ARTICLES

NATURE MEDICINE • VOLUME 5 • NUMBER 9 • SEPTEMBER 1999 1044

Hemophilia A is an X chromosome-linked recessive disorder re-sulting in defective or deficient factor VIII (FVIII) molecules,which, in its severe form, is a life-threatening and crippling hem-orrhagic disease. Infusion of homologous FVIII to patients withsevere hemophilia A results, in 25% of patients, in the emer-gence of alloantibodies against FVIII (inhibitors)(ref. 1) that in-hibit FVIII procoagulant activity by steric hindrance of theinteraction of FVIII either with stabilizing molecules2, with mole-cules essential for its activity3,4 or with activating molecules5.Here, we report on the proteolysis of FVIII by alloantibodies oftwo patients with severe hemophilia A, demonstrating a previ-ously unknown mechanism by which FVIII inhibitors may preventthe pro-coagulant function of FVIII. The kinetic parameters ofFVIII hydrolysis indicate a functional role for the catalytic im-mune response in the inactivation of FVIII in vivo. The characteri-zation of alloantibodies against FVIII as site-specific proteasesmay provide new approaches to the treatment of FVIII inhibitors.

We affinity-purified alloantibodies against FVIII from the plasmaof three hemophilia patients with inhibitor on a FVIII–Sepharosematrix. The yields of IgG antibodies against FVIII were 130, 20and 280 µg per 10 mg of loaded IgG in patients Bor, Che andWal, respectively (143 ± 130 µg/ml of plasma). Affinity-purifiedantibodies against FVIII of patients Bor, Che and Wal inhibitedFVIII pro-coagulant activity up to 57.0, 64.0 and 43.0 Bethesdaunits (BU)/mg of IgG, respectively. 125I-labeled FVIII was incu-bated with purified IgG antibody against FVIII of the patients be-fore SDS–PAGE and autoradiography. Untreated 125I-labeled FVIIIhad a characteristic electrophoretic pattern of migration, withbands of 70, 79, 98, 200 and 300 kDa (Fig. 1a). In two of thethree patients, incubation of FVIII with IgG antibody againstFVIII resulted in hydrolysis of FVIII. Incubation of IgG from pa-tient Bor resulted in the appearance of protein bands of 65, 50,41, 20 and 3 kDa (Fig. 1). For IgG from patient Wal, all bandscharacteristic of FVIII were hydrolyzed and the residual bandswere 58, 21, 16 and 3 kDa. In contrast, the profile of FVIII wasunchanged when 125I-labeled FVIII was incubated with IgG anti-body against FVIII from patient Che. The profile of FVIII was alsounchanged after incubation with the human monoclonal anti-body against cytomegalovirus M061 or with normal polyclonalhuman IgG, which have no inhibitory activity against FVIII.Effluents of the affinity columns, devoid of antibodies againstFVIII, did not hydrolyze 125I-labeled FVIII (Fig. 1b). Removal ofIgG from the eluates containing antibodies against FVIII of pa-tients Wal and Bor by protein G chromatography resulted in the

loss of their hydrolytic activity for FVIII (Fig. 1c). To further ex-clude the possibility of contamination by proteases, we treatedantibodies against FVIII from patient Wal with 8 M urea and sub-jected them to size-exclusion chromatography. The hydrolyzingactivity co-eluted with the IgG fraction, whereas the activity wasnot detected in fractions from which IgG was absent (data notshown). In addition, F(ab′)2 of IgG from patient Wal hydrolyzedFVIII, further demonstrating that proteolytic cleavage was de-pendent on the antibody-combining sites (data not shown). Co-incubation of 125I-labeled FVIII with IgG antibody against FVIIIfrom patients Bor and Wal with the protease inhibitors aprotinin(0.15 µM), E-64 (28 µM), EDTA (1.3 µM), leupeptin (10 µM) orpepstatin (10 µM) for 5 h did not result in inhibition of prote-olytic activity (data not shown). These results confirm thathydrolysis of FVIII is mediated by IgG antibody against FVIII.

We characterized the cleavage sites in the FVIII molecule byamino-acid sequencing of peptide fragments produced by thehydrolysis of FVIII by IgG antibody against FVIII from patientWal. The major scissile bonds were R372–S373, located betweenthe A1 and A2 domains of FVIII; Y1680–D1681, in the N termi-nus of the A3 domain; and E1794–D1795 in the A3 domain.

