A Novel Humanized Hemophilia-a Mouse Model to Facilitate Preclinical In Vivo Studies of Human Specific Fviiia-Mimetic Bispecific Antibodies

Introduction: To improve treatment options for Hemophilia A patients with and without Factor VIII inhibitors, we and others are developing bispecific antibodies that replace the function of FVIII by bringing together activated FIX and FX to activate FX and restore normal hemostasis. Thus far, preclinical in vivo evaluation of these novel reagents has been limited to non-human primates, due to their lack of species cross-reactivity.

To accommodate pharmacokinetic, pharmacodynamic and in vivo efficacy and safety studies of novel FVIIIa-mimetic bispecific antibodies we set out to generate two humanized FIX;FX double knock-in (DKI) mouse models: one with normal murine FVIII activity (hFIX;hFX-DKI) and one with hemophilia A (hFIX;hFX-DKI;mFVIII-KO).

Methods and Results: Human FIX-knock-in (KI) and human FX-KI mice were generated, using classical C57/Bl6 embryonic stem cell targeting. Both hFIX and hFX encoding sequences replaced their murine counterparts with expression of the human genes driven by the intact murine promotor. hFIX;hFX-DKI mice were generated by intercrossing both lines to homozygosity. hFIX;hFX-DKI;mFVIII-KO mice were generated by CRISPR/Cas9 mediated knockout of murine FVIII in hFIX-KI mice, prior to intercrossing with hFX-KI mice. Both hFIX;hFX-DKI and hFIX;hFX-DKI;mFVIII-KO mice are fertile and viable.

Plasma protein expression levels of both hFIX and hFX in single-KI, DKI, and hFIX;FX-DKI;mFVIII-KO mice were determined by human specific ELISAs and were approximately 1 and 2 µg/mL, respectively. These protein concentrations correspond to 20 % FIX and 10 % FX normal human plasma levels. Chromogenic assays showed matching plasma clotting factor activities compared to normal human plasma, i.e. 10-20% activity of hFIX and 5-10% of hFX. Similar data were obtained by one stage assay. Finally, functional FVIII deficiency in the hFIX;hFX-DKI;mFVIII-KO mice was confirmed by both chromogenic and one stage clotting assays.

Importantly, hFIX;hFX-DKI mice showed only slightly reduced whole blood clotting activity compared to wild-type mice by rotational thromboelastometry (ROTEM; mean clotting times: 910 sec and 516 sec, respectively) whereas hemophilic mice showed significant prolonged clotting times (2500 sec). These data suggest a mild hemophilic phenotype due to reduced human clotting factor expression. Nonetheless the hFIX;hFX-DKI mice showed 100% survival in the tail vein transection model compared to nearly 0% survival for HemA and HemB mice. These findings indicate that hFIX;hFX-DKI;mFVIII-KO should be suitable for the tail vein transection bleeding efficacy evaluation of FVIIIa-mimetic antibodies. In contrast, the more severe tail clip bleeding model using hFIX;hFX-DKI mice showed only slightly less bleeding compared to hemophilia mice (median blood loss, 581 µl versus 766 µl), showing that 10% human clotting factor plasma levels compared to wild-type mice (62 µl) is not sufficient to stop bleeding in this efficacy model. The latter finding is in line with reported data showing that near normal FVIII or FIX clotting factor levels are needed in the severe tail clip bleeding model to stop bleeding in HemA or HemB mouse models, respectively (Dumont, Blood, 2012, Toby et al PLOS One 2016).

Conclusions: We generated and characterized novel mouse models enabling pharmacokinetic, pharmacodynamic and in vivo efficacy and safety studies for novel FVIII-mimetic bispecific antibodies. Ongoing and future studies using these novel mouse models aim to elucidate the in vivo mode of action of this new class of non-replacement clotting factor drug candidates as well as their in vivo interactions with other bypass reagents