Cardioblast-intrinsic Tinman activity controls proper diversification and differentiation of myocardial cells in Drosophila

In a wide variety of organisms, both vertebrate and invertebrate,cardiogenesis requires the action of regulators that belong to the NK2, GATA and T-box families of transcription factors(Cripps and Olson, 2002; Buckingham et al., 2005; Reim and Frasch, 2005). In most cases, the expression of each of these cardiogenic factors spans a large developmental window of cardiogenesis and continues in the mature heart. Consequently, the specific function of a cardiogenic factor can change during the progression of cardiogenesis because, for example, it depends on the presence of additional factors and signals that modulate its activity. Although several examples of differential and stage-specific functions of cardiogenic factors have already been described(Pashmforoush et al., 2004; Zeisberg et al., 2005; Oka et al., 2006), our overall knowledge of these dynamic changes in regulatory activities is still very limited.

The apparent conservation of cardiogenic factors during evolution suggests strongly that some of their specific molecular functions within the regulatory network of cardiogenesis and heart differentiation may also be shared among different vertebrate and invertebrate species. Prime examples of conserved cardiogenic factors include the Drosophila NK homeodomain factor Tinman and the homologous Nkx2.5 proteins in vertebrates, which have been shown to play key roles during early stages of heart formation in the respective organisms (Harvey,1996). In Drosophila, the tinman (tin)gene is essential for the specification of all cardiac progenitors in the early dorsal mesoderm, and some of its target genes during this event, and the functional Tin-binding sites in their regulatory sequences have been defined by genetic and molecular analyses (Azpiazu and Frasch, 1993; Bodmer,1993; Xu et al.,1998; Halfon et al.,2000; Gajewski et al.,2001; Knirr and Frasch,2001; Han et al.,2002). The requirement for tin during the earliest steps of cardiogenesis is reflected in its early expression in the mesoderm, which is controlled by two distinct enhancer elements(Yin et al., 1997). Initially,during the invagination and spreading of the mesoderm, tin expression is activated by the bHLH protein Twist through an intronic enhancer in the entire trunk mesoderm (Bodmer et al.,1990; Yin et al.,1997). Thereafter, tin expression becomes dependent on a Dpp-responsive enhancer located downstream of tin, which leads to the restriction of tin expression to dorsal mesodermal cells that receive Dpp signals from the ectoderm (Xu et al.,1998). This corresponds to the stage when tin is required for the specification of myocardial and pericardial progenitors within the dorsally located cardiogenic mesoderm. In addition, tin is essential for the formation of other dorsal mesodermal derivatives during this stage,which include trunk visceral mesoderm precursors and dorsal somatic muscle progenitors. Even later, tin expression is further restricted to cardiac progenitors and this expression persists in myocardial and pericardial cells of the mature dorsal vessel. tin expression in cardioblasts depends on the Tbx20-related genes midline (mid)and H15, and is driven by another downstream enhancer(Yin et al., 1997; Reim et al., 2005). It is thought that tin plays a role in the differentiation of the heart progenitors during this late phase of expression, which is likely to include the direct activation of Mef2 and Hand, which regulate normal differentiation, as well as of cardiac differentiation genes such asβ 3-tubulin (Bour et al.,1995; Lilly et al.,1995; Gajewski et al.,1997; Kremser et al.,1999; Han and Olson,2005). However, genetic analysis of the significance of late tin expression in cardial cells has been hampered owing to the fact that tin mutant embryos lack a heart altogether because of the early requirement of tin for cardiogenesis. Furthermore, conditional alleles that would circumvent this problem have not been available.

