Anti-mitochondrial M5 type antibody represents one of the serological markers for anti-phospholipid syndrome distinct from anti-cardiolipin and anti-β₂-glycoprotein I antibodies

Abstract

The aim of this study was to characterize the antigen specificity and to evaluate the diagnostic and prognostic value of anti-mitochondrial M5 type antibodies (AMA M5). Fifty-eight patients selected on the basis of their AMA M5 positivity were investigated in relationship to their clinical and serological profile. Cross-absorption studies, Western blotting and immunoprecipitation analysis were carried out for AMA M5 antigen specificity characterization. Most patients had a diagnosis of systemic lupus erythematosus (SLE) (65.5%) or of primary anti-phospholipid syndrome (PAPS) (24%); all the patients were positive for IgG or IgM anti-cardiolipin (anti-CL) antibodies and 49% of them also displayed lupus anticoagulant (LA) activity. Anti-β₂-glycoprotein I (β₂-GPI) IgG were detectable in 30/38 sera (78.9%) and IgM in 34/38 (89.4%). While anti-CL and anti-β₂-GPI IgG antibodies were significantly associated with history of thrombosis and fetal loss, AMA M5 displayed a statistical association only for thrombocytopenia and recurrent fetal loss. Absorption with human β₂-GPI both in free solution or in solid phase as well as with CL liposomes or CL/β₂-GPI liposome complexes did not affect AMA M5 fluorescence. While AMA M5 activity is absorbed by whole mitochondrial preparations, no specific reactivities against several human, bovine and rat mitochondrial proteins could be detected in Western blotting and immunoprecipitation studies. AMA M5 appear to be detectable in both primary and secondary APS, displaying a strong association with the presence of thrombocytopenia and fetal loss. Although strictly related to anti-phospholipid antibodies, AMA M5, anti-CL and anti-β₂-GPI antibodies represent distinct serological markers of the APS.

INTRODUCTION

Anti-mitochondrial antibodies (AMA) have been recognized as a heterogeneous group of autoantibodies classified according to their mitochondrial antigen specificity and disease association [1]. AMA of M5 type have been initially described in patients with autoimmune diseases and then identified quite frequently in sera positive for anti-phospholipid antibodies (aPL) [2–7]. The pattern was characterized by a cytoplasmic fluorescence brighter on proximal than on distal renal tubules and by the lack of gastric parietal cell reactivity [2]. Absorption with purified inner mitochondrial membrane fractions abolished anti-M5 activity, suggesting that the specific antigens were located in the inner membranes [2]. It is well known that mitochondria inner membranes contain cardiolipin [8] and that the mitochondrial fluorescence associated with secondary syphilis (AMA M1) is due to the reagins reacting with mitochondrial cardiolipin [1]. So, AMA M5 association with aPL has been suggested to be the simple result of a comparable reactivity. However, absorption and cross-inhibition studies have been unable to demonstrate such a hypothesis [5,6]. In addition, no data are available on the characteristics of the mitochondrial antigens involved, with the single exception of a subgroup of AMA M5. In fact, Klein & Berg described protease-resistant outer mitochondrial membrane antibodies which provide a mitochondrial fluorescence (AMA M5b) distinct from the classical one; these M5b antibodies reacted with two rat liver determinants at 61 and 57 kD by Western blotting [9].

An association between AMA M5 and antibodies directed against β₂-glycoprotein I (β₂-GPI) in aPL⁺ sera has also been reported [7,10]. The demonstration that β₂-GPI can bind to anionic phospholipids and overall that it can react with mitochondrial particles [8] raises the question whether or not AMA M5 fluorescence could be, at least in part, related to anti-β₂-GPI activity.

Whether the occurrence of AMA M5 in anti-phospholipid syndrome (APS) sera is just the result of aPL reactivity with mitochondrial phospholipids and/or phospholipid binding proteins (such as β₂-GPI), or alternatively whether AMA M5 can be regarded as another distinct serological marker for the syndrome is still an open question. In this regard, we thought it useful: (i) to evaluate the diagnostic value of AMA M5 antibodies by investigating their association with several clinical and serological APS markers; and (ii) to characterize the mitochondrial antigens to which AMA M5 are directed, including the possible reactivity against β₂-GPI.

We confirmed the high prevalence of AMA M5 positivity in the largest group of aPL⁺ sera reported in the literature and we found that, although closely associated, anti-mitochondrial, anti-CL and anti-β₂-GPI antibodies do represent distinct serological markers for APS.

