Poster M2 - Elucidating intracellular glycosylation fluxes in CHO cell culture production of Immunoglobulin G

Introduction: Glycosylation is the most common post-translational modification of proteins that involves the attachment of sugar molecules called glycans. Glycans on monoclonal antibody (mAb) drugs affect their therapeutic efficacy, safety, immunogenicity, and bioavailability. Because glycosylation is a non-template driven process, there exist a natural variability of glycans on mAbs, comprising isomers that are called glycoforms. Glycoform distribution is a critical quality attribute in mAb drug production.
Glycosylation involves a network of enzymatic reactions, catalyzed by glycosidases and glycosyltransferases. Because of the lack of technology for measuring glycosylation reaction rates (fluxes) in cells, modeling approaches have been applied to estimate intracellular glycosylation fluxes with the goal of monitoring and controlling the glycoform distribution. We previously developed a constraint-based model, called Glycosylation Flux Analysis (GFA), under the assumption that glycosylation reactions occurring in the Golgi apparatus can be approximated by a well-mixed reactor where the fluxes are balanced. GFA was able to predict intracellular glycosylation flux dynamics in both fed-batch and continuous production of immunoglobulin G (IgG) drug in Chinese Hamster Ovary (CHO) cell cultivation [1].

Materials and

Methods: In this work, we developed improved GFA (iGFA) which significantly improves upon the GFA in several ways: (1) mathematical formulation based on enzymatic kinetics, (2) the use of regularization to improve biological interpretability of the unknown parameters, and (3) computationally efficient, open-source Python implementation using Pyomo-DAE package and IPOPT solver [2, 3]. We applied iGFA to predict intracellular glycosylation fluxes in CHO fed-batch cell cultures under different pH conditions [4]. We constructed a N-linked glycosylation reaction network comprising 16 reactions and 13 glycoforms using GlycoenzDB [5]. The inputs to iGFA were time-series measurements of viable cell density, antibody titer, cell specific productivity and glycoform fractions. We compared the predicted glycosylation fluxes across three different pH conditions and qualitatively analyzed the correlations between model predictions and gene expression and nucleotide sugar measurements.

Results, Conclusions, and Discussions: N-linked glycosylation fluxes predicted by iGFA showed that across all three pH conditions, GlcNAc addition has a marked increase during the cell culture—for example, GlcNAcylation of G0F increased by 50% over the course of 14 days. This trend is correlated with the increase in nucleotide sugar UDP-GlcNAc and gene expression of MGAT4A (Fig. 1C, D, E). We observed a nonlinear influence of media pH on intracellular glycosylation fluxes. Specifically, an increase from pH 6.9 (medium) to 7.1 (high) has a much smaller effect on upstream glycosylation fluxes than an increase from pH 6.7 (low) to 6.9 (medium) (Fig. 1B). This is consistent with the experimental observations where upstream glycans (G0 and high-Mannose structures) were consistently elevated at low pH. Meanwhile, intermediate glycans (G0F and G0F+GlcNAc) tended to increase with time, with greater effect at medium and high pH than low pH. Lastly, iGFA gave an order of magnitude improvement in computational time over GFA.

Conclusions:
In this work, we developed an improved version of our GFA for estimating intracellular glycosylation fluxes. The application of iGFA to examine the impact of media pH on glycosylation revealed that pH has a non-linear effect on glycoform fluxes where upstream N-linked glycoforms are affected more strongly by pH than downstream glycoforms. Open-source implementation of iGFA is available from http://www.github.com/cabsel/.

Acknowledgements (Optional): 1. Hutter, S., et al., Glycosylation flux analysis reveals dynamic changes of intracellular glycosylation flux distribution in Chinese hamster ovary fed-batch cultures. Metab Eng, 2017. 43(Pt A): p. 9-20.
2. Nicholson, B., et al., pyomo.dae: a modeling and automatic discretization framework for optimization with differential and algebraic equations. Mathematical Programming Computation, 2018. 10(2): p. 187-223.
3. Wachter, A. and L.T. Biegler, On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Mathematical Programming, 2006. 106(1): p. 25-57.
4. Lee, A.P., et al., Multi-omics profiling of a CHO cell culture system unravels the effect of culture pH on cell growth, antibody titer, and product quality. Biotechnol Bioeng, 2021. 118(11): p. 4305-4316.
5. Ted Groth, S.N., Yusen Zhou. GlycoEnzDB: Comprehensive glycoEnzyme database and dynamic pathway generation. Available from: http://www.virtualglycome.org/GlycoEnzDB.