Analysis and reconstitution of the genetic cascade controlling early mesoderm morphogenesis in the Drosophila embryo

Mechanisms of Development 124 (2007) 167–179 www.elsevier.com/locate/modo

Analysis and reconstitution of the genetic cascade controlling early mesoderm morphogenesis in the Drosophila embryo Thomas C. Seher, Maithreyi Narasimha 1, Elisabeth Vogelsang, Maria Leptin

*

Institute of Genetics, University of Cologne, Weyertal 121, D-50931 Cologne, Germany Received 21 November 2006; received in revised form 15 December 2006; accepted 19 December 2006 Available online 27 December 2006

Abstract To understand how transcription factors direct developmental events, it is necessary to know their target or ‘effector’ genes whose products mediate the downstream cell biological events. Whereas loss of a single target may partially or fully recapitulate the phenotype of loss of the transcription factor, this does not mean that this target is the only direct mediator. For a complete understanding of the pathway it is necessary to identify the full set of targets that together are sufficient to carry out the programme initiated by the transcription factor, which has not yet been attempted for any pathway. In the case of the transcriptional activator Twist, which acts at the top of the mesodermal developmental cascade in Drosophila, two targets, Snail and Fog, are known to be necessary for the first morphogenetic event, the orderly invagination of the mesoderm. We use a system of reconstituting loss of Twist function by transgenes expressing Snail and Fog independently of Twist to analyse the sufficiency of these factors–a loss of function assay for additional gene functions to assess what further functions might be needed downstream of Twist. Confirming and extending previous studies, we show that Snail plays an essential role, allowing basic cell shape changes to take place. Fog and at least two other genes are needed to accelerate and coordinate shape changes. Furthermore, this study represents the first step in the systematic reconstruction of the morphogenetic programme downstream of Twist.  2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Gastrulation; Cell shape changes; Transcription; Snail; Twist; Fog; Concertina

1. Introduction We have been studying an example of transcriptional regulation of a morphogenetic event, the formation of the ventral furrow in the Drosophila embryo, in which the first steps of the genetic hierarchy are very well understood. The ventral furrow forms in a region covering most of the ventral side of the embryo, the mesoderm primordium, in which the nuclear accumulation of high levels of the transcription factor Dorsal leads to the transcription of two genes, snail and twist. Together these genes define the *

Corresponding author. Tel.: +49 221 4703401; fax: +49 221 4705264. E-mail address: [email protected] (M. Leptin). 1 Present address: Department of Biological Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India.

mesoderm primordium and are essential for all aspects of its differentiation and morphogenesis (Ip et al., 1992; Jiang et al., 1991; Kosman et al., 1991; Simpson, 1984). Snail, a zinc-finger protein, acts mainly as a repressor of ectodermal genes in the region of the prospective mesoderm (Kosman et al., 1991; Leptin, 1991). Twist, a bHLH-protein, acts as an activator for mesodermal genes. Dorsal leads to the initial low level of Snail and Twist expression, but Twist later cooperates with Dorsal to maintain high levels of Twist and Snail throughout the mesoderm. In the absence of Twist, Snail is transiently expressed in a narrow band of cells but can no longer be detected by the beginning of gastrulation (Ip et al., 1992; Leptin, 1991; Sweeton et al., 1991). The invagination of the mesoderm begins with a series of cell shape changes which lead to the formation of an indentation, the ventral furrow, and its subsequent

0925-4773/$ - see front matter  2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.mod.2006.12.004

168

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

internalization (Leptin and Grunewald, 1990; Sweeton et al., 1991). In embryos mutant for snail or twist different aspects of the cell shape changes are impaired and the furrow fails to form properly; in double mutant embryos no cell shape changes occur at all and no signs of furrow formation are seen. None of the molecularly identified target genes of Twist have so far been shown to be required for ventral furrow formation (Casal and Leptin, 1996; Furlong et al., 2001; Stathopoulos et al., 2002; Taylor, 2000). Indeed, apart from snail only one Twist target has been shown to be required for the proper progression of cell shape changes. This gene, folded gastrulation (fog), encodes a secreted peptide and is expressed in the mesoderm and the posterior endoderm (Costa et al., 1994; Sweeton et al., 1991). Whereas fog function in the endoderm is essential, the defect in the mesoderm is transient and by the end of gastrulation the mesoderm has fully invaginated. Thus, the defects are far less severe than those observed in twist embryos, and there must therefore be other genes that act together with fog to mediate the effects of Twist on gastrulation. Yet, no other genes have been found that are essential for gastrulation. The failure so far to identify any zygotically expressed genes other than twist or snail that are essential for the internalization of the mesoderm suggests that the most important role may be played by the genes that are already known to participate in the control of ventral furrow formation. We therefore tested to what degree the two early

Twist targets, snail and fog, are able to mediate cell shape changes. We find that they are able to a significant degree to rescue the defects in twist mutants, but that other genes acting in parallel must be needed to control the timing and increase the efficiency of the cell shape changes occurring in the ventral furrow. 2. Results 2.1. Subtle defects in ventral furrow formation The three known zygotically expressed genes necessary for proper cell shape changes in the ventral furrow are sna, twi and fog. In homozygous snail or twist mutant embryos, the mesoderm does not invaginate, mesodermal differentiation does not take place and no mesodermal tissues form (Leptin and Grunewald, 1990; Simpson, 1984; Sweeton et al., 1991). If the dose of snail is reduced by half (in heterozygous sna/+ embryos), ventral furrow formation is delayed, as well as being uneven and irregular, but the invagination occurs (Fig. 1). No such delay is seen in twist/+ embryos, but the mutant phenotype of embryos in which both twist and snail are reduced by half is stronger than that of sna/+ embryos (not shown). Nevertheless, these embryos produce a fully functional mesoderm and develop into viable and fertile adults, indicating that a delay in furrow formation is not fatal. Similarly, homozygous folded gastrulation (fog) mutant embryos show a delay in ventral furrow formation, but the mesoderm eventually

Fig. 1. Ventral furrow formation in snail/+, fog and fog; hkb::fog embryos. (A,A’) Lateral and ventral view of a wildtype embryo stained with antibodies against Twist (brown) and Even-skipped (blue) (B,B’) heterozygous snail/+ (Df(2L)TE116GW11/CyO–lacZ) embryo stained against Twist (brown) and bgalactosidase (blue) to identify the heterozygous embryos. (C) fog4a6/fog4a6 and (D) fog4a6/fog4a6; hkb::fog embryos hybridized with RNA probes against b-galactosidase (absence of staining identifies mutant embryos) and snail (blue) and, in addition, stained with an antibody against the phosphorylated form of Histone H3 (PH3; brown) for accurate developmental staging, using the spatiotemporal pattern of the mitotic domains as landmarks (Foe, 1989). hkb::fog rescues posterior midgut invagination in the embryo in (D).

