The NK-type homeobox gene tinman and the MADS box gene D-mef2 encode transcription factors required for the development and differentiation of the Drosophila heart, and closely related genes regulate cardiogenesis in vertebrates. Genetic analyses indicate that tinman and D-mef2 act at early and late steps, respectively, in the cardiogenic lineage. However, it is unknown whether regulatory interactions exist between these developmental control genes. We show that D-mef2 expression in the developing Drosophila heart requires a novel upstream enhancer containing two Tinman binding sites, both of which are essential for enhancer function in cardiac muscle cells. Transcriptional activity of this cardiac enhancer is dependent on tinman function, and ectopic Tinman expression activates the enhancer outside the cardiac lineage. These results define the only known in vivo target for transcriptional activation by Tinman and demonstrate that D-mef2 lies directly downstream of tinman in the genetic cascade controlling heart formation in Drosophila.
The phenotypes of tinman and D-mef2 mutant embryos suggest that tinman acts at an early step in the cardiogenic pathway to establish cardiac cell identity, whereas D-mef2 acts later to control the expression of cardiac structural genes. However, the potential hierarchical relationship between these two genes has not been investigated previously. To define further the roles of MEF2 factors in heart development, we have initiated an analysis of the cis-acting sequences that control the expression of D-mef2 in the cardiogenic lineage. Here we describe a novel cardiac enhancer that directs D-mef2 expression in the developing heart and we show that this enhancer is a direct target for transcriptional activation by Tinman.
Tinman binding sites in the regulatory sequences flanking the D-mef2 gene
Sequences controlling D-mef2 gene expression in embryonic cardiac, somatic and visceral muscle cell lineages have been mapped previously to a 13 kb interval between −12.4 and +0.6 relative to the start of the D-mef2 transcription unit (Lilly et al., 1995; Schulz et al., 1996). The 5.4 kb region II DNA (between −8.8 and −3.4 kb) contained enhancer sequences for all three cell types and was the only DNA that directed lacZ reporter gene expression in the developing heart (Figure 1A). To delimit further the D-mef2 heart enhancer element, this DNA was sequenced and scanned for potential Tinman binding sites which might be involved in regulating cardial cell expression. Bacterially produced Tinman binds to the sequence T(C/T)AAGTG (Y.Kim, unpublished data) and two identical, convergently oriented CTCAAGTGG sequences that match the Tinman binding consensus were found between −5.8 and −5.5 kb upstream of D-mef2 in a region we refer to as IIA237 (Figure 1B).
To test whether Tinman can bind to these sites, a DNase I protection assay was performed using a glutathione S-transferase (GST)–Tinman fusion protein and IIA237 DNA labeled specifically on either strand. The fusion protein protected both sites on both strands of the IIA237 DNA (Figure 2). These sequences have, therefore, been designated Tin 1 and Tin 2 for Tinman binding sites 1 and 2.
IIA237 DNA functions as a D-mef2 heart enhancer regulated by the tinman gene
To test the IIA237 region for heart enhancer function, the DNA was cloned upstream of a minimal Hsp43 promoter driving the lacZ gene, and germline transformant lines harboring this construct were established. Heart precursor cells first appear at stage 11, and IIA237 transformant embryos expressed the β-galactosidase marker in these progenitors by stage 12 (Figure 3A). In stage 13 embryos, β-galactosidase was detected in heart precursors located along the length of the dorsal-most part of the mesoderm (Figure 3B) and, in late stage embryos, staining was observed in the inner cardial cells of the newly formed dorsal vessel (Figure 3C). Thus, the 237 bp DNA was sufficient for directing reporter gene expression in the cardiac cell lineage and contained a heart enhancer element. Embryos from a total of six IIA237–lacZ transformant lines were assayed, and five showed expression in heart precursor and cardial cells. However, β-galactosidase expression was consistently lower in these embryos compared with the levels observed in the 5.4 kb DNA II–lacZ transformants, suggesting that an additional sequence(s) in the region II DNA is required for the proper quantitative expression of D-mef2 in the heart.
