Deletion of alanine 2201 in the FVIII C2 domain results in mild hemophilia A by impairing FVIII binding to VWF and phospholipids and destroys a major FVIII antigenic determinant involved in inhibitor development

The factor VIII (FVIII) C2 domain consists of the 170 carboxy-terminal residues of the molecule. It contributes to FVIII binding to von Willebrand factor (VWF)^1-3  and to phospholipids (PLs),^4,5  and thereby determines the stability in plasma and the activity of FVIII. More than 30 mutations associated with mild/moderate hemophilia A have been identified in this domain.⁶  However, in most cases, it remains unclear by which mechanism these mutations reduce plasma FVIII levels.

The crystal structure of the C2 domain was recently determined.⁷  At one side of the domain, hydrophobic residues were predicted to insert into the PL membrane, whereas basic residues above the hydrophobic surface interact with negatively charged PL by electrostatic interactions⁷  (Figure 1). The 2-dimensional crystallography of FVIII bound to PL,⁸  and site-directed mutagenesis⁹  has provided data in agreement with these predictions.

The C2 domain is frequently targeted by FVIII inhibitor antibodies occurring in patients with hemophilia A treated with FVIII concentrates.^10-12  In addition, some point mutations, located in the amino-terminal part of the C2 domain of patients with mild/moderate hemophilia A, are associated with an increased incidence of inhibitor development.¹³ 

Despite the importance of C2 for FVIII function and antigenicity, only limited information is available regarding alterations of structure and function of the C2 domain in hemophilia A. We, therefore, screened plasma from patients with mild/moderate hemophilia A for alterations in the antigenic determinant recognized by a human monoclonal antibody to the C2 domain.¹⁴  We also investigated the interaction of such mutated FVIII molecules with PL and VWF.

Plasma was obtained, under institutionally approved informed consent, from healthy donors and 111 patients with mild hemophilia A.

B-domain–deleted recombinant (BDD-r) healthy Ser2119Tyr and Del2201 FVIII were produced as described.¹⁵  For Del2201 FVIII, mutagenesis was performed by using the primers 5′-ACC-AAT-ATg-TTT-ACC-Tgg-TCT-CCT-TCA-AAA-gCT-CgA-3′ and 5′-TgA-Agg-AgA-CCA-ggT-AAA-CAT-ATT-ggT-AAA-gTA-ggA-3′. BDD-rFVIII production was evaluated by using transiently transfected Chinese hamster ovary (CHO) cells, and BDD-rFVIII preparations were partially purified from VWF-free supernatant of stably transfected CHO cells.¹⁵ 

FVIII activity (FVIII:C) was measured in a 1-stage clotting assay using partial thromboplastin time (PTT) automate (Stago, Parsippany, NJ) or by using a chromogenic assay (Coatest FXa generation assay; Chromogenix-Instrumentation Laboratory SpA, Milano, Italy) as recommended for BDD-rFVIII.¹⁶  FVIII:Ag was measured by enzyme-linked immunosorbent assay (ELISA).¹⁵  The inhibition of plasma FVIII by purified BO2C11 was evaluated as described.¹⁵ 

FVIII binding to VWF in plasma was studied by using a centrifugation assay.¹⁵  The binding of recombinant FVIII (0.5 IU/mL) to VWF was evaluated in the presence of various concentrations of purified VWF, which was bound to anti-VWF antibodies coated on Sepharose.¹⁵  FVIII interactions with PLs were evaluated by using a modification of the Coatest FXa. Aliquots of plasma in 30-μL samples diluted 200-fold in Tris (tris(hydroxymethyl)aminomethane) 50 mM, pH 7.3, 0.2% bovine albumin, or BDD-rFVIII were mixed with 30 μL PL at various concentrations for a 10-minute incubation at 37°C. At each PL concentration, FVIII activity in the test sample was determined by comparing FXa generation with that obtained by using known concentrations of healthy plasma FVIII.¹⁵  For each FVIII variant, results were expressed as the percentage of the FVIII activity measured at the highest PL concentration. Small unilamellar PL microvesicles representative of an activated platelet surface were used.^17-19 

To investigate the effect of mutations responsible for mild/moderate hemophilia A on antigenic determinant(s) in the C2 domain, we checked whether such mutations could prevent FVIII inactivation by BO2C11. The antibody inhibited more than 95% of plasma FVIII from all but 2 patients (Figure 1B). Both patients carried a deletion of Ala2201 (Del2201) in the C2 domain. The effect of this novel mutation is likely explained by the location of Ala2201 close to residues Met2199 and Phe2200, both shown by crystallography to mediate FVIII contact with BO2C11²⁰  (Figure 1A).