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

invaginates. The delay in furrow formation does not interfere with further mesoderm development. The consequence of the early mesodermal defects for viability is less easy to score, because loss of fog also results in the complete failure of posterior midgut invagination and gut morphogenesis. Mosaic studies have suggested that the function of fog at the posterior end is essential, whereas at least part of the ventral region may lack fog without fatal effects for the fly (Zusman and Wieschaus, 1985). We tested directly whether fog function outside the posterior midgut primordium is essential for viability by creating a transgene that allows fog expression exclusively at the posterior pole under control of the huckebein promoter (Barrett et al., 1997), and introducing this hkb::fog transgene into fog mutants. Such embryos show the typical delay in ventral furrow formation, but form a normal posterior midgut invagination (Fig. 1D), extend their germ bands and develop into viable and fertile flies. This shows that fog is essential only for the invagination of the posterior endoderm, but is dispensable for the formation of the mesoderm as well as for all later steps of development. 2.2. Fog and Snail act in parallel pathways downstream of Twist Whereas neither the reduction of snail function by half nor complete loss of fog function in the mesoderm leads to lethality, the combined reduction of gene activities does. fog mutant flies carrying the hkb::fog transgene show severely reduced viability if the dose of snail is halved. If, in addition, the dose of twist is reduced, no flies survive (Table 1). We used two approaches to test whether this was due to defects during gastrulation rather than genetic Table 1 Genetic interactions between fog and other loci involved in gastrulation Cross:

fog hkb::fog fog ; hkb::fog

Genotype of father (mutation)

Male progeny fog Balancer Y ; hkb::fog ð%Þ

fog Mutaion Y ; hkb::fog

twistEY53/SM1 Df(2L)sna/CyO Df(2L)sna Df(2R)twi/SM1 Df(2L)sna/CyO; PE::fog Df(3L)BK10/TM3 Df(3L)rdgCco2/TM6C,Sb Df(3L)ri79c/TM3 Df(3R)Tlp/TM3

51 96 100 41 39 56 46 55

49 4 0 59 61 44 54 45

þ Y;

mutation Balancer

ð%Þ

Total (n) 113 45 63 93 65 64 67 75

Females that were homozygous for fog4a6 and carried the hkb::fog transgene were crossed to males that carried the indicated mutations or deficiencies over a Balancer. The male progeny of these crosses (hemizygous for fog4a6) have no fog function in the mesoderm, whereas the females have a functional copy of fog; half of each group is heterozygous for the other mutation or deficiency being tested, the other half carries the Balancer. The percentage of survivors carrying the deficiency or mutation was scored in each group. The deficiencies for snail and twist were Df(2L)TE116GW11 and Df(2R)S60.

169

interactions at later stages of development. First, if the lethality is exclusively due to the requirement for Fog during ventral furrow formation, viability should be restored if fog function is restored during this stage. We therefore constructed embryos that were heterozygous for snail, homozygous for fog, but in addition to the hkb::fog transgene also carried a transgene expressing fog in the ventral furrow (PE::fog). This additional transgene re-established viability in the fog/Y; sna/+ flies (Table 1). In line with previous conclusions (Morize et al., 1998), this indicates that lack of gene function during early embryogenesis rather than at later stages had been responsible for the observed lethality. We also investigated the phenotype of the mutant embryos. The defects during ventral furrow formation in fog/Y; sna/+ embryos were drastically worse than those in either fog/Y or sna/+ embryos (Fig. 2). Thus, although fog function in the ventral furrow is normally dispensable for viability, it becomes essential when the dose of Snail is reduced, indicating that ventral furrow formation is controlled in a redundant manner, with Snail and Fog acting in parallel pathways. 2.3. Identification of genomic regions required for proper formation of the ventral furrow We sought to identify additional Twist targets by a screen for subtle defects in ventral furrow formation. Since this requires detailed comparison of fixed and stained embryos from narrowly timed, large egg collections (see Section 5) precluding large scale mutagenesis experiments in which thousands of stocks need to be kept, processed and screened, we screened the ‘deficiency kit’, a collection of approximately 200 stocks that are heterozygous for large deletions, each uncovering up to 1% of the genome (Poulson, 1937). Only eight genomic loci are necessary to allow the zygote to develop normally up to gastrulation (Merrill et al., 1988). Any defects seen during gastrulation in embryos that are homozygous for a particular deletion can, therefore, be assumed to be the direct effect of the loss of a gene in this region, rather than a secondary effect of earlier failures in development. The X chromosome was not screened since it had previously been shown that the only zygotic genes on the X chromosome involved in early embryonic morphogenesis are folded gastrulation and nullo (E. Wieschaus, pers. comm., Merrill et al., 1988; (Postner and Wieschaus, 1994; Mu¨ller et al., 1999)). We used antibodies against Twist and Even-skipped to analyze narrowly staged (2–3 h AEL) collections of embryos. Possible genes required for gastrulation located near twi or sna cannot be identified directly by observing deficiencies in these regions, as the effects would be obscured by the sna or twi mutant phenotypes. We therefore introduced transgenes of twi or sna into the mutant stocks to be able to observe homozygous mutants in which the function of sna or twi had been specifically reconstituted (see Section 5). In both cases, the homozygous mutant embryos underwent

170

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

Fig. 2. Effect of reducing the dose of sna in fog/Y embryos. (A–C) fog/Y embryos. (A,B) Cross sections, (C) whole mounts of five randomly chosen fog embryos (judged by the absence of posterior midgut invagination) during early to late gastrulation stages. In the youngest embryo (top; Eve stripes still separated on the dorsal side), the furrow has invaginated only in the anterior region, but in all others it has become at least partly internalized over the whole length of the embryo. (D–F) Sections and whole mounts of fog/Y embryos from a collection of embryos derived from parents that were also heterozygous for snail. (+/Y; sna/CyO · fog/+; sna/+; for selection of embryos see Section 5). The embryos shown here represent the middle of the phenotypic range.

