SEARCH

SEARCH BY CITATION

Keywords:

  • dorsal closure;
  • germ band retraction;
  • JNK signaling;
  • Ricin;
  • Drosophila;
  • embryogenesis

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Dorsal closure in the fruit fly Drosophila melanogaster is a complex morphogenetic process, driven by sequential signaling cascades and involving multiple forces, which contribute to cell movements and rearrangements as well as to changes in cell shape. During closure, lateral epidermal cells elongate along the dorsoventral axis and subsequently spread dorsally to cover the embryonic dorsal surface. Amnioserosal cells, which are the original occupants of the most dorsal position in the developing embryo, constrict during closure; thus, the increase in epidermal surface area is accommodated by a reduction in the amnioserosal surface area. Several of the epidermal requirements for closure have been established in functional assays. In contrast, amnioserosal requirements for closure have remained elusive, in part because laser ablation and clonal approaches are limited to only subsets of amnioserosal cells. Here, we report our use of the UAS-GAL4 system to target expression of the cell autonomous toxin Ricin-A to all cells of the amnioserosa. We show that ablation of the amnioserosa leads to clear defects in dorsal closure and, thus, directly demonstrate a role for the amnioserosa in dorsal closure. We also show that DJNK (Drosophila Jun N-terminal kinase) signaling, an epidermal trigger of closure, is unaffected by amnioserosal ablation. These data, together with our demonstration that amnioserosal ablated and Dpp signaling mutant embryos exhibit shared loss-of-function phenotypes, point to a requirement for the amnioserosa in dorsal closure that is downstream of Dpp, perhaps as part of a paracrine response to this signaling cascade. Developmental Dynamics 232:791–800, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

The directed rearrangement of cells is a major morphogenetic force in the early development of virtually all animal forms. Two well-studied examples of the many developmental processes dependent upon changes in cell shape and position are teleost epiboly (Keller and Trinkaus, 1987) and vertebrate neurulation (Schoenwolf and Alvarez, 1989). In the invertebrate Caenorhabditis elegans, the role that cell rearrangements play in closure of the embryonic hypodermis is also well documented (Williams-Masson et al., 1998), and in the fruit fly Drosophila melanogaster, well-studied cell rearrangements are associated with the morphogenetic processes of germ band retraction and dorsal closure (Young et al., 1993; Campos-Ortega and Hartenstein, 1997). For both germ band retraction and dorsal closure, cell shape transformations effect changes in epidermal surface area. Moreover, accumulating evidence suggests that the epidermal cell shape changes that typify both germ band retraction and dorsal closure are regulated analogously by structural and/or molecular signals that originate in an adjacent tissue—the amnioserosa (Lamka and Lipshitz, 1999; Harden, 2002; Kozlova and Thummel, 2003; Schock and Perrimon, 2003).

Dorsal closure in developing Drosophila embryos occurs between 8 and 12 hr after egg lay (AEL), corresponding to embryonic stages 13–15 (Campos-Ortega and Hartenstein, 1997). During this process, epidermal cells change shape, becoming rectangular in dorsal positions and remaining polygonal in more lateral positions (Young et al., 1993). These cell shape changes are initiated in the row of epidermal cells called the leading edge, which abuts the amnioserosa; elongation of epidermal cells positioned more laterally occurs subsequently in a processive manner from dorsal epidermis to lateral. Widespread epidermal cell elongation leads ultimately to extension of the epidermal sheet over the dorsal surface of the embryo. Amnioserosa, the original occupant of the most dorsal position in the developing embryo, is internalized and is thought to degenerate after closure (Hartenstein and Jan, 1992). Closure is concluded with the juxtaposition and fusion of the dorsal epithelial sheets at the dorsal midline. Upon completion of closure, the epidermis fully envelops the embryo and secretes the larval cuticle; thus, in wild-type larvae, the cuticle encases the animal. In contrast, cuticles derived from dorsal-open group mutants are grossly abnormal, as evidenced by a single, large hole in the dorsal epidermis.

The dorsal-open group is composed mainly of mutants originally identified on the basis of their dorsal-open cuticle (Nusslein-Volhard and Wieschaus, 1980). Molecular analyses of close to all of these loci revealed that they can be separated, based on functional similarity, into at least three classes. The class I and II gene products encode DJNK (Drosophila Jun N-terminal kinase) and Dpp/TGF-β (Decapentaplegic/ transforming growth factor-β) signaling components, respectively; the class III dorsal-open group gene products comprise structural molecules required for closure (for review see Jacinto et al., 2002; VanHook and Letsou, 2003).

