4E10 and 2F5 monoclonal antibodies: binding specificities to phospholipids, tolerance, and clinical safety issues

Small numbers of human monoclonal antibodies have been used as important models by researchers seeking to obtain broadly neutralizing antibodies to HIV-1 for development of vaccines or immunotherapy. Two of these unusual monoclonal antibodies, 4E10 and 2F5, recognize short amino acid sequences in the membrane proximal external region of gp41 envelope protein of HIV-1. These binding sites are close to the lipid bilayer of the virion, and several laboratories have reported that 4E10 and 2F5 also recognize a variety of phospholipids, including cardiolipin ^[1,2]. The lipid-binding properties of the monoclonal antibodies has kindled an interest in the possible role of phospholipids, and perhaps other types of lipid, as additional antigenic targets for HIV-1 vaccine development ^[3]. However, 4E10 and 2F5 were also proposed to be polyreactive autoimmune antibodies of a type that might be subjected to downregulation by immunological tolerance, and it was suggested that this outcome might represent an escape mechanism ^[4]. Two recent papers in this journal ^[5,6] have addressed several important questions relating to these observations. First, are the binding specificities to phospholipids reproducible? Second, is the previously proposed idea correct that immunological tolerance might make induction of antibodies that bind to phospholipids difficult? Third, does a possible clinical risk accompany the use of 4E10 or 2F5?

Both papers confirmed that 4E10 does bind to cardiolipin and to other phospholipids, albeit with apparently lower affinity than the binding to the gp41 epitope ^[5,6]. Interestingly, however, both papers failed to confirm the two previous reports ^[1,2] that 2F5 binds to cardiolipin, although a new finding of slight binding of 2F5 to another anionic phospholipid, phosphatidylserine, was observed by Vcelar et al. ^[6]. This observation might initially seem to call into question whether 2F5 really binds to anionic phospholipid, but it might also simply reflect technical difficulties because of the likelihood that anionic phospholipid might occupy only a small portion of the 2F5 antigen-binding paratope. It could reflect as well methodological differences among studies that have examined phospholipid binding specificity. However, because of differing results among studies from several laboratories, the existence and possible importance of binding of 2F5 to anionic phospholipids remains an open question.

With respect to tolerance mechanisms, although 4E10 and 2F5 each do react with more than one antigen, Scherer et al. ^[5] noted that the reactivity of 4E10 with different unrelated antigens is limited, and downregulation of these types of antibody might not occur. Indeed, such antibodies might be 'oligoreactive' natural antibodies, as described by Notkins ^[7]. Natural antibodies that bind to phospholipids and to numerous other chemical substances are known to be ubiquitous in normal serum ^[8-12]. Titers of antiphospholipid antibodies are also commonly observed in clinical practice, particularly as an epiphenomenon in the course of numerous types of common infectious disease ^[13-15]. In fact, natural antibodies to various lipids are probably responsible for beneficial effects, including opsonization and removal of pathogens, malignant cells, senescent erythrocytes, apoptotic cells, and inflammatory debris ^[7,11,12].

If natural oligoreactive antibodies to lipids are ubiquitous in human serum, one might then ask why they do not damage normal cells, and why are they not subjected to downregulation by immunological tolerance? One answer is that normal cells are covered with proteins that sterically block binding of antibodies to the underlying lipids. In essence, because of overlying protein, lipids are not 'autoantigens' visible on normal cells but are cryptic antigens that are only revealed by proteolytic enzyme treatment ^[9,10,16] or by appearance in abnormal, senescent, or damaged cells or tissues ^[7,11,12]. In support of this possibility, specific oligoreactive antibodies are easily induced against liposomal lipids, and induction of such antibodies is not inhibited by immunological tolerance ^[3]. The available evidence thus suggests that if lipids do emerge as a specificity that is required for broad neutralization of HIV-1, it would seem likely that downregulation of B cells will not be a major difficulty.

Both papers ^[5,6] also addressed the question of whether 4E10 and 2F5 might be similar, from a clinical laboratory standpoint, to antibodies that appear in the course of so-called antiphospholipid syndrome, a disorder associated with a pronounced tendency toward recurrent venous and arterial thrombosis and abortions ^[15,17]. Antiphospholipid syndrome is a primary autoimmune disease, but it sometimes appears as a secondary manifestation of lupus erythematosus or other autoimmune disease. The suggestive name of the syndrome is unfortunate because the term antiphospholipid syndrome is actually a misnomer. The presumed pathological antibodies are not actually directed against phospholipid but rather are autoantibodies that react with circulating autologous protein antigens, particularly β₂-glycoprotein I, a phospholipid-binding protein ^[15,18,19]. Antibodies that bind directly to phospholipid are not part of antiphospholipid syndrome, do not constitute a pathological finding, and are not associated with thrombosis. Both papers ^[5,6] reported that little binding of 4E10 or 2F5 occurred to β₂-glycoprotein I. The clinical diagnosis of antiphospholipid syndrome currently also requires the actual occurrence of a thrombotic episode in addition to the laboratory tests ^[15,17]. Based on passive immunotherapy of 2F5 and 4E10 in four phase I and phase II trials, and based on in-vitro clinical laboratory assays, both papers appropriately concluded that the clinical risk, if any, of causing antiphospholipid syndrome-related thrombotic events with these antibodies appears to be low.

Finally, a further important question does remain: namely, whether binding to phospholipid actually contributes to the neutralizing activities of 4E10 or 2F5? The answer to this question is a matter of current interest, and it may have relevance for the development of neutralizing antibodies. It has been shown that an induced murine monoclonal antibody to a pure anionic phospholipid, phosphatidylinositol-4-phosphate (PIP), did neutralize primary isolates of HIV-1 in a peripheral blood mononuclear cell assay ^[20], and the binding specificities of 4E10 and this monoclonal antibody to lipids were remarkably similar, including similar binding of each to cardiolipin and PIP ^[20,21]. However, anti-PIP did not neutralize virus in the corresponding pseudovirus assay using a Hela cell-derived target for viral infection. This finding, therefore, raises the possibility that more than one specificity of antibody may be necessary in order to achieve neutralizing effectiveness with different types of cells in the course of HIV-1 infection.

Acknowledgements

The author thanks Dr Barbara M. Alving of the National Center for Research Resources, NIH, for review of the concepts relating to APS and for helpful comments.

Note: Alving is an inventor on HIV-1 vaccine-related patents and patent applications that are assigned to the US Army. The views and opinions expressed herein are the private opinions of the author and do not necessarily reflect the views of the US Army or the US Department of Defense.