Fos-related antigen/kayak

DEVELOPMENTAL BIOLOGY

Embryonic

See the embryonic expression pattern of kay at the Berkeley Drosophila Genome Project Patterns of Gene Expression Site.

Fra/Kayak protein is first detectable in the head mesoderm at stage 9, and from stage 11 onwards in additional tissues, including the amnioserosa and the ectoderm. FRA mRNA levels are elevated in the embryonic endoderm of the second gut lobe (Perkins, 1990). During stage 13, endodermal cells begin to show weak Fra staining, with a slightly higher level in a band of the forming midgut epithelium spanning the fusion junction of the two gut primordia. This band of elevated Fas expression becomes more and more prominent, and remains clearly visible from stage 15 onwards throughout late embryogenesis; it stretches throughout the second gut lobe, from the first to the second gut constriction (i.e. through approx. ps6-7). Fra is predominantly (if not exclusively) nuclear in all cell types observed. Fra protein accumulates in all endodermal cells in the second gut lobe. This contrasts with labial, which is expressed only in a subset of the endodermal cells in this lobe; lab expression is not detectable in the most posterior cells within this lobe; throughout the lobe, lab-expressing cells are interspersed with cells not expressing lab (Reise, 1997). Fra staining is strongest in the central region of the lobe, fading slightly towards both constrictions (Reise, 1997), whereas Lab staining shows a striking anteroposterior gradient of expression, with highest levels most posterior (Immergluck, 1990).

Fra-expressing neurons appear to be located near the lateral chordotonal cells and might be extrasensory or multiple dendritic neurons. At the same stages of development, FRA mRNA is localized to a portion of the ectoderm that corresponds with muscle attachment sites. It is also observed in part of the midgut and hindgut and in the anal pad (Perkins, 1990).

Effects of Mutation or Deletion

Drosophila kayak mutant embryos exhibit defects in dorsal closure, a morphogenetic cell sheet movement that takes place during embryogenesis. It is shown that kayak encodes D-Fos, the Drosophila homolog of the mammalian proto-oncogene product, c-Fos. D-Fos is shown to act in a similar manner to Drosophila Jun: in the cells of the leading edge it is required for the expression of the TGFbeta-like Decapentaplegic (Dpp) protein, which is believed to control the cell shape changes that take place during dorsal closure. The kayak expression domain include the cells of the amnioserosa and the lateral epidermis during the process of dorsal closure. At the onset of dorsal closure, elevated levels of D-Fos can be detected in the nuclei of leading-edge cells as they initiate elongation. Subsequently, elevated expression of Fos can also be observed in more ventrally located epidermal cells. Concurrently, cell elongation spreads laterally, until the two edges meet at the dorsal midline. At this stage, Fos is strongly expressed throughout the embryonic epidermis, with the highest levels remaining in the cells of the leading edge. This expression pattern is very similar to that of D-Jun and also correlates with the JNK-pathway-dependent stripe of dpp expression in the leading edge, which becomes apparent during the initiation phase of dorsal closure. Therefore, it is conceivable that Drosophila Fos acts in conjunction with Jun to regulate dorsal closure and dpp expression (Zeitlinger, 1997).

Defects observed in mutant embryos, and adults with reduced Fos expression, are reminiscent of phenotypes caused by 'loss of function' mutations in the Drosophila JNKK homologue, hemipterous. Mutant alleles of D-fos have not previously been described. Based on the potential involvement of D-Fos in the process of dorsal closure, known dorsal open mutants were examined for defects in D-fos. kay1 mutant embryos all die during embryogenesis with large dorsal and anterior holes that indicate failed dorsal closure and head involution. In kay2 embryos, dorsal holes are also observed, but at lower penetrance. Depending on the temperature and genetic background, up to ~1% of kay2 homozygotes even develop to adulthood, as seen after recessive markers had been removed from the kay2 mutant chromosome by recombination. The transheterozygous kay1/kay2 allelic combination displays an intermediate phenotype and embryonic, or early larval, lethality. These observations indicate that kay2 is a weaker allele than kay1, and thus is a hypomorph. To examine the cause of the kay defect at the cellular level, mutant and wild-type embryos were stained with an anti-Coracle serum, which outlines the epidermal cells. It was found that the kay mutant phenotype is caused by a failure of the lateral epidermal cells to elongate. As previously observed in the case of D-jun, leading-edge cells of zygotic kay1 mutants initiate elongation transiently, but fail to maintain it and subsequently resume the unelongated shape. The more lateral epidermal cells elongate to a very minor extent and resume the typical polygonal shape after the process has been terminated prematurely. Thus, the kay mutant phenotype closely resembles those described for hep, bsk and D-jun mutant embryos, also at the cellular level (Zeitlinger, 1997)

