Lupus anticoagulant activity of autoimmune antiphospholipid antibodies is dependent upon beta 2-glycoprotein I.

Abstract

It has been reported that antiphospholipid autoantibodies do not recognize phospholipid alone, but rather the plasma protein beta 2-glycoprotein I (beta 2GPI), or a beta 2GPI-phospholipid complex. In vitro beta 2GPI binds to anionic phospholipids and inhibits the prothrombinase activity of procoagulant membranes. In light of the fact that lupus anticoagulants, a type of antiphospholipid antibody, have similar anticoagulant properties, the relationship of beta 2GPI to lupus anticoagulant activity was investigated. IgG from patients with autoimmune diseases or syphilis were tested for anticardiolipin reactivity and lupus anticoagulant activity in the presence and absence of beta 2GPI. As expected, anti-cardiolipin reactivity associated with autoimmune disease was beta 2GPI dependent. In contrast, IgG from a patient with syphilis recognized cardiolipin alone and binding was inhibited by beta 2GPI. Autoimmune antiphospholipid antibodies prolonged the dilute Russell viper venom time of normal plasma, but had no effect on beta 2GPI-depleted plasma. Antiphospholipid antibodies associated with syphilis had no anticoagulant effect. RP-1, an anti-beta 2GPI mAb, had anticoagulant effects similar to those of autoimmune antiphospholipid antibodies. These data demonstrate that antiphospholipid autoantibodies exert lupus anticoagulant activity via an interaction with beta 2GPI. These antibodies and RP-1 appear to amplify the anticoagulant effect of beta 2GPI itself.

Full text

Page 1

Rapid Publication
Lupus Anticoagulant Activity of Autoimmune Antiphospholipid
Antibodies Is Dependent uponfl2-GlycoproteinI
Robert A. S. Roubey, * Charlotte W. Pratt,t Jill P. Buyon,I and John B. Winfield*
*Division ofRheumatology and Immunology, Department ofMedicine, Thurston Arthritis Research Center, and tDepartment of
Pathology, The University ofNorth Carolina at Chapel Hill, Chapel Hill, North Carolina 27599; and §Department of
Rheumatic Diseases, Hospitalfor Joint Diseases, New York University School ofMedicine, New York 10003
Abstract
It has been reported that antiphospholipid autoantibodies do
not recognize phospholipid alone, but rather the plasma protein
j2-glycoprotein I (,B2GPI), or a ,B2GPI-phospholipid complex.
In vitro #2GPI binds to anionic phospholipids and inhibits the
prothrombinase activity ofprocoagulant membranes. In light of
the fact that lupus anticoagulants, a type of antiphospholipid
antibody, have similar anticoagulant properties, the relation-
ship of,B2GPI to lupus anticoagulant activity was investigated.
IgG from patients with autoimmune diseases or syphilis were
tested for anticardiolipin reactivity and lupus anticoagulant ac-
tivity in the presence and absence of#2GPI. As expected, anti-
cardiolipin reactivity associated with autoimmune disease was
,B2GPI dependent. In contrast, IgG from a patient with syphilis
recognized cardiolipin alone and binding was inhibited by
,B2GPI. Autoimmune antiphospholipid antibodies prolonged
the dilute Russell viper venom time of normal plasma, but had
no effectonf2GPI-depleted plasma. Antiphospholipid antibod-
ies associated with syphilis had no anticoagulant effect. RP-1,
an anti-,B2GPI mAb, had anticoagulant effects similar to those
ofautoimmune antiphospholipid antibodies. Thesedatademon-
strate that antiphospholipid autoantibodies exert lupus antico-
agulant activity via an interaction with ,82GPI. These antibod-
ies and RP-1 appear to amplify the anticoagulant effect of
,B2GPI itself. (J. Clin. Invest. 1992. 90:1100-1104.) Key
words: apolipoprotein H-cardiolipin-phospholipid * systemic
lupus erythematosus-thrombosis
Introduction
Antiphospholipid antibodies (aPL)' occur in certain autoim-
mune diseases, particularly systemic lupus erythematosus
Address correspondence to Dr. R. A. S. Roubey, Division ofRheuma-
tology and Immunology, CB #7280, Room 932 FLOB, University of
North Carolina, Chapel Hill, NC 27599-7280.