To verify that the reporter gene was being expressed in progenitor cells of the heart, we generated a fly strain that was homozygous for the D-mef2–lacZ fusion gene in a tin346/TM3 genetic background. By intermating these flies, we could monitor the expression of the lacZ gene in tinman null embryos that fail to specify cardial cells. β-Galactosidase was not detected in dorsal regions of the mutant embryos, consistent with the conclusion that the 237 bp D-mef2 enhancer is expressed in heart precursor cells (Figure 4A and C). In comparison, the D-mef2–lacZ fusion gene was expressed in the forming and mature heart of tin346 heterozygous embryos (Figure 4B and D).
To address further the relationship of tinman gene function to D-mef2 enhancer expression, we introduced the D-mef2–lacZ reporter into the hsTin9A strain which contains an inducible tinman gene under the control of a heat shock promoter. Ectopic expression of Tinman in heat-shocked embryos resulted in an expanded activation of the enhancer and expression of the β-galactosidase marker, especially in cells of the anterior procephalic region and the dorsal mesoderm of the extended germ band (Figure 4E). Relatively little induced β-galactosidase expression was detected in ventrally positioned cells of the heat-shocked embryos. Overall, these results indicate that Tinman can activate the expression of the D-mef2 heart enhancer, but only certain domains of the germ band extended embryo are permissive for transcriptional activation by Tinman. As a control, non heat-shocked embryos showed the normal pattern of β-galactosidase expression in the heart precursor cells (Figure 4F).
Both Tinman binding sites are required for D-mef2 heart enhancer activity in transformant embryos
The functional importance of the Tinman binding sites in the D-mef2 cardiac enhancer was assessed through the generation of a series of constructs that contained deletions or nucleotide changes in the Tin 1 and Tin 2 sequences (Figure 5). Transformant lines carrying the mutant D-mef2–lacZ fusion genes were established and embryos were stained for β-galactosidase. In the IIA166 construct, both ends of the enhancer including the Tin 1 and Tin 2 sites were deleted. This DNA failed to express in cardial cells. Likewise, deletion of only the 27 bp 5′ region (IIA210) or the 44 bp 3′ region (IIA193) inactivated the enhancer. These results suggested that either both Tinman binding sites were required, or other elements located in the 5′ and 3′ sequences flanking the Tin sites must be present for the functional integrity of the D-mef2 heart enhancer. To discriminate between these possibilities, we created three variant enhancers containing mutations in each Tinman binding site alone or in combination. The mutations in these sites are known to abolish Tinman binding to DNA (Y.Kim, unpublished data). None of the three mutant enhancers were able to direct lacZ expression in heart precursor or cardial cells in transformant embryos, indicating that both Tinman binding sites must be present in the IIA237 DNA for the activity of the D-mef2 enhancer. Together, these results demonstrate that Tinman is a direct activator of D-mef2 transcription in the developing heart of Drosophila and that transcriptional activation by Tinman requires binding to two sites in the cardiac enhancer.
Expression patterns and functions of tinman and D-mef2
The results of this and previous studies (Lilly et al., 1995; Schulz et al., 1996) reveal that D-mef2 transcription in the early mesoderm and in developing muscle cell lineages is regulated by a complex set of independent control regions located upstream of the gene. In the early Drosophila embryo, the expression pattern of D-mef2 is very similar to that of tinman. Both genes are expressed initially throughout the mesoderm within the ventral furrow and in the germ band extended embryo. D-mef2 transcripts accumulate normally in these cells in tinman mutant embryos (Lilly et al., 1994; Nguyen et al., 1994) and tinman is expressed normally in D-mef2 mutant embryos (Lilly et al., 1995), suggesting that the two genes are activated independently in the early mesoderm. The most likely regulator of tinman and D-mef2 is the basic helix–loop–helix transcription factor Twist, as neither gene is expressed in twist mutant embryos (Bodmer et al., 1990; Lilly et al., 1994; Nguyen et al., 1994; Taylor et al., 1995) and ectopic expression of Twist expands the expression pattern of both genes (Taylor et al., 1995).