The phenotypic similarity between D-jun and kay mutants suggests that D-Fos and D-Jun act in concert to mediate dorsal closure. Thus, one may predict that D-fos/kay, like D-jun and the upstream signaling components bsk/JNK and hep/JNKK, is required for the expression of dpp in the leading edge cells. To test this idea, the expression of DPP mRNA was monitored in wild-type and kay-mutant backgrounds. Expression of dpp in the cells of the future leading edge is normally initiated when the germ band is fully extended and is maintained throughout dorsal closure. In contrast, in kay1 homozygous embryos, dpp expression is absent (or reduced in kay2 mutants) in cells of the leading edge. Significantly, other pattern elements of dpp expression are still present in the kay/D-fos mutants, including a more ventral stripe and midgut-specific expression, which have previously been shown to be independent of JNK signaling. Another downstream effect of Bsk signaling, the transcriptional activation of the puckered gene (puc) in the cells of the leading edge is also abrogated in kay1 mutant embryos. Taken together, the requirement of both D-Fos and D-Jun for dpp and puc expression in leading-edge cells suggests that the JNK signal is relayed by a heterodimeric transcription factor composed of D-Jun and D-Fos. These results indicate that D-Fos is required downstream of the Drosophila JNK signal transduction pathway, consistent with a role in heterodimerization with D-Jun, to activate downstream targets such as dpp (Zeitlinger, 1997).

Thorax closure in Drosophila: involvement of Fos and the JNK pathway

Dorsal closure, a morphogenetic movement during Drosophila embryogenesis, is controlled by the Drosophila JNK pathway, D-Fos and the phosphatase Puckered (Puc). To identify principles of epithelial closure processes, another cell sheet movement that can be termed thorax closure was studied: the joining of the parts of the wing imaginal discs that gives rise to the adult thorax during metamorphosis. The genes that are required for dorsal closure give rise to an interesting abnormal adult phenotype, suggesting that there is an additional requirement for these genes during later development: homozygous animals of mutant alleles of D-fos, hep, pannier (pnr) and components of the Dpp pathway show a cleft at the dorsal midline of the thorax and neighbouring bristles are abnormally parted to both sides. In thorax closure a special row of margin cells express puc and accumulate prominent actin fibers during midline attachment. Genetic data indicate a requirement of D-Fos and the JNK pathway for thorax closure, and a negative regulatory role of Puc. Furthermore, puc expression co-localizes with elevated levels of D-Fos; is reduced in a JNK or D-Fos loss-of-function background, and is ectopically induced after JNK activation. This suggests that Puc acts downstream of the JNK pathway and D-Fos to mediate a negative feed-back loop. Therefore, the molecular circuitry required for thorax closure is very similar to the one directing dorsal closure in the embryo, even though the tissues are not related. This finding supports the hypothesis that the mechanism controlling dorsal closure has been co-opted for thorax closure in the evolution of insect metamorphosis and may represent a more widely used functional module for tissue closure in other species as well (Zeitlinger, 1999).

In order to mark and visualize the dorsal parts of the wing imaginal discs that fuse during thorax closure, the UAS-Gal4 system was used to express green fluorescent protein in the expression domain of pnr, a gene encoding a GATA transcription factor whose expression is restricted to dorsal tissues throughout development. The prepupae were then dissected in a way that leaves the entire thorax complex intact and different stages were inspected by confocal microscopy. In addition, actin filaments were visualized by staining with phalloidin to monitor the behaviour of the cytoskeleton during this process. Phalloidin also stains three oblique muscles on each side, a useful marker during thorax closure. Already in third instar wing imaginal discs, pnr expression marks the dorsal part, the future medial notum. At around 6 hours after pupation (AP), after eversion, the dorsal parts of the two wing imaginal discs spread toward the dorsal midline, while the larval epidermis degenerates. When they subsequently meet and attach to each other at around 7 hours AP, filamentous actin becomes visible at the medial edge of the epithelium. These actin bundles at the dorsal midline are most abundant at 8 hours AP and are predominantly localized basally (Zeitlinger, 1999).

