Tinman is a direct activator of midline in the drosophila dorsal vessel

Authors

  • Jae-Ryeon Ryu,

    1. Genes and Development Research Group, Department of Biochemistry and Molecular Biology, Department of Medical Genetics, University of Calgary, Calgary AB, Canada
    Search for more papers by this author
  • Nima Najand,

    1. Genes and Development Research Group, Department of Biochemistry and Molecular Biology, Department of Medical Genetics, University of Calgary, Calgary AB, Canada
    Search for more papers by this author
  • William J. Brook

    Corresponding author
    1. Genes and Development Research Group, Department of Biochemistry and Molecular Biology, Department of Medical Genetics, University of Calgary, Calgary AB, Canada
    • Genes and Development Research Group, Department of Biochemistry and Molecular Biology, Department of Medical Genetics, University of Calgary, 3330 Hospital Drive NW, Calgary AB, Canada T2N 4N1
    Search for more papers by this author

Abstract

Heart development requires a conserved core of transcription factors comprised of Nkx2.5, GATA and T-box family transcription factors. In Drosophila melanogaster, the Nkx2.5 gene tinman acts upstream of many cardiac genes including the Tbx20 homolog midline, a critical regulator of heart development in both flies and vertebrates. By testing genomic fragments containing clusters of consensus Tinman-binding sites, we identified a 4.3 kb fragment 5′ of midline that directs reporter expression in all midline-expressing heart cells and a 1.7 kb subfragment that drives reporter expression in mid-expressing heart cells that maintain tin expression. Both fragments direct reporter gene expression in response to tinman in transgenic embryos and in transient transfection assays in Drosophila S2 cells. Mutation of two Tinman binding sites (Tin1 and Tin2) reduces or abolishes cardiac expression in derivatives of the 1.7 kb fragment. We conclude that Tin is a direct regulator of midline in fly heart development. Developmental Dynamics, 2011. © 2010 Wiley-Liss, Inc.

INTRODUCTION

The Drosophila melanogaster heart, or dorsal vessel, is a simple muscular tube with a single chamber and fewer than 300 cells. It is the only component of the fly's open circulatory system, acting to circulate hemolymph throughout the body cavity. The heart arises from the dorsal embryonic mesoderm, and the lineage of all heart cells is known (Cripps and Olson,2002). Additionally, many of the genetic pathways controlling fly heart development have been identified. A key discovery was the Drosophila Nkx2.5 homeodomain transcription factor gene tinman (tin), which is required for the development of all dorsal mesoderm structures (Azpiazu and Frasch,1993; Bodmer,1993). Tin acts upstream of numerous genes known to be expressed in the heart and the analysis of the regulation and function of tin has uncovered a complex regulatory network that controls the development of the Drosophila heart (Reim and Frasch,2010).

The fly heart is vastly simpler than the vertebrate heart. Yet despite the obvious morphological differences, the fundamental core of signaling molecules and transcription factors controlling heart development is conserved between Drosophila melanogaster and vertebrates (Cripps and Olson,2002; Zaffran and Frasch,2002). The Tin/Nkx2.5, GATA zinc finger, and T-box transcription factors form the core of a gene regulatory network essential not only for fly heart development, but which also act in vertebrate heart development (Olson,2006). The cardiac expression and function of the Tbx20 class of T-box transcription factors is conserved between insects and vertebrates (Stennard and Harvey,2005). The duplicate genes midline and H15 (also referred to as neuromancer 2 and neuromancer 1) are the Drosophila members of the Tbx20 class, and are expressed in all cardioblasts, the cells that form the contractile and valve cells in the dorsal vessel. The two genes are partially redundant and together are essential for the proper polarity and alignment of the cardioblasts in the heart tube (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). They are also required for distinguishing between the valve and contractile cardioblast fates (Reim et al.,2005). In vertebrate model organisms, Tbx20 is also essential for heart development (Cai et al.,2005; Stennard et al.,2005; Takeuchi et al.,2005) and mutations in human Tbx20 are associated with familial congenital heart defects (Kirk et al.,2007; Posch et al.,2010).

In the developing heart, mid is first detected in the dorsal mesoderm of stage 12 embryos with H15 appearing in the same cells in stage 13. The mid-expressing cells are selected from within clusters of cells expressing the transcription factor genes tin, pannier (pnr), which encodes a GATA factor, and Dorsocross 1, 2, and 3 (collectively referred to as Doc), which encode the Drosophila Tbx6 homologs (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005; Zaffran et al.,2006). Lineage studies have shown that two cells in each hemisegment divide to give 6 cardioblasts (Ward and Skeath,2000; Han and Bodmer,2003). One cardioblast progenitor divides twice to give rise to four mid/H15-expressing cardioblasts that maintain tin expression (Tin-CBs). The other cardioblast progenitor divides twice to give rise to two accessory pericardial cells and two cardioblasts, the latter of which lose tin but maintain Doc expression and express the transcription factor seven-up (Svp-CBs). There are different interpretations as to whether the earliest mid expression corresponds to the cardioblast progenitors (Miskolczi-McCallum et al.,2005) or cells arising later in the lineage (Reim et al.,2005). Nevertheless, it is clear that mid and H15 expression is restricted to the six cardioblasts in each hemisegment that fuse to form a single row of cells on each side of the embryo. The rows of cells migrate dorsally and fuse to form a tube that comprises the dorsal vessel or fly heart (Ward and Skeath,2000; Han and Bodmer,2003).

