Evolving enhancer-promoter interactions within the tinman complex of the flour beetle, Tribolium castaneum

Insect genomes are highly dynamic and subject to rapid chromosomal rearrangements. Unlike vertebrate genomes, which contain extensive synteny,insect genomes rarely retain similar linkage arrangements(Bolshakov et al., 2002; Zdobnov and Bork, 2007). A small handful of gene complexes represent the main exceptions to this rule of rapid chromosomal reorganization. However, even these are subject to a variety of chromosomal rearrangements among different insects. For example, the ancestral Hox complex is split into two smaller complexes (ANT-C and BX-C) in Drosophila (Negre and Ruiz,2007). Here, we investigate the possibility that rearrangements within such complexes create an opportunity for evolving novel patterns of gene expression. We present evidence that an internal inversion within the tinman gene complex (Tin-C) redirects a conserved enhancer to different target genes in the flour beetle, Tribolium castaneum and in the fruit fly, Drosophila melanogaster.

The Tin-C contains a series of NK homeobox genes that are evolutionarily ancient, pre-dating even the Hox complex. Phylogenetic analysis of homeobox genes in the sponge Amphimedon indicates the existence of a proto-NK gene cluster prior to the eumetazoan radiation(Larroux et al., 2007). Gene linkage in the Tin-C has been conserved in protostomes such as flies, but is essentially lost in deuterostomes (Luke et al., 2003).

In Drosophila, all members of the Tin-C are involved in muscle cell differentiation, and many of the mesodermal patterning functions of Tin-C genes are conserved between flies, annelids and vertebrates. For example, the founding member, tinman (also known as NK4), is expressed in the cardiac mesoderm of all three major animal groups(Azpiazu and Frasch, 1993; Bodmer et al., 1990; Saudemont et al., 2008; Tanaka et al., 1998). Moreover, bagpipe (NK3) is involved in patterning both fly and vertebrate visceral mesoderm (Azpiazu and Frasch, 1993; Tribioli and Lufkin, 1999), whereas ladybird/Lbx,slouch/Nk1, C15/Txl and Msh(Dr)/Msx regulate the patterning of somatic muscle precursors in both flies and annelids(Jagla et al., 2001; Saudemont et al., 2008).

Here, we examine the role of the Tin-C in patterning the cardiac mesoderm,with a particular emphasis on the regulation of the ladybird gene in the heart field of the Tribolium embryo. In Drosophila,ladybird encodes a transcriptional repressor that is thought to be essential for the subdivision of the cardiac mesoderm into distinct pericardial and cardial lineages (Jagla et al., 1997). ladybird is expressed in the pericardial and cardial cells of Drosophila and Apis mellifera (honeybee)embryos, but is absent from the heart field of Tribolium. Evidence is presented that this loss in expression is replaced by the neighboring C15 gene within the Tribolium Tin-C.

The replacement of ladybird expression by C15 appears to result from altered enhancer-promoter interactions arising from an internal inversion within the Tribolium Tin-C. Evidence is presented that this inversion bypasses an insulator located in the ladybird-C15 region. In Drosophila, the cardiac enhancer is located 3′ of the ladybird early and ladybird late genes, and is unable to activate C15 expression owing to insulator activity at the ladybird early promoter. The chromosomal inversion in Tribolium relocates this enhancer so that the ladybirdpromoter is no longer positioned between the enhancer and C15 gene. As a result, the cardiac enhancer is able to activate C15 expression in Tribolium, but not Drosophila. Thus, a novel pattern of gene expression, C15 expression in Tribolium pericardial and cardial cells, is not due to the modification of gene regulatory networks or the de novo evolution of enhancer sequences, but rather results from the interaction of a conserved enhancer with different target genes: ladybird in Drosophila and the neighboring C15 gene in Tribolium. We propose that the redirection of conserved enhancers might be an important mechanism of regulatory evolution.