PATIENTS AND METHODS

Patients

One hundred and seventy-seven patients (22 males and 155 females; mean age 34.4 years, range 12–68 years) were included in the present study. The main diagnoses were: systemic lupus erythematosus (SLE) according to the American Rheumatism Association criteria [11] and primary anti-phospholipid syndrome (PAPS) according to Hughes [12]. Clinical manifestations of APS were defined as follows: (i) fetal loss as three or more abortions or one or more intrauterine death; (ii) thrombocytopenia when platelet count was < 10¹¹/l and thrombosis when venous and/or arterial thrombotic events were confirmed by venogram, Doppler-echography and angiography or magnetic resonance imaging, respectively [12].

Indirect immunofluorescence

Patients' sera were screened for AMA by standard indirect immunofluorescence technique as previously described [5]. AMA M5 were identified by the characteristic pattern: the fluorescence was brighter on the first portion of the proximal tubules (Fig. 1) and declined in the distal tubules and no reactivity was found on gastric parietal cells. Reference sera positive for AMA M1, M2 and M5 type [5] were included in each assay.

Fig. 1.

Indirect immunofluorescence pattern of anti-mitochondrial M5 type antibodies (AMA M5) on rat kidney section: cytoplasmic fluorescence of the first portion of the proximal tubule near the glomerulus (original mag. × 400).

β₂-GPI and anti-β₂-GPI antibody preparation

Human β₂-GPI was purified according to Polz et al. [13] and characterized as previously described [14,15].

Anti-β₂-GPI polyclonal and monoclonal antibodies (MoAbs) were obtained by immunization of rabbits and BALB/c mice, respectively, with human β₂-GPI, as previously described [15].

Anti-phospholipid and anti-β₂-GPI antibodies

Anti-cardiolipin (CL) and anti-β₂-GPI antibodies were detected by solid-phase ELISAs, as previously described [14]. Lupus anticoagulant activity was detected as previously reported [4].

Absorption experiments

Twenty-one sera positive for both anti-β₂-GPI and AMA M5 antibodies were subjected to absorption with the following antigens: (i) human β₂-GPI (0.25, 0.5 or 1 mg/ml); (ii) CL liposomes (1 mg/ml) incubated with serial concentrations of human β₂-GPI (0.01, 0.1 and 1 mg/ml); and (iii) liposomes at serial CL concentrations (0.1, 0.5, 1 mg/ml).

Cardiolipin micelles were prepared as reported [5]. Cardiolipin liposome–β₂-GPI complexes were prepared incubating the micelles with human β₂-GPI at serial concentrations for 2 h at 4°C on an end-over-side mixer. After incubation, the micelles were extensively washed with cold PBS by centrifugation at 30 000 g for 15 min and adjusted at the final concentration. Sera, at two end points of AMA M5 reactivity dilution, were incubated with an equal volume of the different antigen preparations for 2 h at room temperature and overnight at 4°C. After incubation, the mixtures were centrifuged at 30 000 g for 15 min and the supernatants kept as absorbed sera. Two additional sera positive for AMA M5 and for anti-β₂-GPI antibodies were absorbed on a β₂-GPI-N-hydroxysuccinimide-activated Sepharose (HiTrap NHS activated; Pharmacia Biotech Europe) prepared as previously described [16]. The samples were tested, before and after absorption, for anti-CL and anti-β₂-GPI antibodies by ELISA and for AMA M5 antibody titre by indirect immunofluorescence.

For absorption studies with whole mitochondria, 10 sera were incubated with rat liver mitochondria prepared as described [5]. Absorption experiments were carried out by mixing sera (diluted 1:5 with PBS) with rat liver mitochondria at protein concentrations ranging from 0.5 to 10 mg/ml (v/v); the mixtures were incubated for 1 h at 37°C and overnight at 4°C, then the samples were centrifuged at 30 000 g for 15 min at 4°C and the supernatants kept as absorbed sera [5].

Mitochondrial preparations

Bovine heart mitochondria were kindly provided by Dr Y. Hatefi (The Scripps Research Institute, La Jolla, CA). Human placental and rat liver mitochondrial extracts were obtained as previously described [17]. Briefly, tissues (30 g) were homogenized in 30 ml buffer containing 0.5 m sorbitol, 0.1 mm EDTA and 50 mm Tris pH 7.4 in ice-cold. After two spins at 250 g for 10 min at 4°C with an intermediate filtration through cheesecloth, the supernatants were collected and spun again at 8000 g for 10 min at 4°C. The pellets were then resuspended in 15 ml of buffer, spun twice again as above and resuspended in 4 ml of buffer containing bovine serum albumin (BSA) 0.05% (Sigma Chemical Co., St Louis, MO). Final mitochondrial protein concentrations ranged from 40 to 60 mg/ml depending on the different preparations.

