pannier and pointedP2 act sequentially to regulate Drosophila heart development

Comparative studies between vertebrates and Drosophila highlight a significant conservation in the embryology and molecular regulation of heart development (reviewed by Bodmer and Frasch,1999; Cripps and Olson,2002). For example, in Drosophila and vertebrates the heart develops from those mesodermal cells that migrate most distally from the original point of invagination during gastrulation. In addition, the mature Drosophila heart is a simple linear tube and resembles the primitive heart tube of vertebrates prior to the processes of looping and septation that ultimately create a multi-chambered heart. Furthermore, molecular studies indicate that similar gene cassettes govern heart formation in Drosophila and vertebrates. The homeodomain gene tinman(tin) and the GATA transcription factor pannier(pnr) are necessary for heart formation in Drosophila. Similarly, transcription factors of the Tin and GATA families are also essential for vertebrate heart formation. In both Drosophila and vertebrates, the BMP signaling pathway controls the proper establishment of the expression profiles of Tin and GATA family members in the heart primordium. Thus, Drosophila provides an excellent model for early heart development. Despite this, we know little about the genetic regulatory mechanisms that specify the individual cell types of the Drosophilaheart and we lack a complete knowledge of the lineal relationships of all heart cells.

The Drosophila heart is composed of two cell types (reviewed by Bodmer and Frasch, 1999). Cardioblasts express muscle specific proteins, coalesce to form the linear heart tube and are the contractile cells of the heart. Pericardial cells are loosely associated with and flank the cardioblasts; these cells do not express muscle specific proteins and are thought to filter and detoxify the blood or hemolymph of the fly. Cardioblasts and pericardial cells develop from closely interspersed precursor cells that develop in the dorsalmost region of the mesoderm, termed the cardiac mesoderm. After precursor divisions, heart cells on both sides of the embryo align themselves into two rows of cells, with pericardial cells being displaced slightly ventral and interior to the tightly aligned row of cardioblasts at the dorsalmost extent of the mesoderm. As dorsal closure occurs, the bilaterally symmetric rows of heart cells move toward each other and the two rows of cardioblasts meet at the dorsal midline,align perfectly with one another and coalesce to form a lumen between them. Subsequently, the heart tube becomes divided into two domains: the aorta more anteriorly and the heart proper in the posterior three segments. The heart proper is distinguished from the aorta by a wider bore and the presence at segmental intervals of ostia, the inflow valves of the heart. Here, we collectively refer to the aorta and heart proper as the heart.

Gene expression, cell lineage and morphological studies indicate that distinct subtypes of cardioblasts and pericardial cells populate the heart(see Fig. 7)(Gajewski et al., 2000; Jagla et al., 1997; Lo and Frasch, 2001; Lo et al., 2002; Ward and Skeath, 2000). These studies also distinguish the development and gene expression profiles of heart cells located in the posterior seven segments relative to those found more anteriorly. Each hemisegment of the posterior region contains six cardioblasts and ten pericardial cells. Cardioblasts can be roughly divided into two classes: those that express Svp but not Tin (these are the first two cardioblasts of a hemisegment); and those that express Tin but not Svp, the four remaining cardioblasts of a hemisegment. The lineage of these cardioblasts is known and shown in Fig. 7. Pericardial cells can be divided into three classes: Eve-, Tin-and Odd-positive pericardial cells. The relative position of these cells is shown in Fig. 7. The cell lineage of the Odd-positive pericardial cells is known; however, the lineage of Eve- and Tin-positive pericardial cells remains unclear.

In contrast to the posterior region, all 12 cardioblasts that develop anterior to this domain express Tin but not Svp. In this region, Eve- and Tin-positive pericardial cells develop normally, while Odd-positive pericardial cells are replaced by Odd-expressing lymph gland cells. The differences in gene expression between heart cell types in the same region as well as between different anteroposterior (AP) regions suggest functional specialization of cardioblasts and pericardial cells both within the same region and between different AP regions. Consistent with this model, the Svp-cardioblasts in the posterior three heart segments form the ostia of the larval heart (Molina and Cripps,2001).

Genetic studies have identified the Nk2 type homeodomain protein Tin as a key regulator of heart development (reviewed by Bodmer and Frasch, 1999; Cripps and Olson, 2002). tin expression is the earliest known marker of the cardiac mesoderm and loss-of-function mutations in tin result in the complete absence of all heart cells, as well as all other dorsal mesodermal derivatives. In addition to tin, heart cell development absolutely requires the function of the AP patterning genes wingless and sloppy-paired. Despite the identification of a number of genes required to promote the development of all heart cells, very few genes have been identified that regulate the decision of cells to choose between the cardioblast and pericardial cell fate. One such gene appears to be the GATA zinc-finger transcription factor pnr(Ramain et al., 1993). Loss of pnr function results in a significant loss of cardioblasts and an apparent increase in at least one class of pericardial cells, the Eve-positive pericardial cells (Gajewski et al.,1999). Other genes are likely to act with or in opposition to pnr to regulate the decision of cells to acquire the cardioblast or pericardial cell fate.