Epidermal-specific functions have long been thought sufficient to drive dorsal closure. This idea was largely based on the correlative observation that the first 15 genes of the dorsal-open group to be studied molecularly were all found to be expressed in the closing epidermis. This said, at least seven lines of evidence (albeit largely indirect) are suggestive of a more complex mechanism of closure, one in which the amnioserosa plays a primary role. First, dorsal closure defects in transgenic flies bearing a dominant inhibitory version of Rac1 are associated with disrupted accumulation of F-actin, myosin, and α-spectrin in both leading edge epidermal and amnioserosal cells (Harden et al., 1995; Harden, 2002). Second, transient peripheral actin microspikes that extend from the leading edge over the amnioserosa during closure represent potential focal adhesion sites between the two tissues (Harden et al., 1996). Third, hyperactivation of JNK signaling in the amnioserosa disrupts the ability to form focal adhesion complexes at the boundary between leading edge and amnioserosa, potentially resulting in miscommunication between the two tissues (Reed et al., 2001). Fourth, dorsal closure defects in raw mutants are associated with defects in amnioserosal gene expression (Byars et al., 1999). Fifth, amnioserosal cell constriction is observed during closure, and this tissue specific cell shape change could represent a primary role for the amnioserosa in propelling closure (Jacinto and Martin, 2001). Sixth, laser ablation studies have revealed a structural requirement for intracellular tension in the amnioserosa during dorsal closure. A notable consequence of artificially generated holes in the amnioserosa is localized relaxation of the adjacent epidermis, indicating that forces in the amnioserosa are required to maintain epidermal tension and stretching (Kiehart et al., 2000). Finally, it appears likely that one important function of integrins during dorsal closure is to effect amnioserosal contraction (Narasimha and Brown, 2004; Reed et al., 2004). Taken together, although these data lay a solid foundation for developmental models in which the amnioserosa plays structural and/or signaling roles during dorsal closure, direct tests of the models remain essential. Indeed, this is one important objective of the study reported here.

Germ band retraction is a comparatively less well-studied morphogenetic event that initiates slightly earlier than does dorsal closure. When viewed as a cell biological process, however, retraction resembles closure in many important respects. Retraction, like closure is dependent upon coordinated changes in cell shape and position, albeit along the anteroposterior (AP) rather than the dorsoventral (DV) axis, and the amnioserosa is externalized rather than internalized. During retraction, between 7 and 9 hr AEL, cells of the dorsal epithelium change their shape, contracting along their AP axis, consequently leading to a shortening of the germ band (Campos-Ortega and Hartenstein, 1997). As the germ band retracts over the posterior tip of the embryo, the embryonic dorsal surface and the underlying amnioserosa are exposed. As for dorsal closure, cuticular morphology again provides an excellent marker for identifying mutants defective in germ band retraction. Germ band retraction mutants secrete a distinctive u-shaped cuticle, in which the tail remains up (Frank and Rushlow, 1996). Moreover, it has long been known that each of the germ band retraction (or u-shaped) mutants originally defined by Frank and Rushlow (1996) codes for an amnioserosal gene product (FlyBase, 2003).

In contrast to its presumed role in dorsal closure, a role for the amnioserosa in germ band retraction has been demonstrated directly and convincingly. In this regard, mosaic embryos that lack amnioserosal expression of hnt, a u-shaped gene which codes for a nuclear Zinc finger protein, fail to undergo germ band retraction, whereas mosaic embryos, in which amnioserosal hnt remains unperturbed, complete germ band retraction normally. Furthermore, hnt clones adjacent to the epidermis affect retraction more than similarly sized clones that do not abut the epidermis. These data are suggestive of a special interaction between the amnioserosa and the epidermis that is mediated, at least in part, by the hnt-encoded nuclear factor (Lamka and Lipshitz, 1999).

Despite striking similarities in the morphogenetic processes of retraction and closure, molecular mechanisms driving retraction have not been considered necessary for closure; it is generally accepted that mutations disrupting retraction do not alter patterns of closure. This finding is certainly true for null mutations of the founding members of the u-shaped class, including hnt, tail-up (tup), u-shaped (ush), and serpent (srp). However, it is now clear that hypomorphic alleles of hnt are variably expressed, as a significant fraction manifest defects in dorsal closure (Reed et al., 2001).

In the current report, we describe targeted cell ablation studies that were designed to directly assess the requirements for amnioserosal and leading edge cells during mid-embryonic development. Our studies revealed that amnioserosal cell ablation has striking effects on embryonic development, leading to archetypal defects not only in germ band retraction but also in dorsal closure. In this regard, we have shown that ablation of the amnioserosa (AS) in an AS-GAL4:UAS-ricin transgenic animal leads to defects in germ band retraction and dorsal closure. As a complement to these studies, we used genetic methods to disrupt embryonic morphogenesis. We observed that null mutations in the dorsal-open group gene thickveins (tkv), while fully penetrant with respect to embryonic lethality, are variably expressed and exhibit a range of dorsal closure and germ band retraction phenotypes similar to that which we observe after amnioserosal ablation. Taken together, our targeted cell ablation and genetic disruption data point to a link(s) between the temporally coincident morphogenetic processes of retraction and closure. Furthermore, it is clear that one of these developmental links is the amnioserosa.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Ricin-A–Mediated Ablation Is Cell Autonomous