The proteolytic cleavage of FVIII by IgG antibody against FVIIIfrom patients Bor and Wal is dose-dependent (Fig. 2). Differentpatterns of digestion of FVIII in patients further indicated the de-pendency of hydrolysis on the nature of the antibody bindingsite. Given that the concentration of IgG antibody against FVIIIin the plasma of hemophilic patients is 140 µg/ml (1 µM)(ref. 6),and the plasma concentration of FVIII in patients after infusionof FVIII is 0.13–0.67 nM (0.2–1.0 IU/ml)7, the molar ratio of IgGantibody against FVIII to FVIII in patients’ plasma is1,400–7,200. These molar ratios in the experiments were 0.3–33.Thus, hydrolysis occurred when these molar ratios in the experi-mental samples of those in patients’ plasma, indicating that hy-drolysis is a mechanism of FVIII inactivation by thealloantibodies of the patients in vivo.

We evaluated the kinetics of the hydrolysis of FVIII by alloan-tibodies against FVIII from the patients (Fig. 2c and 2d). The rateof hydrolysis of 125I-labeled FVIII by IgG antibody against FVIIIfrom patient Wal was faster than that of IgG from patient Bor.Furthermore, inhibition assays were done by incubating IgG an-tibody against FVIII from patient Wal with increasing concen-trations of unlabeled FVIII and a fixed concentration of125I-labeled FVIII. The addition of unlabeled FVIII resulted indose-dependent inhibition of hydrolysis of 125I-labeled FVIII(Fig. 3). Saturation of FVIII hydrolysis was not attained with the

Catalytic activity of antibodies against factor VIII inpatients with hemophilia A

SÉBASTIEN LACROIX-DESMAZES1, ALEXANDRE MOREAU1, SOORYANARAYANA1,CÉCILE BONNEMAIN1, NATALIE STIELTJES2, ANASTAS PASHOV1, YVETTE SULTAN2,

JOHAN HOEBEKE3, MICHEL D. KAZATCHKINE1 & SRINIVAS V. KAVERI1

1INSERM U430 and Université Pierre et Marie Curie, Hôpital Broussais, 75014 Paris, France2Natalie Stieltjes, Centre des hémophiles, Hôpital Cochin, Paris, France

3Johan Hoebeke, UPR 9021 CNRS, IBMC, Strasbourg, FranceCorrespondence should be addressed to S.L.D. and S.V.K.; email: kaveri@hbroussais.fr

© 1999 Nature America Inc. • http://medicine.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m

NATURE MEDICINE • VOLUME 5 • NUMBER 9 • SEPTMEBER 1999 1045

ARTICLES

maximum concentration of FVIII used (1.7 µM). However, incu-bation of the antibodies with increasing concentrations of thegeneric PFR-methylcoumarinamide peptide lead to saturationof the hydrolytic activity towards PFR-methylcoumarinamide8

(data not shown). The curves of the reciprocal of the velocityplotted as a function of the reciprocal of the substrate concen-tration were linear (r = 0.99), indicating that the reaction con-formed to simple Michaelis–Menten kinetics, as for polyclonalcatalytic antibodies9,10. The calculated average Km and apparentVmax for the reaction were 9.46 ± 5.52 µM and 85 ± 60 fmol/min,respectively. The nominal Kcat value is the sum of the individualconstants of different antibodies within the polyclonal prepara-

tion and was equal to 0.026 per min. In normal plasma, FVIII isactivated by cleavage by thrombin and activated factor X toform a heterotrimer A2/A1/A3-C1-C2 with full pro-coagulantactivity11. The reported catalytic efficiencies for cleavage of FVIIIby thrombin and activated factor X are 5 × 106 and 1 × 106 per Mper second, respectively12,13. The values for the antibodies frompatient Wal, 2.6 × 103 ± 0.5 × 103 per M per second (Fig. 3), were0.1% of these. FVIII is subsequently inactivated by cleavage byactivated factor IX (refs. 14,15) and activated protein C (ref. 16).The rates of cleavage of FVIII by activated factor IX and acti-vated protein C are 0.3 and 2 nmol per minute per nmol, re-spectively17. Here, the rate of hydrolytic activity of antibodies

