Introduction

‘The art of progress is to preserve order amid change and to preserve change amid order.’ Alfred North Whitehead

Mutations are inevitable. Biological systems maintain their function in the face of genetic changes, while preserving flexibility that allows the system to adapt to new environments. Genomic regulatory networks that control gene expression are no exception. These networks are composed of highly conserved trans-regulatory proteins and cis-regulatory DNA sequences that specify gene expression patterns (Davidson, 2001). Comparisons of cis-regulatory elements among Diptera (ie ‘true’ flies) indicate that their sequences are robust to mutational changes, yet receptive to functional divergence.

Molecular mechanisms that control protein expression facilitate both developmental stability and evolutionary change. Proteins required for the development of characters shared among Diptera typically have conserved expression patterns (eg Averof and Patel, 1997; Panganiban et al, 1997). Identifying cis-regulatory sequences mediating conserved regulatory inputs helps unravel genomic regulatory networks. Traits that differ among Dipteran species, such as body coloration, bristle patterns, and larval hairs, often correlate with divergent expression of developmental proteins (Stern, 1998; Sucena and Stern, 2000; Wulbeck and Simpson, 2000; Pistillo et al, 2002; Wittkopp et al, 2002; Gompel and Carroll, 2003) (Figure 1). Cis-regulatory sequences that control transcription are a common source of divergent protein expression patterns and thus of phenotypic change (Carroll et al, 2001).

Figure 1
figure 1

Divergent protein expression correlates with divergent phenotypes. Dorsal abdominal cuticle from adult flies (a–c) is shown above developing abdominal tissue (d–f) from Drosophila melanogaster, D. subobscura, and D. virilis, respectively. The distribution of the Yellow protein, which controls the production of black pigment, is shown in the color yellow in lower panels. Note that the expression of Yellow correlates with both the pattern and intensity of black pigment in adult flies (Wittkopp et al, 2002).

Full size image

Here, I examine comparative studies of gene regulation among Diptera. For reviews of regulatory evolution that encompass more taxa, see Stern (2000), Tautz (2000), Ludwig (2002), Simpson (2002), Wray et al (2003). After providing an overview of cis-regulatory architecture and molecular evolution, I review case studies that compare the sequence and function of cis-regulatory elements among species. Properties of regulatory systems that allow cis-regulatory sequences and function to evolve at different rates are discussed, illustrated by case studies where available. Understanding the structure–function relationship of cis-regulatory regions is essential for comparative genomic studies of gene regulation. I conclude by examining computational approaches for identifying cis-regulatory regions and arguing that additional biochemical, genetic, and transgenic studies are sorely needed to improve computational tools.

Enhancers control patterns of gene expression

Expression of protein coding sequences is controlled by cis-regulatory regions, which include a ‘basal promoter’ and one or more ‘enhancers’ (Figure 2a). The basic structure of cis-regulatory regions is shared not only among Diptera but among all eukaryotes. For more comprehensive reviews of cis-regulatory architecture, see Arnone and Davidson (1997), Carroll et al (2001), Davidson (2001), Smale (2001), Arnosti (2003) and Wray et al (2003).

Figure 2
figure 2

Reporter genes are used to ascertain enhancer activity. (a) A schematic of the D. melanogaster yellow gene is shown, containing two exons, a basal promoter, and multiple tissue-specific enhancers. (b) Reporter genes contain enhancer sequences, a generic basal promoter, and coding sequences for an easily visualized protein, such as the Green Fluorescent Protein (GFP) or β-galactosidase (β-gal). The composition of a GFP reporter gene for the D. subobscura ‘body’ enhancer is illustrated. (c) When transformed into a fly, the reporter gene provides a read-out of enhancer activity. GFP expression driven by the D. subobscura body enhancer at a late pupal stage is shown in a D. melanogaster transformant fly on the right, with a wild type fly on the left. Note that reporter gene expression within each abdominal segment (bracket) is comparable to expression of the endogenous D. subobscura Yellow protein shown in Figure 1e. (Expression in eyes and ocelli is due to Pax6-GFP transformation marker, not the D. subobscura body enhancer.)

