Gata factor Pannier is required to establish competence for heart progenitor formation

In Drosophila, bilaterally symmetric heart progenitors are specified within the dorsal most region of the mesoderm. These progenitor cells then migrate to the dorsal midline where they form a linear heart tube consisting of two different cell types, the inner contractile myocardial cells and the outer pericardial cells (Rizki,1978), subtypes of which have been identified based on gene expression, function and lineage relationships (Alvares et al., 2003; Han and Bodmer, 2003; Ponzielli et al., 2002; Lo et al., 2002). Several regulatory genes have been identified to be required for the specification of cardiac progenitors within the dorsal mesoderm, including the homeobox transcription factor Tinman (Tin), and the TGFβ and Wnt signaling molecules encoded by dpp and wingless (wg),respectively (Bodmer, 1993; Azpiazu and Frasch, 1993; Frasch, 1995; Wu et al., 1995; Park et al., 1996; Azpiazu et al., 1996; Riechmann et al., 1997). wg, dpp and tin are not only necessary for heart formation,but as overexpression studies suggest the spatial convergence of wgand dpp signaling on cells expressing tin is also sufficient for cardiac-specific differentiation(Lockwood and Bodmer, 2002). A mesodermal mediator of ectodermal wg signaling to the mesoderm is achieved by activation of a transcription factor encoded by sloppy-paired (Lee and Frasch,2000), but it is not known if the cardiogenic dpp signal is also mediated indirectly.

There are striking molecular and developmental similarities between vertebrate and Drosophila heart development(Bodmer, 1995; Bodmer and Venkatesh, 1998; Bodmer and Frasch, 1999). Developmentally, both vertebrate and Drosophila hearts are formed from bilaterally symmetrical rows of mesodermal cells, which will eventually migrate to the midline, where they will fuse to form a linear heart tube. More importantly, tin and dpp, two factors that determine the initial formation of the Drosophila heart, also have vertebrate counterparts (Nkx2.5 and Bmp2/4, respectively) with a similar function in cardiogenesis (Harvey, 1996; Schultheiss et al., 1997). In contrast to Drosophila, canonical Wnt signaling in vertebrates needs to be prevented for promoting heart formation in the anterior lateral plate mesoderm (Schneider and Mercola,2001; Marvin et al.,2001). However, the non-canonical Wnt pathway is required for heart formation in vertebrates (Pandur et al., 2002).

Six Gata transcription factors have been identified in vertebrates,characterized by two conserved DNA-binding zinc fingers(Evans and Felsenfeld, 1989; Tsai et al., 1989; Yamamoto et al., 1990). Gata1, Gata2 and Gata3 are largely expressed in hematopoietic stem cells(reviewed by Orkin, 1998), and Gata4, Gata5 and Gata6 are expressed in several mesoderm- and endoderm-derived tissues, including the developing heart(Arceci et al., 1993; Kelley et al., 1993; Heikinheimo et al., 1994; Laverriere et al., 1994; Jiang and Evans, 1996; Morrisey et al., 1996), where they are thought to regulate cardiac-specific genes(Grepin et al., 1994; Ip et al., 1994; Durocher et al., 1997; Murphy et al., 1997) (reviewed by Molkentin, 2000). Gata4 is already expressed in the early cardiac crescent of the lateral plate mesoderm,and in mice deficient for Gata4, these heart primordia fail to migrate towards the midline where they normally fuse into the primitive heart tube(Molkentin et al., 1997; Kuo et al., 1997). Owing to these ventral closure defects, it has been difficult to discriminate between a direct role for Gata4 in heart formation and an indirect involvement via its function in ventral morphogenesis. Furthermore, Gata4, Gata5 and Gata6 may act in part redundantly, which may further occlude their cardiogenic potential. Consistent with the direct involvement of Gata4 in heart development is the congenital heart disease phenotype observed in individuals heterozygous for deletions of chromosome 8p23.1 region, which includes the GATA4 gene(Pehlivan et al., 1999; Bhatia et al., 1999).