normal gastrulation (not shown), indicating that no other genes required for furrow formation were uncovered by these deficiencies. We found no other regions of the genome that were essential for ventral furrow formation. While this screen only covered 85% of the autosomal genome, we nevertheless consider it very unlikely that further genes with essential functions in furrow formation remain to be found. A similar screen, giving lower resolution but higher coverage did not find any additional loci ((Mu¨ller et al., 1999) A. Mu¨ller and E. Wieschaus, pers. comm.), but we realize that neither screen assayed small regions of the genome directly proximal to the centromeres. However, none of the many other screens conducted in Drosophila have ever identified any likely candidates for such genes (which would be lethal), suggesting that apart from Snail there are no individually essential zygotic components involved in ventral furrow formation. We identified four autosomal regions containing functions contributing to ventral furrow formation, defined by six deficiencies. The cytogenetic interval 24C3–8 is

defined by the overlap of Df(2L)sc19–8 and Df(2L)ed dp, interval 71C3-E5 by Df(3L)BK10, interval 77D1–F1-5 by the overlap of Df(3L)ri79c and Df(3L) rdgCco2, and interval 97A–98A by Df(3R)TlP (Fig. 3). In all four cases the invagination of the mesoderm of homozygous deficient embryos was delayed and less well coordinated than in wildtype embryos. However, in none of the deletions was the invagination of the mesoderm entirely abolished. The observed defects were transient and the mesoderm eventually invaginated completely. The mutant phenotypes were specific for the mesoderm since disturbances of other early morphogenetic events, such as cellularization or germ band extension, were not detected. 2.4. Genetic interactions of the identified deficiencies The genes whose loss is responsible for the defects observed in two of the regions have been identified as the cell cycle regulators tribbles (cytogenetic interval 77D1–77F5) and fru¨hstart (1C3–E5) (Grosshans and Wieschaus, 2000; Mata et al., 2000; Seher and Leptin,

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

171

Fig. 3. Embryos homozygous for deficiencies uncovering loci required for proper ventral furrow formation. Homozygous deficient embryos exhibit transient defects in the invagination of the mesoderm. Lateral and ventral views (A) or transversal sections (B) of wildtype or homozygous mutant embryos. The embryos were stained against Twi and Eve in all cases. Some of the identified regions are represented by different deficiencies (cf. text) for whole mounts and transversal sections. Sections of embryos homozygous for Df(3L)BK10 are not shown as the phenotype is similar to that of embryos deficient for the cytogenetic interval 77D1-F1-5, represented here by Df(3L)rdgCco2.

2000; Grosshans et al., 2003). Both are necessary to delay cell division in the mesoderm until its invagination has been completed. However, in the absence of cell division, tribbles

and fru¨hstart are not required, and thus do not contribute to ventral furrow formation per se (Seher and Leptin, 2000).

172

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

No abnormalities in cell cycle regulation, differentiation, gene expression patterns or cell architecture were observed in embryos that were homozygous for the other two genomic intervals identified. The mutations therefore affected only the early morphogenetic activity of the mesoderm, and not its later differentiation. The early defects resemble the situation in fog mutant embryos, in that no specific aspect of the activities of the prospective mesodermal cells is completely blocked. Rather, they seem to be slowed down and less well coordinated. The gene uncovered by Df(3R)TlP was shown to be t48, which will be described in detail elsewhere (Ko¨lsch et al., 2007). Mutants in t48 are viable, confirming the redundancy of the gene in the mesodermal developmental pathway. The gene uncovered by Df(2L)sc19–8 and Df(2L)ed dp remains to be identified. If, as we postulated, the factors that control the invagination of the mesoderm act redundantly, mutating two or more genes should result in an enhancement of the phenotypes of the single mutants or, in the most extreme case, in the complete absence of ventral furrow formation. However, embryos that were homozygous mutant for any pair of the deficiencies, or that were simultaneously homozygous for Df(2L)ed dp, Df(3R)TlP and tribbles showed no stronger defects than the single mutants. We did not create quadruple homozygous mutant embryos, for the following reasons. First, the low vitality of the quadruple heterozygous parents made this impossible. Second, we suggest that the triple mutant of Df(2L)ed dp, Df(3R)TlP and tribbles

represents the same phenotypic situation as a quadruple mutant also including Df(3L)BK10 or fru¨hstart would, because two of the uncovered genes, tribbles and fru¨hstart, affect the same process (cell division in the mesoderm primordium) and show no redundancy in their function. Thus, the triple mutant disrupts all of the processes affected by the individual mutations, and still does not reproduce the mutant phenotype of twist or snail mutant embryos. The deficiencies also did not reduce the viability of homozygous fog mutants carrying the hkb::fog transgene in the way that snail did (Table 1), suggesting that none of them alone could account for this effect of snail mutations. However, we did observe an exacerbation in the delay of mesoderm invagination when the fog and Df(3R)TlP mutations were combined. Strong defects were seen if embryos were simultaneously mutant for fog and Df(3R)TlP, or if the dose of fog was halved and embryos were also homozygous for Df(3R)TlP. Cell shapes in the mesoderm primordium of these embryos appeared highly abnormal, and the cells were not fully internalized until after the first mesodermal cell division (Fig. 4). The extreme delay in the invagination and the absence of normal cell shape changes in the double mutant show that Df(3R)TlP uncovers a gene that acts together with fog in a redundant manner to enable the invagination of the mesoderm. These defects are reproduced when a small deficiency deleting the gene t48 is used instead of the large deficiency Df(3R)TlP.

Fig. 4. Genetic interaction of fog and Df(3R) TlP leads to a strong gastrulation defect. Whole mount embryos and transversal sections stained against Twi and Eve. (A,B) Ventral and lateral view of the same embryo, which is heterozygous for fog4a6 and homozygous for Df(3R)TlP (fog4a6/+; Df(3R)TlP) (C,C’,D,D’) embryos hemizygous for fog4a6 and homozygous for Df(3R)TlP (fog4a6/Y; Df(3R)TlP) at two different stages of gastrulation: ventral views of whole mount embryos and central transversal sections of the same embryos.

