Co- and posttranslational engineering of the therapeutic glycoprotein erythropoietin with unnatural amino acids

Abstract

Erythropoietin (EPO) is the most important drug to treat different types of anaemia arising from chronic kidney disease, cancer or AIDS. Four of the top ten biotech drugs are a form of recombinant EPO and the market for EPO is steadily increasing. EPO contains three N-linked and one O-linked glycosylations at the positions N24, N38, N83 and S126, which account for 40% of the molecular weight. Glycosylation is very important for the protein stability against thermal and proteolytic degradation. This posttranslational modification enhances the circulatory half-life and therefore the in vivo activity of the protein. Until now, therapeutic glycoproteins are produced in eukaryotic cells resulting in a heterogeneous glycosylation pattern. This structural heterogeneity of N-linked oligosaccharides encumbers the correlation of glycan structure with glycoprotein function. In addition, therapeutic proteins are often PEGylated, as this modification increases the solubility, protects from proteolytic degradation and enhances the circulatory half-life. However, PEGylation has the big disadvantage of being unspecific or even random. Herein, a semi-synthetic approach is presented for the production of EPO with well-defined and uniform glycan structures at specific positions. As bacterial cells do not posttranslationally glycosylate proteins, E. coli cells were used to incorporate non-natural amino acids into EPO by amber stop codon suppression methodology. These non-natural amino acids bear a specific bio-orthogonal chemical function. For example, the pyrrolysine derivative Plk has an alkyne group. After expression and purification, EPO could be coupled at the natural glycosylation sites to defined, synthesized oligosaccharides or purchased PEG chains by copper-catalysed 1,3-dipolar Huisgen cycloaddition between alkynes and azides. Each individual step from synthesizing the unnatural amino acids, amber stop codon suppression in E. coli, purification, refolding to click chemistry had to be optimised in order to produce homogeneously glycosylated or PEGylated EPO in high yields and in a convenient, low-cost manner. The generated EPO variants were characterised biophysically by mass spectrometry and circular dichroism (CD). All of them displayed the correct molecular mass and the secondary structure of EPO is not disturbed by incorporated unnatural amino acids or coupled decorations. Moreover, CD spectra measurements showed a large protective effect of one short glycan (under 2 kDa) or of a 5kDa-PEG on the secondary structure of EPO upon repeated freezing-thawing cycles. The biological activity was investigated by cell differentiation and proliferation assays with different cell lines. It could be shown that all EPO variants have a positive effect on cell differentiation of haematopoietic stem cells from mouse bone marrow. Results that are more detailed were found in cell proliferation assays. In contrast to small PEG-chains, even one coupled glycan with a molecular weight of 1.7 kDa is sufficient to decrease the EC50-value, and consequently, to increase the biological activity of EPO. Two terminal sialic acids further augment this effect. To sum it up, a new procedure for engineering glycoproteins, such as EPO, is presented, which combines non-canonical amino acids and click chemistry. With this new methodology in hand, first steps were undertaken towards elucidating the impact of each glycosylation position and pattern on the function of EPO, as an example for therapeutic glycoproteins.