Antiphospholipid antibodies (apl) are heterogeneous autoantibodies

Transcription

1 Anticardiolipin Antibodies From Patients With the Antiphospholipid Antibody Syndrome Recognize Epitopes in Both 2 -Glycoprotein 1 and Oxidized Low-Density Lipoprotein Sohvi Hörkkö, MD, PhD; Tsaiwei Olee, PhD; Lian Mo, PhD; D. Ware Branch, MD; Virgil L. Woods, Jr, MD; Wulf Palinski, MD; Pojen P. Chen, PhD; Joseph L. Witztum, MD Background We recently suggested that many anticardiolipin antibodies bind only to oxidized cardiolipin (OxCL) and/or to OxCL 2 -glycoprotein 1 ( 2 GP1) adducts but not to a reduced cardiolipin that is unable to undergo oxidation. To test this hypothesis, we investigated 24 sera, 4 protein A purified IgG fractions, and 3 human monoclonal antibodies that were all isolated from patients with antiphospholipid antibody syndrome (APS); testing was also performed in 7 controls. Two monoclonal antibodies (IS3 and IS4) were selected for binding to CL and one was selected for binding to 2 GP1 (LJB8). Methods and Results By chemiluminescent immunoassay, all APS sera samples bound only to OxCL and not to reduced CL, and the binding was inhibited 95% by OxCL but not reduced CL. All purified IgG fractions bound to 2 GP1 but only when the 2 GP1 was plated on microtiter wells coated with OxCL. All 3 monoclonal antibodies bound only to OxCL. On Western blots, IS4 and LJB8 bound to 2 GP1 as well as to delipidated apob of oxidized LDL but not to native apob. IS3 also bound to oxidized apob on Western blot. Covalent modification of 2 GP1 with oxidation products of CL made it more antigenic for APS serum samples, for purified IgG fractions, and for the monoclonal antibodies. Conclusions These data support the hypothesis that oxidation of CL is needed to generate epitopes for many anticardiolipin antibodies and that some of these epitopes are covalent adducts of OxCL with 2 GP1 or apob. (Circulation. 2001;103: ) Key Words: antibodies, antiphospholipid lipoproteins autoimmunity Antiphospholipid antibodies (apl) are heterogeneous autoantibodies detected in vitro by solid-phase immunoassays as antibodies binding to phospholipids such as cardiolipin (CL) or to complexes of phospholipid and 2 - glycoprotein 1 ( 2 GP1). 1 6 Patients with high levels of apl are prone to fetal loss, autoimmune thrombocytopenia, and thrombotic events in either the venous or arterial circulation. 4 7 Elevated levels of apl in combination with one or more clinical features has been termed the antiphospholipid antibody syndrome (APS). The exact nature of the epitope(s) for anticardiolipin antibodies (acl) has been controversial. 4 6 We recently demonstrated that CL is rapidly oxidized when plated on microtiter wells and exposed to air as is done in conventional solid-phase acl immunoassays. 8 We also showed that a few selected reference sera and affinity-purified acl-igg from APS patients progressively bound to CL as it was oxidized but not to a reduced CL (CLred) analogue that was unable to undergo lipid peroxidation (all 4 unsaturated fatty acids in CLred are hydrogenated to saturated fatty acids). We proposed that acl bind to epitopes generated when CL undergoes oxidation. 8,9 2 GP1 is a phospholipid-binding apolipoprotein (also called apolipoprotein H) that seems to be necessary for the binding of some acl. 2,3 It has been proposed that, as a result of noncovalent protein-lipid interactions, novel, conformational epitopes are created on the plated CL, on 2 GP1, or on an admixture of these two, or that 2 GP1 alone is the target antigen. 3 5,10,11 We demonstrated that some apl bind to proteins like 2 GP1 only as a consequence of covalent adduct formation between oxidized phospholipids (OxPL) and the protein. 9 The formation of neoepitopes between OxPL and associated proteins would be analogous to LDL oxidation, which generates immunogenic neoepitopes. 8,12,13 Autoantibodies to OxLDL are present in the sera of animals and humans and are increased in those with increased atherosclerosis. 14,15 Indeed, acl in patients with systemic lupus erythematosus cross-react with oxidized LDL (OxLDL). 16 Received July 12, 2000; revision received October 9, 2000; accepted October 16, From the Department of Medicine, University of California, San Diego, and the Department of Obstetrics and Gynecology, University of Utah, Health Sciences Center, Salt Lake City (D.W.B.). Correspondence to Joseph L. Witztum, MD, or Sohvi Hörkkö, MD, PhD, Dept of Medicine, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA jwitztum@ucsd.edu or shorkko@ucsd.edu 2001 American Heart Association, Inc. Circulation is available at 941