The Drosophila heart is a relatively simple linear tube that, in spite of its overt simplicity, is highly structured and consists of a variety of myocardial and pericardial cell types(Rizki, 1978; Ward and Skeath, 2000). This organization is illustrated by the presence of different chamber-like regions with distinct functions (the heart in the posterior and the aorta in the anterior), and by the controlled posterior-to-anterior flow of the larval hemolymph, which enters through valvular openings (ostia) into the heart region. In addition to its broad anteroposterior (AP) organization, the dorsal vessel maintains a segmental organization, with units that, in most of the tube (i.e. from segments A2 to A7), consist of six pairs of cardioblasts in each segment. The segmental organization and AP polarity within each segment is revealed by the restricted expression of several transcription factor-encoding genes (reviewed by Lo and Frasch, 2003). In fact, the myocardial expression of tinwithin the dorsal vessel is restricted to the four posterior pairs of cardioblasts within each of these segments. Conversely, the anterior two pairs of cardioblasts in each segment are marked by the expression of seven-up (svp), which encodes an orphan nuclear receptor,and of the Dorsocross T-box genes (Doc1, Doc2 and Doc3, henceforth referred to as Doc) that are downstream of svp in these cells (Gajewski et al., 2000; Lo and Frasch,2001; Reim et al.,2003; Reim and Frasch,2005). In the heart region, this differential organization correlates with the formation of distinct subtypes of myocardial cells within each segment. Specifically, the two Svp/Doc-positive cardioblasts in each segment differentiate into ostial cells to form inflow valves, whereas the four Tin-positive cells form `working' cardiomyocytes of the heart(Molina and Cripps, 2001). Currently, it is not known whether the Tin-versus Svp/Doc-positive cells in the major part of the aorta are also distinct with regard to their function or physiology. Likewise, it is still unclear whether the two central pairs of cardioblasts, which express the ladybird (lb)homeobox genes in addition to tin(Jagla et al., 1997), are functionally distinct from the two posterior pairs of Lb^-/Tin⁺ cells within each segment.

In the present study, we test genetically whether tin possesses a later cardiogenic function specifically within the four Tin⁺ pairs of cardioblasts in each segment of the dorsal vessel. Based upon the known arrangement of the different enhancer elements, we generated genomic tin constructs that support normal patterns of tinexpression in the early, or early plus dorsal, mesoderm, without supplying tin expression in any cardioblasts of the dorsal vessel at later stages. By analyzing the phenotype of embryos that carry these constructs in a tin-null mutant background, we show that tin expression during the early stages is sufficient to specify the dorsal mesoderm derivatives, including cardiac progenitors, to generate a dorsal vessel, and to allow the development of adult flies. However, the specific ablation of functional tin in cardioblasts reveals that its function is normally required for the normal diversification of cardial cells and the proper differentiation of the four Tin⁺ cardioblast pairs within each segment. In the heart region, this function of tin is required for preventing `working' myocardial cells from acquiring ostia-like features,which include specific morphological properties and the expression of wingless (wg) (Lo et al., 2002; Ponzielli et al.,2002). We demonstrate that a key role of tin is the repression of Doc and, hence, the restriction of Doc to two segmental pairs that include the future ostial cells. This activity can also be fulfilled by Nkx2.5, thus indicating that the repressive potential of NK2-class homeodomain proteins has been conserved during evolution. In addition, tin promotes the expression of particular differentiation genes specifically in the Tin<sup>+</sup> cardioblasts, which are likely to contribute to the differential properties of the `working'myocardium. We present a model in which svp in cardiac progenitors and their descendents represses tin in two cardioblast pairs within each segment from A2 to A7; this allows the expression of Doc by default, which in turn contributes to the continued repression of tinin these cells. Together, these interactions lead to the establishment and maintenance of two mutually exclusive differentiation states of cardioblasts,which are modulated further by differential homeotic gene activities in the aorta versus heart regions of the dorsal vessel. In addition, we show that the function of tin is required for normal remodeling and growth of the dorsal vessel during metamorphosis in pupal stages and, thus, is needed for generating the enlarged myocardium observed in adult flies.

tin³⁴⁶ (Azpiazu and Frasch, 1993), tin^EC40(Bodmer, 1993) and mid¹ (Buescher et al.,2004), and the deficiency lines Df(3R)GC14(Azpiazu and Frasch, 1993; Bodmer, 1993), Df(3L)DocA and Df(3L)29A6(Reim et al., 2003) and svp^AE127 (svp-lacZ; from Y. Hiromi, National Institute of Genetics, Mishima, Japan) were balanced with CyO,wg-lacZ or TM3, eve-lacZ. Ectopic expression experiments with UAS-tin (Yin and Frasch,1998), UAS-Doc1, UAS-Doc2(Reim et al., 2003),UAS-svp1 (from M. Hoch, Bonn University, Germany) or UAS-Nkx2.5 (lines 1, 2b, and 3a) were carried out at 28°C. UAS-Nkx2.5 was generated from a full-length cDNA of Flag-Nkx2.5 (Kasahara et al.,2001), initially subcloned as a HindIII/NotI blunted fragment into EcoRV of pBS-SK and then into EcoRI of pUAST vector (Brand and Perrimon,1993). S59-Mef2-HtΔD-Gal4 (line 10-2a) contains a minimal cardioblast enhancer of Mef2 (Hanh Nguyen, AECOM, Bronx, NY, unpublished) and a minimal skeletal muscle enhancer of S59/slouch (M.F., unpublished) in pGAL4-221 (from Christian Klämbt, Münster University, Germany), and is active in cardioblasts,in S59-positive somatic muscles and, weakly, in some pericardial cells.