In summary, the process of thorax closure resembles embryonic dorsal closure at a tissue-morphological level: two epithelial sheets with a straight margin approach one another, meet at the dorsal midline, and attach. The actin organization seen along the margin of the epithelium is reminiscent of the accumulation of actin along the leading edge of the closing embryo. However, in contrast to the simple epithelial stretching of embryonic dorsal closure, the morphogenetic movements involved in thorax closure appear to be more complex: most cells are of polygonal shape and not obviously elongated along the dorsoventral axis. Furthermore, the tissue movements also include unfolding (as part of the eversion) and an anterior folding-in during midline fusion with subsequent back folding during head eversion (Zeitlinger, 1999).

Having established a system to monitor the progress of thorax closure, the tissue movements were monitored in a mutant background that gives rise to a cleft phenotype in adults. The hypomorphic mutation in D-fos, kay ² was used in this experiment. It revealed that the dorsomedialward spreading of the epithelium is already abnormal at 6 hours AP in most kay² prepupae. While, in a wild-type background, the pnr expression domain of the wing imaginal disc is found on top of the three oblique muscles and close to the degenerating larval epidermis, the corresponding epithelium in kay² prepupae of this stage has failed to reach this position and is still located more laterally. At 8 hours AP, the spreading epithelium often appears to have retracted and fallen back into its original folded position found at earlier stages, although filamentous actin typical of this stage is detectable. These findings strongly argue that the defects observed in kay² adult animals result from defects in thorax closure during prepupal stages (Zeitlinger, 1999).

The thoracic cleft phenotype observed with hypomorphic mutations in D-fos (kay² ) and hep (hep¹) suggests that D-Fos and the JNK pathway are involved in thorax morphogenesis. To confirm that the cleft phenotype is a result of a D-fos loss-of-function condition, a dominant negative form of D-fos (UAS-D-Fos bZIP) was expressed under the control of pnr-Gal4. This results in the appearance of a marked cleft in the thorax. A similar phenotype is obtained by overexpressing Puc (UAS-Puc) in the pnr domain. In the embryo, overexpression of Puc phenocopies loss-of-function mutations in the JNK pathway, consistent with the proposed function of Puc as a phosphatase that negatively regulates the JNK pathway by dephosphorylation of Basket. The fact that this is also true in thorax closure represents further evidence that the JNK pathway is involved in thorax closure. Next, a test was performed to see whether D-Fos genetically interacts with components of the JNK pathway during thorax closure. In contrast to the D-fos hypomorphic mutant kay², kay¹ represents a D-fos null allele (a deficiency). The heterozygous allelic combination (kay¹ / kay²) is strictly lethal, but can be rescued by ubiquitous expression of D-Fos under a heterologous promoter. Strikingly, the lethality of kay¹ / kay² could also be rescued by eliminating one copy of the wild type puc gene. More than 50% of the expected Mendelian frequency could be recovered. Thus, puc^E69 has a dominant effect in a kay mutant background, even though heterozygosity for puc^E69 has no phenotypic effects in an otherwise wild-type fly. Furthermore, not only the lethality but also the thorax cleft phenotype of kay mutant flies could be dominantly rescued. The cleft phenotype of the rescued kay² / kay¹ puc flies ranges from strong to very mild. Heterozygous puc^E69 in a kay² homozygous background (kay² / kay² puc E69 ) gives rise to a stable stock in which most flies show a very mild or no thorax cleft at all. Therefore, the puc mutation has a dominant effect on thorax closure and two conclusions can be drawn: (1) Puc must be expressed during thorax closure; (2) as in dorsal closure, Puc negatively regulates the pathway in which D-Fos is acting during thorax closure (Zeitlinger, 1999).

