Glycosylation Engineering of Biopharmaceuticals

Sylvie Juliant, Marylêne Lévêque, Pierre Cérutti, Annick Ozil, Sylvie Choblet, Marie-Luce Violet, Marie-Christine Slomianny, Anne Harduin-Lepers, Martine Cérutti

Nowadays, recombinant proteins are used with great success for the treatment of a variety of medical conditions, such as cancer, autoimmune, and infectious diseases. Several expression systems have been developed to produce human proteins, but one of their most critical limitations is the addition of truncated or nonhuman glycans to the recombinant molecules. The presence of such glycans can be deleterious as they may alter the protein physicochemical properties (e.g., solubility, aggregation), its half-life, and its immunogenicity due to the unmasking of epitopes.The baculovirus expression system has long been used to produce recombinant proteins for research. Thanks to recent methodological advances, this cost-effective technology is now considered a very promising alternative for the production of recombinant therapeutics, especially vaccines. Studies on the lepidopteran cell metabolism have shown that these cells can perform most of the posttranslational modifications, including N- and O-glycosylation. However, these glycan structures are shorter compared to those present in mammalian proteins. Lepidopteran N-glycans are essentially of the oligomannose and paucimannose type with no complex glycan identified in both infected and uninfected cells. The presence of short N-glycan structures is explained by the low level of N-acetylglucosaminyltransferase I (GNT-I) activity and the absence of several other glycosyltransferases, such as GNT-II and β1,4-galactosyltransferase I (β1,4GalTI), and of sialyltransferases.In this chapter, we show that the glycosylation pathway of a lepidopteran cell line can be modified via infection with an engineered baculovirus to "humanize" the glycosylation pattern of a recombinant protein. This engineering has been performed by introducing in the baculovirus genome the cDNAs that encode three mammalian glycosyltransferases (GNT-I, GNT-II, and β1,4GalTI). The efficiency of this approach is illustrated with the construction of a recombinant virus that can produce a galactosylated antibody.