From Screening to Production: a Holistic Approach of High-throughput Model-based Screening for Recombinant Protein Production

Efficient and robust screening of production strains in early bioprocess development is usually hampered by the limited cultivation resources and identification of dynamical cell parameters for the complete design space. Even though High-Throughput (HT) liquid handling stations enable a large number of strains to be tested, these experiments provide no insight into the dynamical phenotype of the strains. This is especially critical in scale-up, since cultivations in industrial bioreactors expose the microbial cell factories to significant stresses due to substrate, oxygen, and pH gradients among others. In an effort to address this challenge and reduce the risk of failure during scale-up, new HT scale down systems based on model-based operation strategies have been developed and extended to conditional screening experiments. In this work we further extend the existing platform to enable a feedback control of the 24-parallel mini-bioreactor setting, using a recursive moving horizon parameter estimation combined with a model-predictive control approach to calculate an optimal feeding regime, which exposes the cells to stress conditions similar to those present in large-scale bioreactors. We present a case study showing the advantages of the framework by screening a set of E. coli strains for obtaining highest biomass at the end of the process. The results show that the prediction and selection of the most suitable strain for industrial production is significantly improved.