ESSENTIAL AMINO ACID FORECASTING AND FEEDING METHOD AND NUTRIENT EXCHANGE RATE PREDICTION THROUGH GENOME-SCALE MODEL FOR CHINESE HAMSTER OVARY CELLS
View/ Open
CHEN-THESIS-2020.pdf (1.495Mb) (embargoed until: 2024-05-01)
Date
2020-08-12Author
Chen, Xiaolu
0000-0002-7877-5098
Metadata
Show full item recordAbstract
Constraint-based modeling is an important tool for analyzing metabolic systems in organisms like Chinese hamster ovary (CHO) cells, together with flux balance analysis (FBA) and genome-scale metabolic models (GEMs). Two unconventional FBA approaches were presented in this study and proved their validity: essential nutrient minimization (ENM) approach and uptake-rate objective function (UOF) approach.
An ENM approach based essential amino acid forecast method was presented and three essential amino acids (lysine, leucine, valine) were tested in fed-batch with customized lysine or lysine-leucine-valine knockout media. The ENM-solved predictions presented an overall 16.63% MAPE for total consumption in comparison with experimental measurements through the culture course. However, the accuracy of prediction can be influenced by the estimation of cell dry weight. Furthermore, a four-parameter logistic growth model was presented to fit growth curve in death phase and the comparison with a three-parameter logistic model shows a better performance of four-parameter in both fed-batch and perfusion culture. A death phase essential amino acid calculation formula was also generated for four-parameter logistic model study, which gives a decreasing amino acid consumption simulation. The formula needs more death phase biomass data for evaluation.
Based on ENM approach in predicting essential amino acid concentration, an unconventional FBA approach - UOF approach was applied to predict non-essential nutrient exchange rate and a GUI interface – Model Validation Tool was generated to serve this purpose. An IgG producing CHO-DG44 cell line was tested and a comparison of model predictions and experimental data was provided. As a result, the UOF model showed its ability to give satisfying predictions and correlate a CHO-DG44 model with the experimental cell line. The corresponding shadow prices provided metabolic information about nutrients interactions and different interrelations through different cell lines. The ENM and UOF approach employed in this study have the potential to be expanded and further used in other mammalian cells and different cell cultures.