Biotechnology, the exploitation of biological processes for the benefit of humanity, has become one of the main industries in our current economy. A big part of biotechnology relies on understanding biological microorganisms and how to tune them to best suit the corresponding applications. Metabolism, the interplay of biochemical reactions inside a cell to generate energy and components needed for growth, is an important layer of information for understanding microorganisms. In this thesis, I study metabolism of Saccharomyces cerevisiae (budding yeast), one of the most popular microorganisms in biotechnology, used for the production of bread, beer, wine, biofuels and many other biotechnology products.
As metabolism is very complex, the best way to study it is with the aid of computers. A common approach for doing this is by creating what is known as a genome-scale metabolic model (GEM), which is a mathematical representation of all reactions in metabolism. A problem that this method has is that it lacks information to be able to computationally simulate biological responses under certain experimental conditions. Important information that could be added are the abundances inside the cell of different biomolecules, such as lipids and proteins. In this thesis, I present methods for improving computational simulations of GEMs with so-called “abundance constraints”, together with novel observations about yeast biology.
Among the computational methods, I present i) a way of storing and updating GEMs so we keep track of every change we do; ii) a method for combining GEMs with intracellular levels of lipids; iii) a method for combining GEMs with enzymes, which are proteins that can catalyze metabolic reactions; and iv) an improved method for estimating protein levels based on experimental data.
Among other results, I show that i) lipid metabolism is very flexible, meaning that yeast is able to re-arrange it in different ways without spending too much energy; ii) the total enzyme content in yeast imposes an important constraint on metabolism, as the cell will be limited to certain options based on the amount of available enzyme; iii) specifically, when yeast has to grow fast, it will have to become less efficient in processing substrate; and iv) specific enzyme levels can create limitations in metabolism when the cell is grown under an environmental stress.
The results presented in this thesis show that by accounting for extra levels of information, computational approaches can be improved to better understand biology. Furthermore, the methods developed are expected to be useful for industrial applications of S. cerevisiae, as tools for guiding researchers to develop new strains with improved performance.