2 942 Circulation February 20, 2001 In the present article, we demonstrate that oxidized CL (OxCL) and covalent adducts of OxCL with 2 GP1 are epitopes for many acl. In addition, we show that native 2 GP1 and OxLDL share common epitopes recognized by monoclonal antibodies cloned from APS patients. Methods Materials CL (diphosphatidylglycerol, bovine heart) containing 4 unsaturated fatty acids and hydrogenated CL (CLred) containing 4 saturated fatty acids were obtained from Avanti Polar Lipids. Fatty acid analysis confirmed that linoleic acid accounted for 92% of the fatty acids of CL, whereas all fatty acids in CLred were saturated (18:0). 8 CL was oxidized by air exposure to generate various decomposition products, as previously descibed. 8 Human 2 GP1 was purified as previously described. 17 Human Subjects Serum samples from 21 women and 1 man with APS and from 7 healthy controls were collected at the Department of Obstetrics and Gynecology of the University of Utah Hospital. Two patients had donated a serum sample at 2 different times, 7 and 8 years apart, for a total of 24 serum samples from the APS patients. APS patients had 1 of the following clinical features: (1) a history of either 1 fetal deaths or 3 consecutive pregnancy losses, (2) venous or arterial thrombosis, or (3) autoimmune thrombocytopenia. Six patients had systemic lupus erythematosus. All patients had acl IgG, as measured by a standardized assay 18 in Dr Branch s laboratory (19 samples were 100 IgG phospholipid-binding units and 6 samples were 30, 40, 44, 47, 62, and 75 IgG phospholipid-binding units, respectively). Chemiluminescent Immunoassay for Antibody Binding CL or CLred in 100% ethanol was added at 25 g/ml into white, round-bottomed High Binding Microfluor (Dynex) microtitration plates and exposed to air for the indicated time at room temperature to induce oxidation. Absolute ethanol was added to blank wells. The wells were washed with PBS buffer containing 0.27 mmol/l EDTA and blocked with 10% fetal bovine serum, 1% bovine serum albumin (BSA), or 0.25% gelatin in indicated experiments. The primary antibodies were incubated for 1 hour, and the amount of antibody bound was measured with alkaline phosphatase-labeled goat antihuman IgG (Sigma) using LumiPhos 530 (Lumigen) as the substrate. Luminescence was measured in relative light units (RLU) with a Dynex Luminometer (Dynex Technologies). 8,9 Each point in each of the figures is the mean of triplicate determinations. Preparation of OxCL- 2 GP1 CL was dried and exposed to air for 3 hours. Purified human 2 GP1 (in PBS and 20 mol/l EDTA) and NaCNBH 3 (10 mmol/l) were added and incubated at 37 C for 6 hours. After incubation, 80 mmol/l octylglucoside was added and dialyzed against PBS to remove the unbound CL. Protein A Purification of IgG Whole IgG fractions were purified using ImmunoPure Plus Immobilized Protein A IgG Purification Kit (Pierce). The absence of 2 GP1 in the IgG fractions was verified by a capture assay. Samples were incubated in wells coated with polyclonal goat anti-human 2 GP1 antibody (10 g/ml; Enzyme Research Laboratories) and by detecting the amount of 2 GP1 captured with the biotinylated anti-human 2 GP1 antibody and alkaline phosphatase labeled avidin (Pierce). Figure 1. IgG binding of 24 APS serum samples (from 22 patients) and 7 control samples to OxCL and CLred. Samples were diluted 1:50 with 10% adult bovine serum buffer. Bound IgG was detected with alkaline phosphatase labeled goat anti human-igg antibody. Generation of Monoclonal IgG acl Antibodies Human monoclonal IgG antibodies IS3 and IS4 were generated from a patient with primary APS, as recently described, 17,19 by selecting for their ability to bind to CL. Antibody LJB8 (not previously described) was cloned from another APS patient in a similar manner, but it was selected for binding to 2 GP1. The absence of bovine 2 GP1 in the monoclonal antibody preparations was verified by SDS-PAGE gel electrophoresis and silver staining. Western Blot Proteins were electrophoresed on SDS 4% to 12% trisglycine polyacrylamide gels and electrotransferred to nitrocellulose membranes. Transfer was confirmed with 0.1% Ponceau S (Sigma) staining, and