ladybird, a new component of the cardiogenic pathway in Drosophila required for diversification of heart precursors

During early embryogenesis of Drosophila, prior to cellularization, a maternal regulatory gradient of the Dorsal gene product defines presumptive mesoderm and ectoderm territories (Maggert et al., 1995). In the ventral region, high concentrations of Dorsal induce the expression of mesoderm genes, snail (sna) (Ip et al., 1992; Maggert et al., 1995) and twist (twi) (Thisse et al., 1991) while, in the dorsal ectoderm, low Dorsal levels activate decapentaplegic (dpp) encoding a TGF-β-like signaling molecule (Huang et al., 1993; Maggert et al., 1995). At the beginning of gastrulation, cells of the mesoderm invaginate through the ventral furrow, spread dorsally to form a single layer, and subsequently split into ventral and dorsolateral lineages. This initial subdivision of the mesoderm involves inductive interactions with the dorsal ectodermal cells which express dpp (Staehling-Hampton et al., 1994; Frasch, 1995). Upon Dpp induction, expression of the homeobox gene tinman (tin) becomes restricted to the dorsolateral mesoderm where its activity is required for the development of visceral, cardiac and dorsal somatic lineages (Azpiazu and Frasch, 1993; Bodmer, 1993). The ventral mesoderm forms most other somatic muscle progenitors in a process involving induction by the overlying neuroectoderm (Baker and Schubiger, 1995). The first morphological event in the subdivision of the dorsal mesoderm is the formation of two layers. The outer layer, which contacts the ectoderm, gives rise to heart precursors and dorsolateral body-wall muscles, whereas the inner layer, which is derived from the bagpipe (bap)-expressing cells, contains progenitors of the visceral muscles of the midgut (Azpiazu and Frasch, 1993). Recent data indicate that the specification and subsequent differentiation of mesodermal derivatives requires additional signals from the ectoderm (Volk and Raghavan, 1994; Lawrence et al., 1995; Baylies et al., 1995; Wu et al., 1995; Park et al., 1996; Ranganayakulu et al., 1996; Azpiazu et al., 1996). The segmental origin and character of mesodermal structures, such as the heart (Azpiazu and Frasch, 1993; Lawrence et al., 1995) and the somatic muscles (Bate, 1990; Volk and Raghavan, 1994; Baylies et al., 1995), strongly suggests influences from the segmented ectoderm. Indeed, mutations of the segment polarity genes, which are involved in ectoderm patterning, also affect mesoderm segmentation (Volk and Raghavan, 1994; Baylies et al., 1995; Lawrence et al., 1995; Azpiazu et al., 1996). One member of this class of genes, wingless (wg) (Baker, 1987), encodes a secreted molecule that acts as an inductive signal for somatic muscle precursor formation (Baylies et al., 1995; Ranganayakulu et al., 1996) and the formation of heart precursors (Lawrence et al., 1995, Wu et al., 1995; Park et al., 1996). Epidermal wg activity is able to rescue the heart-less wg⁻ phenotype (Lawrence et al., 1995), suggesting that the Wg signal can act across germ layers. In addition to the dorsoventral and anterior-posterior cues, the segregation of heart and somatic muscle precursors involves cell-cell interactions that are mediated by the neurogenic genes (Corbin et al., 1991; Hartenstein et al., 1992; Bate et al., 1993).

Here, we focus our analysis on the specification of Drosophila heart precursors and the diversification of heart cells. We show that two closely related homeobox genes, ladybird early (lbe) and ladybird late (lbl) (Jagla et al., 1993, 1994, 1997), are specifically expressed in a subpopulation of tin-expressing heart progenitors. Our results suggest that the lb genes are required to specify the identity of heart precursors.

The following fly strains were used: wg^CX4 (Baker, 1987), wg^IL114 (Bejsovec and Martinez-Arias, 1991), hh^9K (Heemskerk and DiNardo, 1991), tin^EC40, tin^Df(3R)GC14 (Bodmer, 1993; Azpiazu and Frasch, 1993) and a set of neurogenic mutants, N^55e11, bib^ID05, mam^Z3, Df(3R)E(spl)^Bx22, Dl^RevF10 and neu^IF65. wg^IL114 and hh^9K are temperature-sensitive alleles that behave as nulls at 29°C. To distinguish between mutant and wild-type embryos, where applicable, the balancer chromosomes were marked with a twi-lacZ P-element and the embryos were stained using anti-β-galactosidase antibody.