It is difficult to interpret whether the effect of Tin on mid and H15 expression is direct or indirect based solely on genetic analysis. Both mid and H15 expression are lost in tin mutant embryos and ectopic expression of tin results in ectopic midline-expressing cells in the cardiac mesoderm (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). Thus, mid and H15 may be directly activated by Tin. However, this interpretation is complicated by two results. First, no heart cells form in tin mutant embryos; therefore, the lack of mid/H15 expression could simply result from the failure of mid-expressing cells to be specified (Azpiazu and Frasch,1993; Bodmer,1993). In tin gain-of-function experiments, the small increase in mid-expressing cells could result from increased cardioblast specification (Qian et al.,2005; Reim et al.,2005). Second, in tin mutant embryos where early tin expression (up to stage 12) is restored with a transgene that allows the specification of cardiac tissues, mid expression is maintained in all six cardioblasts (Zaffran et al.,2006). Thus, tin may be required for the specification of cardioblasts and may not be directly involved in mid expression.

To better define the regulatory relationship between Tin and mid/H15, we searched for cis-regulatory elements (CREs) in the mid/H15 locus by identifying clusters of Tin consensus binding sequences near the mid coding sequences. We present here the characterization of a CRE located close to the mid transcription start site that responds genetically to tin manipulation in vivo and is activated by tin in transient transfection assays. Furthermore, Tin binds to the predicted sites in vitro and these sites are necessary for transgene expression in vivo. Based on these results, we conclude that Tin is a direct activator of mid expression in dorsal vessel development.

RESULTS

Identification of a midline Cardiac CRE

Given the predominant role of Tin in the regulation of cardiac gene expression, we set out to determine if Tin acts directly on mid. We searched for CREs driving mid expression in the developing heart. The mid and H15 locus is large, spanning ∼153 kb (Fig. 1A) from CG14070 to nompC, the two adjacent genes. To narrow our search for heart CREs, we took advantage of previous results, which show that clusters of Tin consensus binding sites (TYAAGTG) are often associated with heart CRE activity (Gajewski et al.,1997,1998; Cripps et al.,1999; Han and Olson,2005; Hendren et al.,2007; Ryan et al.,2007). Using the Genome Enhancer program (www.opengenomics.org; Markstein et al.,2002), we searched for clusters with at least two Tin consensus sites within a 500-bp window. This search identified 11 clusters, 10 with 2 consensus sites and 1 cluster with three sites. The orientation and spacing of the TBEs (Tin binding elements) within each cluster varied from 16 bp to 442 bp and is summarized in Supp. Table S1 (which is available online).

Figure 1.

Identification of cardiac cis-regulatory elements (CREs) in the H15 mid locus. A: Diagram of the H15 mid locus showing the locations of the 11 clusters of at least two TYAAGTG sites in 500 bp identified with Genome Enhancer software. Clusters with heart expression are indicated with solid red rectangles, clusters with no heart expression (open red rectangles) and clusters indicated with black rectangles were not tested. The approximate location of the genomic fragments tested for heart reporter activity are indicated with blue rectangles and the location of the TYAAGTG sites within the fragments with heart reporter activity are indicated with red ovals. B: Expression of mid mRNA in a stage 17 dorsal vessel. C:mid5.7 reporter expression with partial heart expression and ectopic lymph gland expression (arrow). D:mid4.3 reporter expression in a stage 17 embryo. Arrows indicate occasional gaps in CB expression. E:mid1.7 reporter expression in pairs of four CBs in each segment of a stage 17 embryo except for the most anterior two segments, where it is expressed in all six CB pairs. F:mid2.6 reporter expression in a stage 17 embryo with no heart expression. G:mid1.0 reporter expression in a stage 17 embryo is expressed in a pattern similar to mid1.7 (E). All embryos are oriented with anterior to the left. The genomic coordinates for the constructs are listed in Supp. Table S2.

We focused on clusters close to the mid coding region because of its predominant role in heart development compared with its homolog H15. The expression of mid is essential for heart development while H15 is dispensable, although mutating both genes has a stronger effect than loss of mid alone. mid is also activated earlier and at higher levels in the heart compared with H15 (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). We analyzed eight TBE clusters (clusters 4 through 11, Fig. 1) located 5′ and 3′ of the mid gene. Fragments containing the clusters and ranging in size from 4 to 7 kb were cloned into the Pelican lacZ reporter plasmid (Barolo et al.,2000), and several transformants of each line were analyzed and compared with the wild-type expression pattern of mid (Fig. 1B). We found that two of the constructs had heart CRE activity. Cluster 5 was contained in the mid5.7 construct and is located 3′ of the gene but only gave partial staining in the anterior portion of the heart as well as some ectopic expression in lymph gland (Fig. 1C). Cluster 4, contained in fragment mid4.3, is located 5′ of the mid gene, contains 3 Tin consensus sites within 334 bp and has strong expression in the heart that is restricted to cardioblasts in the same manner as mid mRNA (Fig. 1D).