We used rabbit anti-β3-Tubulin (1:1500, TSA; from Renate Renkawitz-Pohl, Philipps University Marburg, Germany), rat anti-Bin (1:500)(Zaffran et al., 2001),anti-α-actinin (Saide et al.,1989), rabbit anti-Homothorax (1:500) (from Richard Mann, Columbia University, New York, NY), monoclonal rat anti-Tropomyosin (1:500), mouse anti-β-galactosidase 40-1a (1:60, TSA), rabbit anti-Toll (1:500, TSA;from Steve Wasserman, UC San Diego, CA) and mouse anti-α-Spectrin 3A9(1:10, TSA) (from the Developmental Studies Hybridoma Bank, University of Iowa, developed under the auspices of NICHD). Other antibodies and the in situ probes for bkh, mid, H15, svp1 and Sur are described by Reim et al. (Reim et al., 2005) and by Lo and Frasch (Lo and Frasch,2001). A Zeiss Axiophot and the confocal Leica TCS-SP and Zeiss LSM 510 META systems were used for analysis.

Restriction fragments from the genomic region of tin in pCaSpeR-Re28 (Azpiazu and Frasch,1993) were subcloned into pBluescript KS (Stratagene). For,tin-AB, the large genomic EcoRI fragment was cloned into pCaSpeR-3. The tin-D enhancer element (Yin et al.,1997) was cloned downstream of the tin-AB fragment to generate tin-ABD. For tin-D, the EcoRV/EcoRI fragment from the tin cDNA was used to replace the corresponding genomic fragment in tin-ABD, leaving intact the minimal promoter. All pCaSpeR constructs were sequenced and several transformant lines from each were analyzed. For rescue with tin-ABD, the insertions T003-1B2 (2nd chromosome) and/or T003-1C1 (3rd chromosome) were used.

Females and males were separated and aged to 2-3 days post-eclosion. Flies were aligned between two electrodes on a glass microscope slide and then paced to 6 Hz for 30 seconds using a square wave stimulator(Wessells and Bodmer, 2004; Wessells et al., 2004). Heart failure rate is defined as the percentage of flies that either arrest or fibrillate during or immediately after pacing. Flies that have undergone heart failure were observed for 2 minutes to calculate the percentage of flies that recover to a normal resting heartbeat [recovery rate; see Wessells et al.(Wessells et al., 2004) for the survival assay].

Embryonic tin expression can be separated into the three major phases (Azpiazu and Frasch,1993; Bodmer,1993), each regulated by a separable enhancer module that is subject to distinct regulatory inputs (Yin et al., 1997; Venkatesh et al., 2000). The earliest tin expression, which is observed in the entire trunk mesoderm during gastrulation, is recapitulated by constructs containing the enhancer element tinB(Fig. 1A,D). The Dpp-dependent expression in the dorsal mesoderm, which includes the primordia of the dorsal vessel, visceral muscles and somatic dorsal muscles, during stage 10-11 is driven by tinD (Fig. 1B,D) (Xu et al.,1998). At later stages, the enhancer tinC mediates expression in a segmental subset of cardioblasts of the developing and mature dorsal vessel (Fig. 1C,D)(Lo and Frasch, 2001).

The maintained tin expression in the dorsal vessel indicates a yet undefined role for tin in the proper differentiation and normal physiology of myocardial cells. To test this possibility, we aimed to restore early tin activity in an otherwise tin-null mutant background in order to permit proper specification of cardiac progenitors without allowing tin expression in cardial cells themselves. To this end, we produced a series of genomic tin rescue constructs containing specific subsets of these enhancer elements without the cardioblast-specific enhancer `C' (Fig. 1D), and brought them into tin mutant backgrounds(Fig. 1E,F; note that the known inactivity of the tinD and tinC enhancers in embryos lacking early tin activity prevented us from performing analogous rescue experiments with tin driven solely by tinD or tinC).

