BIOLOGICAL OVERVIEW

The crucial process of dorsal closure [Images] is one of the more dramatic of all the morphogenetic movements taking place during Drosophila embryogenesis. The closure event occurs late in embryogenesis, and results in the establishment of the dorsal epidermis. Cells in the two lateral epidermal primordia change shape and spread over the amnioserosa [Images], a membranous structure covering the dorsal side of the embryo. Epithelial cells elongate in this spreading, covering process. Neither cell proliferation nor cell rearrangement occurs (Glise, 1995).

Dorsal closure requires cytoskeletal based mechanisms as evidenced by the involvement and dynamics of Myosin II (nonmuscle Myosin), coded for by the zipper locus. Myosin accumulates most conspicuously at the leading edge of the dorsal epithelium; the integrity of this edge is destroyed in zip mutants. At a minimum, Myosin is required for the maintenance and stabilization of cell shape during dorsal closure, no doubt producing the force for cell shape changes during closure (Young, 1993).

Other proteins are involved in the process in what may be considered a "dorsal closure pathway." The hemipterous gene encodes a kinase required for epithelial cell sheet movement. Kinases are enzymes that attach phosphate residues to target proteins, thus transducing signals from upstream proteins which activate the kinase to downstream targets which carry out the instructions arriving by means of the signaling pathway. One such kinase involved in Drosophila morphogenesis is Dsor1 (Downstream of Ras1). This kinase is involved in another signaling pathway, the Ras pathway. HEP occupies an identical position to that of Dsor1 but in the dorsal closure pathway; both proteins serve as mitogen-activated protein kinase kinases (MAPKKs or MEKs). If this sounds rather complex, it is. At least 10 proteins are involved in the Ras pathway, and at least 10 other proteins will be involved in the dorsal closure pathway.

What are some of the other proteins currently implicated in the dorsal closure pathway? The proteins Rac and Cdc42 both belong to a family of small GTP binding proteins; they are important in maintaining cell shape, and are associated with focal adhesion plaques. The discussion will return to Rac and Cdc42 in a moment, but first, a word about the focal adhesion plaque: it is a dynamic F-actin and Myosin II containing structure, necessary for cell adhesion and migration. This structure is involved in the attachment of cells to one another, and is present at the points of adhesion of muscles to gut and epidermal cells. Focal adhesion plaques are marked by the presence of integrins which serve as receptors for extracellular ligands present in the extracellular matrix.

Rac and Cdc42 are GTP binding proteins serve a function similar to that of Ras, that is, all are molecular switches which can activate protein kinases. Both Rac and Cdc42 have important roles in dorsal closure (Harden, 1996). One protein kinase activated by Rac1 and CDC42 is the newly discovered serine/threonine kinase PAK, that binds Rac and CDC42, and in so doing becomes activated (Harden, 1996). For mammalian cells, it has recently been shown that Rac and CDC42 activate signaling by the mitogen-activated protein kinase family member c-Jun amino-terminal kinase (JNK) (Coso, 1995, Vojtek, 1995, and Zhang, 1995). It is now clear that Rac and CDC42 activate JNK signaling in Drosophila. The Drosophila JNK is coded for by basket, also implicated as a mediator of the fly's immune response (Sluss, 1996 and Riesgo-Escovar, 1996). The implication is that the GTP binding proteins activate Hemipterous, which transduces an activation signal to JNK which in turn activates the transcription factor Jun.

CDC42 and Rac1 control actin-dependent processes in the wing disc epithelium of the fly, but their linkage to HEP in this function cannot yet be inferred. Rac1 and CDC42 are involved in cell elongation in the wing disc. The proteins are found in actin associated basal plaques (which may represent classical focal adhesions) and the apical adherens junction containing localized E-cadherin and ß-catenin (Eaton, 1995).

What are the targets of Hemipterous? The mutation puckered was isolated by random insertion of a transcriptional activator into Drosophila genes. One insertion site inactivates puckered, creating a mutation that results in an abnormal morphology of dorsal epidermal cells (Ring, 1993). In spite of this abnormal morphology, dorsal closure occurs to completion The puckered gene is not expressed in hemipterous mutants, indicating that Puckered acts as a downstream target of hemipterous, and that one function of hep is to activate genes expressed in cells involved in dorsal closure. Although the function of puckered is not known, discovery of a target of HEP provides a hint as to the existence of a dorsal closure pathway.

