Differential Glycosylation Between Recombinant Factor VIII Produced in Baby Hamster Kidney and Chinese Hamster Ovary Cells Confers Differences in Immunogenicity in a Humanized Hemophilia Α Mouse Model

Introduction

The formation of factor VIII (FVIII)- neutralizing antibodies is the most critical complication in the treatment of hemophilia A (HA). Recent clinical evidence suggests that recombinant FVIII (rFVIII) produced in baby hamster kidney (BHK-rFVIII) cells is more immunogenic than that produced in Chinese hamster ovary (CHO-rFVIII) cells. This difference in FVIII immunogenicity may be attributed to differences in protein glycosylation, which can impact the removal of FVIII from circulation through mechanisms leading to clearance and antigen presentation. Here, we document significant differences among the 25 potential N-linked glycans between these products, and we provide in vivo animal model-based evidence that supports these clinical observations.

Methods

Factor VIII lectin binding was assessed by ELISA to detect exposed glycans. Commercially-available rFVIII products were adsorbed at 1 ug/mL and specific glycan moieties were detected using a panel of biotinylated lectins and HRP-conjugated streptavidin. Confirmation of differences and determination of N-linked glycan structures was conducted by LC-MS/MS.

Eight to 12 week old transgenic C57BL/6 HA mice were used in these studies. This model contains a murine f8 exon 16 KO and additionally harbors a human F8 transgene with a R593C point mutation. While these mice have undetectable plasma levels of human FVIII antigen and activity, they are tolerant to intravenously infused human rFVIII. Clearance was assessed following a 6 IU (~240 IU/kg) infusion of each rFVIII product, and FVIII activity was measured by chromogenic assay and normalized to a 5-minute time point. The number of interferon (IFN)-γ secreting splenocytes from rFVIII-primed naïve mice was determined by ELISPOT.

FVIII immune responses were elicited by subcutaneous infusion (6 IU twice-weekly for 2 weeks) and adjuvant-coupled intravenous infusion (1 ug lipopolysacharride with the first infusion as described above) with either rFVIII product. Week 5 plasma samples were assessed for FVIII-specific IgG by ELISA, and FVIII inhibitors by one-stage clotting assay.

Results

Lectin binding showed that rFVIII produced in BHK cell lines exhibit lower proportions of high-mannose glycans (p<0.01), and greater levels of sialic acid capping (p<0.01) and fucosylated glycans (p<0.01). Mass spectra confirmed higher levels of sialic acid and identified two additional N-linked sites bearing high mannose glycans on CHO-rFVIII. In this mouse model we observed that BHK-rFVIII had a circulating half-life of 6.06 hours compared to the 10.01 hour half-life of CHO-rFVIII (p<0.0001).

The immunogenicity of the BHK- and CHO-rFVIII products was next evaluated in vivo. In mice primed with a single 6 IU dose of BHK-rFVIII, we identified a higher proportion of FVIII-specific IFN-γ secreting splenocytes after seven days. Furthermore, long-term studies showed that 100% of mice subcutaneously exposed to BHK-rFVIII developed anti-FVIII IgG compared to the 47% that received CHO-rFVIII (p<0.01). Coincidently, when FVIII inhibitors were measured, we observed an incidence of 100% vs 37% (p<0.01), respectively. While the titres of FVIII-specific IgG were higher in mice exposed to BHK-rFVIII (p<0.01), there were no significant differences in the inhibitor concentrations. Similarly, we observed increased titres of FVIII-specific IgG in mice exposed intravenously (1 ug LPS with the first infusion) to BHK-rFVIII compared to CHO-rFVIII. However, there were no differences in the incidence of FVIII-specific IgG, nor in the incidence and concentration of inhibitors between the intravenously-infused mice.

Conclusions

Our results demonstrate that BHK-rFVIII exhibits altered pharmacokinetic and immunogenic properties compared to CHO-rFVIII in humanized hemophilia A mice. The observed early increase in the proportion FVIII-specific IFN-γ producing cells in the spleen suggests an intrinsic immunogenic element of BHK-rFVIII. Similarly, the substantially increased immunogenicity of BHK-rFVIII in mice when treated subcutaneously complements the previously-reported clinical evidence. These differences may be attributed to the significant disparities in N-linked glycosylation, most notably high mannose and sialic acid containing glycans. Additional studies are underway to directly address the role of the these specific glycans and their potential impact on immunogenicity of rFVIII.