Received for publication 30 March 1992 and in revisedform 21
May 1992.
1. Abbreviations used in this paper: aPL, antiphospholipid antibodies;
32GPI, ,32-glycoprotein I; dRVVT, dilute Russell viper venom time.
(SLE) and the primary aPL syndrome, as well as in syphilis and
other infectious diseases. Autoimmune aPL include the lupus
anticoagulant and are associated with thrombosis, recurrent
fetal loss, and thrombocytopenia (1). New insights into the
pathophysiology of autoimmune aPL have been provided by
the recent observation that these antibodies do not react with
phospholipid alone, but rather with the plasma protein 132-gly-
coprotein I(f32GPI,also designated apolipoprotein H) or with
a complex antigen comprised ofboth phospholipid and ,B2GPI
(2, 3).
While its physiological role is not known,fl2GPIbinds to
anionic phospholipids (4) and exhibits anticoagulant proper-
ties in vitro. For example, f32GPI has been shown to inhibit the
contact phase of intrinsic blood coagulation (5), ADP-depen-
dent platelet aggregation (6), and the prothrombinase activity
of platelets (7). ,B2GPI is thought to inhibit the prothrombin-
ase activity of platelets or phospholipid vesicles via binding to
anionic phospholipids expressed on these membranes (7).
f2GPI does not competitively inhibit binding of constituents
of the prothrombin activator complex, i.e., Factors Xa or Va,
but may decrease the number of functional prothrombinase
sites via a structural alteration ofthe phospholipid membrane
(7). Interestingly, lupus anticoagulants are also thought to in-
hibit prothrombinase activity by binding to the phospholipid
portion ofthe prothrombin activator complex (8).
Population and family studies demonstrate that the plasma
level of,2GPI,
genetic control (9-12). Although the in vitro data cited above
imply an in vivo anticoagulant role forf32GPI, no signs or
symptoms of disease have been reported in individuals with
inherited low or absent plasma levels (9). A small number of
patients with disseminated intravascular coagulopathy have
been reported to have very low ,B2GPI levels, although this
appeared to be an acquired phenomenon associated with the
coagulopathy (5). Slightly increased to normal levels of,2GPI
have been observed in a small group ofpatientswith SLE( 13).
The finding that ,2GPI constitutes at least part ofthe anti-
gen targeted by antiphospholipid antibodies suggests that this
protein may play a role in the pathogenesis oflupus anticoagu-
lant activity and the thrombotic diathesis associated with these
antibodies. In the current study aPL from patients with au-
toimmune disease or syphilis were characterized with regard to
their requirement for #2GPI in a modified anticardiolipin
ELISA and in a lupus anticoagulant assay, the dilute Russell
viper venom time (dRVVT). Additionally, the anticoagulant
properties of ,32GPI itself and of a murine mAb to (32GPI,
RP- 1, were investigated.
-200,g/mlin normal individuals, is under
1100
R. A. S. Roubey, C. W Pratt, J. P.Buyon,and J. B. Winfield
J. Clin. Invest.
© The American Society for Clinical Investigation, Inc.
0021-9738/92/09/1100/05
Volume 90, September 1992, 1100-1104
$2.00

Page 2

Methods
Patients and sera. Sera were obtained from two patients with the pri-
mary antiphospholipid antibody syndrome ( 14, 15), one patient with
SLE ( 16), and one patient with primary syphilis, all previously identi-
fied as having high titer anticardiolipin antibodies in a standard assay
( 17), and from a normal individual. Sera were stored at 4°C after the
addition of 0.02% sodium azide. IgG was purified from sera by protein
A-Sepharose 4B (Zymed Laboratories, South San Francisco, CA) col-
umn chromatography under low salt conditions as described ( 18). IgG
preparations contained < 100 ng/mlf32GPIby ELISA.
Purification off32GPI. ,B2GPI was purified from normal human
plasma as a by-product of the protein C inhibitor preparation ( 19).
