Characterization of Interaction of Factor VIII with Engineered Variants of a Single-Chain Variable Antibody Fragment

Introduction

Replacement therapy for Hemophilia A requires frequent infusions of Factor VIII (FVIII) due to its relatively short half-life of ~12 h in plasma. Previous attempts to extend this half-life by genetic and chemical modification of FVIII met the barrier of ~20 h, which is a half-life of von Willebrand factor (VWF), a carrier of FVIII in plasma.

A single-chain variable antibody fragment (scFv) KM33 was shown to inhibit FVIII activity, and interactions with VWF and the low-density lipoprotein receptor-related protein 1 (LRP), the major clearance receptor of FVIII (Bovenschen et al, 2005, Blood, 106:906-12). A study indicated that scFv KM33 may prolong the half-life of FVIII in mice to the level exceeding that of VWF half-life (Mertens et al, US Patent 2008, 20080219983A1). This would make scFv KM33 a promising tool for new designs of the longer-acting FVIII products.

Study objective

We aimed to generate a scFv KM33 variant that can delay FVIII clearance but can be removed from FVIII during its activation by thrombin. Such antibody fragment may extend the half-life of FVIII above that of VWF.

Experimental design

We generated three scFv KM33 variants with different linkers connecting the subunits V_L and V_H of the antibody fragment. The linkers contained variants of thrombin cleavage sites identical to those on FVIII. The proteins were expressed using a baculovirus system, purified by Ni-affinity and size exclusion chromatography (SEC), and tested for their properties.

Results

The engineered scFv variants, along with the unmodified KM33, were tested for binding to FVIII by surface plasmon resonance (SPR). All scFv versions demonstrated similar affinity for FVIII (~1 nM). In addition, a selected variant of scFv inhibited FVIII binding to LRP. These showed that the modifications of scFv did not affect its binding to FVIII.

Thrombin treatment of the engineered scFv variants resulted in dissociation of their V_L and V_H domains, verified by SEC. However, the respective rates of thrombin cleavage were slower than that of FVIII. The preparation of a thrombin-cleaved scFv still inhibited the interaction of FVIII with LRP by SPR, similarly to that observed for the unmodified KM33. All variants of scFv inhibited FVIII activity in a thrombin generation assay suggesting that their moiety remained in complex with FVIII upon its activation.

Conclusions

• The rate of thrombin cleavage of sites within FVIII is higher than that of the identical sites within the scFv. This suggests that additional determinants of FVIII (e. g. sulfated tyrosines adjacent to the sites) contribute to the higher rate of cleavage.

• The cleavage of the linker between the V_L and V_H subunits of scFv KM33 results in dissociation of the subunits and breakdown of the antibody fragment. This mechanism is likely applicable to any scFv, and may be useful in a broad range of applications involving such ligands.

• Both subunits of thrombin-cleaved scFv KM33, most likely, re-assemble on FVIII and form a tertiary complex FVIII/V_L/V_H. In turn, thrombin cleavage of the scFv, complexed with FVIII, does not result in its dissociation from FVIII. These indicate that in such design, an scFv should have lower affinity for FVIII to ensure its release from the complex.