Abstract

Recombinant factor IX (rFIX) expressed in Chinese hamster ovary (CHO) cells has been shown to be safe and effective in clinical studies, but differs in pharmacokinetics from plasma-derived FIX (pdFIX). In clinical studies, CHO-derived rFIX had a 30–50 % lower in-vivo recovery when compared to pdFIX, whereas mean residence time and terminal half-life did not differ between preparations. Although rFIX shows high similarity to pdFIX in structure and function, differences in glycosylation and gamma-carboxylation degree can be detected. Moreover, although experimental proof has yet to be published, the lower degree of phosphorylation of amino acid serine 155, and the lower degree of sulfation of tyrosine 158 have been hypothesized to be causative for the lower in-vivo recovery of rFIX. These two modifications occur at less than 20 % for the tyrosine-sulfation and at less than 1 % for the serine phosphorylation in rFIX, whereas pdFIX has both modifications to more than 90 % completed.

We identified human HEK293 cells to perform rFIX phosphorylation and sulfation to a higher extent than CHO cells. A rFIX-producing cell line derived from HEK293 cells was generated by stable transfection, and was adapted to suspension culture conditions to allow lab-scale fermentation. rFIX was produced and purified from a single fermentation run using two different down-stream process schemes: the first was able to enrich high-phosphorylated and -sulfated rFIX; the second to purify total rFIX from the supernatant at high yield.

For pharmacokinetic comparison, these HEK293 materials, CHO-derived rFIX, and a pdFIX preparation were formulated in the same buffer. Determination of phosphorylation and sulfation by mass spectrometry showed a phosphorylation and sulfation degree of 50 % plus a 20 % single modification (phosphorylation or sulfation) for the HEK293-material purified by the modification enrichment method versus 15 % for both modifications plus a 15 % single modification for the material purified by the high-yield protocol. The values for CHO-derived rFIX and pdFIX were similar to those in the literature. Oligosaccharide mapping revealed glycosylation differences among CHO-, HEK293-, and pdFIX preparations, but high similarity between both HEK293-derived materials.

We compared the pharmacokinetics of the various FIX preparations in FIX-knock-out mice. In-vivo recovery and area under the curve were statistically significantly higher for the high phosphorylated and sulfated HEK293-material than for total rFIX derived from HEK293 cells. However, these two parameters were lower for both HEK293-derived rFIX preparations than for CHO-derived rFIX, and lower for CHO-derived rFIX than for pdFIX. This may be due to glycosylation differences between these FIX preparations. Mean residence times and terminal half-lives were similar for all preparations. In summary, these findings emphasize that the degree of rFIX-sulfation and -phosphorylation influences the pharmacokinetic properties of rFIX.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author