In vitro, Gata4 interacts with a wide array of proteins, including the Tinman homolog Nkx2.5, the bHLH protein Hand and the multiple zinc-finger protein Fog2 (Durocher et al.,1997; Sepulveda et al.,1998; Lee et al.,1998; Lu et al.,1999; Sepulveda et al.,2002; Svensson et al.,1999; Tevosian et al.,1999; Dai et al.,2002). Fog2 apparently modulates Gata-mediated transcriptional regulation not only as a repressor, but also as an activator, depending on the promoter and on cell type (Lu et al.,1999). Fog2 is co-expressed with Gata4 in embryonic and adult cardiomyocytes, and Fog2-deficient mice exhibit severe developmental heart defects, suggesting a direct cardiogenic requirement(Tevosian et al., 2000; Svensson et al., 2000). Moreover, these heart defects are rescued by cardiac-specific transgenic expression of Fog2, providing strong evidence for a cardiac autonomous function (Tevosian et al.,2000).

The three Gata factors found in Drosophila (pannier,serpent and grain) also play important developmental roles(Abel et al., 1993; Ramain et al., 1993; Winick et al., 1993; Lin et al., 1995; Heitzler et al., 1996; Rehorn et al., 1996; Sam et al., 1996; Riechmann et al., 1998; Gajewski et al., 1999; Brown and Castelli-Gair Hombria,2000; Calleja et al.,2000; Herranz and Morata,2001). serpent is required for endodermal gut development, mesodermal fat body formation and hematopoiesis. grainis involved in filzkorper and head skeleton morphogenesis. pannier(pnr) is best known for its requirement during embryonic and adult dorsal closure, and for dorsomedial patterning. The Drosophilacounterpart of Fog2, U-shaped (Ush), can physically interact with Pnr, and (as with Gata4 and Fog2) this interaction is mediated by the N-terminal zinc finger of Pnr, which is thought to antagonize the role of Pnr as a transcriptional activator (Haenlin et al.,1997; Cubadda et al.,1997). At blastoderm, pnr and ush are expressed in response to the dorsal morphogen encoded by dpp(Winick et al., 1993; Jazwinska et al., 1999; Ashe et al., 2000), and are thought to be part of the process that subdivides the dorsal ectoderm(Herranz and Morata,2001).

It has been proposed that pnr promotes myocardial as opposed to pericardial cell fates within the cardiac mesoderm(Gajewski et al., 1999; Gajewski et al., 2001) and that ush antagonizes this function(Fossett et al., 2000; Fossett et al., 2001). Recent lineage studies, however, have indicated that some heart progenitors give rise to mixed myocardial/pericardial progeny, but others do not(Park et al., 1998; Ward and Skeath, 2000; Han and Bodmer, 2003; Alvarez et al., 2003), raising the question of how pnr functions in different heart progenitor populations. We have re-examined the cardiogenic role of these two genes. We find that pnr is required for formation of all tin-expressing cardiac progenitors, and loss of pnr function results in loss of both myocardial and pericardial cell populations. By contrast, loss of ush function did not affect the initial expression of tin in the cardiac mesoderm, but is required for its maintenance of expression as well as for the correct differentiation of both myocardial and pericardial cells. Moreover, specific aspects of early cardiac differentiation were preferentially affected: most of the seven-up(svp)-expressing cells were absent in both mutants, more ladybird (lbe)-expressing cells were absent in pnrthan in ush mutants, and the heart cells expressing even-skipped (eve) were only moderately affected in pnr and virtually not at all in ush mutants. Overexpression of pnr in the entire mesoderm produces ectopic tinexpression, which is strongly antagonized by co-overexpression of ush, suggesting a dual role for ush: one that is necessary for cardiogenesis and another that counteracts pnr function. The heart phenotype of either mutant is rescued by mesoderm-specific expression of wild-type pnr or ush cDNA, respectively; and mesodermal expression of a dominant-negative form of pnr (pnrEnR)mimics the heart defects of pnr mutants when expressed in the mesoderm. Interestingly, dorsal ectodermal dpp expression fades after germband retraction in pnr mutants and cardiac differentiation is also compromised when pnrEnR is overexpressed in the ectoderm. Moreover, mesoderm-specific expression of brinker (brk), a repressor of dpp target genes(Jazwinska et al., 1999; Zhang et al., 2001), has a similar phenotype as pnr mutants or mesodermal pnrEnR expression, suggesting that pnr may be mediating, at least in part,the cardiogenic dpp signal in the mesoderm. Thus, we propose a dual role for pnr in heart development: (1) pnr functions as a mesodermal target and mediator of the ectodermally derived dpp signal by acting in concert with tinman; and (2) pnr is also required in the ectoderm for maintaining dorsal stripe dppexpression.