The D. melanogaster strain used for P-element-mediated transgenesis was yw⁶⁷, as described previously (e.g. Stathopoulos and Levine,2002). pUAST overexpression constructs were balanced using the 612 double balancer (yw; Sp/CyO; Pr Dr/TM3 Sb Ser), and balanced or homozygous males were crossed to twi-GAL4; 24B-GAL4(Lockwood and Bodmer, 2002)virgin females. For promoter blocking, transgenic males were crossed to yw virgin females to prevent two copies of the insulator transgene from pairing in homozygous embryos, and RNA in situs were carried out to detect embryos with the white and lacZ transgenes. All experiments involving P-elements were performed with a minimum of three independent lines. ΦC31-mediated transgenesis was carried out using either 86Fb (Bischof et al.,2007) or an attP2(Groth et al., 2004)pNos-integrase line provided by Michael Eisen (University of California at Berkeley, Berkeley, CA, USA), or usually both. Tribolium castaneumstocks were a gift from Nipam Patel (University of California at Berkeley,Berkeley, CA, USA), and were raised according to standard methods. Apis mellifera hives were obtained from Honeybee Genetics (Vacaville, CA,USA).

DNA fragments were amplified from genomic DNA; primer sequences are available upon request. Drosophila enhancers and the Tribolium ladybird downstream enhancer were cloned into the P-element vector nE2G (Markstein et al.,2004). All remaining fragments were cloned into the attBequivalent of nE2G (eve minimal promoter, lacZ reporter)(Hare et al., 2008). DNA fragments tested for enhancer activity were generally ∼2 kb in length. Full-length Tribolium ladybird and C15 cDNAs were amplified from embryonic Tribolium cDNA generated using a BD Marathon RACE Kit(BD Biosciences). Full-length Drosophila C15 cDNA was generated from an equivalent library using fusion PCR, and all constructs were cloned into pUAST (Brand and Perrimon,1993). Constructs were introduced into the Drosophilagermline as previously described (e.g. Rubin and Spradling, 1982; Groth et al., 2004).

Drosophila, Tribolium and Apis embryo fixation and in situ hybridizations were done as described previously(Kosman et al., 2004; Zinzen et al., 2006). For RNA detection, embryos were hybridized with digoxigenin (Roche) or dinitrophenyl(PerkinElmer) labeled probes and visualized colorimetrically(Jiang et al., 1991) or fluorescently (Kosman et al.,2004), together with the DNA stain DRAQ5 (Biostatus). Probes were generated by in vitro transcription; primer sequences are available upon request. In general, probes were 2-3 kb in length, and encompassed the 5′ regions of the genes. Tc-ladybird and Tc-C15 probes were made using full-length cDNAs, and Apis pannier was made with a 3′ RACE product (Ohara et al.,1989). Eve protein was visualized with the cross-reactive 2B8 antibody from Nipam Patel (Patel et al.,1994), which was diluted to 1:50 and detected using the mouse ABC Elite Kit (Vector Labs), together with fluorescein-conjugated tyramide signal amplification (PerkinElmer).

The Tribolium Tin-C sequence was obtained from the Tcas_2.0 Baylor HSGC assembly. The tiled region consists of ∼240 kb on LG9(12,440,000-12,680,000). NimbleGen designed 50 bp features covering these regions at a density of approximately one feature per 90 bp. Chromatin for immunoprecipitations (IPs) was prepared from ∼15- to 40-hour Tribolium embryos (raised at 30°C) according to Zeitlinger et al.(Zeitlinger et al., 2007b),with the exception that Tribolium embryos were dechorionated for 2 minutes in 100% bleach and were crosslinked for 10 minutes in formaldehyde-buffered hexanes. Tribolium embryo stages were chosen to capture gastrulation and the onset of segmentation, a time when many developmental genes are active. Both ladybird and C15 are expressed at this time, albeit not in large numbers of cells: C15 is expressed in the dorsal ectoderm, ladybird in ectodermal segmental stripes. Chromatin IPs were performed according to NimbleGen's standard protocol(http://www.nimblegen.com),using a combination of anti-Pol II antibodies (H14, H5 and 8WG16, from Covance). Final IP samples were amplified using the WGA2 Kit (Sigma) before being submitted to NimbleGen for labeling and hybridization. Visualization and scaling of tiling data was done with the Integrated Genome Browser(Affymetrix). Drosophila Pol II data was obtained from Zeitlinger et al. (Zeitlinger et al.,2007a). Tribolium microarray data is available from ArrayExpress (Accession number E-TABM-758).