Full size image

Basal (or ‘core’) promoters are necessary for transcription, but do not provide spatiotemporal information for gene expression. They contain binding sites for the general transcription machinery, including the TATA-binding protein and the RNA polymerase II protein complex. As these proteins must bind promoters throughout the genome, and because the assembly of the protein complex is strictly required for the production of mRNA, basal promoter sequences and the proteins that bind to them are under strong functional constraint. This constraint is visible as a reduced level of polymorphism and divergence in these regions among Drosophila species (Kohn et al, 2004). The few polymorphisms that do exist in basal promoters appear to contribute little to variable gene expression among strains of D. melanogaster (Brown and Feder, 2005). Genes can also contain multiple, alternative promoters that are active under different cellular conditions (Ayoubi and Van De Ven, 1996). The contribution of alternative promoters to regulatory divergence remains unclear.

Changes in enhancer sequence are a common cause of cis-regulatory divergence (eg Fang and Brennan (1992), Ross et al (1994), Wittkopp et al (2002). Enhancer sequences specify when, where, and how much mRNA will be transcribed from the associated coding sequence. They are modular (ie function independently) and many genes contain more than one enhancer element, with each directing a subset of the total gene expression pattern. Enhancers are also composed of binding sites for transcription factor proteins, but, unlike basal promoters, each enhancer contains binding sites for a unique combination of transcription factors. Once bound, transcription factor proteins interact with each other and the polymerase protein complex assembled on the basal promoter to activate and sustain transcription. The specific combination of transcription factors assembled on an enhancer determines its activity.

As enhancers are modular, their activity can be determined using transgenic ‘reporter genes’ (Barolo et al, 2000). These constructs contain putative cis-regulatory sequences and a basal promoter that drives expression of an easily visualized reporter protein (Figure 2b). After transforming a reporter gene into a host species, its expression is determined (Figure 2c). D. melanogaster P-elements (Spradling and Rubin, 1982) are the most commonly used transformation system for assaying Dipteran enhancers.

When transformed into D. melanogaster, heterologous cis-regulatory sequences are regulated in trans by D. melanogaster transcription factors. This has both advantages and disadvantages. Assaying orthologous cis-regulatory elements in a common trans-regulatory background allows their functions to be directly compared. However, if properties of the trans-regulators (eg transcription factors) have diverged between the donor and host species, the activity of the cis-regulatory element in D. melanogaster will differ from its activity in the species from which it was derived. Comparing the activity of cis-regulatory sequences transformed into multiple species provides the most complete view of regulatory evolution (Cavener, 1992; Christophides et al, 2000; Wittkopp et al, 2002; Lombardo et al, 2005).

Molecular evolution of cis-regulatory sequences

Enhancers and basal promoters are subject to the same process of molecular evolution as all other regions of the genome. Nucleotide substitutions, insertions, deletions, and rearrangements arise, and the balance of selection and drift determines their survival over time (Li, 1997). Studies of polymorphism and divergence in cis-regulatory regions of Diptera provide evidence both for and against models of neutral sequence evolution (Ludwig and Kreitman, 1995; Hancock et al, 1999; Kohn et al, 2004; Phinchongsakuldit et al, 2004; Andolfatto, 2005). Each of these studies employs different population genetic models and tests for selection; the most appropriate model for the neutral evolution of cis-regulatory regions is not yet established.

cis-Regulatory mutations can influence phenotypes by altering gene expression. Therefore, selection coefficients for cis-regulatory changes should be related to their effects on expression. Mutations that do not alter expression are assumed to be neutral (ie ‘silent’), whereas mutations that disrupt sequences essential for cis-regulatory function are assumed to be deleterious. Sequences comprising transcription factor binding sites may thus be more constrained than sequences not used as binding sites. Surprisingly, the pattern of nucleotide substitutions is similar within characterized binding sites and in surrounding regions of DNA, suggesting this may not be the case (Emberly et al, 2003; Costas et al, 2004; Phinchongsakuldit et al, 2004; Balhoff and Wray, 2005). It remains to be seen how often ‘surrounding’ sequences contain unidentified binding sites, and how sequence divergence affects gene expression.

With few exceptions (Erives and Levine, 2004; Markstein et al, 2004; Senger et al, 2004), the architecture of binding sites in an enhancer, and the nature of interactions among transcription factors that regulate its activity, are not understood well enough to predict the consequences of specific cis-regulatory changes from sequence alone. Empirical tests are required to determine cis-regulatory function and to assess the impact of sequence divergence on gene expression.