The following mutant stocks were used: pnrVX6 is considered to be a null allele, because it contains a small deletion that eliminates all but nine amino acids of the pnr-coding region at the N terminus(Heitzler et al., 1996). Df(2)ush^rev18 is a null allele that deletes the entire gene and some flanking genomic DNA(Cubadda et al., 1997). Misexpression of full-length transgenes was achieved using the Gal4-UAS system(Brand and Perrimon, 1993),using the following stocks: UAS-pnr(Haenlin et al., 1997), UAS-ush (Cubadda et al.,1997), UAS-pnrD4(Haenlin et al., 1997), UAS-tin (Ranganayakulu et al.,1998), UAS-brk(Jazwinska et al., 1999), UAS-pnrEnR (see below), da-Gal4(Wodarz et al., 1995),ZKr-Gal4 (Frasch, 1995),69B-Gal4, 24B-Gal4 (Brand and Perrimon,1993), twi-Gal4 (Greig and Akam, 1993) and twi-Gal4;24B-Gal4(Lockwood and Bodmer, 2002). twi-Gal4, 24B-Gal4 and twi-Gal4;24B-Gal4 drive expression of UAS constructs exclusively within the entire trunk mesoderm, without detectable expression in the ectoderm. twi-Gal4 initiates expression earlier (at least by stage 9) than 24B-Gal4 (stage 11). ZKr-Gal4 drives expression exclusively in the dorsolateral ectoderm, with highest levels in segments T3-A3, whereas 69B-Gal4 drives expression predominantly throughout the ectoderm but with less germ layer specificity than ZKr-Gal4. da-Gal4 drives expression ubiquitously. The following stocks were used for the rescue experiments:

• UAS-pnr;pnrVX6/TM3-P[twi-lacZ]

• twi-Gal4;pnrVX6/TM3-P[twi-lacZ]

• UAS-pnr;pnrVX6,da-Gal4/TM3-P[ftz-lacZ]

• Df(2)ush^rev18/CyO-P[wg-lacZ];UAS-ush

• Df(2)ush^rev18/CyO-P[wg-lacZ];24B-Gal4

• UAS-pnr;pnrVX6,ZKr-Gal4/TM3-P[ftz-lacZ]

All crosses were performed at 29°C. Combinations of transgene insertions were generated using standard genetic crosses. Oregon-R was used as the wild-type reference strain.

The dominant-negative pnr (UAS-pnrEnR) was constructed according to the strategy described by Fu et al.(Fu et al., 1998). Basically the construct contains the repressor domain from engrailed (EnR,amino acid 2-298) (Jaynes and O'Farrell,1991; Smith and Jaynes,1996; Tolkunova et al.,1998) and the two N-terminal zinc-finger domains from pnr(amino acid 153-293) (Ramain et al.,1993). The pnr zinc-finger domains were PCR amplified from the full-length pnr cDNA (5′ primer,CATCTCGAGATGCAGTTCTACTCGCCAAACGCC; 3′primer,GCTCTAGACTACCTCCAAAGTGGAGCCTGTTC) and inserted into XhoI-and XbaI-digested pUAST vector already containing the EnR domain(Fu et al., 1998; Han et al., 2002). Transgenic flies were generated as previously described(Brand and Perrimon,1993).

Immunohistochemistry and in situ hybridization were performed as described(Wu et al., 1995), except that Cy3- or FITC-conjugated secondary antibodies (The Jackson Laboratory)were used for fluorescent confocal microscopy. Fluorescent in situ double labeling was performed as described (Knirr et al., 1999). For Lbe staining the TSA Plus Fluorescence System was used (Perkin Elmer). Embryos were mounted in VectaShield (Vector Laboratories). Fluorescent embryo staining was analyzed by using a Zeiss LSM510 confocal microscope. Primary antibodies were used at the following dilutions: rabbit anti-Eve, 1:300 (Frasch et al., 1987); mouse anti-PC 1:10(Yarnitzky and Volk, 1995);mouse anti-Lbe 1:40 (Jagla et al.,1997); and rabbit anti-Mef2 1:2000(Lilly et al., 1995). Biotinlylated secondary antibodies (Vector Laboratories) were used at 1:200. The following RNA probes were used: the dpp probe was generated from the 2.9 kb dpp E55 fragment(Padgett et al., 1987), the tin probe from a 1.7 kb insert(Bodmer et al., 1990), the svp probe from a 3.1 kb insert(Mlodzik et al., 1990), the pnr probe from a 1.6 kb fragment(Ramain et al., 1993) and the Hand probe from a 0.5 kb insert(Moore et al., 2000).