Late Drosophila embryos (8-10 hours) were chosen for 3C experiments because of the expression of Tin-C genes at this stage. Embryos were dechorionated and fixed in formaldehyde-saturated hexanes, as described for ChIP experiments (Zeitlinger et al.,2007b). The embryos were then treated according to a standard 3C protocol (Lanzuolo et al.,2007). In brief, the crosslinked embryos were resuspended in ice-cold lysis buffer and incubated on ice for ∼15 minutes. The embryos were lysed on ice using a Dounce homogenizer, followed by washes in restriction buffer and the chromatin was then dissolved in the appropriate restriction buffer with SDS. The SDS helps to dissolve the insoluble chromatin for restriction enzyme digestion. This dissolved crosslinked chromatin was subjected to EcoRI digestion to completion. The reaction was stopped by adding an excess of SDS and by heat inactivation of the enzyme. The digested chromatin was diluted 16-fold in ligation buffer and subjected to ligation at 18°C for 2 hours, followed by reversal of the crosslinks and purification of the DNA. This 3C library was then subjected to PCR amplification using primer pairs flanking EcoRI sites near the enhancer and promoter sequences. Similar 3C libraries were constructed as experimental controls from non-crosslinked embryos, genomic DNA (to check primer pair efficiency), and crosslinked embryos but without ligation as essential controls.

Apis and Tribolium orthologs of Drosophila heart genes were identified by a reciprocal BLAST-BLAST strategy, and the Clustal alignment of Tin-C genes was generated in BioEdit. ClusterDraw2 has been described previously (Zinzen et al.,2006). The program and the motifs used in this analysis are available online at http://flydev.berkeley.edu/cgi-bin/cld/submit.cgi,except for dTCF, which is described by Chang et al.(Chang et al., 2008).

The Tin-C is conserved in all insect genomes sequenced to date (Diptera,Coleoptera and Hymenoptera), spanning three orders and an estimated 300 million years of evolutionary divergence (see Fig. S1 in the supplementary material). Tin-C orthologs are readily identified by homology in their homeodomains and TN domains (aligned in Fig. 1A and Fig. S1A in the supplementary material). The TN domain, named after Tinman, is an N-terminal protein interaction domain similar to the eh1 domain present in engrailed family proteins, and is required for interaction with the Groucho corepressor protein (Choi et al., 1999; Jagla et al., 2001). With the exception of Drosophila, there is only a single copy of ladybird, indicating that the ladybird duplication seen in Drosophila melanogaster is a relatively recent occurrence(Fig. 1, compare B and C).

The Tin-C appears to contain a chromosomal breakpoint between C15and slouch in Tribolium, and between ladybird and C15 in Apis. However, this latter break might be an assembly error, as the two Apis genes are located on the ends of contigs in the current Apis genome assembly. Furthermore, the ladybirdand C15 linkage is conserved in the genome of the parasitic wasp and fellow hymenopteran Nasonia vitripennis. The orientation of ladybird is inverted in Tribolium as compared with Drosophila (Fig. 1,compare B and C). The tandem arrangement of the ladybird and C15 genes seen in Tribolium is also present in Nasonia (not shown).

ChIP-chip revealed a large spike of RNA polymerase II (Pol II) binding in the promoter region of the Drosophila ladybird early (lbe)gene in the dorsal ectoderm (Fig. 1C). As lbe and ladybird late (lbl) are not transcribed in this tissue, it appears that Pol II is stalled at the lbe, but not the lbl, promoter(Zeitlinger et al., 2007a). Additional Pol II ChIP-chip assays were consistent with the occurrence of stalled Pol II at the ladybird and C15 promoters in Tribolium (white arrows, Fig. 1B).

The expression patterns of Tin-C genes are broadly conserved in Apis,Tribolium and Drosophila (see Figs S2-S4 in the supplementary material) (Janssen and Damen,2008). For example, tinman is expressed in the developing heart in both Apis and Tribolium, whereas the duplicated Msh genes are expressed in overlapping neurogenic stripes and in specific dorsal somatic muscles, similar to the patterns seen in Drosophila (Azpiazu and Frasch,1993; Bodmer et al.,1990; Lord et al.,1995). A striking exception to this general trend of conserved expression is seen for the neighboring ladybird and C15genes in Tribolium.