The pointed (pnt) locus encodes two protein isoforms,both of which act as transcriptional effectors of the Ras/MAP-kinase pathway(Brunner et al., 1994; Klaes et al., 1994; Klambt, 1993; O'Neill et al., 1994). The two Pnt isoforms, PntP1 and PntP2, are members of the ETS family of transcription factors and arise due to alternative use of two promoters separated by roughly 50 kb. PntP1 and PntP2 contain unique domains at their N terminus but share the identical stretch of 394 amino acids at their C terminus within which resides the ETS DNA-binding domain. The DNA-binding properties of these proteins appear identical; however, PntP1 and PntP2 exhibit crucial differences in their functional properties and transcriptional regulation. PntP1 is a constitutive transcriptional activator and Ras/MAP-kinase pathway activity induces PntP1 transcription(Gabay et al., 1996). By contrast, PntP2 is not a constitutive transcriptional activator and pntP2 transcription appears to be independent of Ras/MAP-kinase activity. Nonetheless, PntP2 activity depends on Ras/MAP kinase activity as Ras/MAP kinase-mediated phosphorylation of PntP2 at Thr151 turns it into a potent transcriptional activator (Brunner et al., 1994; O'Neill et al.,1994). Although PntP1 and activated PntP2 regulate the development of many different cell types and tissues in Drosophila, a role for pnt function in cardioblast development has not been identified.

We present evidence that pnt plays a region specific role in regulating cardioblast and pericardial cell development. Loss of pntfunction results in an approximate twofold increase in cardioblasts and an almost commensurate decrease in pericardial cells. This increase in cardioblasts arises largely from a specific increase in Svp-positive cardioblasts and is restricted to the posterior seven heart segments where Svp-positive cardioblasts normally develop. We demonstrate that this effect of pnt is carried out primarily if not exclusively by the PntP2 isoform,and that in this context PntP2 may act independently of Ras/MAP kinase pathway activity. Contrary to a prior study, we find that pnr acts early in mesoderm development to promote the development of the cardiac mesoderm and thus the development of all heart cells. Phenotypic analyses of pnr pnt double mutant embryos suggest a model whereby pnr acts before pnt to promote the formation of the cardiac mesoderm and that pnt acts subsequently within this domain to distinguish between cardioblast and pericardial cell fates. In addition, we present the completion of the cell lineage of all heart cells. These pedigree analyses identify a clear distinction between the lineage of anterior cardioblast and those that develop in the posterior seven segments. The transition point between these heart cell lineages correlates perfectly with the region specific effect of pnt on heart development. These results suggest independent genetic control of heart cell development in the anteriormost region of the heart relative to the posterior seven segments.

Wild-type patterns of gene expression were examined in Oregon R embryos. Fly lines used include pnt^S012309, pnt², pnt^RM254, pnt^▵88, pnt^RR112, pnr¹, pnr^VX6, spi^IIA, Star^IIN, rhomboid^del-1, aop¹, heartless^AB42, aos^▵7, vein^Ry, vein^γ3, heartbroken^ems6 and svp^H162. svp^H162 is an enhancer trap insert in svp and is referred to as Svp-lacZ (Mlodzik et al.,1990). Gene misexpression was achieved with the Gal4-UAS system and the following lines: Twi-GAL4, UAS-DNEgfr, UAS-DN-heartless, UAS-DNras,UAS-pntP1 and UAS-pntP2. Standard genetic crosses were used to create the multiply mutant fly lines noted in the text.

Random lacZ-expressing clones were created using the FLP/FRT lineage tracing system as described previously(Ward and Skeath, 2000) with the following modifications. Three to four-hour-old embryos of the appropriate genotype were heat-shocked for 20 minutes at 33°C to induce flprecombinase, placed at 18°C and aged until stage 15-16, at which point they were fixed and stained.

Amino acids 1-230 of PntP1 were cloned into pET (Novagen) for protein expression and purification. This protein domain is unique to PntP1. This antigen was used to immunize rabbits at Pocono Rabbit Farm. The PntP1 antibody is specific for PntP1 because it detects a protein expressed in a pattern identical to pntP1 RNA and because the antibody does not detect antigen in embryos that delete the pntP1-specific exons and downstream exons of the pnt locus.

Single- and double-label immunohistochemistry analyses were performed as described previously (Skeath,1998). We used the following antibodies at the indicated dilutions: mouse anti-Zfh1 (1:1000) (Lai et al., 1991); rabbit anti-Mef2 (1:1000)(Lilly et al., 1995); rabbit anti-Eve (1:2000) (Frasch et al.,1987); mouse anti-β-gal (1:2000; Promega); rabbit anti-β-gal (1:2000; Jackson); rabbit anti-Tin (1:500)(Azpiazu and Frasch, 1993);rabbit anti-Pnr (1:400) (Herranz and Morata, 2001); rabbit anti-Odd (1:500)(Ward and Skeath, 2000);rabbit anti-PntP1 (1:500).

RNAi was prepared as described previously(Kennerdell and Carthew,1998). We used dsRNA probes specific for pntP1 or pntP2 to target each isoform independently. For pntP1 we made dsRNA corresponding to nucleotides 1-690 of the pntP1-coding region. This region encodes for the entire pntP1-specific domain. For pntP2, we made dsRNA for corresponding to nucleotides 1-906 of the pntP2 coding region. This region encodes ∼90% of the pntP2-specific domain. dsRNA was injected into the posterior region of pre-cellular blastoderm embryos at a concentration of 2 μg/μl, and the embryos were allowed to develop until stage 15 to 16 at which point embryos were collected and fixed for immunohistochemistry.