Fig. 1 Hydrolysis of 125I-labeled FVIII by affinity-purified IgG antibodiesagainst FVIII of hemophilia A patients. a, 125I-labeled FVIII was incubated for10 h in the presence or absence of IgG antibody against FVIII from patientsBor, Che and Wal. mAb, human monoclonal IgG antibody against CMVM061; IVIg, normal unfractionated human polyclonal IgG (negative con-trols). Incubation in buffer alone did not result in hydrolysis (lanes 1 (0 h)and 2 (10 h), nor did incubation with mAb or IVIg. b, Hydrolysis is depen-dent on antibodies specific for FVIII. 125I-labeled FVIII was incubated for 10 hin buffer alone (Control), with IgG antibody against FVIII from patient Wal

Fig. 2 Dose- and time-dependency of proteolysis of 125I-labeled FVIII byaffinity-purified antibodies against FVIII from hemophilia A patients. 125I-la-beled FVIII was incubated with IgG antibody against FVIII; factor VIII wasthen analyzed by 7.5% SDS–PAGE. Molecular weight standards between

gels. a and b, Increasing amounts (below gels) of antibodies from patientsWal (a) and Bor (b) were incubated for 2 h. c and d, The kinetics of 125I-la-beled FVIII proteolysis; anti-FVIII antibodies from patients Wal (c) and Bor(d) were incubated up to 4 h (time below gels).

(the acid-eluate of the column; Eluate) or with IgG in the effluent of the affin-ity column (Flow-through). c, Hydrolysis is dependent on antibodies of theIgG isotype. Acid-eluted fractions containing IgG antibody against FVIII frompatients Wal and Bor were depleted of IgG by chromatography on protein Gbefore incubation for 2 h with 125I-labeled FVIII. IgG-depleted fractions (–)were tested at a similar dilution to the dilution that resulted in a 50-µg/mlconcentration of IgG in the nondepleted (+) samples. Control, FVIII incu-bated for 2 h in buffer alone. Samples were separated by 7.5% (a and c) and15% (b) SDS–PAGE. Left margins, molecular weight standards.

a b c

a b c d

IgG (nM)

Time (min)

© 1999 Nature America Inc. • http://medicine.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m

1046 NATURE MEDICINE • VOLUME 5 • NUMBER 9 • SEPTEMBER 1999

ARTICLES

against FVIII was estimated to be 0.02–0.2 nmol per minute pernmol, as assessed using methylcoumarinamide peptides8 (datanot shown). The apparent catalytic efficiency, Vmax and rate ofhydrolysis of the antibodies from patient Wal are probably un-derestimated because of the assumption that all antibodiesagainst FVIII are catalytic. This is certainly not true, as the pop-ulation of antibodies against FVIII of the patients also containsantibodies without catalytic activity. Furthermore, whereas en-zymes are more efficient than catalytic antibodies in hydrolyz-ing FVIII, they are rapidly inactivated in plasma. In contrast,catalytic antibodies could exert a biological effect for hours ordays. Assuming that the plasma concentration of antibody-combining sites is 2 µM in patient Wal, and that the concentra-tion of FVIII after infusion to patients is 0.13–0.67 nM (ref. 7), anominal catalytic constant of 0.026 per minute would result inthe hydrolysis of 52 nM FVIII per minute, and the total amountof infused FVIII would be hydrolyzed within 0.15–0.77 seconds.Thus, the kinetic parameters of hydrolysis calculated in vitro in-dicate that proteolysis may be a mechanism of FVIII inactiva-tion by the alloantibodies of the patients in vivo. However, thecontribution of catalytic antibodies to the inactivation of FVIIIin patients remains undetermined. Thus, antibodies againstFVIII in a patient are a mixture of catalytic and noncatalytic an-tibodies; and the inhibitory activity results from several simul-taneous mechanisms, such as steric hindrance inhibiting theinteraction of FVIII with relevant molecules, and proteolysis of

FVIII by the subset of catalytic antibodies.The association of FVIII with von Willebrand factor (vWF) in-

creases the catalytic rate of thrombin for FVIII, whereas it pro-tects FVIII from hydrolysis by activated protein C (ref. 18). Weassessed the effect of the binding of FVIII to vWF on the cleavageof FVIII by IgG antibody against FVIII. 125I-labeled FVIII bound tovWF in a dose-dependent manner. The addition of vWF to FVIIIpartially inhibited the hydrolysis of FVIII by antibodies, by36.9%, when vWF and FVIII were mixed using a weight/weightratio similar to that in normal plasma: 30 µg/ml of vWF and 300ng/ml of FVIII (ref. 19)(data not shown). The protective effect ofvWF may be secondary to the ‘masking’ of some of the cleavagesites for catalytic antibodies on the FVIII molecule after complexformation with vWF, or to competition for cleavage betweenFVIII and vWF. Indeed, residues Y1680–D1681, which define thescissile bond located in the N terminus of the A3 domain, are es-sential for the interaction of FVIII with vWF (ref. 20).