We chose to focus on cluster 4 because it had complete heart expression and it was closest to the transcription start site of mid. The mid4.3 construct extends 4,377 bp upstream of mid and mimics mid heart expression by driving lacZ in all six cardioblasts in each hemisegment of the fully formed heart. Other staining was detected in the gonad mesoderm, a site of endogenous mid mRNA expression (Miskolczi-McCallum et al.,2005), and ectopically in cells dispersed throughout the embryo where we do not detect mid mRNA expression (not shown). We subdivided the 4.3 kb fragment into two fragments (2.6 kb and 1.7 kb) to find smaller Tin responsive elements. The proximal 1.7 kb fragment expressed lacZ in four out of six cardioblast pairs in each segment of the dorsal vessel, except for the two most anterior segments where it is expressed in all six cardioblast pairs (Fig. 1E). We confirmed that this expression corresponds to tin-expressing cells by comparing mid1.7 expression to svp-lacZ, a marker of Tin-negative cardioblasts (not shown). We found no cardiac CRE activity in a distal 2.6 kb fragment. This fragment also exhibits considerable ectopic expression (Fig. 1F). The inability of the distal 2.6 kb to drive lacZ expression in the heart suggests that it contains elements that are necessary but not sufficient for expression in the Svp-CBs. We have not identified an element that is sufficient to drive expression in the Svp-CBs and thus we focused our analysis to the 4.3 kb and the 1.7 kb fragments.

We compared the mid4.3 expression pattern with mid mRNA using fluorescent in situ hybridization and found that mid and lacZ expression are largely coincident. At stage 12, when mid expression is initiated, virtually all mid-expressing cells also express lacZ (Fig. 2A). Occasional mid-expressing cells do not co-express lacZ, but we interpret this as being due to the embryo-to-embryo variability in expression of the transgene that we observe at all stages of expression. Transgene expression is maintained during stages 12 through 17 during which period mid-expressing cells arise, divide, and give rise to the six cardioblasts in each hemisegment (Fig. 2B, data not shown). This result suggests that the mid4.3 CRE contains the elements required to initiate and maintain mid expression at the correct developmental stage. Comparison of mid1.7 to mid mRNA at stage 12 gives similar results in that most mid-expressing cells co-express lacZ (Fig. 2C). By stage 14, some cells begin to lose lacZ (Fig. 2D). From stage 15, a regular pattern of four lacZ-positive and two lacZ-negative cells is observed (Fig. 2E). As with the mid4.3 construct, there is variability in mid1.7 lacZ mRNA expression between embryos. However, we conclude that the mid1.7 fragment is sufficient to activate mid expression in all or most mid-expressing cells but is only able to maintain that expression in the Tin-CBs at later stages.

Figure 2.

Comparison of mid mRNA and mid-reporter lacZ mRNA. Comparison of mid mRNA (green) and lacZ mRNA (red) driven by the mid4.3 and mid1.7 reporter constructs. A,B: Stage 12 (A) and stage 14 (B) embryos showing strong correspondence between mid and mid4.3 lacZ. C: Stage 12 embryo showing strong correspondence between mid and mid1.7 lacZ. D,E: Stage 14 (D) and stage 15 (E) embryos showing regular gaps in mid1.7 lacZ corresponding to Svp-CBs compared with mid mRNA. Embryos are viewed laterally in panels A–D and dorsally in panel E.

The midline Cardiac CRE Responds to tin, pnr, and mid In Vivo

We manipulated the expression of tin, pnr, and mid and found that both the mid4.3 and mid1.7 transgenes responded in the same manner as mid mRNA. We present the data for mid1.7 here because our biochemical analysis (see below) focuses on this transgene and its derivatives. The results for mid4.3 are presented in Supp. Fig. S1. As is the case for mid mRNA (Miskolczi-McCallum et al.,2005), we found no mid1.7 reporter expression in the hearts of tin homozygous mutants (Fig. 3B). This result was not surprising because of the absence of all dorsal mesoderm derivatives in tin mutants. Conversely, ectopic expression of UAS-tin throughout the embryonic mesoderm of mid1.7 embryos using the How-Gal4 driver led to a small number of cells that ectopically express lacZ (Fig. 3C). This is similar to the effects of ectopic tin on endogenous mid (Qian et al.,2005), indicating that mid1.7 faithfully reports the effects of manipulating tin expression (Fig. 3C,D). The limited effects of ectopic tin may be due to the broad endogenous tin expression present in stage 12 when mid expression is first activated in the mesoderm. The endogenous tin domain may encompass most of the cells competent to express mid, thus How-Gal4 may not be able to drive sufficient ectopic tin to have a large effect on expression of the mid transgenes.

Figure 3.

