(69g) Bioprocess Control Using Stoichiometric Models of Metabolism

Mammalian cells produce up to 80% of the commercially available therapeutic proteins, including monoclonal antibodies, with Chinese Hamster Ovary (CHO) cells being the primary production host. Manufacturing involves a train of reactors, the last of which is typically operated in fed-batch mode, where cells grow and produce the required protein. The reactor's product yield is typically low (up to 200 mL L^-1 day^-1), which complicates downstream purification, accounting for up to 80% of the total production cost. Commonly employed optimization strategies include optimizing specific cell culture conditions (temperature and pH) and the feeding strategy. The feeding strategy is often decided a priori, from either prior knowledge or through design of experiments, and rarely considers the current state of the process. Advanced process control strategies like model predictive control are not typically used as online monitoring of samples is limited. In this work, we propose an advanced control strategy that leverages a metabolic model to optimize process yield and that requires only offline measurements commonly available for most fed-batch processes.

The model predictive control formulation proposed herein is built on a generic hybrid kinetic-stoichiometric reactor model and is therefore agnostic to particular cell lines or products. In this sense, the model is readily transferable across CHO cell culture systems. The control strategy (graphically represented in the image below) leverages a reduced metabolic network model of a CHO cell to calculate glucose and amino acid uptake rates that will lead to optimal cell growth for each culture phase. These fluxes are then used as control targets to compute the optimal feeding strategy. The optimal feeding strategy is then simulated in a high-fidelity model of the production reactor to predict the outcome in silico. We compare our closed-loop results against experimental data of a feeding strategy designed for antibody maximization [1]. Our results suggest that antibody production can be increased even further.

The proposed strategy is a first step towards integrating Industry 4.0 tools in bioprocessing. It does not require extensive computational power and can be implemented without extra infrastructure costs. It also bypasses the need to develop a complete dynamic metabolic model for CHO cells, which would be computationally challenging.

References:

[1] Kyriakopoulos S, Kontoravdi C. A framework for the systematic design of fed-batch strategies in mammalian cell culture. Biotechnol Bioeng. 2014;111(12):2466-2476. doi:10.1002/bit.25319