN Ab IN THE APS

FIGURE 4. Some anti-thrombin Ab inhibit thrombin inactivation by AT. A, CL1, CL15, and CL24 signiﬁcantly reduce the thrombin inactivation by AT. Thrombin was preincubated with test mAb. Then, AT was added to reaction mixtures (in the presence of heparin), which was followed immediately by the addition of the thrombin chromogenic substrate S-2238. OD after 1 min was measured. The results are expressed as the percentage of thrombin inactivation by AT. An asterisk denotes p Ͻ 0.05. B, The effect of CL24 on accumulated thrombin activity in the presence of AT and heparin. The experiment was performed in a similar manner to that in A except that generation of p-nitroaniline was monitored Downloaded from continuously for 5 min. A representative result from three experiments is given. C, CL24 reduces the thrombin inactivation by AT over a range of AT concentrations. The experiment was performed in a similar manner to that

in A except that AT was used at the http://www.jimmunol.org/ indicated ﬁnal concentrations. D, CL15 and CL24 reduce thrombin inactivation by AT over a range of heparin concentrations. The experiment was performed in a similar manner to that in A except that heparin was used at the indicated concentrations (0–0.4 U/ml). The mean and the range are given for A, C, and D (n ϭ 2). by guest on September 28, 2021

Preliminary characterization of thrombin epitope recognized by Importantly, of the six monoclonal anti-thrombin mAb, CL15 and IS3 CL24 inhibited thrombin activity per se (Fig. 3), whereas CL1, CL15, and As a first step to define the epitope recognized by CL24, we com- CL24 significantly reduced the AT inactivation of thrombin in the ␮ paratively analyzed the effects of CL24 on ␣-thrombin and presence of heparin (Fig. 4A), and CL24 at 10.7 g/ml could inhibit ␥-thrombin, a proteolytic variant of ␣-thrombin with impaired ex- AT inactivation of thrombin in the presence of AT at 533 nM (Fig. osite I. The exosites are patches of positively charged amino acid 4C). Of note, the human plasma concentrations of AT and IgG are 2 ␮ residues on the thrombin surface that interact with the thrombin M and 10 mg/ml, respectively (38). Moreover, although the reduc- substrate, inhibitor, and modifier (i.e., thrombomodulin; Ref. 40). tion of AT inactivation was only moderate, the effect of such an Ab As can be seen in Fig. 6, AT alone inactivated 67–69% of on cumulative thrombin cleavage products increases rapidly over time ␥-thrombin activity in the presence of either polyclonal human IgG (Fig. 4B). In addition, the binding of three functionally representative or a monoclonal human IgG3 isotype control. Under the same mAb (i.e., CL15, CL24, and IS3) to PT and thrombin was inhibited conditions, CL24 reduced the degree of ␥-thrombin inactivation to more effectively by soluble thrombin than by PT (Fig. 5), indicating 40%, equivalent to a 42% reduction (based on the 69% for the that these three mAb are more specific for thrombin than PT. Com- monoclonal human IgG3 isotype control) in AT inactivation of bined, these data show that some anti-thrombin Ab in APS patients thrombin. Importantly, this CL24-induced reduction on AT inac- (like CL24 at ϳ0.1% of the plasma IgG concentration) could interfere tivation of ␥-thrombin is similar to the 42% reduction of AT in- with the AT inactivation of thrombin, thus allowing for prolonged activation of the ␣-thrombin by CL24 (decreasing from 88% AT coagulation in blood once clotting is initiated in patients carrying such inactivation of ␣-thrombin to 51%). The results suggest that CL24 autoantibodies. Obviously, prolonged and unchecked thrombin activ- is unlikely to react with thrombin at its exosite I and/or the sur- ity is most likely to promote and/or sustain thrombosis. rounding regions. Fig. 4D shows that CL24-mediated inhibition of AT inactivation of thrombin was decreased by increasing concentrations of hepa- Discussion rin. As noted earlier, the addition of heparin is to approximate the To test our hypothesis that some thrombogenic aPT may bind to in vivo inactivation of thrombin by AT, which often binds to thrombin and thus interfere with thrombin inactivation