The ladybird expression patterns seen in Drosophila were found to be conserved in Tribolium, except for the cardiac component of the pattern, which was absent (arrow, Fig. 2B). In addition, ladybird was expressed in the developing larval limbs in Tribolium, an expression pattern that is frequently seen with genes expressed in Drosophila leg imaginal discs (J.D.C. and M.L.,unpublished). The loss of ladybird expression in the Tribolium heart is surprising, as Ladybird is thought to play a crucial role in patterning pericardial and cardial cells in Drosophila (Jagla et al.,1997).

The C15 gene was more broadly expressed in Tribolium than in Drosophila or Apis(Fig. 2, Tribolium;Fig. S3 in the supplementary material, Apis). In Drosophila,C15 is expressed in the dorsal ectoderm(Lin et al., 2006) and in the leg discs, where it works with aristaless to pattern the distal tip of the leg (Campbell, 2005). Drosophila C15 is additionally expressed in a subset of dorsal mesodermal cells outside the cardiac field (M. Frasch, personal communication). In Tribolium, C15 was expressed in the dorsal ectoderm, starting at the onset of gastrulation (not shown), and in the tips of the larval legs, consistent with its role in limb patterning(Fig. 2C). C15 was also broadly expressed along the edge of the germ band encompassing the presumptive heart (arrow, Fig. 2D).

The Tribolium C15 expression pattern partially overlapped tinman staining (Fig. 2E) and was expressed in cells alongside the Eve-positive pericardial cells (Fig. 2F; see Fig. S5 in the supplementary material), as seen for the Drosophila ladybird expression pattern (Cripps and Olson, 2002). However, unlike either the Drosophila ladybird or C15 pattern, Tribolium C15 staining extended beyond the Tribolium tinman domain, and included the cardiac mesoderm and dorsal ectodermal cell types. By contrast, the Apis C15expression pattern was restricted to the imaginal discs and a small number of dorsal cells (arrows, Fig. S3C,D in the supplementary material). Thus, it appears that ladybird expression is replaced by C15 in the cardiac mesoderm of Tribolium, although the remaining components of the expression patterns (such as expression in the central nervous system and imaginal discs) are conserved.

Drosophila and Tribolium are sister groups, whereas Apis is the outgroup of this comparison(Savard et al., 2006), and the gene duplication event that gave rise to ladybird and C15occurred before the protostome/deuterostome split(Larroux et al., 2007). The Apis ladybird and C15 genes are orthologs of the vertebrate Lbx and Tlx genes, respectively(Saudemont et al., 2008). This observation, along with the conservation of expression between Drosophila and Apis, suggests that the loss of ladybird and gain of C15 expression in the cardiac mesoderm is specific to Tribolium.

ladybird and C15 encode similar (but not identical)transcription factors, and contain well-conserved homeodomains, as well as related TN domains (Fig. 1A). It is therefore possible that one protein might be able to substitute for the other. In Drosophila, overexpression of Ladybird throughout the mesoderm represses its direct transcriptional target, even-skipped(eve) (Jagla et al.,2002), whereas eve is expanded in ladybirdmutants (Jagla et al., 1997). Tribolium Ladybird, Tribolium C15 and DrosophilaC15 were overexpressed throughout the mesoderm using the GAL4/UAS system(Lockwood and Bodmer, 2002). All three proteins repressed the late mesodermal eve expression pattern (arrows, Fig. 3). As an internal control, eve expression was largely unaffected in the neurogenic ectoderm, which resides outside the limits of the specific GAL4 mesodermal driver used in these assays. However, there might be some divergence in protein function: Drosophila C15 attenuated Eve expression only in the dorsal somatic muscles and not in the pericardial cells when misexpressed with a weaker GAL4 driver. By contrast, the Tribolium C15 protein repressed Eve equally well in both cell types(see Fig. S6 in the supplementary material).

To determine the basis for the switch in the ladybird and C15 expression patterns in Tribolium, it was necessary to identify the cis-regulatory DNAs within the ladybird-C15 region of the Tin-C in both Drosophila and Tribolium. In Drosophila, mesodermal expression of lbe and lbl is regulated by two enhancers, which were identified by ChIP-chip assays for the pan-mesodermal transcription factor Twist(Sandmann et al., 2007). One of the enhancers is located within the lbl intron(Fig. 4A, fragment 2), and the other is located downstream of the two ladybird genes(Fig. 4A, fragment 1). These enhancers drove nearly identical patterns of reporter gene expression in transgenic Drosophila embryos in the cardiac mesoderm(Fig. 4C,E), starting around stage 9 (see Fig. S7 in the supplementary material).