Heat-shock (Hs) stocks were used to study the effects of lb gene over-expression. Hs-lbe-4 and Hs-lbl-5 (Jagla et al., 1997) are inserted on the X chromosome. Double Hs-lbe;lbl transgenic flies were generated as described previously (Jagla et al., 1997). A modified deficiency stock (Df(3R)e^D7, P(w+)tinre58/TM3ftzlacZ; Azpiazu and Frasch, 1993), in which both tin and lb loci are missing while tin function is restored with a rescue construct containing genomic tin sequences, was used to analyse the role of lb in heart formation.

The transgenic embryos were aged at 25°C and heat-shock treated, at different times of development (indicated in the text), for 15 minutes at 37°C in water. The temperature-sensitive wg and hh mutants were aged at 18°C and then raised to the non-permissive temperature (29°C) at 4, 5.5, 6.5, 8 or 9 hours AEL. The rate of development at 18°C is 0.5 times lower than at 25°C and 1.4 times faster at 29°C (Wu et al., 1995).

Detections of the antibody stainings were performed using ABC-Elite-peroxidase or ABC-alkaline phosphatase kits (Vector Laboratories). Antibody dilutions were as follows: monoclonal anti-Lbe (1:1), rabbit anti-Lbl (1:5000), rabbit anti-Eve (1:5000), guinea pig anti-Eve (provided by Dave Kosman) (1:500), rabbit anti-β-galactosidase (1:8000), rabbit anti-Tin (1:2000), rabbit anti-Wg (provided by Roel Nusse) (1:1000) and rabbit anti-MEF2 (provided by Hanh Nguyen) (1:1000). Colour reactions were developed using diaminobenzidine (for peroxidase) or NBT (for alkaline phosphatase) as substrates. Fluorescent microscopy with secondary antibodies conjugated with FITC, Texas Red or Cy5 (1:200; Jackson Immuno Research) was used to determine the position of Lb-positive cells with respect to Eve, MEF2 or Tin-positive heart precursors. The double or triple stainings were analysed using a Leica confocal microscope TCS 4D with 40× or 100× objectives.

lbe and lbl were previously shown to be expressed in the same heart progenitors, but the exact identity of these cells was not known (Jagla et al., 1997). Therefore, we examined the spatial relationship between heart progenitors expressing lb, tin, even-skipped (eve) or mef-2 using confocal microscopy (Frasch et al., 1987; Bodmer, 1993; Azpiazu and Frasch, 1993; Bour et al., 1995; Lilly et al., 1995). As shown in Fig. 1A, at early stage 12, lb is expressed in clusters of about four cells per hemisegment in the developing heart region. These cells represent a segmental subset of tin-expressing heart progenitors, which form a continuous row at the dorsal crest of the mesoderm at this stage (Fig. 1G). eve expression begins at a slightly earlier time than lb in similar clusters of cells. It appears that two cells from each segmental eve cluster develop into a particular type of pericardial cells, termed e-PCs. Double stainings for lb and eve expression demonstrate that the e-PC progenitors are distinct from the lb-expressing heart progenitors and located posteriorly adjacent to them in each segment (Fig. 1B). Similar stainings of embryos at later stages show that the lb-expressing cells give rise to a subpopulation of cardioblasts (CBs) and a second type of pericardial cells, termed l-PCs. As shown in Fig. 1C-F, cell rearrangements during stage 12, which involve a 90°, clockwise rotation of the heart progenitor clusters within each segment, place the lb-expressing cells at the dorsal side and move the eve-expressing cells ventrally to them. This morphogenetic process results in a dorsal row of cardioblasts and ventrolaterally adjacent rows of pericardial cells on either side of the embryo. At stage 14, generally four out of six cardioblasts per hemisegment express both tin and mef-2 (Fig. 1H-J). Double stainings with Lb antibodies show that the two anterior tin- and mef-2-expressing cardioblasts in each hemisegment co-express lb (Fig. 1H-J). In addition, tin and lb are co-expressed in the l-PCs, which are located ventrally below the cardioblasts (Fig. 1H). However, lb is not expressed in the e-PCs, which are found in more lateral positions at this stage (Fig. 1J). These results indicate a diversification among cardioblasts of each segment, as well as among the pericardial cells, that is already apparent during stage 11.

Because of the important role of wg in heart development, we compared its domains of expression with the locations of lb- and eve-expressing heart progenitors. As shown in Fig. 2, in embryos before and during germ-band retraction, both lb- and eve-labeled heart-progenitors are localized in the mesodermal areas below each ectodermal wg stripe. This arrangement is compatible with a role of wg in the specification and/or maintenance of the developmental fates of these cells.