For expression analysis, 25-50 embryos were used as a sample size. Embryos were placed in categories based on expression: +, less than 1/4 staining or expression when compared with wild type; ++, 1/4 to 1/2; +++, 1/2 to 3/4;++++, 3/4. When ZKr-Gal4 was used, only the segments T3-A3 were assayed.

pnr and ush are both expressed in the mesoderm at the time of cardiac mesoderm formation (Fig. 1), in addition to their expression in the dorsal ectoderm. Mesodermal expression of pnr is restricted to the dorsal cardiogenic margin, whereas ush extends more laterally(Fig. 1D)(Gajewski et al., 1999; Fossett et al., 2000). In order to assess the requirement for pnr and ush in initiating cardiac mesoderm and cardiac cell type-specific differentiation, we first examined tin expression at progressively later developmental stages in null mutants for both pnr and ush. During mid-stage 11, tin is expressed segmentally in two regions of the mesoderm (Fig. 2A). The dorsal clusters of cells correspond to the cardiac precursor cells, whereas the lateral clusters will become part of the visceral mesoderm. In same stage pnr mutant embryos, tin expression is dramatically reduced in the clusters that correspond to the cardiac precursors, indicating that cardiogenesis is not being initiated (Fig. 2B,D). tin expression in the visceral mesodermal clusters, as well as tin expression earlier in development, is unaffected, suggesting the heart is a focal point for pnr function,which is consistent with its cardiac-restricted expression in the mesoderm(Fig. 1)(Gajewski et al., 1999). By contrast, ush mutant embryos initially seem to exhibit normal tin expression (Fig. 2C,D). At later stages, when tin expression is solely restricted to the heart cells, ush mutants display a progressively more severe reduction in tin expression, approaching the phenotype of pnr mutants (Fig. 2E-L). Thus, both pnr and ush are required for heart-specific tin expression, although ush seems to be initially dispensable.

Even though tin is initially expressed in all heart progenitors,its expression is later turned off in some specific lineages, but continues to be expressed in many myocardial and pericardial cells(Bodmer, 1993; Ward and Skeath, 2000; Venkatesh et al., 2000; Han et al., 2002). To determine which heart cells are affected in pnr and ushmutants, we examined mutant embryos with various markers. eve, for example, is co-expressed with tin in 11 clusters of heart progenitors(Fig. 2M), and these lineages give rise to a subset of pericardial cells(Frasch et al., 1987). eve expression is only moderately reduced in pnr and hardly at all in ush mutants at early as well as later stages(Fig. 2M-S; note, however, the patterning defects at progressively later stages in Fig. 2R,S). By contrast, the lbe-expressing heart progenitors, which produce both myocardial and pericardial cells, are dramatically reduced in pnr but less so in ush mutants (Fig. 2Q-T). Moreover, the svp-expressing cells, which also give rise to a mixed lineage, but cease to co-express tin at later stages, are dramatically reduced in both mutants(Fig. 2U-X). Thus, all lineage markers we assayed are reduced in both mutants, but each is affected with disproportional severity, which is consistent with the idea that the formation of each cell type has a direct requirement for pnr and ush.

By stage 16, dorsal closure is complete and the linear heart tube has assembled beneath the dorsal midline. A general marker for pericardial cells shows a severe reduction in these cells in both mutants(Fig. 2Y1-Y4). As pnr and ushmutants fail to undergo dorsal closure, we wanted to determine if this process was a prerequisite for cardiac cell-type specification, by perhaps causing heart defects indirectly. As a test of this hypothesis, we examined another dorsal closure mutant, raw (Byars et al., 1999), in which we observe pericardial cell staining that is normal or in excess along the dorsal mesoderm(Fig. 2Z). This increase in cardiac differentiation is probably due to an excess in dppsignaling. Thus, a dorsal open phenotype in itself is insufficient to compromise cardiac differentiation.

Analysis of pnr and ush mutants suggests that both genes functions are required for heart formation. In order to explore their functional relationship in heart development further, we performed overexpression studies. When pnr is expressed throughout the mesoderm, tin expression is no longer confined to the heart precursors by late stage 11, but is expanded laterally throughout the mesoderm, suggesting that pnr is sufficient to ectopically initiate tin expression within the mesoderm(Fig. 3A,B). This is in contrast to mesodermal overexpression of tin, which does not seem to cause significant initiation of cardiogenesis without spatially intersecting with dpp (and wg) signaling(Lockwood and Bodmer, 2002). Much of this lateral expansion of tin driven by ectopic pnrdoes not persist beyond stage 13, where ectopic tin is reduced to small ventrolateral cell clusters (Fig. 3H). These results suggest that pnr can activate early ectopic expression of tin, but by itself is insufficient to maintain it at significant levels.