Plasma was made 0.2 M in barium chloride, centrifuged at 4,000 g, and
the supernatant collected. Material precipitated by ammonium sulfate
(50-70% saturation) was collected by centrifugation, resuspended in
50 mM Tris, 100 mM NaCI, pH 7.0, and applied to a dextran sulfate-
agarose column. Proteins were eluted with a linear salt gradient ( 100-
600 mM NaCI). Fractions containing putative
by SDS-PAGE, pooled, dialyzed against 50 mM Tris, pH 9.0, applied
to DEAE-Sephacel (Pharmacia-LKB, Piscataway, NJ) and eluted with
50mM Tris-phosphoric acid, pH 6.0.f32GPIwas dialyzed againstHNP
buffer (20 mM Hepes, 150 mM NaCl, 0.1% polyethylene glycol, pH
7.4) and stored at -20°C. A single protein band was observed by SDS-
PAGE and Coomassie blue staining. The identity of the protein was
confirmed by NH2-terminal amino acid sequencing (12 cycles) (20)
performed by the Protein Chemistry Laboratory, The University of
North Carolina at Chapel Hill.
AMonoclonal antibody preparation. mAbs to ,B2GPI were produced
by immunizing BALB/c mice with a crude ,B2GPI preparation, boost-
ing twice, and performing the fusion as described by Harlow and Lane
(18) using X63Ag8.653 myeloma cells. Colonies were screened for
binding to solid-phase f2GPI in an ELISA. Positive clones were sub-
cloned and expanded. mAbs were purified from cell culture superna-
tants by affinity chromatography on immobilized protein A (Beckman
Instruments, Inc., Fullerton, CA) and stored in HNP buffer. The mAb
of highest affinity, RP- 1, was isotyped (HyClone Laboratories, Inc.,
Logan, UT) as IgG 1. RP-1 did not bind to solid-phase cardiolipin in
the absence of human
32GPI. Murine myeloma protein MOPC-21
(Organon Teknika, Durham, NC) and mAb 792 to complement com-
ponent C5 (provided by Dr. William Kolb, Quidel, San Diego, CA)
were used as isotype controls.
Modifiedanticardiolipin antibody ELISA. The ELISA for antibod-
ies to cardiolipin was performed with modifications similar to those of
McNeil et al. (3). The standard anticardiolipin ELISA ( 17) uses serum
samples, which contain endogenous human ,B2GPI (- 2 gg/ml, as-
suming sera are diluted
bovine f2GPI. In contrast, the modified assay uses purified IgG sam-
ples, containing < 100 ng/ml
32GPI, and 0.3% gelatin (Sigma Chemi-
cal Co., St. Louis, MO)/PBS as the blocking agent and sample diluent.
Neither FCS nor BSA were present in the assay to avoid contamination
with bovine
32GPI. Purified humanf32GPI, 30 ,ug/ml, was added to
the samples in certain experiments, as indicated.
(J2GPI ELISA. f2GPI was quantitated in a sandwich ELISA. Mi-
crotiter plates (ICN/Flow Laboratories, Inc., McLean, VA) were
coated with goat anti-human ,B2GPI IgG (Atlantic Antibodies, Scar-
borough, ME), 10 ,ug/ml in PBS, and blocked with 1% BSA/PBS.
Samples diluted in 1% BSA/PBS were applied to the wells and incu-
bated for 2 h at room temperature. The amount of,2GPI bound to the
wells was quantitated by the addition of RP-1, 2 ztg/ml, followed by
alkaline phosphatase-conjugated goat anti-mouse IgG (Zymed) and
p-nitrophenyl phosphate (Sigma). Optical density at 405 nm was mea-
sured with an Emax microplate reader and SOFTmax analysis software
(Molecular Devices Corp., Menlo Park, CA). Standard concentrations
of f2GPI, 0.2 ng/ml to
Preparation of J2GPI-depleted plasma. Anti-fl2GPI mAb RP-1
was coupled to Affi-Gel Hz hydrazide gel (Bio-Rad Laboratories, Rich-
mond, CA) according to the manufacturer's protocol. The RP-1 col-
32GPI were identified
1:100), and 10% FCS/PBS, which contains
1 ,ug/ml, were included on each plate.
umn was equilibrated with PBS and normal platelet-poor plasma was
infused at 0.1 ml/min. Peak effluent fractions were collected with min-
imal dilution, pooled, and stored at -20°C. 32GPI concentration of
depleted plasma was < 1jig/mlby ELISA.