We identified the molecular lesions in pnt², pnt^RR112 and pnr¹ by PCR-based sequencing of the entire coding region and intron/exon boundaries of the appropriate locus from genomic DNA obtained from each mutant background. pnt² contains G to A conversion at base 2653 of the pntP1 cDNA (base 2667 of the pntP2 cDNA)(Klambt, 1993). This mutation converts the tryptophan (W) at amino acid 536 of PntP1 (amino acid 631 of PntP2) to a premature stop codon, truncating both PntP1 and PntP2 roughly one third of the way through the shared ETS-DNA-binding domain. pnt^RR112 contains a G to A conversion in the splice donor site of exon IV in pntP2. This lesion converts the GT donor site to AT, and is expected to disrupt splicing of pntP2 but not pntP1 because this exon is specific for pntP2. pnr¹ contains a G to A conversion at nucleotide 1034. This mutation converts a W to a premature stop codon, truncating the pnrmidway through the first zinc finger.

Pioneering work in C. elegans established the importance of elucidating cell lineages to obtain a thorough understanding of animal development (Sulston and Horvitz,1977). Our prior work has established the cell lineage of 10 out of the 16 heart cells that arise in each hemisegment of the posterior seven heart segments (Ward and Skeath,2000). To define the lineage of the remaining heart cells, we used the FLP/FRT lineage tracing system (Struhl and Basler, 1993) to determine the lineal relationship of the two Eve-positive pericardial cells and four Tin-positive pericardial cells that arise in each hemisegment. This system creates random clones marked by tau-lacZ reporter gene activity. Briefly, we induced clones during stage 8 just as the pan-mesodermal divisions are being completed(Borkowski et al., 1995). This allowed us to induce clones in mesodermal cells prior to the emergence of heart precursors. To identify the lineage of Eve-positive pericardial cells we double labeled embryos for β-galactosidase to mark clones, and Eve to identify Eve-positive pericardial cells. To identify the lineage of Tin-positive pericardial cells we double labeled embryos forβ-galactosidase to mark clones, and Tin to identify Tin-positive pericardial cells. In addition to the Tin-positive pericardial cells, Tin labels cardioblasts and Eve-positive pericardial cells. However, based on position and morphology one can unambiguously distinguish Tin-positive pericardial cells from Tin-positive cardioblasts and Eve-positive pericardial cells.

We identified eighteen clones that contained at least one Eve-positive pericardial cell. Eleven of these clones (61.1%) consisted solely of two Eve-positive pericardial cells (Fig. 1A), six clones (33.3%) consisted of two Eve-positive pericardial cells and one or two nearby heart or other mesodermal cells, and one clone(5.6%) consisted of a single Eve-positive pericardial cell. Thus, when we observe one Eve-positive pericardial cell within a clone of two or more cells a second Eve-positive pericardial cell always exists within this clone. These data demonstrate that the two Eve-positive pericardial cells within a hemisegment are siblings and arise from an Eve-positive pericardial cell precursor.

Our data on the lineage of Eve-positive pericardial cells contrasts with a prior lineage study (Park et al.,1998). This study also used the FLP/FRT lineage tracing system but determined that Eve-positive pericardial cells are not obligate siblings. We attribute the difference in our results to the different stages during which clones were induced in the two studies. We induced clones during stage 8, well before the division of Eve-expressing mesodermal cells during late stage 10. However, Park et al. (Park et al.,1998) induced clones around stage 10, in close proximity to the time during which Eve-positive mesodermal cells divide. We speculate that the timing of flp induction in the prior study was too late to identify an obligate sibling relationship between Eve-positive pericardial cells. Our results agree with those of Carmena et al.(Carmena et al., 1998) who argue for a sibling relationship between Eve-positive pericardial cells based on gene expression studies.

Tin-positive pericardial cell clones fall into two classes: those that contained two Tin-positive pericardial cells (n=23), and those that contained one Tin-positive pericardial cell and one cardioblast(n=18). These two classes of clones arise in mutually exclusive regions of the heart. Clones that contain two Tin-positive pericardial cells arise in the posterior seven segments of the heart (we refer to this region as the posterior heart domain), whereas clones that contain one Tin-positive pericardial cell and one cardioblast arise anterior to this domain (we refer to this region as the anterior heart domain). The point of demarcation between these clonal types coincides precisely with the location of the first pair of Svp-positive cardioblasts (see below). These data demonstrate that heart cells exhibit distinct cell lineages as a function of position along the anteroposterior axis.

We identified a total of 24 Tin-positive pericardial cell clones in the posterior heart domain. Fifteen of these clones (62.5%) consisted solely of two Tin-positive pericardial cells (Fig. 1B), eight clones (33.3%) consisted of two Tin-positive pericardial cells and two nearby mesodermal cells, and one clone (4.2%)consisted of a single Tin-positive pericardial cell. Thus, when we observe one Tin-positive pericardial cell within a clone of two or more cells, a second Tin-positive pericardial cell always exists within this clone. These data indicate that the four Tin-positive pericardial cells found in each hemisegment of the posterior domain arise from two Tin-positive pericardial cell precursors. Our inability to identify any clones that contain four Tin-positive pericardial cells indicates that adjacent Tin-positive pericardial cell precursors are unlikely to share a common lineage.