Conserved sequence and function

Expression patterns conserved across species are specified by cis-regulatory elements that have preserved their function over time. Cis-regulatory sequences from other Dipteran species often retain their activity when introduced into D. melanogaster using transgenes. This is true for sequences taken from other Drosophila species, distantly related flies (including the house fly Musca domestica and the black fly Simulium vittatum) and even animals outside Diptera (Mitsialis and Kafatos, 1985; Martin et al, 1988; Langeland and Carroll, 1993; Magoulas et al, 1993; Lukowitz et al, 1994; Pan et al, 1994; Xiong and Jacobs-Lorena, 1995; Tortiglione and Bownes, 1997; Ludwig et al, 1998; Wolff et al, 1999; Wittkopp et al, 2002).

The simplest mechanism for maintaining activity of a cis-regulatory element is to conserve the sequences that determine its function. Sequence comparisons of orthologous cis-regulatory elements generally show blocks of conserved sequence surrounded by more divergent sequences (Kassis et al, 1985; Wilde and Akam, 1987; Kassis et al, 1989; Langeland and Carroll, 1993; Lukowitz et al, 1994; Pan et al, 1994; Sackerson, 1995; Ludwig et al, 1998; Wolff et al, 1999; Kim, 2001; Dellino et al, 2002; Emberly et al, 2003; Berman et al, 2004; Costas et al, 2004). Sequence similarity to D. melanogaster cis-regulatory regions has been used to identify enhancers in other Dipterans, including Anopheles gambiae (Papatsenko and Levine, 2005), Scaptodrosophila lebanonensis (Papaceit et al, 2004), and Calliphora vicina (Gibert and Simpson, 2003).

Evolutionary comparisons of well-characterized D. melanogaster enhancers were used to help motivate the sequencing of genomes from other Dipterans. The expectation a priori was that most cis-regulatory regions would be easily identified based on sequence similarity to noncoding sequences of the D. melanogaster genome (Hardison, 2000; Bergman et al, 2002). Unfortunately, this does not appear to be the case. With the completion of the D. pseudoobscura genome sequence, researchers found that computational searches for conserved sequences only identified a small fraction of the enhancers in the genome (Richards et al, 2005). Ascertainment bias in early empirical studies of cis-regulatory regions may have overestimated the requirement for sequence conservation; Dipteran enhancers were often identified based on sequence similarity to D. melanogaster cis-regulatory regions, and many D. melanogaster enhancers were recognized precisely because they evolved slower than surrounding sequences.

Conserved function despite divergent sequence

Comparisons of cis-regulatory elements among Diptera demonstrate that enhancer activity can be maintained despite extensive sequence divergence (Martin et al, 1988; Magoulas et al, 1993; Tortiglione and Bownes, 1997; Wolff et al, 1999; Ludwig et al, 2000). This phenomenon has been most clearly illustrated in the stripe 2 enhancer of the even-skipped gene, described below.

Even-skipped (eve) encodes a transcription factor that plays a similar role in embryonic patterning of Drosophila, Anopheles, and presumably all Diptera (Goltsev et al, 2004). Eve protein is expressed in seven transverse stripes along the embryo, which are controlled by five independent enhancers. The D. melanogaster eve stripe 2 enhancer includes binding sites for five transcription factors (including two activators and three repressors) that are required for expression of the Eve protein in stripe 2 (Stanojevic et al, 1991). Although these binding sites were required for activity in the D. melanogaster enhancer that was dissected experimentally, sequence variation within and between species is comparable to other noncoding regions and fits a model of neutral sequence evolution (Ludwig and Kreitman, 1995).

To determine the functional consequences of sequence divergence in the eve stripe 2 enhancer (Figure 3), Ludwig et al isolated DNA orthologous to the D. melanogaster enhancer from D. yakuba, D. erecta, and D. pseudoobscura, and assayed its activity in transgenic D. melanogaster using reporter genes. Despite little sequence similarity (including divergence of binding sites essential for expression of the D. melanogaster enhancer) orthologous enhancers were able to drive gene expression in a pattern comparable to the D. melanogaster eve stripe 2 enhancer (Ludwig et al, 1998). Chimeric enhancers were constructed between the D. pseudoobscura and D. melanogaster alleles, each containing the 5′ and 3′ regions from different species, and introduced into D. melanogaster (Ludwig et al, 2000). The chimeric enhancers did not function properly, indicating that compensatory changes have evolved since the split of D. melanogaster and D. pseudoobscura. Orthologous enhancers that produce the same expression pattern despite differences in the arrangement of binding sites have presumably evolved under stabilizing selection.