The role of the Drosophila TAK homologue dTAK during development

TGF-ß activated kinase 1 is required during morphogenetic changes and the fusion of the epithelial wing disc cell layers that takes place in thoracic closure, acting in the context of JNK signaling. JNK signaling is required in thoracic closure. The notum of the adult animal is formed by tissue of the two collateral wing imaginal discs, which undergo extensive morphogenetic rearrangements during metamorphosis. LOF in hep/JNKK and kayak/D-Fos results in aberrant wing disc morphogenesis and failure of wing disc fusion, giving rise to a thoracic cleft along the dorsal midline in the adult. To test whether Tak1 can also act in this context, DN (kinase dead) forms of Tak1 (UAS-Tak1^K46R or UAS-Tak1^D159A) were overexpressed with ap-GAL4 and pnr-GAL4 in the thoracic parts of the wing discs. Examination of such flies shows incomplete closure of the thorax, giving rise to a mild thoracic cleft at the dorsal midline of the notum. Although this phenotype is relatively weak, it has a very high penetrance of 91% and is highly reminiscent of that observed in hypomorphic allelic combinations of either hep/JNKK or kay/d-fos. This is also in agreement with the phenotypic result of expressing Puckered, a negative regulator of JNK signaling at the time when wing disc fusion occurs. Puc overexpression driven by pnrGAL4 leads to the same phenotypic thorax cleft defects. These observations suggest that Tak1 is required during morphogenetic changes and the fusion of the epithelial wing disc cell layers, acting in the context of JNK signaling (Mihaly, 2001).

Top-DER- and Dpp-dependent requirements for the Drosophila fos/kayak gene in follicular epithelium morphogenesis

The Drosophila fos/kayak gene is a key regulator of epithelial cell morphogenesis during dorsal closure of the embryo and fusion of the adult thorax. It is also required for two morphogenetic movements of the follicular epithelium during oogenesis: (1) it is necessary for the proper posteriorward migration of main body follicle cells during stage 9; (2) it controls, from stage 11 onwards, the morphogenetic reorganization of the follicle cells that are committed to secrete the respiratory appendages. Egfr pathway activation and a critical level of Dpp signaling are required to pattern a high level of transcription of kayak at the anterior and dorsal edges of the two groups of cells that will give rise to the respiratory appendages. In addition, evidence is provided that, within the dorsal-anterior territory, the level of paracrine Dpp signaling controls the commitment of follicle cells towards either an operculum or an appendage secretion fate. kayak is required in follicle cells for the dumping of the nurse cell cytoplasm into the oocyte and the subsequent apoptosis of nurse cells. This suggests that in somatic follicle cells, kayak controls the expression of one or several factors that are necessary for these processes in underlying germinal nurse cells (Dequier, 2001).

The earliest requirement for kayak in egg chamber development occurs during stage 8/9. The Kayak protein is thus the first factor to be identified that controls the posteriorward migration of main body follicle cells at these stages. It has been suggested that this migratory process involves adhesion molecules, possibly integrins, located within the basal membrane of migrating main body follicle cells. In addition adhesion molecules may be involved in the establishment of a small region of strong adhesion between the posterior-most follicle cells and the posterior region of the oocyte. However, Shotgun (DE-Cadherin), which is necessary for the migration of both border cells and centripetally migrating follicle cells is not required for this migratory process. It could be postulated that the narrowing of main body follicle cells as they migrate posteriorly plays an active role in this process by creating a driving force for the migration of these main body follicle cells towards this posterior region of strong adhesion. In mosaic egg chambers with all follicle cells homozygous for the kay¹ mutation, the migration of main body follicle cells initiates at the correct stage but stops rapidly (Dequier, 2001).

As a consequence, the morphology of the follicular epithelium at late stage 9 is indistinguishable from that of a wild-type egg chamber at early stage 9. This phenotype may reflect the requirement for kayak in the expression of the somatic components of adhesive complexes involved in the migration of main body follicle cells. Alternatively, this migratory defect may reveal its necessity for the reorganization of the shape of main body follicle cells. The data also suggest that the JNK encoding gene basket is also required for this morphogenetic movement. Interestingly, the migration of border cells, which occurs also during stage 9, neither requires kayak nor the JNK pathway. Therefore, these two morphogenetic movements, which are temporally co-ordinated, do not involve the same pathways (Dequier, 2001).

The results strongly suggest that the thin, 'paddleless' or shortened shapes of the respiratory appendages of eggs derived from kay² or kay^1351.3 mosaic egg chambers reflect the requirement for kayak in the reorganization and migration of the respiratory appendage secreting follicle cell (RASFC) prior to and during appendage secretion, and are not an indirect consequence of the partial 'dumpless' phenotype of these egg chambers. This leads to a proposal that the cells displaying the kayak columnar expression pattern together with those accumulating BR-C Z1 characterize a functional respiratory appendage secretion unit whose identity can be traced as early as stage 10B by the expression of these two genes. However, because of the present limitations of clonal analysis in the follicular epithelium, it was not possible to determine whether kayak is required for appendage morphogenesis in the entire unit or solely in the 'G-shaped' rows of columnar follicle cells (CFC) that first start expressing a high level of the gene at stage 10B (Dequier, 2001).