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

It is striking that the mesoderm in the double mutant embryos succeeded in establishing a cell layer on the inner side of the ectoderm in spite of the severe disruptions in ventral furrow formation. This had also been observed in tribbles mutant embryos, but the disruptions we show here are even more severe. Thus, surprisingly, an orderly progression through the sequence of cell shape changes seen during ventral furrow formation appears not to be essential for the establishment of the mesodermal germ layer, which is probably directed by FGF-mediated attachment and migration of mesodermal cells on the ectoderm. 2.5. Ventral furrow formation without Twist activity: replacement of Twist function by its target genes snail and fog The nature of the phenotypes observed in the mutants isolated in the screen, together with results from similar efforts in other laboratories ((Mu¨ller et al., 1999), A. Mu¨ller and E. Wieschaus, pers. comm.) make the existence of other zygotic genes essential for ventral furrow formation unlikely. Thus, apart from the essential function of Snail,

173

there appear to be several non-essential genetic processes contributing to early mesoderm morphogenesis downstream of Twist, including that of Fog. This raises the question whether the genes that have been identified so far are sufficient to execute the programme of cell shape changes controlled by Twist, in particular, to what extent its effects are mediated by Snail and Fog alone. If these two factors are the only targets of Twist mediating the invagination of the mesoderm, they should be able to rescue the defects in twist mutants. Indeed, it has been shown previously that raising the level of either Snail or Fog in twist mutants ameliorates the defects (Ip et al., 1994; Morize et al., 1998), but it has not been demonstrated to what extent, and whether their combination is sufficient to generate a ventral furrow. To test this we used constructs that allowed us to express fog and snail in the absence of twist by fusing the fog and snail coding regions to a part of the twist promoter (Ip et al., 1994; Jiang et al., 1991; Jiang and Levine, 1993). We tested the capacity of these constructs (named PE::fog and 2xPE::sna) to rescue the phenotypes of fog and snail, respectively. Both constructs were able to rescue the mesodermal defects of the corresponding mutants (Fig. 5 and (Maggert et al., 1995)). Specifically, Snail was

Fig. 5. Rescue of mesoderm morphogenesis in sna and fog embryos by 2xPE::sna and PE::fog. Lateral midsagittal (A,B) and superficial (C,D) views of whole mount embryos. The embryos are either double labeled with a Dig labeled snail RNA probe and an antibody against Eve (A–D), or hybridized with a Dig labeled fog RNA probe (e–g). (A–D) Rescue of mesoderm invagination in snail (Df(2L)TE116GW11) mutants. Wildtype or heterozyogous embryos (A,C) or embryos homozygous for snail (Df(2L)TE116GW11) (B,D) carrying the 2xPE::sna transgene. After its first phase of expression in the mesoderm, Snail is normally activated in the neuroectoderm in the progenitors of the central nervous system. This expression becomes detectable during germ band extension as the expression in the mesoderm begins to fade (A,C). The lack of snail expression in the ectoderm identifies embryos homozygous mutant for snail (B,D). The expression of snail in the mesoderm of these embryos is due to the expression of the transgene. Mesoderm invagination in such embryos is fully restored. Older homozygous Df(2L)TE116GW11 embryos (D) nevertheless look morphologically abnormal, presumably because of the failure of neurogenesis, for which snail and its paralogs escargot and worniu are required, all of which are deleted in the deficiency used here (Ashraf et al., 1999). (C,D, insets). Optical section at the level of the midline at higher magnification. Eve-positive midline neurons are seen in the wildtype butnot in the snail mutant. (E–G) Expression of fog in mutant and transgenic embryos. RNA in situ hybridization with a Dig labelled RNA probe on (E) wildtype, (F) twistEY53, or (G) twistEY53 embryos carrying the PE::fog transgene. The mesodermal expression of fog in twistEY53 mutant embryos is abolished. Wildtype levels of fog can be restored by the PE::fog transgene.

174

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

Fig. 6. Effect of expression of snail and fog in the mesoderm of twist mutant embryos. Top row, whole mount embryos stained against Eve. Six stages of gastrulation were defined by the position of the Even-skipped stripes (of which the first four are shown). The positions of the Eve stripes depend on morphogenetic events unconnected to ventral furrow formation. This allows the analysis of precisely the same stages in wildype embryos and mutants with perturbed mesoderm invagination. The bars below indicate the trunk regions from which transversal sections were analysed. Rows 2–7, cross sections of wildtype, snail twist double mutant (Df(2L)TE116GW11 Df(2R)S60) and twistEY53 mutant embryos, and twistEY53 mutant embryos expressing the indicated transgenes. The developmental stages of the sectioned embryos correspond to those of the whole mount embryos above. Rows 2–4, the comparison of wildtype and and twist mutants shows the range of residual morphogenetic movements in the mesodermal primordium of embryos without twist activity. This residual activity is due to the twist independent initial expression of snail, as indicated by the comparison to snail twist double mutants. For more details see text. Rows 5–7, twistEY53 mutants expressing the PE::fog and 2xPE::sna transgenes show cell shape changes or even invagination of the mesodermal primordium. The latter requires the activities of snail. For more details see text. Cell outlines were traced by hand to visualize cell shapes more clearly. The original photographs without tracings are shown in Supplementary Figure 2.

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

175

Fig. 7. Statistical analysis of the morphogenetic activity in twist mutants with or without the expression of twist targets in the mesodermal primordium. (A) Classification of morphogenetic activity within the mesodermal primordium. Furrows were classified as type 1 (no activity), type 2 (shallow furrow) or type 3 (deep invagination in which the apices of opposite cells make contact to each other). The number of sections on which the furrows had reached the various degrees at each of the stages shown in Fig. 5 are plotted. The mutant allele of twist was twistEY53. (B) Depths of ventral furrows of the indicated genotypes were evaluated according to the classification in (A). At least 15 sections from the central region of each embryo were scored, and at least 5 embryos of each genotype and stage were analysed. The values for all embryos of one stage and genotype were added up, and the proportion of Type I, Type II and Type III furrows is indicated in the bar diagrames and tables. All wildtype embryos show Type III furrows already at stage 0.