Figure 3
figure 3

Enhancer activity is conserved despite divergent binding sites. (a) The function of the even-skipped (eve) stripe 2 enhancer is conserved between D. melanogaster and D. pseudoobscura (Ludwig et al, 1998, 2000, 2005). Expression patterns of reporter genes containing eve stripe 2 enhancers from both species are depicted in schematic D. melanogaster embryos. (b) Despite functional conservation of eve enhancers, >80% of characterized binding sites have diverged between species. Locations of binding sites for the Bicoid (circle), Hunchback (square), Kruppel (oval), Giant (rectangle), and Sloppy-paired (inverted triangle) transcription factor proteins are shown for the D. melanogaster and D. pseudoobscura eve stripe 2 enhancers. Binding sites with divergent sequences are shaded; arrows indicate binding sites unique to the D. melanogaster enhancer. Adapted from Ludwig et al (2005).

Full size image

Recently, the D. melanogaster, D. yakuba, D. erecta, and D. pseudoobscura eve stripe 2 enhancers were tested for their ability to rescue an eve mutant phenotype (Ludwig et al, 2005). The D. yakuba and D. pseudoobscura eve alleles restored a wild-type phenotype, but the stripe 2 enhancer from D. erecta, a species which is more closely related to D. melanogaster than is D. pseudoobscura, failed to complement the mutation. Sequence divergence of the D. erecta stripe 2 enhancer may be such that the D. erecta allele requires sequences outside of the region orthologous to the D. melanogaster enhancer. Alternatively, the activity and/or expression level of transcription factors regulating the element may have diverged between species. Further experimentation will distinguish among these possibilities.

If transcription factor binding sites can diverge between species while maintaining enhancer function, then polymorphisms in binding sites may also be segregating within species. Indeed, analysis of another enhancer of the eve gene revealed an experimentally confirmed and phylogenetically conserved binding site segregating in natural populations (A Palsson, M Ludwig, and M Kreitman, personal communication), indicating that empirically validated biding sites are not necessarily fixed within species. A cluster of binding sites was also recently found to be polymorphic in a sea urchin cis-regulatory element (Balhoff and Wray, 2005). Such intraspecific variation provides raw material for changing enhancer sequences while maintaining enhancer function.

Uncoupling enhancer sequence and function

How can enhancer activity be maintained despite overall sequence divergence? Molecular mechanisms that translate cis-regulatory sequences into gene expression patterns allow them to evolve at different rates. Features of regulatory mechanisms that can separate the evolution of cis-regulatory sequence and function include: biochemical properties of transcription factors, redundant binding sites and enhancers, changes in transcription factor inputs, and coevolution of transcription factors and their binding sites.

Properties of transcription factors

Flexibility in transcription factor binding (ie ‘degeneracy’) as well as flexibility in the arrangement and spacing of transcription factors permit many sequence changes to evolve without altering enhancer function (Arnone and Davidson, 1997). Degenerate binding sites allow a transcription factor to continue regulating an enhancer despite sequence divergence. Flexible cis-regulatory architecture allows binding sites to be reshuffled while maintaining cis-regulatory function.

Redundant binding sites and enhancers

Redundant transcription factor binding sites within an enhancer also facilitate sequence divergence. If individual binding sites can be mutated with minimal disruption to enhancer function, compensatory binding sites can produce a fluid restructuring of cis-regulatory regions. For example, expression of the Drosophila spalt and knot genes is repressed in the developing haltere by the Ultrabithorax (Ubx) homeodomain protein. Multiple Ubx binding sites are present in enhancers for both genes and the loss of individual binding sites has minimal effect on cis-regulatory activity (Galant et al, 2002; Hersh and Carroll, 2005). Distinct Ubx binding sites have evolved to repress knot expression in the halteres of D. melanogaster and D. pseudoobscura (Hersh and Carroll, 2005). Redundant binding sites may also promote sequence divergence and reorganization of the eve stripe 2 enhancer (Ludwig et al, 2000) and yolk protein genes (Piano et al, 1999). Redundancy among enhancer modules (Buttgereit, 1993; Piano et al, 1999; Pappu et al, 2005) also permits sequence changes to accumulate with out affecting cis-regulation; if one element is altered, the redundant element can compensate for its function.

Changing transcription factor inputs

Developmental system drift (DSD, True and Haag, 2001) can create enhancers with conserved functions but greatly diverged sequences. DSD occurs when the output of a developmental system remains the same despite the evolution of underlying developmental mechanisms. An apparent case of DSD is embryonic patterning between the mosquito, Anopheles gambiae, and the fruit fly, D. melanogaster. Expression of the Even-skipped protein appears to be conserved between species, but expression patterns of the genes that regulate eve in D. melanogaster are different in Anopheles (Goltsev et al, 2004). These data suggest that Anopheles cis-regulatory elements of eve are controlled by different transcription factors than the D. melanogaster eve enhancers. DSD may also contribute to the restructuring of the D. erecta eve stripe 2 enhancer (Ludwig et al, 2005).

