Recent studies revealing how the gain, loss and repositioning of transcription factor binding sites within regulatory sequences can alter gene expression with observable phenotypic consequences 
have focused efforts to understand the molecular basis for organismal diversity on the evolution of regulatory DNA. However, a growing body of work has demonstrated that alterations of binding-site composition and organization often leave regulatory sequence function unchanged 
The potential for significant changes in regulatory sequences to have no functional consequences complicates efforts to identify sequence changes that are likely to affect gene expression and phenotype. But precisely because many of these changes do not affect regulatory output, they provide a powerful opportunity to understand how the arrangement of transcription factor binding sites in a regulatory sequence determines its output. We believe that identifying divergent enhancers that drive similar patterns of expression, and distilling the common principles that unite them, will allow us to decipher the molecular logic of gene regulation.
We began to explore the effectiveness of this approach with the extensively studied regulatory systems of the early D. melanogaster
, using the recently sequenced genomes of 12 Drosophila
species to document the evolutionary fate of transcription factor binding sites in early embryonic enhancers (Peterson, Hare, Iyer, Eisen, unpublished). A consistent pattern emerged: while binding site turnover is common, a large fraction of the binding sites in most enhancers are conserved across the genus (see ).
Binding site conservation and turnover in Drosophila even-skipped stripe 2 enhancer.
The extent to which variation in enhancers from sequenced Drosophila species represented all of the possible variation in these sequences was unclear. Perhaps the conserved sites were an imperturbable core essential for each enhancer's function. Or, perhaps, there had simply not been enough time since the divergence of the genus for mutation to have generated alternative configurations that would produce identical expression patterns. To resolve this ambiguity it was necessary to reconstruct binding site turnover events that occurred over longer evolutionary timescales by comparing Drosophila enhancers to their counterparts in species from outside the genus. The appropriate species for such comparisons would share basic patterning mechanisms with Drosophila species, but be sufficiently diverged from Drosophila to provide significant additional data on the constraints on binding site turnover. Ideally, these species would be amenable to experimental analysis and have fully sequenced genomes.
Unfortunately, the closest available genome sequences were from several very distantly related mosquito species 
, whose most recent common ancestor with Drosophila
lived approximately 220 million years ago. These sequences were unlikely to be informative because of several important differences between early-embryonic patterning in Drosophila
and mosquitoes. Mosquitoes, for example, lack the primary anterior morphogen in Drosophila
, the modified Hox
gene Bicoid, which is found only in higher cyclorrhaphan Diptera (the “true flies”) 
With essentially no information on non-coding sequences and regulatory networks from flies outside the Drosophilidae, we reasoned that other groups within the Acalyptratae, the speciose 100 million year-old division of Diptera that includes Drosophila, represented the best compromise between our aims to maximize sequence divergence and minimize regulatory network divergence.
We selected three families, Sepsidae, Diopsidae and Tephritidae, that span acalyptrate diversity, have well-characterized phylogenies, and contain multiple species whose specimens could be readily obtained. In this paper we present results on gene regulation in sepsids, which, due to their small genomes, were the most amenable to genome analysis.
Specifically, we report the sequence and experimental characterization of the even-skipped
locus from six sepsid species. The particular species were selected to include the major sepsid lineages, and, in several cases, because of the amenability of the species for embryological study. We chose to characterize multiple sepsid species to facilitate the identification of sepsid enhancers by intra-family comparisons 
and to enable comparisons of enhancer evolution between sepsids and drosophilids.