able to repress its ectodermal target genes within the mesoderm (not shown). To allow a precise evaluation of the effect of the Fog and Snail transgenes in twist mutants we first documented the twist mutant phenotype using a scoring system for the quality of the furrows (Figs. 6 and 7). It is particularly important to be able to judge the age of the embryo. Since

gastrulation movements are extremely fast during this period, and shapes change within minutes, even the most precisely timed egg collections contain too large a range of developmental stages. We therefore used the position of the pair-rule stripes marked by staining for Evenskipped as an internal measure for developmental age. These stripes are shifted by the movements of germ band

176

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

extension (cf. Fig. 6, top row) and provide a precise readout of developmental timing. Mesodermal morphogenesis in twist mutants is variable, ranging from embryos that show no detectable furrow to embryos showing regions with clear invaginations, albeit with highly abnormal cell shapes (Fig. 6). Cells become narrow at their apical ends and the nuclei may move towards the basal side, but strong constrictions are not seen, nor does apical flattening occur. The outer surface of the furrow is therefore never smooth. Coordinated cell behaviour along the whole length of the ventral side of the embryo leading to the characteristic structure of the furrow was never found in twist mutants. twist mutant embryos, although unable to form any mesoderm, show a characteristic residual invagination of the mesodermal primordium (Ip et al., 1994; Leptin, 1991; Leptin and Grunewald, 1990), which is transient and irregular. Twist mutants show residual snail activity deriving from an early, transient burst of Twist-independent expression of snail in a narrow band of mesodermal cells (Kosman et al., 1991; Leptin, 1991). When this residual expression is abolished in snail twist double mutants (Fig. 6, third row) no signs of furrow formation can be seen (Ip et al., 1994; Leptin and Grunewald, 1990). Thus, low levels of Snail are able to induce weak cell shape changes in the absence of Twist. The expression of high levels of Snail resulting from the 2xPE::sna transgene caused a significant improvement in ventral furrow formation in twist mutants. Most notably, furrows were regular and often so deep that cells from the two sides of the furrow faced each other with tightly apposed apical surfaces (Fig. 6). Cells showed clear apical constrictions and ventral nuclei moved away from the apical side of the cells. However, the furrows formed later than in wildtype embryos and were transient, as seen from the statistical evaluation of a large number of sections (Fig. 7). Although in more than half of the cases the sections of twi; 2xPE::sna embryos showed a deep furrow by stage 2, these furrows had mostly disappeared by stage 3. When Fog was expressed in the mesoderm of twist mutants, apical flattening of the mesodermal cells was restored, consistent with previous reports that expression of fog in blastodermal cells is sufficient to induce flattening (Morize et al., 1998). In addition, apical constriction was enhanced. The furrows in twi; PE::fog embryos thus look more regular with smoother apical surfaces than in twi mutant embryos and in many embryos the cells take on shapes that resemble the normal cell shapes seen in early wildtype ventral furrows (Fig. 6). It is important to remember that the observed furrows are the result of Fog in combination with the low level, Twist-independent, early Snail expression. When Fog was expressed in sna twi double mutants, no shape changes were observed (not shown), as has also been observed for fog expression in snail mutant embryos (Morize et al., 1998). This suggests that Snail is a permissive factor for induction of cell shape changes by Fog.

Finally, in twist mutant embryos expressing both transgenes together, a much larger number of embryos show deep invaginations of the mesoderm primordium. These are formed earlier than in embryos with only one transgene and they last longer (Figs. 6 and 7). Cells show the typical morphological characteristics of invaginating ventral furrow cells, such as strong apical constrictions and movement of the nuclei away from the apical sides of the cells. However, the furrows still appear later than a wildtype ventral furrow, and eventually disappear, cells failing to make contact with the basal side of the ectoderm (a process dependent on two Twist targets, htl and dof, that are not expressed in these embryos). Consistent with this, we find that the defects in embryos lacking all function of Snail and Fog are not as strong as those mutant for snail and twist (not shown), indicating that Snail and Fog together do not account for all functions downstream of Twist. Conversely, Fog expression in the absence of snail and twist function (i.e. in snail twist double mutants) is unable to induce any cell shape changes beyond apical flattening. In summary, this analysis shows that in the absence of twist function the activity of snail is able to direct some aspects of cell shape changes necessary to make an invagination, and the appearance of the furrow is accelerated if fog is also expressed. However, the stability and the timing of the mesoderm invagination are not fully restored, suggesting that these functions are mediated by additional Twist targets. In which way the genes uncovered by the deficiencies described above fulfill such a function can be tested in this system when the necessary reagents become available. 3. Discussion 3.1. Zygotic genes involved in ventral furrow formation We have found that in addition to Twist, Snail and Fog, there are genes in four regions of the autosomal genome which upon deletion lead to abnormalities during ventral furrow formation. Is it likely that we have detected all zygotically active genes that participate in normal mesoderm invagination? Although our assay proved to be sufficiently sensitive to identify the mutants described above, it is conceivable that further genes with even less pronounced mutant phenotypes were missed. Further, genes with completely redundant functions, for example, because duplicates exist in distinct regions of the genome, might not give a loss-of-function phenotype. Such genes might be identifiable only via sophisticated genetic screens (modifier screens) or appropriate molecular approaches. The loss of the genes uncovered by the deficiencies results only in a delay of furrow formation, and not in the complete failure of invagination. If these genes are Twist targets, there are different possible explanations for the mutants showing such weak phenotypes. The genes might control an essential process parallel to that controlled by Snail, but act in a redundant manner in the