As ush is required to maintain tin expression in the heart-forming region, we wanted to see if ush can also provide a maintenance role ectopically. Pan-mesodermal ush expression, however,does not cause an expansion but rather a reduction of cardiac-specific tin expression (Fig. 3C,I), similar to ush loss of function(Fig. 2G,K). These findings suggest that a correct amount of ush activity is crucial for heart development, which is consistent with a model in which Ush and Pnr act in a multiprotein complex. To examine this idea further, we co-overexpressed both genes throughout the mesoderm. Similar to overexpressing ush alone,co-overexpression results in a reduction in tin expression(Fig. 3D,J), unlike what is observed with overexpression of pnr alone, suggesting that excess ush inhibits the level of tin activation by pnr in normal as well as ectopic locations. This repressor function of ushis reminiscent of its role in adult mechanosensory bristle formation and thorax development (Cubadda et al.,1997; Sato and Saigo,2000; Tomoyasu et al.,2000). These results further support the idea that the appropriate level of ush activity is crucial for correct heart development.

Previous data suggest that Ush exerts its inhibitory activity by binding to the N-terminal zinc finger of Pnr, an interaction that is blocked in the allele pnrD4, which has an amino acid substitution in this domain and thereby abolishes Ush binding to Pnr(Haenlin et al., 1997). When we overexpressed this gain-of-function allele of pnr in the mesoderm,we also observed ectopic induction of ventrolateral tin expression in late stage 11 embryos (Fig. 3E), as with overexpression of the wild-type form of pnr(Fig. 3B). At later stages,however, ectopic tin levels increase dramatically in the ventrolateral mesoderm and exceed those of wild-type pnr mesodermal overexpression (Fig. 3H,K). Unlike co-overexpression of wild-type pnr and ush, using pnrD4 in conjunction with ush does not cause a ush-like phenotype but rather one like pnrD4, which produces ectopic tin expression (Fig. 3L), suggesting that ush is unable to inhibit the gain of function of this pnr allele. Taken together, these data are consistent with a dual function of ush: (1) a positive role in maintaining tin expression within the cardiogenic region and (2) a negative role in limiting the level and spatial distribution of pnractivity (see Fig. 1D for normal patterns of expression).

To determine if pnr cooperated with tin in heart formation, we examined other markers of cardiac-specific differentiation. Similar to the presence of ectopic tin(Fig. 3H), ectopic expression of Hand, a general heart marker(Fig. 3M)(Kolsch and Paululat, 2002),is also observed ventrolaterally when pnr is induced throughout the mesoderm (Fig. 3N). Interestingly, more ectopic Hand expression is induced by co-overexpressing pnr as well as tin(Fig. 3O), similar to the extent of ectopic tin with pan-mesodermal pnrD4(Fig. 3K). This indicates that pnr and tin act synergistically in their ability to induce heart formation (overexpression of tin alone does not cause ectopic heart induction) (see Lockwood and Bodmer,2002), and that the presence of `activated' PnrD4 is sufficient to sustain heart formation.

It is well established that both ectodermal and mesodermal patterning information is required for heart development(Bodmer, 1993; Azpiazu and Frasch, 1993; Frasch, 1995; Wu et al., 1995; Park et al., 1996; Azpiazu et al., 1996; Lockwood and Bodmer, 2002). As pnr and ush are expressed in both of these germlayers(Fig. 1)(Winick et al., 1993; Heitzler et al., 1996; Calleja et al., 2000; Gajewski et al., 1999; Fosset et al., 2000; Herranz and Morata,2001), it is possible they are required for heart development in either or both germlayers. We already showed that mesodermal overexpression of pnr and ush alters tin expression, demonstrating that these two genes can influence heart development within the mesoderm. In order to test for a specific germlayer requirement directly, we overexpressed these genes in the respective mutant background either in the mesoderm or the ectoderm (see Materials and Methods). We then assayed for restoration (i.e. rescue) of tin expression within the heart-forming mesoderm of these rescue embryos. When pnr or ush is rescued in the mesoderm specifically, 57% and 52% of the embryos, respectively, show cardiac-specific tin expression that is restored close to wild-type levels(Fig. 4A-C). The ubiquitous da-Gal4 driver confers similar levels of rescue(Fig. 4A), which suggests that forced mesodermal expression of these genes is sufficient to initiate proper heart formation. However, this interpretation does not exclude the possibility that ectodermal pnr and ush expression is also a contributor to heart-specific tin expression. Ectoderm-specific rescue of pnr, using ZKr-Gal4 (Frasch,1995) (see Materials and Methods), restores a considerable amount of tin expression in a small but significant number of pnrmutant embryos (Fig. 4A,D),suggesting that pnr activity in the ectoderm can also contribute to cardiogenesis. Because the level of ectodermal rescue is low, we cannot rule out that this ectodermal driver also allows low levels of mesodermal expression, which may be sufficient to achieve considerable rescue. Nevertheless, these results are consistent with the hypothesis that pnr and ush are mediators of an ectodermal cardiogenic signal within both germlayers.