Ricin-A–mediated genetic cellular ablation can be a useful tool for developmental studies in Drosophila (Sentry et al., 1993; Hidalgo et al., 1995). To confirm both the efficiency and the cell autonomy of Ricin-A–induced toxicity in Drosophila, we examined Ricin-A–induced cell ablation in the fly eye. To this end, we used the GMR-GAL4 fly line to drive expression of either a Ricin-A toxin gene, UAS-ricin, or a reporter gene, UAS-GFP, in the fly eye. In this system, cells killed by Ricin-A corresponded to those in which the UAS-GFP reporter was induced; transgenic animals harboring both the GMR-GAL4 and UAS-ricin transgenes were eyeless (Fig. 1). Furthermore, although all cells of the eye and ocelli in which the toxin was expressed were ablated, surrounding tissues were unaffected. This one-to-one correspondence of Ricin-A–induced ablation with GFP reporter gene expression confirms and extends results of previous experiments exploiting Ricin-A to induce cell death in Drosophila embryos (Hidalgo et al., 1995; Hidalgo and Brand, 1997; Allen et al., 2002; Roman and Davis, 2002).

thumbnail image

Figure 1. Cell-specific ablation in the eye. The GMR-GAL4 insertion line was used to drive expression of either the toxin gene UAS-ricin or the reporter gene UAS-GFP in the eye. A,C: Transgenic animals carrying the GMR-GAL4 driver alone retained wild-type eye morphology (A) and did not express GFP (C). B: Animals harboring both the GMR-GAL4 and the UAS-ricin transgenes were eyeless. D: Cells killed by Ricin-A correlated with those in which the UAS-GFP reporter was induced, including the eyes and the ocelli (arrows). Photographs in A and B were taken using standard light microscopy; images in C and D were obtained by confocal microscopy.

Download figure to PowerPoint

Targeted ricin Expression Induces Ablation of the Amnioserosa

Having confirmed the cell autonomous toxicity of the UAS-ricin transgene, we next coupled UAS-ricin with amnioserosal drivers to assess the effects of amnioserosal integrity on embryonic development. We used two independent amnioserosal-GAL4 (AS-GAL4) lines, C381-GAL4 and Kr-GAL4 (Krüppel-GAL4), to drive expression of either UAS-ricin or a reporter gene, UAS-lacZ, in the amnioserosa.

After introduction of the UAS-lacZ reporter into each of the AS-GAL4 fly lines by mating, β-galactosidase expression was monitored using an anti-β-galactosidase–specific antibody. We found that both drivers are highly effective in driving UAS-lacZ expression in the amnioserosa. Amnioserosal β-galactosidase activity was observed in both C381-GAL4/UAS-lacZ and Kr-GAL4/UAS-lacZ stage 9/10 transgenic embryos (47% and 74% of stage 9/10 embryos, respectively; Fig. 2A–C,G). Moreover, in both transgenic lines, β-galactosidase expression remained high until the end of dorsal closure, at stage 15. Potentially informative differences in the reporter gene expression profiles included staining in the head of C381-GAL4/UAS-lacZ transgenics. In addition to amnioserosal expression, Kr-GAL4 epidermal expression was localized to abdominal domains of the extended germ band. We considered the possibility that this pattern might represent residual Kr gap gene expression; however, there is no evidence of Kr-GAL4 driver expression before embryonic stage 9. Finally, we also determined that the C381- and the Kr-GAL4 drivers similarly promote amnioserosal expression of the UAS-ricin transgene, even though Ricin is not normally found in this tissue (Fig. 2D–F).

thumbnail image

Figure 2. C381-GAL4 and Kr-GAL4 drive expression of UAS-lacZ and UAS-ricin in the amnioserosa. A–F: Antibodies directed against β-galactosidase (A–C) or Ricin (D–F) were used to assess gene activity in wild-type (wt, A,D) and mutant embryos (B,C,E,F). Whereas wild-type embryos express neither β-galactosidase (A) nor Ricin (D), both are expressed in the amnioserosa of C381-GAL4 (B,E) and Kr-GAL4 (C,F) transgenics. G: A table documents quantitation of lacZ expression patterns. In these and subsequent panels of embryos, dorsal is up, and anterior is to the left.