Dilute Russell viper venom time. The effect of antibodies and
f2GPI on the dRVVT of platelet-poor plasma was assessed (21 ). 100
,il normal or J2GPI-depleted plasma, 50/ilThrombofax (Ortho Diag-
nostic Systems, Raritan, NJ), diluted 1:4, 50 ,ul Russell viper venom
(Wellcome Diagnostics, Dartford, UK), diluted 1:100, and 100jtl
HNP buffer containing antibodies and/or CaCl2 purified 32GPI were
incubated 30 s at 37°C. 100 ,l 25 mM CaC12 (Organon Teknika) was
added and the clotting time measured using a fibrometer (Fibrosystem;
BBL Microbiology Systems, Becton, Dickinson, and Co., Cockeysville,
MD). All assays were performed in duplicate (< 5% variability).
Results
In initial studies the anticardiolipin reactivity of purified IgG
from patients with the aPL syndrome, SLE, and syphilis was
characterized with respect to the requirement for#2GPI (Fig.
1). In the absence off2GPI, the binding ofIgG from patients
with the aPL syndrome (aPL-1 and aPL-2) and SLE did not
differ from the normal control, in agreement with previous
reports (2, 3). In contrast, IgG from a syphilitic patient recog-
nized cardiolipin in the absence of f2GPI. In the presence of
p2GPI, aPL-1 and aPL-2 IgGs bound strongly, and SLE IgG
bound moderately, to cardiolipin. The anticardiolipin reactiv-
ity of the syphilis IgG was partially inhibited by f32GPI. The
binding of normal IgG was higher in the absence of 32GPI,
implying that f2GPI is providing additional blocking of the
ELISA plate. (An alternative interpretation, that the higher
"background" ofnormal IgG represents low level specific bind-
ing that can be inhibited by f2GPI, cannot be excluded.) In no
instance was binding detected to solid-phase f2GPI alone (data
not shown). These data suggest that the "anticardiolipin" reac-
tivity ofIgG from patients with autoimmune disease is signifi-
cantly different from that ofpatients with syphilis. The former
apparently binds to a cardiolipin-f32GPI complex, while the
latter recognizes cardiolipin alone and is inhibited by 32GPI.
2.0
1.5-
0
0.5-
0.0
aPL-1aPL-2 SLE
IgG
Syph
Normal
Figure 1. Binding ofautoimmune aPL IgG to cardiolipin requires the
presence off2GPI. Purified IgG, 50,g/mlin 0.3% gelatin, from pa-
tients with the primary aPL syndrome (aPL-1 and aPL-2), SLE, pri-
mary syphilis (Syph), and from a normal individual were assayed.
Each IgG was tested in the absence (open bars) or presence (cross-
hatched bars) of 30tig/mlhumanfl2GPI.Results are the mean±SD
of three samples. o, No fl2GPI; *, + ,B2GPI.
Lupus Anticoagulant and f32-Glycoprotein I
1101

Page 3

In view ofthis differential reactivity with ,B2GPI, the lupus
anticoagulant activity of IgG from patients and controls was
assessed in a modified dRVVT assay. As shown in Fig. 2, IgG
from patients with the aPL syndrome and SLE demonstrated
lupus anticoagulant activity when added to normal plasma in a
dose-dependent fashion. As expected, IgG from syphilitic or
normal inclividuals did not exhibit an anticoagulant effect.
Thus, antibody reactivity with the f2GPI-cardiolipin complex,
but not reactivity with cardiolipin alone, was correlated with
anticoagulant activity.
To determine the role of,2GPI in the lupus anticoagulant
activity ofautoimmune aPL, the effect of these antibodies on
the dRVVT of normal plasma, f32GPI-depleted plasma, and
,B2GPI-depleted plasma partially reconstituted with f32GPI was
studied. The ,B2GPI dependence oflupus anticoagulant activity
is demonstrated in Fig. 3. Antibodies from patients with au-
toimmune disease prolonged the dRVVT of normal plasma
but had little or no anticoagulant effect when added to fl2GPI-
depleted plasma. Partial reconstitution ofdepleted plasmawith
purified human 32GPI restored the anticoagulant effect ofau-
toimmune aPL. Depletion of,2GPI from normal plasma did
not in itself significantly affect the dRVVT.