To test the idea further that pnr is normally required in both germ layers, we interfered with pnr activity by expressing a dominant-negative form of pnr (pnrEnR, see Materials and Methods) in the ectoderm or the mesoderm. When pnrEnR is expressed throughout the mesoderm, a dramatic decrease in tin expression is observed (Fig. 4E). When pnrEnR is expressed in the ectoderm using the 69B-Gal4 driver, which is broader but slightly less ectoderm-specific than ZKr-Gal4, early tin expression is undiminished (data not shown), but at later stages a moderate decrease is observed (Fig. 4F, typical of 10% of the embryos). Similar observations were obtained when assayed for Eve staining(Fig. 4G-L), except that Eve is affected less than tin at early stages, similar to pnr null mutants (Fig. 4E,H, compare with Fig. 2D,P). The fact that interference with pnr function predominantly in the ectoderm leads to a reduction in cardiac differentiation indicates strongly that pnrfunction is normally required not only in the mesoderm, but also in the ectodermal germlayer in order to achieve wild-type levels of cardiogenesis. As pnr codes for a transcription factor, its ectodermal role in heart formation is probably indirect, requiring induction across germlayers.

As previously described, the expression patterns of pnr and ush are initially broadly induced by dpp in the dorsal ectoderm (Winick et al., 1993; Ashe et al., 2000). Later,these expression patterns are further refined but continue to overlap spatially with ectodermal dpp (as well as wg) expression,but their genetic relationship at later stages is not known. The maintenance of dpp expression in a thin dorsal ectodermal stripe(Fig. 5A) is thought to be essential for controlling dorsal morphogenesis and closure by regulating a number of target genes (Winick et al.,1993; Heitzler et al.,1996; Calleja et al.,2000; Herranz and Morata,2001). As pnr also exhibits an ectodermal requirement for heart development, we hypothesized that pnr may be needed for maintaining late dpp expression(Herranz and Morata, 2001),which in turn contributes to the progression of cardiogenesis(Lockwood and Bodmer, 2002). A late role for dpp in maintaining cardiogenesis has been difficult to ascertain, because the stage 11 dorsal stripe expression could not be abolished easily or selectively. When we examined dpp expression in pnr and ush mutant embryos, we find that dorsal ectodermal stripe expression of dpp is present at stage 11, but is progressively reduced after germband retraction (Fig. 5). This finding is consistent with the idea that ectodermal pnr/ush function acts via maintenance of dpp in a dorsal stripe overlaying the forming heart. Thus, pnr/ush is likely to play a crucial role in a crossregulatory network of the cardiogenic function of dpp: first by mediating the early Dpp signal within the mesoderm and later by maintaining ectodermal dpp expression.

To test if the immediate target genes of the cardiogenic Dpp signal transduction pathway are activated within the mesoderm or in the ectoderm or both, we examined the cardiogenic role of brk, a transcriptional repressor of dpp targets(Sivasankaran et al., 2000; Kirkpatrick et al., 2001; Rushlow et al., 2001; Saller and Bienz, 2001; Zhang et al., 2001). When brk is expressed throughout the mesoderm, there is a considerable reduction in cardiogenesis as assayed by tin and eveexpression (Fig. 6A-D), similar to what is observed in pnr mutants and mesodermal expression of pnrEnR. This suggests that mesodermal pnr maybe a primary target of the cardiogenic Dpp signal. Moreover, brk overexpression in the ectoderm with the early onset ZKr-Gal4 driver selectively reduces cardiac-specific tin expression as early as stage 11(Fig. 6E, compare with Fig. 2B). This is unlikely to be due solely to an elimination of ectodermal pnr expression, which causes a weaker and later-onset reduction of cardiac tin(Fig. 4F). Therefore, we examined if dpp expression itself is inhibited by ZKr-Gal4>UAS-brk. Indeed, dpp expression is significantly reduced within the ZKr-Gal4 expression domain already at stage 11(Fig. 6G), which is much earlier than is the case in pnr mutants, suggesting that dppis a direct target of brk. By contrast, when brk is overexpressed with the later onset ectodermal driver, 69B-Gal4, tinexpression appears to be reduced later and only slightly(Fig. 6F), accompanied by a weak and late reduction of dpp expression(Fig. 6H). These data suggest that the Dpp pathway directly affects targets in the mesoderm, and that pnr/ush (along with tin) are likely mediators and effectors of dpp signaling that is necessary for proper heart development(illustrated in Fig. 7).

