Glycoremodeling of monoclonal antibodies

Abstract

Monoclonal antibodies (mAbs) are therapeutic glycoproteins mostly used in the areas of oncology and immunology. The structure of an mAb can be divided into the Fab fragment and the Fc region. The Fc region can be subdivided into the CH1 and the CH2 domain, which contains the glycosylation site. Effector ligands like Fcγ-receptors (FcγRs) and C1q bind to the CH2 domain and instigate a pro-inflammatory immune response that will eventually lead to target cell death. Glycans exert influence on the affinity of the effector ligands for the CH2 domain and are therefore target for modulation. Manipulation of the glycans may be achieved through deployment of chemoenzymatic strategies and the results of those undertakings are described in this dissertation. The introductory chapter provides an overview of the current knowledge of the mechanisms of the pro-inflammatory response, the effect of mutations in the protein backbone of mAbs and the influence of the composition of the glycan are described. Glycans are built up from N-acetylglucosamine (GlcNAc), fucose, mannose, galactose and sialic acid and are generally biantennary in mAbs. The first step in glycan remodeling is often the removal of terminal galactose from the glycan. The galactosidase BgaA from S. pneumoniae was found to be particularly active towards the mAb glycan when compared with other galactosidases from E.coli and A. oryzae. An assay to determine the release of nanomolar quantities released galactose was developed and provided insight into the enzyme kinetics of the galactosidases: BgaA was most active towards the G2-glycan as well as fully galactosylated infliximab, while the other galactosidases preferentially catalyzed lactose. Through a glycosynthase (mutant endoglycosidases) methodology using semisynthetic glycans several homogenous glycoforms of trastuzumab were synthesized carrying a bisecting GlcNAc either with or without core-fucose. Previous research hampered drawing accurate conclusions about the absolute contribution of a bisecting GlcNAc in the mAb glycan as mixtures were used. Binding of the newly synthesized mAbs to a panel of FcγRs was assessed using surface plasmon resonance. It appeared that fucosylation has major impact on the FcγR binding, especially to FcγRIIIa (which is known for a long time), yet bisecting GlcNAc independently had no added effect on FcγR binding. The glycosynthase methodology has several limitations: 100 molar equivalents of the activated glycan to 1 mol of mAb are needed, activated glycans tend to be labile and glycosynthases often possess residual hydrolytic properties. We developed a fully enzymatic procedure to modify the glycans directly on the intact mAb. Both arms of the biantennary glycan can be modified independently to generate asymmetric structures. To demonstrate the versatility of the method a panel of 17 well-defined infliximab and cetuximab glycovariants have been created. Some of the structures bear an alpha-gal epitope, generally considered to be undesired due to its immunogenicity potential. Patient derived IgE binding to the glycoforms gave new, structural insights into the alpha-gal immunogenicity. Lastly, all chemoenzymatic synthesis routes are discussed and a blueprint for future experiments is provided, including strategies to modify glycoprotein with triantennary glycans.