Download figure to PowerPoint

Both C381- and Kr-GAL4 drivers effectively lead to ablation of the amnioserosa in UAS-ricin transgenic embryos. This determination is based upon our examination of two molecular markers: the first for cell death—reaper (rpr), the second for amnioserosal fate—Krüppel (Ashburner, 1989; Gaul and Weigel, 1990; Yip et al., 1997). rpr expression, although not evident in the amnioserosa of wild-type embryos, is robust in germ band retraction mutants and precedes cell death (Fig. 3A,C,G). In similar manner, Krüppel protein is differentially expressed in wild-type and germ band retraction class mutant embryos. Whereas Krüppel is localized to wild-type amnioserosal nuclei during germ band retraction and dorsal closure stages of embryogenesis, the absence of amnioserosal cells in mutants belonging to the u-shaped class is correlated with an absence of Krüppel staining (Fig. 3B,D,H).

thumbnail image

Figure 3. A–J: Markers of amnioserosal cell death. rpr expression was used as a marker for apoptosis (A,C,E,G,I) and Kr expression was used to mark amnioserosal cells (B,D,F,H,J) in wild-type (A,B), ush2 (C,D), hnt1 (G,H), C381-GAL4/UAS-ricin (E,F), and Kr-GAL4/UAS-ricin (I,J) embryos. A,C,E,G,I: rpr expression, undetectable in the amnioserosa of wild-type embryos (A), is induced in the germ band retraction mutants (C,G) and in response to Ricin-A (E,I). B,D,F,H,J: Kr expression normally seen in the amnioserosa (B) is lost in germ band retraction (D,H) and transgenic (F,J) mutant embryos.

Download figure to PowerPoint

As we observed for u-shaped mutant controls, the coincident induction of rpr and loss of Krüppel was clearly evident in AS-GAL4/UAS-ricin transgenic embryos. In particular, we observed rpr induction in the amnioserosa of both C381-GAL4/UAS-ricin and Kr-GAL4/UAS-ricin transgenics, although rpr expression, like that of lacZ, was evident earlier in Kr-GAL4/UAS-ricin animals (Fig. 3E,I). This temporal distinction is reminiscent of that which differentiates rpr expression in the u-shaped class mutant hnt from that in ush. In this regard, rpr expression is robust in the amnioserosa of stage 9 hnt and Kr-GAL4/UAS-ricin mutants but not detectable in the amnioserosa of most ush and C381-GAL4/UAS-ricin embryos until stage 11. Thus, it appears that the amnioserosa dies earlier in hnt and Kr-GAL4/UAS-ricin mutants than it does in ush and C381-GAL4/UAS-ricin mutants.

In accordance with standard convention, we also used Krüppel expression as a measure of amnioserosal viability in AS-GAL4/UAS-ricin transgenics. Results from this analysis were in agreement with those from our rpr analysis. We found that both C381-GAL4/UAS-ricin and Kr-GAL4/UAS-ricin animals exhibited little to no Krüppel staining in embryonic domains fated for amnioserosal development (Fig. 3F,J). As we observed to be the case for rpr, there were subtle differences in Krüppel expression in the two transgenic lines: Kr-GAL4/UAS-ricin mutants, which exhibit the earlier pattern of rpr expression, were devoid of Krüppel staining and, thus, we infer devoid of amnioserosal cells as well. In contrast, in C381-GAL4/UAS-ricin animals, where robust rpr induction occurs slightly later, a small number of Krüppel-positive amnioserosal cells occasionally persisted.

Ablation of the Amnioserosa Leads to Defects in Germ Band Retraction and Dorsal Closure

Embryos in which ricin was expressed under the control of either of the amnioserosal drivers exhibited a fully penetrant embryonic lethality that was not observed in the absence of ricin expression. Analyses of cuticles derived from C381-GAL4/UAS-ricin transgenic animals revealed that virtually all of this embryonic lethality could be traced to defects in germ band retraction. This finding was not unanticipated, as amnioserosal ablation is expected to yield phenotypes identical to those associated with null mutations in amnioserosal-specific gene products. Of C381-GAL4/UAS-ricin embryos, 95% exhibited a u-shaped embryonic lethal phenotype, indistinguishable from that which is seen in animals homozygous for loss-of-function alleles of ush, the prototype of the u-shaped group (Fig. 4C,D). Intriguingly, the remaining 5% of C381-GAL4/UAS-ricin embryos exhibited defects in dorsal closure (Fig. 6D).

thumbnail image

Figure 4. Expression of UAS-ricin in the amnioserosa produces germ band retraction and dorsal closure defects. A: A wild-type cuticle pattern is shown. C,D: Cuticles derived from C381-GAL4/UAS-ricin transgenic embryos revealed a germ band retraction phenotype (C) that is identical to that of the u-shaped group mutant ush2 (D). In both cases, posterior segments fail to retract (arrows). E–G:Kr-GAL4/UAS-ricin transgenics displayed a range of defects including a slight pucker of the dorsal cuticle (E, asterisk) to a dorsal hole (G, arrow), reminiscent of the defects observed in the dorsal-open group mutant put62 (F, asterisk, H arrow). B: Cuticles derived from Kr mutants display a gap phenotype that is different from the defects seen in Kr-GAL4/UAS-ricin transgenics (E,G).