To investigate further the anticoagulant activity ofantibod-
ies reactive with j2GPI, RP-1, a murine mAb to ,B2GPI, was
studied. RP- 1 was found to have anticoagulant properties simi-
lar to those ofautoimmune aPL. RP- 1 prolonged the dRVVT
of normal plasma in a dose-dependent fashion (Fig. 4, solid
line). Isotype control myeloma protein MOPC-21 did not af-
fect the dRVVT (data not shown). The anticoagulant effect of
RP-1 wasdependent uponthepresence off2GPI. In a represen-
tative experiment RP-1, 62.5ugg/ml,prolonged the dRVVT of
normal plasma by 53% (41.5 s vs. 27.2 s), had no affect on the
dRVVT of,2GPI-depleted plasma (27.7 s vs. 27.3 s), and
prolonged the dRVVT of reconstituted plasma (f32GPI-de-
pleted plasma + 24 ,ug/ml fl2GPI) by 36% (36.9 s vs. 27.2 s).
While depletion of,2GPI from normal plasma did not signifi-
cantly affect the dRVVT, addition of supraphysiological con-
centrations of exogenous ,B2GPI prolonged the dRVVT in a
dose-dependent fashion, as shown in Fig. 4. Fig. 4 also demon-
40
38
36
tn34
z 32
30
28
42
40
38
4)-)36
U)
~34
VD.3
aPL-1
SLE
IgG
Normal
Figure 3. Anticoagulant activity of autoimmune aPL IgG is depen-
dent uponf#2GPI.The dRVVT assay was performed using normal
plasma (open bars), f2GPI-depleted plasma (single-hatched bars),
andfl2GPI-depletedplasma partially reconstituted with 40,tg/ml
,B2GPI (cross-hatched bars). Patient and normal IgG were tested at 1
mg/ml. Data shown are representative ofat least three independent
experiments. o, Normal plasma; m, depleted plasma; ., depleted
plasma +f32GPI.
strates that the anticoagulant effects of RP-1 and exogenous
f2GPI were approximately additive. Prolongation of the
dRVVT was not thought to be due to a nonspecific immune
complex effect as mAb 792 to another plasma protein, comple-
ment component C5, had no anticoagulant activity (data not
shown).
Discussion
These data demonstrate a critical role for p2GPI in the patho-
physiology oflupus anticoagulant activity. Antibodies from pa-
ci)
(0)
V-
0.25
0.50
0.75
1.00 1.25
IgG (mg/ml)
Figure 2. Prolongation of the dRVVT of normal plasma by autoim-
mune aPL IgG. Samples tested are the same as in Fig. 1. Data shown
are representative of at least two independent experiments.-
aPL; - -- m - --, aPL-2; ... A
lis;-*
-
* v ***, SLE;- - o --, primary syphi-
-o -* *-, normal.
0 20 40
60
RP-1 (pg/mi)
80
100
120
140
Figure 4. Effects ofmAb RP-l and exogenous fl2GPI on the dRVVT
ofnormal plasma. The solid line represents the effect ofmAb RP- I
on normal plasma without any exogenous,#2GPI.As indicated, ex-
ogenous ,B2GPI was added to normal plasma at final concentrations
of 35, 105, and 175 gg/ml. The final concentration of endogenous
f2GPI was
792 did not affect the dRVVT (not shown). Data shown are repre-
sentative ofat least two independent experiments. Exogenous#2GPI
Cg/ml):
175.
-50,g/ml.Isotypecontrol MPOC-2 1 and anti-C5 mAb
-, none;- -- a - --, 35;...
, 105; - - o -
1102
R. A. S. Roubey, C. W. Pratt, J. P.Buyon,and J. B. Winfield
32

Page 4

tients with the aPL syndrome orSLE reacted with a fl2GPI-car-
diolipin complex in ELISA and exhibited f32GPI-dependent
anticoagulant activity. In contrast, aPL associated with syphilis
reacted with cardiolipin and had no anticoagulant effect. Addi-
tionally, RP-1, ananti-fl2GPImAb, demonstrated lupus anti-
coagulant-like activity. These findings provide an explanation
for the clinical observation that autoimmune aPL, but not
those associated with syphilis, are associated with the lupus
anticoagulant (22). Although these experiments were not per-
formed using affinity-purified antibodies, the clear require-
ment for ,32GPI in both the modified anticardiolipin ELISA
and thedRVVT suggests that the antibodies possessing anticar-
diolipin activity and lupus anticoagulant activity are either the
same or very closely related. Previous data suggesting different
antibody specificities (23, 24) may need to be reinterpreted
with regard to the presence of,B2GPI in the respective assays. A
recent report by Bevers et al. (25) indicates that there may be
two types oflupus anticoagulants. In 11 of 16 patients studied,
incubation of plasma with liposomes, followed by centrifuga-
tion, reduced the anticardiolipin antibody titer of the plasma
supernatants without affecting lupus anticoagulant activity.
