Cryo-EM Structure of BIVV001 Reveals Coagulation Factor VIII-Von Willebrand Factor D'D3 Interaction Mode

Introduction

Coagulation Factor VIII (FVIII) is a serine protease cofactor that directly interacts with coagulation factors IXa and X on activated platelets, and enhances FIXa activity toward FX by 10⁵. von Willebrand Factor (VWF), via its D'D3 domains, interacts with FVIII and prevents premature deposition on phospholipids until activation by thrombin. Thrombin cleavage at Arg1689 of FVIII promotes VWF dissociation by disrupting the FVIII a3 high affinity interaction with the VWF D' domain. VWF extends the half-life of circulating FVIII from less than 3 hours to ~11 hours in humans.

While crystal structures of FVIII and VWF D'D3 alone have been solved, the atomic details of a formed complex are unknown. We sought to determine the FVIII-VWF D'D3 complex structure by using BIVV001, our investigational new drug currently in clinical trials for the treatment of Hemophilia A. BIVV001 (rFVIIIFc-VWF-XTEN) is a novel fusion protein consisting of single chain B-domain deleted (BDD) human FVIII, the Fc domain of human immunoglobulin G1 (IgG1), the FVIII-binding D'D3 domain of human von Willebrand factor, and 2 XTEN polypeptide linkers. The Fc, VWF, and XTEN linker portions of the molecule are each designed to extend the half-life of FVIII. We anticipated that the tethering of FVIII to D'D3 through the Fc dimer in BIVV001 would stabilize the complex for structural studies. Given the large size of BIVV001, at 312 kDa, we thought it an ideal target for structure determination by single particle cryo-EM.

Methods

We collected a total of 3955 micrographs of BIVV001 embedded in vitreous ice at 81,000x magnification using a Titan Krios electron microscope equipped with a Gatan BioQuantum K3 energy filter and camera operating in super-resolution mode. Preferential particle orientation was a major challenge that was overcome through a variety of methods. Micrograph movies were motion-corrected and summed, and over 2 million candidate particle coordinates were extracted. Repeated rounds of reference-free 2D classification resulted in a set of 1.2 million particles that generated a reasonable ab initio/de novo 3D model. Initial full 3D refinements of this model produced a map at approximately 5 Å resolution, into which available crystal structures can be readily fit. Subsequent iterative 3D refinement and 3D classification resulted in a final map at high resolution, into which an atomic model was built.

Results

The structure of BIVV001 was solved by single particle cryo-EM. D' of VWF interacts with the front face of the C1 and A3 domains of FVIII, consistent with a lower resolution, negative stain EM map (Yee et al. 2015. Blood). Interface residues on FVIII identified in an HDX-MS dataset (Chiu et al. 2015. Blood.) largely correspond to this high affinity interaction. D' protrudes upward from the VWF D3 domain, which sits centrally located between the C1 and C2 domains of FVIII at a 45° tilt. By occupying this position, D3 likely sterically blocks the FVIII C domains from binding to membrane. The VWD3 module of the D3 domain contacts the base of the C1 domain, whereas C8-3 binds to the bottom of the C2 domain. The conserved Ca²⁺ site in VWD3 identified previously (Dong et al. 2019. Blood.) is in the interface with C1. This is consistent with Yee et al., where docking placed D3 below the C domains. In that study, a lack of density between FVIII and VWF D3 in the 3D reconstruction, due to flexibility, prevented the detailed analysis that is possible here. In this study, flexibility in this region is also apparent, as C2 is less well ordered than the rest of FVIII and VWF D3 is the least well-ordered portion of the resolved structure. The XTEN linkers are not visible in the final map and were not apparent in any 2D class averages. The Fc is absent in most 2D class averages, due to a lack of consistent positioning relative to FVIII. In the rare cases where the Fc is visible, it adopts a preferred position on the back side of FVIII below the A3 protrusion.

Conclusions

The structure of BIVV001 has been solved by cryo-electron microscopy to high resolution. Alignment with previous results and the averaging out of BIVV001 elaborations suggests the structure obtained here likely represents WT FVIII-D'D3. This structure demonstrates how VWF D'D3 prevents premature FVIII deposition on phospholipids. The structural basis of type 2N von Willebrand Disease mutations in D'D3 can be readily interpreted. Next steps include solving a FVIII-D'D3 dimer structure at high resolution.