McCann, Meghan2021-10-252021-10-252021-09https://hdl.handle.net/11299/225106University of Minnesota Ph.D. dissertation. September 2021. Major: Chemical Engineering. Advisor: Wei-Shou Hu. 1 computer file (PDF); xviii, 204 pages.Mammalian cells are inherently complex, possessing unknown levels and details of regulation that dictate their behaviors in the context of their ever-changing environments. Despite that, Chinese Hamster Ovary (CHO) cells are a workhorse of the pharmaceutical industry; their products – therapeutic proteins –generate > $40 billion per annum. Cell metabolism drives cell growth and other behaviors. In establishing an industrial process for biologic generation with established cell lines, culture conditions are used to control cell growth, productivity, and product quality, including protein glycosylation. I have taken systems engineering approaches to increase our understanding and control of complex, but important, cellular metabolic behavior. To create new cell lines that produce desired glycoforms for protein therapeutics, the metabolic reactions supporting this critical quality attribute were systematically perturbed. An antibody producing cell line was characterized and found to produce IgG with low galactosylation. A library of genes intended to adjust enzyme kinetics and substrate availability within glycosylation was synthesized and the genes stably introduced to the cells. At the culmination of an iterative design strategy, three-gene cassettes containing galactosyltransferase, UDP-sugar transporters, and UDP-Galactose synthesis genes were found to substantially increase galactosylation on the final product. Dihydrofolate reductase (DHFR) and glutamine synthesis (GS) systems are the two major methods used to generate recombinant protein producing CHO cell lines. The DHFR selection system relies on the intervention of the folate synthesis pathway, while the GS system perturbs the supply of an essential amino acid. Cell lines created with the two methods differ metabolically while exhibiting general similarities. I have taken a system engineering approach, building data processing, aggregation, and analysis pipelines to facilitate big data compilation and process characterization. Comprehensive measurements using metabolomic and transcriptomic tools were employed, expanding beyond traditional analytes like glucose, lactate, and ammonia, and into extracellular and intracellular amino acids, metabolic by-products, trace metals, vitamins, and N-glycan composition of the product protein. Additionally, mRNA transcript levels at different metabolic states were obtained by RNAseq, while flux distribution in key metabolic nodes during key metabolic cell-states were evaluated by stoichiometric MFA. The comprehensive data set enabled the computation of intracellular metabolic flux distribution and further facilitated the adoption of a generic metabolic model for the cells. The metabolic model obtained can be extended to a process model for in silico process optimization and process performance prediction in scaling up and scaling down, presenting a systems approach demonstrated in this study that should be generally applicable to cell lines used in the production of biologics. By increasing our understanding of complex biological systems, we get closer to the goal of controlling desired metabolic behaviors in cells used for therapeutic protein production. A challenge towards this goal can be gathering and analyzing process data in a meaningful way. This work presents frameworks for methodically increasing our ability to obtain desired glucose and glycan metabolic phenotypes in industrially relevant CHO cell culture processes through integrated experimental and modeling frameworks.enThesis or Dissertation