Although cytoskeletal mechanics provide an immediate explanation for defects in dorsal closure, there is a direct, though complex link between cell surface phenomena and gene expression regulated by cytoskeletal changes. One such link between the cytoskeleton and gene expression is Hemipterous, operating in a pathway resembling the well characterized Ras pathway. What actions or feedback do nuclear regulatory events have on cytoskeletal based morphological changes? Pinpointing the downstream targets of hemipterous will yield answers to this question.

hemipterous and the JNK pathway are also required later in development, for correct morphogenesis of other epithelia, the imaginal discs. During metamorphosis, the imaginal discs undergo profound morphological changes, giving rise to the adult head and thoracic structures, including the cuticle and appendages. hep mutant pupae and pharate adults show severe defects in disc morphogenesis, especially in the fusion of the two lateral wing discs. These defects are accompanied by a loss of expression of puckered (puc), a JNK phosphatase-encoding gene, in a subset of peripodial cells that ultimately delineates the margins of fusing discs. In further support of a role for puc in disc morphogenesis, pupal and adult hep phenotypes are suppressed by reducing puc function, indicative of a negative role for puc in disc morphogenesis. Furthermore, the small GTPase Dcdc42, but not Drac1, is an activator of puc expression in a hep-dependent manner in imaginal discs. Altogether, these results demonstrate a new role for the JNK pathway in epithelial morphogenesis, and provide genetic evidence of a role for the peripodial membrane in disc morphogenesis. A general model is discussed whereby the JNK pathway regulates morphogenesis of epithelia with differentiated edges (Agnès, 1999).

Loss of both maternal and zygotic hep function leads to embryonic dorsal closure (DC) defects, a process of epithelial morphogenesis by which lateral ectodermal cells elongate dorsalward and ultimately fuse at the dorsal midline. By contrast, zygotically deficient hep animals die later during the pupal stage, thus indicating hep requirement in at least two different processes and stages. The anterior (head and thorax) and abdominal regions of the adult body derive from different structures: the imaginal discs and the histoblasts, respectively. Dissected null hep mutant pupae (greater than 10 hours after puparium formation; h APF) show incomplete metamorphosis of the anterior body region, whereas abdominal structures seem to form properly. These observations suggest that hep is involved in imaginal disc morphogenesis specifically. Interestingly, a viable P insertion allele of hep (hep¹) causes poorly penetrant (approx. 5%) adult head and thoracic abnormalities. The phenotypes range from slight to strong unilateral deletions of imaginal disc-derived adult structures like the wings (hence the name of the gene), legs and eyes. The most frequent and typical phenotype is a cleft of variable width and depth at the dorsal midline of the thorax. In strongly affected adults, a large cleft can separate two deformed dorsal hemithoraces, and, more rarely, all wing disc derivatives (wing and dorsal mesothorax) can be missing: a dramatically bent fly results. In this extreme case, the wing disc is improperly located in the abdominal cavity, with well differentiated adult wing tissues. These observations suggest that hep controls the correct morphogenesis and/or positioning of imaginal discs. Advantage was taken of the unique and late lethality associated with strong, lethal hep alleles (hep^r75and hep^r39) to study the origin of these defects during pupal development (Agnès, 1999).

Dissected discs from third instar (L3) hep^r larvae show various phenotypes. (1) There is a delay in hep^r larval development; (2) there is a variable reduction in the size of mutant discs as compared to wild type (approximately 30% reduction in strong cases). This phenotype is observed in staged larvae and white pupae, indicating that the delayed development may not be the cause of this defect. (3) Though the overall disc morphology is correct, malformed and misfolded discs can also be observed, with a higher frequency in smaller discs, suggesting a growth origin for these defects. Because the major events in disc morphogenesis take place during pre-pupal (extending from 3 hours before to 12 hours APF) and early pupal development, the development of hep^r mutant discs undergoing morphogenesis was studied in staged prepupae. Overall, metamorphosis is strongly affected in hep^r mutants, as evidenced by an important disorganization of pupal tissues. The pattern of defects varies from individual to individual. Therefore, only those defects are described that are the most frequently observed and which best reflect the hep pupal phenotype, i.e., the morphogenesis of the wing imaginal discs. To better visualize morphogenetic defects and have a view of the overall organization in each larva, different markers were used to label specific disc domains (puc-lacZ and dpp-lacZ). The analysis of dissected hep mutant pupae reveals one major defect: the absence or aberrant spreading and fusion of the two lateral wing discs. Wing discs contribute to the wings and almost all the dorsal thorax. By 8 hours APF, the wing discs meet and fuse at the dorsal midline, a region that is frequently affected in hep adult mutants. In strong hep alleles (hep^r75or hep^r39), the two wing discs remain in their initial position in the pre-pupae, and do not spread and meet at the dorsal midline. In these mutants, the morphology of the discs is strongly affected, suggesting that folding and/or eversion are not completely normal. In some cases, eversion does take place, indicating that hep is not absolutely required for this step. However even in these cases spreading and fusion did not take place, leading to the open thorax phenotype. If hep function is only partially absent, as in the viable hep¹/hep^r75allelic combination, wing disc morphogenesis can proceed almost normally. In these conditions, discs can spread and fuse, though in some cases one disc does not reach the midline, a behavior that may lead to the unilateral defects (Agnès, 1999).

