422442 A Model-Based Strategy for Understanding and Improving Recombinant Protein Glycosylation in CHO Cells

Thursday, November 12, 2015: 4:43 PM
151D/E (Salt Palace Convention Center)
Sarah Fadda1, Philip M Jedrzejewski1,2,3, SI Nga Sou1,2,3, Ioscani J. del Val1, Sarantos Kyriakopoulos1, Karen M. Polizzi2,3 and Cleo Kontoravdi1, (1)Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London, United Kingdom, (2)Department of Life Sciences, Imperial College London, London, United Kingdom, (3)Centre for Synthetic Biology and Innovation, Imperial College London, London, United Kingdom

Glycoprotein therapeutics represent an increasing sector of the pharmaceutical industry. Glycoproteins are large complex molecules, which are heterogeneous in terms of their glycan moieties. This heterogeneity arises from a sequence of complex post-translational enzymatic modifications that take place in the ER and Golgi apparatus and play an important role in drug safety, efficacy and half-life. It has been shown that the process of protein glycosylation can be partially driven by the intracellular availability of nucleotide sugars (NSs), which are the co-substrates to the glycosylation reactions. It is also known that the availability of the NSs is affected by process conditions including cell culture mode, temperature, dissolved oxygen tension, nutrient availability and addition of precursor molecules [1,2].

In this work, we describe a model-based platform for understanding the culture factors that affect recombinant protein glycosylation in Chinese hamster ovary (CHO) cells, the workhorse of the biopharmaceutical industry. Our platform constitutes a multi-scale model describing the process at three different scales and through different approaches. Specifically, the higher scale model is a cell culture dynamics (CCDyn) model, which describes cell growth and the variation of the nutrient and product concentrations over time in the extracellular medium. The specific production and consumption rates (qi) in the extracellular medium are used as inputs for an intracellular compartment (cytosol) model.

The cytoplasmic compartment has been described following different approaches. The first approach, termed the Jedrzejewski model [3], consists of a semi-structured purine and pyrimidine synthesis network, which describes the synthesis of nucleotides based on the concentrations of extracellular metabolites, and a structured and mechanistic representation of the NSD synthesis pathway, based on the murine metabolic network as depicted in the Kyoto Encyclopaedia of Genes and Genomes and depending on the concentrations of sugars in the extracellular medium. The latter model follows the bottom-up approach because the kinetic pathway has been described in its completeness according to the individual enzyme data as found in BRENDA.

On the contrary, the second approach known as the del Val model follows a top-down modelling method because the kinetic pathway has been reduced through the adoption of a macroscopic reaction scheme (macroreactions). Both models describe the temporal profiles of nucleotides as well as NSDs concentrations in the cytoplasm in response to extracellular feeding.  They both are theory-based models that are based on biophysical equations that require many, sometimes difficult-to-measure kinetic parameters.

For this reason, the intracellular compartment has been described also through a fully determined dynamic metabolic flux model (dMFA) which describes, in a parameter-free manner, the most significant metabolite fluxes in the cytoplasm comprising a more reduced version of a genome-scale model. The metabolic flux network consists of 55 material balances and 55 metabolite fluxes. Eight NSs and outlets for host cell protein N-linked and O-linked glycans as well as mAb glycosylation have been considered. The outputs are the intracellular metabolite fluxes varying with respect to time as a function of changes in the extracellular environment.

The lowest scale model is the del Val glycosylation model [4] describing the N-linked glycosylation process of the antibody heavy chain through the Golgi apparatus viewed as a plug flow reactor. Given the concentrations or fluxes of the NSDs from the higher scale model in the cytosol, this model is capable of reproducing glycosylation profiles of commercial mAbs.

Our platform has been validated against experimental data from two IgG-producing CHO cell lines representing all aforementioned scales. The platform allows one to capture the effect of different amino acid feeding strategies [5], as well as that of hexose and nucleotide precursor additions to culture media and their impact on NSD concentrations in the first place, and consequently on the product glycoform. Finally, based on this platform, we present the computational design of novel strategies for improving the glycan profile of antibody products and their experimental implementation.


  1. del Val, I. J., Kontoravdi, C. and Nagy, J. M. (2010), Towards the implementation of quality by design to the production of therapeutic monoclonal antibodies with desired glycosylation patterns. Biotechnol. Prog., 26: 1505–1527.

  2. Sou, S. N., Sellick, C., Lee, K., Mason, A., Kyriakopoulos, S., Polizzi, K. M. and Kontoravdi, C. (2015), How does mild hypothermia affect monoclonal antibody glycosylation? Biotechnol. Bioeng., 112: 1165–1176.

  3. Jedrzejewski, P. M., del Val, I. J., Constantinou, A., Dell, A., Haslam, S. M., Polizzi, K. M., and Kontoravdi, C. (2014), Towards Controlling the Glycoform: A Model Framework Linking Extracellular Metabolites to Antibody Glycosylation. Int. J. Mol. Sci., 15: 4492-4522.

  4. del Val, I. J., Nagy, J. M. and Kontoravdi, C. (2011), A dynamic mathematical model for monoclonal antibody N-linked glycosylation and nucleotide sugar donor transport within a maturing Golgi apparatus. Biotechnol. Prog., 27: 1730–1743.

  5. Kyriakopoulos, S. and Kontoravdi, C. (2014), A framework for the systematic design of fed-batch strategies in mammalian cell culture. Biotechnol. Bioeng., 111: 2466–2476.

Extended Abstract: File Not Uploaded