Coevolution of transcription factors and binding sites

Evolutionary changes in the DNA binding domains of transcription factors promote the divergence of cis-regulatory sequences. An example from Diptera is the coevolution of the binding domain of the Bicoid transcription factor and the cis-regulatory sequences of the hunchback (hb) enhancer. The developmental function of the Hunchback protein in early embryonic patterning is conserved among Drosophila, the housefly Musca domestica, and blowflies Lucilia sericata and Calliphora vicina (Sommer and Tautz, 1991; Bonneton et al, 1997; McGregor et al, 2001a). Despite this conservation, hb cis-regulatory elements have undergone changes in primary sequence that affect the number and organization of binding sites, especially for Bicoid (Hancock et al, 1999; McGregor et al, 2001b). Biochemical and transgenic experiments demonstrate that the DNA binding domain of the Bicoid protein coevolved with the binding sites in the hb promoter to maintain their regulatory interaction (Shaw et al, 2002). Similar coevolution of transcription factor specificity and cis-regulatory binding sites has been invoked to explain divergent bristle locations between D. melanogaster and D. simulans (Skaer et al, 2002) and as a possible cause for the divergence of the D. erecta eve stripe 2 enhancer (Ludwig et al, 2005). However, DNA binding domains of transcription factors are among the most conserved sequences in animal genomes, and it is unclear whether coevolution of transcription factors and their binding sites is a common feature of regulatory evolution.

Sources of divergent cis-regulatory activity

Although many cis-regulatory elements maintain their function over time, changes in gene expression among Dipteran species are also common. During the 2 million years since the divergence of the D. melanogaster and D. simulans lineages, up to half of the genes in the genome have evolved differences in their expression level (Ranz et al, 2003; Rifkin et al, 2003). The majority of these changes appear to be caused by functional divergence of cis-regulatory sequences associated with the affected gene (Wittkopp et al, 2004). Cis-regulatory elements that specify new or altered expression patterns can evolve (1) de novo, (2) by divergence of parologous enhancers following duplication, or (3) through the modification of existing enhancers. Studies providing evidence for these modes of cis-regulatory divergence are reviewed below.

Enhancer evolution de novo

Theoretically, cis-regulatory elements controlling evolutionarily novel patterns of gene expression may arise de novo. Stone and Wray (2001) simulated neutral sequence evolution by point mutations, while MacArthur and Brookfield (2004) simulated enhancer evolution using a model that incorporates positive selection. Both studies concluded that transcription factor binding sites appear frequently and can be fixed in a population over relatively short periods of time. Currently, there is no empirical evidence of an enhancer evolving de novo, but lineage-specific enhancers derived from neutral sequences may be very difficult to identify.

Duplication and divergence

The function of an enhancer can be altered following gene duplication. When a gene, including its cis-regulatory sequences, is duplicated, the two copies are redundant and one is free to change its expression pattern (Li and Noll, 1994; Lynch and Force, 2000). Indeed, Gu et al (2004) found that duplicated genes are more likely to have evolved expression differences between Drosophila species than single copy genes. Parologous genes have been identified in D. melanogaster for which the expression change seems to be the primary difference among duplicates; the protein functions remain interchangeable (Rodriguez et al, 1990; Li and Noll, 1994). A pair of duplicated genes with divergent expression has also been identified in the medfly, Cereatitis capitata (Christophides et al, 2000). In a rare experiment using transgenic flies other than D. melanogaster, the authors showed that despite extensive sequence similarity, the parologous regulatory elements had evolved differences in tissue-specific and sex-specific expression by altering cis-regulatory activity. The two, ∼280 basepair (bp) cis-regulatory sequences differ by only 12 divergent bases and seven deleted nucleotides, indicating that patterns of gene expression can be dramatically altered with minimal differences in enhancer sequence.

