Bioengineering Factor VIII B-Domain Sequences Improves Function and Efficacy in Hemophilia A Models.

Abstract 2208

We previously reported that recombinant canine B-domain deleted factor VIII (FVIII) was expressed predominantly (>75%) as a single-chain protein and, upon activation with thrombin, yielded an active cofactor species with increased stability due to delayed A2 domain dissociation. We hypothesized that this could be in part due to a unique recognition sequence (HHQR) for intracellular PACE/Furin protease present in the canine FVIII, but absent in other mammals including human, porcine and murine FVIII where the sequence is RHQR. Here, we changed position 1645 from R to H in the B-domain deleted human FVIII (hFVIII-RH) and found that much more of the protein was expressed in a single chain form (3-fold) compared to hFVIII-wild type (hFVIII-WT). Importantly, the variant hFVIII-RH was biologically active exhibiting a 2-fold higher activity measured in a 2-stage activated partial thromboplastin time (p<0.05) likely due to slower dissociation of the A2-domain upon thrombin activation compared to hFVIII-WT (t_½ 3.48 vs. 1.48 minutes, respectively).

We next sought to determine the potential in vivo efficacy and safety of circulating hFVIII-RH in murine models of hemophilia A (HA) by hepatocyte-restricted transgene expression using adeno-associated viral (AAV serotype 8) vector and by exogenous administration of the recombinant protein. HA mice received varying doses (8×10¹²-4×10¹³ vg/kg) of AAV-hFVIII-WT or AAV-hFVIII-RH and exhibited a corresponding dose-dependent response of FVIII with plateau antigen levels 2–3 fold higher in hFVIII-RH than hFVIII-WT expressing mice (n = 5/group, p < 0.05 at all doses). At the low dose cohort, circulating FVIII antigen levels were 79±6.6 and 50±9 ng/ml for the FVIII-RH and FVIII-WT, respectively. In the mid dose cohort, FVIII levels were 165±54 and 84±9 ng/ml and in the high dose 274±39 and 120±35 ng/ml. We monitored the blood loss following tail-clip assay in groups of mice stratified into groups determined by expression to be low (28–60 ng/mL), mid (60–100 ng/mL), or high (above 100 ng/mL) for analysis (n= 3–9/group). Expression of either hFVIII variant was capable of decreasing the blood loss of mice in the low group compared to untreated HA mice, but did not reach that of hemostatically normal mice. Notably, at the mid expression group only hFVIII-RH expressing mice had corrected blood loss to normal hemostasis (hFVIII-RH vs hFVIII-WT p < 0.02). In the high dose all mice (FVIII-RH and FVIII-WT) exhibited blood loss similar to that of hemostatically normal mice. The increased effect of hFVIII-RH is a result of more stable clot as seen in the FeCl₃- carotid artery injury model. At comparable FVIII levels, 15/16 mice expressing hFVIII-RH form stable occlusive thrombi whereas only 14/21 in hFVIII-WT group.

Laser-induced injury at microcirculation resulted in increased fibrin deposition by 2–3 fold in HA mice expressing FVIII-RH (n=4 mice, 20 injuries) compared to those expressing FVIII-WT (n=3, 10 injuries). Similar findings were obtained upon injection of recombinant protein to achieve ∼40% of normal for both FVIII-RH (n=3, 13 injuries) and FVIII-WT (n=3, 18 injuries). Moreover, we assessed the immunogenicity of hFVIII-RH by administering AAV vectors to transgenic HA mice tolerant to B-domain deleted hFVIII-WT. At 4 weeks post AAV administration all mice exhibited stable hFVIII-WT/hFVIIl-RH levels (n=5/group) with no evidence of inhibitor by Bethesda assay or by mouse specific-hFVIII-IgG titers. In summary, FVIII-RH is superior to FVIII-WT in terms of expression levels and total hemostastic function without evidence of increased immunogenicity. Therefore FVIII-RH is an attractive molecule for protein replacement and gene/cell-based strategies for HA.