Close parallels can be drawn between the roles played by the kayak gene in migration of main body and respiratory appendage secretory follicle cells and in dorsal closure of the embryo. During dorsal closure, cells from the two lateral ectodermal sheets elongate, migrate dorsally and then fuse along the dorsal midline of the embryo. This process is controlled by the JNK signal transduction pathway and kayak. During embryonic stages 11 and 12, kayak is expressed at a high level in a single row of cells corresponding to the dorsal-most epidermal cells that form the leading edge of the lateral ectoderm, and at a lower level in cells located more ventrally. In kay¹ mutant embryos, these groups of cells initiate a dorsalward stretching which they then fail to maintain and they subsequently resume an unelongated shape after premature termination of the migration process (Dequier, 2001).

In mosaic egg chambers with kay¹ mutant follicular epithelium, the migration of main body follicle cells towards the posterior pole starts properly at early stage 9 but stops almost immediately. Moreover, during the reorganization of the RASFC territories that takes place from stage 10B/11 onwards, Kayak accumulates to a high level first in single rows of cells at the anterior and median edges of these territories, and expands to all RASFCs by stage 12/13. By analogy with the expression pattern of the kayak gene in leading-edge cells of the lateral epidermis during embryonic dorsal closure, it is proposed that expression of kayak in follicle cells located at the edges of the presumptive respiratory appendage territories controls the proper elongation and migration of these cells prior to and during secretion of the respiratory appendages (Dequier, 2001).

Mosaic egg chambers comprised of follicle cells homozygous for the hypomorphic kay² or kay^1351.3 mutations give rise to small and occasionally deflated eggs. In addition, the disappearance of nurse cell nuclei during the latest stages of oogenesis is delayed in these egg chambers. This is reminiscent of the phenotypes of mutations in 'dumpless' genes such as chickadee (chic), quail (qua) or singed (sn), which are required for the rapid transfer of the nurse cell cytoplasm into the oocyte during stages 10B and 11. It has been shown that 'dumpless' eggs display short and broad respiratory appendages. This is likely due to the inhibition of RASFC migration by residual nurse cell material. In contrast, the thin and 'paddleless' phenotype of respiratory appendages of eggs derived from kay² or kay^1351.3 follicular mosaic egg chambers is not the mere consequence of the dumping defect, because no correlation could be drawn between the strength of this defect, as determined by egg length or deflation, and the alterations in respiratory appendages (Dequier, 2001).

It has been shown that the dumping process is driven by nurse cell contractions induced by their subcortical actin filaments that form a dense network. This is in good agreement with the fact that the three 'dumpless' genes chic, qua and sn encode actin-binding proteins. Moreover, these three genes are required in germinal cells only, like the bullwinkle (bwk) gene, which is necessary for proper completion of the dumping process. In contrast, clonal analysis has demonstrated that the requirement for kayak in nurse cell dumping depends on its transcription in the follicular epithelium. This suggests that the transfer of the cytoplasm of nurse cells and the disappearance of their nuclei both involve one or several kayak-dependent somatic signals emanating from the nurse-cell-associated follicle cells. However, the precise role played by the kayak gene in this process remains unclear at present since neither the subcortical network of actin fibers nor the cytoplasmic actin bundles that anchor nurse cell nuclei at stage 10B display any detectable alterations in mosaic kay² or kay^1351.3 egg chambers. However, preliminary experiments show that overexpression of kayak in all CFC using the GAL4-UAS method induces large gaps in the subcortical actin network of the oocyte. This observation suggests that kayak expression in the nurse-cell-associated follicle cells controls subtle aspects of the organization of the actin cytoskeleton in underlying nurse cells (Dequier, 2001).

