Mathematical modelling of metabolism and enzyme-level regulation to understand and infer functional regulators of metabolism

Microbial metabolism is feedback regulated on many layers. Feedback control can be executed by metabolite-protein interactions to allosterically control enzyme activity, and to transcriptionally control enzyme abundance, providing cells with robustness to withstand, and adapt to perturbations. However, these interactions contribute to the complexity of metabolism and prohibit an intuitive understanding. To gain a deeper understanding of metabolism, mechanistic mathematical models are useful tools to simplify complex interrelationships. In this thesis, we developed mathematical models to study allosteric feedback, transcriptional feedback, and the interplay of both mechanisms. Then, we used this knowledge to develop a method to map feedback regulation between metabolism and transcription in E. coli metabolism. Since high-quality metabolite data are crucial to study metabolism, we finish this thesis with two chapters on mass spectrometry-based metabolomics. After providing a general introduction in Chapter 1, we develop a mathematical model of amino acid biosynthesis in Chapter 2 to study the interplay between allosteric feedback and transcriptional feedback. We showed that both feedbacks act in concert to balance robustness and efficiency. In Chapter 3, we developed a mathematical model of glycolysis that is transcriptionally regulated by the transcription factor Cra, to study metabolic burden in glycerol-producing E. coli. A robustness analysis revealed that Cra regulation causes growth defects and low glycerol titers in E. coli, and that this burden can be solved by engineering Cra regulation into the glycerol pathway. In Chapter 4, we developed a mathematical model to understand the implications of an ornithine-based allosteric activation at the branch point between arginine and pyrimidine biosynthesis. We showed that the feedback activation buffers upstream perturbations and thereby stabilizes pathway end products. In Chapter 5 we investigated causes for periodic pyruvate oscillations using a mathematical model of glycolysis. We show that feed forward activation of the pyruvate kinase and high saturation of the pyruvate dehydrogenase contribute to pyruvate oscillations. In Chapter 6 we performed knockdowns of 283 genes of E. coli metabolism and measured proteome and metabolome of the perturbed strains. A pathway-based analysis allowed us to map feedback regulation between metabolism and transcription using proteome and metabolome data. In Chapter 7 we analysed and validated the mass spectrometry-based flow-injection metabolomics approach with 160 spike-in samples. We showed that flow-injection causes complex MS1 spectra that can lead to false positive peak annotations. Finally, we concluded this thesis in Chapter 8 by developing an approach to generate reference fragments for low-abundant, or commercially unavailable metabolites to complement reference databases. We showed a proof of principal for two metabolites.