the membranes were blocked with Super Block (Pierce) or 0.25% gelatin and immunostained with either the human monoclonal antibodies or goat anti-human 2 GP1 antibody. Antibody binding was detected by appropriate alkaline phosphatase labeled secondary antibodies and visualized with Alkaline Phosphatase Conjugate Substrate Kit (Bio-Rad). Results Antibody Binding to OxCL and CLred Figure 1 demonstrates that IgG binding was substantially higher to OxCL than to CLred for each of the APS serum samples. Control samples had low binding to both antigens. When the assay was repeated in the absence of 10% bovine serum, virtually identical results were observed (data not shown). Because the acl binding has been suggested to depend exclusively on the presence of 2 GP1, we tested whether 2 GP1 binds to both OxCL and CLred to the same extent. Figure 2A shows an example in which there is increased IgG binding from one APS serum sample to the OxCL but no binding to CLred. In the same experiment, we used an anti- 2 GP1 antibody to demonstrate that substantial amounts of human 2 GP1 (from the added serum) bound to both OxCL and CLred when measured in parallel wells under identical conditions (Figure 2A). This suggests that the binding of acl to phospholipids does not depend exclusively on the presence of 2 GP1.

3 Hörkkö et al acl Recognize 2 GP1 and OxLDL 943 Figure 2. A, IgG binding from 1 APS serum sample (1:50 dilution in 1% BSA-PBS) to OxCL ( ) and CLred (Œ) plated on microtiter wells. Amount of 2 GP1 bound to OxCL ( ) and CLred ( ) in parallel wells using goat anti-human 2 GP1 antibody is also shown. B, Competition immunoassay of IgG binding from 1 APS serum sample. Serum sample (1:50 in 1% BSA-PBS) was preincubated for 60 minutes with increasing amounts of OxCL ( ) or CLred (Œ), and then immune complexes were pelleted by centrifugation ( rpm for 30 minutes) and the supernatants were tested for IgG binding to OxCL. Amount of 2 GP1 remaining in supernatant after absorption with OxCL ( ) or CLred ( ) is also shown. Samples were incubated in microtiter wells coated with anti-human 2 GP1 antibody, and amount of 2 GP1 captured was measured with biotinylated anti-human 2 GP1 antibody. Results are expressed as percentage remaining after absorption when compared with a control incubation performed under same conditions in absence of any phospholipid. Specificity of acl IgG Binding The serum samples were preincubated with OxCL or CLred, and the supernatants were then tested for binding to OxCL. Figure 2B shows that the preincubation of 1 APS serum sample with OxCL but not CLred removed 95% of the original acl binding. Using a capture assay (see Methods) we found that 20% of the total 2 GP1 content was absorbed from the 1:50 dilution of serum during the preincubation with OxCL or CLred (Figure 2B). Figure 3 demonstrates that preincubation of all APS serum samples with OxCL removed 95% of the IgG binding to OxCL. A control incubation for each sample without phospholipid did not remove any IgG binding to OxCL (data not shown). To examine whether a population of antibodies binding to OxCL could also bind to another OxPL epitope, we tested the IgG binding to copper-oxidized LDL (CuOx- LDL). Figure 3 also shows that preincubation with OxCL absorbed 40% of the binding to CuOx-LDL. Preincubation with CLred did not remove any IgG binding to either OxCL or CuOx-LDL. Using similar competition assays, we demonstrated that even the slight degree of binding to CLred observed with a few of the APS samples (Figure 1) was nonspecific (data not shown). Binding of Human Monoclonal IgG acl Antibodies Figure 4 shows that monoclonal antibodies IS4 (selected for binding to CL) and LJB8 (selected for binding to 2 GP1) had high binding to OxCL but not to CLred. Both of these monoclonals also bound to CuOx-LDL and to another model epitope, malondialdehyde-modified LDL (MDA-LDL) but not native LDL. Monoclonal antibody IS3 (selected for Figure 3. Absorption immunoassay for IgG binding of 24 APS serum samples. Each serum sample was diluted 1:50 with 2% BSA triethanolamine-buffered saline and preincubated with 25 g of OxCL or CLred. Immune complexes were pelleted by centrifugation, and supernatants were tested for IgG binding to OxCL or CuOx-LDL. Results are expressed as percentage remaining after absorption when compared with control incubations performed under same conditions in absence of