The expression patterns directed by the ladybird enhancers were complementary to the eve pattern (e.g. Fig. 4E; see Fig. S7 in the supplementary material), consistent with their repression by Eve(Jagla et al., 2002). The expression patterns of the reporter genes were slightly broader than the endogenous patterns (see Fig. S8 in the supplementary material), in that they persisted in most cardial cells (Fig. 4C,E), rather than being restricted to just two cardial cells per segment (Cripps and Olson,2002). This discrepancy can be explained by the loss of repressor sites in the minimal enhancer sequences used in this analysis. Expression of both ladybird enhancers was also drastically reduced or absent in tin^-/- mutant embryos (see Fig. S9 in the supplementary material), which lack the heart and associated dorsal mesodermal lineages(Azpiazu and Frasch, 1993; Bodmer, 1993; Yin and Frasch, 1998).

Previous Twist ChIP-chip assays(Sandmann et al., 2007)identified a putative enhancer downstream of the C15 transcription unit (Fig. 4A, fragment 4; Fig. 4G). When tested in transgenic embryos, this downstream enhancer appeared to recapitulate the endogenous expression pattern in the dorsal ectoderm (see Fig. S7 in the supplementary material), although it was weaker than the previously published intronic C15 enhancer (Fig. 4A, fragment 3) (Lin et al.,2006).

Computational methods were used to identify potential ladybird and C15 enhancers within the Tribolium Tin-C. Particular efforts focused on the identification of binding site clusters for transcription factors known to be involved in the differentiation of the cardiac mesoderm,including Tinman, Eve and GATA factors such as Pannier, Smads, TCF (also known as Pan), Twist and Mef2 (Fig. S10 in the supplementary material shows the cluster analysis) (Cripps and Olson,2002). Putative enhancers were tested for activity in transgenic Drosophila embryos (Fig. 4D,F,H; see Fig. S8 in the supplementary material). Genomic DNA fragments 3 and 4 (indicated in Fig. 4B) appeared to function as C15 enhancers(Fig. 4H; see Fig. S7C in the supplementary material). The downstream Tribolium C15 enhancer(fragment 4) directed reporter expression in the Drosophila dorsal ectoderm and a small number of dorsal mesodermal cells, just like the endogenous Drosophila C15 gene (Fig. 4H; compare with 4G). The intronic enhancer(fragment 3) directed expression primarily in the dorsal ectoderm (see Fig. S7C in the supplementary material). These enhancers are located in roughly the same relative positions as the two Drosophila C15 enhancers (Fig. 4B; compare with 4A). Neither enhancer appeared to direct reporter gene expression in the Drosophila cardiac field,so it would appear that the expression of C15 in the Tribolium heart might depend on enhancers located elsewhere in the Tin-C.

Two potential enhancers were also identified in the vicinity of the Tribolium ladybird gene. One is located downstream of ladybird and upstream of C15(Fig. 4B, fragment 2), and the other is located within the ladybird intron(Fig. 4B, fragment 1). This is similar to the arrangement seen in Drosophila (Fig. 4B; compare with 4A). The Tribolium ladybird intronic enhancer mediated broad reporter expression throughout the developing Drosophila dorsal somatic musculature(Fig. 4F). This intronic enhancer might be a bona fide non-cardiac mesodermal Tribolium ladybird enhancer that is misexpressed in Drosophila embryos due to evolutionary changes in crucial upstream regulators. The downstream 3′ enhancer (fragment 2) directed expression throughout the cardial cells (Fig. 4D), as well as in additional cells located dorsal to the developing heart. These expression profiles closely resemble the endogenous Tribolium C15 expression pattern, and those produced by the corresponding lbe and lblenhancers in Drosophila (Fig. 4D; compare with 4C,E). An additional six tested genomic DNA fragments failed to produce specific patterns of expression in transgenic Drosophila embryos (gray bars in Fig. S10 in the supplementary material; data not shown).