Genetic regulation of mid heart transgenes. A,B:mid1.7 heart expression (A) is completely absent in tinec40 null embryos (B). C: UAS-tin expression driven by How-GAL4 induces ectopic expression of mid1.7. Arrows indicate examples of ectopic expression. D: UAS-pnr and UAS-tin expression driven by How-GAL4 induces ectopic expression of mid1.7. Arrows indicate examples of ectopic expression. E:mid1.7 cardioblast expression is absent in pnrvx6 embryos. F: UAS-mid expression driven by How-GAL4 induces ectopic expression of mid1.7. Arrows indicate examples of ectopic expression. G: UAS-tin, UAS-pnr, and UAS-mid expression driven by How-GAL4 induces ectopic expression of mid1.7. H:mid1.7 heart expression is maintained in H15x4 mid1a5 null embryos. All embryos are postdorsal closure (stage 17). The pnrvx6 embryo in panel E is arrested in dorsal closure due to the requirement for pnr function in morphogenesis. Panels A, C, D, F–H are dorsal views. Panel B is a lateral view and panel E is a dorsal lateral view.

The GATA factor pnr acts in parallel with tin to regulate tin responsive genes including mid. Null mutations in pnr greatly reduce the number of cardioblasts (Gajewski et al.,1999; Alvarez et al.,2003; Klinedinst and Bodmer,2003) causing mid expression to be almost completely absent and ectopic pnr has effects similar to ectopic tin expression (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). Similar to the effects on mid, mid1.7 is expressed in a small number of ectopic cells in response to pnr alone (not shown) and co-expression of tin and pnr had an increased effect on mid1.7 compared with tin or pnr alone (Fig. 3D). In embryos homozygous for the strong loss-of-function mutant pnrvx6, mid1.7 expression is greatly reduced or missing altogether (Fig. 3E). We conclude that mid transgenes respond to pnr in a manner similar to endogenous mid.

Previously, we (and others) have demonstrated that ectopic mid expression can induce the expression of an enhancer trap inserted in the H15 mid locus, suggesting that mid may regulate its own transcription (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). We found that ectopic mid can induce ectopic mid1.7 expression (Fig. 3F). Furthermore, this effect was greatly enhanced when mid was expressed ectopically in conjunction with tin and pnr (Fig. 3G). However, despite the striking effects of mid gain-of-function, we found mid 1.7 was not lost in mid/H15 null mutants (Fig. 3H) indicating that mid and H15 are not essential for the expression of mid cardiac CREs.

The mid Cardiac CRE Responds to tin in Transient Expression Assays

Our results show that the mid4.3 construct recapitulates mid expression during dorsal vessel development and that both mid4.3 and mid1.7 respond in the same manner as mid mRNA to manipulations of tin, pnr, and mid. However, the effects on mid mRNA and mid transgene expression are quite small despite the ectopic expression of the cardiogenic transcription factors throughout the mesoderm raising the possibility that the effects of ectopic expression could simply be due to an increase in the numbers of cardioblasts. The loss-of-function effects of tin and pnr could also be explained in terms of a decrease of cardioblasts.

To test whether there are direct effects on mid transcription, we examined the ability of tin, pnr, and mid expressed in Drosophila S2 cells to activate mid CRE-luciferase reporter genes. We transfected S2 cells with a luciferase reporter under the control of the mid 4.3 kb CRE with or without tin driven by the actin promoter. We found that expressing tin resulted in a ∼ three-fold increase in the levels of mid4.3 luciferase compared with controls transfected with empty vector (not shown). Although this is a weak effect, a mid1.7 luciferase construct is activated more than 40-fold by Tin compared with control transfected cells (Fig. 4A) and we find similar effects with the shorter mid1.0 construct (data not shown). In contrast to the strong effects observed with tin, expression of either mid or pnr had only a slight effect on mid1.7 luciferase expression in S2 cells. These results suggest that only Tin is able to activate the mid1.7 CRE on its own.

Figure 4.

A: Analysis of mid1.7 expression in transient transfection assays. mid1.7 luciferase construct is activated ∼40-fold by Tin compared with empty vector controls. Mid and Pnr alone had weak effects on mid1.7 luciferase. Tin and Mid, or Tin and Pnr increased luciferase activity by ∼40% or ∼300%, respectively, when compared with Tin alone. Tin, Mid, and Pnr expressed together showed a strong effect on mid1.7 luciferase but less than Tin and Pnr. Duplicate transfections were performed in each experiment and experiments were replicated three times. Luciferase activity was normalized as described in the methods. The data are the means of each normalized luciferase assay and standard deviations are indicated. B: Tin, Pnr, and Mid bind in pull down assays. Glutathione-Sepharose 4B beads coupled with GST, GST-Tin, GST-Mid, or GST-Mid-Tbox proteins were incubated with 35S-methionine labeled Tin, Mid, and Pnr. Each protein was pulled down by GST fused proteins, but not by GST controls. GST-Tin bound to Pnr and Mid. GST-mid bound Tin and Pnr. GST-Tbox showed binding to Tin, Pnr, and Mid.