Download figure to PowerPoint

thumbnail image

Figure 6. Germ band retraction defects are evident in tkv mutants. A–C: Cuticles derived from the dorsal-open group mutant display phenotypic variability similar to that seen in AS-GAL4/UAS-ricin transgenic animals, including weak dorsal closure defects (A), strong dorsal closure defects (B), and germ band retraction defects (C). D: Phenotypic variation in tkv and transgenic mutant animals is quantified and compared.

Download figure to PowerPoint

Although fully penetrant with regard to embryonic lethality, the Kr-GAL4/UAS-ricin lethality differed significantly from that of C381-GAL4/UAS-ricin transgenics. Whereas 16% of Kr-GAL4/UAS-ricin cuticles exhibited the u-shaped phenotype, 84% revealed clear defects in the process of dorsal closure (Figs. 4E,G, 6D). Dorsal closure defects in Kr-GAL4/UAS-ricin mutants ranged from weak (a puckered dorsal midline) to strong (a dorsal hole) and were indistinguishable from those observed in embryos homozygous for loss-of-function alleles of punt (put), a well-characterized member of the dorsal-open group encoding the Drosophila type II Dpp receptor (Fig. 4F,H). For both C381-GAL4/UAS-ricin and Kr-GAL4/UAS-ricin mutants, cuticular phenotypes were clearly distinct from those of wild-type and Kr mutants (Fig. 4A,B).

Embryos Complete Development After Ricin Is Expressed in Leading Edge Epidermal Cells

Cumulative evidence supports a model in which interactions between the amnioserosa and the leading edge of the epidermis are essential for dorsal closure and germ band retraction (Byers et al., 1987; Lamka and Lipshitz, 1999; Reed et al., 2001; Stronach and Perrimon, 2001). To investigate the effect of specific ablation of the epidermal leading edge cells, UAS-ricin was expressed under the control of a leading edge GAL4 (LE-Gal 4) driver. This driver has been used previously in studies of dorsal closure to examine ectopic expression of genes required for closure (Lu and Settleman, 1999). Although β-galactosidase activity was detected in leading edge cells in LE-GAL4/UAS-lacZ transgenic animals, we saw no effect of ricin-A–induced toxicity in the leading edge on either germ band retraction or dorsal closure in LE-GAL4/UAS-ricin transgenic animals (data not shown).

Leading Edge JNK Signaling Is Independent of Amnioserosal Integrity

We next sought to test the idea that a signal emanating from the amnioserosa is required to regulate DJNK signaling in adjacent leading edge epidermal cells. Whereas DJNK signaling is restricted to leading edge cells in wild-type animals, it is abolished in leading edge cells in hep (hemipterous, which codes for DJNKK), bsk (basket, which codes for DJNK), and Jra (Jun-related antigen, which codes for DJun) dorsal-open group mutant embryos (Glise and Noselli, 1997; Hou et al., 1997; Riesgo-Escovar and Hafen, 1997), and it is ectopically activated in raw (a novel protein), rib (ribbon, which codes for a putative BTB-type transcription factor), and puc (puckered, which codes for a DJNK phosphatase) dorsal-open group mutant embryos (Blake et al., 1998; Martin-Blanco et al., 1998; Byars et al., 1999). It is notable that both the loss and the gain of DJNK signaling can be causative with respect to gross defects in dorsal closure.

For the studies described here, we exploited expression of dpp, a well-characterized transcriptionally regulated target of DJNK signaling to monitor DJNK activity during closure. Animals homozygous for AS-GAL4 drivers were mated to UAS-ricin/CyO-lacZ flies, and embryos resulting from these crosses were doubly stained, allowing us to compare dpp expression patterns in lacZ-positive, wild-type embryos and lacZ-negative, amnioserosal ablated mutant embryos. In 97% (n = 35) of stage 11–12 lacZ-positive, wild-type embryos, dpp transcription was restricted to leading edge epidermal cells (Fig. 5A, page 796). In analogous manner, in C381-GAL4/UAS-ricin mutant embryos (n = 61), we found that leading edge dpp expression was either indistinguishable from wild-type (44% of mutants) or that dpp expression was evident not only in leading edge epidermal cells but in epidermal cells positioned one to two rows more ventrally as well (56% of mutants; Fig. 5B,D). Similar results were obtained in our analysis of Kr-GAL4/UAS-ricin mutants: 11 of 11 lacZ-positive, wild-type embryos exhibited leading edge restricted dpp transcription. Mutants (n = 8) similarly exhibited wild-type patterns of leading edge restricted dpp transcription, often, however, within the context of epidermal expression in abdominal domains of the extended germ band (Fig. 5C). Although defects in DV patterning can block leading edge formation (Stronach and Perrimon, 2001), our results lead us to conclude that later blocks to amnioserosal development have no effect on establishment of a dpp-expressing leading edge. From the perspective of tissue interactions that might be important for dorsal closure, these data indicate that, if the amnioserosa is the source of a signal required for closure, this signal must be generated at or earlier than stage 9 of embryogenesis, when we observe amnioserosal cell death in Kr-GAL4/UAS-ricin mutants.

thumbnail image

Figure 5. Leading edge dpp expression is evident in amnioserosal ablated animals. In situ hybridization to whole mount embryos with a digoxigenin-labeled probe for dpp. A: Transcription of dpp in germ band extended stage wild-type (wt) embryos includes specific expression at the leading edge of the dorsal epidermis. C: This same pattern of leading edge expression was seen in Kr-Gal4:UAS-ricin mutant animals. B,D: A wider band of epidermal cells, which expresses dpp, was observed in rawIG (B) and C381-Gal4:UAS-ricin (D) mutant embryos.

