90808 Clin Pathol 1993;46:908-9 11 Role of/h-glycoprotein I and anti-phospholipid antibodies in activation of protein C in vitro J Clin Pathol: first published as 10.1136/jcp.46.10.908 on 1 October 1993. Downloaded from

D M Keeling, A J G Wilson, I J Mackie, D A Isenberg, S J Machin

Abstract hypotheses suggesting mechanisms of action Aims-To investigate the effect of P2- whereby these antibodies might cause or be glycoprotein I (A GPI) on the thrombin/ associated with thrombosis. One of the possi- thrombomodulin dependent activation of bilities is interference with the activation of protein C; and to determine whether P2 protein C by the thrombin/thrombomodulin GPI dependent anticardiolipin anti- (IIa/TM) complex,8 9 a reaction that is bodies have any effect. enhanced by phospholipids."0 Activated pro- Methods-Protein C was activated by tein C inactivates factors Va and VIIIa and thrombin in the presence of thrombo- stimulates fibrinolysis," 12 and deficiency is a modulin and phospholipid vesicles in an risk factor for thrombosis."3 in vitro system. The effect of adding Recently it has been shown that purified purified P2 GPI to this system was aPL only bind to anionic phospholipids in the observed. Affinity purified anticardio- presence of a serum cofactor,14-'6 identified as lipin antibodies and total IgG from fl2-glycoprotein I (J GPI).'4 A protein that patients with anticardiolipin antibodies forms a complex with activated protein C via and the lupus anticoagulant were studied its active site has been described'7 and this for their effects on protein C activation seems to be /2 GPI.'8 We therefore decided to in the presence and absence of P2 GPI. investigate whether f2 GPI affected either the Results-/h-Glycoprotein I had no effect rate of protein C activation by the Ila/TM on the activity of preformed activated complex or the activity of activated protein C, protein C. When the phospholipid vesi- and to determine the effect of aPL on protein cles were incubated with I2 GPI before C activation both in the presence and the the addition of protein C, the activation absence ofP2 GPI. of protein C was inhibited in a dose dependent manner. With phosphati- dylserine:phosphatidylcholine vesicles at Methods http://jcp.bmj.com/ a concentration of 1 4uM:2 jfM, 1P2 GPI Plasma was obtained from five female and began to inhibit the reaction at a concen- one male patient who all had aCL and lupus tration of 15 nM, and at 4,uM (the anticoagulants. Their details have already normal plasma concentration) the been published.'9 activation of protein C was reduced to Cardiolipin, phosphatidylserine, phos- 40%. Anticardiolipin antibodies had no phatidylcholine, bovine serum (B-2771), demonstrable effect. bovine albumin (7030), human thrombin and on September 28, 2021 by guest. Protected copyright. Conclusions-fl2-Glycoprotein I inhibits hirudin were obtained from Sigma Chemical protein C activation in an in vitro sys- Company Ltd, Poole, Dorset; rabbit lung tem. Its physiological role is unknown thrombomodulin from American Diagnostica but it has potential procoagulant as well Inc., Greenwich, Connecticut, USA; human as anticoagulant properties. An effect of protein C from CRTS, Lille, France; protein antiphospholipid antibodies on protein C G, S sepharose fast flow, heparin sepharose, activation, which might explain their and sephacryl S200 HR from Pharmacia, association with thrombosis, could not Milton Keynes, Bucks; Protein C Substrate be shown. (2AcOH-H-D-Lys(Cbo)-Pro-Arg-pNA) and Department of Haematology, the reagents used in the diluted Russell's University College and (7 Clin Pathol 1993;46:908-91 1) viper venom test (DRVVT) were from Middlesex School of Unicorn Diagnostics, London. Medicine, London D M Keeling For cardiolipin affinity chromatography, A J G Wilson Antibodies to phospholipid (aPL) are associ- the method of McNeil et al'4 was used, I J Mackie ated with venous and arterial thrombosisl1 though with step rather than gradient elution. S J Machin and recurrent fetal loss.5 6 Fetal loss may be The cardiolipin affinity column was equili- Department of due to thrombosis