(133f) Dynamic Model for CHO Cell and Process Engineering

Industrial CHO cell fed-batch processes have progressed significantly over the past decade, with recombinant protein production titers consistently reaching the gram per liter level.  Such improvements have largely resulted from separate advances in process and cell line development. In contrast, metabolite overproduction using microbial cells has spectacularly benefited from metabolic engineering efforts combining model-based design and genetic manipulations. Model-based selection of targets for genetic engineering in CHO cells is confounded by the dynamic nature of the fed-batch process. Due to transitions such as the lactate shift, inferior targets may be selected by relying on steady-state models used to design and optimize microorganisms during log-phase growth.

In this work, we use a dynamic model of CHO cell metabolism to simultaneously identify both process and cell modifications that improve antibody production. Model simulations were used to characterize the dynamic responses of metabolic gene knockdowns in the context of variations in process variables over the full course of a multi-day fed-batch. The simulations explored ca. 9,200 combinations of process variables (shift temperature, shift day, seed density, and harvest day) and knockdowns (8 metabolic enzymes). The simulation results were screened for improvements in product titer as well as other quality control criteria, including peak cell density and final lactate and ammonia concentrations in the medium. Approximately 0.5% of the tested combinations met these criteria and produced an antibody titer that is an increase greater than 1.5-fold compared to the control condition (unmodified cell and base process). Among these combinations, 88% involved knockdown of enzymes involved in lactate metabolism. The improvement in antibody titer with reduced lactate production was corroborated by experimental observations. Interestingly, depending on the process conditions, modulating the lactate enzymes yielded varying productivities, ranging from a reduction in final titer to greater than 2-fold improvement.

Our results underscore the benefit of combining process and cell optimization, and highlight the need to develop dynamic models capable of accurately simulating experimentally documented metabolic shifts in fed-batch cultures of mammalian cells.