Tempo and mode of regulatory evolution in Drosophila

Genetic changes affecting gene expression contribute to phenotypic divergence; thus, understanding how regulatory networks controlling gene expression change over time is critical for understanding evolution. Prior studies of expression differences within and between species have identified properties of regulatory divergence, but technical and biological differences among these studies make it difficult to assess the generality of these properties or to understand how regulatory changes accumulate with divergence time. Here, we address these issues by comparing gene expression among strains and species of Drosophila with a range of divergence times and use F₁ hybrids to examine inheritance patterns and disentangle cis- and trans-regulatory changes. We find that the fixation of compensatory changes has caused the regulation of gene expression to diverge more rapidly than gene expression itself. Specifically, we observed that the proportion of genes with evidence of cis-regulatory divergence has increased more rapidly with divergence time than the proportion of genes with evidence of expression differences. Surprisingly, the amount of expression divergence explained by cis-regulatory changes did not increase steadily with divergence time, as was previously proposed. Rather, one species (Drosophila sechellia) showed an excess of cis-regulatory divergence that we argue most likely resulted from positive selection in this lineage. Taken together, this work reveals not only the rate at which gene expression evolves, but also the molecular and evolutionary mechanisms responsible for this evolution.Understanding the relationship between tempo (the rate at which a trait evolves) and mode (the manner in which a trait evolves) is essential for understanding the evolutionary process (Simpson 1944). This is true not only for organismal phenotypes, but also for the molecular phenotypes that produce organismal traits. Gene expression is one such molecular phenotype (Gordon and Ruvinsky 2012); it is essential for organismal form, fitness, and function, and frequently varies within and between species. Comparative studies using genomic surveys of gene expression in yeast (Busby et al. 2011), Drosophila (Rifkin et al. 2003), and mammalian species (Brawand et al. 2011) with a range of divergence times have provided insight into the tempo of gene expression evolution, but the mode and its relationship to tempo remain less well understood.Elucidating the mode of gene expression evolution includes identifying the types of regulatory changes that have evolved as well as how interactions among divergent regulatory alleles affect gene expression. F₁ hybrids, in which divergent regulatory alleles interact in the same cellular environment, can be used to investigate these issues. Allele-specific expression in F₁ hybrids separates the effects of cis- and trans-regulatory changes affecting a gene’s expression by providing a readout of relative cis-regulatory activity in a common trans-regulatory environment (Cowles et al. 2002). Expression differences between genotypes not attributed to cis-regulatory changes are inferred to be caused by trans-regulatory divergence (Wittkopp et al. 2004). In addition, the net effects of interactions among divergent regulatory alleles are revealed by comparing levels of total expression in F₁ hybrids to parental genotypes.This approach was initially used to separate cis- and trans-regulatory effects of divergence affecting expression of dozens of genes. These studies suggested that (1) cis-regulatory changes are more common than trans-regulatory changes between species (Wittkopp et al. 2004); (2) genes with cis- and trans-acting changes favoring expression of opposite alleles are more likely than other types of changes to cause misexpression in F₁ hybrids (Landry et al. 2005); (3) environmental factors modulate relative cis-regulatory activity (de Meaux et al. 2006); (4) cis-regulatory variation is abundant in natural populations (Osada et al. 2006; Genissel et al. 2007; Campbell et al. 2008; Gruber and Long 2009); and (5) the amount of expression divergence attributable to cis-acting changes is greater between than within species (Wittkopp et al. 2008).More recently, microarrays and RNA-seq have been used to extend these analyses to the genomic scale (Wang et al. 2008; Graze et al. 2009; Tirosh et al. 2009; Zhang and Borevitz 2009; Fontanillas et al. 2010; McManus et al. 2010; He et al. 2012; Shi et al. 2012; Coolon and Wittkopp 2013; Levy et al. 2013; Schaefke et al. 2013). In some cases, relationships seen in the smaller scale studies were replicated. For example, cis- and trans-regulatory changes with effects in opposite directions were overrepresented among misexpressed genes (Tirosh et al. 2009; McManus et al. 2010; Schaefke et al. 2013) and cis-regulatory changes explained more of the expression differences between than within species (Tirosh et al. 2009; Emerson et al. 2010). Other observations, such as the relative proportion of genes with evidence of cis- and/or trans-regulatory changes, were much more variable among studies. Finally, novel patterns, such as the relationship between dominance and cis/trans-regulatory changes (Lemos et al. 2008; McManus et al. 2010) and the frequency of compensatory cis- and trans-regulatory variants (Tirosh et al. 2009; Goncalves et al. 2012; Shi et al. 2012), were identified.Despite this growing collection of case studies examining the types of changes responsible for expression differences within and/or between particular pairs of species, the use of different organisms (flies, yeast, plants, and mice), techniques (pyrosequencing, microarrays, RNA-seq), and analysis methods (linear models, exact tests, and Bayesian approaches) among these studies precludes the type of meta-analysis needed to determine how the mode of regulatory evolution changes with divergence time and to robustly assess the generality of relationships reported in previous studies. To address these issues, we examined the tempo and mode of regulatory evolution in concert using strains and species of Drosophila with a range of divergence times.