We identified 18 Tin-positive pericardial cell clones in the anterior heart domain. All 18 clones consisted of one Tin-positive pericardial cell and one cardioblast (Fig. 1C). These data indicate that within this region Tin-positive pericardial cells arise from bi-potent heart precursors, each of which produces one Tin-positive pericardial cell and one cardioblast. These data also demonstrate that cardioblasts and Tin-positive pericardial cells in the anterior heart domain develop via a different cell lineage than cardioblasts and Tin-positive pericardial cells that develop in the posterior domain.

The analysis of ten additional cardioblast clones in the anterior heart domain support a distinct cell lineage for anterior versus posterior cardioblasts. Nine clones consisted of one cardioblast and one non-Tin-expressing pericardial cell (Fig. 1D), whereas a single clone consisted of two cardioblasts. Thus,most, if not all, anterior domain cardioblasts share a sibling relationship with a pericardial cell. In addition, all anterior domain cardioblasts exhibit cell lineages distinct from posterior domain cardioblasts. Together with the lineage data on Tin-positive pericardial cells, these results support the idea that cardioblasts and Tin-positive pericardial cells in the anterior heart domain carry out distinct functions from those found in the posterior heart domain.

Interestingly, the lineage of the twelve cardioblasts in the anterior heart domain appears fixed with respect to whether they share a sibling relationship with a Tin-positive or Tin-negative pericardial cell. We numbered these cardioblasts 1-12 from anterior to posterior with cardioblast 12 being immediately anterior to the first Svp-positive cardioblast. We identified four clones that contained cardioblast 12 and in each clone this cardioblast shared a sibling relationship with a Tin-negative pericardial cell. By contrast,cardioblasts 10 and 11 each share a sibling relationship with a Tin-positive pericardial cell (n=3/3 and 5/5 clones, respectively). We have not obtained multiple clones for all twelve cardioblasts; nonetheless, these data suggest a fixed relationship between the position of a cardioblast and whether its sibling pericardial cell expresses Tin. We speculate that the differences in gene expression between different pairs of sibling cardioblasts and pericardial cells in the anterior domain may reflect functional differences between such pairs of heart cells.

We identified pnt as an inhibitor of cardioblast development in a screen for mutations that affect cardioblast and/or pericardial cell development. To identify genes that regulate heart development we screened∼2000 third chromosomal lethal P element lines obtained from the Hungarian P element Stock Collection for defects in the expression of Mef2 a protein expressed in all cardioblasts and Eve. We uncovered two P element mutations that cause an approximate twofold increase in cardioblasts(Fig. 2; X. Tian and J.B.S.,unpublished). One of these P elements [l(3)S012309] maps to cytological position 94F1-3 and was known to be an allele of pnt(FlyBase, 2003). To verify that lesions in pnt result in the formation of ectopic cardioblasts, we assayed the phenotype of five additional pnt alleles. Although the severity of the phenotype varies for each pnt allele, all alleles display a significant increase in cardioblast number relative to wild-type embryos (Fig. 2). With respect to the excess cardioblast phenotype, we can group these alleles into the following allelic series: pnt^S012309, pnt² > pnt^RR112, pnt^RM254 > pnt^▵88, pnt⁰⁷⁸²⁵. The presence of excess cardioblasts in embryos homozygous mutant for each pnt allele indicates that pntnormally functions in heart development to repress cardioblast development.

Two distinct types of cardioblasts exist in the heart: Svp-positive cardioblasts and Tin-positive cardioblasts. To determine whether mutations in pnt result in an increase in one type of cardioblast or a general increase in all cardioblasts we assayed the relative percentage of the two types of cardioblasts in wild-type and pnt embryos. In wild-type embryos, a total of 52 cardioblasts develop per embryo side: 14 of these cardioblasts express Svp-lacZ and 38 express Tin but not Svp. In pnt^S012309 embryos an average of 98.5 cardioblasts develop per embryo side (n=10): 58.6 express Svp (a 318% increase relative to wild-type) and 41.3 express Tin but lack Svp (a 8.6% increase relative to wild-type). In pnt^S012309/pnt² embryos, an average of 99 cardioblasts develop per embryo side (n=5): 53 express Svp-lacZ (a 278% increase) and 46 express Tin but lack Svp (a 21%increase). Thus, the pnt excess cardioblast phenotype arises predominantly from an increase in Svp-positive cardioblasts.

Our analysis of heart development in pnt embryos indicated that the effect of pnt on cardioblast development is region specific. In wild-type embryos 12 cardioblasts develop anterior to the first pair of Svp-positive cardioblasts on each side of the embryo. As detailed in our lineage studies, these cardioblasts define the anterior heart domain. Interestingly, cardioblast development in the anterior domain is essentially normal in pnt mutant embryos (Fig. 3). In pnt^S012309 and in pnt^S012309/pnt² embryos, an average of 10.8(n=10) and 12.4 (n=5) cardioblasts develop anterior to the first pair of Svp-lacZ-positive cardioblasts. Thus, in pntembryos the ectopic cardioblast phenotype is restricted to the region of the heart, the posterior domain, that normally contains endogenous Svp-lacZ-positive cardioblasts.

To investigate whether pnt promotes pericardial cell development,we followed pericardial cell and lymph gland development in pntembryos. In wild-type embryos, we detect 60.1 pericardial cells per embryo side (n=17) while in pnt^S012309 embryos we observe 31.4 pericardial cells per side (n=18; Fig. 4). All other pntalleles exhibit significant reductions in pericardial cell number (not shown). We observe no effect of pnt on pericardial cell or lymph gland development in the anterior heart domain. Together with the data on cardioblast development, these results reveal that loss of pnt causes reciprocal effects on cardioblast versus pericardial cell development and that these effects are restricted to the posterior heart domain.