The identification of FVIII inhibitors as catalytic antibodies ex-tends the spectrum of catalytic immune responses, with reportsof hydrolyzing antibodies against vasoactive intestinal peptide,DNA and thyroglobulin8,21–23. This is the first report, to ourknowledge, of the induction of catalytic antibodies in humans inresponse to exogeneous administration of a protein antigen. Thekinetic parameters of FVIII hydrolysis by catalytic antibodies andthe estimated amounts of these antibodies in plasma indicate afunctional role for the catalytic immune response in inactivatingFVIII in vivo. Identification of target epitopes for proteolytic anti-bodies against FVIII may be essential for understanding thepathophysiology of the FVIII inhibitor response. Furthermore,the characterization of FVIII inhibitors as site-specific proteasesmay provide new approaches for the treatment of inhibitors.

MethodsAffinity-purification of antibodies against FVIII. Antibodies were isolatedfrom plasma by ammonium sulfate precipitation. Antibodies reactive withFVIII were then affinity-purified on a CNBr-activated Sepharose 4B matrix towhich immunopurified FVIII had been coupled (25,000 U per 3 g of gel;Laboratoire Français des Biotechnologies, Les Ulis, France). The ‘flow-throughs’ of the columns were collected. After samples were washed exten-sively with PBS, pH 7.4, antibodies against FVIII were eluted using 0.2 Mglycine, pH 2.8, dialyzed against PBS and concentrated with Centriprep.‘Flow-throughs’ and eluates were divided into aliquots and stored at –20ºC. F(ab′)2 fragments of antibodies against FVIII were prepared asdescribed24.

FVIII-neutralizing activity. The FVIII-neutralizing activity of antibodiesagainst FVIII was determined by a published method25 and expressed asBethesda units (BU). BU were defined as the inverse of the concentration ofIgG that causes 50% inhibition of FVIII procoagulant activity. Residual FVIIIactivity was measured in an one-stage assay by determining the activatedpartial thromboplastin time using human plasma depleted of FVIII as sub-strate and human placental pathromtin® as activators. Heated plasma or im-munopurified IgG antibody against FVIII were incubated with pooledcitrated human plasma for 2 h at 37 ºC. The clotting times of four serial di-lutions of a reference plasma pool (Immuno AG, Wien, Austria) were com-pared with the clotting times of three dilutions of each sample. Dilutionswere made in Owren-Koller buffer (Diagnostica Stago, Asnieres, France).Inter-assay variation ranged between 1 and 2.5%.

Assay for hydrolysis of FVIII. Human recombinant FVIII (Hyland Baxter,Glendale, California) was labeled with 125I to a specific activity of 11.6nCi/µg, using the iodogen method. 125I-labeled FVIII (150 ng, except wherenoted otherwise was incubated 5 min–10 h at 38 ºC with the samples (50µg/ml IgG) in 50 µl of 50 mM Tris-HCl, pH 7.7, 100 mM glycine, 0.025%Tween-20 and 0.02% NaN3. Human monoclonal IgG antibody againstCMV M061 (a gift from M. Ohlin, Lund, Sweden) and normal unfraction-

Fig. 3 Hydrolysis of 125I-labeled FVIII by IgG antibodies against FVIII IgGin the presence of increasing amounts of unlabeled FVIII. 125I-labeled FVIII(1.5 ng) was incubated for 4 h with IgG antibodies against FVIII from pa-tient Wal in the presence of increasing concentrations of unlabeled FVIII.Factor VIII was then analyzed by 7.5% SDS–PAGE, and autoradiographswere scanned. a, Rate of hydrolysis of labeled FVIII, calculated by scan-ning FVIII bands of apparent molecular weight 200 and 300 kDa, whichare consistently hydrolyzed by IgG antibodies against FVIII. Data are rep-resentative of three experiments. Open circles, empirical data; curve, datafitted to the Michaelis–Menten equation (r2=0.998). b, Mean kinetic para-meters of the reaction.

b

a

Vmax (fmoles/min)