This anticoagulant activity was attributed to antibodies reac-
tive with lipid-bound prothrombin. In the remaining five pa-
tients, lupus anticoagulant activity co-sedimented with the an-
ticardiolipin antibodies. These patients may be similar to those
in our study, with f32GPI-dependent lupus anticoagulant ac-
tivity.
Based on our data, the anticoagulant effects of autoim-
mune aPL and anti-,B2GPI mAb are mediated by the interac-
tion ofthree molecular species: immunoglobulin, 32GPI, and
phospholipid. Autoimmune aPL and mAb RP- 1 appear to ex-
hibit lupus anticoagulant activity by amplifying the anticoagu-
lant activity of ,B2GPI. These antibodies may act either via a
steric effect, i.e., blocking phospholipid catalytic sites in addi-
tion to those blocked by ,B2GPI, or by increasing the affinity of
thef32GPI-phospholipidinteraction. The latter mechanism is
supported by the observation that "true" aPL associated with
syphilis bind phospholipids but do not act as anticoagulants. It
is therefore postulated thatautoimmune antiphospholipid anti-
bodies, or more precisely anti-,32GPI/phospholipid antibod-
ies, augment the interaction of fl2GPI with anionic phospho-
lipid surfaces, possibly by cross-linking bound f32GPI mole-
cules, further inhibiting phospholipid-dependent reactions.
Two alternative hypotheses are proposed to explain the
strong association of aPL and the lupus anticoagulant with
thrombosis in vivo. First, if one assumes that fl2GPI plays a
physiological role as an anticoagulant, then autoimmune aPL
could act to decrease the plasma level of,2GPI thereby causing
a thrombotic tendency. However, inherited low or absent
plasma levels of#2GPIhave not been associated with thrombo-
sis, and the patients with primary aPL syndrome and SLE re-
ported in this study had normal levels of#2GPI by ELISA
(data not shown). Even if autoimmune aPL do not lower
,B2GPI levels, these antibodies could theoretically inhibit
f2GPI function in vivo. Our in vitro data, however, suggest the
contrary, that autoimmune aPL enhance the function ,B2GPI.
A second hypothesis to explain the paradoxical in vivo pro-
coagulant effect ofautoimmune aPL is that 32GPI may affect
phospholipid-dependent reactions ofthe hemostatic system in
addition to theprothrombinase activity measured in lupus anti-
coagulant assays. For example, f2GPI may inhibit the throm-
bin/thrombomodulin-dependentactivation ofprotein C(26)
or the protein C/protein S inactivation ofFactors Va and VIIIa.
In such instances f32GPI would exert a procoagulant and/or
antifibrinolytic effect by binding to phospholipid surfaces, the
same mechanism by which it inhibits prothrombinase activity.
In the protein C pathway, enhancement of f2GPI-phospho-
lipid interactions by autoimmune aPL would favor thrombo-
sis. p2GPI, therefore, may inhibit both procoagulant and anti-
coagulant/fibrinolytic pathways, with the predominant clini-
cal effect of autoimmune aPL being on the latter pathway.
Inhibition of the protein C/protein S system by autoimmune
aPL has been proposed as a mechanism of thrombosis
(27-29).
The precise nature ofthe antigen targeted by autoimmune
aPL remains to be determined. These antibodies may recog-
nize an epitope comprised ofboth 32GPI and cardiolipin or a
cryptic epitope on one of the molecules expressed only when
the two are bound together. Nimpf et al. (7) studied the inter-
action off2GPI with phosphatidylcholine/phosphatidylserine
vesicles and observed structural alteration ofthe phospholipids
but not f32GPI. On the other hand, Galli et al. (2) reported that
anticardiolipin antibodies bound to solid-phase #2GPI alone
and speculated that the antigenic epitope on ,B2GPI might be
present only when the molecule is bound to phospholipid or
absorbed to plastic. Moreover, mAb RP-1, which mimics au-
toimmune aPL, recognizes ,B2GPI alone. Our data and that of
others (3, 30) indicate that autoimmune aPL bind to a ,B2GPI-
phospholipid complex and not to solid-phase ,B2GPI alone.