The reciprocal effect pnt has on the development of cardioblasts versus pericardial cells suggests that pnt may normally function during heart development to repress the cardioblast fate in cells destined to acquire the pericardial cell fate. To test this model, we carefully followed cardioblast development in pnt embryos. In wild-type embryos, all pericardial cells become detectable by late stage 12/early stage 13 in a position just ventral and slightly interior to cardioblasts(Fig. 4). At this stage and all subsequent stages, cardioblasts are clearly distinguished as a single line of cells at the dorsal most extreme of the mesoderm. In pnt embryos by late stage 12/early stage 13 we observe a clear excess of cardioblasts many of which are found in locations normally occupied by pericardial cells(Fig. 4). These results are consistent with a model in which pnt represses cardioblast development in presumptive pericardial cells thereby promoting pericardial cell development.

Owing to the specific increase in Svp-lacZ cardioblasts in pnt embryos, we also performed a detailed analysis of the development of Svp-lacZ cardioblasts and pericardial cells. In wild-type embryos,two Svp-lacZ heart precursors arise in each hemisegment during stage 11 (Ward and Skeath, 2000). Each precursor divides during stage 12 to yield one Svp-lacZcardioblast and one Svp-lacZ pericardial cell. In pntembryos the formation and division of each endogenous Svp-lacZ heart precursor is normal and wild-type numbers of Svp-lacZ pericardial cells persist throughout embryogenesis (data not shown). Thus, Svp-lacZ pericardial cells appear to develop normally in pntembryos. However, we observe ectopic Svp-lacZ cardioblasts during late stage 12/early stage 13 in pnt embryos and these cells are found in locations normally occupied by pericardial cells(Fig. 4; not shown). These results, together with those detailed above, suggest that pntnormally functions in presumptive non-Svp-lacZ-expressing pericardial cells to repress the development of the Svp-lacZ cardioblast fate.

Our quantification of cardioblasts and pericardial cells in pntembryos suggests the excess cardioblast phenotype does not arise solely from a conversion of pericardial cells into cardioblasts, as we observe a net loss of∼30 pericardial cells and a net gain of ∼50 cardioblasts per embryo side. To investigate whether loss of pnt affects the proliferative potential of heart cells, we created and analyzed cardioblast clones in pnt embryos. We identified 124 clones in pnt embryos that contained at least one cardioblast. 97 clones (78%) consisted of either two cardioblasts or one cardioblast and one pericardial cell; 21 (17%) clones consisted of two cardioblasts and one or two pericardial cells whereas two(1.6%) consisted of one cardioblast. These clone types as well as their frequencies are similar to that observed for cardioblast clones in wild-type embryos (Ward and Skeath,2000) (data not shown). However, in addition to these clones, we identified four (3.2%) that consisted of between six and 12 cardioblasts. We have never observed clones of more than four cardioblasts in wild-type embryos(n>200 clones). These data suggest that loss of pnt leads to a slight but perceptible increase in the proliferative capability of cardioblast precursors. However, the weak increase in cardioblast proliferation and the apparent conversion of pericardial cells into cardioblasts still appear insufficient to account for the approximate twofold increase in cardioblasts in pnt embryos. We hypothesize that loss of pnt also causes other dorsal mesodermal cells to acquire the cardioblast fate inappropriately. Consistent with this, we observe loss of specific dorsal muscles in pnt embryos.

Through the use of alternative promoters, the pnt locus encodes two distinct protein isoforms: PntP1 and PntP2. Both isoforms act as effectors of the Ras/MAP kinase pathway in multiple developmental contexts(Brunner et al., 1994; Klambt, 1993; O'Neill et al., 1994). This raises the possibility that the role of pnt during heart development is mediated through Ras/MAP kinase activity. Thus, we examined whether loss or reduction in the function of different members of the Ras/MAP kinase pathway also increased cardioblast number. We assayed cardioblast development in homozygous embryos singly mutant for spitz, Star, rhomboid, heartlessand heartbroken. We also assayed cardioblast number in embryos in which we expressed dominant-negative forms of ras as well as the EGF- and FGF-receptors specifically in the mesoderm to reduce the activity of these genes in this tissue (see Materials and Methods). In all genetic backgrounds tested, we never observed an increase in cardioblast number. For the experiments involving dominant-negative constructs, we verified dominant-negative activity of the expressed protein by assaying ras-dependent developmental events that occur in the mesoderm prior to the role of pnt in cardioblast development. In all cases, the ras-dependent developmental events were perturbed (data not shown). Thus, we are confident that our failure to observe an effect on cardioblast number is not simply due to an inability of the dominant-negative proteins to inhibit the function of the targeted proteins in a timely manner. We interpret these results to suggest that pnt may regulate cardioblast development in a Ras-MAP kinase-independent manner.