We also expressed the factors in combination to compare activation of the mid1.7 transgene in vitro with the effects we observed in the embryo. We found that expressing mid and pnr together caused only a weak effect. However, expressing mid and tin together resulted in a nearly 40% increase in expression compared with expressing tin alone, suggesting that there may be synergy between tin and mid. The effects of expressing both tin and pnr were more pronounced, resulting in a ∼120-fold increase compared to controls and a 3-fold enhancement over the 40-fold increase caused by tin alone. Finally, the effects of expressing mid, tin, and pnr together resulted in an intermediate effect between the effects of tin and pnr together and the effects of mid and tin together, suggesting that mid may have an inhibitory effect on tin and pnr. Thus, the severity of the effects of tin and pnr with or without mid in in vitro assays were opposite those observed in the embryo, where mid had a much greater effect than tin and pnr alone and greatly enhanced the effects of tin and pnr, instead of inhibiting them as occurred in vitro. This suggests that the effects of mid on mid1.7 expression in the embryo are likely to be indirect.

In general, the effects of mid, tin, and pnr on mid1.7 reporter expression are similar to those observed for the vertebrate homologs Tbx20, Nkx2.5, and GATA factors, and suggest that the transcription factors may somehow act cooperatively in the regulation of cardiac gene expression (Stennard et al.,2003). Indeed, in addition to the synergistic effects on reporter gene expression, the three vertebrate factors are known to bind one another in vitro suggesting a possible mechanism for the synergy in transcription assays. We find that this is also the case for mid, tin and pnr. GST-tagged Tin is able to bind in vitro translated Mid and Pnr. Similarly, GST-Mid is able to bind in vitro translated Tin and Pnr protein (Fig. 4B). We have not been able to efficiently express tagged versions of the Pnr protein and so we have not performed the reciprocal experiments.

The mid Cardiac CRE Depends on Tin Binding Sites In Vivo

The preceding data strongly suggest that mid is a direct target of the Tin transcription factor in vitro, possibly acting in conjunction with other factors including Pnr. For this analysis, we focused on mid1.0, a 958-bp subfragment of mid1.7 that is also expressed in the Tin-CBs (Fig. 1G). We searched for consensus binding sequences in the mid1.0 fragment and found that, in addition to the three Tin consensus sites used in the initial identification of the CRE, there are also two GATA factor consensus binding sites (WGATAR) in the fragment (Fig. 5A). To test the requirement for the Tin and GATA sites for CRE activity in vivo, we performed a series of deletions and site-specific mutations of the mid1.0 CRE. Deletion of ∼300 bp from the 3′ end to generate a fragment containing the three Tin sites and a single GATA site (Fig. 5B) or ∼700 bp to generate a fragment that lacked all GATA sites but retained two (Fig. 5C) Tin sites (Tin1 and Tin2) did not alter the dorsal vessel pattern of 4 pairs of cardioblasts staining. A further deletion that contained only the 5′ Tin site had variable expression with the majority of cells not expressing lacZ (Fig. 5D). A 5′-deletion series also highlighted the necessity of Tin1 and Tin2 for cardiac expression. Deletion of Tin1 (Fig. 5E) caused a reduction in the levels of CB staining that was further enhanced when Tin2 was deleted (Fig. 5F). A 180-bp fragment bounded by Tin1 and Tin2 also had strong expression in the four Tin-CBs (Fig. 5I). This deletion series suggests that the Tin sites are critical for mid CB expression while the GATA sites are not.

Figure 5.

Deletion and mutation analysis of mid1.0. A:mid1.0 retains strong expression in Tin-CBs. Red ovals indicate Tin sites and green ovals indicate potential Pnr binding sites with the WGATAR GATA factor consensus binding-site. B,C:mid3T1P is a ∼ 300-bp 3′ deletion of mid1.0 (B) and mid2T is a ∼ 700-bp 3′ deletion (C) and both retain strong expression in Tin-CBs. D:mid1T, a 3′ deletion retaining only a single Tin site has weak, variable expression. E:mid2T2P, a 5′ deletion removing the Tin1 site, has weak variable expression. F:mid1T1P, a 5′ deletion retains a single Tin site and has very weak expression. G: Mutation of the three Tin binding sites in mid1.0 (mutated sites are represented by a white oval) greatly reduces expression. Similar results are seen in a construct where only Tin1 and Tin2 are mutated (not shown). H: Mutation of both Tin sites in mid2T** results in a very strong reduction in expression. I,J: The mid180 construct expresses in the Tin-CBs (I) and mutation of the Tin1 and Tin2 sites blocks all heart expression (J). The genomic coordinates for the constructs are listed in Supplementary Table 2.

Tin Directs Mid Expression Through Two Predicted Binding Sites

Our 5′ and 3′ deletions suggested that fragments containing Tin1 and Tin2 were able to drive robust cardiac expression in the Tin-CBs. Mutating all three TBEs in the mid1.0 fragment caused a strong reduction in cardiac expression (Fig. 5G). Similar results were obtained when only the Tin1 and Tin2 sites were mutated in the mid1.0 fragment (not shown) or in the 300-bp mid2T fragment (Fig. 5H), indicating that Tin binding was critical for strong expression of the transgene. Mutating both sites in the 180-bp fragment blocked all CB expression (Fig. 5J). We tested both of these sites for the ability to bind Tin protein expressed in a rabbit reticulocyte lysate. Oligonucleotides representing the two sites in their genomic context were found to bind specifically to in vitro translated Tin, with the binding able to be competed by a specific oligonucleotide but not by a nonspecific competitor (Fig. 6). These results indicate that Tin can directly regulate Mid by means of the Tin1 and Tin2 sites.

