Evolution in changing environments : heavy metal adaptation in experimental yeast populations as a case study

Abstract

Directional environmental change in the form of global climate change and human-induced pollution is one of the most pressing problems facing society today. While species can sometimes adapt to such change by means of phenotypic plasticity and range shifts, there is considerable concern that these mechanisms are insufficient for long-term population persistence for at least some species. Evolutionary adaptation could potentially offer a solution, but it is unclear what the potential and limitations of this process are for populations exposed to deteriorating conditions. Specifically, we have limited knowledge about the factors promoting evolutionary population rescue, and we know even less about how directional change affects evolutionary dynamics and outcomes. In this thesis, I investigated how the rate of directional environmental change affects evolutionary dynamics and outcomes using experimental evolution of heavy metal tolerance in the baker’s yeast Saccharomyces cerevisiae as a model system. Heavy metals are important pollutants that primarily originate from mining and industry and play a key role in both essential biological processes and toxicity due to their highly reactive chemical properties. While some of these metals, such as cadmium (Cd), lead (Pb), and mercury (Hg), are non-essential and have harmful effects even at low concentrations, others, such as manganese (Mn), iron (Fe), nickel (Ni), and zinc (Zn), are essential micronutrients for a wide range of organisms and are harmful only at high concentrations. Baker’s yeast is the most widely studied eukaryote model organism and shares numerous features with both plants and animals. Due to its short generation time, large population sizes, and readily manipulated genetics it is also used extensively in experimental evolution studies. In such studies, organisms are cultured under controlled conditions for many generations, mostly to study fundamental evolutionary questions, but increasingly also to characterize the evolutionary response of an organism to a particular selection pressure and thus increase insight into the genetic architecture of the trait under selection. I cultured replicate populations of yeast for 500 generations in the presence of either gradually increasing or constant high concentrations of Cd, Ni, and Zn, and analysed the evolutionary response at the fitness (Chapter 2) and genomic (Chapter 3) level. Additionally, I generated several mutants carrying various combinations of mutations based on the genomic information from Chapter 3, and determined their fitness at different metal concentrations (Chapter 4). I used these data to address a number of hypotheses, which were based on a genotype-by-environment (GxE) framework that I developed to aid my thinking about evolution in changing environments. This framework is based on the concept of the fitness landscape, and distinguishes between two general patterns of change in landscape topography across conditions. Magnitude GxE refers to the case where the height of the fitness landscape increases along an environmental gradient, while its shape remains roughly constant. This implies that the fitness ranking of genotypes is the same at all concentrations, but that fitness differences between genotypes are larger at high metal concentrations. This scenario predicts that the same mutations will be selected under gradual and abrupt change, but that adaptation will be delayed under gradual change. By contrast, reranking GxE refers to the case where the shape of the fitness landscape changes along an environmental gradient. This implies that the fitness ranking of genotypes changes across metal concentrations. This scenario predicts that different mutations will be selected under gradual and abrupt change, and that adaptation will be delayed under gradual change. Moreover, it predicts different evolutionary outcomes following gradual and abrupt change, and lower repeatability of evolution under gradual change. I anticipated that the relative importance of magnitude and reranking GxE would depend on the nature of the stressor. That is, I expected that magnitude GxE would be more important for the non-essential metal Cd, because in that case the same evolutionary solution — minimizing internal Cd concentrations — should be favoured at all concentrations, but more strongly so at high concentrations. This is consistent with directional selection with increasing intensity. On the other hand, for the presumably essential metals Ni and Zn, I expected the reranking GxE scenario to be more important, because in these cases different evolutionary solutions should be favoured at different external metal concentrations to maintain a constant internal metal concentration. This is consistent with stabilizing selection to maintain low, but non-zero intracellular metal concentrations. My results provide support for these hypotheses at several levels. In Chapter 2, I determined relative fitness of populations isolated at different time points from the evolution experiment in the presence of different concentrations of the selective metal. These phenotypic assays showed that for Cd, the fitness ranking of isolates was the same at all metal concentrations, but that fitness differences were larger at high concentrations. Conversely, for Ni and Zn, the fitness ranking of isolates changed across metal concentrations. For all metals, this resulted in a delay in fitness increase under gradual relative to abrupt change. However, fitness of evolved populations from the final time point of the experiment was the same following gradual and abrupt change. In Chapter 3, I performed whole-genome sequence analysis on a single clone isolated from each replicate population at the final time point of the evolution experiment. This revealed that adaptation to the selective environments occurred via a complex combination of SNPs, small indels, whole-genome duplications, and other large-scale structural changes. Furthermore, these analyses confirmed the phenotypic results of Chapter 2 and showed that for Cd, mutations in the same genes were selected under gradual and abrupt change, whereas for Ni and Zn, mutations in different genes were selected in response to different rates of environmental change. Additionally, I found that evolution was less repeatable at the genomic level following gradual change in the case of Ni and Zn, as predicted by my GxE framework. Finally, in Chapter 4, I reconstructed local fitness landscapes for each metal by deleting all repeatedly mutated genes — as identified by whole genome sequencing — both by themselves and in combination. I used deletions because most mutations that I found in the evolved lineages were predicted to result in loss of function. Fitness assays on these landscapes at different metal concentrations were then used to evaluate how the height and shape of each landscape changed as a function of concentration. For Cd, I found that the height, but not the shape, of the landscape changed across concentrations. Conversely, for Ni, the shape of the landscape changed considerably across concentrations, and I made predictions about the consequences that this likely had for the selective dynamics of mutations in my evolution experiment. Deep sequencing of evolved population samples from different time points supported these predictions, demonstrating the power of landscape reconstruction approaches for understanding evolutionary dynamics. Taken together, this multi-level evidence makes a strong case for the usefulness of my GxE framework for understanding evolution in changing environments. Moreover, my results confirm that the rate of environmental change and the nature of the selection pressure can have crucial consequences for the types and selective dynamics of the mutations that are used for adaptation. These findings imply that if we are to fully understand and anticipate the consequences of important societal problems such as climate change and pollution, we need to take into account not only the total magnitude of the projected change, but also its rate. My results are explained remarkably well by the GxE framework that I developed, which captures some of the generic principles that determine how local fitness landscapes may change across environments. This framework can be readily extended to other selection pressures and scenarios of change. Therefore, my work provides a potentially useful basis for future studies aimed at understanding evolution in changing environments. Although some of the specifics of my model system and experimental approach preclude a straightforward extrapolation of my results to smaller populations of larger organisms, there are in fact more parallels between evolution under both these settings than may seem obvious. While microbial laboratory evolution experiments make simplifying assumptions about reality, it is exactly for this reason that they capture the essence of adaptation to directionally changing environments in a thus far unmatched manner.