First, we tested whether the early phases of tin expression in the entire trunk mesoderm and/or the dorsal trunk mesoderm would be sufficient for cardiac specification and dorsal vessel formation. Surprisingly, tinmutant embryos that express Tin protein from a transgenic rescue construct driven by the twist-dependent enhancer (tin-AB) only in the early mesoderm are able to specify progenitors of all derivatives of the dorsal mesoderm. For example, staining for Biniou, a FoxF-related factor expressed in the visceral mesoderm(Zaffran et al., 2001), shows that the visceral muscle precursors are present in tin-AB; tin mutant embryos (Fig. 2A,B). The progenitors of cardioblasts and pericardial cells, as marked by their expression of brokenheart (bkh)(Fremion et al., 1999) and even-skipped (eve)(Frasch et al., 1987), are also present in these mutants, although their number is variably reduced when compared with wild-type embryos (Fig. 2C-F). Also at later stages, cardioblasts and all types of pericardial cells are present in tin-AB; tin mutant embryos,although the dorsal vessel appears discontinuous, the arrangement of cardioblasts and pericardial cells is irregular, and lymph gland formation is impaired (Fig. 2E-H).

Altogether, these and other data suggest that the early phase of tin expression is necessary and sufficient to specify the progenitors of all cell types generated from the dorsal mesoderm, but specification occurs with reduced efficiency. This activity of tin in tin-AB; tin mutant embryos in the absence of the second, dorsal-specific phase of tin expression is probably due to Tin protein perduring in the dorsal and cardiac mesoderm from its earlier phase of twist-driven mRNA expression. However, the combination of the first two phases of tin expression, in both early in the entire early mesoderm and subsequently in the dorsal mesoderm, leads to an improved rescue of cardiac specification and allows the formation of lymph glands (shown below). This demonstrates that Dpp-induced tin does contribute to the full biological activity of tin. We conclude that the absolute requirement for dpp during the specification of dorsal mesodermal derivatives is mainly due to the requirement of synergistic Tin and activated Smads during the induction of Tin/Dpp targets, and to a lesser degree to the induction of tin expression itself by Dpp.

Next, we used a transgenic construct expressing tin throughout its two early phases (twist and Dpp dependent, respectively) with the aim of obtaining robust expression close to wild-type levels of Tin in the cardiogenic mesoderm in otherwise tin mutant animals. Staining of such embryos for several markers that normally show expression in all cardiomyocytes shows that the absence of cardiac tin expression has little effect on dorsal vessel formation as such. Notably, an almost normal number of cells expressing the pan-myogenic transcription factor Mef2(Bour et al., 1995; Lilly et al., 1995) is detectable along the dorsal midline of tin-ABD; tin³⁴⁶ mutant embryos (compare Fig. 3B with 3A). The slight and variable decrease in cell numbers when compared with the wild type could be due to variable and still suboptimal tin expression from the transgenic tin-ABD rescue construct. In addition, cardioblasts of tin-ABD; tin³⁴⁶ mutant embryos show nearly normal expression of general cardioblast markers, including the Tbx20-homologs H15 and midline (mid)(Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim et al.,2005), the bHLH gene Hand(Kölsch and Paululat,2002), the sarcomeric protein Tropomyosin(Wolf et al., 2006) and the transmembrane receptor encoding gene Toll(Wang et al., 2005)(Fig. 3A-H). These data lead to the conclusion that cardiac precursors differentiate into myocardial cells in the absence of cardioblast-specific tin expression. However, staining for Toll, which marks the basolateral membrane of the cells, and for the cytoskeletal protein α-Spectrin demonstrate that in absence of tin, the morphology of the posterior `heart' region in particular is abnormal (Fig. 3H, and data not shown). The cardioblast rows are uneven, some cardioblasts appear intercalated, and the shape of the cardioblasts is less regular, indicating that tin is needed for some aspects of proper specification or differentiation of myocardial cells.

