Second Generation Adeno-Associated Viral Vector Using a Novel Factor VIII Variant with Improved Secretion Achieves Higher Factor VIII Expression in Non-Human Primates

Abstract

Hemophilia A (HA) is an X-linked bleeding disorder caused by a deficiency in functional factor VIII (FVIII). Adeno-associated virus (AAV)-mediated gene therapy for HA is actively in clinical development, primarily utilizing codon-optimized (CO) wild type (WT) B-domain deleted human FVIII (hFVIII) transgenes. Limitations of these approaches highlight the need to develop novel second generation AAV approaches that improve transgene expression while reducing the vector dose with the fundamental goal of achieving durable levels of FVIII in the normal range while minimizing the potential risk of anti-AAV immune responses and genotoxicity. One of the main challenges of HA gene therapy development is that hFVIII is not expressed efficiently in vitro or in vivo compared to other proteins of similar size. Notably, the FVIII proteins of other species, such as canine and porcine, have superior expression. Previously, we reported that canine FVIII is expressed at higher levels than hFVIII which prompted us to conduct a comparative study between these two proteins (Sabatino 2009). Those studies led us to engineer and characterize a series of FVIII variants around the furin cleavage recognition site (1645-48)(Nguyen 2017) and a secondary cleavage site at 1657-58 in the acidic region 3 (a3). After investigating 15 variants, we identified our best performing variant, hFVIII-Δ3-SP/DE, carrying a deletion of three residues at the furin site (1645-47) (Δ3) (Nguyen 2017) and two amino acid substitutions S1657P and D1658E at the a3 cleavage site. With these minimal modifications, hFVIII-Δ3-SP/DE is secreted significantly more efficiently than WT hFVIII with and without codon-optimization and in conjunction with other improvements in the expression cassette. In the setting of AAV gene therapy, we used AAV8 to deliver WT hFVIII-CO or variant hFVIII-Δ3-SP/DE-CO driven by the modified transthyretin (TTRm) promoter and compared levels of FVIII expression in vivo. Immuno-deficient HA/CD4 KO mice administered AAV8-TTRm-hFVIII-Δ3-SP/DE-CO expressed FVIII levels that were 2-3-fold higher than animals delivered WT FVIII-CO. To investigate this variant in a large animal model, we previously injected AAV8-hFVIII-CO and AAV8-hFVIII-Δ3-SP/DE-CO (2x1012 vg/kg) in an HA dog model tolerized to hFVIII. We observed that the levels of hFVIII expression using the variant were 2 to 4-fold higher than WT hFVIII. Given the success of our variant in HA mice and dogs, we investigated hFVIII-Δ3-SP/DE in non-human primates (NHPs). Animals were negative for neutralizing antibodies to AAV8 prior to vector administration. We delivered AAV8-TTRm-hFVIII-CO or AAV8-TTRm-hFVIII-Δ3-SP/DE-CO to male African green monkeys via intravenous infusion at three vector doses (2x1012, 5x1012, 1x1013 vg/kg)(n=3 per dose per AAV construct). hFVIII antigen levels were measured by hFVIII specific ELISA and are expressed as a percentage of normal levels. Two weeks after AAV administration, NHPs treated with low dose (2x1012 vg/kg) of AAV8-hFVIII-CO expressed a median of 7.4% (range 5.2-8.8%) and AAV8-hFVIII-Δ3-SP/DE-CO expressed 19.2% (range 11.4-72.0%). The mid-dose cohort (5x1012 vg/kg) expressed 17.7% (range 11.8-128.2%) after AAV8-hFVIII-CO and 38.2% (range 12.1-97.0%) after AAV8-hFVIII-Δ3-SP/DE-CO delivery. At 1x1013 vg/kg the AAV8-hFVIII-CO treated NHPs expressed 17.9% (range 12.6-54.3%) while the AAV8-hFVIII-Δ3-SP/DE-CO expressed 114.2% (range 20.2-195.6%). As noted in clinical studies, there is some variability in hFVIII expression in our cohorts. Nevertheless, our data suggests that the delivery of the WT hFVIII transgene did not show a dose response. In contrast, there is a dose-dependent increase in hFVIII-Δ3-SP/DE expression, implying improved secretion correlating with higher levels of FVIII expression. As anticipated, anti-hFVIII antibodies developed to the non-species specific transgene in 16 of 18 animals after 3 weeks post vector administration, leading to a decline in hFVIII expression in most animals. In summary, consistent with our observations in HA mice and dogs, AAV delivery of the Δ3-SP/DE variant in NHPs results in 2 to 5-fold higher dose-dependent levels of FVIII expression than the WT hFVIII. This novel hFVIII variant provides the opportunity to overcome the challenge of inefficient FVIII secretion to improve the efficacy of AAV gene therapy for hemophilia A.