Many biotechnological efforts in plants have been proposed to address issues around climate change, sustainability, and food security. These include the modification of the oilseed crop Camelina sativa to improve the efficiency with which it produces oil from captured carbon, to the suppression of photorespiration and enhancement of yield in crops by engineering a Carbon Concentrating Mechanism into them. However, research and development efforts to use biotechnological interventions to improve plants and microbes have been hampered by their extreme complexity. Many of these bioengineering efforts seek to modify the rates of in vivo biochemical reactions – referred to hereafter as fluxes – in order to improve the efficiency or yield with which a desired product(s) is produced. Therefore, these efforts require the characterization and modification of the organism’s metabolic activity. As in other areas of engineering, these processes can be aided by the use of quantitative modeling. In the case of metabolic modeling, multiple approaches already exist. These include the use of simplified compartmental models (Chapter 2) and enzyme-based modeling (Chapter 4), as well as constraint-based approaches such as the linear-optimization-based Flux Balance Analysis (Chapter 3) and the nonlinear regression based Metabolic Flux Analysis (Chapter 2). Although these methods are rarely used in tandem, they are mathematically interrelated and can be used to validate and/or corroborate one another’s findings. Indeed, previous literature in the area of metabolic modeling has frequently paid short shrift to the importance of validation and model selection in this area of study, calling into question the biological relevance and accuracy of many modeling studies. I begin with a discussion of challenges and prospects for future development in the statistical evaluation of metabolic models (Chapter 1), emphasizing the need for cross-comparison of multiple techniques, which I put into practice in later chapters. Emphasis is put on the need for careful validation of 13C-Metabolic Flux Analysis findings using multiple lines of evidence and on the usefulness of validating Flux Balance Analysis flux predictions using estimates from 13C-Metabolic Flux Analysis. I apply these methodological insights to studies of Camelina sativa and its relative Arabidopsis thaliana. I start with a 13C-Metabolic Flux Analysis study of the metabolism of photosynthesizing leaves of Camelina sativa (Chapter 2). By modeling the stable isotopic labeling levels of Calvin-Benson intermediates with a series of polyexponential models, I corroborate the study’s 13C-Metabolic Flux Analysis findings, resulting in a more detailed model of C. sativa’s leaf metabolism and resolving a decades-old mystery in the labeling of these metabolites. Following this, I present a novel method of incorporating multi-omic datasets into FBA predictions of metabolic fluxes in the closely related organism A. thaliana (Chapter 3). I demonstrate that this new method successfully improves agreement between FBA and 13C-MFA flux maps of A. thaliana, setting the stage for improved FBA and metabolic engineering insight into the related C. sativa.Finally, I turn my attention to reaction-diffusion modeling. In Chapter 4, I apply enzyme-based and spatial modeling techniques to understand the net CO2 fixation and light-use efficiency implications of incorporating a biophysical Carbon Concentrating Mechanism into a C3 plant. After some concluding remarks in Chapter 5, I present two additional studies that are related either to the quantitative modeling of plants or metabolic networks, but which are not directly related to the rest of the investigations in this thesis. First, I investigate the extent to which the kinds of tissue-specific gene expression patterns utilized in Chapter 3 are conserved across all flowering plants, providing evidence that such an approach may be broadly usable in plant metabolic modeling. Finally, I present interactive educational materials that teach the underlying theory for all of the metabolic modeling approaches used in the above studies. I implemented these materials into an intensive workshop series put on at Michigan State University and demonstrate that participants’ self-assessed confidence in the techniques taught increased significantly.
Read