Based on the expression patterns of various genes, cardioblasts are thought to acquire at least three different identities within each segment of the dorsal vessel, the Tin-positive cells, the Tin/Lb-positive cells and the Svp/Doc-positive cells (Lo and Frasch,2003). The Svp/Doc-positive cells in the posterior three segments of the dorsal vessel become different from the remaining cardioblasts morphologically and functionally, as these are the cells that form the ostia(inflow valves) of the larval heart(Molina and Cripps, 2001). In addition, based on the restricted expression of certain genes specifically in the Tin-positive cells such as Sulfonylurea receptor (Sur),which encodes a K⁺ channel subunit(Nasonkin et al., 1999), and the structural protein β3-tubulin(Kremser et al., 1999),Tin-positive and Tin-negative cells are thought to possess different physiological properties. To test for possible alterations in the identities of myocardial cells in mutants lacking tin expression in the developing dorsal vessel, we examined the expression of the available identity markers in tin-ABD; tin³⁴⁶ mutant embryos(Fig. 4). Strikingly, Doc,which is normally expressed in the two Tin^- pairs of cells within each segment, is expressed ectopically and detected in all cardioblasts in these embryos (compare Fig. 4B with 4A). Conversely, the expression of Lbe, which is normally co-expressed with Tin in subsets of cardioblasts, is lost(Fig. 4C,D)(Jagla et al., 1997). The expanded expression of Doc is highly reminiscent of what has recently been described for mid mutants (Reim et al., 2005). These mutants also lose Tin from developing cardioblasts, which explains the similarities of the phenotypes with regard to Doc expression and of other phenotypes such as disrupted cardioblast diversification in the `heart' region (Fig. 3H) (Reim et al.,2005). Furthermore, in both cases high-level expression of the Sur gene, a likely target of Tin in the Tin-positive cells, is lost(Fig. 4E,F)(Reim et al., 2005; Akasaka et al., 2006). Surprisingly, β-Tubulin at 60D(βTub60D, also known as β3-tubulin), which has also been reported to be a target gene activated by Tin in the dorsal vessel(Kremser et al., 1999), is still expressed in the tin-ABD; tin³⁴⁶ mutant background and maintains its intrasegmental modulation, albeit with a slightly disturbed pattern (Fig. 4G,H). This observation suggests that the β3-tubulin gene is driven by both tin-dependent and tin-independent enhancers in the Tin⁺ cardioblasts. In addition, the segmental pattern ofβ 3-tubulin in tin-ABD; tin³⁴⁶mutants shows that there must be factors other than Tin and Doc that can still generate a patterned expression of certain differentiation genes in the dorsal vessel. svp, which is activated during early stage 12 possibly in response to Hh signaling (Ponzielli et al., 2002), also maintains its segmental expression in this genetic background (Fig. 4I,J),and hence may act as a repressor that antagonizes the activation ofβ 3-tubulin by a uniformly expressed activator in myocardial cells. We predict that other markers like β3-tubulin exist that are regulated differentially by svp but do not depend on tin.

The anterior aorta, which lacks the Svp⁺ cardioblasts(Fig. 4I)(Lo and Frasch, 2003), also shows ectopic Doc expression in the absence of myocardial Tin(Fig. 4B,D), whereas the expression of the anterior-specific marker Homothorax (Hth; Perrin et al., 2004) is unaffected (see Fig. S1 in the supplementary material). In addition, the near absence of Tin in pericardial cells causes a loss of pericardial ladybird and a reduction of pericardial zfh-1 expression,but the only known marker specific for Tin^- pericardial cells, Odd,does not exhibit any concomitant ectopic expression (see Fig. S2 in the supplementary material).

The observed expansion of Doc expression, as well the morphological abnormalities in the heart region, suggest a switch in cardial cell identities in dorsal vessels lacking Tin. In order to test this notion further and to establish the underlying regulatory mechanisms, we examined the expression of regulatory factors associated with cell diversifications into different cardioblast subtypes under various genetic conditions. Fig. 5A visualizes the expression of the three known regulatory genes, tin, Doc and svp (as detected by svp-lacZ expression from a heterozygous lacZ enhancer trap insertion in svp), which are expressed in the described `4+2' pattern in an otherwise wild-type embryo. Whereas in the absence of Tin in developing cardioblasts Doc expression is expanded into all cardioblasts, svp-lacZ maintains its restriction to two pairs of cardioblasts per segment (Fig. 5B, yellow arrows). Although we have shown previously that svp is normally required for Doc expression in the two Svp⁺ cells in each hemisegment(Lo and Frasch, 2001), our present observation shows that Doc expression can occur in Svp-negative cardioblasts, provided that Tin is not present. Hence, we propose that Doc is under the repressive control of Tin rather than being activated directly by Svp. To confirm this hypothesis, we analyzed Doc expression in double mutants lacking Tin and Svp in cardioblasts. Although the dorsal vessels of such embryos (tin-ABD; svp^AE127 tin³⁴⁶) display some morphological defects, we still observe an expansion of Doc expression into all cardioblasts present (Fig. 5C). Thus, in the normal situation, the presence of Tin in four cardioblasts per hemisegment is essential for the repression of Doc in these cells. Accordingly, analysis of embryos in which tin is ectopically expressed in most (although not all) cardioblasts using S59-Mef2-HtΔD-Gal4 confirms the repressive activity of Tin towards Doc (Fig. 5D,E; the S59 enhancer-driven component serves as an internal expression control). The presence of svp-lacZ-positive Tin⁺/Doc^- cells suggests that Tin did not repress svp in these experiments. In conclusion, these data demonstrate that Tin acts as a repressor of Doc but not svp expression in cardioblasts.

