Options
Automated kinetic model identification of biocatalysts under process conditions
Citation Link: https://doi.org/10.15480/882.13090
Publikationstyp
Doctoral Thesis
Date Issued
2024
Sprache
English
Author(s)
Advisor
Referee
Title Granting Institution
Technische Universität Hamburg
Place of Title Granting Institution
Hamburg
Examination Date
2024-05-24
Institute
TORE-DOI
Citation
Technische Universität Hamburg (2024)
Kinetic model identification is important for the optimization and scale-up of enzymatic processes. The modelling process is strongly influenced by human error, experimental design and model selection, which can be overcome by automation. Therefore, in this work, an automated reactor platform has been developed to perform kinetic model identification for a model enzymatic reaction. This reaction is catalyzed by a formate dehydrogenase from the yeast Candida boidinii. The reduction of NAD+ to NADH with the hydride donor sodium formate is monitored by inline UV/Vis spectroscopy at 340 nm in the design space of 0.35 to 2 mM NAD+ and 50 to 300 mM formate. The enzyme is utilized in a packed bed reactor, where it is immobilized on functionalized porous resin particles. The steady-state concentration of NADH is measured to determine the corresponding reaction rate at various initial substrate concentrations, providing data for identifying a kinetic model from eight candidate models. These kinetic models serve as the basis for a model-based design of experiments using the Akaike weight design criterion. This method uses Akaike weights to measure the relative goodness of fit for candidate models and identify experiments with maximum information content. Once a successful model discrimination is achieved, parameter estimation is improved through additional experiments according to the e-optimal design criterion to reduce the confidence intervals of the model parameters of the discriminated kinetic model. The model identification process is based on six fully factorial designed experiments initially, which is then improved through nine additional experiments. This approach leads to the discovery of a kinetic model within 15 experiments, with standard deviations of less than 10% within the range of measurement error. The double substrate Michaelis-Menten model accurately represented the packed bed reactor data after implementing a multi-start heuristic and extending the design space. The kinetic model identification was optimized and applied to data from an enzyme membrane reactor process with formate dehydrogenase in solution. The resulting model showed non-competitive product inhibition, in contrast to the model for the immobilized enzyme. The reactor configuration and immobilization resulted in a distinct kinetic model and confidence intervals. This emphasizes the impact of experimental design on identifying kinetic models and the advantages of automation in this field.
Subjects
automation
Enzyme kinetics
immobilized biocatalysis
model-based design of experiments
DDC Class
572: Biochemistry
Loading...
Name
Hennecke_Leon_Dissertation_2024.pdf
Size
4.85 MB
Format
Adobe PDF