Genome-wide gene expression and the epigenetic landscape

Integrating development into evolutionary theory requires understanding how sequence variation is translated into variation in the activities of genes and in turn how this generates variation in other phenotypic characters and in fitness. New technologies have made it possible to study variation in gene expression at the whole genome scale. In this dissertation, I develop a geometric framework in which to analyze the dynamics of genome-wide gene expression during physiological processes and development and investigate variation in gene expression in Saccharomyces cerevisiae and during two timepoints at the start of Drosophila metamorphosis. In S. cerevisiae, I show that significant variation in gene activity is restricted to a low-dimensional subspace of the high-dimensional gene expression space, and I partition these subspaces to isolate distinct processes driving gene expression. Different strains of yeast after different perturbations relax back into the same cell-cycle. In Drosophila, I find extensive intra- and inter-specific variation in gene expression trajectories at the start of metamorphosis that is related to gene function and developmental context. There are stronger constraints on the activation of gene expression than on the down-regulation or degradation of transcripts during development and stronger constraints on the expression of genes that encode regulatory molecules than on the expression of genes that encode structural factors or enzymes. In order to build a foundation for quantitative genetic analyses of gene expression, I measure the neutral rate of gene expression evolution in twelve D. melanogaster mutation accumulation lines and find that mutational heritabilities for gene expression range from 10 -7 to 10-3 (median 10-5). Around 75% of the genes expressed at the two timepoints show significant mutational variance across the lines. Environmental and mutational variances are strongly correlated, hinting at widespread canalization. Developmental timing is important; genes have lower mutational heritabilities at the timepoint in which they are more highly expressed. If gene expression were evolving strictly neutrally, there should be far more variation between species than I observe. Similarities in patterns of variation in mutation accumulation lines and between species suggest that biases in the generation of gene expression variation affect the course of evolution.