Previous loss- and gain-of-function studies have shown that Svp acts as a repressor of tin (Gajewski et al., 2000; Lo and Frasch,2001; Molina and Cripps,2001) (see also Fig. 6B,C). This repressive effect of Svp towards tin is confirmed upon ectopic expression of Svp in the cardioblasts of the dorsal vessel (Fig. 5F). In addition,ectopic Svp leads to an expansion of svp-lacZ expression into some,but not all, of the Doc-positive cells, arguing for the presence of a positive feedback loop in those cells.

We further considered the possibility that Doc can also act as a repressor of tin. When UAS-Doc2 is ectopically expressed throughout the dorsal vessel (Fig. 5G), we observe a reduction or loss of tin expression in most of the Doc-expressing cells, particularly in the posterior aorta and heart regions. Repression of tin by Svp in the misexpression assay was consistently more robust throughout the dorsal vessel when compared with repression of tin by Doc. However, the repression of tin by Doc is independent of svp, as svp is not activated downstream of Doc in Tin-negative cardioblasts(Fig. 5G), and tinrepression is also seen when Doc2 is misexpressed in a homozygous svp^AE127 mutant background (see Fig. S3 in the supplementary material). In addition, analyses of embryos with genotypes in which Doc1 is deleted and the gene dose of Doc2 and Doc3 is reduced by half (Reim et al., 2003; Reim and Frasch,2005) also support the notion that Doc acts as a repressor during cardioblast diversification. In these embryos, we detect ectopic expression of tin in a number of Svp-positive cells, presumably because the lowered dose of Doc provided insufficient repressive activity towards tin (Fig. 5H). Taken together these findings show that Doc and Tin have mutually repressive functions in cardioblasts of the dorsal vessel.

The expression of Wingless (Wg) in the late stage embryonic heart in three segmentally repeated double pairs of cardioblasts(Fig. 6A) marks the differentiation of the Svp/Doc-positive cardioblasts within the heart region into ostia (inflow valves) (Lo et al.,2002). Previously, it has been shown that wg expression depends on svp and abd-A activities; it is missing in svp mutants, but expanded if svp is expressed ectopically(Lo et al., 2002; Perrin et al., 2004) (see also Fig. 6B,C). However, based on these data, it is not possible to discriminate between a direct activation of wg by Svp, an indirect activation by Svp via Doc or an absence of repression of wg by Tin in the Svp⁺cardioblasts. In order to distinguish between these possibilities, Doc2 was activated ectopically in the dorsal vessel(Fig. 6D), which produced a moderate expansion of Wg, with some but not all Doc⁺ cardioblasts in the `heart' region being positive for Wg. tin-ABD; tin<sup>346</sup> mutant embryos, in which svp is not affected, display ectopic activation of wg in all cardioblasts of the`heart' region (Fig. 6E). Furthermore, svp^AE127 tin³⁴⁶ double mutant embryos carrying the tin-ABD transgene also show expanded wgexpression (Fig. 6F). Hence, wg may either be activated by Doc, or it may be repressed by Tin while being activated by another factor that could be active in all cardioblasts. In summary, these data prove that svp is not directly required for Doc and wg gene activation, but rather as a repressor of tin, which in turn represses Doc and perhaps wg. These findings agree with a role of Svp as a transcriptional repressor, a function also fulfilled by its mammalian homolog COUP-TFII(Nr2f2) (Pereira et al.,1999).

To test directly whether the tin-related Nkx2.5 gene from the mouse possesses a repressive function equivalent to tin, and to assess whether such a function is active in the heterologous system, we expressed Nkx2.5 throughout all cardioblasts in late stage embryos. Although Nkx2.5 could not promote cardiogenesis in Drosophila (Park et al.,1998; Ranganayakulu et al.,1998), we observed that Doc and Wg expression are strongly repressed when Nkx2.5 is expressed in cardioblasts, similar to the effects seen with ectopic tin (compare Fig. 6H with 6I). These data indicate that Nkx2.5, like Tin, can have a repressive function towards some cardiac regulators.

