CHO cell productivity improvement by genome-scale modeling and pathway analysis: Application to feed supplements

• Published in: Biochemical engineering journal 2020-08, Vol.160, p.107638, Article 107638
• Main Authors: Huang, Zhuangrong, Xu, Jianlin, Yongky, Andrew, Morris, Caitlin S., Polanco, Ashli L., Reily, Michael, Borys, Michael C., Li, Zheng Jian, Yoon, Seongkyu
• Format: Article
• Language: English
• Subjects: Cellular metabolism; Feed improvement; Genome-scale model; GS CHO-K1 cells; Model simulation; Pathway analysis
• ISSN: 1369-703X; 1873-295X
• DOI: 10.1016/j.bej.2020.107638

•A genome-scale CHO model was employed to analyze the metabolism of CHO cells.•Flux analysis results from genome-scale model were compared with transcriptomics analysis.•In silico simulations using genome-scale model were performed for feed optimization.•This is the first study to apply genome-scale CHO model to improve the IgG production. Effective bioprocess development using Chinese hamster ovary (CHO) cells as hosts is hampered by the limited understanding of cellular metabolism under process conditions in bioreactors. Systematic tools such as genome-scale models have been developed, but their value has not been satisfactorily demonstrated and exploited for process development. In this study, we proposed a method using a genome-scale model to analyze existing process studies for bioprocess optimization. First, we used existing industrial CHO cell culture experiments to systematically gain metabolic insights for bioprocess development. Two fed-batch cultures, using the same cell line and process, resulted in different titers by supplementing two different types of feed media. A genome-scale model was applied to calculate fluxomics (i.e., intracellular fluxes) from extracellular metabolomics and then the metabolic differences were further analyzed through pathway analysis between these two cell culture conditions. Transcriptomics data from RNA-Seq were employed at this point for comparison and found to be consistent in pathway analysis with the flux analysis results. At the second stage, we developed a modeling-based approach for media optimization to increase antibody production based on the understanding from the first stage. The new design was tested in silico using a genome-scale model and then verified experimentally, confirming the applicability of this modeling-based approach for bioprocess optimization. The framework proposed in this study can maximize the utilization of existing process studies and minimize the time consumed for empirical work in developing new processes.