The presence of two Pnt isoforms raises the question as to whether PntP1 and/or PntP2 carry out the function of the pnt locus during heart development. To address this issue, we used isoform-specific RNAi and isoform-specific rescue of the pnt cardioblast phenotype. We first generated double-stranded RNA probes to the unique 5′ regions of the PntP1 and PntP2 transcripts, and injected these separately into presynctial stage Drosophila embryos. We then labeled all such embryos either for Mef2 (to follow cardioblast development) or Mef2 and PntP1 protein (to follow cardioblast development and PntP1 protein levels). Embryos treated for pntP1 RNAi exhibit severe morphological defects and a complete loss of PntP1 protein expression. In many embryos the extent of the morphological defects preclude a clean analysis of cardioblast development; however, it is possible to score cardioblast number in a subset of these embryos. We only scored cardioblast number in embryos devoid of detectable PntP1 protein. In these embryos, we observe an average of 53.2 cardioblasts per embryo side(n=11), nearly identical to the 52 cardioblasts that develop on each side of wild-type embryos. These results suggest that pntP1 does not play a significant role in the regulation of cardioblast number by pnt.

By contrast, PntP2-RNAi indicates that pntP2 function is necessary to regulate cardioblast number. Embryos treated for PntP2 RNAi exhibit wild-type morphology, a clear excess of cardioblasts and an essentially normal pattern of PntP1 expression (Fig. 5). In these embryos, we observe an average of 80.1 cardioblasts per embryo side (n=22; ranging from 62 to 108 cardioblasts). The most severe pntP2 RNAi phenotypes are as severe as those observed for pnt² or pnt^S012309 embryos. We attribute the variable expressivity of the pntP2 RNAi phenotype to the technique of RNAi as we observe a large variance in expressivity of the RNAi phenotype for all genes we have assayed in this manner.

Allele sequencing of pnt^RR112 supports the idea that the PntP2 isoform is necessary to regulate cardioblast number. pnt^RR112 is an allele of pnt we identified in an EMS screen to identify genes that control CNS and heart development. We find that pnt^RR112 contains a molecular lesion that converts a G to A at the splice donor site immediately 3′ to the PntP2 specific fourth exon (converting the GT site to AT). This lesion should disrupt splicing of pntP2 but not pntP1. Consistent with this, PntP1 expression is normal in pnt^RR112 embryos. pnt^RR112 embryos exhibit a strong excess cardioblast phenotype, indicating that the PntP2 isoform regulates cardioblast development(Fig. 2).

We also addressed the relative roles of pntP1 and pntP2by assaying the effect generalized mesodermal expression of each pntisoform has on cardioblast development in wild-type and pnt embryos. In these experiments, we used the Twist-GAL4 driver line to drive either pntP1 or pntP2 under UAS control throughout the mesoderm of homozygous wild-type or pnt^S012309 embryos. We find that mesodermal expression of pntP2 in pnt embryos is sufficient to rescue to wild-type the pnt cardioblast phenotype(Fig. 5). In addition, we find that mesodermal expression of pntP2 in otherwise wild-type embryos has no effect on cardioblast development (not shown). By contrast, we find that mesodermal expression of pntP1 in wild-type or pntembryos leads to a near complete loss of all cardioblasts and pericardial cells (Fig. 5). This drastic effect of pntP1 on heart development may arise because of an effect of pntP1 overexpression on early steps of mesodermal development prior to heart cell development. This possibility makes interpretation of whether pntP1 can rescue the pnt cardioblast phenotype difficult. Nonetheless, these experiments clearly show that pntP2 is sufficient to rescue the pnt heart phenotype. Together with the RNAi experiments and the phenotypic analysis of a pntP2-specific allele,these results demonstrate that pntP2 is necessary and sufficient for the cardioblast and pericardial cell development, and suggest that pntP1 is irrelevant in this developmental context.

The published heart phenotype of the GATA transcription factor pnris opposite to that of pnt. In pnr mutant embryos, too many pericardial cells and too few cardioblasts are thought to develop(Gajewski et al., 1999). As a first step towards examining the potential regulatory interactions between pnr and pnt, we carried out a detailed analysis of heart development in pnr mutant embryos. We used pnr^VX6, a null allele that contains a small deletion that removes all but the N-terminal nine amino acids of pnr(Ramain et al., 1993), as well as pnr¹, a molecularly uncharacterized allele. In contrast to a prior study, we find a loss of both cardioblasts and pericardial cells in pnr embryos (Fig. 6). We quantified the dorsal mesodermal phenotypes for Eve-positive pericardial cells as well as for all pericardial cells using the pan-pericardial marker Zfh1. In wild-type embryos we observe an average of 22.7 Eve-positive pericardial cells (n=32) and 61.1 Zfh1-positive pericardial cells(n=17) per embryo side. pnr^VX6 embryos exhibit the most severe effect with an average of 9.4 (n=11) and 16.9(n=21) Eve- and Zfh1-positive pericardial cells, respectively. pnr^VX6/pnr¹ embryos exhibit an intermediate phenotype with an average of 16.4 Eve-positive pericardial cells(n=25) and 27.4 Zfh1-positivepericardial cells (n=18), while pnr¹ embryos exhibit the mildest phenotype with an average of 21.2 and 37.4 Eve- (n=10) and Zfh1- (n=11) positive pericardial cells, respectively. We also observed a severe loss of cardioblasts and Odd-positive pericardial cells in these backgrounds although we did not quantify these phenotypes. The loss of cardioblasts and Odd-positive pericardial cells is most severe in pnr^VX6embryos and least severe in pnr¹ embryos where short stretches of cardioblasts are still visible(Fig. 6). These results indicate that pnr normally functions to promote the development of all heart cells.