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

pathway, such that disruption of only one of their functions does not lead to the disruption of furrow formation. Alternatively, only the pathway controlled by Snail may be essential, with other genes acting in parallel affecting only the speed and efficiency of furrow formation. The severe phenotype seen in the double mutants of Df(3R)TlP and fog as well as the enhancement of the fog phenotype by the loss of one copy of Snail argue for the latter scenario, i.e. two parallel pathways, both of which are essential. 3.2. snail and fog are sufficient to initiate mesoderm invagination We have created embryos in which the function of the mesodermal transcription factor twist was replaced by two of its downstream targets, snail and fog. We have concentrated our analysis of these embryos on the first phase of mesoderm morphogenesis, during which cell shape changes internalize the prospective mesoderm. The subsequent epithelial–mesenchymal transition, cell division and cell migration depend on other Twist targets, such as string, htl and dof (Edgar et al., 1994; Shishido et al., 1993; Vincent et al., 1998). As these are not expressed in the twist,PE::fog;2xPE::sna embryos, this aspect of morphogenesis cannot occur in the reconstituted twist embryos. The fog and snail transgenes had distinguishable effects in twist embryos. The PE::fog transgene induces the earliest event of the typical cell shape changes, apical flattening, and enhanced apical constrictions of ventral cells. By contrast, nuclear movement away from the apical cell surface was not significantly improved, nor was cell shortening observed. While the 2xPE::snail transgene also led to some improvement in apical flattening, it had additional effects. A larger number of cells showed distinctive nuclear movement, and a higher frequency of deep invaginations were scored, suggesting a role for Snail in releasing nuclei. With the combination of both transgenes nearly normal cell shape changes occurred which resulted in the formation of proper furrows in a substantial number of embryos. Specifically, many cells assumed the typical wedge-shape of ventral furrow cells, showing that snail and fog are sufficient to induce this shape in the absence of further Twist targets. However, it appears that snail and fog cannot be the only targets downstream of twist to control mesoderm invagination. If they were, they should be able to replace twist function completely and fully restore furrow formation. We therefore conclude that besides snail and fog other twist target genes must exist which are necessary to orchestrate the formation of the ventral furrow in the accurate, fast and stable fashion in wildtype embryos. The as yet unknown targets must be involved in those events which were not restored in twist,PE::fog;2xPE::sna embryos: the speed of the process, adhesion between apical surfaces, and cell shortening during invagination. The latter process occurs efficiently in snail mutants, confirming that it is not Snail-dependent (Leptin, 1991; Sweeton et al., 1991). Since fog does not contribute to this process

177

it is likely that one or more other twist targets are involved in cell shortening. 4. Conclusion In summary, we have shown that the zygotic control of ventral furrow formation branches into separable functions downstream of Twist, the induction of the basic cell shape changes of ventral cells, and the control of the speed, accuracy and coordination of the shape changes. One of the known targets of Twist, the repressor Snail, is necessary to allow the shape changes to occur, whereas Fog and probably other Twist targets are responsible for accuracy and efficiency. Together, they ensure the rapid and regular formation of the ventral furrow. Ventral furrow formation may be an adaptation to the rapid early embryogenesis of long germ insects, serving to position mesodermal cells at a site where they can efficiently begin their FGF-dependent spreading on the inner surface of the ectoderm. The experimental system used here may be extended to test the function of the whole set of Twist targets, once they have been identified, for their ability to re-establish mesoderm invagination in the absence of Twist, and thereby reconstruct fully the pathway from a ‘selector’ gene to the cell biological processes it controls. 5. Materials and methods 5.1. Fly stocks OregonR was used as the wildtype stock. Most of the fly stocks carrying autosomal deletions were obtained from the Bloomington Stock Center, Indiana, USA (Deficiency Kits for chromosomes 2 and 3 (DK2 and DK3)) and are described in FlyBase (http://flybase.bio.indiana.edu/). Additional fly stocks with deficient chromosomes, single mutations or Pelement insertions were obtained from T. Hummel & K. Schimmelpfeng (Kla¨mbt-Lab, Germany), the European Stock Centers in Umea˚, Sweden and Szeged, Hungary and from the stock collection of the MPI, Tu¨bingen, Germany. The twist null allele used for the analysis of twist function was twi[EY53] and was obtained from the stock collection of the MPI, Tu¨bingen, Germany. The snail null allele used to test the 2xPE::sna construct was Df(2L)TE116GW11. The snail twist double null mutants used for comparison to twist mutants carried a double deficient chromosome of the genotype: Df(2L)TE116GW11 Df(2R)S60. The latter three stocks were obtained from R. Reuter, Tu¨bingen, Germany. The amorphic allele of folded gastrulation to test the PE::fog construct was fog[4a6] and obtained from E. Wieschaus, Princeton, USA. The stock carrying the 2xPE::sna construct was obtained from Ip et al. (1994).

5.2. Molecular genetics The PE::fog transgenic construct was generated from a 2.9 kb fog cDNA in Bluescript (Costa et al., 1994) and the ‘pCaSper 2xPE beta Gal transformation vector’ (=‘‘2xPE vector’’) (Jiang et al., 1991; Jiang and Levine, 1993). The complete fog cDNA was excised from pBS using PvuII and inserted into the SmaI site of pSP72. It was re-excised from the vector with BglII and XbaI. This fragment was used to replace the BamHI–XbaI fragment containing the lacZ gene in the ‘‘2xPE vector’’ (Jiang et al., 1991; Jiang and Levine, 1993). This construct was then used for germline transformation. The fog cDNA was a gift from Mike Costa and Eric Wieschaus. The hkb::fog transformation construct was built in the pW8 transformation vector. First, the 0.8 kb XbaI/PstI fragment of

178

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179

PE was inserted between the XbaI and PstI site of the pW8 polylinker, giving rise to pW8-SV40. The 1 kb EcoRI/PstI fragment of the hkb-promoter (2 kb upstream of the hkb transcription start) was inserted into pBS, and the PstI site was destroyed and converted to a KpnI site using linkers. The hkb promoter was re-excised from this plasmid with EcoRI and KpnI and inserted between the EcoRI and KpnI site of the pW8 poly linker in pW8SV40, resulting in pW8-hkb-SV40. This vector was cut with KpnI and XbaI to accept a DNA fragment containing the heatshock basal promoter (a 0.4 kb XhoI/HindIII fragment) and the fog cDNA (a 2.9 kb HindIII/ XbaI fragment, which contains the polylinker fragment from EcoRI to XbaI of pBS after the 3 0 end of the fog cDNA). The huckebein cDNA was a gift from Bro¨nner et al. (1994). Transgenic flies were generated by germ line transformation by standard methods (Rubin and Spradling, 1983).

