Marco Thomann

In Vitro Glycoengineering of IgG1 and Its Effect on Fc Receptor Binding and ADCC Activity

The importance and effect of Fc glycosylation of monoclonal antibodies with regard to biological activity is widely discussed and has been investigated in numerous studies. Fc glycosylation of monoclonal antibodies from current production systems is subject to batch-to-batch variability. If there are glycosylation changes between different batches, these changes are observed not only for one but multiple glycan species. Therefore, studying the effect of distinct Fc glycan species such as galactosylated and sialylated structures is challenging due to the lack of well-defined differences in glycan patterns of samples used. In this study, the influence of IgG1 Fc galactosylation and sialylation on its effector functions has been investigated using five different samples which were produced from one single drug substance batch by in vitro glycoengineering. This sample set comprises preparations with minimal and maximal galactosylation and different levels of sialylation of fully galactosylated Fc glycans. Among others, Roche developed the glycosyltransferase enzyme sialyltransferase which was used for the in vitro glycoengineering activities at medium scale. A variety of analytical assays, including Surface Plasmon Resonance and recently developed FcγR affinity chromatography, as well as an optimized cell-based ADCC assay were applied to investigate the effect of Fc galactosylation and sialylation on the in vitro FcγRI, IIa, and IIIa receptor binding and ADCC activity of IgG1. The results of our studies do not show an impact, neither positive nor negative, of sialic acid- containing Fc glycans of IgG1 on ADCC activity, FcγRI, and RIIIa receptors, but a slightly improved binding to FcγRIIa. Furthermore, we demonstrate a galactosylation-induced positive impact on the binding activity of the IgG1 to FcγRIIa and FcγRIIIa receptors and ADCC activity.

Fc-galactosylation modulates antibody-dependent cellular cytotoxicity of therapeutic antibodies.

The therapeutic activity of monoclonal antibodies can involve immune cell mediated effector functions including antibody-dependent cellular cytotoxicity (ADCC), an activity that is modulated by the structure of Fc-glycans, and in particular the lack of core fucose. The heterogeneity of these glycostructures and the inherent variability of traditional PBMC-based in vitro ADCC assays, have made it challenging to quantitatively assess the impact of other glycostructures on ADCC activity. We applied a quantitative NK cell based assay to generate a database consisting of Fc-glycostructure and ADCC data from 54 manufacturing batches of a CHO-derived monoclonal antibody. Explorative analysis of the data indicated that, apart from afucosylation, galactosylation levels could influence ADCC activity. We confirmed this hypothesis by demonstrating enhanced ADCC upon enzymatic hypergalactosylation of four different monoclonal antibodies derived using standard CHO manufacturing processes. Furthermore we quantitatively compare the effects of galactosylation and afucosylation in the context of glycan heterogeneity and demonstrate that while galactose can influence ADCC activity, afucosylation remains the primary driver of this activity.

Multi-Angle Effector Function Analysis of Human Monoclonal IgG Glycovariants

Therapeutic performance of recombinant antibodies relies on two independent mechanisms: antigen recognition and Fc-mediated antibody effector functions. Interaction of Fc-fragment with different FcR triggers antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity and determines longevity of the antibody in serum. In context of therapeutic antibodies FcγRs play the most important role. It has been demonstrated that the Fc-attached sugar moiety is essential for IgG effector functionality, dictates its affinity to individual FcγRs and determines binding to different receptor classes: activating or inhibitory. In this study, we systematically analyze effector functions of monoclonal IgG1 and its eight enzymatically engineered glycosylation variants. The analysis of interaction of glycovariants with FcRs was performed for single, as well as for antigen-bound antibodies and IgGs in a form of immune complex. In addition to functional properties we addressed impact of glycosylation on the structural properties of the tested glycovariants. We demonstrate a clear impact of glycosylation pattern on antibody stability and interaction with different FcγRs. Consistent with previous reports, deglycosylated antibodies failed to bind all Fcγ-receptors, with the exception of high affinity FcγRI. The FcγRII and FcγRIIIa binding activity of IgG1 was observed to depend on the galactosylation level, and hypergalactosylated antibodies demonstrated increased receptor interaction. Sialylation did not decrease the FcγR binding of the tested IgGs; in contrast, sialylation of antibodies improved binding to FcγRIIa and IIb. We demonstrate that glycosylation influences to some extent IgG1 interaction with FcRn. However, independent of glycosylation pattern the interaction of IgG1 with a soluble monomeric target surprisingly resulted in an impaired receptor binding. Here, we demonstrate, that immune complexes (IC), induced by multimeric ligand, compensated for the decreased affinity of target bound antibody towards FcRs, showing the importance of the IC-formation for the FcR- mediated effector functions.