It has been previously reported that pnr promotes myocardial cell fates and opposes that of the Eve pericardial cells(Gajewski et al., 1999),whereas the function of ush was to limit the development of both by interfering with dorsal spreading of the ventrally invaginated mesoderm(Fossett et al., 2000). In this study, we present evidence that pnr and ush are part of the initiation and maintenance process of cardiogenesis, respectively, and that they are required for the formation of both myocardial and pericardial cell fates. In pnr mutants, tin expression is normal until early stage 11, but by mid- to late-stage 11 becomes dramatically reduced along the dorsal mesodermal edge, where the heart precursors normally form,indicating a failure to specify cardiac mesoderm. By contrast, cardiac tin expression in ush mutants appears normal initially, and only later begins to exhibit a pronounced decrease in tin expression,considerably after dorsal mesodermal migration is complete, unlike what was observed in migration-defective heartless mutants(Gisselbrecht et al., 1996),indicating ush is involved in maintaining cardiac differentiation.

Even though tin expression is dramatically reduced in early and late stage pnr and ush mutants, respectively, cardiac subtype-specific gene expression is not affected equally. In stage 13 embryos, eve-, lbe- and svp-expressing cells were more affected in pnr than in ush mutants, presumably because the reduction in cardiac tin expression occurs earlier in pnrthan in ush mutants. The largest difference in susceptibility to pnr relative to ush was observed with lbeexpression. Of the three cardiac cell type-specific markers, Eve is the least sensitive to pnr loss-of-function. We speculate that this difference may be due to direct versus indirect (via tin) regulation of the relevant enhancers by pnr.

Ectopic ventrolateral tin expression is observed when pnris overexpressed in the mesoderm. This expansion in tin expression is reminiscent to what is observed when dpp is expressed throughout the mesoderm (Lockwood and Bodmer,2002). This raises the question of whether pnr directly regulates tin expression, or indirectly through dpp (or both). As shown previously, global overexpression of pnr causes ectopic dpp expression in the ectoderm(Herranz and Morata, 2001). However, we find that pan-mesodermal overexpression of pnr does not cause an expansion of dpp expression in the mesoderm or the ectoderm(data not shown). This suggests that pnr must be able to activate the expression of tin either by itself or with some other factors,excluding dpp, in this overexpression assay. This does exclude the possibility that normally pnr and ectodermal Dpp signaling could act in parallel to activate tin expression in the heart primordial (see below). The ability of pnr to activate tin is likely to be direct, as a heart-specific enhancer of tin(Venkatesh et al., 2000)contains several consensus Gata sites (M. Liu and R.B., unpublished). As shown by transcription assays (Gajewski et al.,2001), pnr is also a likely direct target of tin, suggesting that they both contribute to maintaining each other's expression. Both tin and pnr have been shown to be targets of Dpp signaling at stage 9/10 (Xu et al., 1998; Ashe et al.,2000). We propose that dpp is necessary again at stage 11 to activate and maintain pnr and tin expression in the cardiogenic region of the mesoderm (Fig. 7). First, pnr is activated with the help of early stage 11 tin, which is expressed broadly throughout the dorsal mesoderm,and dpp, which is expressed in a narrow dorsal ectodermal stripe. Then, at mid-stage 11, tin is restricted to the cardiogenic region with the help of mesodermal pnr as well as continuous ectodermal Dpp signaling. Once both are activated in the cardiogenic mesoderm, they are likely to contribute to the maintenance of each other's expression, probably aided again, but only moderately, by ectodermal Dpp signaling. This interpretation is consistent with mesodermal versus ectodermal expression of dominant-negative pnrEnR (Fig. 4) and the dpp target repressor encoded by brk(Fig. 6). They are both equally effective in reducing cardiac-specific tin when expressed in the mesoderm, but ectodermal repression is more effective when dorsal-stripe dpp at stage 11 is also affected (as in the case of ZKr-Gal4>UAS-brk shown in Fig. 6G, but not with ZKr-Gal4>UAS-pnrEnR, data not shown).