Later, as the mesoderm differentiates, tinman expression becomes regulated by Dpp and is restricted to dorsal mesodermal cells located adjacent to Dpp-expressing ectodermal cells (St Johnston and Gelbart, 1987; Bodmer, 1990; Azpiazu and Frasch, 1993; Frasch, 1995). The function of tinman in the dorsal mesoderm is to determine the formation of heart precursor cells and to activate the expression of bagpipe, another NK-type homeobox gene required for specification of visceral muscle precursor cells (Azpiazu and Frasch, 1993; Bodmer, 1993). During this refinement of tinman gene expression, D-MEF2 is expressed in cardiac, somatic and visceral muscle cell precursors and later is expressed in the differentiated heart, body wall, and gut musculature (Bour et al., 1995; Lilly et al., 1995). Despite the expression of D-mef2 throughout the course of mesoderm formation and differentiation, gene function is required only for the differentiation of heart, somatic and visceral muscle cells (Bour et al., 1995; Lilly et al., 1995; Ranganayakulu et al., 1995). Thus, genetic studies have demonstrated an early function for tinman, and a late function for D-mef2, in the development of specific derivatives of the mesoderm.
Tinman regulation of D-mef2 in the cardial cell lineage
D-MEF2 is expressed in all of the contractile cardial cells of the heart tube where one of its functions is to activate the transcription of the three myosin subunit genes (Bour et al., 1995; Lilly et al., 1995; Ranganayakulu et al., 1995). The IIA237 region upstream of D-mef2 functions as a transcriptional enhancer in most, but not all of these D-MEF2-positive cells as enhancer activity was not observed in short, segmentally repeated sections of the heart tube (Figures 3C and 4D). However, this result is consistent with the regulation of the IIA237 heart enhancer by Tinman, as tinman is expressed in this pattern, i.e. in most but not all of the cardial cells of the mature heart (Bodmer, 1993). Thus, D-mef2 requires Tinman and at least one other regulatory input for its expression in the complete complement of cardial cells.
The requirement for two Tinman binding sites for the activity of the D-mef2 heart enhancer suggests some manner of cooperativity between Tinman proteins bound to these sites, perhaps mediated by the interaction of Tinman molecules with each other. Other NK-type homeodomain proteins can also bind this consensus sequence. For example, the NK-2 protein, which is the product of the ventral nervous system defective (vnd) locus of Drosophila (Jimenez et al., 1995), has a consensus binding site of T(T/C)AAGT(G/A)G (Tsao et al., 1994) which matches the Tinman sites in the D-mef2 cardiac enhancer. vnd is expressed in cells of the developing nervous system and midgut of the embryo (Jimenez et al., 1995; Mellerick and Nirenberg, 1995) and it is conceivable that NK-2 could bind to the IIA237 DNA in these cells. However, since the IIA237 enhancer is not expressed in CNS or midgut cells, it seems likely that additional factors must work with the NK proteins to regulate target genes in the appropriate cell types. Thus, the selective activation of the D-mef2 enhancer by Tinman may require a co-activator protein that is expressed in the cardial cells. Because the IIA237 region contains all the information necessary for transcriptional activation by Tinman, it will be of particular interest to screen for other potential regulatory elements within this enhancer to identify the transcription factors that may cooperate with Tinman to control cardial cell gene expression.