5.3. Antibody staining and RNA in situ hybridization Embryo collections and whole mount antibody staining were performed as described elsewhere (Reuter et al., 1990). We used the following antisera at the indicated dilutions: rabbit anti-Twi (from Roth et al., 1989, 1:4000), rabbit anti-Eve (from Frasch and Levine, 1987, 1:5000), monoclonal rat anti-N-Cadherin (from Iwai et al., 1997, 1:20), rabbit anti-Dof (Vincent et al., 1998, 1:200), rabbit anti-b-galactosidase (Cappel Research Products, USA , 1:5000), monoclonal mouse anti-Phospho-Tyrosine (#4G10, Deborah Morrison, 1:20). The presence of the antibodies in whole mount embryos was detected histochemically by using biotinylated secondary antibodies (anti-rabbit biotinylated, 1:500 and anti-rat biotinylated, 1:500, both from Dianova, Germany) and subsequent usage of the Vectastain ABC-Kit ‘Elite’ (Vector Labs., USA) with DAB (diamino-benzidine) as substrate. Stained embryos were transferred into Araldite as described previously (Leptin and Grunewald, 1990) and stored at 20 C. Embryos mounted in Araldite were examined under Zeiss microscopes with differential interference contrast optics. Photographs were taken using digital cameras (Kontron ProgRes 3008, Photometrix Quantix) and processed with Adobe Photoshop 5.5–7.0. Standard protocols were used for in situ hybridization with digoxigenin-labelled RNA probes (based on Tautz and Pfeifle, 1989).

5.6. Analysis of the enhancement of the embryonic phenotype of fog/Y by reduction of snail To create fog/Y; sna/+ embryos while avoiding non-specific effects on the phenotype by additional transgenes we used unmarked stocks (+/Y ; sna/+ x fog/+; sna/+), and the sna genotype of the embryos was therefore, inferred from statistics. Embryos in a narrowly staged collection stained with antibodies against Twist and Even-skipped were sorted into age and phenotypic classes. fog-embryos were identified by their failure to make a posterior midgut invagination. From a collection in which 133 embryos were at the correct stage, 33 were phenotypically fog, of which a randomly chosen subset of 19 were photographed and subsequently sectioned. Nine or ten of the embryos are expected to be sna/+, with the remaining embryos evenly split between +/+ and sna/sna. Five embryos (from the middle of the phenotypic range, and therefore most likely sna/ +) are shown in Fig. 2. In a separate experiment the embryos were hybridized with an antisense probe for singleminded, which allows the unambiguous identification of the sna/sna embryos because singleminded is expanded into the mesodermal domain. When this was used to sort out the sna/sna embryos from among the fog/Y embryos the majority of the remaining embryos showed the same phenotype as those shown in Fig. 2.

Acknowledgements We are grateful to the Bloomington, Umea˚ and Szeged stock centers and the MPI in Tu¨bingen for supplying fly stocks. We thank T. Ip, R. Reuter, A. Mu¨ller, E. Wieschaus, S. Roth, M. Frasch, T. Uemura, D. Morrison, G. Bro¨nner, H. Ja¨ckle for DNA, flies and antibodies. We are grateful to J. Hancke for help with the artwork. We thank A. Mu¨ller and E. Wieschaus for communicating data prior to publication, A. Csiszar, V. Riechmann, S. Roth, F. Sprenger and R. Wilson for critical reading of the manuscript and valuable suggestions. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 243 and SFB 572).

5.4. Screening for mutant phenotypes

Appendix A. Supplementary data To be able to detect subtle and transient disturbances of ventral furrow formation it was important to use a sufficiently sensitive assay. We used antibodies directed against Twist and Even-skipped (Eve) to stain the embryos. The Eve expression pattern is useful to judge developmental timing independent of mesodermal behaviour (Fig. 1A, Fig. 3, top, Supplementary Fig. 1). Eve is expressed in seven stripes in the ectoderm (Frasch and Levine, 1987; Ingham et al., 1988) whose spatial relationships to each other begin to change soon after gastrulation begins, due to the ectodermal movements during germ band extension. Furthermore, the first stripe of Eve is located in the cephalic furrow, which invaginates at the same time as the ventral furrow (Vincent et al., 1997), and therefore, serves as a good temporal marker. The combination of these two antibodies provides a coordinate system of the blastoderm embryo that allows to detect irregularities of the invagination of the mesoderm and the specificity of putative mesodermal phenotypes in contrast to general morphological defects. Flies from each deficiency stock were reared in large numbers to allow the collection of sufficient embryos over one-hour collection periods. The embryos were then aged for another 2 h, fixed and stained, and embedded for detailed observation.

5.5. Embryo sectioning Embryos embedded in Araldite were positioned and incubated over night at 60 C to polymerise the Araldite and sectioned with a Leica 2065 Supercut Microtome. Sections were mounted in Araldite, incubated again at 60 C over night and photographed as described for whole mount embryos.

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.mod. 2006.12.004. References Ashraf, S.I., Hu, X., Roote, J., Ip, Y.T., 1999. The mesoderm determinant snail collaborates with related zinc-finger proteins to control Drosophila neurogenesis. EMBO J. 18, 6426–6438. Barrett, K., Leptin, M., Settleman, J., 1997. The Rho GTPase and a putative RhoGEF mediate a signalling pathway for cell shape changes in Drosophila gastrulation. Cell 91, 905–915. Bro¨nner, G., Chu-LaGraff, Q., Doe, C.Q., Cohen, B., Weigel, D., Taubert, H., Jackle, H., 1994. Sp1/egr-like zinc-finger protein required for endoderm specification and germ-layer formation in Drosophila. Nature 369, 664–668. Casal, J., Leptin, M., 1996. Identification of novel genes in Drosophila reveals the complex regulation of early gene activity in the mesoderm. Proc. Natl. Acad. Sci. USA 93, 10322–10327. Costa, M., Wilson, E., Wieschaus, E., 1994. A putative cell signal encoded by the folded gastrulation gene coordinates cell shape changes in Drosphila gastrulation. Cell 76, 1075–1089. Edgar, B.A., Lehman, D.A., O’Farrell, P.H., 1994. Transcriptional regulation of string (cdc25): a link between developmental programming and the cell cycle. Development 120, 3131–3143.