phospholipid. binding to CL) also showed identical results (data not shown). To test if these monoclonals specifically recognized lipidprotein adducts, we performed Western blot analyses. Figure 5A shows that IS4 bound to the protein of both MDA-LDL (lane E) and CuOx-LDL (lane F), but not to native LDL (lane D) or BSA (lane B). In addition, IS4 showed strong binding to 2 GP1 (lane C). Figure 5A also demonstrates the absence of human 2 GP1 on native-ldl (lane I) or CuOx-LDL (lane J). Figure 5B shows that LJB8 bound to 2 GP1 (lane B), MDA-LDL (lane D), and CuOx-LDL (lane E) but not to native-ldl (lane C) or BSA (lane F). IS3 (Figure 5B) also showed binding to MDA-LDL (lane I) and CuOx-LDL (lane Figure 4. Binding of human monoclonal antibodies from APS patients to OxCL ( ), CLred ( ), MDA-LDL (ƒ), CuOx-LDL ( ), or native LDL ( ). IS4 was originally selected for binding to CL, and LJB8 was selected for binding to 2 GP1. Antigen-coated wells were blocked with 0.25% gelatin, and monoclonal antibodies (5 g/ml) were diluted in 0.25% gelatin. Amount of bound IgG was detected with alkaline phosphatase labeled anti-human IgG antibody.

4 944 Circulation February 20, 2001 Figure 6. Immunoassay of protein A purified IgG fractions from sera of 4 APS patients (APS-IgG #1 through #4) and 1 control subject (human IgG). Wells were coated with OxCL or CLred (25 g/ml) or ethanol only. After 4 washes, wells were incubated with 2GP1 (10 g/ml in PBS buffer) or 1% BSA-PBS buffer for 1 hour at room temperature, and amount of IgG or 2GP1 bound to wells was measured with appropriate antibodies. Figure 5. A, Western blots of binding of human monoclonal antibody (IS4) originally selected for binding to CL (lanes A through F) and binding of goat anti-human 2GP1 antibody (lanes G through J). Lane A, standard; lane B, BSA; lane C, 2GP1; lane D, native LDL; lane E, MDA-LDL; lane F, CuOxLDL; lane G, BSA; lane H, human 2GP1; lane I, native LDL; and lane J, CuOx-LDL. B, Western blots of binding of monoclonal antibody (LJB8) originally selected for binding to 2GP1 (lanes A through F) and monoclonal antibody (IS3) originally selected for binding to CL (lanes G through J). Lane A, standard; lane B, 2GP1; lane C, native LDL; lane D, MDA-LDL; lane E, CuOxLDL; lane F, BSA; lane G, 2GP1; lane H, native LDL; lane I, MDA-LDL; and lane J, CuOx-LDL. acl. To further test this hypothesis, we prepared covalent adducts between OxCL and 2GP1 (OxCL- 2GP1). Figure 7 demonstrates that although most of the APS serum samples showed increased binding to the native unmodified 2GP1 compared with BSA (mean, versus 6445 RLU/100 ms, respectively; P when using Student s paired t test), the binding to OxCL- 2GP1 (mean, RLU/100 ms) increased 4-fold (P 0.001). The control samples showed very little binding to either native 2GP1 or OxCL- 2GP1 (Figure 7). There was a positive correlation between the measurements of IgG binding to OxCL and to OxCL- 2GP1 among the APS samples (r 0.84, P , linear regression analysis). The protein A purified IgG fractions IS4 and LJB8 also had increased binding to OxCL- 2GP1 compared with the native 2GP1 (data not shown). J) but not to native LDL (lane H) or to 2GP1 (lane G). These data clearly demonstrate that IS4 and LJB8 recognize similar oxidatively-modified protein moieties of CuOx-LDL, MDALDL, and 2GP1 but that IS3 seems to recognize a different epitope. Antibody Binding to Oxidatively Modified Human 2GP1 To test whether OxCL on microtiter wells could modify 2GP1 and create epitopes for acl antibodies, we first dried OxCL, CLred, or solvent only (ethanol) in microtiter wells before adding 2GP1. Figure 6 demonstrates that protein A purified IgG fractions isolated from different APS patients bound to the 2GP1 added to OxCL-coated wells but not to the 2GP1 added to CLred- or ethanol-coated wells. Note that equal amounts of 2GP1 were present in the wells under different conditions. Under these conditions, these purified IgGs did not bind to OxCL alone (eg, BSA wells). The data in Figure 6 strongly suggest that OxCL can modify 2GP1 in such a way that it forms epitopes for some Figure 7. Binding of IgG from 24 APS serum samples and 7 control samples to BSA, 2GP1, or 2GP1 covalently modified with CL oxidation products (OxCL- 2GP1). Serum samples were diluted 1:50 with 2% BSA triethanolamine-buffered saline, and bound IgG was detected with alkaline phosphatase labeled anti-human IgG antibody.

