Hyperglycosylation prolongs the circulation of coagulation factor IX


Gert Bolt, Mammalian Cell Technology, Novo Nordisk A/S, Novo Nordisk Park, 2760 Måløv, Denmark.
Tel.: +45 30756654; fax: +45 44444008.
E-mail: bolt@novonordisk.com

Mutations abolishing or substantially reducing coagulation factor IX (FIX) function lead to hemophilia B, a hereditary bleeding disorder that can be managed by replacement therapy using plasma-derived or recombinant FIX. For prophylaxis, 2-weekly infusions are traditionally recommended. Thus, a modified FIX protein with prolonged circulatory half-life allowing less frequent infusions may improve patient convenience and compliance. Several strategies for reducing clearance of FIX have been described. Recombinant FIX compounds prolonged by glycoPEGylation [1], genetic fusion to albumin [2] or genetic fusion to the IgG Fc domain [3] are currently in clinical development.

In the present study, we investigated the possibility of prolonging FIX by introducing new N-glycans. Previous studies suggest a possible role for N-glycans in determining the clearance of FIX. Enzymatic removal of N-glycans increased clearance of wild-type recombinant FIX [4,5], and a correlation between N-glycan sialic acid content and terminal half-life of recombinant FIX has been reported [6]. N-glycosylation is an intracellular process carried out by the oligosaccharyl transferase complex, which glycosylates asparagine residues in N-X-S/T motifs of nascent proteins. Thus, N-glycans can be added to a recombinant protein by introducing mutations that establish N-glycosylation motifs in the amino acid chain. This strategy has previously proven efficient in prolonging the circulatory half-lives of erythropoietin, follicle stimulating hormone, interferon alpha and growth hormone [7].

Wild-type FIX has two N-glycans at positions 157 and 167. We introduced additional N-glycosylation sites in the FIX protein by site-directed mutagenesis of human FIX-coding cDNA. The extra N-glycosylation sites were introduced at positions that we believed allowed for the presence of a glycan without severely interfering with the biological functions of FIX. An expression vector encoding FIX-T172N-K228N-I251T-A262T was established. Besides the wild-type glycosylation sites, the FIX-T172N-K228N-I251T-A262T variant also has N-glycosylation sites at amino acid positions 172, 228, 249 and 260 and thus contains a total of six N-glycosylation sites. Clonal CHO-K1 cell lines producing FIX-T172N-K228N-I251T-A262T were generated using the procedures previously described for FVII variants [8]. A clone exhibiting a favourable combination of FIX productivity and degree of Gla domain gamma-carboxylation was selected, and FIX-T172N-K228N-I251T-A262T was purified from cell culture supernatant as previously described for wild-type FIX [1]. This purification procedure combines an immunoaffinity step using a monoclonal antibody recognizing the functional conformation of the FIX Gla domain and an anion exchange chromatography step, allowing selective isolation of fractions with a desired gamma-carboxylation profile. Only FIX with 10–12 gamma-carboxy groups have fully functional Gla domains [9], and according to anion exchange HPLC as previously described [1], the Gla domains of the FIX-T172N-K228N-I251T-A262T proteins in the present preparation did indeed harbour 12 (82%), 11 (17%) or 10 (1%) gamma-carboxy groups. In SDS-PAGE and HPLC, the FIX-T172N-K228N-I251T-A262T proteins were homogenous, with a molecular weight increase of 11.3 kDa compared with wild-type recombinant human FIX (BeneFIX®; Pfizer, New York, NY, USA). Thus, all four extra N-glycosylation sites in FIX-T172N-K228N-I251T-A262T appeared to be fully utilized. Degraded or activated FIX was not seen in the preparation of hyperglycosylated FIX. According to one-stage FIX clot analysis carried out as described by Østergaard et al. [1], the specific activity of FIX-T172N-K228N-I251T-A262T was reduced to 18% when compared with wild-type recombinant human FIX (42 U mg−1 compared with 235 U mg−1 for BeneFIX®, Pfizer). The mechanisms behind the reduced specific activity of the present FIX variant remain to be determined. Both steric hindrance and repulsion due to the negative charge of sialylated glycans have been implicated in the reduced activity of other hyperglycosylated proteins.

For pharmacokinetic analysis, groups of 15 hemophilia B mice (strain B6.129P2-F9tm1Dws/J) were given bolus injections of 1.0 mg kg−1 FIX-T172N-K228N-I251T-A262T or 1.5 mg kg−1 wild-type FIX (BeneFIX®, Pfizer) in the tail vein. The animal part of this study was performed according to guidelines from and was approved by the Danish Animal Experiments Council, The Danish Ministry of Justice. Plasma samples were collected according to a staggered design (three mice per time point and three samples per mouse) and quantitatively analysed for FIX antigen by ELISA. Standard pharmacokinetic parameters were estimated using non-compartmental analysis (NCA) (WinNonlin, Pharsight Corporation, Mountain View, CA, USA) (Fig. 1). The terminal half-life of the hyperglycosylated FIX variant was increased 2.4-fold relative to wild-type FIX (from 9 to 22 h), the clearance was reduced 5.4-fold (from 81 to 15 mL h−1 kg−1), and the mean residence time was increased 3.4-fold (from 8 to 27 h). Thus, addition of new N-glycosylation sites substantially prolonged the in vivo circulation of human FIX in hemophilia B mice.

Figure 1.

 Mean ± standard deviation FIX antigen content in plasma of FIX-deficient mice versus time after injection of FIX-T172N-K228N-I251T-A262T (solid line) or recombinant wild-type human FIX (dotted line).

The present study raises questions regarding the mechanism(s) behind the reduced clearance induced by adding N-glycans to FIX. For erythropoietin and growth hormone, the reduced clearance of hyperglycosylated variants is mediated by sialic acid residues at the terminus of N-glycan antennae, and correlation between sialic acid content and circulatory half-life was also noticed for hyperglycosylated follicle stimulating hormone and interferon alpha. The negative charge of sialic acids may reduce both cellular uptake and renal clearance due to repulsion from other negatively charged polysaccharides on clearance receptors and on cell surfaces, including those forming the glomerular filter [7].

The exact mechanisms involved in in vivo clearance of FIX remain to be determined. Binding to collagen IV on endothelial cells seems to play a role in the distribution and clearance of FIX [10,11]. The finding by Peters et al. [3] that circulation of FIX-Fc fusion protein is prolonged compared with wild-type FIX by a mechanism mediated by the neonatal Fc receptor also suggests that uptake by endothelial or other cells is involved in the clearance of FIX.

The amino acid replacements required for introducing glycosylation sites may increase the immunogenicity of FIX. However, the glycans may also shield the new epitopes and reduce the overall immunogenicity by increasing the solubility of the protein.

In the present study, we identify hyperglycosylation as a method for prolonging the in vivo circulation of FIX. Further studies on the role of position, number and composition of extra N-glycans for the clearance, biological activity and immunogenicity of hyperglycosated FIX will be required to assess the full potential of hyperglycosylation as a strategy for reducing the dosing frequency of pharmaceutical FIX.


We thank D. Riis, A.M.B. Marnow, M. Nielsen, P. Justesen and K. Vinther for excellent technical assistance.

Disclosure of Conflict of Interests

All authors are employees of Novo Nordisk.