Anti-phosphatidylserine/prothrombin antibodies: an additional diagnostic marker for APS?

Among the diagnostic assays for anti-phospholipid syndrome (APS), lupus anticoagulant (LA) is the strongest predictor of thrombosis; however, it presents several limitations as interference with anticoagulant therapy and poor inter-laboratory agreement. Two-thirds of LA activity is apparently due to antibodies against prothrombin (PT), usually detectable by ELISA. Binding of PT to phosphatidylserine (PS) has been shown to enhance solid-phase anti-PT assay sensitivity. To determine the prevalence of antibodies against PS/PT (aPS/PT) in APS, we tested the semiquantitative QUANTA Lite^® aPS/PT ELISA in a cohort of 80 APS patients. The prevalence of aPS/PT was 81.3 %, rising to 87.6 % when considering LA-positive subjects only. We observed a strong correlation between aPS/PT and LA (p = 0.006). To note, APS patients with thrombotic manifestations displayed significantly higher IgG aPS/PT titers compared to 20 aPL asymptomatic carriers (p = 0.012). To rule out a possible cross-reactivity of anti-β2 glycoprotein I antibodies (aβ2GPI) with PS/PT complex, we tested two monoclonal aβ2GPI antibodies and an affinity-purified (AP) polyclonal aβ2GPI IgG obtained from the serum of a patient reacting against both β2GPI and PS/PT. The two monoclonal antibodies did not show any reactivity against PS/PT complex, similarly the AP IgGs did not react toward PS/PT antigen while preserved their aβ2GPI activity. Our findings suggest that aPS/PT are a definite antibody population in APS. Moreover, the good correlation between aPS/PT ELISA and LA may support its use as a surrogate test for LA, particularly useful to overcome the technical limitations of the functional assay.     

Immunologic Detection of Phosphatidylserine Externalization during Thrombin-Induced Platelet Activation

The antiphospholipid antibody syndrome is characterized by circulating antiphospholipid antibodies against cardiolipin (CL) and phosphatidylserine (PS) and clinically associated with a high risk of spontaneous thrombosis. Three monoclonal antibodies that differentiate between CL or PS were tested against resting and thrombin-activated platelets by flow cytometry. Each antibody reacted differently with CL and PS; 3SB9b reacted with PS, D11A4 reacted with CL, and BA3B5C4 reacted with both CL and PS. Activated platelets bound BA3B5C4 and 3SB9b, but not D11A4. The BA3B5C4-reactive epitope appeared earlier during activation than the epitope reactive with 3SB9b. These data suggest that antibodies against PS are reactive with activated platelets and that two immunoreactive forms of PS are sequentially expressed on platelets during activation.     

Human monoclonal anti-DNA antibodies react as lymphocytotoxic antibodies

Two out of 25 monoclonal anti-DNA autoantibodies that were produced by human-human hybridoma were found to have lymphocytotoxic activity. The antibodies reacted with normal B and T lymphocytes at cold (4 degrees C) as well as at warm (37 degrees C) temperatures. The lymphocytotoxic activity of the monoclonal anti-DNA antibodies could be inhibited by prior incubation of the antibodies with either polynucleotides, e.g. poly(I), poly(dT) or anti-idiotypic antibodies, that had been raised against a dominant anti-DNA antibody. The cross-reactivity between nuclear material and lymphocyte membrane raises the question whether these apparently diverse materials have a shared epitope. The cross-reactivity between anti-DNA antibodies and lymphocyte membrane may account in part for the lymphopenia observed in systemic lupus erythematosus patients.     

Phospholipid-bound beta 2-glycoprotein I induces the production of anti-phospholipid antibodies