Regulatory network controlling cardiac gene expression in Drosophila
Our analysis of D-mef2 expression and function in the heart, coupled with the findings of others, suggests a genetic hierarchy controlling gene expression during Drosophila cardiogenesis (Figure 6). Wg is required for the specification of cardiac progenitors (Wu et al., 1995), while Dpp produced in the ectoderm determines the expression domain of tinman in the dorsal mesoderm (Staehling-Hampton et al., 1994; Frasch, 1995). Tinman is a direct activator of D-mef2 in heart precursor cells and this regulation probably requires a Tinman co-activator protein, since Tinman expression in the dorsal mesoderm is much broader than D-mef2, and ectopic Tinman expression can activate the D-mef2 cardiac enhancer in some but not all regions of the embryo. Additional factor(s) are required for the expression of D-mef2 in the full complement of cardial cells because the Tinman-dependent enhancer is active in only a subset of cells of the dorsal vessel. Late-expressed contractile and structural protein genes can be divided into D-mef2-dependent and D-mef2-independent subgroups. The cardial cell transcription of the three myosin subunit genes requires D-mef2 function (Ranganayakulu et al., 1995) while the βTub60D and TmI genes are expressed normally in the heart of D-mef2 mutant embryos (Bour et al., 1995; Ranganayakulu et al., 1995; Lin et al., 1996). Whether these latter genes are activated directly by Tinman or require other as yet unidentified factor(s) for their cardial cell expression remains to be determined.
Conservation of genes and mechanisms controlling heart formation
DNAs were sequenced using the Sanger dideoxy method combined with automated fluorescence detection (Spurgeon et al., 1995). The 5.4 kb D-mef2 genomic DNA II (Schulz et al., 1996) was sequenced using a series of internal primers. Overlapping sequences were linked together and scanned for potential Tinman binding sites using the GCG DNA sequence analysis program. Wild-type and mutated D-mef2 DNAs to be tested for enhancer function in the CaSpeR-Hsp43-AUG-βgal (Chab) transformation vector were sequenced using a primer from the Hsp43 promoter region. Primers were obtained from Genosys Biotechnologies (Woodlands, TX).
DNase I protection assay
Footprinting assays were performed as described (Hoey et al., 1988). The IIA237 D-mef2 heart enhancer in pCRII (Invitrogen, San Diego, CA) was digested with either HindIII (minus-strand labeling) or XbaI (plus-strand labeling) and digested DNAs were treated with alkaline phosphatase. Dephosphorylated DNAs were redigested with either XbaI (minus-strand labeling) or HindIII (plus-strand labeling), eluted from a gel, and 32P-end labeled with T4 kinase. Binding reactions were then performed with a GST–NK-4 fusion protein and 12.5 fmol of labeled DNA in 10 μl of binding buffer containing 25 mM HEPES (pH 7.5), 3 mM MgCl2, 1 mM EDTA, 0.5% NP-40, 10% glycerol, 1 μg of poly[d(I–C)]. After incubation for 15 min at room temperature, CaCl2 (final 0.5 mM) and DNase I (0.05 U) were added to the binding reactions for DNase I digestion. Reactions were terminated after 1 min incubation and analyzed on an 8% denaturing polyacrylamide gel.