Since heart development requires tin function (Azpiazu and Frasch, 1993; Bodmer, 1993) and lb-positive heart progenitors emerge from tin-expressing dorsal mesodermal cells, we tested for lb activity in tin mutants. We find that segregation of both cardioblasts and pericardial cells, as monitored by lb expression, does not occur in tin⁻ embryos (compare Fig. 3A and Fig. 3C) and lb-heart cells are missing after germ-band retraction (compare Fig. 3B and Fig. 3D). To investigate the influence of Wg and Hh signaling on lb-positive cardiac cells, we shifted wg^IL114 and hh^9K thermosensitive mutant embryos at 4 or 6 hours of development to non-permissive temperatures. The influence of wg (Jagla et al., 1997) and hh (our unpublished observation) on lb expression in the dorsal epidermis allowed us to identify the homozygous mutant embryos. In wg^IL114 embryos lacking either early (from 4 hours AEL) or late (from 6 hours AEL) wg function, lb expression in heart precursors is absent (Fig. 3E,F), indicating that initiation of lb activity requires both tin and wg action and is maintained by Wg signaling from the dorsal epidermis. In contrast, loss of early hh function does not prevent lb expression (Fig. 3G), while, in the absence of late Hh signaling, lb expression is expanded (Fig. 3H).

Thus, epidermal Wg and Hh signaling have opposing influences on lb activity in heart progenitors. Wg signal is required to induce lb in anteriorly located cardiac cells, while the Hh signal in the cells underlying posterior epidermal compartments acts to inhibit lb expression (Fig. 3I).

Our previous analysis of the epidermal lb function revealed that late wg activity in dorsal and caudal epidermis is maintained by a wg-lb regulatory feedback loop (Jagla et al., 1997). Since lb-dependent dorsal epidermal domains of wg expression overlie the heart mesoderm (see Fig. 2), the Wg signal sent by epidermal cells may continue to reach the cardiac cells during this period. The time window during which wg can influence heart development was determined with the thermosensitive wg^IL114 allele and three different antibodies as markers, anti-Eve (Fig. 4A-C), anti-Tin (Fig. 4D-F) and anti-MEF2 (Fig. 6G-I). The loss of early wg activity, starting from 3.5-4.5 hours of development (Fig. 6B,E,H), abolishes dorsal mesoderm differentiation and determination of eve-, tin- and mef2-positive cardiac lineages, suggesting that the Wg signal is required for specifying both pericardial cells and cardioblasts. Later, during germ-band retraction, when cardiac lineages have been defined, wg activity continues to play a role in heart development (Fig. 4C,F,I). wg^IL114 embryos shifted to the non-permissive temperature at 7.5 hours AEL have a reduced number of eve-positive pericardial cells (Fig. 4C), mef2-cardioblasts (Fig. 4I) and disrupted heart pattern as revealed by Tin staining (Fig. 4F). There is a correlation between the severity of the heart defects and the start of the temperature shift, such that earlier loss of wg function causes stronger defects in heart formation (data not shown). Since late wg activity in the dorsal epidermis is maintained by lb (Jagla et al., 1997), we conclude that the formation of a normal heart pattern requires the wg-lb regulatory loop.

Previous data have shown that, similar to the neural cells, the neurogenic genes are required for the segregation of a proper number of somatic muscle founder cells and heart progenitors (Corbin et al., 1991; Hartenstein et al., 1992; Bate et al., 1993). The analysis of lb expression patterns revealed that lb-expressing heart progenitors contribute to the increased number of cardiac precursors in neurogenic mutants (e.g., N, Dl, E(spl), mam – Fig. 5) (bib, neu – data not shown). Thus, in extendedgerm-band embryos, the expansion of lb expression in the CNS is accompanied by the formation of enlarged clusters of lb-positive heart cells (Fig. 5A,C,E). Supernumerary lb-heart precursors appear both in embryos displaying weak (Fig. 5A) and severe (Fig. 5E) neurogenic phenotypes. At this early stage, the (abnormal) segregation of cardiac cells does not seem to be disturbed by the neurogenic phenotypes in the dorsal epidermis (these embryos lack for instance lb expression in dorsal epidermal cells – Fig. 5E). As the germ band retracts, the degree of disruption of the heart pattern in different mutant backgrounds correlates with the degree of CNS expansion. Thus, E(spl) or mam embryos with moderate neurogenic phenotypes show heart hyperplasia accompanied by ectopic lb expression (Fig. 5B,D) while, in Dl (Fig. 5F), N and neu (data not shown) mutants, dorsal vessel morphogenesis is strongly disrupted and the expanded lb expression in the heart decays in late stage embryos. This late loss of lb is most likely due to the progressive disruption of the dorsal epidermis and the loss of epidermal Wg signaling that is required for lb expression.