In both larval and adult dorsal vessels of wild-type animals, the myofibrils are arranged spirally with essentially transverse orientations around the heart and aortic tubes (Fig. 7A,A′) (Molina and Cripps, 2001; Monier et al.,2005). By contrast, larval dorsal vessels from animals in which tin was never expressed in myocardial cells (tin-ABD,tin³⁴⁶/tin^EC40) show a very different ultrastructure. In these dorsal vessels, the myofibrils are arranged almost exclusively in an AP orientation, which leads to the appearance of striations(Fig. 7B,B′). The only exceptions are seen in the heart, where several abnormal cross-shaped or`knotted' patterns of myofibrils are present, particularly near the posterior end. In addition, the aortae in these larvae appear thinner when compared with the wild type, whereas the heart frequently has a wider diameter(Fig. 7B,B′, see Fig. S4 in the supplementary material; data not shown). Hence, myocardial activity of tin is required for establishing the normal ultrastructure of the contractile fibers and is, perhaps, related to this function, for generating a morphologically normal dorsal vessel. An almost identical alteration of the myofiber orientation has been reported for adult animals with ectopic expression of abd-A in the aorta(Monier et al., 2005);however, it is presently not known whether there is any mechanistic connection with our observation.

In adult animals lacking cardiac tin expression, we observe a much thinner heart tube (Fig. 7D,F, compare with 7C,E). As in larvae, the myofibrils are arranged longitudinally and transverse spirally arranged myofibrils are almost completely absent in these mutants (Fig. 7F, compare with 7E) (see also Molina and Cripps, 2001; Monier et al.,2005). The adult heart is generated by remodeling of the posterior larval aorta, which is accompanied by a significant widening of the tube and myocardial cell growth and the histolysis of the larval heart(Molina and Cripps, 2001; Monier et al., 2005; Sellin et al., 2006). Obviously, the absence of tin activity in myocardial cells impedes this process of heart remodeling. As a consequence, the small adult dorsal vessel in tin-ABD, tin³⁴⁶/tin^EC40flies probably represents the largely unchanged larval aorta after the larval heart has been histolyzed.

The altered ultrastructure of myocardial cells, as well as the observed changes in the expression of cardiac differentation genes and in cardial cell identities, would be expected to affect the functionality of the dorsal vessel. In order to examine the role of tin in the function of the adult heart, we measured cardiac responses to acute stress in tin-depleted adult flies and in controls. External electrical stimuli were applied to pace the heart of the fly to an elevated rate compared with wild type (Wessells et al.,2004). In response to this stress, flies either recover to a regular heart beat or else fail (defined as cardiac arrest or fibrillation). tin-deficient hearts showed a dramatic increase in heart failure rate, while those that are heterozygous for tin in the heart failed at the same rate as wild-type controls (20-30%; Fig. 7G; see Movies 1 and 2 in the supplementary material). In parallel, we monitored the recovery of flies that underwent either fibrillation or arrest. Two minutes after the pacing protocol, almost all wild-type and heterozygous flies recovered to a normal resting heartbeat, whereas in the absence of cardiac tin, only 40% of flies were able to recover (Fig. 7H). This suggests that tin function is required for a properly functioning adult heart. Moreover, flies without cardiac tinhave a reduced lifespan (Fig. 7I), consistent with a possible link between cardiac function and aging. Taken together, these data suggest that tin is required for the formation of the adult heart, in addition to its early requirement for the embryonic dorsal vessel. As a consequence, the lack of cardiac tincauses severe disruptions in cardiac contractility and rhythmicity (see Movies 1 and 2 in the supplementary material), and leads to a much elevated risk of heart failure in response to stress.