Apoptotic-cell-bound beta2-glycoprotein I (beta2GPI), but not apoptotic cells or beta2GPI alone, can induce the production of anti-phospholipid (anti-PL) antibodies (Ab) in normal mice. Although it is presumed that beta2GPI binds to anionic phospholipid (PL) exposed on the apoptotic cell membrane, the precise nature of this complex and its immunogenicity is unclear. To address these issues, we investigated the structure and immunogenicity of human beta2GPI in the presence of different PL that may be expressed on the surface of apoptotic cells. BALB/c mice were immunized intravenously (iv) with beta2GPI in the presence of cardiolipin (CL), phosphatidylglycerol (PG), phosphatidylserine (PS), phosphatidylcholine (PC), or PS/PC (25%/75%) vesicles. Cardiolipin+beta2GPI induced the highest levels of anti-beta2GPI and anti-CL IgG Ab and lupus anticoagulant (LA) activity, while beta2GPI with PC or PS/PC vesicles produced no significant anti-PL Ab. PS+beta2GPI was somewhat immunogenic, but less so than PG+beta2GPI. beta2GPI was immunogenic in the presence of native (CL(N)), but not hydrogenated (CL(H)), CL. Circular dichroism analysis demonstrated that the structure of beta2GPI was altered specifically by interaction with CL(N), but not other anionic PL, including CL(H). Similarly, the structure of CL(N)was affected by interaction with beta2GPI, as detected by(31)P nuclear magnetic resonance. These findings demonstrate that beta2GPI complexed with CL(N)is structurally altered, highly immunogenic, and induces the production of IgG anti-PL Ab. Furthermore, the structural modification and the generation of immunogenic epitopes on beta2GPI upon interaction with CL(N)require the presence of unsaturated fatty acid chains, suggesting a role for oxidation in this process.     

Homogeneous immunoassay for alpha 2 plasmin inhibitor (alpha 2PI) and alpha 2PI-plasmin complex. Application of a sandwich liposome immune lysis assay (LILA) technique

The measurement of the alpha 2 plasmin inhibitor (alpha 2PI) and alpha 2PI-plasmin complex is important for a complete understanding of fibrinolytic conditions. Using monoclonal antibodies against alpha 2PI, a sandwich liposome immune lysis assay (LILA) has been used for the quantitation of alpha 2PI and the alpha 2PI-plasmin complex. In the assay system for the measurement of alpha 2PI, anti-alpha 2PI monoclonal antibodies were covalently coupled to liposomes and specific lysis of liposomes was observed when the liposomes were incubated with the alpha 2PI antigen, TNP haptenized second monoclonal antibody against alpha 2PI and complement activating rabbit anti-TNP antibody. The same liposomes and rabbit anti-plasminogen antibody could be used for the homogeneous determination of the alpha 2PI-plasmin complex. The former assay suggests that monoclonal antibodies lacking complement-activating ability can be used in the sandwich LILA technique. The second application suggests that the LILA technique is capable of measuring heterocomplexes. These assays, which involve the same analytical system, are simple, fast and highly sensitive. They are potentially useful in determining the fibrinolytic status of patients.     

Cross-reactions of nucleic acids with monoclonal antibodies to phosphatidylinositol phosphate and cholesterol

Four monoclonal IgM antibodies to phosphatidylinositol phosphate (PIP), four antibodies to cholesterol and one antibody to liposomes containing phosphatidylcholine, cholesterol and dicetyl phosphate were tested for reactivity with denatured DNA. Three of four antibodies to PIP cross-reacted strongly with denatured DNA. The other antibodies did react with denatured DNA but only very weakly. The binding to DNA was competed by synthetic polynucleotides. In competitive assays, one of the anti-PIP antibodies was particularly reactive with poly(dT) and another with poly(I) and poly(dG). Binding of an anti-cholesterol antibody to ssDNA was also inhibited by poly(I) and poly(dG). Two of the anti-PIP antibodies were also reactive with mononucleotides, and all four bound inositol hexaphosphate. High concns of nucleosides did not compete for binding, indicating that phosphate is involved in the binding site. Phospholipids, particularly those containing inositol phosphate, also competed for binding to DNA, but to varying extents, indicating a variable overlap in the antibody binding site for DNA and phospholipid determinants. These antibodies, induced by immunization with liposomes, showed cross-reactivity characteristics often found with certain types of autoantibodies, but they did not bear the H130 idiotype, which was identified on IgM anti-DNA autoantibodies from MRL-lpr/lpr mice.     

Anti-phospholipid antibodies in patients with Plasmodium falciparum malaria.

Plasma levels of antibodies against phosphatidylinositol (PI), phosphatidylcholine (PC) and cardiolipin (CL) were measured by enzyme-linked immunosorbent assay (ELISA) in patients from malaria endemic area of Sudan and The Gambia. Some Sudanese adults produced IgM antibodies against all three types of phospholipids (PL) during an acute Plasmodium falciparum infection. The anti-PL antibody titre returned to preinfection levels in most of the donors 30 days after the disease episode. IgG titres against PI, PC and CL were low. In Gambian children with malaria, IgM antibody titres against PI and PC were significantly higher in those with severe malaria than in those with mild malaria. These results show that a proportion of malaria patients produce anti-PL antibodies during infection and that titres of these antibodies are associated with the severity of disease.     

Specific Antiphospholipid Antibodies as a Predictive Variable in Patients With Recurrent Pregnancy Loss

PROBLEM: Antiphospholipid antibodies (APLs) consist of very heterogenous autoantibodies. It has not been fully explored what kind of specificities are most relevant to recurrent pregnancy loss. Thus, we investigated the effects of specific APLs on recurrent aborters. METHOD: IgG and IgM antibodies against PE (treated with 1% acetic acid) and five negatively-charged phospholipids were measured by ELISA among 334 recurrent aborters without autoimmune disease. The relationships between APL specificities and subsequent pregnancy outcome were prospectively investigated in 38 recurrent aborters with positive APL who did not receive treatment with prednisolone and aspirin. Antibody levels exceeding the 99th percentile of 280 healthy women were considered positive. RESULTS: Positive IgG and/or IgM APLs were detected in 14%, IgG APLs in 12%, and IgG antibodies against PA, PG, PI, PS, CL and PE, respectively, in 9%, 7%, 7%, 7%, 8%, and 8%. In a prospective study of the 38 untreated patients, fetal loss recurred in 82% of the 33 IgG APL-positive patients, but in 40% of the five patients positive for only IgM APLs. The incidence of fetal loss in the next pregnancy of patients with IgG specific APL-positive against PE, PI, PS, or CI was even higher at 90% and over, and fetal loss recurred in all of 21 patients with two or more IgG APL-positive against PE, PI, PS, or CL. CONCLUSION: These results suggest the possibility that two or more IgG APL-positive value against treated PE, PI, PS, or CL, may be more accurate as a predictive variable than that of only one IgG APL-positive in patients with recurrent pregnancy loss.     

Evidence for different receptor sites in mouse spleen cells for the Sendai virus hemagglutinin-neuraminidase (HN) and fusion (F) glycoproteins

Liposomes were reconstituted from phosphatidylcholine and Sendai virus glycoproteins HN or F and their interaction with mouse spleen cells was studied. Both the HN and F liposomes were able to stimulate chemiluminescence (CL), indicating that the glycoproteins were able to interact with the cell membrane independently of each other. The induction of CL in cells which had been pretreated with liposomes by monoclonal antibodies to either HN or F demonstrated that HN and F bind to the cells independently. The presence of F liposomes on the cell surface was also confirmed by immunoelectron microscopy. Cells pretreated with HN and F liposomes revealed a different pattern of CL when challenged with intact virus or the calcium ionophore A23187 indicating that HN and F bind to different receptor sites.     

Monoclonal IgM Antiphosphatidylserine Antibody Reacts Against Cytoskeleton-Like Structures in Cultured Human Umbilical Cord Endothelial Cells

PROBLEM: It has been proposed that antibodies against phospholipid-dependent antigens (aPLs), induce recurrent pregnancy loss and thrombosis through modulation of endothelial cell function, yet aPLs have not been conclusively shown to bind with endothelial cells. METHOD: Using indirect immunofluorescence we investigated the anti-endothelial cell reactivity of three monoclonal antibodies that differentiate between the phospholipids cardiolipin (CL) and phosphatidylserine (PS): BA3B5C4 (CL+/PS+); 3SB9b (CL-/PS+); and D11A4 (CL+/PS-). Cultured umbilical cord endothelial cells were prepared without fixation or with cold acetone fixation. RESULTS: None of the aPLs reacted with endothelial cells prepared without fixation. 3SB9B reacted strongly with cytoskeletal-like components in acetone-fixed cells, whereas BA3B5C4 and D11A4 were unreactive. The cytoskeletal-like binding of 3SB9b was completely blocked by a monoclonal antibody against vimentin, whereas antibodies against tubulin or actin were not inhibitory. Lipid extraction of the cells destroyed the 3SB9b reactive antigen without affecting the reactivity of anti-vimentin. CONCLUSION: These results suggest that phospholipid-dependent antigenic determinants are not expressed on the surface of resting endothelial cells but that a PS-dependent antigenic determinant is associated with endothelial cell intermediate filaments.     

Monoclonal antibodies reveal extensive antigenic differences between the hemagglutinin-neuraminidase glycoproteins of human and bovine parainfluenza 3 viruses

A panel of twenty monoclonal antibodies to the hemagglutinin-neuraminidase (HN) glycoprotein of human parainfluenza 3 (PI3) virus has been obtained and tentatively classified into four different groups based on reactivity in hemagglutination inhibition (HI), neuraminidase inhibition (NI), and plaque neutralization (NT) tests. The antibodies were tested for cross-reactivity with bovine PI3 virus, Sendai virus, and simian virus 5 (SV5). Only two of these antibodies showed similar reactivities with human and bovine PI3 viruses in HI and NT tests; a few other antibodies showed low levels of reactivity with the heterologous viruses in HI tests. A competitive binding assay further suggested that the two cross-reactive antibodies are directed against the same domain of the HN molecule. Therefore, the HN glycoproteins of human and bovine PI3 viruses appear to be antigenically dissimilar, although they share at least one common epitope.     

Cross-neutralization of staphylococcal and streptococcal pyrogenic toxins by monoclonal and polyclonal antibodies.

We evaluated cross-reactivity of antibodies against staphylococcal and streptococcal pyrogenic toxins. Monoclonal antibodies against staphylococcal enterotoxin (ET) C1 and streptococcal pyrogenic exotoxin (SPE) A were tested for reactivity with homologous and heterologous pyrogenic toxins in vitro. Ten immunoglobulin G1 anti-ET C1 monoclonal antibodies showed little or no cross-reactivity in an enzyme-linked immunosorbent assay, but many of these could neutralize the mitogenic effect of ET B, SPE A, or both. Two immunoglobulin M anti-ET C1 monoclonal antibodies and eight immunoglobulin M anti-SPE A monoclonal antibodies showed extensive cross-reactivity in the enzyme-linked immunosorbent assay and the mitogenicity neutralization assay. No cross-reactivity was observed with SPE C or toxic shock syndrome toxin 1. Rabbits immunized against ET B, ET C1, or SPE A were resistant to challenge with the immunizing toxin. In addition, reciprocal immunity was stimulated by the two ETs, and immunity to SPE A provided protection against ET B but not ET C1. These results show that staphylococcal and streptococcal pyrogenic toxins which share sequence homology have common antigenic determinants which may not be detected in Ouchterlony immunodiffusion assays.     

Anti-mitochondrial type M5 and anti-cardiolipin antibodies in autoimmune disorders: studies on their association and cross-reactivity.

In a series of 42 positive sera, anti-mitochondrial type M5 antibodies (AMA-M5) were found most frequently in patients with SLE (24) and SLE-like syndromes. Patients with AMA-M5 displayed a higher prevalence of thrombocytopenia, thrombosis, biological false positive seroreactions for syphilis, lupus-like anticoagulant activity and anti-cardiolipin antibodies in comparison with a group of 43 SLE AMA-M5 negative patients. The strong association between anti-phospholipid and AMA-M5 antibodies cannot be explained entirely by cross-reactivity between these two groups of antibodies, as indicated by absorption experiments and studies using affinity purified antibody preparations. However, cardiolipin liposomes were able to reduce partially the titres of AMA-M5 sera, suggesting that a small population of AMA-M5 antibodies exists that cross-reacts with cardiolipin. The existence of this population was further substantiated by our demonstration that an IgM monoclonal antibody, from a patient with Waldenström''s macroglobulinaemia, displayed both anti-cardiolipin and AMA-M5 activity, and AMA-M5 activity was completely inhibited by cardiolipin.     

Antiphospholipid Antibody Prevalence in Patients With IVF Failure

Antiphospholipid antibodies (APAs) have been associated with reproductive wastage. The purpose of this study was to establish the prevalence of APAs in women who have had at least 12 embryos transferred during several in vitro fertilization (IVF) cycles without ensuing pregnancy. Sera from 42 women with IVF failure and 42 women who successfully conceived after IVF were tested for the presence of APAs by ELISA. Successful post-IVF pregnancy was determined by obtaining two consecutive rising beta-hCG levels followed by an ultrasound to confirm a viable conceptus. The sera were tested for three isotypes of antibody: IgA, IgG, and IgM against seven phospholipids: cardiolipin (CL), phosphatidyle-thanolamine (PE), phosphatidylinositol (PI), phosphatidic Acid (PA), phosphatidyl-glycerol (PG), phosphatidylcholine (PC), and phosphatidyl-serine (PS). From the IVF failure group, 11/42 (26.2%) were positive for APAs. From the control group, 2/42 (4.8%) were found positive only for IgA against PE. The difference between IVF failure and successful IVF groups was significant (P=0.01). These results suggest that antiphospholipid antibodies should be considered an important marker for increased risk of IVF failure. Patients who are involved with an IVF program should be tested for the presence of APAs prior to initiation of an IVF cycle.     

Modulation of Nonisduced and Phorbol Ester-Induced Generation of Superoxide Anion by Free Liposomes and Liposomes Containing Dexamethasone.

We tested the relative efficacy of free dexamethasone, dexamethasone containing liposomes and free liposomes in preventing superoxide anion, O^-₂ generation by neutrophils. O^-₂ production by 5 ± 10⁵ neutrophils, whether primed or not with lipopolysacchaiide, was stimulated by phorbol 12-myristate 13-acetate (PMA) to 13.4 ± 1.3 nmoles after 15 minutes compared to 1.2 ± 0.3 nmoles with nonstimulated cells. Free liposomes but not dexamethasone (dexa) decreased non-stimulated as well as PMA-induced O^-₂ generation. Dexa-containing phosphatidylcholine from egg yolk: phosphatidylserine from bovine brain (POPS 7:3) liposomes, unlike free dexa, diminished PMA-stimulated O^-₂ production in a dose-dependent manner with a maximal effect at 37.5 μg/ml phospholipid (6.6 ± 1.6 nmoles). The kinetics of cytochrome-c reduction revealed that decreased O^-₂ production resulted from an extended lag-time of release to almost 8 minutes with PMA induction and consequently led to the conclusion that liposomes modified the activity of NADPH oxidase as well as that of protein kinase C. Liposomes prepared with PC and PS of natural origin had a greater inhibitory effect on O^-₂ generation by neutrophils than dipal-mitoylphosphatidylcholine (DPPC) and phosphatidylethanolamine from egg yolk (PE):PC (3:1) liposomes. When 100 μM of Ca²⁺ was added to the medium, the inhibitory action of liposomes prepared with egg yolk PC and DPPC was increased by 30 and 60% respectively, while that of PS and PE:PC was prevented. We also verified that liposomes by themselves, even if phagocytized, did not induce O^-₂ generation or its concentration was too low to be detected by this technique. From the clinical point of view, some formulations delayed non-induced and PMA-induced O^-₂ generation, thus adding to the anti-inflammatory effect of the glucocorticoid they transported.     

Relationship of ability of phospholipids to stimulate growth and bind to macrophages

Among the phospholipids normally present in mammalian cell membranes, the negatively charged phospholipids, phosphatidylserine (PS), phosphatidylglycerol, and cardiolipin, and the neutral phospholipid phosphatidylethanolamine--but not phosphatidylcholine (PC) or sphingomyelin--were found to induce growth of peripheral macrophages. By use of liposomes prepared from PS, which stimulated growth, and PC, which did not, the relation between growth-stimulating activity and binding of the phospholipids to macrophages was studied. The growth-stimulating activity of PS/PC liposomes decreased with increase in their relative content of PC. The amount of PS bound to macrophages also decreased with increase in the proportion of PC in PS/PC liposomes. These decreases in growth-stimulating and binding activities were both partly recovered by additional incorporation into the PS/PC liposomes of cholesterol, which is also a cell membrane component. Phospholipids that stimulated macrophage growth showed high binding ability to macrophages, whereas those that did not stimulate growth scarcely bound to macrophages. Thus the macrophage growth-stimulating activities of phospholipids correlated well with their ability to bind to macrophages. These liposome models should be useful in elucidating the early mechanism of induction of macrophage growth by lipid materials such as cell debris and lipoproteins.     

Monoclonal antibodies to cholera toxin with special reference to cross-reactions with Escherichia coli heat-labile enterotoxin.

Seventy monoclonal antibodies to cholera toxin were prepared and characterized. All were of immunoglobulin G (IgG) isotypes (39 IgG1, 29 IgG2, and 2 IgG3). A total of 61 clones produced antibody directed against the B subunit, and 9 clones produced antibodies with specificity for the cholera toxin A subunit. Among both the anti-B and anti-A antibodies, there were representatives which showed full cross-reactivity with the heat-labile enterotoxin of Escherichia coli (14 clones), others which gave partial cross-reactions (12), and still others (44) which did not cross-react. Although 24 of 25 tested anti-B monoclonal antibodies could neutralize cholera toxin, none of the 9 anti-A clones had any detectable neutralizing ability. Among the anti-B antibodies, those which cross-reacted completely with E. coli heat-labile enterotoxin all had strong cholera toxin-neutralizing capacity, whereas those with lesser or no degree of cross-reactivity varied more in their neutralizing potency. The isolation of monoclonal antibodies that distinguish between enterotoxins of different bacterial origin suggests the possibility of developing immunodiagnostic methods allowing species-specific enterotoxin detection in stools of patients with diarrheal disease.     

Identification of monoclonal antibodies to pancreatic islet cells by immunoperoxidase staining

Immunoperoxidase staining and enzyme-linked immunosorbent assay (ELISA) were used to identify monoclonal antibodies that reacted with pancreatic islet cells. All monoclonal antibodies produced against isolated human or rat pancreatic islets including one mouse autoantibody reacted with pancreatic islets in formalin-fixed pancreas sections, but not with rat kidney or thyroid. Reactivity was also found with suspensions of normal rat islet cells and rat insulinoma cells using a 3-stage immunoperoxidase procedure and an ELISA technique. Differences were observed in staining intensity between the various antigenic substrates tested suggesting variable cross-reactivity and/or number of epitopes. The sensitivity of the immunoperoxidase technique proved to be favourable for identification of monoclonal antibodies that recognize cellular constituents such as islet cell antigens present in low concentrations.     

In Vitro and In Vivo studies of monoclonal antibodies with prominent bactericidal activity against Burkholderia pseudomallei and Burkholderia mallei

Our laboratory has developed more than a hundred mouse monoclonal antibodies (MAbs) against Burkholderia pseudomallei and Burkholderia mallei. These antibodies have been categorized into different groups based on their specificities and the biochemical natures of their target antigens. The current study first examined the bactericidal activities of a number of these MAbs by an in vitro opsonic assay. Then, the in vivo protective efficacy of selected MAbs was evaluated using BALB/c mice challenged intranasally with a lethal dose of the bacteria. The opsonic assay using dimethyl sulfoxide-treated human HL-60 cells as phagocytes revealed that 19 out of 47 tested MAbs (40%) have prominent bactericidal activities against B. pseudomallei and/or B. mallei. Interestingly, all MAbs with strong opsonic activities are those with specificity against either the capsular polysaccharides (PS) or the lipopolysaccharides (LPS) of the bacteria. On the other hand, none of the MAbs reacting to bacterial proteins or glycoproteins showed prominent bactericidal activity. Further study revealed that the antigenic epitopes on either the capsular PS or LPS molecules were readily available for binding in intact bacteria, while the epitopes on proteins/glycoproteins were less accessible to the MAbs. Our in vivo study showed that four MAbs reactive to either the capsular PS or LPS were highly effective in protecting mice against lethal bacterial challenge. The result is compatible with that of our in vitro study. The MAbs with the highest protective efficacy are those reactive to either the capsular PS or LPS of the Burkholderia bacteria.     

Antibodies against phosphatidylinositol and inositol monophosphate specifically inhibit tumour necrosis factor induction by malaria exoantigens.

The active component of the exoantigens of malarial parasites which stimulates macrophages to secrete tumour necrosis factor (TNF) has been shown to depend upon a phospholipid, the activity of which was blocked by phosphatidylinositol (PI) and inositol monophosphate (IMP) in competitive inhibition studies. Antisera made against the exoantigens of Plasmodium yoelii, which inhibited their induction of TNF, were found by an ELISA assay to contain antibody against several other phospholipids. However, the inhibitory antibody was removed specifically by adsorption with liposomes containing PI, but not other phospholipids. Furthermore, PI was the only phospholipid in non-liposomal form which induced the production of inhibitory antisera. Mice immunized with IMP, but not inositol, also produced inhibitory antisera. When incorporated into liposomes several other phospholipids did give rise to inhibitory antibodies but, in contrast to the antisera against parasite exoantigens, PI and IMP, the inhibitory activity was removed by adsorption with heterologous phospholipid liposomes, suggesting that it was directed against a common determinant, presumably the phosphate ester head group. Inhibitory antibodies in the antisera tested were predominantly IgM and titres were not increased after repeated injections. Antisera raised against PI, IMP or the cross-reacting phospholipid liposomes also inhibited TNF secretion by macrophages stimulated by exoantigens of the human parasites P. falciparum and P. vivax, but not by bacterial lipopolysaccharide. These findings confirm our conclusion that exoantigens from these different species contain phosphate bound to inositol in their TNF-inducing moiety.