High-quality production of human α-2,6-sialyltransferase in Pichia pastoris requires control over N-terminal truncations by host-inherent protease activities

Backgroundα-2,6-sialyltransferase catalyzes the terminal step of complex N-glycan biosynthesis on human glycoproteins, attaching sialic acid to outermost galactosyl residues on otherwise fully assembled branched glycans. This “capping” of N-glycans is critical for therapeutic efficacy of pharmaceutical glycoproteins, making the degree of sialylation an important parameter of glycoprotein quality control. Expression of recombinant glycoproteins in mammalian cells usually delivers heterogeneous N-glycans, with a minor degree of sialylation. In-vitro chemo-enzymatic glycoengineering of the N-glycans provides an elegant solution to increase the degree of sialylation for analytical purposes but also possibly for modification of therapeutic proteins.ResultsHuman α-2,6-sialyltransferase (ST6Gal-I) was secretory expressed in P.pastoris KM71H. ST6Gal-I featuring complete deletion of both the N-terminal cytoplasmic tail and the transmembrane domain, and also partial truncation of the stem region up to residue 108 were expressed N-terminally fused to a His or FLAG-Tag. FLAG-tagged proteins proved much more resistant to proteolysis during production than the corresponding His-tagged proteins. Because volumetric transferase activity measured on small-molecule and native glycoprotein acceptor substrates did not correlate to ST6Gal-I in the supernatant, enzymes were purified and characterized in their action on non-sialylated protein-linked and released N-glycans, and the respective N-terminal sequences were determined by automated Edman degradation. Irrespective of deletion construct used (Δ27, Δ48, Δ62, Δ89), isolated proteins showed N-terminal processing to a highly similar degree, with prominent truncations at residue 108 - 114, whereby only Δ108ST6Gal-I retained activity. FLAG-tagged Δ108ST6Gal-I was therefore produced and obtained with a yield of 4.5 mg protein/L medium. The protein was isolated and shown by MS to be intact. Purified enzyme exhibited useful activity (0.18 U/mg) for sialylation of different substrates.ConclusionsFunctional expression of human ST6Gal-I as secretory protein in P.pastoris necessitates that N-terminal truncations promoted by host-inherent proteases be tightly controlled. N-terminal FLAG-Tag contributes extra stability to the N-terminal region as compared to N-terminal His-Tag. Proteolytic degradation proceeds up to residues 108 – 114 and of the resulting short-form variants, only Δ108ST6Gal-I seems to be active. FLAG-Δ108ST6Gal-I transfers sialic acids to monoclonal antibody substrate with sufficient yields, and because it is stably produced in P.pastoris, it is identified here as an interesting glycoengineering catalyst.

Rec. ST6Gal-I variants to control enzymatic activity in processes of in vitro glycoengineering