D-mef2–Chab enhancer test constructs
The D-mef2 IIA237, IIA166, IIA210 and IIA193 DNAs were generated by PCR amplification using the DNA II template, the terminal 20 nucleotide 5′ and 3′ primers, and dNTP solutions, LA PCR buffer and LA Taq enzyme provided in the TaKaRa LA PCR kit (PanVera, Madison, WI). PCR products were cloned into the pCRII vector using the TA cloning kit from Invitrogen (San Diego, CA). To generate the D-mef2 IIA237mTin1, IIA237mTin2 and IIA237mTin1/2 mutant DNAs, a PCR-based method was also used. For IIA237mTin1, two separate DNA fragments (A1 and A2) were amplified by PCR with specific primers (A1 DNA: SP6 and mT2R; A2 DNA: mT25 and T7) and IIA237 template DNA. The A1 and A2 DNAs were mixed, denatured and reannealed. DNAs were gap-filled and reamplified by PCR using 30 amplification cycles (94°C, 1 min; 55°C, 1 min; 72°C, 1 min) with SP6 and T7 primers. The amplified DNA fragments were then cloned into pCRII. For the IIA237mTin2 construct, the same method was employed using different primers (A3 DNA: SP6 and mT1R; A4 DNA: mT15 and T7). For the IIA237mTin1/2 DNA, an A5 DNA was generated by PCR using the mT15 and T7 primers and IIA237mTin1 template DNA. The A5 and A3 DNAs were then mixed, reamplified by PCR and subcloned. Primers used were as follows: SP6, 5′-ATTTAGGTGACACTATA-3′; mT1R, 5′-GCAGCTAAAGGATGCGCCgtCTacAGACCGGGCTCGCTATC-3′; mT15, 5′-GATAGCGAGCCCGGTCTgtAGacGGCGCATCCTTTAGCTGC-3′; mT2R, 5′-AATGGAGGCACCCAAGCgtCTacAGCCCCTTATCCGTGTC-3′; mT25, 5′-GACACGGATAAGGGGCTgtAGacGCTTGGGTGCCTCCATT-3′; T7, 5′-TAATACGACTCACTATAGGG-3′. Lower case letters correspond to the introduced base changes. The wild-type or mutant D-mef2 sequences were isolated from the pCRII vector and cloned into Chab.
Drosophila germline transformation
The D-mef2–Chab DNAs were stably introduced into the Drosophila genome by P element-mediated germline transformation (Rubin and Spradling, 1982). The y w67c23 strain was used for embryo injections and transformed flies were identified by w+ eye color selection. A minimum of four lines were established for each construct. The insertion of the IIA237 D-mef2–Chab DNA on chromosome II in the IIA237-9 line was determined using standard linkage analysis. IIA237-9 flies were crossed with tin346/TM3,ftz-lacZ flies to generate a IIA237-9/IIA237-9;tin346/TM3,ftz-lacZ stock. Upon intermating of these flies, the expression of the IIA237–lacZ DNA was assayed in tin346 heterozygous or homozygous embryos, with the former identified by the expression of the marker linked to the balancer chromosome. tin346 has been classified as a null allele of the tinman gene (Azpiazu and Frasch, 1993). IIA237-9 flies were also crossed to hsTin9A/TM3,ftz-lacZ flies to generate embryos that could be heat treated to induce ectopic expression of the Tinman protein. The heat shock regimen was as follows. Embryos were collected for 2 h at 25°C on yeasted grape-agar plates and then aged for 4 h at 25°C. Embryos were heat-shocked for 30 min by placing the plates in a water bath maintained at 37°C, followed by recovery for 30 min at 25°C. The heat shock/recovery cycle was repeated twice and the embryos were aged an additional 3 h at 25°C. Under these conditions, only the IIA237-9;hsTin9A embryos showed an altered distribution of the β-galactosidase expression marker.
Embryo β-galactosidase assays
Processing and immunostaining of whole mount embryos for β-galactosidase activity was as described (Patel et al., 1987) using the Vectastain Elite ABC Kit (Vector, Burlingame, CA). The anti-β-galactosidase primary antibody (Promega, Madison, WI) was used at a 1:1000 dilution while the horse anti-mouse–biotin conjugate secondary antibody (Vector) was used at a 3:1000 dilution. Stained embryos were mounted in 50% glycerol under cover slips for photography using a Nikon Optiphot photomicroscope and Kodak Ektachrome 64T film.
We are grateful to C.Chromey and M.-F.Lu for technical assistance, R.Bodmer and M.Frasch for providing the hsTin9A and tin346 fly stocks, and T.Mohun for communicating results prior to publication. DNA sequences were determined by the M.D.Anderson DNA Core Sequencing Facility supported by NIH grant CA16672. K.G. was supported by NIH postdoctoral training grant T32 HD 07324. This research was supported by grants to R.A.S. from the National Science Foundation and Muscular Dystrophy Association and a grant to R.A.S., E.N.O. and R.P.Harvey from the Human Frontier Science Program.