Western blotting analysis

Anti-mitochondrial M5-positive sera were tested by Western blotting analysis against bovine heart, human placental and rat liver mitochondrial preparations as described by Tobwin et al. [18]. Each mitochondrial preparation was solubilized in Laemmli's sample buffer, heated at 100°C for 5 min and than separated on 10% SDS polyacrylamide resolving gel with a 6% stacking gel using 30 V for 12 h at room temperature. Separated proteins were electrotransferred to nitrocellulose membranes at 50 V for 4 h at 4°C. The gel was calibrated using commercial molecular weight markers ranging from 6 to 200 kD. Nitrocellulose membranes were cut into strips and blocked in 5% non-fat milk in PBS containing 0.05% Tween 20 for 30 min. Individual strips were incubated for 1 h with samples diluted 1:200 in blocking solution. Normal controls were represented by blood donors. Bound antibodies were detected by chemiluminescence using peroxidase-conjugated goat anti-human immunoglobulin (Cappel, Cochranville PA) and ECL (Amersham Int. plc, Aylesbury, UK) followed by autoradiography.

Immunoprecipitation

HEp2 cells were grown in monolayers and radiolabelled with ³⁵S-methionine. The cells were washed in PBS and harvested with a rubber policemen in ice-cold buffer containing 10 mm Tris–HCl pH 7.4, 150 mm NaCl, 1.5 mm MgCl₂, 0.5% Nonidet P40, supplemented with 2 mm PMSF. The cells were frozen-thawed twice, centrifuged at 10 000 g, and the supernatant was used for immunoprecipitation. Ten microlitres of each serum were incubated with 100 μl of protein A Sepharose CL-4B (Pharmacia) at 10% (v/v) suspension in buffer containing 50 mm Tris–HCl, 150 mm NaCl, 5 mm EDTA, 0.5% Nonidet, 0.5% sodium deoxycholate, 0.1% SDS and supplemented with 2 mg/ml BSA. Radiolabelled HEp2 cell protein extract (100 μg) was added to the beads and incubated for 1 h with constant mixing. After centrifugation the pellet was washed five times, boiled in Laemmli sample buffer and the proteins were analysed by SDS–PAGE, followed by autoradiography.

Statistical analysis

A cross-tabulation for the main clinical and laboratory signs of the syndrome (thrombosis, recurrent fetal loss and thrombocytopenia) was performed according to: (i) the presence or absence of anti-CL, anti-β₂-GPI and AMA M5 antibodies; and (ii) the joint concurrent presence or absence of AMA M5 plus anti-CL or AMA M5 plus anti-β₂-GPI antibodies. The Pearson χ² for association, the odds-ratio (OR) and the Cornfield 95% confidence intervals were also estimated (P < 0.05 taken as significant). Since IgG anti-CL have been reported to display the strongest diagnostic and prognostic value [19], the statistical analysis was carried out on anti-CL, anti-β₂-GPI and AMA M5 IgG-positive samples only.

RESULTS

Clinical and serological associations

Among the included patients we selected 58 AMA M5-positive sera (six males and 52 females; mean age 31.5 years, range 12–66 years) with titres ranging from 1/10 up to 1/1280 (53/58 samples displayed an AMA M5 staining with a titre > 1/40). Forty out of 58 samples were positive for IgG and the remaining for both IgG and IgM AMA M5. As shown in Table 1, most patients were diagnosed as SLE (38/58) or as definite PAPS (14/58). In the remaining patients the following diagnoses were made: three rheumatoid arthritis (RA) according to Arnett et al. [20], one drug addict, one Waldenström's macroglobulinaemia and one systemic autoimmune vasculitis. All the AMA M5 sera were positive for anti-CL antibodies and 26/53 (49%) also for lupus anticoagulant (LA). In our series both IgG and IgM anti-β₂-GPI antibodies were detectable in the majority of investigated samples (Table 1). As a whole, the group of AMA M5-positive patients was characterized by a high prevalence of thrombosis, recurrent fetal loss and thrombocytopenia (Table 1). Since most AMA M5-positive patients were diagnosed as SLE or PAPS, 83 AMA M5-negative SLE (10 males and 73 females; mean age 36.8 years) (46/83 positive for anti-CL antibodies and 22/44 for LA) and 36 AMA M5-negative PAPS (six males and 30 females; mean age 34.9 years) were also investigated (Table 1).

