Despite its centrality to Darwin’s theory of evolution by natural selection, the process of adaptation is still not fully understood. In particular, the dynamics of the genotypes and phenotypes associated with an adaptive response remain to be fully elucidated. In my dissertation, I utilized laboratory evolution experiments to study how the genotypes and phenotypes of Escherichia coli change over time as they adapted to high temperature.
Chapter 1 explored how metabolic phenotypes of 115 evolved E. coli clones changed as a result of 2,000 generations of adaptation to 42.2°C. Using phenotypic microarrays (Biolog plates), I quantified 94 phenotypes of these evolved clones, as well as their ancestor under stressed (42.2°C) and unstressed (37.0°C) conditions. Comparing the evolved phenotypes to the ancestral phenotypes revealed that adaptation was predominantly restorative, shifting evolved phenotypes from the stress state toward the unstressed state. I also uncovered associations among common genotypic changes found in the evolved clones and their phenotypes.
Chapter 2 investigated the different mutational dynamics in populations traversing two different adaptive pathways typified by mutations in the rpoB and rho genes, respectively. These genes were predicted to be differentially pleiotropic, and were therefore expected to create differences in compensatory evolution when mutated. I used temporal sequencing data of four rpoB and four rho populations to reconstruct their mutational trajectories over the course of adaptation to 42.2°C. These trajectories revealed that rpoB and rho mutations occurred early on during adaptation, canalizing the adaptive process. Furthermore, rpoB populations accumulated more mutations and experienced more clonal interference over the course of adaptation than rho populations.
Chapter 3 was a study of the Lazarus effect, a phenomenon of population recovery under lethal selection conditions. I evolved ~300 E. coli populations to the lethal temperature of 43.0°C and measured their cell density over five days. I sequenced those populations that recovered and found mutations in two operons—hslUV and rpoBC—to be the major drivers of Lazarus events. These mutations differed in their frequency in the experiment, degree of parallelism within and between weeks, and fitness tradeoffs at 37.0°C, suggesting different origins and adaptive dynamics between them.