Modified enhancer activities

Gene expression can diverge by altering the function of existing cis-regulatory elements. For example, the Alcohol dehydrogenase (Adh) gene shows differences in its spatiotemporal expression pattern among Drosophila species that are caused by modifications of cis-regulatory sequences (Dickinson et al, 1984; Fang et al, 1991; Papaceit et al, 2004). Similarly, changes in expression of the Yellow protein that correlate with differences in abdominal pigmentation among D. melanogaster, D. subobscura, and D. virilis, are caused by functional divergence of orthologous enhancers (Wittkopp et al, 2002). Cis-regulatory changes are also responsible for differences in the expression of the glucose dehydrogenase (gld) gene among flies (Schiff et al, 1992). The presence and absence of a five base pair sequence (TTAGA) in the gld enhancer correlates with the expression of Gld protein in the ejaculatory duct among Drosophila species (Ross et al, 1994), implying that enhancer functions may be modified by only a few changes in cis-regulatory sequence.

Enhancers controlling lineage-specific patterns of gene expression can also evolve from existing cis-regulatory elements, taking advantage of transcription factor binding sites already present. The D. biarmipes enhancer regulating expression of the pigmentation gene yellow in a male-specific wing spot was recently identified (Gompel et al, 2005). This enhancer sequence is orthologous to the cis-regulatory element controlling ubiquitous wing expression in D. melanogaster. Only 6 bp of the 675 bp enhancer are essential for activation in the wing spot (B Prud'homme and S Carroll, personal communication). Transcription factor(s) that bind to these sequences have not yet been identified. Repression of the spot expression in the posterior compartment is controlled by two, 8 bp binding sites for the Engrailed transcription factor that have evolved in the D. biarmipes lineage. Again, changing only a handful of nucleotides produces major changes in cis-regulatory activity.

Computational enhancer prediction

Understanding the relationship between sequence and function is essential for developing computational methods to study cis-regulatory elements. Comparative genomic approaches hold great promise for accelerating studies of gene regulation in Diptera. Current methods can be broken into two general classes (Bulyk, 2003): (1) phylogenetic footprinting, which uses sequence conservation among two or more species to identify putative cis-regulatory sequences (eg Moses et al, 2004; Siepel et al, 2005) and (2) motif detection, in which statistical models are used to recognize binding sites for specific transcription factors or sequences shared among co-regulated genes (eg Markstein et al, 2002). Phylogenetic footprinting may miss many cis-regulatory elements because, as discussed in this review, sequence conservation is not strictly required for cis-regulatory function. Indeed, the first comparison of two Drosophila genome sequences failed to uncover many known cis-regulatory elements (Richards et al, 2005). Motif detection algorithms, especially those that look for clusters of experimentally defined binding sites, can identify new enhancers without searching for linear sequence conservation (eg Bergman et al, 2002; Markstein et al, 2002; Berman et al, 2004; Erives and Levine, 2004; Papatsenko and Levine, 2005). However, this approach is limited to finding enhancers regulated by transcription factors with known binding sites. The two strategies can also be combined to improve the accuracy of enhancer prediction (Grad et al, 2004).

Back to the bench

Despite the promise of computational approaches for studying cis-regulatory evolution, analyses of additional enhancers using biochemical, genetic, and transgenic tools is essential for refining these methods. Empirical studies that elucidate enhancer structure and function will allow powerful motif finding algorithms to be used more broadly. The experimental characterization of cis-regulatory regions has historically been time-consuming and labor-intensive. Fortunately, high-throughput techniques for studying transcription factor binding on a genomic scale are now available (Sun et al, 2003; van Steensel et al, 2003; Bergman et al, 2005), putting a complete list of transcription factor binding sites in the D. melanogaster genome within reach. Functional, in vivo, tests of enhancer function in transgenic flies remain the rate-limiting step. Methods that automate the production, isolation and characterization of transgenic flies carrying reporter genes will expedite this work. In addition, tests of enhancer function in D. melanogaster must be supplemented with tests of enhancer function in other Dipteran hosts to uncover changes in trans-regulation (Cavener, 1992; Christophides et al, 2000; Wittkopp et al, 2002; Lombardo et al, 2005), minos (Loukeris et al, 1995), piggyBac (Handler et al, 1998), mariner (Coates et al, 1998), and Hermes (Jasinskiene et al, 1998) transposable elements, engineered with dominant fluorescent transformation markers (Berghammer et al, 1999), are now available for transforming diverse insect species (Handler, 2002; Wimmer, 2003).

The complex relationship between cis-regulatory sequence and function is only beginning to be uncovered. With the sequence of Anopheles and 12 Drosophila genomes completed or in progress – and advanced transgenic and functional genomic tools already available – studies of Dipteran cis-regulatory sequences are poised to continue providing valuable insights into the process of regulatory evolution.