Functional Assessment of VWD Type 2N Variants Using Mammalian Cell Displayed Surface-Tethered Von Willebrand Factor

Background von Willebrand factor (VWF) is a large plasma protein critical in hemostasis and thrombosis. Defects in VWF cause the inherited autosomal bleeding disorder von Willebrand disease (VWD). One critical function of VWF is to bind circulating coagulation factor VIII (FVIII) and protect it from rapid clearance. Type 2N VWD is a recessive form of VWD characterized by low FVIII levels due to impairment in VWF-FVIII binding. Type 2N VWD clinically is similar to mild-to-moderate hemophilia A, an X-linked bleeding disorder caused by deficiency in FVIII. It is critical to distinguish type 2N VWD from hemophilia A because the treatments are different. VWF genotyping can inform accurate diagnoses, treatment, and testing in relatives. With the rapid expansion of DNA sequencing, a growing number of DNA variants are being identified in the VWF gene. However, the interpretation of significance of these variants is often challenged by insufficient evidence to inform confident classification regarding pathogenicity. In vitro functional assays provide an avenue to develop the evidence needed to accurately assign functional significance to DNA variants. We adapted multiplexed surface tethering of extracellular proteins (MultiSTEP), our mammalian cell surface display system for secreted proteins, to display VWF and test the functional impact of VWF variants on VWF-FVIII binding.

Methods We constructed a VWF cDNA of common VWF coding sequences by mutating rare variants to the more common nucleotides. The regions encoding the VWF propeptide were deleted (∆Pro-VWF). ∆Pro-VWF was cloned into our MultiSTEP cassette that provides a C-terminal flexible linker, Strep II tag, and CD28 transmembrane domain. Three VWF missense variants implicated in type 2N VWD (Table 1) that had varying levels of evidence to support classification of pathogenicity were chosen for study – VWF c.2561 G>A (p.R854Q), c.2572 T>A (p.C858S), and c.2635 G>A (p.D879N). Variants were subcloned into our ∆Pro-VWF cassette and then recombined into the genomically integrated MultiSTEP landing pad in HEK293 Freestyle cells. Following induction of expression, VWF-displaying cells were incubated with recombinant FVIII. VWF expression was measured using a polyclonal anti-VWF antibody, and VWF-FVIII binding was detected with two different monoclonal antibodies to FVIII. Cells were incubated with fluorescently-conjugated secondary antibodies and cell surface VWF and FVIII abundance determined by flow cytometry.

Results We show that ∆Pro-VWF robustly displays in our MultiSTEP system and binds FVIII. All three VWF missense variants express VWF at levels similar to ∆Pro-VWF. Relative to ∆Pro-VWF all three variants show marked impairment in VWF-FVIII binding (Figure 1), providing strong in vitro evidence for pathogenicity in type 2N VWD useful in variant reinterpretation (bold column in Table 1).

Conclusions We have developed an in vitro mammalian cell surface display system for the interrogation of VWF variants and adapted it to study VWF-FVIII binding. The Pathogenic variant p.R854Q shows decreased VWF-FVIII binding in this assay, as expected. The Likely Pathogenic variant p.D879N and the Variant of Uncertain Significance p.C858S both exhibit a profound loss of VWF-FVIII binding in this assay. Our results provide evidence for reclassification of p.C858S to Likely Pathogenic for type 2N VWD. We expect our system can be used at scale to similarly interrogate the functional significance of a multiplexed library of VWF variants and improve variant interpretation in VWD.