Insights into the genome-scale metabolic modelling framework based on Pseudomonas taiwanensis VLB120
MetadataVis full innførsel
Any living organism - complex or single-cell - consists of multiple layers of inter connected components. This starts on an atomic scale all the way through enzymes and cells, and in some cases organs, to make up organisms, which are in turn connected in the food-web. The study of any of these individual components provides important biological insight. However, the impact of a change in a specific compound onto other compounds or layers is not necessary clear: the knowledge that certain genes and enzymes are not essential or limit the production of metabolites in vivo will not necessarily tell us the impact of a combination of such factors. Consequently, a high-level view, drawing the connections between various components, is required. In this dissertation I present the work in systems biology I performed during my doctoral studies. The focus is on the field of constraint-based modelling, in particular genome-scale metabolic modelling. The thermoplastic nylon is one of the most important polymers required for a variety of daily-use products. However, the non-biological synthesis of the organic compounds, specifically the monomers is an environmentally harmful process. During the last decades, the biotechnological production of various molecules such as antibiotics, vanillin, and a variety of bio-active compounds has increased due to advances in research. The development and improvement of eco-friendly production of nylon precursors using the host organism Pseudomonas taiwanensis VLB 120 was the starting point of this dissertation: generating and curating a high-quality genome-scale metabolic reconstruction for the organism to predict and optimize nylon precursor production (Paper 3 and Chapter 2.4). During this process, the software tool AutoKEGGRec was educed (Paper 1 and Chapter 2.3.2). AutoKEGGRec generates genome-scale draft reconstructions based on the well-known KEGG database, besides community and consolidated models. AutoKEGGRec was further used as a basis for a metabolic functionality clustering of 975 organisms across the Tree of life (Paper 2 and Chapter 4). This allows for an evaluation of relationships between organisms based on their metabolic capability instead of conserved genes. This methodology was applied to identify organisms metabolically similar to P. taiwanensis. A fundamental assumption in constraint-based modelling is that the in vivo organism is optimally adapted to its environment to grow as fast as possible, outcompeting possible intruders. Typically, this is represented by optimizing the in silico model for growth, so that in vivo and in silico organism have the same objective - replication / growth. A pseudo reaction in the model, the biomass objective ii function (BOF), consumes all resources required to generate more of itself and therefore correlates with growth. This objective function is commonly inferred from similar organisms, or extensive literature research if possible. However, it is a known fact that the BOF changes for various environmental conditions: the biomass differs for various organisms, environments, and even growth rates. Depending on the conditions, the organisms are starving or can store certain molecules. While these are not required to make a functional copy, they are potentially an important part of all the organisms living in this environment. For generating a high-quality model we started the laborious laboratory effort to generate an exactly determined in vivo biomass composition for various environments for any organism (Chapter 3). In Paper 4 we discuss how linear optimization can approach the existence of multiple BOFs in a single model, especially when the choice of a certain BOF is not obvious. The BOFs are and will be determined for multiple complex and single-cell organisms.