The data show that determination and localization of the kayak columnar expression pattern requires both Egfr pathway activation and a precise level of paracrine Dpp signaling. The alteration of kayak expression in mutants affecting different components of the Egfr pathway shows clearly that Grk-dependent Egfr activation and secondary Spitz- dependent Egfr amplification and refinement are necessary to determine the kayak columnar expression pattern. Nonetheless, colchicine feeding experiments demonstrate that Grk-dependent Egfr activation is not sufficient to induce kayak transcription in CFC, as is the case for the Egfr target gene kekkon. However, alteration of kayak expression resulting from either a reduction of the Dpp level or its overincrease throughout the columnar epithelium, provides direct evidence that this signaling process is also required for proper patterning of kayak expression (Dequier, 2001).

In C532-GAL4/UAS-dpp females grown at 18°C, a slight increase in the level of Dpp accumulation in CFC induces multiple patches of cells showing a pattern of BR-C Z1 and Kayak accumulation reminiscent of that of respiratory appendage secreting units in wild-type egg chambers. Strikingly, these patches are located at the lateral and posterior peripheries of the dorsal-anterior follicle cell territory, which is consistent with the hypothesis that the central-most CMFC are the localized source of a Dpp gradient. In addition, these results indicate that ectopically provided Dpp in FLP-out clones represses BR-C Z1 and Kayak accumulation in both dpp-expressing cells and those located within a radius of one to two cells, thus providing a direct evidence that Dpp acts in a paracrine manner to repress expression of the BR-C Z1 and kayak genes. The observation that the Dpp-dependent repression of BR-C Z1 is restricted to DAFC suggests that it is mediated by a component of the Dpp-signaling pathway, i.e., either a Dpp receptor or a Smad cofactor expressed differentially in DAFC. It has been shown that among the known Dpp receptors, Saxophone and Punt are ubiquitously expressed in CFC whereas Thick-vein is expressed in a row of anterior follicle cells. In a preliminary investigation of the pattern of expression of the Drosophila Smad genes in follicle cells, it has been observed that medea is expressed from stage 11 onwards in two patches of CFC that may correspond to RASFC. However, whereas the medea gene is required for kayak transcription in the main body follicle cells during stage 9, it appears to be fully dispensable for the kayak columnar expression pattern. Work is currently in progress to investigate the pathway involved in the restriction of the Dpp-dependent repression of BR-C Z1 to DAFC (Dequier, 2001).

JNK signaling pathway is required for efficient wound healing in Drosophila

Efficient wound healing including clotting and subsequent reepithelization is essential for animals ranging from insects to mammals to recover from epithelial injury. It is likely that genes involved in wound healing are conserved through the phylogeny and therefore, Drosophila may be a useful in vivo model system to identify genes necessary during this process. Furthermore, epithelial movement during specific developmental processes, such as dorsal closure (DC), resembles that seen in mammalian wound healing. Since puckered (puc) gene is a target of the JUN N-terminal kinase signaling pathway during DC, puc gene expression was investigated during wound healing in Drosophila. puc expression is induced at the edge of the wound in epithelial cells and Jun kinase is phosphorylated in wounded epidermal tissues, suggesting that the JUN N-terminal kinase signaling pathway is activated by a signal produced by an epidermal wound. In the absence of the Drosophila c-Fos homologue, puc gene expression is no longer induced. Finally, impaired epithelial repair in JUN N-terminal kinase deficient flies demonstrates that the JUN N-terminal kinase signaling is required to initiate the cell shape change at the onset of the epithelial wound healing. It is concluded that the embryonic JUN N-terminal kinase gene cassette is induced at the edge of the wound. In addition, Drosophila appears as a good in vivo model to study morphogenetic processes requiring epithelial regeneration, such as wound healing in vertebrates (Ramet, 2002).

In most cases, flies were anesthetized and then mechanically wounded with iridectomy scissors to cut adult abdomen vertically between the third and the seventh tergites. Semi-thin sections were used to examine the histology of wound healing at the cellular level. The first response to epithelial wound is the formation of a clot at the initial site. Subsequently, the clot becomes melanized making the location of the wound clearly visible. The clot appears to consist of an accumulation of melanin and by hemocytes that aggregated at the site of injury. Hemocytes may be involved also in the clearance of cellular debris and invading microbes (Ramet, 2002).