Download figure to PowerPoint

Dorsal-Open and U-Shaped Mutants Exhibit Shared Loss-of-Function Phenotypes

Our demonstration that germ band retraction and dorsal closure defects are observed in C381- and Kr-GAL4/UAS-ricin transgenic embryos led us to speculate that the amnioserosa is required for both types of morphogenetic movements. To address whether this phenotypic duality was specific to our targeted ablation approach, or if a similar phenotypic duality occurs in genetic mutants, we examined the cuticle phenotypes of a classically defined dorsal-open group mutant. Because punt is haploinsufficient (Simin et al., 1998) and because null alleles of dpp produce a very early embryonic lethality due to defects in DV patterning (Irish and Gelbart, 1987), we chose tkv, which codes for the type I Dpp receptor, for this study. Upon examination of cuticles derived from animals that are homoallelic for two different null alleles of tkv, we confirmed an earlier report of Terracol and Lengyel (Terracol and Lengyel, 1994) documenting both germ band retraction and dorsal-open phenotypes in tkv mutant embryos (Fig. 6A–C). We went on to quantify this phenotypic duality and found that the range of phenotypes seen in the two tkv null alleles, tkv5 and tkv7, is similar to that seen in the Kr-GAL4/UAS-ricin transgenic embryos (Fig. 6D). Whereas the tkv and hnt phenotypic dualities underscore the link between dorsal closure and germ band retraction, presumably at the level of the amnioserosa, it is also true that these phenotypic data undermine the view that genetic approaches can be used to cleanly tease apart the morphogenetic processes of closure and germ band retraction.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Amnioserosal Integrity Is Required for Dorsal Closure

To directly assess amnioserosal requirements for morphogenesis, we turned to the UAS-GAL4 mosaic expression system and the cell autonomous toxin Ricin-A to ablate the amnioserosa in its entirety. Ricin, a potent inhibitor of translation in eukaryotic cells, consists of two polypeptide chains. The Ricin-A polypeptide is both necessary and sufficient to mediate cellular toxicity; it is the A chain that catalytically inactivates ribosomes by means of depurination of 28S ribosomal RNA. The Ricin B polypeptide is dispensable for toxicity; its function enables Ricin to cross cell membranes and enter and kill neighboring cells (Endo and Tsurugi, 1988). These observations indicate that Ricin-A alone can function as a cell autonomous toxin. In the studies reported here and elsewhere (Hidalgo et al., 1995; Hidalgo and Brand, 1997; Allen et al., 2002; Roman and Davis, 2002), targeted cell ablation was simply and efficiently achieved by crossing GAL4 insertion lines to a transgenic line (UAS-ricin), carrying the GAL4-responsive toxin gene UAS-ricin.

By using two independent AS-GAL4 lines, we observed a range of germ band retraction and dorsal closure phenotypes in response to Ricin-A–induced toxicity and, thus, established that amnioserosal integrity is required not only for germ band retraction but for dorsal closure as well. In one line, C381-GAL4/UAS-ricin, we observed primarily germ band retraction defects, whereas in the other line, Kr-GAL4/UAS-ricin, we observed roughly equal numbers of u-shaped and dorsal-open group mutants. In both mutant lines, DJNK signaling, which is required for dorsal closure, was evident in leading edge epidermal cells. The unique distribution of phenotypes seen in each of the AS-GAL4 transgenic lines could represent differences in the timing, location, or abundance of ricin-A expression. However, we believe it most likely that temporal differences in amnioserosal cell death account for the phenotypic differences that we observe, as (1) ush and C381-GAL4/UAS-ricin mutants exhibit identical temporal patterns of amnioserosal cell death in conjunction with predominantly germ band retraction failures and as (2) hnt and Kr-GAL4/UAS-ricin mutants exhibit identical temporal patterns of amnioserosal cell death in conjunction with germ band retraction/dorsal closure phenotypic duality. Moreover, we suggest that early amnioserosal cell death is causative with respect to dorsal closure defects in hnt and Kr-GAL4/UAS-ricin mutants. Viewed from this fresh perspective, we suggest that we have revealed an amnioserosal requirement for dorsal closure that precedes those which have so far been defined for germ band retraction.