This discrepancy may be due to differences in immobilization
ofthe antigen, i.e., some methods may cause ,32GPI to assume
a configuration similar to that of phospholipid-bound 32GPI
whereasothersmay not, orto true differencesofantigenic speci-
ficity among the patients studied.
In summary, our data suggest that autoimmune aPL ex-
press in vitro lupus anticoagulant activity by enhancing the
interaction off2GPI with phospholipid. Further elucidation of
the physiological functions off2GPI should be helpful in un-
derstanding the pathophysiology ofthe clinical phenomena as-
sociated with "antiphospholipid" antibodies, e.g., thrombosis
and fetal loss.
Acknowledgments
The authors thank James Meade, Ph.D., of the Clinical Coagulation
Lab, University of North Carolina Hospitals, for technical assistance,
Dr. Jay Lozier for helpful suggestions, and Dr. Gilbert C. White for a
thoughtful critique ofthis manuscript.
This research was supported in partby National Institutes ofHealth
grants AR-30863, AR-7416, AR-30701, and HL-06350, and by a Bio-
medical Research Center grant from the Arthritis Foundation.
References
1. McNeil, H. P., C. N. Chesterman, and S. A. Krilis. 1991. Immunology and
clinical importance ofantiphospholipid antibodies. Adv. Immunol. 49:193-280.
2. Galli, M., P. Comfurius, C. Maassen, H. C. Hemker, M. H. De Baets,
P. J. C. van Breda-Vriesman, T. Barbui, R. F. A. Zwall, and E. M. Bevers. 1990.
Anticardiolipin antibodies directed not to cardiolipin but to a plasma protein
cofactor. Lancet. 335:1544-1547.
3. McNeil, H. P., R. J. Simpson, C. N. Chesterman, and S. A. Krilis. 1990.
Anti-phospholipid antibodies are directed against a complex antigen that in-
cludes alipid-binding inhibitor ofcoagulation: 82-glycoprotein I (apolipoprotein
H). Proc. Natl. Acad. Sci. USA. 87:4120-4124.
4. Wurm, H. 1984. ,32-glycoprotein I (apolipoprotein H) interactions with
phospholipid vesicles. Int. J. Biochem. 16:511-515.
5. Schousboe, I. 1985. &2-glycoprotein I: a plasma inhibitor of the contact
activation ofthe intrinsic blood coagulation pathway. Blood. 66:1086-1091.
Lupus Anticoagulant and/32-GlycoproteinI
1103

Page 5

6. Nimpf, J., H. Wurm, and G. M. Kostner. 1987. 132-glycoprotein-I (apo H)
inhibits the release reaction ofhuman platelets during ADP-induced aggregation.
Atherosclerosis. 63:109-114.
7. Nimpf, J., E. M. Bevers, P. H. H. Bomans, U. Till, H. Wurm, G. M.
Kostner, and R. F. A. Zwaal. 1986. Prothrombinase activity ofhuman platelets is
inhibited by B2-glycoprotein I. Biochim. Biophys. Acta. 884:142-149.
8. Shapiro, S., and P. Thiagarajan. 1982. Lupusanticoagulant. Prog. Hemosta-
sis Thromb. 6:263-285.
9. Haupt, H., H. G. Schwick, and K. Storiko. 1968. Uber einen erblichen
02-glycoprotein-I-mangel. Humangenetik. 5:291-293.
10. Cleve, H. 1968. Genetic studies on the deficiency of 32-glycoprotein I of
human serum. Humangenetik. 5:294-304.
11. Cleve, H., and C. Rittner. 1969. Further family studies on the genetic
control of (2-glycoprotein I concentration in human serum. Humangenetik.
7:93-97.
12. Koppe, A. L., H. Walter, V. P. Chopra, and M. Bajatzadeh. 1970. Investi-
gations on the genetics and population genetics ofthe /32-glycoprotein I polymor-
phism. Humangenetik. 9:164-171.
13. Cohnen, G. 1970. Immunochemical quantitation of02-glycoproteinI in
various diseases. J. Lab. Clin. Med. 75:212-216.
