Abstract

The formation of inhibitors against Factor VIII (FVIII) is one of the major adverse effects of FVIII substitution therapy in hemophilia A patients. Previously we have cloned a panel of antibodies from a phage display library from the VH gene repertoire of a single hemophilia A patient (Van den Brink et al, Blood 2002). One single-chain variable fragment (scFv), designated KM33, was found to display remarkably high affinity FVIII binding and FVIII nhibition. The epitope of this scFV was located in the FVIII light chain, in particular within the C1 domain. Using computational epitope prediction methods we identified three putative antibody binding C1 loops. By mutagenesis we found one of these (K2092-S2094) critical for KM33 binding. This region is the counterpart of one of the lipid-binding loops in the FVIII C2 domain. This suggest that KM33 inhibits FVIII activity by interfering in FVIII with membrane binding. Sequence analysis revealed that the scFV fragment KM33 was derived from the germ-line sequence VH 3–30 with a number of substitutions. As somatic mutations have the potential of contributing to affinity maturation and antigen neutralization, we addressed the contribution of individual mutations to the inhibitory properties of scFV KM33. Somatic mutations were reversed to germ-line residues, and purified KM33 variants were analyzed by Surface Plasmon Resonance (SPR) and by solid-phase binding and inhibition assays. Somatic mutations in positions H56 (S to N) and H57 (N to D) within the Complementarity Determining Region H2 (CDR H2) proved neutral with regard to affinity for FVIII. In contrast, the somatic mutations near the CDR H1 (H23 A to V, H24 A to D, and H27 F to L) proved of major impact. The equilibrium binding constant (Kd) of mutant VH-D24A (4.4 nM) and the triple mutant VH-(V23A-D24A-L27F) (10 nM) were 2–3 orders of magnitude higher than that of the wild-type KM33 (Kd = 0.05 nM). The SPR sensorgrams revealed that the difference in affinity resulted mainly from higher dissocation rates for the mutants. Molecular modeling and molecular dynamics suggested that Asp in position H24 plays a key role in that it forms a hydrogen bond network that stabilizes the CDR H1 in the mature KM33 antibody. The stabilization of CDR H1 in KM33 proved essential for its inhibitory properties and its detection as an anti-FVIII antibody. First, the KM33 D24A variant failed to display any inhibitory activity in the classical Bethesda assay, irrespective whether residual FVIII activity was assessed by one-stage assay or by chromogenic assay. Second, this variant tested negative in ELISA based assays for FVIII-binding, but non-inhibitory antibodies, presumably due to its relatively high dissociation rate from immobilized FVIII. This finding may have major implications for the detection of FVIII inhibitors. First, anti-FVIII antibodies may exist that are directed against functional epitopes on the FVIII molecule that escape detection in functional Bethesda assays. The same may hold for binding assays that are designed to detect typical non-inhibitory antibodies. While patients having such antibodies seem virtually inhibitor-free, a single somatic mutation within a pre-existing non-inhibitory antibody may result in the a strong FVIII inhibitory response. This may be particularly relevant in linking FVIII inhibitor development in individual patients to exposure to specific FVIII products.

Author notes

Disclosure: No relevant conflicts of interest to declare.