Figure 6.

The Tin1 and Tin2 sites bind Tin specifically. An electromobility shift assay shows that both the Tin1 and Tin2 sequences are able to bind Tin protein specifically. The Tin-DNA complexes were formed in the presence of IVT Tin protein and Tin1 and Tin2 sites, respectively, but were not formed in the IVT blank (unprogrammed). The Tin-DNA complexes were competed by addition of 100-fold molar excess unlabeled specific competitor, but not by addition of nonspecific competitor.

DISCUSSION

We have identified and characterized CREs in the H15 mid locus of Drosophila melanogaster that direct reporter gene expression in the cardioblasts (CBs) of the developing dorsal vessel. In the embryo, H15 and mid are expressed in the same pattern in several tissues including the ectodermal segments, the central nervous system, and dorsal vessel (Buescher et al.,2004; Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). Previous studies had demonstrated that clusters of TYAAGTG Tin binding sites are often predictive of cardiac CREs (Gajewski et al.,1997,1998; Cripps et al.,1999; Han and Olson,2005; Hendren et al.,2007; Ryan et al.,2007) and we used this approach to identify the mid4.3 CRE from within the 153 kb H15 mid locus. Direct comparison with mid mRNA showed that the 4.3 kb CRE is able to re-capitulate all mid expression in the cardiac mesoderm, driving reporter expression only in mid expressing cells. Like mid mRNA in mesoderm, mid4.3 is also induced at stage 12 and it responds to genetic manipulation of tin, mid, and pnr. Our analysis of CREs in H15 mid is not exhaustive, and there are likely other sequences that contribute to H15 mid cardiac expression, for example the 3′ mid5.7 fragment that has partial expression in the heart. Nevertheless, our results suggest that mid4.3 is an important element in H15 mid regulation in the heart.

H15 and mid are expressed in all CBs, although the four Tin-CBs derive from one progenitor and the two Svp-CBs derive from another (Ward and Skeath,2000; Han and Bodmer,2003). Our results show that mid expression is controlled by different CREs in the two lineages. Subfragments of mid4.3, ranging in size from 1.7 kb (mid1.7) to 180 bp (mid180) and which contain the Tin1 and Tin2 sites, are expressed exclusively in the Tin-CBs. Similar Tin-CB specific CREs have been reported in other genes expressed in all cardioblasts including Dmef2 and Toll (Gajewski et al.,1997; Wang et al.,2005), in genes with dorsal vessel expression restricted to the Tin-CBs such as Sulfonyl-urea-receptor (Sur; Akasaka et al.,2006; Hendren et al.,2007) and in tin itself (Lo and Frasch,2001). The distal 2.6-kb fragment of mid4.3 is not sufficient to drive any heart expression, suggesting that mid expression in the Svp-CB lineage requires sequences from both the proximal and distal halves of the mid4.3 fragment. Indeed, our results suggest that the mid1.7 CRE is initially expressed in all mid-positive cells and then is lost or repressed in the Svp-CBs. Thus, sequences in the proximal 1.7-kb fragment may be required to initiate the expression in both cardioblast lineages but are only able to maintain Tin-CB expression, whereas sequences in the distal 2.6-kb fragment are required to maintain mid expression in Svp-CBs following activation. This is also consistent with the observation that mid expression is maintained in tin null embryos that have early tin expression rescued by a transgene (Zaffran et al.,2006). In these embryos, we presume that mid expression is activated by tin but maintained by tin-independent elements, like those that we propose are found in the mid2.6 fragment. There is precedent for this sort of regulation as Tin is required to activate the expression of seven-up in the Svp-CBs, but is repressed in those cells following svp activation and thus is not required to maintain svp expression (Ryan et al.,2007).

Our results implicate Tin as a direct activator of mid. Previous work has shown that mid expression is lost in tin mutants and expressed ectopically in response to tin overexpression in the cardiac mesoderm (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). However, it has been impossible to distinguish whether these changes in mid expression result from the direct action of Tin on mid or if they were indirect effects due to tin-dependent changes in cardioblast specification. The strong activation of the mid1.7 fragment by tin in transient transfection assays is consistent with the direct activation of mid by Tin. The Tin1 and Tin2 sites also appear to be functional as they bind Tin specifically in electrophoretic mobility shift assays (EMSA) and deletion or mutation of the sites causes a great reduction in reporter gene expression in Tin-CBs. Our results are supported by a genome wide chromatin immunoprecipitation analysis of Tin binding in Drosophila embryos that shows Tin is enriched in a genomic region containing the mid1.7 fragment (Zinzen et al.,2009). It is also of interest that of the eight Tin binding site clusters we examined in this study, Tin was located by ChIP only to clusters 4 and 5, and only these clusters had cardiac CRE activity. There were genomic sites enriched for Tin binding, notably upstream of H15, that did not correspond to the clusters of TYAAGTG sites but which do contain single Tin consensus binding sites. Overall, our results suggest that clusters of TYAAGTG sites are not strictly predictive of cardiac CRE activity.

