High cell density induces spontaneous bifurcations of dissolved oxygen controllers during CHO cell fermentations

• Published in: Biotechnology and bioengineering 2003-10, Vol.84 (2), p.224-232
• Main Authors: Chung, John D., Chang, Conway C., Groves, James Ashley
• Format: Article
• Language: English
• Subjects: Algorithms; Animals; bifurcation; Bioreactors; Cell Count; Cell Division - drug effects; Cell Division - physiology; CHO; CHO Cells; Computer Simulation; Cricetinae; Cricetulus; dissolved oxygen control; Fermentation; high cell density; Kinetics; Models, Theoretical; Oxygen - metabolism; Oxygen - pharmacology; stability
• ISSN: 0006-3592; 1097-0290
• DOI: 10.1002/bit.10763

High cell density cultures of CHO cells growing in a bioreactor under dissolved oxygen control were found to undergo spontaneous bifurcations and a subsequent loss of stability some time into the fermentation. This loss of stability was manifested by sustained and amplified oscillations in the bioreactor dissolved oxygen concentration and in the oxygen gas flow rate to the reactor. To identify potential biological and operational causes for the phenomenon, linear stability analysis was applied in a neighborhood of the experimentally observed bifurcation point. The analysis revealed that two steady state process gains, KP1 and KP2, regulated kla and gas phase oxygen concentration inputs, respectively, and the magnitude of KP1 was found to determine system stability about the bifurcation point. The magnitude of KP1, and hence the corresponding open‐loop steady state gain KOL1, scaled linearly with the bioreactor cell density, increasing with increasing cell density. These results allowed the generation of a fermentation stability diagram, which partitioned KC‐N operating space into stable and unstable regions separated by the loci of predicted critically stable controller constants, KC,critical, as a function of bioreactor cell density. This consistency of this operating diagram with experimentally observed changes in system stability was demonstrated. We conclude that time‐dependent increases in cell density are the cause of the observed instabilities and that cell density is the critical bifurcation parameter. The results of this study should be readily applicable to the design of a more robust controller. © 2003 Wiley Periodicals, Inc. Biotechnol Bioeng 84: 224–232, 2003.