Table 1.

Clinical and serological characteristics of patients

SLE, Systemic lupus erythematosus; PAPS, primary anti-phospholipid syndrome; LA, lupus anticoagulant; β₂-GPI, β₂-glycoprotein I.

The simultaneous presence of several well accepted markers for APS [21] raises the question, what is the prognostic value of AMA M5 in comparison with that due to anti-CL or anti-β₂-GPI antibodies? In accordance with previous reports, anti-CL and anti-β₂-GPI IgG antibodies were significantly associated with the presence of thrombotic events in the patients' clinical records [19,22,23] (Table 2a). In contrast, AMA M5 antibodies did not associate with thrombosis. The positivity for AMA M5 was significantly more frequent in patients displaying platelet values < 10¹¹/l (P = 0.015); the simultaneous presence of anti-CL or anti-β₂-GPI IgG antibodies did not affect the statistical analysis (P = 0.029 and P = 0.036, respectively) (Table 2b). In agreement with previous reports, anti-CL and anti-β₂-GPI antibodies both displayed a statistically significant association with recurrent fetal loss (reviewed in [23,24]) (Table 2c). As reported in Table 2c, only a few AMA M5-positive women, in whom both anti-CL and anti-β₂-GPI antibodies were simultaneously determined, satisfied the inclusion criteria and were therefore investigated. The positivity for AMA M5 formally displayed a significant association and a high OR value for fetal loss (P = 0.039; OR = 7.82).

Table 2a.

a. Association between thrombosis and the presence of IgG anti-cardiolipin (CL), IgG anti-β₂-glycoprotein I (β₂-GPI) and IgG anti-mitochondrial M5 type antibodies (AMA M5) individually evaluated or between the joint presence of IgG AMA M5 plus IgG aCL or IgG AMA M5 plus IgG anti-β₂-GPI. b. Association between thrombocytopenia and the presence of IgG anti-CL, IgG anti-β₂-GPI and IgG AMA M5 individually evaluated or between the joint presence of IgG AMA M5 plus IgG aCL or IgG AMA M5 plus IgG anti-β₂-GPI. c. Association between recurrent fetal loss and the presence of IgG anti-CL, IgG anti-β₂-GPI and IgG AMA M5 individually evaluated or between the joint presence of IgG AMA M5 plus IgG aCL or IgG AMA M5 plus IgG anti-β₂-GPI

Characterization of the antigen(s) responsible for AMA M5 reactivity

Absorption studies

Absorption with whole rat liver mitochondria (10 mg/ml) virtually abolished the fluorescence reactivity of the tested AMA M5-positive sera (data not shown). Twenty-one additional AMA M5, anti-CL and anti-β₂-GPI-positive sera were subjected to extensive absorption with the following antigens: (i) CL-liposomes, (ii) human β₂-GPI and (iii) CL-liposomes incubated with different concentrations of human β₂-GPI in order to obtain PL–cofactor complexes. In agreement with previous data [5,6], CL-liposomes were unable to affect at all AMA M5 titres (data not shown). Absorption with β₂-GPI–CL-liposomes as well as with β₂-GPI alone inhibited anti-β₂-GPI activity (> 85% inhibition), but did not reduce anti-mitochondrial reactivity even with high β₂-GPI protein concentrations known to be able to absorb anti-β₂-GPI antibodies in fluid-phase systems [16]. Comparable results have also been obtained after absorption of anti-β₂-GPI antibodies on immobilized β₂-GPI columns (data not shown).

Immunofluorescence studies

Two murine MoAbs and a rabbit polyclonal IgG anti-β₂-GPI were tested by standard indirect immunofluorescence on rat tissue sections and on HEp2 cells: no cytoplasmic immunofluorescence reactivity was found, suggesting that anti-β₂-GPI activity was not responsible for mitochondrial fluorescence.

In addition, using isotype-specific anti-human immunoglobulin, we found the same immunoglobulin isotype pattern for AMA M5 and anti-β₂-GPI only in two out of 16 sera positive for both autoantibodies (data not shown). These latter findings once more suggested that anti-β₂-GPI and AMA M5 reactivities were due to different antibody populations.