T.C. Seher et al. / Mechanisms of Development 124 (2007) 167–179 Frasch, M., Levine, M., 1987. Complementary patterns of even-skipped and fushi tarazu expression involve their differential regulation by a common set of segmentation genes in Drosophila. Genes Dev. 1, 981– 995. Furlong, E.E., Andersen, E.C., Null, B., White, K.P., Scott, M.P., 2001. Patterns of gene expression during Drosophila mesoderm development. Science 293, 1629–1633. Grosshans, J., Wieschaus, E., 2000. A genetic link between morphogenesis and cell division during formation of the ventral furrow in Drosophila. Cell 101, 523–531. Grosshans, J., Mu¨ller, H.A., Wieschaus, E., 2003. Control of cleavage cycles in Drosophila embryos by fru¨hstart. Dev. Cell 5, 285–294. Ingham, P.W., Baker, N.E., Martinez-Arias, A., 1988. Regulation of segment polarity genes in the Drosophila blastoderm by fushi tarazu and even skipped. Nature 331, 73–75. Ip, Y.T., Park, R.E., Kosman, D., Yazdanbakhsh, K., Levine, M., 1992. dorsal-twist interactions establish snail expression in the presumptive mesoderm of the Drosophila embryo. Genes Dev. 6, 1518–1530. Ip, Y.T., Maggert, K., Levine, M., 1994. Uncoupling gastrulation and mesoderm differentiation in the Drosophila embryo. EMBO J. 13, 5826–5834. Iwai, Y., Usui, T., Hirano, S., Steward, R., Takeichi, M., Uemura, T., 1997. Axon patterning requires DN-cadherin, a novel neuronal adhesion receptor, in the Drosophila embryonic CNS. Neuron 19, 77–89. Jiang, J., Levine, M., 1993. Binding affinities and cooperative interactions with bHLH activators delimit threshold responses to the dorsal gradient morphogen. Cell 72, 741–752. Jiang, J., Kosman, D., Ip, Y.T., Levine, M., 1991. The dorsal morphogen gradient regulates the mesoderm determinant twist in early Drosophila embryos. Genes Dev. 5, 1881–1891. Kosman, D., Ip, Y.T., Levine, M., Arora, K., 1991. Establishment of the mesoderm-neuroectoderm boundary in the Drosophila embryo. Science 254, 118–122. Ko¨lsch, V., Seher, T., Fernandez-Ballester, G.J., Serrano, L., Leptin, M., 2007. Control of Drosophila gastrulation by apical localization of adherens junctions and RhoGEF2. Science 315, in press. Leptin, M., 1991. twist and snail as positive and negative regulators during Drosophila mesoderm development. Gene. Dev. 5, 1568– 1576. Leptin, M., Grunewald, B., 1990. Cell shape changes during gastrulation in Drosophila. Development 110, 73–84. Maggert, K., Levine, M., Frasch, M., 1995. The somatic-visceral subdivision of the embryonic mesoderm is initiated by dorsal gradient thresholds in Drosophila. Development 121, 2107–2116. Mata, J., Curado, S., Ephrussi, A., Rorth, P., 2000. Tribbles coordinates mitosis and morphogenesis in Drosophila by regulating string/CDC25 proteolysis. Cell 101, 511–522. Merrill, P.T., Sweeton, D., Wieschaus, E., 1988. Requirements for autosomal gene activity during precellular stages of Drosophila melanogaster. Development 104, 495–509.

179

Morize, P., Christiansen, A.E., Costa, M., Parks, S., Wieschaus, E., 1998. Hyperactivation of the folded gastrulation pathway induces specific cell shape changes. Development 125, 589–597. Mu¨ller, H., Samanta, R., Wieschaus, E., 1999. Wingless signaling in the Drosophila embryo: zygotic requirements and the role of the frizzled genes. Development 126, 577–586. Postner, M.A., Wieschaus, E.F., 1994. The nullo protein is a component of the actin-myosin network that mediates cellularization in Drosophila melanogaster embryos. J. Cell Sci. 107, 1863–1873. Poulson, D.F., 1937. Chromosomal deficiencies and the embryonic development of Drosophila melanogaster. PNAS 23, 133–137. Reuter, R., Panganiban, G.E., Hoffmann, F.M., Scott, M.P., 1990. Homeotic genes regulate the spatial expression of putative growth factors in the visceral mesoderm of Drosophila embryos. Development 110, 1031–1040. Roth, S., Stein, D., Nusslein-Volhard, C., 1989. A gradient of nuclear localization of the dorsal protein determines dorsoventral pattern in the Drosophila embryo. Cell 59, 1189–1202. Rubin, G.M., Spradling, A.C., 1983. Vectors for P element-mediated gene transfer in Drosophila. Nucleic Acids Res. 11, 6341–6351. Seher, T.C., Leptin, M., 2000. Tribbles, a cell-cycle brake that coordinates proliferation and morphogenesis during Drosophila gastrulation. Curr. Biol. 10, 623–629. Shishido, E., Higashijima, S., Emori, Y., Saigo, K., 1993. Two FGFreceptor homologues of Drosophila: one is expressed in mesodermal primordium in early embryos. Development 117, 751–761. Simpson, P., 1984. Mutations and chromosomal rearrangements affecting the expression of Snail, a gene involved in embryonic patterning in Drosophila melanogaster. Genetics 108, 347–360. Stathopoulos, A., Van Drenth, M., Erives, A., Markstein, M., Levine, M., 2002. Whole-genome analysis of dorsal-ventral patterning in the Drosophila embryo. Cell 111, 687–701. Sweeton, D., Parks, S., Costa, M., Wieschaus, E., 1991. Gastrulation in Drosophila: the formation of the ventral furrow and the posterior midgut invaginations. Development 112, 775–789. Tautz, D., Pfeifle, C., 1989. A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback. Chromosoma 98, 81–85. Taylor, M.V., 2000. A novel Drosophila, mef2-regulated muscle gene isolated in a subtractive hybridization-based molecular screen using small amounts of zygotic mutant RNA. Dev. Biol. 220, 37–52. Vincent, A., Blankenship, J.T., Wieschaus, E., 1997. Integration of the head and trunk segmentation systems controls cephalic furrow formation in Drosophila. Development 124, 3747–3754. Vincent, S., Wilson, R., Coelho, C., Affolter, M., Leptin, M., 1998. The Drosophila protein Dof is specifically required for FGF signaling. Mol. Cell 2, 515–525. Zusman, S.B., Wieschaus, E.F., 1985. Requirements for zygotic gene activity during gastrulation in Drosophila melanogaster. Dev. Biol. 111, 359–371.