Further glycoengineering of the Drosophila S2 insect cell expression system for improved production of vaccines and therapeutics

Glycosylation is one of the main post translational modifications (PTMs) of proteins and can have significant impact on protein folding, stability, activity, half-life, or immunogenicity, and can mediate many important biological events. In this project we used Drosophila S2 cells expression system and tailored glycoengineering with the purpose of improving vaccines and therapeutics. It has been well established that glycosylation of proteins plays an important role in antigen recognition by Antigen Presenting Cells (APCs) through C-type lectin receptors (CLRs) which recognize carbohydrate structures. This has been explored by many research studies for targeting the APCs with the CLRs ligands or with antibodies targeting CLRs. Mannose containing glycans are present on many pathogens and there are several CLRs recognizing various mannose structures.
In Chapter 2 we investigated immunogenicity of high-mannose N-glycans (HM, Man5-Man9) on Receptor Binding Domain (RBD) of the Spike protein of SARS-CoV-2 produced in α-Mannosidase-Ia knock-out (KO) cell line compared to the wild-type S2 cells paucimannosidic (Man3) N-glycans. We compared the glycan structure influence on soluble antigen vaccine formulations as well as when displayed on capsid Virus-Like Particles (cVLPs) known to be an immunogenic antigen delivery system, to see possible synergistic effect. We compared the vaccines by antibody (Abs) titers they elicited in immunized mice, followed by virus neutralization capacity of the sera. We found a strong indication that high-mannose leads to better polarization of immune system when comparing soluble antigen vaccine formulations. The titers measured on ELISA coated with more native-like, HEK293 produced Spike protein, have shown high titers from soluble HM-RBD vaccine and low titers from soluble Man3-RBD. However, titers measured on ELISA with Man3-RBD coat showed much high titers from Man3-RBD vaccine, although still lower than from HM-RBD vaccine. The virus neutralization data has shown correlation with HEK293 Spike coat ELISA antibody titers measurements, rather than with Man3-RBD coat ELISA measurements. This has shown that even though both vaccines elicited high absolute antibody titers, within the soluble vaccine formulation, only HM-RBD led to protective antibodies.
We additionally considered that since β1,2-xylose on N-glycans is not present in humans, it could also be recognized by some CLRs as foreign, and lead to increased immunogenicity. In Chapter 3, we established a cell line with 57% of xylosylated Man3 N-glycans on purified human erythropoietin by stable expression of plant β1,2-xylosyltransferase. This cell line yielded 42% of xylosylated N-glycans on purified RBD. We then tested it in a similar set-up as the high mannose study and showed that soluble xylosylated RBD gave much higher Abs titers than paucimannose RBD on HEK293 Spike coat ELISA. This indicated that high-mannose and xylosylated N-glycans lead to different way of polarizing the immune system in comparison to paucimannose. In both studies, the glycosylation pattern of antigen displayed on cVLPs did not have an influence on vaccine immunogenicity.
In Chapter 5 we made the first steps towards glycoengineering S2 cells to produce short, cancer-like O-GalNAc glycans. Tn antigen and STn antigen O-glycans are prevalent on cancer cells, but not on healthy cells and are therefore a good target for cancer targeting strategies. We have obtained a KO of Core 1 Galactosyltransferase A (C1GalTA) leading to indication of prevalence of Tn antigen glycans in S2 cells.
Many of the expression hosts used for therapeutic protein production has been subjected to efforts of creating complex, human like N-glycans. Especially, glycans terminated with sialic acid contribute to longer serum-half life since they protect the protein from clearance through galactose-recognizing receptor. The most widely used insect cell line, Sf9, has been successfully engineered to produce branched, tetra-antennary galactosylated as well as bi-antennary sialylated N-glycans. However, glycoengineering of S2 cells has not gone this far yet. Natural N-glycosylation of insect cells is of short, paucimannosidic structures, due to active β-N-acetylhexosaminidase (fdl) that cleaves GlcNAc moieties. In Chapter 4, we built on previous work in S2 cells where fdl KO and expression of N-acetylglucosaminyltransferase I and II (Mgat1 and Mgat2) led to 86% of GlcNAcated N-glycans. In this study we have further expressed B4GALT1 with transmembrane domain changed to that of human FUT7 (α-1,3-fucosyltransferase 7) which prevented enzyme cleavage and secretion leading to addition of galactose to the N-glycans. Surprisingly, addition of galactose made a substrate glycan for addition of glucuronic acid and revealed that S2 cells express high levels of glucuronyltransferase(s). Current efforts lie in KO of genes involved in glucuronidation. Nevertheless, in this study we obtained a cell line producing 92% GlcNAcated N-glycans and a cell line producing 78% of N-glycans with at least one galactose.

In summary, in this thesis we showed higher immunogenicity of high-mannose and xylosylated N-glycans on soluble RBD vaccine formulations which acted as a cis-adjuvant. We also created a new O-GalNAc glycosylation altered cell line and made a significant step towards complex human N-glycans. Altogether these results contributed to the glycoengineering efforts of S2 cells and provided insights into glycan-based immunomodulation.