Closed-loop identification of enzyme kinetics applying model-based design of experiments

Abstract

Accurate kinetic models for enzyme catalysed reactions are integral to process development and optimisation. However, the collection of useful kinetic data is heavily dependent on the experimental design and execution. In order to reduce the limitations associated with traditional statistical design and manual experiments, this study introduces an integrated, automated approach to identifying kinetic models based on model-based optimal experimental design. The immobilised formate dehydrogenase of Candida boidinii catalyses the enzymatic reduction of NAD⁺ to NADH and is used as a model system. Continuous collection of UV/Vis data under steady-state conditions is employed to determine the kinetic parameters in a packed bed reactor. Automation of the experimental work was utilised in Python to compensate for the need for more time-consuming data collection. A completely automated closed-loop system was created and appropriate kinetic models for anticipating process dynamics were identified. The automated platform was able to identify the correct kinetic model out of eight candidate models with only 15 experiments. Further extension of the design space improved model discrimination and led to a properly parameterized kinetic model with sufficeintly high parameter precision for the conditions under examination.