Because the thorax does not close in hep^r75pupae, the internal tissues, including the gut and larval tissues, become extruded; this presents a phenotype similar to that found in hep embryos, which do not close dorsally. Although eye-antenna discs can fuse in hep^r75, the head does not form. It is not clear whether this is an indirect consequence of aberrant thorax formation or if hep eye-antenna discs are not able to complete morphogenesis. On the ventral side, leg discs also appear to evert and fuse, but their shape is frequently abnormal. These results show that hep is required for the formation of the head and thoracic structures during metamorphosis, with a strong effect on the most distant imaginal tissues, the wing discs. This suggests that the initial position of discs (the distance separating contralateral discs is different from disc to disc) contributes to their hep-dependent fusion pattern (Agnès, 1999).

The puckered (puc) gene encodes a JNK MAPK phosphatase that negatively regulates the activity of the JNK pathway during dorsal closure. It is specifically expressed in a population of lateral cells (the leading edge) that delineates the boundary between the ectoderm and the amnioserosa. Importantly, puc expression is activated by hep in the leading edge, to initiate a negative feedback loop. Therefore, puc is a marker of both JNK activity and the margins of moving epithelia during dorsal closure. It was therefore of particular interest to analyze the expression pattern of puc during larval and pupal development, using the puc^E69 allele, a P lacZ enhancer-trap line inserted in the puc gene. In the wild type, expression of puc becomes detectable in the third instar larva (epidermis and spiracles) and slightly increases throughout this stage. It then becomes stronger during prepupariation and decreases by the end of this stage. Interestingly, puc is specifically expressed in particular cell populations in all thoracic discs. In the proximal part of the wing, haltere and leg discs, puc is strongly expressed in the stalk region, where imaginal discs connect to the larval epidermis. Further, puc is also expressed in rows of cells in wing, eye-antenna, haltere and leg discs, in a pattern that is reminiscent of the margin expression observed in the ectoderm during dorsal closure. These cells are on the dorsal side of imaginal discs, and are part of a particular structure of the discs, the peripodial epithelium or peripodial membrane. The peripodial membrane is an epithelium made of squamous cells easily distinguishable from those of the disc proper due to their larger and more widely spaced nuclei. During metamorphosis, cells of the peripodial membrane play an active role through the dramatic change of their shape, either by intensive stretching or contraction. In addition to this role, the peripodial membrane also contributes to some parts of the adult integument, especially in regions where adjacent discs will suture. To confirm that puc is expressed in the peripodial epithelium, double immunostaining was performed using an antibody directed against beta-galactosidase (reflecting puc expression) and another against the conserved N terminus of the Broad-complex (Br-C) protein isoforms. Br-C is ubiquitously expressed in the nuclei of disc cells, and serves here as a marker of peripodial membrane vs columnar epithelial cells. The results show that puc is expressed in the peripodial membrane of all thoracic discs, at the boundary between peripodial and columnar epithelia. Due to a more sensitive detection using antibody staining, it was found that puc is expressed in a larger subset of peripodial membrane cells than that revealed using X-gal stainings. These experiments thus confirm the restriction of puc expression in the peripodial membrane. Later during prepupariation, puc expression is maintained in the peripodial membrane and marks the presumptive suture sites of imaginal discs with their neighbors. By the end of this stage, puc staining is found at the frontier between sutured discs. These data suggest that puc, and the JNK pathway, are required in a specific subset of peripodial cells for morphogenesis of imaginal discs (Agnès, 1999).