Western blotting analysis

Anti-mitochondrial M5-positive sera were analysed by Western blotting analysis against bovine heart, rat liver and human placental mitochondrial protein preparations. Figure 2 shows that AMA M5-positive sera did not display any reactivity against bovine heart mitochondrial preparation, while two sera from patients suffering from primary biliary cirrhosis (PBC) recognized molecules at 74, 50 and 41 kD, consistent with the expected AMA M2 antigens [25–28]. Comparable results were also obtained by testing AMA M5-positive sera against rat liver and human placental mitochondria (data not shown).

Fig. 2.

Western blotting analysis of anti-mitochondrial M5 type antibody (AMA M5) and AMA M2-positive sera against bovine heart mitochondrial preparation. Anti-mitochondrial OGDC-BCOADC prototype sera displayed a reactivity with a 50-kD antigen (lane 2) or with PDC-E2 and PDC-E1α antigens (lane 3). No reactivity was found when normal human serum (NHS) (lane 1) or AMA M5-positive sera (lanes 4, 5 and 6) were analysed.

Immunoprecipitation studies

To investigate further AMA M5 reactivity with mitochondrial proteins, positive reference sera were studied for their ability to immunoprecipitate ³⁵S-methionine-labelled HEp2 cell extract. Figure 3 shows that AMA M5-positive sera displayed the same pattern found with the control human serum (NHS), further supporting the inability to demonstrate a specific reactivity against mitochondrial proteins. On the other hand, a well characterized AMA M2-positive control serum from a patient with PBC immunoprecipitated the expected mitochondrial polypeptides.

Fig. 3.

Immunoprecipitation analysis of radiolabelled HEp2 cell protein extract performed with anti-mitochondrial M5 type antibody (AMA M5) and AMA M2-positive sera: AMA M5 sera (lanes 3 and 4) gave an immunoprecipitation pattern quite comparable to that displayed by the control normal human serum (NHS) (lane 1). On the other hand, an AMA M2-positive serum immunoprecipitated the expected autoantigens (PDC-E2 70 kD, OGDC and BCOADC 50 kD, PDC-E1α 41 kD) and other uncharacterized antigens.

DISCUSSION

APS is characterized by peculiar clinical manifestations in the presence of persistent positivities for aPL [22]. However, it is general knowledge that the clinical picture is not always homogeneous and patients with recurrent arterial but not venous thrombosis (or vice versa), or women with only recurrent fetal loss but without any vascular involvement, can be identified. No specific markers predictive for the different clinical manifestations have been reported.

Although the relationship between APS and aPL antibodies does represent a milestone for diagnosis, several reports state that other autoantibodies can be detected in the syndrome. Besides anti-β₂-GPI antibodies whose diagnostic and pathogenic value for APS is now increasing, other autoantibodies in fact have been found in aPL⁺ sera, such as anti-erythrocyte, anti-endothelial, anti-platelet or even anti-nuclear autoantibodies [29].

In the present study, we confirm the association between AMA M5 and aPL antibodies in the largest analysis reported in the literature to our knowledge. In agreement with previous studies, AMA M5 reactivity was found significantly associated with the presence of anti-CL, anti-β₂-GPI antibodies and LA [3–7,10]. We did not find aPL⁻ samples in the present AMA M5 populations, further supporting the data in the literature that AMA M5-positive patients without any aPL activity are an exception [3–7,10]. In addition, a high proportion of our patients, selected on the basis of M5 positivity only, displayed clinical features characteristic of the syndrome. Due to the formal lack of a population positive for AMA M5 but negative for aPL, we investigated the prognostic value of AMA M5 for thrombocytopenia and recurrent thrombosis or fetal loss versus that displayed by anti-CL or by anti-β₂-GPI IgG antibodies individually considered. Statistical analysis has been carried out in the whole AMA M5-positive population. Actually, the distinction between primary and secondary APS would lead to an investigation of AMA M5 association weight as a function of a given diagnosis (i.e. PAPS), which itself is based on the presence of thrombocytopenia, recurrent thrombosis or fetal loss (and persistent aPL positivity) [12]. We found that anti-mitochondrial reactivity was not associated with recurrent thrombosis, in contrast to anti-CL or even anti-β₂-GPI antibodies, which behaved as independent markers for vascular involvement. While in our study the presence of anti-CL and anti-β₂-GPI antibodies was not associated with thrombocytopenia, AMA M5 seemed to be an independent marker for platelet values < 10¹¹/l. These findings better define previous reports that found a high prevalence of haematological abnormalities (thrombocytopenia, haemolytic anaemia, Coombs' positivities) in AMA M5 sera [2,3,7]. AMA M5 antibodies appeared to be associated also with recurrent fetal loss and they displayed a high relative risk for miscarriages (OR = 7.82). However, it is important to point out the small number of M5-positive women who satisfied the inclusion criteria and in whom both anti-CL and anti-β₂-GPI antibodies were simultaneously investigated.

