Evolutionary dynamics of metabolic adaptation

Abstract

In this thesis we study how organisms adapt their metabolism to a changing environment. Metabolic adaptation occurs at different timescales. Organisms adapt their metabolism via metabolic regulation, which happens in the order of minutes to hours and via evolution, which takes many generations. Here we study the interplay between these timescales by studying the evolution of metabolic adaptation. In chapter 2, we develop an individual-oriented, spatial model of cells that evolve their lac operon promoter function. In this way we study what kind of promoter function evolves. The promoter function that evolves in the simulations is surprisingly similar to the experimentally observed promoter function. Furthermore we show that, in contrast to the current belief, the lac operon is not a bistable switch when induced by the natural inducer lactose. Only when induced by artificial inducers, the lac operon behaves bistably. We modified this model to incorporate the effect of stochasticity in gene expression. In this way we study the effect of stochasticity by comparing the results between the deterministic and stochastic model. We find that in our model cells avoid stochastic gene expression by evolving to larger repressed transcription rates and therefore we conclude that stochasticity in the lac operon is detrimental for the cells. In chapter 4 we study the mutational dynamics of massive gene loss. We find that the pattern of gene loss in S. cerevisiae cannot be explained by random single gene loss. If we however assume that large base pair deletions cause simultaneous deletion of neighboring genes, we can explain the pattern of gene loss in S. cerevisiae satisfactorily. This model predicts that genes that have smaller intergenic regions in a yeast species that did not underwent WGD are more likely to be both deleted in S. cerevisiae and this is exactly what we find in the data. In chapter 5 we study the effect of a WGD on the metabolic network of S. cerevisiae using metabolic modeling. We find that we can satisfactorily predict which genes are retained in duplicate after WGD. Furthermore, we predict that both transporter genes and genes functioning in glycolysis are more likely to be retained in duplicate after WGD, which we also found to be true in S. cerevisiae. Finally we show that a WGD can lead to a fitness increase in environments to which the cells are not yet perfectly adapted. Concluding, we have shown that combining detailed, quantitative modeling with an evolutionary, individual based approach is a promising way to study the evolution of biological systems. This approach has not only given us insight how, from an evolutionary point of view, such systems came about, but also how such systems work. Ignoring the evolutionary aspects of biological systems has, in the case of the lac operon, led to a wrong understanding of the dynamics of the system. Furthermore we have shown that even for such a large scale system as the metabolic network of S. cerevisiae, this approach leads to accurate predictions of the evolutionary outcome.