Catalytic constant (per min)

Catalytic efficiency (per M min)

FVII (nM)

Rate

of h

ydro

lysi

s (f

mol

es/m

in)

© 1999 Nature America Inc. • http://medicine.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m

NATURE MEDICINE • VOLUME 5 • NUMBER 9 • SEPTMEBER 1999 1047

ARTICLES

ated human polyclonal IgG (SandoglobulinR; Central Laboratory of theSwiss Red Cross, Bern, Switzerland) were used as negative controls.Samples were mixed 1:1 with Laemmli’s buffer without β-mercaptoethanoland were separated by SDS electrophoresis without being boiled; 20 µl ofeach sample was loaded per lane. Samples were separated by 7.5% and15% SDS–PAGE in nonreducing conditions at room temperature in a mini-PROTEAN II system at 25 mA/gel, until the dye front reached the bottom ofthe gel. The gels were then dried and protein bands were detected using X-OMAT AR. Autoradiographs were scanned to allow calculation of the rate ofhydrolysis of labeled FVIII. The data were fitted to the Michaelis–Mentenequation by means of the Sigma-plot program (Sigma).

For experiments assessing the effect of vWF on the cleavage of FVIII byIgG antibody against FVIII, purified vWF and mouse monoclonal antibody487 against vWF were a gift from J. P. Girma (Le Kremlin-Bicêtre, France).

Fast protein liquid chromatography gel filtration. Antibodies againstFVIII in 8 M urea were gel-filtered on a superose-12 column equilibratedwith 0.01% azide in PBS at a flow rate of 0.2 ml/min. Fractions (500 µleach) were collected and assayed for the presence of IgG by sandwichELISA and for FVIII proteolytic activity after ten-fold dilution.

Analysis of N-terminal sequences. Unlabeled recombinant FVIII (300 µg;Octocog Agfa; Bayer, Berkeley, California) was treated for 24 h at 38 ºCwith IgG antibody against FVIII from patient Wal (74 µg) in 1,500 µl of 50mM Tris-HCl, pH 7.7, 100 mM glycine, 0.025% Tween-20 and 0.02%NaN3. The resultant FVIII fragments were separated by 10% SDS–PAGE at50 mA in nonreducing conditions and were transferred for 2 h at 100 mAonto a Hybond-P PVDF membrane (Amersham) in 10 mM CAPS, 10%ethanol at pH 11.0. After being stained with Coomassie blue, visible bandswere cut and subjected to N-terminal sequencing, using an automatic pro-tein microsequencer Prosize 492 cLC (PE-Applied Biosystems, Foster City,California). The amount of protein sequenced ranged from 0.5 to 2 pmoles,depending on the fragment.

AcknowledgmentsThe authors thank Drs. J.-P. Bouvet, V. Frémaux-Bacchi and T. Croughs forsuggestions and J. Reinbolt and C. Lichté for help in microsequencing. Thiswork was supported by Institut National de la Santé et de la RechercheMédicale (INSERM) and Centre National de la Recherche Scientifique (CNRS),France and by the Bayer Pharma, France. S.L.D. is a recipient of a grant fromBayer Pharma. A.M. is a recipient of a fellowship from the Ministère de laRecherche et de la Technologie, France. A.P. is a recipient of a fellowship fromNATO.

RECEIVED 13 MAY; ACCEPTED 11 JUNE 1999

1. Ehrenforth, S. et al. Incidence of development of factor VIII and factor IX in-hibitors in haemophiliacs. Lancet 339, 594–598 (1992).

2. Saenko, E.L., Shima, M., Gilbert, G.E. & Scandella, D. Slowed release of throm-bin-cleaved factor VIII from von Willebrand factor by a monoclonal and a human

antibody is a novel mechanism for FVIII inhibition. J. Biol. Chem 271,27424–27431 (1996).

3. Arai, M., Scandella, D. & Hoyer, L.W. Molecular basis of factor VIII inhibition byhuman antibodies: Antibodies that bind to the factor VIII light chain prevent theinteraction of factor VIII with phospholipid. J. Clin. Invest. 83, 1978–1984(1989).

4. Zhong, D., Saenko, E.L., Shima, M., Felch, M. & Scandella, D. Some human in-hibitor antibodies interfere with factor VIII binding to Factor IX. Blood 92,136–142 (1998).