Tin1 and Tin2 are critical for mid transgene expression in Tin-CBs, but it is obvious that other CREs and other trans-acting factors are required. For example, mutating Tin1 and Tin2 in several fragments greatly diminished, but did not completely abolish CB expression. One possibility is that there are noncanonical Tin sites in the fragment, but it is also likely that sites for other cardiogenic transcription factors in the fragment contribute to mid expression. Several previously characterized cardiac CREs have been shown to require the additive effects of multiple transcriptional inputs. For example, the even-skipped pericardial cell CRE requires input from several signaling pathways in addition to Tin and mutations of sets of binding sites diminish but do not completely reduce reporter expression (Halfon et al.,2000).

In addition to tin, we also assessed the role of mid and pnr in the regulation of mid expression. Previous genetic results showed that gain-of-function for pnr induced ectopic mid expression in embryos (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005), and we found that this was also the case for mid4.3 and mid1.7 CRE expression. Pnr is a strong candidate to regulate mid, although we cannot determine from our results that it is either an essential or direct regulator of mid expression. The mid1.7-kb fragment contains several consensus GATA-binding sites and while pnr expression alone does not have a strong effect on mid1.7 luciferase activity, it does show a ∼three-fold increase in luciferase activity when co-expressed with tin compared with tin alone. Thus, Tin and Pnr may have a synergistic effect on the activation of mid. However, deletions that remove the GATA consensus sites in derivatives of mid1.7 have little effect on the expression pattern of mid transgenes. Furthermore, while strong pannier mutants block almost all CB development (Gajewski et al.,1999; Alvarez et al.,2003; Klinedinst and Bodmer,2003), the remaining CBs express mid mRNA (Miskolczi-McCallum et al.,2005) and the mid1.7 transgene (Fig. 3E). This suggests that pnr is not essential for mid, although we cannot rule out an essential role for pnr in modulating the levels, timing or maintenance of mid.

In the case of mid, ectopic expression of mid alone in the embryonic mesoderm has strong effects on mid1.7 expression and ectopic mid greatly enhances the effects of ectopic tin and pnr. However, we argue that this is likely to be an indirect effect due to increased CB specification, possibly resulting from the ability of mid to regulate endogenous tin expression (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005). Indeed, loss of mid and H15 has little effect on mid transgene expression. In transient transfection assays, mid has a weak effect on its own and only a mild synergy in combination with tin, when compared with the effects of tin and pnr. The effects of mid could be indirect and related to the ability of Mid and Tin to bind, allowing Mid to enhance Tin transactivation without interacting directly on the mid CRE. Indeed, we find no sites similar to a Mid binding site determined by in vitro site selection (N.N. and W.J.B., unpublished data). In contrast, it has an inhibitory effect on the synergy between tin and pnr in mid1.7 activation. Since we have shown that Mid can bind both Tin and Pnr in vitro, the expression of Mid may block an interaction between Tin and Pnr responsible for the synergy.

In summary, we have presented several lines of evidence that, in addition to acting upstream of mid in the genetic pathway for cardioblast development, Tin is a direct activator of mid transcription. Because tin is expressed outside the mid domain, for example in the pericardial cells of the dorsal vessel, and mid expression is maintained in the Svp-CBs that do not maintain tin, there are certainly more factors required to control mid CB transcription. One candidate that may restrict the expression of mid is the Islet-1 homolog tail-up (tup), which is expressed in all CBs (Tao et al.,2007). Hypomorphic tup1 embryos have reduced expression of the H15-lacZ enhancer-trap and thus Tup is a potential cofactor for Tin in mid activation. Similarly, both svp and Doc are potential upstream regulators that may act to maintain mid expression in the Svp-CBs (Reim and Frasch,2005). Further analysis of the mid CB CREs will be necessary to test these possibilities.

EXPERIMENTAL PROCEDURES

Drosophila Culture and Strains

Flies were grown on yeast-agar-cornmeal-sucrose medium at 25°C. The following fly stocks were used: yw for making transgenic flies, HowGal4 (Brand and Perrimon,1993), UAS-tin (Ranganayakulu et al.,1998), UAS-pnr (Haenlin et al.,1997), UAS-mid (Buescher et al.,2004), UAS-tin/UAS-pnr, UAS-tin/UAS-pnr; UAS-mid for ectopic expression, tinec40 (Bodmer,1993), pnrvx6 (Ramain et al.,1993), H15X4 mid1a5 (Svendsen et al.,2009) for loss-of-function crosses.