of placental vessels.7 These brated with 0-02 M TRIS (pH 7.2). Plasma Rheumatology D A Isenberg antibodies are identified in two different was diluted 1 in 5 with this buffer and applied ways, as to the column. Correspondence to: either immunoglobulins reacting After washing, bound protein Dr D M Keeling, with cardiolipin (aCL) or other anionic phos- was eluted with 0-02 M TRIS/0-75 M NaCl Department of Haematology, pholipids (usually detected in solid phase (pH 7.2). Addenbrooke's Hospital, enzyme linked immunosorbent assay), or by For affinity purification of anticardiolipin Hills Road, Cambridge CB2 2QQ. their ability to prolong phospholipid depen- antibodies 10 ml plasma from the six patients Accepted for publication dent coagulation tests, the so-called lupus were chromatographed on the cardiolipin 5 May 1993 anticoagulants. There have been many affinity column. The protein eluted was dia- 132 GPI antiphospholipid antibodies and activation ofprotein C 909

lysed against 0-02 M TRIS (pH 7 2) and composition. The mixture was dried in nitro- applied on to a protein G column. IgG was gen and the residue dispersed in TBS to give eluted off the protein G with 0 1 M the following concentrations: phosphatidyl glycine/HCl (pH 7-2). The antibodies were serine:phosphatidylcholine 25,uM:50,uM; J Clin Pathol: first published as 10.1136/jcp.46.10.908 on 1 October 1993. Downloaded from concentrated in 0-02 M TRIS/0I15 M NaCl cardiolipin:phosphatidylcholine 50,uM:200 (pH 7-4) (TBS) using an Amicon Mini- ,uM; cardiolipin alone 200,uM. The emulsion Ultrafiltration Cell with a PM30 membrane was sonicated for 30 seconds every minute for to 0-15 g/l (1 ,uM) assuming an extinction 40 minutes under nitrogen at 27°C using an coefficient of 13-6 at 280 nm for a 1% solu- MSE Soniprep 1500 at an amplitude of 6 tion. ,uM. The resulting unilamellar vesicles were fl2-glycoprotein I (IA GPI), 0 5 g/l (10 uM) kept at room temperature before use. in TBS was purified by a previously described Activation of protein C was carried out in method,20 modified from that of McNeil et TBS (pH 7 4) containing 0 5 g/l bovine albu- al.14 min: 100 ,ul CaCl2 (25 mM), 10 ul phospho- Total IgG was prepared from the plasma of lipid vesicles, and 10 41l TM (1 -25 U/ml) the six patients using the protein G column were incubated for 15 minutes. TBS (11Opl) and concentrated to 3 g/l (20 ,uM). with or without fl2 GPI or IgG was then The antiphospholipid enzyme linked added and after 15 minutes 10 1ul IIa (25 immunosorbent assay (ELISA) was based on U/ml) and 104pl protein C (25 U/ml) were that of Gharavi et al,21 using 10% (v/v) bovine added to start the reaction. Final concen- serum as the blocking agent and sample dilu- trations were CaCl 10 mM, TM 0.05 U/ml, ent. Blanks obtained from uncoated wells on IIa 1 U/ml, protein C 1 U/ml; and for the the same plate were subtracted to account for different phospholipid vesicles, phospha- non-specific binding. The reference serum tidylserine:phosphatidylcholine 1 uM:2,uM, was calibrated at 105 GPL against a standard cardiolipin:phosphatidylcholine 2 pM:8 pM, prepared at an international workshop22 cardiolipin alone 8 uM. After 15 minutes (GPL are arbitrary units derived from the 80,ul samples were taken in duplicate and activity of an affinity purified serum). added to 20,ul hirudin (10 U/ml) in a micro- In our modified DRVVT 0 1 ml affinity titre plate to inactivate the thrombin. Protein purified aCL (AP-aCL), or total IgG, or C (50 pl) substrate (1-8 mM) were added buffer, was mixed with 0 1 ml normal plasma and after incubating for 15 minutes at 37°C and 0 1 ml of dilute phospholipid reagent (or cleavage of substrate by preformed activated freeze-thawed platelets); 0 1 ml Russell's protein C was stopped by adding 50 p1 of viper venom was then added and after incu- 1 M citric acid. The activated protein C con- bation for 30 seconds clotting was initiated by tent was then proportional to the absorbance adding 0 1 ml of 25 mM CaCl2. measured at 405 nm. The amount of acti- Stock solutions of phospholipids in organic vated protein C formed in an experiment was