5. Neuenschwander, P.F. & Jesty, J. Thrombin-activated and factor Xa-activatedhuman factor VIII: differences in cofactor activity and decay rate. Arch. Biochem.Biophys. 296, 426–434 (1992).

6. Gilles, J.G.G., Arnout, J., Vermylen, J. & Saint-Remy, J.M. Anti-factor VIII antibod-ies of hemophiliac patients are frequently directed towards nonfunctional deter-minants and do not exhibit isotypic restriction. Blood 82, 2452–2461 (1993).

7. Rickard, K.A. Guidelines for therapy and optimal dosages of coagulation factorsfor treatement of bleeding and surgery in haemophilia. Haemophilia 1, 8–13(1995).

8. Li, L., Paul, S., Tyutyulkova, S., Kazatchkine, M.D. & Kaveri, S. Catalytic activityof anti-thyroglobulin antibodies. J. Immunol. 154, 3328–3332 (1995).

9. Gallacher, G. et al. A polyclonal antibody preparation with Michaelian catalyticproperties. Biochem. J. 279, 871–881 (1991).

10. Paul, S. et al. Characterization of thyroglobulin-directed and polyreactive cat-alytic antibodies in autoimmune disease. J Immunol 159, 1530–1536 (1997).

11. Eaton, D.L. et al. Construction and characterization of an active factor VIII vari-ant lacking the central one-third of the molecule. Biochemistry 25, 8343–8347(1986).

12. Lollar, P., Knutson, G.J. & Fass, D.N. Activation of porcine factor VIII:C by throm-bin and factor Xa. Biochemistry 24, 8056–8064 (1985).

13. Hill-Eubanks, D.C. & Lollar, P. von Willebrand factor is a cofactor for thrombin-catalyzed cleavage of factor VIII light chain. J. Biol. Chem. 265, 17854–17858(1990).

14. O’Brien, D.P., Johnson, D., Byfield, P. & Tuddenham, E.G.D. Inactivation of fac-tor VIII by factor IXa. Biochemistry 31, 2805–2812 (1992).

15. Lamphear, B.J. & Fay, P.J. Proteolytic interactions of factor IXa with human fac-tor VIII and factor VIIIa. Blood 80, 3120–3126 (1992).

16. Eaton, D., Rodriguez, H. & Vehar, G.A. Proteolytic processing of human factorVIII. Correlation of specific cleavages by thrombin, factor Xa, and activated pro-tein C with activation and inactivation of factor VIII coagulant activity.Biochemistry 25, 505–512 (1986).

17. Regan, L.M., Lamphear, B.J., Huggins, C.F., Walker, F.J. & Fay, P.J. factor IXa pro-tects factor VIIIa from activated protein C. J. Biol. Chem. 269, 9445–9452 (1994).

18. Koedam, J.A., Meijers, J.C., Sixma, J.J. & Bouma, B.N. Inactivation of human fac-tor VIII by activated protein C. Cofactor activity of protein S and protective effectof von Willebrand factor. J. Clin. Invest. 82, 1236–1243 (1988).

19. Wion, K.L., Kelly, D., Summerfield, J.A., Tuddenham, E.G. & Lawn, R.M.Distribution of factor VIII mRNA and antigen in human liver and other tissues.Nature 317, 726–729 (1985).

20. Leyte, A. et al. Sulfation of Tyr1680 of human blood coagulation factor VIII is es-sential for the interaction of factor VIII with von Willebrand factor. J. Biol. Chem266, 740–746 (1990).

21. Paul, S., Volle, D.J., Powell, M.J. & Massey, R.J. Site specificity of a catalytic va-soactive intestinal peptide antibody. J. Biol. Chem. 265, 11910–11913 (1990).

22. Shuster, A.M. et al. DNA hydrolyzing autoantibodies. Science 256, 665–667(1992).

23. Bronshtein, I.B. et al. DNA-specific antiidiotypic antibodies in the sera of patientswith autoimmune diseases. FEBS Lett. 314, 259–263 (1992).

24. Kaveri, S. et al. Antibodies to a conserved region of HLA class I molecules, capa-ble of modulating CD8 T cell-mediated function, are present in pooled normalimmunoglobulin for therapeutic use (IVIg). J. Clin. Invest. 97, 865–869 (1996).

25. Kasper, C.K., et al, A more uniform measurement of factor VIII inhibitors.Thromb. Diath. Haemorrh. 34, 869–872 (1975).

© 1999 Nature America Inc. • http://medicine.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m