14. Asherson, R. A., M. A. Khamashta, J. Ordi-Ros, R. H. Derksen, S. J.
Machin, J. Barquinero, H. H. Outt, E. N. Harris, M. Vilardell-Torres, and G. R.
Hughes. 1989. The "primary: antiphospholipid syndrome: major clinical and
serological features. Medicine (Baltimore). 68:366-374.
15. Sammaritano, L. R., A. E. Gharavi, and M. D. Lockshin. 1990. Antiphos-
pholipid antibody syndrome: immunologic and clinical aspects. Semin. Arthritis
Rheum. 20:81-96.
16. Tan, E. M., A. S. Cohen, J. F. Fries, A. T. Masi, D. J. McShane, N. F.
Rothfield, J. G. Schaller, N. Talal, and R. J. Winchester. 1982. The 1982 revised
criteria for the classification of systemic lupus erythematosus. Arthritis Rheum.
25: 127 1-1277.
17. Gharavi, A. E., E. N. Harris, R. A. Asherson, and G. R. V. Hughes. 1987.
Anticardiolipin antibodies: isotype distribution and phospholipid specificity.
Ann. Rheum. Dis. 46:1-6.
18. Harlow, E., and D. Lane. 1988. Antibodies: A Laboratory Manual. Cold
Spring Harbor Laboratory, Cold Spring Harbor, NY. 726 pp.
19. Pratt, C. W., B. G. Macik, and F. C. Church. 1989. Protein C inhibitor:
purification and proteinase activity. Thromb. Res. 53:595-602.
20. Lozier, J., N. Takahashi, and F. W. Putnam. 1984. Complete amino acid
sequence of human plasma j32-glycoprotein
8 1:3640-3644.
21. Thiagarajan, P., V. Pengo, and S. S. Shapiro. 1988. The use ofthe dilute
Russell viper venom time for the diagnosis of lupus anticoagulants. Blood.
68:869-874.
22. Johansson, A. E., and A. Lassus. 1974. The occurrences of circulating
anticoagulants in patients with syphilitic and biologically false positive antilipoi-
dal antibodies. Ann. Clin. Res. 6:105-108.
23. Exner, T., N. Sahman, and B. Trudinger. 1988. Separation ofanticardioli-
pin antibodies from lupus anticoagulant on a phospholipid-coated polystyrene
column. Biochem. Biophys. Res. Commun. 155:1001-1007.
24. McNeil, H. P., C. N. Chesterman, and S. A. Krilis. 1989. Anticardiolipin
antibodies and lupus anticoagulants comprise antibody subgroups with different
phospholipid binding characteristics. Br. J. Haematol. 73:506-513.
25. Bevers, E. M., M. Galli, T. Barbui, P. Comfurius, and R. F. A. Zwaal.
1991. Lupus anticoagulant IgG's(LA) are notdirected to phospholipids only, but
to a complex of lipid-bound human prothrombin. Thromb. Haemostasis.
66:629-632.
26. Keeling, D. M., A. J. G. Wilson, 1. M. Mackie, D. A. Isenberg, and S. J.
Machin. 1991. 132-glycoprotein I inhibits the thrombin/thrombomodulin depen-
dent activation ofprotein C. Blood. 78:184a. (Abstr.).
27. Malia, R. G., S. Kitchen, M. Greaves, and F. E. Preston. 1990. Inhibition
ofactivated protein C and its cofactor protein S by antiphospholipid antibodies.
Br. J. Haematol. 76:101-107.
28. Lo, S. C., H. H. Salem, M. A. Howard, M. J. Oldmeadow, and B. G.
Firkin. 1990. Studies of natural anticoagulant proteins and anticardiolipin anti-
bodies in patients with the lupus anticoagulant. Br. J. Haematol. 76:380-386.
29. Marciniak, E., and E. H. Romond. 1989. Impaired catalytic function of
activated protein C: a new in vitro manifestation oflupus anticoagulant. Blood.
74:2426-2432.
30. Sammaritano, L. R., M. D. Lockshin, and A. E. Gharavi. 199 1. Antiphos-
pholipid antibodies from different patients differ in aPL cofactor requirement.
Arthritis Rheum. 34:S42. (Abstr.).
I. Proc. Natl. Acad. Sci. USA.
1104
R. A. S. Roubey, C. W Pratt, J. P.Buyon,and J. B. Winfield