In embryos, hep activates puc and dpp expression in ectodermal margins. puc activation serves to initiate a negative feedback loop that controls the levels of JNK activity during dorsal closure. To test whether these important regulatory links also exist during disc morphogenesis in larvae, the expression pattern of puc and dpp in L3 larvae were analyzed using strong hep alleles. Both in hep^r39 and hep^r75male larvae, puc expression in imaginal tissues is dramatically reduced. Since the loss of puc staining may reflect a delay in the onset of puc expression, white-pupae were dissected and stained to reveal puc-lacZ expression. As in L3 larval stage, puc expression is absent in hep white-pupa discs, whereas wild-type puc expression slightly increases at this developmental stage. In contrast, puc expression is still detectable in the epidermis, mouthparts, and spiracles both in the wild type and in hep mutants, indicating a hep-independent expression of puc in these tissues as well as providing an internal control. In contrast, a dpp-lacZ line, which is expressed in a pattern very similar to that of puc in the peripodial membrane, is not controlled by hep function in L3 and 5 h APF discs. Similarly, in later stage discs (L3 to 8 hours APF), dpp expression is not changed in regions close to the dorsal midline, like the notum. Together, these results show that hep is required for the normal expression of puc in the peripodial membrane of thoracic discs, suggesting that the subset of puc-expressing cells in the peripodial membrane is the primary site of JNK activity during disc morphogenesis. They also highlight an important difference with the situation found during dorsal closure, where the JNK and dpp pathways are coupled (Agnès, 1999).

During the process of dorsal closure, hep and puc have opposing effects on their common target, basket/DJNK. To begin to analyse the role of puc during disc morphogenesis, genetic interactions between hep and puc were examined. In addition to dominantly suppressing the embryonic lethality associated with the hep¹allele, puc mutations also strongly suppress the adult phenotypes associated with this allele. The extent of suppression allows a normally embryonic lethal stock (hep¹/hep¹) to be kept. Moreover, reduction by half of puc activity also suppresses hep^r early pupal development arrest, allowing a large proportion of hemizygous males to proceed to late pupal stage. In the presence of only one copy of puc, defects associated with mutant hep are strongly reduced. These results indicate that puc has a negative regulatory function in discs, suggesting that the regulatory link established between hep and puc during dorsal closure is well conserved in imaginal discs during metamorphosis (Agnès, 1999).

The small GTPases of the Rho family, DRac1 and Dcdc42, can positively regulate the Drosophila JNK pathway in embryos. To further investigate the conservation of JNK pathway function in disc morphogenesis, activated forms of DRac1 (UAS-Drac1V12) and DCdc42 (UAS-DCdc42V12) were expressed in different subsets of imaginal disc cells using the UAS-GAL4 system. Because GAL4 lines driving specific expression in the peripodial membrane do not exist, GAL4 lines that are expressed in different domains of the columnar epithelium were used to test for a potential activating role of Drac1 and Dcdc42 in discs; among those, some are also expressed in the peripodial epithelium. Targetted expression of either Drac1 or Dcdc42 results in a strong ectopic expression of puc in patterns specific of each GAL4 driver used. In the wing, expression is prominent along the anteroposterior boundary. The effects of Drac1 and Dcdc42 on puc are similar, although not exactly the same in terms of the activation levels and pattern of activated cells. In a hep^r mutant background, only Dcdc42-mediated puc ectopic expression is strongly suppressed, indicating that hep is required downstream of Dcdc42 to activate puc. However, suppression is not complete, since puc expression remains in some cells of the discs, indicating a hep-independent activation of puc. Similar results were also observed in embryos. In addition to its potent activity on puc expression, Dcdc42 also induces dramatic defects in the overall disc morphology, including a reduction of the size and an aberrant shape. These defects are also strongly suppressed by hep, reinforcing the notion that a Dcdc42, hep, puc cascade plays an important morphogenetic activity in imaginal discs. These data suggest that Dcdc42 may play a role in disc morphogenesis in combination with hep and puc. In addition, they provide evidence that cells of the columnar epithelium are competent for JNK activity, and that a limiting factor for JNK activation lies upstream of the small GTPases. In this respect, the disc epithelium is very similar to the lateral ectodermal cells during dorsal closure (Agnès, 1999).