by AT, we anticoagulantly active heparan sulfate proteoglycans on the vas- searched for the presence of anti-thrombin Ab and studied their effects cular endothelium (31, 41). During the in vitro heparin-dependent on the AT inactivation of thrombin. The results showed that anti- AT inactivation of thrombin, heparin binds simultaneously to thrombin Ab were detected in several APS patients (Fig. 1, A and B), thrombin and AT, and thus bridges thrombin and AT to form a and several patient-derived IgG mAb bound to thrombin (Fig. 2B). trimolecular complex of heparin-thrombin-AT, leading to a several The Journal of Immunology 7197 Downloaded from http://www.jimmunol.org/

FIGURE 5. Thrombin is more effective than PT in inhibiting the chosen mAb from binding to PT (A, top) and to thrombin (B, bottom). The results are by guest on September 28, 2021 expressed in the percentage of inhibition, and a representative result from four experiments is shown. thousand-fold increase in the rate of AT inactivation of throm- consistent with the finding that CL24 also inhibited the AT bin (42). In this context, the above data suggest that CL24 may inactivation of ␥-thrombin, which contains the intact exosite II bind to thrombin’s heparin binding site (the exosite II) and thus but impaired exosite I (Fig. 6). interfere with the binding of heparin to thrombin, and this effect Similar to the in vitro heparin-mediated enhancement of AT inac- is overcome by increased concentrations of heparin, which tivation of thrombin, it is thought that in vivo acceleration of AT could compete for the same binding sites. This contention is inactivation of thrombin occurs through the binding of AT and thrombin to the anticoagulantly active endothelial glycosaminoglycans (41). Although the effective activity of anticoagulantly active heparan sulfate proteoglycans on the EC surface is unknown, it can be speculated that in some in vivo circumstances, anti-thrombin Ab similar to CL24 may interfere with the binding of thrombin to these heparan sulfates and may therefore inhibit the formation of thrombin-heparan sulfate-AT complexes and the accelerated AT inactivation of thrombin. Of the above three likely prothrombotic anti-thrombin Ab, CL15 also inhibits thrombin activity (Fig. 3). In contrast, IS3 did not reduce AT inactivation of thrombin, but did inhibit thrombin activity per se (Fig. 3). The observed inhibitory activity of these two anti-thrombin mAb was similar to a previously reported monoclonal IgG anti-thrombin Ab that apparently caused severe bleeding in a patient with monoclonal gammopathy (43). This latter mAb did not bind to PT, and thus was similar to the anti-thrombin Ab observed in plasma sample P1b (Fig. 1). In light of patient-derived anti-thrombin mAb with different (and FIGURE 6. The AT inactivation of ␣-thrombin and ␥-thrombin (a pro- even opposite) functional activities, it would be fruitless to assess the teolytic variant of ␣-thrombin with the impaired exosite I) is inhibited to clinical significance of all anti-thrombin Ab in APS by association similar degrees by CL24. Each thrombin was preincubated with test mAb studies of the presence of anti-thrombin Ab to APS, or to study the and the other procedures are the same as in Fig. 4A. functional activities of affinity-purified polyclonal anti-thrombin Ab 7198 ANTI-THROMBIN Ab IN THE APS from patients. Instead, it will be first necessary to delineate the throm- 15. Roubey, R. A. S. 1996. Immunology of the antiphospholipid antibody syndrome. bin epitopes recognized by various anti-thrombin Ab with different Arthritis Rheum. 39:1444. 16. Roubey, R. A. S., and M. Hoffman. 1997. From antiphospholipid syndrome to functional activities such as CL24 (which inhibits AT inactivation of antibody-mediated thrombosis. Lancet 350:1491. thrombin), IS3 (which inhibits thrombin activity), and IS4 (which 17. Roubey, R. A. S., C. W. Pratt, J. P. Buyon, and J. B. Winfield. 1992. Lupus anticoagulant activity of autoimmune antiphospholipid antibodies is dependent binds to thrombin but neither inhibits thrombin activity nor interferes ␤ upon 2-glycoprotein I. J. Clin. Invest. 