Identification of Active Coagulation Factor IX Variants with Insertion or Fusion of Unstructured Polypeptide Xten

Background: A variety of approaches have been used to extend the half-life of human factor IX (FIX) in the circulation, including PEGylation and recombinant fusion of either the Fc domain of IgG or albumin to the C-terminus of FIX. Introduction of XTENs, unstructured hydrophilic polypeptides of defined amino acid composition and low immunogenic potential, represents a promising alternative approach for generating FIX variants with improved pharmacokinetic properties. Moreover, the hydrophilic properties of XTEN have the potential to increase the solubility of FIX and thereby render it suitable for subcutaneous administration, which is currently an unmet need. Because XTEN is introduced into FIX by DNA recombination, the location, length, composition and number of XTEN modifications can be readily varied, and impact of these modifications on the activity and clearance of FIX can be evaluated.

Aim: To identify sites in FIX that can accommodate the introduction of XTENs without abrogating FIX activity and apply this approach to both otherwise non-modified FIX and a recombinant FIX-Fc fusion protein.

Methods: The highly active FIX Padua variant (R338L) was used as a scaffold to counter FIX activity loss due to reduced activity caused by the introduction of XTENs. XTEN insertion within the Gla domain was avoided due to the essential role of the Gla domain in anchoring FIX to phospholipid surfaces and subendothelial type IV collagen. XTEN insertion sites were selected by analysis of available FIX structures in the Protein Data Bank in conjunction with the following criteria: 1) calculated accessible surface area, 2) solvent accessibility assessed by hydrogen/deuterium exchange mass spectrometry (H/DX-MS), 3) exclusion of sites within defined secondary structural elements, 4) preference for positions with significant inter-species protein sequence variability, and 5) exclusion of sites proximal to known hemophilia B mutations. A 42-residue XTEN element (AE-42) was inserted at sites selected by using these criteria or fused at the C-terminus of FIX. FIX activities of these variants were evaluated in conditioned medium of transfected HEK293 cells. Longer XTENs (AE-72, -144 and -288) were then similarly tested at sites shown to be permissive for AE-42 insertion. Finally, based on results obtained with these single XTEN variants, FIX variants with multiple XTEN insertions of varying lengths and at representative sites were evaluated for FIX activity.

Results: Based on the criteria described above, a total of 33 sites in FIX were selected and evaluated by insertion of AE-42. Of these, two in EGF domain 2 (EGF2), one in the EGF2-activation peptide (AP) linker region, four in the AP, and four in the catalytic domain, including the C terminus, were identified as permissive sites by FIX activity assay. Only sites in AP and sites at or close to the C-terminus of FIX tolerated longer XTENs (AE-144, AE-288 or AE-864). FIX activity detected in conditioned medium inversely correlated with the length of XTEN introduced. Four representative sites in FIX were selected to generate a combinatorial library of 79 FIX variants. Three groups, FIX with a single XTEN, FIX with dual XTENs and FIX-Fc with a single XTEN insertion, showed detectable activity, while combination of insertion/fusion at three or more sites abolished FIX activity.

Conclusions: Several permissive sites for XTEN insertion are present in FIX and select combinations of XTEN insertions variants retain FIX activity. Active FIX-XTEN variants identified here are candidates for pharmacokinetic characterization in hemophilia B mice.