The fact that hep regulates puc expression in cells of the peripodial membrane indicates a requirement of the JNK pathway in these cells, and reveals at least one site of JNK activity during metamorphosis. The role of this structure, which covers one side of the larval imaginal discs, has been controversial and unclear. It has been proposed to play an active role during metamorphosis, by undergoing successive stretching and contraction. The results provide a genetic confirmation of a role for the peripodial membrane, by showing a clear link between abnormal disc morphogenesis and spreading, and cell fate determination in the peripodial membrane mediated by the JNK pathway (Agnès, 1999).

One important question is whether the defects observed in hep mutants are only the result of a lack of peripodial membrane function. This question could be addressed directly by generating mutant hep clones in the peripodial membrane specifically, using directed mosaics. If hep function is indeed restricted to the peripodial membrane, then it would strongly argue in favor of a model in which the peripodial membrane would orchestrate the process. Such a role has been proposed in dorsal closure of the leading edge. Interestingly, a Hedgehog-mediated organizing role of the peripodial membrane during regeneration of the T2 leg disc has recently been reported, as well as a novel structural basis for the long-range activity of signaling molecules in the wing disc. Whether hep and the JNK pathway contribute to metamorphosis through related mechanisms will be further investigated (Agnès, 1999).

By several criteria, the so-called leading edge of the embryonic ectoderm is very similar to the peripodial membrane cells expressing puc: it expresses puc in a hep-dependent manner, and marks the future sites of epithelial suture. In addition, it is found at the boundary between a columnar and a stretched/squamous epithelium. These similarities suggest that the same signal(s) may activate hep and the JNK pathway in both developmental processes. Consistent with this view, it is found that other members of the JNK pathway, including puc and probably Dcdc42, play a role in metamorphosis. A role for Dcdc42 during pupal development has been reported to control epithelial cell shape. Finally, the cleft thorax phenotypes displayed by other genes involved in dorsal closure, like Dfos and ZO-1, reinforces the notion that most of the members of the JNK pathway have a role in metamorphosis. The observation that Drac1V12 can activate puc expression independent of hep suggests that another JNK-related activity is present in discs. In support of this view, a MAPKK of the JNKK family has been isolated, DMKK4, that may have a partially redundant function with hep. Further, Drosophila homologs of another stress-activated MAPK pathway, the p38 pathway, have been identified recently in flies that may also play a role in pupal development, a question that can now be addressed using the Drosophila p38 MAPKK licorne mutants (Agnès, 1999 and references therein).

The study of dorsal closure and metamorphosis provides two complementary examples of epithelial morphogenesis, where dorsal closure can be viewed more like a two-dimensional process. In contrast, disc morphogenesis represents a more complex process involving movements in a third dimension. Despite their apparent differences, these two processes share several features, including gene-specific expression of puc and dpp, and their genetic requirement for members of the JNK pathway. In both processes, margins are established that develop at the boundary between two epithelia, one stretched (peripodial membrane and amnioserosa), and the other columnar in shape (lateral ectoderm and columnar imaginal tissue). Based on these similarities, it is proposed that the JNK pathway regulates the determination of the leading cells in moving epithelia. As evidenced by the mutant phenotypes of hep at embryonic and pupal stages, the making of margins is crucial for the morphogenesis of entire epithelial structures (dorsal ectoderm and imaginal discs), especially during the spreading and fusion of sheets of cells that ultimately make a continuous tissue. This general view may be extended to related processes in other multicellular organisms. In the worm C. elegans for example, the morphogenetic process known as ventral enclosure displays several common features with dorsal closure. The leading edges of the ectoderm accumulate F-actin to form a purse-string, move toward one another and suture at the ventral midline. As in dorsal closure, ventral enclosure relies on cell elongation along the DV axis, suggesting that the underlying mechanisms may be conserved. In vertebrates, several observations suggest that the process of wound-healing is controlled by a network of molecules involved in the JNK and TGF-beta pathways, leading to the proposal that both dorsal closure, and imaginal disc closure, may represent workable genetic models for wound-healing. In conclusion, dorsal closure, metamorphosis and related processes in other organisms will provide complementary models to unravel the intimate mechanisms whereby modulation of a single conserved pathway may create the fascinating diversity of epithelium movement accompanying metazoan development (Agnès, 1999 and references therein).

GENE STRUCTURE

Bases in 5' UTR -643

Exons - 7

Bases in 3' UTR - 600

PROTEIN STRUCTURE

Amino Acids - 487

date revised: 12 December 98

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.