The modulation of the activity of genes involved in heart formation such as tin, wg and hh leads to a deficit or hyperplasia of heart precursors (Wu et al., 1995; Park et al., 1996). Since lb is specifically expressed in a subset of heart progenitors, we wondered whether increased levels of lb gene products may influence specification of the heart lineages. To address this question, we used transgenic flies carrying lbe, lbl or both lb cDNAs driven by a heat-shock promoter (Jagla et al., 1997) and antibodies labeling cardioblasts (anti-MEF2), pericardial cells (anti-Eve) or both types of heart precursors (anti-Tin). Overexpression of lbe or lbl at 5-6 hours AEL, during the segregation of cardioblasts and pericardial cells, leads to a significant hyperplasia of heart progenitors, as monitored by tin expression (compare panels A, E and B, F in Fig. 6). Surprisingly, the number of eve-expressing pericardial cells is strongly reduced (compare panels C, E and D, F in Fig. 6). The analysis of embryos in which ubiquitous expression of both lb transgenes was maintained between 5 and 9 hours AEL confirms that ectopic lb activity can recruit additional cardiac cells. After germ-band retraction, we observe expanded Tin domains in the majority of these embryos (compare panels G, H and M, N in Fig. 6). (Fig. 6I, J), as well as supernumerary mef-2-cardioblasts (compare panels K, L and M, N in Fig. 6). Since, in the same embryos, eve-pericardial cells are missing, this raises the possibility that lb determines the identity of a distinct type of pericardial cells and that misexpression of lb in e-PCs leads to their transformation into l-PCs. To further investigate this eventuality, we analysed eve expression in Df(3R)e^D7, tin rescue embryos in which both lbe and lbl genes are deleted, while tin function is restored by a P-insertion (Azpiazu and Frasch, 1993). We find that, in embryos lacking lb, eve expression is expanded (Fig. 7). At stage 12, instead of 2-3 cells per hemisegment, the eve clusters contain 4-5 heart precursors (Fig. 7A). After germ-band retraction, additional eve-cells contribute to the formation of the heart and are found at positions where lb-cells are normally located (Fig. 7B). Since heart hyperplasia was not observed in Df(3R)e^D7, tin rescue embryos (Azpiazu and Frasch, 1993 and our unpublished observations), these supernumerary eve-cells result most likely from the transformation of l-PCs into e-PCs. To exclude that this is due to the absence of genes located distally to lb, the Df(3R)e^D7, tin rescue flies were crossed with a stock carrying a shorter deficiency, Df(3R)e^F1 (Jagla et al., 1997). The analysis of transheterozygous embryos revealed the expansion of e-PCs similar to that presented in Fig. 7 (data not shown). Taken together, these data indicate that lb cooperates with tin to determine a subset of tin-positive cardiac precursors, thus providing spatial information along the anterior/posterior axis during differentiation of the heart mesoderm.

In both vertebrates and invertebrates, the heart originates from bilateral mesodermal primordia and there is increasing evidence suggesting that molecular control mechanisms in the specification of cardiac lineages have been conserved during evolution (for review see Bodmer, 1995 ; Olson and Srivastava, 1996). However, the subsequent development of the heart differs between vertebrates and invertebrates. Cells in the linear heart tube of vertebrates assume different characteristics along the anterior/posterior axis, an important prerequisite for later morphogenesis that includes looping and chamber formation. In contrast, the morphology of the insect heart appears rather uniform along the anterior/posterior axis and the heart remains as a linear tube. Here, we show that the heart precursors in Drosophila assume different identities along the anterior/posterior axis within each segment. The lb genes, which encode homeodomain-containing transcription factors, are expressed in a specific subset of cardioblasts, as well as in a particular type of pericardial cells within each segment, and play a role in diversification of heart precursors. Like other Drosophila genes involved in important developmental decisions, lb genes have been evolutionarily conserved and have homologs in mouse and humans (Jagla et al., 1995).