Based on these results, we used Tin expression to determine the earliest stage at which we could identify a defect in dorsal mesoderm development in pnr embryos. In wild-type embryos, Tin expression becomes restricted to the dorsal mesoderm by stage 10 (Azpiazu and Frasch, 1993). During stage 11, Tin expression resolves to two stripes of cells: a dorsal stripe that labels the cardiac mesoderm and a more ventral undulating stripe that labels the primordia of the visceral mesoderm(Fig. 6)(Azpiazu and Frasch, 1993). In pnr embryos, Tin expression is normal until stage 11. However, during stage 11 Tin expression is lost from the cardiac mesoderm while it is maintained normally in the visceral mesoderm(Fig. 6). These results demonstrate that the earliest manifestation of cardiac mesoderm development is defective in pnr embryos. Furthermore, they suggest that the general lack of heart cells in pnr embryos arises indirectly via a defect in the specification of the cardiac mesoderm.

The above results indicate that pnr¹ is a hypomorphic allele. Sequence analysis identified a single mutation in the pnr¹-coding region that converts a tryptophan residue at amino acid 180 to a premature stop codon roughly halfway through the first zinc finger. Using an antibody specific to epitopes N-terminal to this premature stop codon (Herranz and Morata,2001), we find that the pattern and level of the mutant Pnr¹ protein in homozygous pnr¹ embryos are identical to those of the wild-type Pnr protein. However, while wild-type Pnr protein localizes predominantly to the nucleus, we find that Pnr¹protein localizes predominantly to the cytoplasm (not shown). These data together with the mild pnr¹ phenotype relative to the pnr^VX6 null allele suggest that the truncated Pnr¹ protein retains residual activity.

Our studies on pnr and pnt suggest these genes act in a developmental pathway in which the prior function of pnr to promote cardiac mesoderm formation is required for the subsequent action of pnt to specify between pericardial cell and cardioblast fates. If this model is correct, pnr pnt double mutants should display the pnr phenotype, as neither pericardial cells nor cardioblasts will arise in the absence of cardiac mesoderm. Consistent with this, pnr^VX6 pnt^S012309 embryos lack cardioblasts and pericardial cells, and appear phenotypically indistinguishable with respect to heart development from pnr^VX6 embryos (not shown). To test our model more stringently, we assayed heart development in pnr¹ pnt^S012309 embryos. We used pnr¹ embryos because small regions of cardiac mesoderm develop in pnr¹ embryos and these regions produce short strings of cardioblasts that maintain the wild-type 1:2 ratio of Svp-positive:Svp-negative cardioblasts(Fig. 6). We reasoned that if pnr and pnt act in a developmental pathway, then we should observe the pnt mutant phenotype in those regions of pnr¹ pnt^S012309 embryos in which cardiac mesoderm develops. In agreement with this, we observe local overproduction of cardioblasts in pnr¹ pnt^S012309 embryos and the vast majority of these cardioblasts are Svp-lacZ positive. These double mutant studies support the model that pnr acts upstream of pnt in a developmental pathway.

The results in this paper indicate that pnr and pnt act sequentially to regulate heart development(Fig. 7, bottom panel). pnr acts early in mesoderm development to enable the cardiac mesoderm to form. Subsequent to this event, pnt acts within the cardiac mesoderm to regulate the ability of cells to choose between the pericardial or cardioblast fate. In this context, pnt inhibits the development of the Svp-class of cardioblasts and appears to function independently of Ras/MAP kinase pathway activity.

The effect of pnt on heart development is restricted to the posterior seven heart segments where Svp cardioblasts normally develop. Interestingly, our lineage studies identify a clear difference in the cell lineage of cardioblasts that develop in the posterior seven heart segments versus those that develop more anteriorly(Fig. 7)(Ward and Skeath, 2000). These results identify a genetic and developmental distinction between these two regions of the heart. In addition, they suggest that cells in different regions of the heart carry out different functions and that these functions are probably under homeotic gene control. Future work that addresses the physiological role of these cells in heart function and the control of their development by homeotic genes should provide a more comprehensive understanding of heart development.

Our data suggest that PntP2 may regulate cardioblast and pericardial cell development independently of Ras/MAP kinase activity. Given that every other developmental function of pnt has been traced back to receptor tyrosine kinase/Ras signaling activity, the apparent Ras independent activity of PntP2 is puzzling. As PntP2 is expressed broadly throughout the mesoderm(data not shown) (Klambt,1993), a number of models can explain the apparent Ras-independent activity of PntP2 in the heart. For example, PntP2 may not require MAP-kinase-mediated phosphorylation to carry out a subset of its function. Consistent with this, phosphorylation of PntP2 does not appear to affect its DNA-binding ability (O'Neill et al.,1994). Thus, in the absence of MAP-kinase stimulation, PntP2 is still probably able to bind target promoters alone or in complexes with other proteins. Such an activity of PntP2 could on its own regulate target gene expression by blocking the ability of other transcriptional effectors to bind to and activate target gene transcription, or through an obligate association with other proteins required to activate (or to repress) target genes. Significant precedent exists for such activity. For example, the Su(H)/CSL and pangolin/TCF proteins are the transcriptional effectors of the Notchand wingless pathways, respectively, and in the absence of Notch or wingless activity these proteins can repress target gene transcription (Cavallo et al.,1998; Li et al.,1997; Mumm and Kopan,2000; van de Wetering et al.,1997).