solvents were mixed to give the desired molar expressed as a percentage of that obtained in http://jcp.bmj.com/ a control, run concurrently, under these standard conditions with no fl2 GPI or immunoglobulin added. Each of the phos- pholipid vesicles at these concentrations Figure 1 Protein C activation (as a percentage enhanced the reaction about five-fold-that of that without 32 GPI) v ;-o--R is, in the absence of phospholipid vesicles 32 GPI concentration (og protein C activation as defined above was on September 28, 2021 by guest. Protected copyright. scale) for the three 0 phospholipid vesicles, 4- about 20%. phosphatidylserine: co phosphatidykholine (0, ), cardiolipin U (+,- -) and cardiolipin: Results phosphatidylkholine When fl2 GPI was added at the end of the (A, ---). Results are the reaction, at a final concentration of 4,uM, it mean offive experiments. 0~ had no effect on the activity of preformed activated protein C (data not shown). P2 GPI (M) When the phospholipid vesicles were incu- bated with J2 GPI before the addition of pro- tein C, the activation of protein C was inhibited in a dose dependent manner. This Figure 2 Protein C inhibition was seen with all three phospho- activation (as a percentage All further of that with 1 p1M lipid vesicles (fig 1). experiments phosphatidylserine without were performed with the phosphatidylserine: 132 GPI) v phospholipid phosphatidylcholine vesicles. With these vesi- concentration (log scale), cles at a concentration of 1 uM:2 uM, GPI without 12 GPI (0) and fl2 with 0 1iuM fi2GPI (0). began to inhibit the reaction at a concentra- Results are the mean of two tion of 15 nM, and at 4,uM the activation of experiments. protein C was reduced to 40%. At a f62 GPI concentration of 0 1 pM the inhibition could not be completely overcome by increasing the phospholipid concentration (fig 2). Phosphatidylserine: phosphatidylcholine The activities of the IgG preparations in (,uM Phosphatidylserine) the aCL ELISA and in our modified DRVVT 910 Keeling, Wilson, Mackie, Isenberg, Machin

Table 1 ACL levels and DRVVTratios with dilute phospholipid (PL) andfreeze vesicles in case the phospholipid phase had thawed platelets (platelets) for AP-aCL and total IgG purifiedfrom six patients any effect. Pure cardiolipin vesicles in the

aCL (GPL) DRVVTAP-aCL DRVVT total IgG bilayer phase undergo transition to the hex- J Clin Pathol: first published as 10.1136/jcp.46.10.908 on 1 October 1993. Downloaded from Case No AP-aCL Total IgG PL Platelets PL Platelets agonal phase when calcium is added; the cardiolipin:phosphatidylcholine mixture is 1 139 89 1-42 0-92 1-43 0-91 2 135 126 1-53 1-00 1-14 0-99 stabilised and remains in the bilayer phase 3 33 18 1-18 1-06 2-01 1-25 when calcium is added as do the more 4 110 72 1-34 0.99 1-27 1-07 5 38 19 1-06 0-93 1-11 0-99 physiological phosphatidylserine:phospha- 6 23 16 0 94 0 95 1-04 0-89 tidylcholine vesicles. Whichever vesicles were Results of the mean of the two experiments. used, fl2 GPI resulted in decreased protein C activation at low concentrations (the normal plasma concentration is 4 ,uM). These results contrast with those of Oosting et al,30 who Table 2 Protein C activation in the presence ofaPL with found no effect of f32 GPI on endothelial cell and without 112 GPI, expressed as percentage ofthat mediated protein C activation, using a more obtained in absence ofimmunoglobulin and 1J2 GPI physiological in vitro system. Clearly, our AP-aCL Total IgG results do not mean that /2 GPI inhibits pro- Case No Alone +f2 GPI Alone +fl2 GPI tein C activation in vivo, but they show that 1 107 70 104 72 /32 GPI has the potential to interfere with anti- 2 96 68 92 63 coagulant as well as procoagulant pathways. 