Altogether, these findings indicate the usefulness of AMA M5 screening in patients with a suspected diagnosis of the syndrome, and suggest its prognostic value for thrombocytopenia and fetal loss.

The antigenic determinants against which the antibodies react are still undefined. In line with the literature [5,6], extensive absorption of anti-CL antibodies did not affect at all the reactivity of positive AMA M5 sera with mitochondria.

β₂-GPI is a plasma anionic phospholipid binding glycoprotein that reacts with mitochondrial particles [8]. Furthermore, we recently reported that IgG anti-β₂-GPI, affinity-purified from APS sera, recognizes epitopes on the molecule that are preserved during species evolution, being able to react with cofactor preparations from different animals, including rat β₂-GPI [16]. These findings raise the hypothesis that AMA M5 fluorescence could be, entirely or in part, related to anti-β₂-GPI antibodies able to recognize the glycoprotein adhered to mitochondrial structures in rat tissues. Our previous results and the present ones do not support such a possibility, since anti-mitochondrial and anti-β₂-GPI activities do not always coexist in the same sera; in addition, when present in the same samples frequently they belong to different isotypes [10]. Although anti-β₂-GPI antibodies are low-affinity antibodies that display a higher binding to β₂-GPI coated to negatively charged surfaces than to the cofactor in solution, absorption of the antibody activity can be achieved by incubating whole sera or IgG fraction with β₂-GPI concentrations > 150 μg/ml [16]. When sera positive for the two specificities were incubated with both immobilized human glycoprotein and with β₂-GPI in free solution at high protein concentrations [16], anti-β₂-GPI activity was significantly inhibited but no changes in AMA M5 titre could be found. The lack of association between mitochondrial fluorescence and anti-β₂-GPI antibodies was further supported by the inability of two murine monoclonal and a rabbit polyclonal anti-human β₂-GPI antibody to give any mitochondrial fluorescence when incubated on HEp2 cells or on rat tissue substrates.

Some authors suggest that the antigen for many aPL antibodies is a complex between anionic PL and plasma β₂-GPI. According to them, the target for the autoantibodies would be a shared epitope between β₂-GPI and PL or conformational changes in β₂-GPI after its binding to PL [30,31]. Our results show that even the complex between β₂-GPI and CL was unable to affect the mitochondrial fluorescence, suggesting once again that AMA M5 was not due to antibodies reacting with the PL–β₂-GPI complex.

Previous studies clearly demonstrated that AMA M5 react with mitochondrial antigen(s) [1]. On the other hand, AMA of different types have been reported to display specific patterns when tested by Western blotting or immunoprecipitation against cytoplasmic components [1]. Western blotting analysis of AMA M5-positive sera against mitochondrial proteins from human, bovine and rat tissues did not reveal any band, while reference AMA-positive PBC sera identified molecules consistent with the expected antigens of the pyruvate dehydrogenase complex enzyme [25–28]. Comparable results were also found with radiolabelled HEp2 cell extracts analysed by immunoprecipitation. Some hypotheses can be put forward to explain our inability to identify a proteinic M5 antigen: it is in fact possible that the antigen to be recognized (i) requires a conformational structure that is potentially altered by SDS treatment, or alternatively that (ii) it could require the simultaneous presence of both a proteinic molecule and PL to make a complex antigen. SDS might be responsible for lipid detachment from such a complex. Finally, it is also possible that the conformational antigen epitope might be complexed with other proteins becoming unavailable for the autoantibody in the immunoprecipitation assay.

In conclusion, our results confirm and extend the association of AMA M5 with LA, anti-CL and anti-β₂-GPI antibodies, but clearly demonstrate that they belong to separate antibody populations. At variance with anti-CL and anti-β₂-GPI antibodies, AMA M5 appears to be a prognostic marker for thrombocytopenia and fetal loss, but not for arterial or venous thrombosis. Both laboratory and clinical associations support that AMA M5 detection should be included in the antibody panel for APS diagnosis.