Cloning Mid Cardiac CRE Fused lacZ and Luciferase Constructs

Various mid locus fragments were amplified by polymerase chain reaction (PCR) and cloned into the lacZ enhancer tester pHpelican (Barolo et al.,2000). The mid4.3, mid1.7, and mid1.0 fragments were also cloned into the p2TLuc vector for S2 cell luciferase reporter assays (Ryu and Arnosti,2003).

Transgenic Flies, Immunohistochemistry, and In Situ Hybridization

P-element–mediated germ line transformation followed standard protocol using the Delta2-3 helper plasmid (Robertson et al.,1988) and three to five independent transgenic lines were analyzed for each construct. The embryos from transgenic flies were collected, fixed, and labeled with mouse anti-β-galactosidase (1:500, Promega) as a primary antibody. The primary antibody was then detected with goat anti-mouse IgG, and stained with diaminobenzidine (DAB) using Vectastain ABC kit (Vector Laboratories Inc.). Fluorescent in situ hybridizations were carried out using a protocol modified from (Lecuyer et al.,2008).

Transient Transfection and Luciferase Assays

Drosophila S2 cells were plated in 12-well plates at a density of 1 × 106 cells per well and transient transfection was performed as described previously (Ryu and Arnosti,2003). Briefly, each transfection cocktail contained 100 ng of mid CRE-Luc, 100 ng of Actin-expressor, and 20 ng of pAc/lacZ internal control for each well. pAx DNA was used to equalize the total amount of DNA. After 40 hr, the cells were harvested and lysed with reporter assay lysis buffer (Promega). Luciferase assays were performed following the manufacturer's instructions (Promega) and analyzed with a luminometer. Each experiment was performed in duplicate at least three times. Luciferase activity was corrected for the transfection efficiency using β-gal activity as an internal control and to a normalized Bradford protein assay (Bio-Rad).

GST Pull Down Assay

We cloned GST-Tin, GST-Mid, and GST-Tbox in frame and purified the GST-fused proteins following the original protocol (GE Health Care). The binding partner proteins, pET-Tin, pET-Mid, and pET-Pnr were cloned into pET21a vector (Novagen) and confirmed by sequencing. We used GST or GST-Tin or GST-Mid or GST-Tbox fusion proteins immobilized on glutathione-Sepharose 4B beads and in vitro translated Tin, Mid, and Pnr labeled with 35S-methionine with the TNT Quick Coupled transcription/translation Systems (IVT reaction; Promega). The proteins were mixed in binding buffer (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, 2.5 mg/ml bovine serum albumin (BSA), 10 μM ZnSO4, 0.5% NP40) for 1 hr at 4°C. We separated the GST fused protein bound Tin, Mid, and Pnr by 4–15% gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel and visualized them by autoradiography.

Electrophoretic Mobility Shift Assays

We made the Tin protein using IVT (Promega) following the manufacturer's protocol. Complementary oligonucleotides having Tin consensus sites were end-labeled with 32P at the 5′ sites. The oligonucleotides are described below (wild-type, bold; mutant, underlined).

Tin1 top: 5′CTTCAGGGCGGCTCA AGTGCTCGAAAAACG3′; Tin1 bottom: 5′CGTTTTTCGAGCACTTGAG CCGCCCTGAAG3′; Tin1* top: 5′CTT CAGGGCGGCTGCCTGGCTCGAAAA ACG3′; Tin1* bottom: 5′CGTTTTTC GAGCGAGGCAGCCGCCCTGAAG3′; Tin2 top: 5′CAGGGGAATCTCAAG TGAGCAGAGTGAAAA3′; Tin2 bottom: 5′TTTTCACTCTGCTCACTTGA GATTCCCCTG3′; Tin2* top: 5′CAG GGGAATCTGTCGACAGCAGAGTGA AAA3′; Tin2* bottom; 5′TTTTCACTC TGCTGTCGACA GATTCCCCTG3′.

A total of 3 μl of IVT synthesized Tin and double-stranded 32P-end-labeled (10,000 cpm) oligonucleotides were mixed in binding buffer (15 mM Hepes [pH 7.6], 100 mM KCl, 0.1 mM ethylenediaminetetraacetic acid (EDTA), 1 mM dithiothreitol (DTT), 12% glycerol, 0.25 mg/ml sonicated salmon sperm DNA, 0.25 mg/ml BSA) and incubated for 15 min on ice. One hundred-fold molar excess unlabeled specific or nonspecific oligonucleotides were added before adding IVT expressed Tin. Unprogrammed IVT reactions were performed without Tin expressing DNA as a negative control. DNA–protein complexes were resolved on 6% nondenaturing polyacrylamide gels in 40 mM Tris/glycine (pH 8.3)/2 mM EDTA at 25°C (6–7 cm/V). Gels were analyzed with a PhosphoImager (Molecular Dynamics).

Acknowledgements

We thank Drs. Jeb Gaudet, Savraj Grewal, and Pia Svendsen for comments on the manuscript. Nekky Jamal, Sharayu Jangam, and William Emery provided technical assistance. This work was funded by the CIHR MOP-84444. N.N. was supported by the ACHRI/CIHR Training Program in Genetics Child Health and Development. W.J.B. was a Senior Scholar of the AHFMR.

Ancillary