(675c) Multiscale Modeling of Monoclonal Antibody (mAb) Production and Glycosylation in a Chinese Hamster Ovary (CHO) Cell Culture Process

Authors: 
Luo, Y., University of Delaware
Price, J. V., University of Delaware
Lovelett, R. J., Princeton University
Radhakrishnan, D., University of Delaware
Barnthouse, K., Janssen Research and Development
Schaefer, E., Janssen Pharmaceutical Companies of Johnson and Johnson
Hu, P., Janssen Research and Development
Lee, K. H., University of Delaware
Shivappa, R., Janssen Research and Development
Ogunnaike, B. A., University of Delaware
The production of recombinant therapeutic monoclonal antibodies (mAbs) using cultured mammalian cells accounts for approximately $80 billion in global sales annually. These antibodies are often produced using Chinese hamster ovary (CHO) cell lines that execute the necessary post-translational modifications (e.g., glycosylation) for the drug to be therapeutically efficacious. Glycosylation is an intracellular, enzymatic process by which glycans (i.e., sugar molecules) are attached to a specific location on the antibody. The structure of the glycans attached to the mAb affects the therapeutic function of the molecule, making glycan distribution a critical quality attribute. Consequently, the ability to predict how variations in process conditions affect both product formation and glycosylation is important from both a process development and process control viewpoint. A multiscale, mathematical model describing CHO cell growth and antibody production was developed in MATLAB to provide a quantitative understanding of how to manipulate a cell culture process to improve antibody titer and control glycosylation effectively. At the macroscopic (bioreactor) scale, the model uses Monod growth kinetics to describe cell growth, nutrient/metabolite concentrations, and mAb production; at the microscopic (antibody) scale, the glycosylation process in the Golgi apparatus is modeled using a reaction network governed by the Michaelis–Menten enzyme kinetics. Although both macro- and micro-scale processes are dynamic, disparate time scales make it possible to solve the (fast) glycosylation model as a static function of the (slowly changing) macro-scale state variables. In this multidisciplinary study, we will present a design of experiments (DoE) approach for (1) identifying significant factors affecting glycosylation—including concentrations of asparagine, glutamate, and copper in the media, and (2) using these factors as process inputs. Model predictions are validated against an independent data set from a representative industrial mammalian cell culture process. Ultimately, the models we discuss will improve process development time by identifying media components and/or process parameters that are most influential to glycosylation during antibody production and will be valuable for designing robust, model-based control systems for the production of biopharmaceuticals.