3 114 84 90 66 4 101 78 103 73 On the basis of their in vitro data, workers 5 113 84 102 71 have naturally speculated that /32 GPI may act 6 95 83 90 56 Normal IgG as a physiological anticoagulant,2529 but this mean (SD) 97 (8-6) 68 (8-2) 95 (7 8) 64 (7 8) may not be so. Indeed, the plasma concentra- tion of 32 GPI depends on genotype; homo- Results for the patient IgG are the mean of two experiments, for normal IgG at the appropriate concentration the mean zygotes for the normal gene (BgN, gene (SD) of five experiments. frequency 0 937) comprise 88% of the popu- lation and have concentrations of 150-300 mg/l. Homozygotes for the BgD gene (gene are shown in table 1. The effect of each on frequency 0 063) have concentrations of 0-50 protein C activation was studied. In each mg/l and would be expected to comprise case 100 ,ul of the immunoglobulin was used 0-4% of the population, and heterozygotes to give a final concentration of 04 ,uM for the have intermediate values, 50-150 mg/l, and AP-aCL and 8 pM for the total IgG. comprise 12% of the population. Despite this Experiments were done without fl2 GPI and there are no reports of thrombosis associated with /)2 GPI at a final concentration of 0 04 with low /32 GPI concentrations and one ,uM, which in the absence of immunoglobulin report which suggests such an association is gave a protein C activation of 74% (n = 5, unlikely.31 http://jcp.bmj.com/ SD = 6-8%) and was at a point on the steep In 1986 Freyssinet et a18 isolated a lupus part of the inhibition curve (fig 1). Neither anticoagulant fraction from a patient's plasma the AP-aCL nor the total IgG preparations by gel filtration. They studied the effect of had a demonstrable effect (table 2). In the this IgM containing fraction on the IIa/TM absence of fi GPI the mean (SD) protein C dependent activation of protein C, using activation for the six AP-aCL and the six total human derived purified proteins. They found IgG preparations was 104% (8-3%) and 97% that the anticoagulant could neutralise the on September 28, 2021 by guest. Protected copyright. (6-8%) and in the presence of 32 GPI 78% enhancement of protein C activation seen (7-2%) and 67% (6 6%), respectively. These with phospholipids. In 1988 Cariou et a19 results do not differ significantly from those purified IgG from eight patients with the without any immunoglobulin, nor from those lupus anticoagulant on protein A sepharose. with normal IgG at the appropriate concen- They studied the activation of protein C by tration (t test; all p > 0 05). thrombin in the presence of human endothe- lial cells or rabbit thrombomodulin. In both systems the IgG with lupus anticoagulant Discussion activity inhibited activation of protein C. In l2 GPI binds to phospholipids,2'24 platelets,2' contrast, Oosting et aP' could not detect an and perhaps activated protein C,17 18 and has effect for aPL on endothelial cell mediated been found to have several effects on the protein C activation in the presence or coagulation system. It inhibits the contact absence of/2 GPI. They studied 46 sera from activation of the intrinsic pathway of coagula- patients with systemic lupus erythematosus, tion25-28 and it interferes with the assembly of 19 of whom had aCL or lupus anticoagulant the prothrombinase complex on platelet activity, and purified total IgG on protein G membranes and phospholipid vesicles.29 We sepharose from 12 of them, six of whom had could not detect any inhibition of preformed aCL or lupus anticoagulant activity. We, too, activated protein C at physiological concen- could not show an effect on protein C activa- trations of fl2 GPI. It might, however, be tion with either AP-aCL or total IgG from expected to inhibit the IIa/TM/phospholipid our patients. Some of these disparate results dependent activation of protein C, very much may be due to the heterogeneity of aPL, an like it inhibits prothrombinase, simply by understanding of which is beginning to occupying the phospholipid surface. We ini- emerge. Bevers, Galli, and colleagues have tially used the three different phospholipid proposed that the term lupus anticoagulant /32 GPI antiphospholipid antibodies and activation ofprotein C 911

