Antiphospholipid antibodies (aPL) occur in patients with antiphospholipid syndrome (APS), systemic lupus erythematosus (SLE), and related autoimmune disorders (1–5). The presence of these antibodies is associated with venous and arterial thrombosis, recurrent intrauterine fetal death, cardiac valvular lesions, and thrombocytopenia. The mechanisms responsible for the production of aPL are not clearly understood.
In 1990, 3 groups independently reported that binding of aPL to cardiolipin (CL) requires the presence of a normal plasma protein cofactor (6–8). This protein was identified as β2-glycoprotein I (β2GPI), also known as apolipoprotein H. β2GPI is a heavily glycosylated single-chain protein consisting of 326 amino acids, with a molecular weight of ∼50 kd (9). It is a non-complement-binding member of the complement control protein repeat superfamily that has in common 4 short consensus repeats (SCRs) of ∼60 amino acids each, also known as Sushi domains. Each domain is formed by 4 disulfide-bonded cysteines in a pattern of Cys1–3 and Cys2–4 (10). β2GPI also has a fifth SCR that forms a modified Sushi domain and contains a PL binding site (11). β2GPI is believed to be part of the epitope to which aPL bind.
In 1992, it was reported that mice and rabbits immunized with heterologous β2GPI produced high levels of aPL in addition to antibodies against β2GPI (12). These antibodies had characteristics similar to those of the autoimmune human aPL, and their binding to PL was enhanced by addition of β2GPI (13). These findings were reproduced by other investigators (14, 15), and β2GPI-induced aPL were shown to be pathogenic in certain strains of mice (16–19).
We hypothesized that in vivo binding of foreign β2GPI to self PL formed immunogenic complexes against which aPL would be produced. To better understand the mechanisms involved in aPL production following immunization with β2GPI, a region of β2GPI that is responsible for the induction of pathogenic aPL was identified. Antiphospholipid antibodies were induced in mice by immunization with GDKV, a 15-amino acid synthetic peptide that spans Gly274–Cys288 of β2GPI, which represents a major PL binding region of the protein (11). To induce immunogenicity, GDKV was conjugated to carrier proteins such as bovine serum albumin (BSA). These aPL were subsequently shown to be pathogenic, as evidenced by their ability to enhance thrombus formation and increase leukocyte adherence to endothelial cells in a mouse model (20).
These findings led us next to hypothesize that viral peptides resembling GDKV might induce aPL, and that these antibodies might themselves be pathogenic. Therefore, aPL were induced in mice by immunization with synthetic peptides of viral and bacterial origin that had sequence and functional similarity to the phospholipid-binding site of β2GPI (TADL, spans Thr77–Glu96 of the 72-kd human adenovirus type 2 DNA binding protein; TIFI, spans Thr101–Thr120 of ULB0-HCMVA [Swiss Protein Database designation] from human cytomegalovirus [CMV]; VITT, spans Val51–Ile70 of US27-HCMVA; and SGDF, spans Ser237–Ser256 of TLPA-BACSU from Bacilus subtilis) (21). The next step was to determine whether these antibodies have functional and pathogenic properties similar to those found in aPL in patients with APS and/or SLE. The question was addressed in the present study by examining the functional properties in vitro and the thrombogenic effects in vivo of monoclonal antibodies (mAb) induced by the viral peptide TIFI (TIFI-induced aPL).
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
Synthetic peptide. TIFI, a 20-amino acid peptide that spans Thr101–Thr120 of ULB0-HCMVA from human CMV, was synthesized with the ABI 430 solid-phase multiple synthesizer (Applied Biosystems, Foster City, CA) at the Biochemistry Department of Emory University, Atlanta, GA. Purity of the peptide was determined by micro high-performance chromatography analysis and matrix-assisted laser desorption/ionization time mass spectrometric analysis. This peptide, like the PL binding site of β2GP (GDKV), contains a lysine-rich region flanked by hydrophobic residues (Table 1). Binding of this peptide to PL was determined in an assay in which the competition of the peptide with β2GP was measured in PL-coated plates, as previously described (21).