5 Hörkkö et al acl Recognize 2 GP1 and OxLDL 945 Discussion On the basis of previous observations, 8,9 we proposed that some acl bind to neoepitopes of OxPL or to neoepitopes generated by adduct formation between reactive breakdown products of OxPL and associated proteins. We now demonstrate that most acl in the tested APS sera require CL oxidation for the CL to be an antigen. Furthermore, our data using purified IgG fractions indicate that CL oxidation is required even in the presence of 2 GP1 (ie, 2 GP1 alone or bound to CLred does not yield epitopes). Two monoclonal antibodies from two APS patients (IS4 and LBJ8) bound to both 2 GP1 and to the apob of OxLDL but not to native apob, implying similar oxidized lipid-protein epitopes on these different proteins. Antibody IS3 bound to both OxCL and oxidized apob but not to 2 GP1, implying variations in oxidized lipid-protein epitopes. We postulate that a large number of different lipid-protein and even lipid-lipid adducts could form when phospholipids undergo oxidation. This would be analogous to the adduct formation between oxidized lipids and apob that occurs during LDL oxidation. 20,21 Indeed, the recent observation that CL is found in the circulation 22 suggests that OxCL-apoB adducts form on OxLDL. 2 GP1 has been reported to be the primary serum cofactor or target antigen for many acl. 2 5,10,11 We and others have reported that other proteins, such as polylysine, LDL, and apoai, also have cofactor activity. 9,23 2 GP1 seems to be an excellent cofactor because of its high avidity to phospholipids and its ability to form adducts with OxCL (and possibly other phospholipids 9 ). To test this idea, we demonstrated that the binding of the purified IgG fractions to 2 GP1 occurred only when the 2 GP1 was plated with OxCL (Figure 6). Also, we demonstrated that APS sera, purified IgG fractions, and human monoclonals all showed strong binding to the OxCL- 2 GP1 adduct. Furthermore, among all APS sera, the binding to OxCL- 2 GP1 correlated well with the binding to OxCL. These data suggest that not only does 2 GP1 readily form adducts with OxPL in the microtiter wells, but it may already contain some oxidized lipid-protein epitopes. In support of this, we demonstrated that monoclonal antibodies cloned from APS patients recognized epitopes not only on native 2 GP1 but also on oxidatively modified apob on Western blots. These data strongly indicate that these epitopes are covalently oxidized lipid-protein adducts on 2 GP1 and apob. In the case of 2 GP1, this may occur in plasma in vivo or during the isolation procedure ( 2 GP1 is often isolated with a method involving perchloric acid precipitation that generates a strong pro-oxidant condition 24 ). There has been considerable difficulty in generating reliable and reproducible clinical assays for measuring acl, 25,26 which we believe is partly because of the oxidation of CL. In fact, there is variability not only between different preparations but even in the same CL preparation depending on its age (even if stored at 70 C under argon). Because the rate of CL oxidation is extremely difficult to control or standardize, an alternative approach might be to use an adduct between OxCL and 2 GP1 (or other protein) as an antigen. If many apl are, in fact, directed against oxidationdependent epitopes and because many OxPL products can form, it is likely that some apl are against unique oxidationspecific structures whereas other apl are against more common oxidation-dependent structures. These data imply heterogeneity even among apl to OxCL epitopes. Moreover, our data do not address the observations that there are many oxidation-independent antibodies that bind exclusively to conformational changes or even primary sequences of 2 GP1, independent of any bound lipid or lipid- 2 GP1 adducts. 