In concept and objective, our studies are highly reminiscent of those of Kiehart and coworkers (Kiehart et al., 2000). This said, inherent methodological differences exist between laser- and Ricin-A–mediated ablations, and the difference in outcome between the previously published ablation study and ours is not unanticipated. Ricin-A activity induced apoptotic cell death throughout the amnioserosa, in contrast to the effects of laser ablation, which were localized. Thus, it is likely that in the study of Kiehart et al., signals emanating from the amnioserosa were not completely eliminated, nor were the tensile forces of the tissue. In contrast, ablation of the entire tissue, as we describe here, disrupted all mechanical contributions and all but the earliest signals. In this regard, the amnioserosa may be the site of production not only of germ band retraction triggers, such as Hnt and Ecdysone (Lamka and Lipshitz, 1999; Kozlova and Thummel, 2003), but of dorsal closure regulating molecular signal(s) and/or tensile force(s) as well.

DJNK Signaling Initiates in the Absence of the Amnioserosa

Having established that amnioserosal ablation is associated with defects in dorsal closure, we next sought a molecular explanation for the morphogenetic defects that we observed. As a first step toward this objective, we tested whether loss of epidermal Dpp signaling or whether gain or loss of epidermal DJNK signaling (all shown to be causative with respect to dorsal closure defects) was evident in amnioserosal ablated mutant embryos. Our finding that dpp is expressed in the epidermis of amnioserosal ablated mutant embryos indicates that a molecular trigger of closure is generated even in the absence of a viable amnioserosa. Thus, it appears unlikely that the amnioserosa serves as the source of an as yet unidentified DJNK activator, at least after stage 9 of embryonic development. It is notable that we sometimes observed short range expansion of the epidermal DJNK signaling domain in C381-GAL4/UAS-ricin mutants. Although this range of signaling is not extensive enough to be causal for gross defects in dorsal closure, it is consistent with models postulating a role for the amnioserosa in restricting DJNK signaling to leading edge cells (Byars et al., 1999).

It remains crucial to figure out how the amnioserosa and the epidermis coordinate their roles in effecting dorsal closure. Some additional insight into this important question might be drawn from phenotypic comparisons as shared loss-of-function phenotypes point to sequential function in single molecular pathways (Nusslein-Volhard and Wieschaus, 1980). In this regard, cuticles derived from animals harboring mutations in Dpp signaling components exhibit centrally positioned dorsal holes in contrast to those derived from animals harboring mutations in DJNK signaling components, which exhibit dorsal holes that span central and anterior domains (for review, see Jacinto et al., 2000). Our demonstration that cuticles generated by amnioserosal ablation and by mutations in Dpp pathway components (both tkv and punt) are strikingly similar, in combination with our molecular data showing DJNK signaling to be unaffected after amnioserosal ablation, points to an amnioserosal function in the process of dorsal closure that is downstream of Dpp. It is tempting to speculate that leading edge Dpp elicits a paracrine response not only in lateral epidermal cells but also in amnioserosal cells abutting and underlying the closing epidermis, although we cannot yet say whether this response is molecular or mechanical in nature. Consistent with this idea are (1) our previous demonstration that mutations affecting epidermal Dpp signaling disrupt gene expression in the amnioserosa (Byars et al., 1999), and (2) the demonstration that, during vertebrate wound healing (a process analogous to dorsal closure), the Dpp-related cytokine TGF-β elicits paracrine responses in both the epidermal cells adjacent to the leading edge, as well as in the exposed mesenchymal cells underlying the wound (Martin et al., 1993).

Three signaling cascades have recognized leading edge epidermal roles in dorsal closure (for review see Jacinto et al., 2000). It is clear that the two we have studied here, the DJNK and Dpp signaling cascades, play central roles in closure as mutations in genes encoding components of either of these pathways severely disrupt the process. A third signaling cascade (mediated by Rho and PKN) directs assembly of cytoarchitectural components of closure in leading edge cells. Null mutations in the genes coding for Rho and PKN disrupt dorsal closure but not as severely as loss-of-function mutations in genes coding for DJNK or Dpp pathway members—indeed the holes are both smaller and shifted anteriorly (Lu and Settleman, 1999). We have yet to address the role of the amnioserosa in relation to the Rho/PKN signaling molecules; thus, it remains possible that one function of the amnioserosa is to trigger Rho/PKN signaling in leading edge cells. We do not favor this model, however, given the differences in the rho/PKN and amnioserosal ablation phenotypes.