In principle, the Tribolium cardiac enhancer located between ladybird and C15 (fragment 2, Fig. 4B), along with the dorsal ectodermal enhancers within the C15 locus, can account for the complete Tribolium C15 expression pattern. These enhancers are organized in a similar manner in Drosophila and Tribolium. However, the ladybird locus in Tribolium is inverted with respect to Drosophila (Fig. 4B; compare with 4A). This flips the orientation of the Tribolium ladybird gene with respect to C15, although the order of the genes in the Tin-C is unchanged. As a result, the cardiac enhancer that resides downstream of ladybird is now located 5′of the C15 gene in Tribolium. We performed a number of assays to determine whether the novel cardiac C15 expression pattern seen in Tribolium could arise from the interaction of the`ladybird' downstream enhancer with the Tribolium C15promoter.

In Drosophila, the 3′ lbe enhancer might not be able to activate C15 expression in the cardiac mesoderm owing to an insulator DNA sequence located somewhere in the lbe-C15 genomic interval (see summary in Fig. 4A). A recent study suggests that promoters containing stalled Pol II can function as insulators (Chopra et al., 2009). Consequently, the lbe promoter, which contains stalled Pol II (Fig. 1C), might be responsible for blocking C15 expression in the cardiac mesoderm. Thus, the inversion of ladybird in Tribolium might be sufficient to account for the activation of C15 in the cardiac mesoderm, as the downstream ladybirdenhancer is no longer positioned between the ladybird promoter and the C15 transcription start site (see summary in Fig. 4B). This model for the switch in the ladybird and C15 expression patterns in Drosophila and Tribolium was tested using a combination of enhancer-blocking and chromosome conformation capture (3C) assays.

The Tribolium ladybird promoter possessed a robust enhancer-blocking activity when positioned between the IAB5 enhancer and distal white gene (Fig. 5A,B; see also Fig. S11 in the supplementary material). The IAB5 enhancer (Busturia and Bienz,1993) activates lacZ expression in the presumptive abdomen (Fig. 5A), but fails to activate white (Fig. 5B). By contrast, the Tribolium C15 promoter did not function as an insulator, and consequently, the shared IAB5 enhancer activated both the lacZ and white reporter genes in the presumptive abdomen (Fig. 5C,D). We also tested the Drosophila lbl, lbe and C15 promoters for insulator activity using the same assay. In Drosophila, the lbe promoter acted as an insulator(Fig. 6C,D), whereas the lbl (Fig. 6A,B) and C15 (Fig. 6E,F)promoters did not. This is consistent with the occurrence of stalled Pol II at the lbe promoter, but not at the lbl or C15promoters (Fig. 1C). However,there is not always a perfect correlation between stalled Pol II and enhancer blocking.

The Tribolium ladybird and C15 promoters appear to contain stalled Pol II (see Fig. 1B). Nonetheless, only ladybird exhibits insulator activity in these assays. Similar results were observed in a recent study using the same white/lacZ expression vectors(Chopra et al., 2009). Most,but not all, stalled promoters were found to possess an insulator activity. For example, the gooseberry promoter contains stalled Pol II, whereas the linked gooseberry-neuro promoter does not. Neither promoter was found to contain insulator activity, suggesting that only a subset of stalled promoters mediate enhancer blocking(Chopra et al., 2009) (see Discussion).

Chromosome conformation capture (3C) permits the direct detection of long-range enhancer-promoter interactions(Dekker et al., 2002; Lanzuolo et al., 2007). This technique involves crosslinking embryos, restriction digestion of the crosslinked chromatin, ligation of noncontiguous sequences and PCR amplification. We used 3C assays to examine the regulation of the two ladybird genes in Drosophila. These assays suggest that the ladybird enhancers interact with the lbe and lblpromoters, but not the C15 promoter(Fig. 6G). The lblintronic enhancer appeared to activate lbl(Fig. 6G, lane 1), whereas the 3′ enhancer bypassed the lbl promoter (which lacks insulator activity) to interact with the distal lbe promoter(Fig. 6G, lane 2). Similar preferential enhancer-promoter interactions have been documented in the Drosophila Hox complexes (Ohtsuki et al., 1998).