3,4,10,11,27,28 Furthermore, there may also be antibodies against conformational changes in either CL, 2,29 2 GP1, 30 or prothrombin. 31 There is controversy over which type of antibodies are best associated with various aspects of clinical disease. 4 6 Knowledge that many apl can be oxidation-dependent may give insight into some of the pathogenic events underlying the clinical manifestations of APS. These data suggest that inflammatory conditions and an attendant pro-oxidant state are associated with the generation of epitopes to many apl. Recently, Iuliano and colleagues 32 tested this hypothesis and reported a strong correlation between acl and lipid peroxidation (isoprostane excretion) in patients with systemic lupus erythematosus; treatment with vitamin E led to a reduction in isoprostane excretion. 33 If antiphospholipid antibodies are indeed pathogenic or even simply a marker of enhanced lipid peroxidation, then therapies aimed at the underlying inflammation and, in particular, at ameliorating the pro-oxidant state may be beneficial. Acknowledgments These studies were supported by NIH grants HL57505 (to J.L.W.), HL56989 (La Jolla SCOR), and AR42506 (to P.P.C.) and by a Postdoctoral Research Fellowship from the American Heart Association, California Affiliate (to S.H.). References 1. Harris EN, Gharavi AE, Boey ML, et al. Anticardiolipin antibodies: detection by radioimmunoassay and association with thrombosis in systemic lupus erythematosus. Lancet. 1983;2: McNeil HP, Simpson RJ, Chesterman CN, et al. Anti-phospholipid antibodies are directed against a complex antigen that includes a lipid-binding inhibitor of coagulation: beta 2-glycoprotein I (apolipoprotein H). Proc Natl Acad Sci U S A. 1990;87: Galli M, Comfurius P, Maassen C, et al. Anticardiolipin antibodies (ACA) directed not to cardiolipin but to a plasma protein cofactor. Lancet. 1990;335: Kandiah DA, Sali A, Sheng Y, et al. Current insights into the antiphospholipid syndrome: clinical, immunological, and molecular aspects. Adv Immunol. 1998;70: Roubey RA, Hoffman M. From antiphospholipid syndrome to antibodymediated thrombosis. Lancet. 1997;350: Harris EN, Pierangeli SS, Gharavi AE. Diagnosis of the antiphospholipid syndrome: a proposal for use of laboratory tests. Lupus. 1998;7(suppl 2):S144 S Silver RM, Draper ML, Scott JR, et al. Clinical consequences of antiphospholipid antibodies: a historic cohort study. Obstet Gynecol. 1994;83: Hörkkö S, Miller E, Dudl E, et al. Antiphospholipid antibodies are directed against epitopes of oxidized phospholipids: recognition of cardiolipin by monoclonal antibodies to epitopes of oxidized low-density lipoprotein. J Clin Invest. 1996;98: Hörkkö S, Miller E, Branch DW, et al. The epitopes for some antiphospholipid antibodies are adducts of oxidized phospholipid and beta-2 glycoprotein 1 (and other proteins). Proc Natl Acad Sci U S A. 1997;94:

6 946 Circulation February 20, Matsuura E, Igarashi Y, Fujimoto M, et al. Heterogeneity of anticardiolipin antibodies defined by the anticardiolipin cofactor. J Immunol. 1992;148: Keeling DM, Wilson AJ, Mackie IJ, et al. Some antiphospholipid antibodies bind to beta 2-glycoprotein I in the absence of phospholipid. Br J Haematol. 1992;82: Palinski W, Ylä-Herttuala S, Rosenfeld ME, et al. Antisera and monoclonal antibodies specific for epitopes generated during oxidative modification of low density lipoprotein. Arteriosclerosis. 