90:1100. with AT inactivation of thrombin) (Figs. 3 and 4). If anti-thrombin Ab 18. Permpikul, P., L. V. M. Rao, and S. I. Rapaport. 1994. Functional and binding with different functional activities are found to recognize different studies of the roles of prothrombin and ␤2-glycoprotein I in the expression of lupus anticoagulant activity. Blood 83:2878. thrombin epitopes, then specific assays for each type of anti-thrombin 19. Galli, M., G. Finazzi, E. M. Bevers, and T. Barbui. 1995. Kaolin clotting time and Ab may be developed and used to study the roles of CL24-like anti- dilute Russell’s viper venom time distinguish between prothrombin-dependent ␤ thrombin Ab in thrombosis in APS patients, as well as the roles of and 2-glycoprotein I-dependent antiphospholipid antibodies. Blood 86:617. 20. Arvieux, J., L. Darnige, C. Caron, G. Reber, J. C. Bensa, and M. G. Colomb. IS3-like Ab in Ab-mediated bleeding disorders. 1995. Development of an ELISA for autoantibodies to prothrombin showing their It is intriguing that five of seven mAb generated by screening prevalence in patients with lupus anticoagulants. Thromb. Haemost. 74:1120. 21. Horbach, D. A., E. Van Oort, R. C. Donders, R. H. Derksen, and P. G. de Groot. against cardiolipin in the presence of bovine serum bind to PT and 1996. Lupus anticoagulant is the strongest risk factor for both venous and arterial thrombin. Of these five mAb, IS3, IS4, CL1, and CL24 also react thrombosis in patients with systemic lupus erythematosus: comparison between with ␤ GPI, the major autoantigen or cofactor for autoantibodies different assays for the detection of antiphospholipid antibodies. Thromb. Hae- 2 most. 76:916. detected by the conventional aCL ELISA (37). Viewed as a whole, 22. Rao, L. V. M., A. D. Hoang, and S. I. Rapaport. 1996. Mechanism and effects of these data suggest that the latter four mAb may recognize an the binding of lupus anticoagulant IgG and prothrombin to surface phospholipid. epitope shared by PT and ␤ GPI, and that such an epitope may be Blood 88:4173. 2 23. Galli, M., G. Beretta, M. Daldossi, E. M. Bevers, and T. Barbui. 1997. Different Ј Downloaded from analogous to the one revealed by P11 monoclonal aPT F(ab )2 anticoagulant and immunological properties of anti-prothrombin antibodies in isolated by panning a phage display Ab library on PT (36). How- patients with antiphospholipid antibodies. Thromb. Haemost. 77:486. 24. Horbach, D. A., E. Van Oort, R. H. W. M. Derksen, and P. G. de Groot. 1998. ever, CL24, IS3, and IS4 differ in their effects on thrombin activity The contribution of anti-prothrombin-antibodies to lupus anticoagulant activity: and AT inactivation of thrombin. In this context, the combined discrimination between functional and non-functional anti-prothrombin- data may suggest that there is more than one thrombin epitope that antibodies. Thromb. Haemost. 79:790. ␤ 25. Bertolaccini, M. L., T. Atsumi, M. A. Khamashta, O. Amengual, and is shared among PT, thrombin, and 2GPI. Clearly, further exper- G. R. V. Hughes. 1998. Autoantibodies to human prothrombin and clinical man- imentation to test these hypotheses is warranted. ifestations in 207 patients with systemic lupus erythematosus. J. Rheumatol. 25: http://www.jimmunol.org/ 1104. 26. Galli, M., L. Ruggeri, and T. Barbui. 1998. Differential effects of anti- Acknowledgments ␤2-glycoprotein I and antiprothrombin antibodies on the anticoagulant activity of We thank Paifei Chen for her technical assistance. activated protein C. Blood 91:1999. 27. de Groot, P. G., D. A. Horbach, M. J. A. Simmelink, E. Van Oort, and R. H. W. M. Derksen. 1998. Anti-prothrombin antibodies and their relation with References thrombosis and lupus anticoagulant. Lupus 7:S32. 28. Sorice, H., V. Pittoni, A. Circella, R. Misasi, F. Conti, A. Longo, G. M. Pontieri, 1. Hahn, B. H. 2001. Pathogenesis of systemic lupus erythematosus. In Kelley’s and G. Valesini. 1998. Anti-prothrombin but not “pure” anti-cardiolipin antibod- Textbook of Rheumatology. S. Ruddy, E. D. Harris, and C. B. Sledge, eds. ies are associated with the clinical features of the antiphospholipid antibody syn- W. B. Saunders Company, Philadelphia, p. 1089. drome. Thromb. Haemost. 80:713.

2. Greaves, M. 1999. Antiphospholipid antibodies and thrombosis. Lancet 29. Galli, M., and T. Barbui. 1999. Antiprothrombin antibodies: detection and clin- by guest on September 28, 2021 353:1348. ical significance in the antiphospholipid syndrome. Blood 93:2149. 3. Lockshin, M. D. 1999. Pregnancy loss in the antiphospholipid syndrome. 30. Esmon, C. T. 2000. Regulation of blood coagulation. Biochim. Biophys. Acta Thromb. Haemost. 82:641. 1477:349. 4. Boey, M. L., C. B. Colaco, A. E. Gharavi, K. B. Elkon, S. Loizou, and 31. Pearson, J. D. 1999. Endothelial cell function and thrombosis. Bailliere’s Best G. R. V. Hughes. 1983. Thrombosis in systemic lupus erythematosus: striking Pract. Res. Clin. Haematol. 12:329. association with the presence of circulating lupus anticoagulant. Br. Med. J. 287: 32. Dahlback, B. 2000. Blood coagulation. Lancet 355:1627. 1021. 33. Zhao, Y., R. Rumold, A. E. Ahmed, D. T. Le, B. H. Hahn, V. L. Woods, Jr., and 5. Harris, E. N., A. E. Gharavi, M. L. Boey, B. M. Patel, C. G. Mackworth-Young, P. P. Chen. 1999. An IgG anti-prothrombin antibody enhances prothrombin bind- S. Loizou, and G. R. Hughes. 1983. Anticardiolipin antibodies: detection by ing to damaged endothelial cells and shortens plasma coagulation times. Arthritis radioimmunoassay and association with thrombosis in systemic lupus erythem- Rheum. 42:2132. atosus. Lancet 2:1211. 34. Friguet, B., A. F. Chaffotte, L. Djavadi-Ohaniance, and M. E. Goldberg. 1985. 6. Pierangeli, S. S., M. Stewart, L. K. Silva, and E. N. Harris. 1998. An antiphos- Measurements of the true affinity constant in solution of antigen-antibody com- pholipid wet workshop: 7th International Symposium on Antiphospholipid An- plexes by enzyme-linked immunosorbent assay. J. Immunol. Methods 77:305. tibodies. J. Rheumatol. 25:156. 35. Bock, P., S. Olson, and I. Bjork. 1997. Inactivation of thrombin by antithrombin 7. Triplett, D. A. 1998. Many faces of lupus anticoagulants. Lupus 7:S18. is accompanied by inactivation of regulatory exosite i. J. Biol. Chem. 272:19837. 8. Wilson, W. A., A. E. Gharavi, T. Koike, M. D. Lockshin, D. W. Branch, 36. Chukwuocha, R. U., E. T. Hsiao, P. Shaw, J. L. Witztum, and P. P. Chen. 1999. J. C. Piette, R. Brey, R. Derksen, E. N. Harris, G. R. V. Hughes, et al. 1999. Isolation, characterization, and sequence analysis of five IgG monoclonal anti- ␤ International consensus statement on preliminary classification criteria for defi- 2-glycoprotein-1 and antiprothrombin antigen-binding fragments generated by nite antiphospholipid syndrome. Arthritis Rheum. 42:1309. phage display. J. Immunol. 163:4604. 9. McNeil, H. P., R. J. Simpson, C. N. Chesterman, and S. A. Krilis. 