The lack of a blocking activity by Drosophila lbl can account for the long-range activation of lbe by the 3′ enhancer(Fig. 6G). Moreover, this enhancer should be `captured' by the lbe promoter/insulator, and therefore fail to activate the neighboring C15 gene. The inversion of the ladybird-C15 region in Tribolium causes the 3′cardiac enhancer to be positioned on the other side of the ladybirdpromoter/insulator, where it is equidistant between the promoters for the ladybird and C15 genes. As a result, the Tribolium ladybird promoter is not expected to block activation of C15expression in the cardiac mesoderm (Fig. 7).

There are a growing number of examples of enhancer modifications resulting in changes in gene expression and morphology during animal evolution. For example, regulatory changes in pitx1 lead to a reduction in the pelvic fins of certain populations of stickleback fish(Shapiro et al., 2004). Moreover, regulatory changes in the yellow locus have been implicated in the diverse patterns of wing spots among divergent drosophilids(Prud'homme et al., 2006). Additionally, pre-existing enhancers are known to evolve through binding site turnover. Such changes can be neutral and cause little or no alterations in gene expression (Ludwig et al.,1998). Alternatively, such changes can lead to novel patterns of gene expression as binding sites for new transcription factors are gained or lost, as seen for the honeybee sim midline enhancer(Zinzen et al., 2006).

In the present study, we have presented evidence for a distinctive mechanism of regulatory evolution. Namely, a chromosomal inversion within the Tin-C results in a novel pattern of C15 gene expression in the cardiac mesoderm of Tribolium (summarized in Fig. 7). Previous studies provide evidence that some of the most dramatic developmental mutants result from chromosomal rearrangements. For example, the dominant Antennapedia mutation in Drosophila is caused by a chromosomal inversion that results in the ectopic expression of Antennapedia in the head, thereby transforming antennae into legs(Frischer et al., 1986).

What are the consequences of replacing Ladybird with C15 in the Tribolium heart? Although ladybird and C15 probably arose from the duplication of a single ancestral gene, the two have diverged significantly in function and protein sequence in flies and vertebrates(Jagla et al., 2001). Outside the DNA-binding and TN domains, the two proteins are extensively different,and might interact with different protein partners. It is therefore likely that exchanging one protein for the other is not a neutral change, even though Tribolium C15 can repress the Drosophila Ladybird target eve in the fly cardiac mesoderm (see Fig. 3). Expression arrays in Drosophila have demonstrated that Ladybird has a wide variety of transcriptional targets, including not only transcription factors involved in specifying cell identity during heart patterning, but also genes required for cell shape changes, adhesion and motility, and late-acting genes involved in myogenesis (Junion et al.,2007). Thus, the substitution of Ladybird with C15 might subtly change the functional or morphological properties of the Triboliumheart, such as alterations in contractile strength, beating frequency, cell migration or attachment.

We have presented evidence that the switch in expression in the Tribolium heart field is due to a chromosomal inversion in the ladybird-C15 region of the Tin-C(Fig. 7). This inversion appears to foster the activation of C15 by a cardiac enhancer located 3′ of the ladybird locus. In Drosophila, this interaction appears to be inhibited by an insulator located at the intervening lbe promoter. In Tribolium, the altered orientation of ladybird and C15 places the conserved cardiac enhancer between the two genes. As a result, the ladybird insulator cannot block enhancer-C15 interactions, as it no longer resides between the enhancer and the C15 promoter, as seen in Drosophila. In principle, the cardiac enhancer should be able to activate both ladybird and C15 expression because it is located at an equidistant position between the two genes. It is possible that preferential activation of C15 expression results from promoter competition,whereby C15 sequesters the cardiac enhancer (e.g. Choi and Engel, 1986). Alternatively, the C15 promoter region might contain `tethering'elements that preferentially recruit the cardiac enhancer (e.g. Calhoun and Levine, 2003).

We do not believe that the switch in enhancer-promoter interactions observed in this study is a unique attribute of the Tin-C. Inversions are quite common in insect genomes. By a recent estimate, only 10% of all genes share the same neighbors in the Drosophila and Apis genomes(Zdobnov and Bork, 2007). In principle, genome rearrangements have the potential to create novel patterns of gene expression by bringing new promoters under the influence of pre-existing enhancers. Unlike the more popularly documented examples of enhancer evolution through binding site turnover, changes in enhancer-promoter interactions not only produce novel patterns of gene expression, but also have the potential to introduce a new protein into a pre-existing developmental gene network in a single step.