Our current data demonstrate that a major function of tin during later stages of embryonic cardiogenesis is the establishment of non-ostial,i.e. `working' myocardial cells. Doc and Svp restrict this function by repressing tin and, in the heart region, enable Tin-negative cells to form ostia (summarized in Fig. 6G). These data fill an important gap in the clarification of the full pathway of the regulation and function of tin during cardiogenesis. In this pathway, the early functions of tin (together with the GATA gene pnr and the Doc T-box genes) are required for activating the Tbx20 genes mid and H15 within early cardioblasts (Miskolczi-McCallum et al.,2005; Qian et al.,2005; Reim and Frasch,2005; Reim et al.,2005), and mid then feeds back to activate tinexpression specifically in myocardial cells(Reim et al., 2005). However,the expression of the COUP-TFII-related gene svp, which is activated during stage 12 in two cardioblast progenitors in each hemisegment between A2 and A7, prevents mid from activating tin in these cells(Gajewski et al., 2000; Lo and Frasch, 2001). tin then represses Doc, which appears to be activated by default (via yet unknown activators) in the Tin<sup>-</sup> cardioblasts and in turn contributes to the continued repression of tin in them. These interactions lead to the stabilization of two mutually exclusive differentiation states of myocardial cells that are defined presumably to a large extent by their differential expression of tin versus Doc and of their respective target genes. The particular features of these two segmental types of myocardial cells are further modulated by the activities of Hox genes such as abd-A, which leads to the differential formation of `working' (Tin⁺) versus ostial myocardial cells (Doc⁺, Wg⁺) that form the inflow valves in the heart region (Fig. 6G).

Whereas previous data have documented that Tin functions as a direct activator of specific target genes, our present data implicate Tin strongly in the repression of certain myocardial genes, including Doc and potentially wg. Indeed, interactions of Tin with the corepressor Groucho have been demonstrated in biochemical and cell culture experiments(Choi et al., 1999), and the N-terminal TN domain of Tin is proposed to function as an EH1 repressor domain(Copley, 2005). We propose that the activity of Tin either as an activator or a repressor is context-specific and is ultimately determined by the enhancer architecture of a particular target gene. For example, during early stages of cardiac induction by Dpp,combinatorial binding of Tin and Smads promotes activating functions of Tin,whereas in the Tin⁺ cardioblasts during later stages, Tin can either activate (e.g. Mef2, Sur) or repress (e.g. Doc). These opposite activities in the same cells are probably determined by the presence of additional binding sites for tissue-specific or ubiquitous co-factors that can switch Tin activity.

Our findings with the `cardiac-specific knockout' of tin are reminiscent of some of the phenotypes reported for mouse mutant phenotypes of Nkx2.5. For example, in both situations, cardiomyocytes and a linear embryonic heart tube with a largely normal morphology are formed(Lyons et al., 1995; Tanaka et al., 1999) (our present data). Although at first it was thought that, in the mouse, closely related Nkx genes could partially compensate for the functional loss of Nkx2.5, it is now widely assumed that cardiogenic genes encoding other classes of transcription factors, particularly those of the GATA and T-box families, exert most of the compensatory effects(Tanaka et al., 1998; Olson and Schneider, 2003). The latter type of compensation appears also to operate upon cardioblast-specific ablation of tin function, where the derepressed Doc T-box genes are presumably able to regulate myocardial development in the cells that would normally be under the control of tin. It will be instructive to determine the phenotype of animals in which the cardioblast expression of both tin and Doc has been ablated simultaneously. Likewise, the analysis of cardiogenesis with analogous double mutants in the mouse will be highly informative.

Additional phenotypic similarities between our tin mutant animals and mouse Nkx2.5 mutants include the failure to remodel the linear heart tube, which in the mouse leads to defects in looping morphogenesis and chamber formation and in Drosophila to a failure of converting the larval aorta into an adult heart. As in adult flies, a conditional knockout of mouse Nkx2.5 also leads to defects in ventricular cell lineage specification and maturation, which is accompanied by the aberrant down- or upregulation of cardiac differentiation genes(Pashmforoush et al., 2004). Notably, a major difference between the vertebrate and Drosophilasystems is the early and broad mesodermal expression of Tin, which is essential for Drosophila cardiac induction and is not compensated for by other factors.

We thank Seigo Izumo and Patrick Jay for the Flag-Nkx2.5 plasmid. We are grateful to Margaret Buckingham for allowing Stéphane Zaffran to carry out part of this work in her laboratory. The research was supported by grants from NIH-NICHD (HD30832) to M.F and from NIH-NIA (2P01AG015434) to R.B. Confocal laser-scanning microscopy at the MSSM-Microscopy Shared Resource Facility was supported by NIH-NCI (R24 CA095823) and NSF (DBI-9724504). I.R. was supported by an AHA/Harriet and Robert Heilbrunn Research Fellowship in Cardiovascular Genetic Research. S.Z. is a fellow researcher from the `Centre National de la Recherche Scientifique' in France.