The paired basic amino acid cleaving enzyme (PACE)/Furin is a protein convertase system that plays a vital role in several biological processes, including coagulation. The propeptide processing of human FIX by PACE/Furin is a critical posttranslational modification, so cells co-expressing PACE/Furin and FIX are used for production of clinical recombinant protein. In the development of recombinant B-domain deleted (BDD) FVIII for hemophilia A (HA), a 14 amino acid B-domain sequence containing a putative cleavage site for PACE/Furin was retained because it was believed to be critical for intracellular processing and secretion. In contrast to FIX, we report here a surprising detrimental effect of PACE/Furin in FVIII activity and intracellular processing and secretion.
We engineered a human FVIII variant where the PACE/Furin site at residues 1645-1648 was deleted from FVIII-BDD (FVIII-ΔP/F). Notably, FVIII-ΔP/F exhibits a 3-fold increased activity over FVIII-BDD (p=0.0004) in a 2-stage APTT assay. Moreover, the A2-domain dissociation of activated FVIII-ΔP/F was also 3-fold longer compared to FVIII-BDD, suggesting a more stable activated FVIII molecule. The amount of FVIII secreted from stably transduced BHK cells was about 3-fold higher for FVIII-ΔP/F than for FVIII-BDD. Conversely, the amount of intracellular FVIII antigen was lower for FVIII-ΔP/F than for FVIII-BDD.
To confirm that PACE/Furin was implicated in the underlying mechanisms for the observed differences in FVIII secretion, we inhibited PACE/Furin in FVIII-BDD producing BHKs by transducing them with vectors expressing an engineered α1-antitrypsin variant (haat-PDX) that specifically inhibits PACE/Furin. This resulted in a 40% increase in FVIII secretion (p=0.017) and a decrease in intracellular FVIII-BDD, whereas transduction with the haat-wild type control, which does not inhibit PACE/Furin, did not significantly change the amount of FVIII secreted (p=0.32). Importantly, the secretion and intracellular levels of FVIII-ΔP/F were not affected by the inhibition of PACE/Furin by haat-PDX, indicating that the secretion of this FVIII variant does not benefit from further inhibition of PACE/Furin cleavage. Together these data suggest that the increased secretion of FVIII-ΔP/F compared to FVIII-BDD is due to the former circumventing PACE/Furin.
Furin is ubiquitously expressed in mammal tissues. In a stringent cellular model, we used LoVo, a unique human cell line that lacks functional Furin to determine whether expression of FVIII-ΔP/F and FVIII-BDD would differ, as we have observed in cells expressing Furin. Interestingly, the secretion of FVIII-ΔP/F and FVIII-BDD were comparable. This result confirms that FVIII-BDD is secreted better in the absence of Furin. In summary, our novel variant FVIII-ΔP/F exhibits enhanced secretion primarily by bypassing PACE/Furin cleavage; inhibiting this cellular process also enhances the secretion of FVIII. Futhermore in vivo experiments also demonstrated a beneficial effect of FVIII-ΔP/F: HA mice (n=4-7/dose) given adeno-associated viral 8 (AAV8) vectors for liver gene expression of FVIII-ΔP/F resulted in a 3-fold higher circulating FVIII levels than FVIII-BDD-expressing mice (p=0.025).
These exciting results from human FVIII-ΔP/F prompt us to test this variant HA canine model. First we found that recombinant canine FVIII with the entire PACE/Furin site deleted (cFVIII-ΔP/F) had increased activity in a 2-stage aPTT assay compared to wild-type cFVIII-BDD. Injection of cFVIII-ΔP/F effectively corrects the hemophilia coagulopathy in two HA dogs. Further, AAV8 liver gene therapy with cFVIII-ΔP/F in additional two HA dogs at doses of ~6 x 1012 vg/kg, a log lower than previously used for canine FVIII-BDD AAV8 gene therapy, resulted in therapeutic levels of cFVIII and shortening of clotting times. Preliminary data on injection of cFVIII-BDD protein was well tolerated in cFVIII-ΔP/F-expressing dogs.
In conclusion, these data suggest that PACE/Furin cleavage of FVIII hampers protein biological activity. FVIII variants lacking PACE/Furin recognition sequences are secreted more efficiently and exhibit improved hemostatic effects in both protein- and gene-based strategies. Inhibition of PACE/Furin in manufacturing systems for recombinant human FVIII may increase the yields of protein production. Thus these strategies have a strong rationale for translation to HA therapy.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.