The 2 patients carrying Del2201 are unrelated at least up to the second generation. FVIII:C ranged between 11% and 33%, with FVIII:Ag between 35% and 58%. No difference in FVIII:C activity was recorded by using either the 1-stage or the chromogenic FVIII assay. One patient (43 years old) experienced several bleeding episodes, including spontaneous and postoperative ocular angioma bleeding and gingivorrhagia, whereas the other patient (18 years old) suffered from frequent bruising and epistaxis. In case of minor surgery, both patients could always be efficiently treated with DDAVP (1-deamino-8-D-arginine vasopressin) and, therefore, never received FVIII concentrates.

Because BO2C11 inhibits FVIII binding to VWF and PLs, we investigated Del2201 FVIII interactions with VWF and PLs. In control assays, the fraction of healthy FVIII bound to VWF was 85% and that of a FVIII with mutation Ser2119Tyr in the C1 domain was 15%, in agreement with published values.¹⁵  For both patients with Del2201, plasma FVIII bound to VWF was less than 55% (Figure 1C). A reduced binding of FVIII to VWF was also observed when FVIII was measured by using a FVIII:Ag ELISA (data not shown).

The functional alteration of Del2201 FVIII was further studied by using BDD-rFVIII. In the presence of VWF, the rate of production of Del2201 BDD-rFVIII was lower than that of healthy BDD-rFVIII. By contrast, in VWF-free medium, the rates of production and the FVIII:C/FVIII:Ag ratio of both FVIII proteins were similar (Figure 2A-B). In a FVIII 1-stage assay, the activity of both BDD-rFVIII proteins was 50% lower than in the chromogenic assay, in agreement with observations made with another BDD-rFVIII.¹⁶  To evaluate the binding of BDD-rFVIII to VWF, we measured the capture of rFVIII by various concentrations of VWF bound to Sepharose. Between 10- and 500-fold more VWF was required to capture 25% of Del2201 and Ser2119Tyr FVIII, respectively, than of healthy FVIII (Figure 2C). Similar results were obtained when FVIII bound to VWF was evaluated by measuring FVIII:Ag (data not shown). These observations confirm the role of the C2 domain in FVIII binding to VWF.^1-3  Given the importance of VWF for FVIII stability, the mild reduction of Del2201 FVIII binding to VWF could contribute to low plasma FVIII levels.

Del2201 FVIII interaction with PLs was evaluated in a FXa generation assay, using synthetic PL vesicles representative of activated platelet surface. There was a good correlation between the FVIII:C obtained at 4 μM PL and that measured in standard FVIII:C assays. However, reducing PL concentrations resulted in a progressive decrease in FVIII activity in plasma of both patients carrying the deletion compared with the effect on healthy plasma samples. At the lowest PL concentration still allowing a significant FXa generation (1.5 μM PL), Del2201 FVIII:C was reduced 3-fold relative to healthy plasma (Figure 1D). Comparison of healthy and Del2201 BDD-rFVIII provided the same results as with healthy and patient plasma FVIII (Figure 2D).

In conclusion, Del2201 in the FVIII C2 domain eliminates a major antigenic determinant of the FVIII molecule and impairs FVIII interaction with both VWF and PLs. This first report of a mutation occurring in patients with mild/moderate hemophilia A and altering FVIII binding to PLs raises the question of whether mutation in the FVIII PL binding site can lead to mild hemophilia A, while escaping detection using conventional FVIII:C assays. However, as previously shown by site-directed mutagenesis,⁹  it is possible that any mutation in the FVIII PL binding site impairs FVIII binding to VWF, thereby resulting in low FVIII:C levels detectable with routine FVIII:C assays.