Development of mathematical modelling for the glycosylation of IgG in CHO cell cultures
Time: Thu 2020-06-04 15.00
Subject area: Biotechnology
Doctoral student: Liang Zhang , Industriell bioteknologi, AdBIOPRO, VINNOVA Competence Centre for Advanced Bioproduction by Continuous Processing
Opponent: Professor Michael J. Betenbaugh, The Johns Hopkins University / Department of Chemical & Biomolecular Engineering
Supervisor: Docent Véronique Chotteau, Industriell bioteknologi, AdBIOPRO, VINNOVA Competence Centre for Advanced Bioproduction by Continuous Processing
Chinese hamster ovary (CHO) cells are the most popular expression system for the production of biopharmaceuticals. More than 80% of the approved monoclonal antibodies (mAbs) or immunoglobulin G (IgG) are produced with these cells. Glycosylation is a usual post- translational modification important for therapeutic mAbs. It affects their stability, half-life and immunological activities. Substantial studies have shown that glycosylation can be affected by the culture conditions in manufacturing, e.g. pH, temperature and media components. To achieve a good control of the glycosylation, a number of mathematical models have been developed. However, most of them have been developed for the cell line engineering, while very few can be used to design the media components for matching a given glycoprofile.
This thesis presents developments of mathematical modelling for glycosylation prediction and experimental design of feeding different combinations of carbon sources in CHO cell cultures. The first study investigates the impacts of mannose, galactose, fructose and fucose to the IgG glycoprofile. Specifically, we look at intracellular nucleotide sugars in fed-batch cultures, where glucose is absent and lactate is used as complementary carbon source. The second study is based on the concept of elementary flux mode (EFM) and the mass balance of the glycan residues. A mathematical model named Glycan Residue Balance Analysis (GReBA) is developed for the prediction of the glycosylation profiles of IgG in pseudo perfusion cultures by feeding combinations of glucose, mannose, galactose and lactate. The model is further optimized for a feeding strategy design of perfusion cell cultures to obtain a desired glycoprofile. In the last study, a probabilistic graphic model based on Bayesian network (BN) is developed for glycosylation prediction in cultures under different multiple variable factors affecting the glycosylation.
The results show that the manipulation of different sugars in the media can be used to control the glycosylation. Both the GReBA and PGM models exhibit abilities for glycosylation prediction and experimental design.