Mesodermal overexpression of ush and co-overexpression with pnr results in a decrease in the amount of cardiac-specific tin expression, suggesting that ush may not only be required along with pnr for heart development, but also play an inhibitory role. To test this hypothesis further, we overexpressed pnrD4, an allele that abolishes Ush binding to Pnr, and found not only ectopic tin expression at early stages of cardiogenesis, but also undiminished and even increased levels of expression at later stages. A similar phenotype was observed when both pnrD4 and ush were expressed throughout the mesoderm, suggesting that ush plays an anti-cardiogenic role by antagonizing the activity of wild-type Pnr, but not that of PnrD4. It would be interesting to see if pan-mesodermal overexpression of wild-type pnr in a ush mutant background results in ectopic tin expression similar to pnrD4, or if a minimal amount of ush activity is required to maintain normal and ectopic tin expression even with forced pnr expression. Interestingly, overexpression of both pnr and tin together in the mesoderm also causes a pnrD4-like phenotype, as assayed with Hand expression, suggesting that pnr and tincollaborate during initiation and subsequent differentiation of the heart progenitors.

Although in vitro the Ush-related FOG factors are primarily known for their role as transcriptional repressors(Svensson et al., 1999; Tevosian et al., 1999), they apparently can also function as co-activators: Fog2 can synergistically activate or repress the transcriptional activity of Gata4, depending on the(cardiac) promoter and cell line used (Lu et al., 1999), and FOG-1 can cooperate with Gata1 to transactivate NF-E2, an erythroid cell-specific promoter(Tsang et al., 1998). Moreover, the ventricular hypoplasia and other heart defects observed in Fog2-deficient mice suggest a deficit rather than an excess in heart development (Tevosian et al.,2000; Svensson et al.,2000). In addition, mice with an equivalent mutation to PnrD4 knocked into the Gata4 locus, thus eliminating binding to Fog2, exhibit in many ways a similar phenotype to Fog2-deficient mice(Crispino et al., 2001). These data are consistent with the idea that Fog2 is normally involved in promoting rather than antagonizing cardiogenesis, similar to what we have found with our genetic studies during Drosophila heart development.

The dual role of Ush suggests that the amount of Ush may be crucial for whether it exerts its function as a an activator or repressor, perhaps by binding to different sets of co-factors in a concentration-dependent manner. Alternatively, the mode of transcriptional regulation by Ush could be stage-dependent: at stage 11, Pnr and Ush cooperate as transcriptional activators in initiating cardiac-specific tin expression and heart development, but later Ush becomes a repressor to limit the transcriptional activation of tin by Pnr

pnr and ush are initially broadly expressed in the dorsal ectoderm of the early embryo, but by germband retraction the ectodermal expression of pnr is confined to a narrow stripe of cells along the border of the amnioserosa, which overlaps with the thin dorsal dppstripe (Fig. 1D). The early ectodermal expression of ush is restricted to the presumptive amnioserosa, and by germband extension, ush also overlaps with the dorsalmost region of the ectoderm (Fossett et al., 2000; Herranz and Morata, 2001). These patterns of expression suggest that pnr and ush may be acting in both germ layers. Our genetic data, including germ layer-specific expression of wild-type and dominant-negative pnr constructs, as well as germ layer-specific rescue experiments suggest strongly that pnr and ushfunction is not only needed in the mesoderm, but also in the ectoderm for heart formation (see model in Fig. 7). The ectodermal requirement for pnr and ushin heart development is probably achieved via the maintenance of dppexpression, as dorsal stripe dpp expression diminishes in pnr and ush mutants and ectodermal interference with pnr, ush and/or dpp-signaling function compromises the normal progression of heart development.

We thank Ken Cadigan, Tom Kerppola and Jim Skeath for their helpful discussions. We also thank Manfred Frasch, Marc Haenlin, Anthea Letsou, Marek Mlodzik, Adrian Moore, Philippe Ramain, Christine Rushlow and the Bloomington Drosophila Stock Center for providing fly stocks and reagents essential for this work. S.K. is supported by a predoctoral fellowship from the American Heart Association. This work was funded by grants from the National Institutes of Health (NHLBI) and the American Heart Association to R.B.