1990;10: Palinski W, Hörkkö S, Miller E, et al. Cloning of monoclonal autoantibodies to epitopes of oxidized lipoproteins from apo E-deficient mice: demonstration of epitopes of oxidized LDL in human plasma. J Clin Invest. 1996;98: Witztum JL, Palinski W. Autoimmunity to oxidized lipoproteins. In: Hansson G, Libby P, eds. Immune Functions of the Vessel Wall. Amsterdam: Harwood Academic Publishers; 1996: Ylä-Herttuala S. Is oxidized low-density lipoprotein present in vivo? Curr Opin Lipidol. 1998;9: Vaarala O, Alfthan G, Jauhiainen M, et al. Crossreaction between antibodies to oxidised low-density lipoprotein and to cardiolipin in systemic lupus erythematosus. Lancet. 1993;341: Olee T, Pierangeli SS, Handley HH, et al. A monoclonal IgG anticardiolipin antibody from a patient with the antiphospholipid syndrome is thrombogenic in mice. Proc Natl Acad Sci U S A. 1996;93: Harris EN, Pierangeli SS, Birch D. Anticardiolipin Wet Workshop report: fifth international symposium on antiphospholipid antibodies. Am J Clin Pathol. 1994;101: Zhu M, Olee T, Le DT, et al. Characterization of IgG monoclonal anti-cardiolipin/anti- 2GP1 antibodies from two patients with antiphospholipid syndrome reveals three species of antibodies. Br J Haematol. 1999;105: Gillotte K, Hörkkö S, Witztum JL, et al. Oxidized phospholipids, linked to apolipoprotein B of oxidized LDL, are ligands for macrophage scavenger receptors. J Lipid Res. 2000;41: Steinberg D, Witztum JL. Lipoproteins, lipoprotein oxidation, and atherogenesis. In. Chien KR, ed. Molecular Basis of Cardiovascular Disease. Philadelphia: Saunders; 1999: Deguchi H, Fernandez JA, Hackeng TM, et al. Cardiolipin is a normal component of human plasma lipoproteins. Proc Natl Acad Sci U S A. 2000;97: Dinu AR, Merrill JT, Shen C, et al. Frequency of antibodies to the cholesterol transport protein apolipoprotein A1 in the patients with SLE. Lupus. 1998;7: Polz E, Wurm H, Kostner GM. Investigations on 2-glycoprotein-I in the rat: isolation from serum and demonstration in lipoprotein density fractions. Int J Biochem. 1980;11: Peaceman AM, Silver RK, MacGregor SN, et al. Interlaboratory variation in antiphospholipid antibody testing. Am J Obstet Gynecol. 1992;166: Favaloro EJ, Silvestrini R, Mohammed A. Clinical utility of anticardiolipin antibody assays: high inter-laboratory variation and limited consensus by participants of external quality assurance programs signals a cautious approach. Pathology. 1999;31: Roubey RA, Eisenberg RA, Harper MF, et al. Anticardiolipin autoantibodies recognize beta-2-glycoprotein I in the absence of phospholipid. J Immunol. 1995;154: Tincani A, Spatola L, Prati E, et al. The anti-beta-2-glycoprotein I activity in human anti-phospholipid syndrome sera is due to monoreactive lowaffinity autoantibodies directed to epitopes located on native beta-2- glycoprotein I and preserved during species evolution. J Immunol. 1996; 157: Rauch J, Janoff AS. Phospholipid in the hexagonal II phase is immunogenic: evidence for immunorecognition of nonbilayer lipid phases in vivo. Proc Natl Acad Sci U S A. 1990;87: Matsuura E, Igarashi Y, Yasuda T, et al. Anticardiolipin antibodies recognize beta 2-glycoprotein I structure altered by interacting with an oxygen modified solid phase surface. J Exp Med. 1994;179: Permpikul P, Rao LV, Rapaport SI. Functional and binding studies of the roles of prothrombin and beta2-glycoprotein I in the expression of lupus anticoagulantactivity. Blood. 1994;83: Iuliano L, Pratico D, Ferro D, et al. Enhanced lipid peroxidation in patients positive for antiphospholipid antibodies. Blood. 1997;90: Pratico D, Ferro D, Iuliano L, et al. Ongoing prothrombotic state in patients with antiphospholipid antibodies: a role for increased lipid peroxidation. Blood. 1999;93: