Factor VIII–antibody structure and membrane binding

The cocrystal structure of antibody 2A9 and factor VIII (fVIII), reported by Gish et al¹  in this issue of Blood, delineates an epitope on the C1 domain and novel mobility of the C2 domain, both relevant to membrane binding.

An autoimmune response to fVIII is more common than to other blood-coagulation proteins. This response is more frequent following infections, inflammatory diseases, and cancer but can occur with no triggering illness. Inhibitory antibodies from this autoimmune response cause bleeding in a disorder called acquired hemophilia A. For another group of patients with inborn hemophilia A, an immune response to infused fVIII is also common. Development of anti-fVIII antibodies leads to frequent and inadequately treated hemorrhages. Thus, the immune response to native and infused fVIII appears to be anomalously strong and is of interest because of the clinical impact, as well as the insights into biology and the biochemistry of fVIII.

fVIII is composed of repeating regions with an A1-A2-B-A3-C1-C2 domain structure. The C1 and C2 domains of fVIII mediate platelet membrane binding, as well as binding to von Willebrand factor (VWF).²  In addition, the C1 and C2 domains have poorly characterized effects on the interaction with fIXa and fX. Further, the C1 domain is the primary motif that interacts with scavenger receptors on dendritic cells, critical to the development of anti-fVIII antibodies.³  The antibody-fVIII structure in this article sheds light on a functional motif of the C1 domain and how an inhibitory antibody may interfere with binding VWF and phospholipid membranes, as well as affecting function of the C2 domain.

The C1 domain of fVIII appears to be the most frequent target of inhibitory antibodies in acquired hemophilia A.⁴  Some of the anti-C1 antibodies inhibit binding to VWF, presumably by interfering with a known VWF-interactive facet on the C1 domain. Other antibodies interfere with membrane binding or interaction with fIXa or fX.⁵ 

The major epitopes of the C1 domain have been mapped into groups A, B, and overlapping AB.⁵  An intensively studied anti-C1 antibody, LE2E9, belongs to the A group.⁶  It was cloned from the lymphocytes of a patient with mild hemophilia A who developed an inhibitor to infused fVIII. LE2E9 binds the native fVIII C1 domain but not the C1 domain with the Arg→His mutation present in the proband. The antibody interferes with fVIII binding to VWF and decreases fVIII activity by 80% to 90% in 1- and 2-stage assays.

The variable region of LE2E9 has an Asn-linked glycosylation site.⁷  Mutagenesis of Asn to Gln eliminated the carbohydrate moiety, with the mutant antibody named TB402. TB402 has the same affinity for fVIII C1 but no longer interferes with VWF binding and impairs fVIII activity by only ∼50% in an fVIII assay. TB402 was developed as a human therapeutic for postoperative thromboprophylaxis. A phase 3 clinical trial showed promising efficacy, but the bleeding rate was somewhat higher than with the fXa inhibitor. The exact mechanism of partial fVIII inhibition for TB402 is not fully understood, and the increased bleeding rate in comparison with an fXa inhibitor is not readily rationalized with the degree of inhibition in a functional assay. Thus, there is a residual degree of mystery about the clinical lack of success associated with TB402. The antibody in this costructure, 2A9, has an epitope that largely overlaps LE2E9 and has similar interference with VWF binding and fVIII function. The reported structure should be largely illustrative for both antibodies.

fVIII binds to membranes via cooperative action of the C1 and C2 domains.²  An fVIII construct lacking the C2 domain retains phosphatidyl-l-serine specificity, presumably via the C1 domain. The membrane-binding affinity of the isolated C2 domain is ∼40-fold lower than intact fVIII and is inhibited by physiologic saline.⁸  Because membrane binding of fVIII is a multistep kinetic process in which a slow second step follows rapid association,⁹  it is tempting to speculate that insertion of the protruding hydrobic spikes of the C2 domain constitutes the second kinetic step. Another interesting function of C2 is its binding to fibrin. Because fVIII binds to membrane phosphatidylserine and fibrin complexed to the α_IIbβ₃ integrin on thrombin-stimulated platelets, flexibility and conformational change of C2 are relevant to this process.¹⁰  Thus, any flexibility or conformational change in the C2 domain that may help to explain the cooperative binding of the C1 domain to thrombin-stimulated platelets or phosphatidylserine-rich membranes is of interest.

The 2A9 epitope is on the same surface and just superior to the membrane-interactive facet of the C1 domain so that it is easy to rationalize modest interference in membrane binding (see figure).²  It is easy to envision 2A9 acting as a wedge between C1 and the membrane, altering the angle at which fVIII binds to a membrane. A mechanism of this type would impact function without directly interfering with fIXa or fX binding. This structure helps to suggest experiments that might illuminate the active mechanism(s).

The cocrystal of 2A9 and fVIII also provides a novel position for the C2 domain relative to the rest of the molecule. It is translated some 20 Å relative to other structures. It is not clear whether the novel position is related to binding of 2A9 or simply illustrative of a previously unknown degree of flexibility for the C2 domain attachment. In either case, it is of interest for the stepwise binding of fVIII to membranes or the cooperative attachment of fVIII to phosphatidylserine and ligated fibrin on platelet membranes.