Background Glycosylation is an important posttranslational modification of proteins influencing protein folding, stability and regulation of the biological activity. The sialyl mojety (sialic acid, 5-N-acetylneuramic acid) is usually exposed at the terminal position of N-glycosylation and therefore, a major contributor to biological recognition and ligand function, e.g. IgG featuring terminal sialic acids were shown to induce less inflammatory response and increased serum half-life. The biosynthesis of sialyl conjugates is controlled by a set of sugar-active enzymes including sialyltransferases which are classified as ST3, ST6 and ST8 based on the hydroxyl position of the glycosyl acceptor the Neu5Ac is transferred to [1]. The ST6 family consists of 2 subfamilies, ST6Gal and ST6GalNAc. ST6Gal catalyzes the transfer of Neu5Ac residues to the hydroxyl group in C6 of a terminal galactose residue of type 2 disaccharide (Galb1-4GlcNAc). To our knowledge, the access to recombinant ST6GalI for therapeutic applications is still limited due to low expression and/or poor activity in various hosts (Pichia pastoris, Spodoptera frugiperda and E. coli). The present study describes the high-yield expression of two variants of human beta-galactoside alpha-2,6 sialyltransferase 1 (ST6Gal-I, EC 2.4.99.1; data base entry P15907) by transient gene expression in HEK293 cells with yields >100 mg/L featuring distinct mono(G2 +1SA) as well as bi(G2+2SA) sialylation activity. Materials and methods Two N-terminally truncated fragments of human ST6Gal-I (delta89, residues 89-406, and delta108, residues 109-406) were designed for transient gene expression (TGE): Instead of the natural leader sequence and N-terminal residues, both ST6Gal-I coding regions harbor the Erythropoietin (EPO) signal sequence in order to ensure correct processing of the polypeptides by the secretion machinery. Following cloning into pM1MT, expression of the ST6Gal-I coding sequences is under control of a hCMV promoter followed by an intron A. Sialyltransferase assays: 1. Asialofetuin was used as acceptor and CMP-9F-NANA as donor substrate. Enzymatic activity was determined by measuring the transfer of 9F-NANA to asialofetuin. 2. Recombinant humanized IgG1 and IgG4 monoclonal antibodies (mabs), characterized as G2+0SA, as well as desialylated EPO were used as targets in sialylation experiments (30 μg enzyme/300 μg target protein). Both enzyme variants of ST6Gal-I (delta89 and delta108) were used under identical reaction conditions and the sialylation status was analyzed by mass spectrometry.

Two N-terminally truncated variants of human β-galactoside α2,6 sialyltransferase I with distinct properties for in vitro protein glycosylation.

Sialic acid groups of protein N-glycans are important determinants of biological activity. Exposed at the end of the glycan chain, they are potential targets for glycan remodeling. Sialyltransferases (STs; EC 2.4.99) are the enzymes that catalyze the sialic acid transfer from a CMP-activated donor on to a carbohydrate acceptor in vivo. Recombinant expression of the full-length human β-galactoside α2,6 sialyltransferase I (ST6Gal-I) was hampered and therefore variants with truncated N-termini were investigated. We report on the distinct properties of two N-terminally truncated versions of ST6Gal-I, namely Δ89ST6Gal-I and Δ108ST6Gal-I, which were successfully expressed in human embryonic kidney cells. The different properties of these enzymes result most probably from the loss of interactions from helix α1 in the Δ108ST6Gal-I variant, which plays a role in acceptor substrate binding. The Km for N-acetyl-d-lactosamine was 10-fold increased for Δ108ST6Gal-I (84 mM) as compared to Δ89ST6Gal-I (8.3 mM). The two enzyme variants constitute a suitable tool box for the terminal modification of N-glycans. While the enzyme Δ89ST6Gal-I exhibited both ST (di-sialylation) and sialidase activity on a monoclonal antibody, the enzyme Δ108ST6Gal-I showed only ST activity with specificity for mono-sialylation.

Oxidation of M252 but not M428 in hu-IgG1 is responsible for decreased binding to and activation of hu-FcγRIIa (His131)

Oxidation of monoclonal therapeutic antibodies (mAbs) can affect binding to Fc-receptors and potentially influence pharmacokinetics or effector functions like e.g. antibody dependent cellular phagocytosis (ADCP). Recently, it has been demonstrated that binding to FcγRIIa (H131) is affected by methionine oxidation of the Fc-portion but it is currently unknown which methionine is responsible for decreased binding. We separated an oxidized IgG1 monoclonal antibody based on the oxidation state of methionine 252 and analyzed fractionated material in receptor binding experiments as well as in functional (cell-based) assays. Although the unfractionated mixture demonstrated weaker interaction/activation of the receptor, differently oxidized isolated subspecies can lead both to stronger as well as weaker binding and activation of the histidine variant of FcγRIIa.