A second model is that PntP2 requires MAP kinase activation but that this activity is carried out by one of the other MAP kinase pathways in Drosophila: the JNK pathway or the p38 pathway. Preliminary phenotypic analyses indicate that heart development is normal in embryos mutant for basket, the Drosophila JNK-kinase (J.B.S.,unpublished). Analysis of p38 kinase activity is presently limited because of the absence of suitable genetic backgrounds. A third possibility is that a novel Ras-dependent pathway does in fact activate PntP2 during heart development. This model is consistent with the recent identification of a novel receptor tyrosine kinase expressed in the developing visceral mesoderm(Loren et al., 2001). Our experiments that failed to identify a pnt-like excess cardioblast phenotype upon mesodermal overexpression of a dominant-negative form of Ras argue against this model. However, Ras is maternally loaded and it is extremely difficult to eliminate all Ras activity in this manner. Thus, even though we observed Ras-like mesodermal phenotypes in these experiments, we still may have missed a role for Ras in regulating cardioblast number because of differential sensitivity of different developmental pathways to partial Ras inactivation. Future work that (1) addresses the ability of MAP-kinase insensitive forms of PntP2 to regulate heart development, and (2) identifies PntP2 target genes in the heart and elucidates how PntP2 regulates such genes should help clarify the molecular basis through which PntP2 governs heart development.

Our phenotypic analysis of pnr conflicts with a prior study that showed an increase in pericardial cells in pnr mutants(Gajewski et al., 1999). This study used Eve to identify a subset of pericardial cells in pnr¹ embryos. We attribute the difference in our results to our use of the pnr^VX6 null allele, our ability to distinguish unambiguously Eve-positive pericardial cells from Eve-positive somatic muscle progenitors, and to specific defects in dorsal closure exhibited by pnr embryos that result in the local aggregation of cells in the dorsal region of the embryo. Our genetic results identify pnr¹ as a hypomorphic allele and we find that Eve-positive pericardial cell formation is almost wild type in this background. In these experiments, we unambiguously identified Eve-positive pericardial cells via their co-expression of Zfh1 and were thus able to quantify precisely Eve-positive pericardial cell number in pnr¹ embryos. This is important as one can observe local increases in Eve-positive mesodermal cells in pnr embryos. However, such apparent increases arise from the local aggregation of dorsal mesodermal cells in pnr¹embryos caused by defects in dorsal closure and not by an overall increase in Eve-positive mesodermal cells.

The genetic identification of pnr¹ as a hypomorphic allele is intriguing given that molecular and expression analyses indicate the pnr¹ lesion results from a premature stop codon in the middle of the first zinc finger and that the Pnr¹ protein localizes predominantly to the cytoplasm. This lesion is expected to abrogate the DNA-binding ability of the Pnr protein. However, our genetic experiments indicate that the Pnr¹ protein retains residual activity at least with respect to heart development. These results raise the possibility that Pnr may be able to carry out some of its functions independently of DNA binding. Precedence for such an activity comes from studies on a genetically engineered form of the homeodomain transcription factor Fushi-tarazu that lacks the homeodomain but retains significant biological activity(Copeland et al., 1996). Future work that focuses on a detailed structure function analysis of the Pnr protein should clarify whether Pnr can act independently of its DNA-binding ability in some developmental contexts.

We should also note that our pnt allelic series indicates that pnt^▵88 exhibits a milder excess cardioblast phenotype than pnt^S012309, pnt², and pnt^RR112. This result is surprising as pnt^▵88deletes the exons pntP2 shares with pntP1 and as a result was assumed to be an amorphic allele of the pnt locus(Scholz et al., 1993). Using antisense RNA probes specific for the unique exons of pntP2, we observe an essentially wild-type pattern of pntP2 transcription in pnt^▵88 mutant embryos(data not shown). These data raise the possibility that the N-terminal regions of pntP2 may also retain partial activity. Studies along the lines of those suggested for Pnr should also help elucidate whether truncated forms of PntP2 retain residual activity.

As noted, significant similarity exists between the embryology and molecular regulation of early heart development in Drosophila and vertebrates. In this context, the identification of a role for pnt, a member of the evolutionarily conserved ETS transcription factor family, in Drosophila heart development raises the possibility that ETS family proteins regulate vertebrate heart development. Consistent with this, ETS1 and ETS2, the two most closely related vertebrate ETS proteins to pnt,are expressed in the developing vertebrate heart; functional studies indicate these genes regulate the expression of specific genes in the heart(Majka and McGuire, 1997; Macias et al., 1998). However,knockout studies have not yet revealed a clear role for ETS1 or ETS2 in the morphological development or differentiation of the vertebrate heart. The existence of multiple vertebrate ETS-family members highly homologous to pnt, as well as a total of 25 ETS family members in humans suggests the possibility of functional redundancy in ETS protein function during vertebrate and mammalian heart development. Thus, a full understanding of ETS protein function during heart development awaits construction and analysis of animals multiply mutant for different ETS family members.

We are especially grateful to Manfred Frasch, Rolf Bodmer and Petra Levin for invaluable comments on the manuscript and various reagents. We are also indebted to Alan Michelson for providing many fly lines and helpful advice,and Scott Wheeler for Fig. 7. We thank Bruce Paterson, Gines Morata, Zhi-Chun Lai and the Developmental Studies Hybridoma bank for antibodies and/or fly stocks. As always we greatly appreciate the help received from Kathy Matthews and the Bloomington Stock Center. This work was supported by a grant to J.B.S. from NSF(IBN-0077727).