is used to describe those aPL which anticoagulant on antithrombogenic properties of endothelial cells-inhibition of thrombomodulin-depen- prolong coagulation tests because they recog- dent protein C activation. Thromb Haemostas 1988; nise a prothrombin-Ca2+-phospholipid com- 60:54-8. 10 Freyssinet J-M, Gauchy J, Cazenave J-P. The effect of J Clin Pathol: first published as 10.1136/jcp.46.10.908 on 1 October 1993. Downloaded from plex32 and that aCL which recognise a /32 phospholipids on the activation of protein C by the GPI-phospholipid complex are divided into human thrombin-thrombomodulin complex. Biochem Jf 1986;238:151-7. aCL-type A which prolong phospholipid 11 Stenflo J. Structure and function of protein C. Semin dependent coagulation reactions and aCL- Thromb Haemostas 1984;10:109-21. 12 Van Hinsberg VWM, Bertina RM, Van Wijngaarden A, type B which have no anticoagulant activity.3 Van Tilburg NH, Emeis JJ, Haverkate F. Activated pro- It seems likely that some of these patients will tein C decreases plasminogen activator inhibitor activity in endothelial cell conditioned medium. Blood 1985; have antibodies against other protein-phos- 65:444-51. pholipid complexes, such as factor X-Ca2+- 13 Griffin JH, Evatt B, Zimmermann TS, Kleiss AJ, Wideman C. Deficiency of protein C in congenital phospholipid (which would be a lupus thrombotic disease. JClin Invest 198 1;68: 1370-3. anticoagulant) and protein C-Ca2+-phospho- 14 McNeil HP, Simpson RJ, Chesterman CN, Krilis SA. Anti-phospholipid antibodies are directed against a lipid; indeed antibodies against prothrombin, complex antigen that includes a lipid binding inhibitor factor X, and protein C could be cross-reac- of coagulation: fl,-Glycoprotein I (apolipoprotein H). Proc NadAcad Sci USA 1990;87:4120-4. tive. Our six patients all had aCL and lupus 15 Galli M, Comfurius P, Maassen C, et al. Anticardiolipin activity. Their total IgG could antibodies (ACA) directed not to cardiolipin but to a anticoagulant plasma protein cofactor. Lancet 1990;335:1544-7. contain any of the antibody types mentioned. 16 Matsuura E, Igarashi Y, Fujimoto M, Ichikawa K, Koike of T. Anticardiolipin cofactor(s) and differential diagnosis We affinity purified aCL in the absence of autoimmune disease. Lancet 1990;336:177-8. calcium, so antibodies which recognise a pro- 17 Canfield M, Kisiel W. Evidence of normal functional lev- complex could not els of activated protein C inhibitor in combined factor temi-Ca2+-phospholipid V/VIII deficiency disease. J Clin Invest 1982;70: copurify. The resulting antibodies have 1260-72. 18 Van der Meer FJM, Van Tilburg NH, Van Wijngaarden already been characterised. They recognise A, Van der Linden IK, Briet E, Bertina R. A second phospholipid bound 82 GPI and in cases 1-4 plasma inhibitor of activated protein C: a,-antitrypsin. lupus anticoagulant activity is probably Thromb Haemostas 1989;62:756-62. the 19 Keeling DM, Wilson AJG, Mackie IJ, Isenberg DA, due to aCL-type A.1920 This could not be Machin SJ. Lupus anticoagulant activity of some shown in cases 5 and 6, and in these patients antiphospholipid antibodies against phospholipid bound /h-glycoprotein I. J7Clin Pathol 1993;46:665-7. it may have been that the lupus anticoagulant 20 Keeling DM, Wilson AJG, Mackie IJ, Machin SJ, activity was not due to antibodies against /32 Isenberg DA. Some "anti-phospholipid antibodies" bind to fl,-glycoprotein I in the absence of phospholipid. GPI. Unfortunately the total IgG from case 5 BrJ Haematol 1992;82:571-4. weak lupus anticoagulant 21 Gharavi AE, Harris EN, Asherson RA, Hughes GRV. had extremely Anticardiolipin antibodies: isotype distribution and activity and that from case 6 none at all (the phospholipid specificity. Ann Rheum Dis 1987;46:1-6. 22 Harris EN, Gharavi AE, Patel SP, Hughes GRV. antibodies may have been inactivated during Evaluation of the anticardiolipin antibody test: report of purification or they may have been IgM). We an international workshop held 4 April 1986. Clin Exp effect Immunol 1987;68:215-22. have therefore not been able to show an 23 Schousboe I. Characterization of the interaction between with aPL which recognises a P2 GPI- fl,-glycoprotein I and mitrochondria, platelets, lipo- phospholipid complex, but cannot exclude somes and bile acids. IntlJBiochem 1983;15:1393-401. 24 Wurm H. fi,-Glycoprotein I (apolipoprotein H) interac-

the possibility that antibodies to other protein- tions with phospholipid vesicles. Int J Biochem 1984; http://jcp.bmj.com/ phospholipid complexes affect protein C 16:511-5. 25 Schousboe I. h,B-Glycoprotein I: a plasma inhibitor of the activation. contact activation of the intrinsic blood coagulation of aPL to impair pro- pathway. Blood 1985;66:1086-91. Finally, an inability 26 Schousboe I. In vitro activation of the contact activation tein C activation does not mean that they do system (Hageman factor system) in plasma by acidic not interfere with the activated protein C/pro- phospholipids and the inhibitory effect of /,-glycopro- tein I on this activation. IntJIBiochem 1988;20:309-15. tein S mediated inactivation of factors Va and 27 Schousboe I, Rasmussen MS. The effect of /,-glycopro-

tein I on the dextran sulfate and sulfatide activation of on September 28, 2021 by guest. Protected copyright. VIIIa, as has been suggested.'s36 the contact system (Hageman factor system) in the blood coagulation. IntJ Biochem 1988;20:787-92. 28 Henry ML, Everson B, Ratnoff OD. Inhibition of the acti- vation of Hageman factor (factor XII) by fl,-glycopro- tein I. YLab Clin Med 1988;111:519-23. 29 NimpfJ, Bevers EM, Bomans PHH, et al. Prothrombinase 1 Bowie WEJ, Thompson JH, Pascuzzi CA, Owen GA. activity of human platelets is inhibited by fl,-glycopro- Thrombosis in systemic lupus erythematosus despite tein I. Biochim BiophysActa 1986;884:142-9. circulating anticoagulant. J Clin Invest 1963;62:416-30. 30 Oosting JD, Derksen RHWM, Hackeng TM, et al. In 2 Mueh JR, Herbst KD, Rapaport SI. Thrombosis in vitro studies of antiphospholipid antibodies and its patients with the "lupus"-type circulating anticoagulant. cofactor, fl,-glycoprotein I, show negligible effects on Ann Intern Med 1980;92:156-60. endothelial cell mediated protein C activation. Thromb 3 Boey ML, Colaco CB, Charair AE, Eckon KB, Loizou S, Haemostas 1991;66:666-71. Hughes GRV. Thrombosis in systemic lupus erythe- 31 Bancsi LFJMM, van der Linden IK, Bertina RM. fl,2- matosus: striking association with the presence of circu- Glycoprotein I deficiency and the risk of thrombosis. lating lupus anticoagulant. BMJ 1983;287:1021-3. Thromb Haemostas 1992;67:649-53. 4 Elias M, Eldor A. Thromboembolism in patients with 32 Bevers EM, Galli M, Barbui T, Comfurius P, Zwaal RFA. "lupus"-like circulating anticoagulant. Arch Int Med Lupus Anticoagulant IgGs (LA) are not directed to 1984;144:510-5. phospholipids only, but to a complex of lipid-bound 5 Nilsson IM, Astedt B, Hedner U, Berezin D. Intrauterine human prothrombin. Thromb Haemostas 1991;66: death and circulating anticoagulant ("antithromboplas- 629-32. tin"). Acta Med Scand 1975;197:153-9. 33 Galli M, Comfurius P, Barbui T, Zwaal, RFA, Bevers 6 Firkin BG, Howard MA, Radford N. Possible relationship EM. Anticoagulant activity of /h-glycoprotein I is poten- between lupus inhibitor and recurrent abortion in young tiated by a distinct group of anticardiolipin antibodies. women. Lancet 1980;ii:366. Thromb Haemostas 1992;68:297-300. 7 De Wolf F, Carreras LO, Moerman P, Vermylen J, Van 34 Marciniak E, Romond EH. Impaired catalytic fimction of Assche A, Renaer M. Decidual vasculopathy and exten- activated protein C: a new in vitro manifestation of sive placental infarction in a patient with repeated lupus anticoagulant. Blood 1989;74:2426-32. thromboembolic accidents, recurrent fetal loss and a 35 Malia RG, Kitchen S, Preston FE. Inhibition of activated lupus anticoagulant. Am Y Obstet Gynecol 1982;142: protein C and its cofactor protein S by antiphospholipid 829-34. antibodies. Br J Haematol 1990;76:101-7. 8 Freyssinet J-M, Wiesel M-L, Gauchy J, Boneu B, 36 Borrell M, Sala N, de Castellarnau C, Lopez S, Gari M, Cazenave J-P. An IgM lupus anticoagulant that neutral- Fontcuberta J. Immunoglobulin fractions isolated from izes the enhancing effect of phospholipid on purified patients with antiphospholipid antibodies prevent the endothelial thrombomodulin activity-a mechanism for inactivation of factor Va by activated protein C on thrombosis. Thromb Haemostas 1986;55:309-13. human endothelial cells. Thromb Haemostas 1992;68: 9 Cariou R, Toblem G, Bellucci S, et al. Effect of lupus 268-72.