Double-staining experiments with different markers revealed that lb-expressing cardiac cells correspond to a subset of cardioblasts and pericardial cells. At the beginning of the extendedgerm-band stage, these cells segregate from the dorsal mesodermal cells in a process that depends on mesodermal tin function. Co-localisation with the epidermal wg domains, detected by confocal microscopy, shows that they are positioned below the anterior compartments. In addition to tin activity, the formation of lb-heart cells requires Wg signaling, which is consistent with the findings of Wu et al. (1995). Using a thermosensitive wg allele, we show that lb expression in the heart depends on Wg that is secreted from the dorsal epidermis. Since lb expression is limited to the heart cells that directly contact overlying wg domains, it seems that only high Wg concentrations are able to induce and maintain lb activity. In contrast, cardiac cells located below posterior compartments are under the influence of Hh signals which suppresses lb. This opposing action of Wg and Hh on mesodermal lb expression is reminiscent of the regulation of bap activity during visceral mesoderm development at earlier stages (Azpiazu et al., 1996). These results indicate that wg and hh, in addition to their earlier requirements for the formation of all heart progenitors, play important roles in the diversification of heart cells within each segment.

Our analysis of lb expression provides further evidence that the formation of a normal heart requires neurogenic gene function. Hyperplasia of lb-cardiac precursors in any of the mutants for neurogenic genes indicates that the previously observed increase of heart cell number, as monitored with enhancer trap markers (Hartenstein et al., 1992), does not occur at the expense of lb-expressing cells. It is possible that both e-PCs and l-PCs fail to segregate in neurogenic mutants, thus giving rise to supernumerary lb-expressing cardioblasts.

The pattern of lb expression in progenitors of cardioblasts and pericardial cells suggests a role for the Lb gene products in the specification and diversification of cardiac lineages. To investigate this hypothesis, we have used transgenic embryos expressing lbe, lbl or both lb genes ubiquitously. Overexpression of lb during segregation of the cardiac lineages leads to a hyperplasia of heart precursors. This may indicate that lb can activate tin and perhaps other early heart determination genes to recruit ectopic mesodermal cells into the cardiogenic pathway. Since lb expression itself depends on tin function and is associated with a subpopulation of tin-positive heart cells, this activity of ectopic lb may reflect an auto- and crossregulatory loop between tin and lb to maintain their mutual expression in heart progenitors during normal development. Moreover, ubiquitous lb expression leads to a loss of eve- positive pericardial cells suggesting that ectopically expressed lb can change the identity of heart cells. Since, in embryos lacking lb function, eve expression is expanded, presumably because of a transformation of l-PCs into e-PCs, we conclude that lb is required for the diversification of cell fates within the heart. In addition, late wg activity in the dorsal epidermis, which depends on ectodermally expressed lb (Jagla et al., 1997), plays a role in maintaining lb expression in heart progenitors and in heart pattern formation. Together, our results provide new insights into Drosophila cardiogenesis, showing that lb genes exert at least two functions. First, autonomous, during the segregation of the heart precursors, where lb genes act to specify a subset of cardioblasts and pericardial cells, and second, nonautonomous, in heart pattern formation by maintaining late Wg signaling from epidermal cells (Fig. 8).

As we have shown previously (Jagla et al., 1994, 1997), lb homeobox genes reside in the 93E region of the third chromosome, distally to tin and bap (Azpiazu and Frasch, 1993) and proximally to S59 and C15 (Dohrmann et al., 1990; P. Andermann, M. Frasch and E. Weinberg, unpublished data). All these genes code for homeodomain-containing transcription factors that are involved in various aspects of mesoderm differentiation. tin and bap, located in the most proximal part of the cluster, determine the formation of the heart and the visceral mesoderm primordia (Azpiazu and Frasch, 1993; Bodmer, 1993) whereas the distally located S59 and C15 homeobox genes appear to play roles in founder cell specification of body wall and alary muscles (Dohrmann et al., 1990; P. Andermann, M. Frasch and E. Weinberg, unpublished data). lb genes occupy a central position in the cluster and are expressed in a subset of tin-heart cells, corresponding to anterior cardioblasts and pericardial cell precursors in each segment. In addition, lb expression is specific to the segmental border muscle founder cells (Jagla et al., 1997, and data not shown). Interestingly, mesodermal bap, S59, C15 and lb expression all require tin function, suggesting that these 93E homeobox genes act in the tin-dependent cascade of genetic interactions. Since the members of the 93E cluster are evolutionarily conserved (for review see Harvey, 1996; P. Andermann, M. Frasch and E. Weinberg, unpublished data), it would be interesting to determine whether the homologs are also involved in a similar pathway in vertebrates.

We are grateful to Pierre Chambon for the support, Dave Kosman for giving helpful advice for the confocal microscopy and Hanh Nguyen for valuable comments on the manuscript. We thank also Pascal Heitzler for discussion and Marie-Louise Nullans for excellent technical assistance. This work was supported by grants from the Ministère de la Recherche, the CNRS, the INSERM, the Fondation pour la Recherche Médicale, the Association pour la Recherche sur le Cancer as well as by a grant from the NIH and a Pew Award to M. F.