Table 1. Amino acid sequences of GDKV and TIFI*
Production of aPL mAb. To study the functional and binding characteristics and the pathogenic effects of TIFI-induced aPL, mAb were developed in PL/J mice (The Jackson Laboratory, Bar Harbor, ME) by immunizing them with TIFI conjugated to BSA (TIFI-BSA) in adjuvant, as previously described (21). Spleen cells from a TIFI-BSA-immunized aPL-producing mouse were fused with Sp2/0 (CRL-2016; American Type Culture Collection, Rockville, MD), a nonsecreting myeloma cell line, using polyethylene glycol (22). Antibody-secreting hybridoma clones were rendered monoclonal by using the limiting dilution method, and mAb were tested for aPL activity with an enzyme-linked immunosorbent assay (ELISA). Ten mAb (IgM isotype) were obtained and characterized for aPL, lupus anticoagulant (LAC) activity, and binding to human umbilical cord endothelial cells (HUVECs) (Table 2). Two TIFI-induced mAb with aPL activity (D3/AC10 and F3/AA4) were used in the in vivo thrombus enhancement and endothelial cell activation studies. The IgM concentrations of the hybridoma supernatants were determined by quantitative radial immunodiffusion using a BIND A RID kit (The Binding Site, Birmingham, UK) (Table 2).
Table 2. Binding and functional characteristics of the 10 TIFI-induced antiphospholipid (aPL) monoclonal antibodies*
|mAb†||aCL||aPS||aPA||aPI||aPG||aPE||LAC‡||IgM concentration, μm/ml|
Determination of aPL activities. Binding to CL and to other anionic phospholipids (phosphatidylserine, phosphatidylinositol, phosphatidylglycerol, phosphatidylethanolamine, and phosphatidic acid) in the mAb preparations was determined by standard ELISA, using alkaline phosphatase-conjugated anti-mouse IgM, as described previously (4, 5). Supernatants (1:5 dilutions) were tested in the presence of β2GPI, using 10% adult bovine serum (ABS) in phosphate buffered saline (PBS) (10% ABS-PBS) as a diluent. The color reaction was stopped when a positive control sample with an activity of ∼100 IgM phospholipid units reached an optical density (OD) of 1.0.
Inhibition experiments. Inhibition of the binding of mAb to CL was determined by ELISA. Antibody samples (0.1 ml) were treated with serial dilutions (6.25–250 μg/ml) of CL in either PBS or 100 μg/ml purified human β2GPI. Samples were incubated for 1 hour at room temperature, then overnight at 4°C, and then were spun at 10,000g for 30 minutes. The anticardiolipin antibody (aCL) activity of the supernatants was determined by ELISA, as previously described (4, 5).
The percentage inhibition of aCL was calculated as follows:
LAC activity. The LAC activity of the TIFI-induced aPL mAb was determined using modified kaolin clotting time (KCT) testing, as previously described (23). Briefly, 50 μl of the test mAb (∼15 μg/ml) was mixed with 50 μl of normal human plasma. Subsequently, 50 μl of a 2% kaolin suspension (Sigma, St. Louis, MO) was added, and the mixture was incubated for 3 minutes at 37°C. The clotting reaction was initiated by adding 100 μl of 0.03M CaCl2, and the clotting time was determined in a semiautomatic BBL fibrometer (Becton Dickinson, Franklin Lakes, NJ). LAC activity was considered positive when the ratio of the clotting time of a test mAb to that of normal control was >1.2 (23).
Detection of anti-endothelial cell antibodies (AECA). For detection of AECA, a cyto-ELISA that uses unfixed HUVECs was performed, as described previously (24). Briefly, HUVECs were seeded in gelatin-coated 96-well microtiter plates (2.5 × 104 cells/well) and allowed to grow to confluence for 1–2 days. Cells were washed with Hanks' balanced salt solution (HBSS) and incubated with blocking buffer (HBSS/0.5% BSA) for 30 minutes at 37°C to prevent nonspecific binding. After additional washing, HUVECs were exposed to the mAb preparations in duplicate for 60 minutes at room temperature. Cells were washed again and incubated with alkaline phosphatase-conjugated goat anti-mouse IgM (Sigma) followed by 3 washes. The substrate p-nitrophenyl phosphate disodium was added to obtain the color reaction. The OD at 405 nm was read in an ELISA plate reader (Bio-Rad, Costa Mesa, CA) after 20 minutes of incubation with the substrate. A positive control and a negative control (mAb with irrelevant specificity) were included in each run. Samples were run in duplicate, and “net” OD values were obtained by subtracting the mean OD readings of blank wells.
In vivo experiments.Animals and injection protocol. Normal male CD-1 (outbred) mice weighing 30–40 gm (Charles River, Wilmington, MA) were used for these studies. The animals were housed in the Animal Care (American Association of Laboratory Animal Care-approved) facilities of the Morehouse School of Medicine. Animals were handled by trained personnel according to Institutional Animal Care and Use Committee guidelines. Mice, in groups of 9, were initially injected intraperitoneally with 1 ml of mAb D3/AC10 or FF3/AA8 at 0 and 48 hours. The control group received equal amounts of a murine mAb of irrelevant specificity. The surgical procedures were performed 72 hours after the first antibody injection.
Analysis of thrombus dynamics: effects of aPL on thrombus formation. Analysis of thrombus dynamics in a mouse model has been described previously (25–27). Briefly, mice were anesthetized 72 hours after receiving the first injection of aPL or control IgM, and the right femoral vein was exposed. The vein was pinched using a standard pressure to introduce injury and induce a clot. Clot formation and dissolution in the transilluminated vein were visualized with a microscope equipped with a closed-circuit video system (including a color monitor and a recorder). When a thrombus reached maximum size, it was measured (in μm2) by digitizing the image and tracing the outer margin of the thrombus. Three to 5 thrombi were induced in each animal, and the mean thrombus area was computed for each group of injected animals. The person performing the surgery and measurements (XL) was blinded as to what treatment had been given to each animal.
Analysis of endothelial cell activation in the microcirculation of exposed cremaster muscle of mice. Activation of endothelial cells was assessed by direct visualization and quantitation of the adhesion of white blood cells (WBCs) to endothelial cells in the microcirculation of the exposed cremaster muscle of mice, as described elsewhere (28–30). Briefly, 72 hours after the first injection, mice were anesthetized and placed in the dorsal position. Each animal's right scrotum was incised, and the cremaster muscle and testicle were gently exposed and placed on a microscope slide. The dynamic events in the microcirculation of the exposed cremaster muscle (thickness 120 μm) can be directly visualized without the use of vital dyes. Thus, the lumens of venules and capillaries can be directly viewed, and the interaction of individual blood cells (e.g., WBCs) with the luminal surface can be quantitatively assessed. After a stabilization period of 30 minutes, the leukocytes that remained stationary on the endothelium for a period of at least 30 seconds were considered to be adhering. Adhering WBCs in 5 different venules (diameter 25–35 μm) were counted under the microscope. The mean values of treated and control groups were calculated and compared.
Statistical analysis. An independent t-test was used to compare the antibody levels in different groups of mice. Student's unpaired t-test was used to compare the mean values of thrombus sizes and number of adhering WBCs between treated and control groups. P values less than 0.05 were considered significant.
- Top of page
- MATERIALS AND METHODS
Antiphospholipid activity and its inhibition. All 10 mAb showed significantly high binding to CL (OD 0.551–0.823) and varying degrees of binding to other acidic PLs (OD 0.070–1.15) in the presence of β2GPI (Table 2). The binding of the mAb to CL was dose dependent. Figure 1 shows the binding of serial dilutions of a representative TIFI-induced aPL mAb preparation (F3/AA4) to CL. Furthermore, binding of F3/AA4 to CL was inhibited by CL micelles in a dose-dependent manner. Liposomes containing 250 μg/ml of CL inhibited binding of F3/AA4 antibody by 59%. When β2GPI was added, inhibition increased to 90% (Figure 2).
Figure 1. Titration of anticardiolipin antibody (aCL) activity. TIFI-induced antiphospholipid (aPL) monoclonal antibody (mAb) F3/AA4 hybridoma supernatants (see Table 1 for concentrations) were serially diluted in 10% adult bovine serum (ABS) in phosphate buffered saline (PBS) (10% ABS-PBS) and tested for aCL activity by enzyme-linked immunosorbent assay. Binding to CL was determined by subtracting the mean duplicate optical density (OD) at 405 nm of blank wells from the mean duplicate OD at 405 nm of mAb for each dilution point.
Download figure to PowerPoint
Figure 2. Inhibition of aCL activity by CL and CL/β2 liposomes. F3/AA4 TIFI-induced aPL mAb were diluted 1:5 in 1% bovine serum albumin-PBS and incubated with increasing concentrations of CL liposomes in PBS (up to 250 μg/ml CL) (•) or with CL liposomes in 100 μg/ml β2-glycoprotein I solution (○) for 1 hour at room temperature and then overnight at 4°C. Samples were centrifuged, and aCL activity in supernatants was determined. Percentage inhibition of binding to CL was determined as described in Materials and Methods. ELISA = enzyme-linked immunosorbent assay. See Figure 1 for other definitions.
Download figure to PowerPoint
LAC activity and AECA. Five of 10 aPL mAb had LAC activity (KCT aPL:control ratio ≥1.2) (Table 2). The 10 mAb preparations showed various degrees of binding to HUVEC by cyto-ELISA (Figure 3).
Figure 3. Binding of TIFI-induced antiphospholipid (aPL) monoclonal antibodies (MoAb) to human umbilical cord endothelial cells (HUVECs) by cyto-enzyme-linked immunosorbent assay (ELISA). Binding of TIFI-induced aPL MoAb to HUVECs by ELISA, expressed as net optical density units at 405 nm, was determined for each MoAb tested (1:5 dilution). Results are expressed as the mean ± SEM (2 determinations).
Download figure to PowerPoint
Effect of aPL mAb on thrombus formation and on activation of endothelial cells in vivo. The pathogenic effects of the TIFI-induced aPL mAb were further evaluated using the in vivo mouse model of thrombus enhancement and the in vivo mouse model of microcirculation described previously (25–30) and in Materials and Methods. Two representative mAb preparations, D3/AC10 and F3/AA4, both of which have aCL and LAC activity, were chosen for these experiments. Mice injected with TIFI-induced aPL mAb showed significantly elevated levels of aPL as determined by ELISA (0.366 ± SEM 0.139 OD units for D3/AC10 and 1.085 ± SEM 0.313 for F3/AA4). Animals injected with control monoclonal preparations were negative for aPL (mean net OD units <0.085).
Mice injected with TIFI-induced aPL mAb (D3/AC10 or F3/AA4) had significantly larger clots when compared with controls (2,608 ± 1,142 and 1,820 ± 911 μm2 for D3/AC10 and F3/AA4, respectively, versus 657 ± 134 for controls; P = 0.0001 and P = 0.002, respectively) (Figure 4). Adhesion of leukocytes to endothelial cells was significantly increased in mice injected with D3/AC10 or F3AA4 compared with controls (14.7 ± 3.0 and 9.5 ± 2.3 versus 5.4 ± 2.1; P = 0.002 and P = 0.003, respectively) (Figure 5). This increased adhesion of leukocytes to endothelial cells is an indication of endothelial cell activation in vivo.
Figure 4. Effects of TIFI-induced antiphospholipid monoclonal antibody (MoAb) on dynamics of thrombus formation in vivo. Mice, in groups of 9, were infused with D3/AC10, F3/AA4, or control MoAb, and surgical procedures were performed to determine the thrombus size in the exposed femoral vein of anesthetized animals, as described in Materials and Methods. Results are expressed as the mean ± SD (n = 9). * = P < 0.05 versus control animals.
Download figure to PowerPoint
Figure 5. Effects of TIFI-induced antiphospholipid monoclonal antibodies (MoAb) on endothelial cells in vivo. White blood cell adhesion (number of adhering leukocytes) to endothelial cells of the postcapillary venules in the cremaster muscle of mice treated with D3/AC10, F3/AA4, or control MoAb was measured, as described in Materials and Methods. Results are expressed as the mean ± SD. * = P < 0.05 versus control mice.
Download figure to PowerPoint
- Top of page
- MATERIALS AND METHODS
This study clearly shows that pathogenic aPL can be generated by immunizing mice with TIFI, a peptide of viral origin that spans Thr101–Thr120 of ULB0-HCMVA from human CMV. Antiphospholipid antibodies generated in this way showed binding characteristics and functional properties similar to those of aPL found in the serum of patients with APS; namely, they bound to CL and to other negatively charged phospholipids in the presence of β2GPI. Furthermore, binding of TIFI-induced aPL mAb to CL was inhibited by CL liposomes in a dose-dependent manner, and this inhibition increased when β2GPI was added to the liposomes.
The 10 mAb showed different degrees of binding to various phospholipids, and only 5 of the 10 mAb had LAC activity. The differences in binding patterns and LAC activities may be attributable to the fact that, although the mAb were obtained by immunization with the same peptide, each aPL-producing hybridoma that was rendered monoclonal by limiting dilution was derived from a different B cell. Therefore, differences in affinity for various phospholipids and differences in LAC activity are expected.
The main goal of this study was to determine whether pathogenic aPL could be generated after immunization with a viral peptide. Therefore, we chose the 2 mAb preparations F3/AA4 and D3/AC10 because they had binding and functional characteristics similar to those found in “pathogenic human aPL”; i.e., strong binding to CL and to other negatively charged phospholipids and positive LAC activity.
In these experiments, all 10 mAb were of the IgM isotype. In humans, IgG aPL rather than IgM aPL are more commonly associated with thrombosis and other clinical complications of APS (31). However, there have been several reports of APS in patients who have only IgM aPL. We have described a woman who had APS and clinical features resembling those of transverse myelitis but had only IgM aPL (32). Furthermore, affinity-purified IgM aPL from 3 patients with APS showed thrombogenic properties in a mouse model (27). Reverter et al reported platelet activation and increased tissue factor expression by 2 pathogenic human IgM monoclonal aPL and 2 affinity-purified IgM aPL (33).
Our findings were confirmed in a recent study reported by Blank et al (34). Using a peptide phage display library, the authors identified 3 hexapeptides that specifically reacted with anti-β2GPI antibodies and inhibited the in vivo properties of anti-β2GPI antibodies (34). They found high homology between the hexapeptides with bacterial products derived from Haemophilus influenzae or Neisseria gonorrhoeae. Mice immunized with products from those bacteria containing the peptides had clinical manifestations resembling those of experimental APS (34).
Previous studies showed that immunization of mice with heterologous β2GPI induced aPL with properties similar to those of aPL found in patients with autoimmune diseases (13, 14). β2GPI-induced aPL have been shown to be pathogenic in certain strains of mice, causing intrauterine fetal death, thrombocytopenia, and thrombosis (15–19). Induction of aPL is not dependent on immunization with the whole β2GPI molecule. Peptides representing the PL binding site of β2GPI, such as GDKV, may also induce aPL production (20). This PL-binding peptide is too small to be immunogenic by itself, and it had to be coupled to larger protein molecules (e.g., albumin) to become immunogenic. We hypothesize that a PL-binding peptide must be combined with a carrier protein for the induction of aPL in vivo. The β2GPI molecule is large enough that no carrier protein is needed for it to be immunogenic. It is possible that the PL-specific site (GDKV) in the fifth domain of β2GPI binds to PL, and that the rest of the β2GPI molecule serves as a carrier protein, thus inducing production of aPL and anti-β2GPI antibodies. This study demonstrates that pathogenic aPL and anti-β2GPI antibodies with properties similar to those of aPL and anti-β2GPI antibodies found in APS patients can be induced by immunization with β2GPI-like, PL-binding, CMV-derived peptides.
It has been demonstrated that aPL can be induced in patients infected with many bacteria, other microorganisms (35–44), and viruses, including CMV (45), varicella zoster virus (46), human immunodeficiency virus (47), hepatitis C (48), and Epstein-Barr virus (49). In addition, transient LAC activity in patients with Epstein-Barr virus infections has been reported (49). In most cases, the presence of these antibodies was not associated with clinical manifestations of APS, and the antibodies did not require β2GPI for binding to CL.
Recently, the belief that infection-associated aPL are β2GPI-independent, do not bind β2GPI in the absence of phospholipids, and are not associated with thrombosis or other clinical complications of APS has been challenged in a significant number of reports (50–52). In one study, aCL present in leprosy patients were shown to be heterogeneous with respect to their β2GPI requirement: in 10 of 31 leprosy sera, the aCL were β2GPI dependent, and 21 of 31 did not require β2GPI for binding to CL (50). Other investigators have shown that these β2GPI-dependent aPL were associated with thrombosis in leprosy (53). In another study, parvovirus B19-associated aCL antibodies were shown to be β2GPI dependent and to behave in a manner similar to that of “autoimmune” aPL (51). Furthermore, some cases of viral hepatitis C-associated aPL have been reported to be complicated with thrombosis (48). A recent report described a young woman with acute CMV infection who developed APS manifested as a common iliac vein thrombosis. IgM aCL appeared with the onset of infection, followed later by IgG aCL (54). Five months later, both IgM and IgG aCL disappeared from her serum. To our knowledge, this was only the second case of APS associated with CMV infection reported in the literature. The above observations support the hypothesis that aPL induced by infectious agents may cause thrombosis.
Because human subjects are commonly exposed to viral products, tolerance to self antigens may be compromised by a molecular mimicry mechanism, and production of aPL and anti-β2GPI may occur. Molecular mimicry has been reported to occur in other autoimmune syndromes and is a particularly attractive model, because it offers an explanation for the initial activation of autoreactive B cells that does not require priming by autoantigen. In support of this model, studies have demonstrated the presence of anti-double-stranded DNA (anti-dsDNA) antibodies in individuals after the onset of microbial infections (55). Furthermore, antiidiotype studies have shown structural similarities between antibacterial and anti-dsDNA antibodies (56–58). Molecular mimicry has also been studied in rheumatic fever, where there is well-documented cross-reactivity between streptococcal M protein and cardiac myosin, and mAb that bind both Streptococcus and myosin and can induce histopathologic alterations in myocytes (59). Autoimmune phenomena, particularly antinuclear antibodies and rheumatoid factor, have also been reported to occur in leprosy (59). A recent report indicated that a correlation exists between chlamydial infection and autoimmune response (60). The study showed evidence of molecular mimicry between an RNA polymerase major sigma subunit from Chlamydia trachomatis and one of the ribosomal proteins frequently targeted by autoantibodies in rheumatic diseases.
Although aPL are present in a large number of infectious diseases, in the majority of cases these antibodies are either transient, or their presence is not associated with clinical manifestations of APS. Cross-reactions between host autoantigens and parasitic microorganism antigens is quite common, but it is perhaps only in the genetically programmed individuals that the antibody production persists and pathologic damage occurs. Genetic factors such as HLA type and complement C4 deficiency (61), sex, and age may also be involved in development of autoimmune disease. Our recent experiments with GDKV-induced aPL mAb suggest that the tertiary structure of these peptides is involved in their recognition by aPL.
In summary, our results indicate that aPL can be generated in animals by active immunization with several phospholipid-binding peptides/proteins, suggesting a possible mechanism of induction of APS (17, 20, 21). These findings suggest that incidental immunization with these viral and bacterial proteins during a subclinical infection may trigger aPL production. The sequence homology of these foreign proteins to the PL-binding region of β2GPI may break tolerance to self β2GPI and induce pathogenic “autoimmune” anti-β2GPI/aPL, though the mechanisms involved in this process are not completely known. However, not all aPL are pathogenic (34), as confirmed by other investigators. Further studies are needed to evaluate the pathogenicity of aPL and to identify the agents responsible for induction of pathogenic aPL. Based on clinical experience suggesting that not all aPL are pathogenic, it is predicted that only a limited number of aPL induced by certain viral products would be pathogenic in certain genetically predisposed individuals. Identification of those agents may help in finding strategies for their elimination, protection from them, or methods for the prevention of aPL production. Alternatively, free peptides may be used to induce tolerance against aPL production.