(632h) Utilising a Genome Scale Metabolic Model to Design High-Producing CHO Cell Lines through Bi-Level Optimisation

Chinese hamster ovary (CHO) cells are the dominant expression platform to produce therapeutic monoclonal antibodies (mAbs). However, despite increasing global demand and persistent societal needs, mammalian cell-based production systems suffer from low product yields as specific productivity is inversely proportional to cellular growth rates. In this work, bilevel optimization techniques are applied to the CHO-K1 cell genome-scale metabolic model (GEM) [1], to detect genetic regulations that will increase specific mAb productivity during the growth phase of the cell culture. Initially, the GEM was manually curated by removing the blocked reactions, limiting the cellular uptake and secretion rates based on four different experimental exometabolomic datasets and applying elemental balance algorithms to obtain realistic bounds for the intracellular reactions. Then, based on the OptReg algorithm [2], a bi-level optimization algorithm was formulated to maximise both the antibody production and cellular growth of the CHO-K1 GEM for the four experimental datasets. The optimisation resulted in a plethora of scenarios, each containing a different combination of regulations on the reaction bounds corresponding to enzyme knock-out and/or overexpression strategies that would theoritacally increase the antibody specific productivity during the exponential growth phase of the cell culture. The resulting scenarios of genetic regulations were then analysed using sensitivity analysis to identify a minimum set of genetic interventions that could achieve our goal and a select few were screened experimentally.

[1] Hefzi H et al. A Consensus Genome-scale Reconstruction of Chinese Hamster Ovary Cell Metabolism. Cell Systems. 2016; 3:434-43. e8.

[2] Pharkya P and Maranas C. An optimization framework for identifying reaction activation/inhibition or elimination candidates for overproduction in microbial systems. Metabolic Engineering. 2006; 8:1-13