Ricin-A Expression in Leading Edge Cells Does Not Affect Embryonic Development

Ablation of the leading edge, in contrast to ablation of the amnioserosa, appeared to have no affect on either dorsal closure or germ band retraction. This finding was an unexpected result given the unique role attributed to the leading edge epidermis during closure. From a signaling perspective, leading edge restricted DJNK activity is essential for proper closure, and from a cell biological perspective, leading edge epidermal cells are the first to elongate during closure, contracting along the AP axis and elongating along the DV axis (Knust, 1997; Noselli, 1998). Two interpretations, which are not mutually exclusive, underscore the complexities of defining tissue-specific requirements for morphogenesis. First, it is possible that, after Ricin-A–induced ablation, second-row lateral epidermal cells assume leading edge cell fates and thereby functionally substitute for the original leading edge. Fate mapping of individual epidermal cells will be required to determine whether this transformation indeed occurs in LE-GAL4/UAS-ricin transgenics. Alternatively, leading edge expression of ricin-A may be initiated too late in LE-GAL4/UAS-ricin transgenics to disrupt dorsal closure. This said, the LE-GAL4 driver has been shown to be effective in at least one functional study of dorsal closure (Lu and Settleman, 1999).

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Molecular and genetic characterizations of the dorsal-open group mutants have revealed requirements for epidermal gene products in dorsal closure, whereas similar analyses of the u-shaped class of mutants have revealed the requirements for amnioserosal gene products in germ band retraction. Together, these studies have led us to models of closure and germ band retraction that are largely distinct. Now, in combination with a re-examination of the phenotypes of canonical dorsal-open and u-shaped class mutants reported here and elsewhere (Reed et al., 2001), genetic cell ablation studies reveal the requirement for the amnioserosa not only in germ band retraction, but in dorsal closure as well. With respect to dorsal closure, which serves as a model for morphogenetic events dependent upon changes in cell shape and position in higher eukaryotes, this finding is particularly significant, as it highlights the role of the tissue that abuts and/or underlies the closing epidermis.

In considering the relationship of fly dorsal closure to vertebrate closure processes such as wound healing, it is notable that the fly and vertebrate processes exhibit requirements for identical signaling pathways and cytoskeletal elements. Just as in dorsal closure, aggregation of actin at the leading edge of the epithelium surrounding a wound is postulated to provide a force necessary to draw the edges of the wound together in a purse-string like manner (Martin and Lewis, 1992). Up-regulation of TGF-β molecules in epithelial cells at the wound margin also parallels the induction of dpp in the leading edge (Martin et al., 1993), and it is JNK signaling that triggers TGF-β/Dpp expression in both processes (Martin and Nobes, 1992). With respect to the studies reported here, it is notable that, during wound healing, a paracrine response is elicited by TGF-β both in epidermal cells adjacent to the leading edge and in exposed mesenchymal cells underlying the wound (Martin et al., 1993). In the long-term, further investigation of the role of the amnioserosa during dorsal closure will allow us to extend functional parallels between the amnioserosa and the wound mesenchyme, between dorsal closure and wound healing, and between dorsal closure and other vertebrate closure processes, including neurulation, palate formation, and eyelid fusion.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Fly Strains

Mutants fly lines included in this study (tkvv5, tkv7, ush2, and hnt1) as well as fly lines expressing GAL4 in a cell-specific pattern (GMR-GAL4, LE-GAL4, Kr-GAL4, and C381-GAL4), fly lines harboring UAS transgenes (UAS-lacZ, UAS-GFP, and UAS-ricin), and balancer lines have been described (Ring and Martinez Arias, 1993; Castelli-Gair et al., 1994; Hidalgo et al., 1995; FlyBase, 2003). Flies were reared on standard culture medium at 22°C.

Phenotypic Analyses

Embryonic lethal cuticular phenotypes were viewed after mounting devitellinized samples in One-Step Mounting Medium (30% CMCP-10 [Masters Chemical Company, Bensenville, IL], 13% lactic acid, 57% glacial acetic acid). Dorsal closure defects were characterized by the presence of a hole in the dorsal surface of the cuticle or a puckering of dorsal epidermis. Germ band retraction defects were classified by the presence of at least one abdominal segment retained on the dorsal side of the embryo, resulting from failure of the germ band to fully retract posterior segments.

Hybridizations in situ were performed as described (Tautz and Pfeifle, 1989), with the exception that proteinase K steps were omitted. Antisense probes labeled with digoxigenin-UTP were generated using the dpp cDNA clone pMBO-1457 (from M. O'Connor) as template for T3 transcription and the pBS-rpr cDNA clone (from C.S. Thummel) as template for T7 transcription. β-Galactosidase activity assays and antibody stainings were performed as described (Ashburner, 1989). Antibodies used for these studies included anti–β-galactosidase (from S. Carroll), anti-Krüppel (from M. B. O'Connor), and commercially available anti-Ricin. Antibody reaction products were visualized in embryos using the Vectastain ABC Kit (Vector Laboratories).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

We thank Cherie Byars and other members of our laboratory for valuable discussions; Sean Carroll, Mike O'Connor, and Carl Thummel for antibodies and clones; Andrea Brand, Michelle Lamka, Stephane Noselli, and Carl Thummel for fly lines; and Diana Lim for figure preparation. A.L. and A.S. were funded by the National Institutes of Health.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES