Introduction

Antiphospholipid syndrome (APS) is a systemic autoimmune disease characterized by vascular thrombosis and/or pregnancy morbidity, associated with a persistent positivity for serum antiphospholipid antibodies (aPL). aPL are currently evaluated by use of two solid-phase assays, detecting antibodies against cardiolipin (aCL) and β2–glycoprotein-I (anti-β2GPI antibodies), and by use of a functional assay, lupus anticoagulant (LA) test [1].

aPL were initially believed to react against negatively charged phospholipids (PL); however, it soon became clear that aPL bind to proteins with affinity for PL. In particular, β2–glycoprotein-I (β2GPI), together with prothrombin, is the most important epitope targeted by aPL. Antibodies reacting against β2GPI have also been identified as the main pathogenic subset of aPL in both in-vivo and in-vitro experiments.

Consistent with laboratory findings, anti-β2GPI antibodies have been associated with increased risk of developing clinical manifestations of APS [2]. However, not all patients carrying anti-β2GPI antibodies develop aPL-related clinical manifestations. This observation agrees with increasing evidence that anti-β2GPI antibodies are rather heterogeneous. Within this autoantibody family are multiple antibody subsets, each with a different pathogenic potential. Such heterogeneity among anti-β2GPI antibodies might be partially explained by evidence that several epitopes of β2GPI can be targeted by specific autoantibodies. However, a specific epitope in domain (D) I, a positively charged discontinuous structure located in the N-terminal of β2GPI, has been identified as the most relevant antigenic target involved in β2GPI/anti-β2GPI antibody binding [3]. Over recent years, much research into APS has had the objective of better characterizing the pathogenic function and the clinical significance of antibodies specifically targeting DI. The implications of such a finding might have a strong effect on APS clinical diagnosis and management in the near future. It is therefore an appropriate time to critically review the available evidence regarding anti-DI antibodies and APS.

Beta-2 Glycoprotein-I: The Protein

β2GPI, also known as apolipoprotein H, is a single-chain 43 kDa glycoprotein found in human plasma at concentrations in the range 50–400 μg mL−1. This evolutionary conserved protein is synthesized by, among others, endothelial cells, hepatocytes, and trophoblast cells [4]. The physiological function of β2GPI was determined only recently, when two independent groups revealed that the C-terminal of the protein interacts specifically with lipopolysaccharide (LPS). This observation led the investigators to hypothesize that β2GPI may act as a carrier or as a scavenger for LPS [5••, 6]. Further evidence for an interaction between β2GPI and LPS was provided by the in-vivo observation that LPS injection induces a 25 % reduction of baseline β2GPI serum levels. Moreover, plasma levels of β2GPI were found to inversely correlate with inflammatory markers, including tumor necrosis factor (TNF) α, interleukin (IL)-6, and IL-8 [5••]. A member of the complement control protein (CCP) family, β2GPI consists of 326 aminoacidic residues arranged in five CCP repeat domains, termed “sushi” domains. DI–IV comprise 60 amino acids and each contain two disulfide bridges. DV is aberrant, including 82 amino acids resulting from a six-residue insertion and a 19-residue C-terminal extension crosslinked by an additional disulfide bond. DV is responsible for β2GPI binding to PL via a cluster of positively charged amino acids (282–287); the same cluster also mediates the adhesion of β2GPI to cells targeted by aPL, including the trophoblast and endothelial cells [2, 7].

Three configurations of β2GPI have been described. Circulating plasma β2GPI exists in a circular form, observed by use of electronic microscopy [8]. Upon binding to suitable anionic surfaces, for example to cardiolipin (CL) and other PL, or to LPS, the molecule opens up to a J-shaped fish-hook configuration, as revealed by its crystal structure [9, 10]. An intermediate S-shape of β2GPI has been recently observed by use of small-angle X-ray scattering [11] (Fig. 1).

Fig. 1
figure 1

Schematic representation of the three conformations of β2glycoprotein I and the relative exposure of domain I

Full size image

Beta-2 Glycoprotein-I and Reactivity Toward its Five Domains

In the late 90s, research focused on identifying the β2GPI binding site for anti-β2GPI antibodies. Some groups claimed that the epitope for binding of anti-β2GPI antibodies was located on DIV, whereas other investigators suggested DV was involved [12]. It was observed that binding of anti-β2GPI antibodies to β2GPI was prevented when a clip was positioned at 316–317 in DV; Arvieux reported that anti-β2GPI antibodies inhibited the binding of β2GPI to CL [13–15]. Lastly, others reported anti-β2GPI antibodies reacting with peptides that cover sequences in DI–IV [16]. These findings clearly revealed that anti-β2GPI antibodies can bind to each of the five domains of β2GPI. However, consistent evidence has led to the identification of DI as the immunodominant epitope.

Domain I as the Immunodominant Epitope

The reactivity of antibodies against DI was first described in 1998, when Iverson developed domain-deletion mutants of β2GPI [17]. By use of surface plasmon resonance, it was consistently observed that the immunodominant binding epitope for anti-β2GPI antibodies was localized in DI of β2GPI [18]. Creation of human β2GPI variants with point mutations in DI has enabled observation of the discontinuous nature of the main epitope: it involves arginine 39-arginine 43, aspartic acid 8-aspartic acid 9, and possibly the interlinking region between DI and DII, with R39 being the most important residue [19–21]. This epitope was later revealed to be a cryptic and conformation-dependent structure. In the circular conformation of β2GPI, DI interacts with DV and the critical epitope is thus hidden. When β2GPI adopts the S-shape, the epitope is covered by DIII–IV carbohydrate chains. These residues form a shield over DI, thus preventing antibodies from binding β2GPI. It has been consistently observed that antibodies against DI are able to bind β2GPI only when the carbohydrate chains have been removed; the antibodies have no reactivity toward the intact molecule [22]. Upon β2GPI opening to a J-configuration, the critical epitope arginine 39-glycine 43 is exposed, thus becoming available for antibody binding. The three conformations of β2GPI and the relative exposure of DI to the surface are represented in Fig. 1. The hypothesis that the immunogenicity of β2GPI depends upon its conformation is supported by in-vivo evidence. Mice developed antibodies against DI only when injected with misfolded β2GPI or with β2GPI-CL, and production of anti-DI antibodies was observed when mice were injected with DI but not with DII–V [23••].

The conformation-dependent binding of anti-β2GPI antibodies to their target antigen, together with the low avidity of these antibodies, might explain why β2GPI/anti-β2GPI antibody immune complexes are not easily isolated from serum samples from APS patients. Overall, it is clear that β2GPI conformation strongly affects anti-β2GPI antibodies binding to the target epitope. Several factors might lead to the surface exposition of the critical epitope, including oxidative stress. In healthy persons, a free thiol form of β2GPI, characterized by a broken disulfide bridge, predominates in the plasma. Under conditions of oxidative stress, disulfide bonds form at these sites, possibly exposing the critical B-cell epitope [24]. Compared with asymptomatic aPL carriers and healthy volunteers, APS patients were consistently found to have significantly higher oxidized plasma β2GPI. Furthermore, anti-β2GPI antibodies purified from β2GPI-immunized animals and from APS patients had reduced binding to β2GPI treated with oxidoreductase [25].

However, in rats primed with LPS, deposition of β2GPI-dependent aPL human IgG on the endothelium was observed immediately after the infusion, implying that in-vivo anti-β2GPI antibodies adhere to the vessel wall [26]. Therefore, β2GPI might undergo a conformational change upon binding to the surface of target cells, exposing the critical epitope, which thus becomes available to specific autoantibodies. As a whole, the above evidence suggests that β2GPI becomes immunogenic as a consequence of a conformational change: it could be hypothesized that, being cryptic, DI, in contrast with the other domains, does not induce immune tolerance. Consequently, DI could induce specific autoantibodies more easily than do other domains [27].

Anti-Domain I Antibodies in APS: Evidence of their Pathogenetic Function

Evidence of the pathogenicity of anti-DI antibodies comes from both in-vitro and in-vivo studies. First, anti-DI antibodies were repeatedly observed to induce in-vitro prolongation of clotting time [23••, 28]. Proof of their thrombogenic function was obtained in 2009, when passive infusion of a synthetic DI peptide in naïve mice was revealed to prevent, in a dose-dependent manner, the thrombus enhancement mediated by polyclonal aPL human IgG fractions. Furthermore, the infusion of the peptide inhibited, although not completely, the expression of adhesion molecules on aortic endothelial cells and the production of tissue factor (TF) by murine macrophages. Interestingly, mutations in DI associated with either an increase or a reduction in its affinity for IgG purified from APS patients correspondingly affected the ability of the mutant peptide to reverse the effects mediated by the same aPL fractions [29]. More recently, it was revealed that a greater increase in TF activity and significantly larger thrombi were induced by eluted fractions rich in anti-DI antibodies, obtained from an APS patient, than by the anti-DI-antibody-poor serum recollected after affinity-purification [30]. Direct observation of the pathogenic effects of anti-DI antibodies has been recently obtained by use of a human monoclonal anti-DI IgG, infusion of which induced clotting and fetal loss in naïve mice. The anti-DI monoclonal was found to induce clotting only after the concomitant administration of LPS, in agreement with the “two-hit” hypothesis [31••].

Tests Detecting Anti-Domain I Antibodies

The conformational changes of β2GPI might directly affect detection of anti-β2GPI antibodies. It has been clearly revealed that β2GPI might adopt peculiar conformations in different assays, caused by heterogeneity in purification methods or in coating procedures. Most recently, DI exposure has been revealed to be highly heterogeneous across commercially available ELISA assays, potentially affecting test results [32].

It has been consistently observed that anti-DI antibodies have reactivity toward their target epitope only when DI is coated onto hydrophobic, but not hydrophilic, plates, suggesting that β2GPI undergoes conformational changes leading to DI being exposed to the surface only when coated on hydrophilic microtiter plates [28]. This conformation challenge contributes greatly toward the inter-assay variability regarding detection of anti-DI antibodies. To date, there is no commercial kit to detect anti-DI antibodies available on the market; however, several research assays have been developed.

Other than the two-step ELISA test using both hydrophilic and hydrophobic plates, a few additional ELISA assays have been developed to detect anti-DI antibodies, each using different molecular antigenic targets. Recently, a β2GPI-DI chemiluminescence immunoassay (CIA, INOVA Diagnostics, San Diego, USA) has been developed, which uses a recombinant DI coupled to paramagnetic beads by use of the BIO-FLASH technology (Biokit, Barcelona, Spain) [33]. The ELISA and CIA research assays, both developed by INOVA Diagnostics (San Diego, CA, USA), have been directly compared, revealing that the two methods have the same specificity but a different sensitivity [34]. The CIA immunoassay has also been evaluated in comparison to a UK in-house ELISA test: a good agreement between the two tests was observed [35]. Although these preliminary data seem to suggest comparability between the solid phase assays and the CIA, multi-center prospective studies are warranted to more fully investigate the reproducibility of the different anti-DI antibody assays. The clinical significance of anti-DI antibodies and of anti-β2GPI antibodies detected by CIA has been directly compared; the two assays had good qualitative and quantitative agreement and similar discrimination between APS patients and controls [36]. Lastly, two additional tests to detect anti-DI antibodies have been described: a capture ELISA method using N-terminally biotinylated DI on streptavidin plates, and a liquid phase inhibition assay using whole β2GPI immobilized on the solid phase and synthetic β2GPI-DI as inhibitor [37, 38•].

Anti-Domain I Antibodies and aPL-related Clinical Manifestations

To date, few studies have addressed the subject of anti-DI antibodies and aPL-related clinical manifestations. Details of these studies, i.e. study population, anti-DI antibody positivity, the assay used, and the observed association between anti-DI antibodies and aPL-related clinical manifestations, are shown in Table 1.

Table 1 Studies assessing clinical function of anti-DI antibodies
Full size table

Positivity for DI-targeting antibodies among APS patients varies widely, depending on the selection of the study population and the test performed. In a cohort of 144 APS patients, anti-DI antibodies were detected in 85 % of cases by use of CIA, while Hollestelle has recently reported positivity as low as 33.3 % for anti-DI antibodies among APS patients [39, 40]. Positivity for anti-DI antibodies has been reported to be higher in primary APS than in APS associated with other systemic autoimmune diseases [36]. Despite these discrepancies, antibodies targeting DI—but notably not those reacting against the whole molecule—were found to be significantly associated with diagnosis of APS [40]. Furthermore, anti-DI antibodies as detected by use of CIA provide good specificity and high sensitivity for diagnosis of APS, both being approximately 85 % [41]. This observation fits well with the documented strong association between anti-DI antibodies and a positive LA test, which is the most powerful predictor of clinical events in APS [42].

Most of the available clinical studies on anti-DI antibodies focused on the association with thrombosis. Many authors confirmed the relationship between anti-DI antibodies and thrombotic events affecting the venous and the arterial vascular tree [28, 39, 43–46]. In particular, such an association emerged in the largest study published to date, a multi-center cohort comprising 442 APS patients. The investigators reported an odds ratio (OR) of 3.5 for anti-DI antibodies to predict thrombosis [43]. In contrast with the findings of the first study conducted by de Laat in 2005, anti-DI antibodies were also found to be related to pregnancy complications, although to a lesser extent than to thrombosis (OR 2.4) [28, 43]. Concordantly, anti-DI antibodies have been identified as the prevalent antibody specificities among APS patients who also have pure obstetric morbidity. Although the positivity was slightly lower among women with obstetric APS than among subjects with thrombosis (61.3 % versus 78.2 %), no significant differences in anti-DI antibody frequency were observed between the two subgroups of patients with different APS clinical manifestations. In addition, the same study reported no significant difference in anti-DI antibody titers between patients with thrombotic manifestations and women with pure obstetric complications [46]. In contrast, a few years ago other authors claimed that patients with thrombotic events had higher anti-DI antibody titers than those with non-vascular manifestations, although this finding was not replicated in further studies [41].

It is interesting to note that anti-DI antibodies provide the main epitopic specificities even for patients with autoimmune conditions other than APS. Indeed, aPL-positive subjects with systemic lupus erythematosus (SLE) or undifferentiated connective tissue diseases (UCTD), but no clinical features of APS, have positivity for anti-DI antibodies comparable to that of APS patients, whereas antibodies against DIV or DV are less frequent. On the other hand, in control populations the positivity for anti-DI antibodies has been revealed to be rather low. Anti-β2GPI IgG isolated from sera obtained from aPL-positive asymptomatic carriers, individuals with leprosy, or children with atopic dermatitis have been revealed to preferentially recognize epitopes on DIV or V [46, 47]. Overall, these observations suggest that anti-DI antibodies may cluster in patients with systemic autoimmune diseases. As a consequence, it has been suggested that the ratio between antibodies targeting DI and those reacting against DIV and DV might be of use in discriminating between pathogenic anti-β2GPI antibodies and those autoantibodies that, occurring in association with a wide variety of non immune-mediated clinical conditions, are a mere epiphenomenon with no pathogenic potential. This hypothesis needs to be tested by appropriately sampled studies; if confirmed, it would imply that screening for reactivity against specific domains of the protein would enable predictive and non-predictive anti-β2GPI antibodies to be distinguished.

Further evidence for the pathogenic potential of anti-DI antibodies is provided by a good correlation with annexin A5 resistance assay, observed for cohorts of APS subjects and of adult and pediatric SLE patients [48–50]. Annexin A5 resistance is detected by use of a novel two-stage coagulation assay, and has been revealed to be reduced for aPL-positive patients with a history of thrombosis [51]. Annexin A5 is a potent anticoagulant protein mainly found in trophoblasts and vascular endothelial cells; β2GPI-dependent aPL have been revealed to interfere with the protective shield that annexin A5 provides over the endothelium, thus favoring thrombosis [52].

Given that anti-DI IgG may have a more predictive aPL profile, anti-DI antibodies might be useful for risk-stratification of APS patients. APS patients at higher risk, i.e. those with triple aPL-positivity at medium-high titers, have been consistently observed to have a higher frequency and higher titers of anti-β2GPI-DI antibodies [38•, 45, 46].

Anti-Domain I Antibodies: Potential use for Clinicians

Anti-β2GPI antibodies specifically reacting with DI have a particular clinical importance, being more commonly detected among patients with APS and other autoimmune diseases than among those with transient aPL positivity caused by polyclonal B-cell activation, e.g. infections or atopic dermatitis. This observation implies that, compared with antibodies targeting the whole molecule, anti-DI antibodies have higher specificity for APS. As a result, routine testing for anti-DI antibodies in clinical practice would enable easy differentiation of subjects carrying clinically meaningful anti-β2GPI antibodies from those individuals with a benign autoantibody profile.

Routine availability of an additional laboratory test would be valuable for risk-stratification of APS patients, because it is well established that APS subjects might be classified into different risk categories by their aPL profile [1, 53]. Indeed, antibodies targeting DI not only are more frequently detected in patients at highest risk but also, when at high titers, enable identification of patients with a more aggressive clinical presentation [38•, 45, 46].

No study has systematically addressed the persistency of anti-DI antibodies and the resultant need to confirm positivity 12 weeks apart, as currently recommended for the three criteria tests by international guidelines [1]. It has recently emerged that triple aPL-positivity at first detection is later confirmed for 98 % of cases, suggesting that robust and consistent autoantibody profiles do not require retesting [54]. Consequently, given the higher anti-DI antibody frequency among triple-positive patients, it can be inferred that positive anti-DI antibodies may have diagnostic value even when tested on a single occasion. This critical subject warrants further clarification in future prospective studies. Despite the high specificity for APS of anti-DI antibodies, it is far too soon to state whether isolated low-titer anti-DI antibodies have diagnostic significance, or whether they instead have limited clinical meaning as isolated low-titer aCL and anti-β2GPI antibodies.

Conclusions

Increasing evidence suggests DI is the most relevant epitope targeted by anti-β2GPI antibodies in patients with autoimmune conditions. Anti-DI antibodies have been consistently revealed to be clinically interesting, being significantly associated with both vascular and obstetric aPL-related events.

Despite the many corroborating findings, it is far too soon to recommend replacement of anti-β2GPI antibody testing with anti-DI antibody assay. There are critical challenges caused by the current lack of standardization of the test, and, compared with the assay detecting antibodies against the whole molecule, tests for anti-DI antibodies have lower sensitivity for APS. Indeed, a low percentage of anti-β2GPI-antibody-positive patients with a formal diagnosis of APS present with autoantibodies reacting with β2GPI epitopes other than DI. These subjects could be misdiagnosed if testing for antibodies against the whole β2GPI was not available.

Although anti-DI antibodies are significantly associated with APS clinical events and with a high-risk aPL profile, there is still no definite prospective evidence that this test may provide stronger risk factor than anti-b2GPI antibodies for aPL-related manifestations. However, anti-DI antibodies can enable more straightforward diagnosis and risk-stratification, possibly leading to a treatment strategy tailored to the individual clinical and laboratory characteristics of each patient.

As a whole, anti-DI antibodies are a very promising tool for managing APS; the coming years will be essential for clearly defining the diagnostic and prognostic value of anti-DI antibodies. We believe that, within a few years, testing for anti-DI antibodies will be part of routine clinical practice.