1990. Anti- 37. Zhu, M., T. Olee, D. T. Le, R. A. S. Roubey, B. H. Hahn, V. L. Woods, Jr., and phospholipid antibodies are directed against a complex antigen that includes a P. P. Chen. 1999. Characterization of IgG monoclonal anti-cardiolipin/anti␤2GP1 ␤ lipid binding inhibitor of coagulation: 2-glycoprotein I (apolipoprotein H). Proc. antibodies from two patients with the anti-phospholipid syndrome reveals three Natl. Acad. Sci. USA 87:4120. species of antibodies. Br. J. Haematol. 105:102. 10. Galli, M., P. Comfurius, C. Maassen, H. C. Hemker, M. H. de Baets, 38. Greenber, C. S., and C. L. Orthner. 1999. Blood coagulation and fibrinolysis. In P. J. C. van Breda-Vriesman, T. Barbui, R. F. A. Zwaal, and E. M. Bevers. 1990. Wintrobe’s Clinical Hematology. G. R. Lee, J. Foerster, J. Lukens, F. Paraskevas, Anticardiolipin antibodies (ACA) directed not to cardiolipin but to a plasma J. P. Greer, and G. M. Rogers, eds. Williams & Wilkins, Baltimore, MD, p. 684. protein cofactor. Lancet 335:1544. 39. Becker, D., J. Fredenburgh, A. Stafford, and J. Weitz. 1999. Exosites 1 and 2 are 11. Fleck, R. A., S. I. Rapaport, and L. V. Rao. 1988. Anti-prothrombin antibodies essential for protection of fibrin-bound thrombin from heparin-catalyzed inhibi- and the lupus anticoagulant. Blood 72:512. tion by antithrombin and heparin cofactor ii. J. Biol. Chem. 274:6226. 12. Bevers, E. M., M. Galli, T. Barui, P. Comfurius, and R. F. A. Zwaal. 1991. Lupus 40. Stubbs, M. T., and W. Bode. 1993. A player of many parts: the spotlight falls on anticoagulant IgGs (LA) are not directed to phospholipids only, but to a complex thrombin’s structure. Thromb. Res. 69:1. of lipid-bound human prothrombin. Thromb. Haemost. 66:629. 41. Marcum, J. A., and R. D. Rosenberg. 1989. Role of endothelial cell surface 13. Oosting, J. D., R. H. W. M. Derksen, I. W. G. Bobbink, T. M. Hackeng, heparin-like polysaccharides. Ann. NY Acad. Sci. 556:81. B. N. Bouma, and P. G. de Groot. 1993. Antiphospholipid antibodies directed 42. Olson, S. T., I. Bjork, R. Sheffer, P. A. Craig, J. D. Shore, and J. Choay. 1992. against a combination of phospholipids with prothrombin, protein C, or protein S: Role of the antithrombin-binding pentasaccharide in heparin acceleration of an- an explanation for their pathogenic mechanism? Blood 81:2618. tithrombin-proteinase reactions: resolution of the antithrombin conformational 14. Arvieux, J., G. Pernod, V. Regnault, L. Darnige, and J. Garin. 1999. Some an- change contribution to heparin rate enhancement. J. Biol. Chem. 267:12528. ticardiolipin antibodies recognize a combination of phospholipids with thrombin- 43. Colwell, N., D. Tollefsen, and M. Blinder. 1997. Identification of a monoclonal modified antithrombin, complement C4b-binding protein, and lipopolysaccharide thrombin inhibitor associated with multiple myeloma and a severe bleeding dis- binding protein. Blood 93:4248. order. Br. J. Haematol. 97:219.