During the first 2 h after wounding no sign of epithelial cell movement can be seen. In most cases, the edges of the cut epidermis are found far away from the broken cuticle. As for the wounded embryonic epidermis, the adult epidermal layer may be submitted to an intrinsic isotropic epidermal tension that retracts it upon any break injury. By 4 h, the epithelial cells of the edge of the wound seem to shed from the disrupted cuticle. These cells appear larger than the epithelial cells lining the normal cuticle, and exhibit cytoplasmic protrusions. By 12 h, the protrusive cytoplasmic extensions extend from the cells of the edge of the wound and 'migrate' toward each other under the melanin clot, Subsequently, they cause the epidermis to form a suture. These cytoplasmic extensions suggest that adult epidermis is healed by the activity of dynamic lamellipodia or filipodia. Correspondingly, cytoskeleton reorganization has been previously described in wound healing model of cultured Madin-Darby canine kidney cell (MDCK) and during Drosophila DC. At this point, the epithelial cells are still enlarged but start to return to their initial shape. The suture of the epithelium is normally achieved within 18 h after injury. By this time, the wounded epithelium has healed, and cells have returned to their original shape (Ramet, 2002).

To ascertain the importance of melanin production in wound healing, the survival rate of Black cells (Bc) homozygous flies was measured after a transversal wound of the adult abdomen cuticle. Bc/Bc flies lack hemolymphatic phenoloxydase activity and therefore, do not produce melanin. By 24 h after wounding, wild type flies and Bc/+ heterozygous flies, present a dominant melanized crystal cell phenotype, but have wild-type phenoloxidase activity, both of which have about 20% mortality, suggesting an efficient wound repair. In contrast, the vast majority (91%) of wounded Bc/Bc mutants died. 50% mortality of Bc flies was already seen by 6 h, suggesting that phenoloxydase activity is essential early in the wound healing process. Similarly, lozenge (lz) mutants, which lack crystal cells and hence present a weak hemolymphatic phenoloxydase activity, exhibit a poor ability to recover from the injury (Ramet, 2002).

The wound clots differently in Bc flies compared to wild type. In wild type flies, a melanin deposit is observed as early as 10 min after wounding and it is still visible 6 h after injury. In contrast, there is no evidence of melanin formation in the wounded integument of Bc flies, indicating that the latter is of hemolymphatic origin. Furthermore, the two edges of the wound are found apart in Bc flies. This failure to keep the edge of the wound in close proximity leads to death due to bleeding. These results underlie the essential role of the phenoloxydase activity, or an associated phenomenon, when it comes to efficient clot formation and the prevention of bleeding (Ramet, 2002).

To ascertain that the Drosophila JNK pathway is activated in wounded epidermis, DJun N-terminal kinase activity was assayed using anti-phospho-JNK antibodies. Protein extracts from adult abdominal integument from control and wounded flies were assayed for anti-phospho-JNK reactivity. Phospho-JNK can be detected only in the protein extracts from the wounded epidermis, whereas the unphosphorylated form of DJNK is present in all of the samples. This indicates that DJNK is phosphorylated in response to wounding in epidermal cells, and therefore, suggests that the JNK signaling pathway is activated (Ramet, 2002).

In contrast to embryonic DC where only the most dorsal cell row at the leading edge is expressing puc gene, several rows of adult epidermal cells show a strong ß-galactosidase activity during wound healing. This result is consistent with high DJNK activity in the vicinity of the wound. Indeed, the extent of the area expressing puc-lacZ clearly depends on the size of the wound (up to 8 cell rows). Furthermore, puc gene expression showed a decreasing gradient from the edge of the wound towards healthy epithelium. This suggests that a newly formed signal emerges from the wound and diffuses through the epidermal layer (Ramet, 2002).

During embryonic DC, the DJNK pathway activates puc expression in the LE and as a negative feedback loop, puc itself down-regulates the DJNK. To further test if the DJNK pathway also mediates puc induction during wound healing, puc expression pattern in adult epidermis was analyzed in fly mutants of this pathway. In hep, a hypomorphic allele of hemipterous, (encoding the MAPKK), puc-lacZ expression is almost completely normal. This result is not totally unexpected since hep flies present no epidermal defects. Stronger hep allele mutations are lethal and could not be studied. However, a heteroallelic combination of kay mutations leads to viable flies. Since the incidence of dorsal thoracic cleft phenotype of this mutant is higher than that in hep, it is more likely to affect the regulation of puc expression. In kay¹/kay² animals, puc gene induction is drastically reduced at the site of injury compared to wild-type. This result demonstrates that puc regulation is dependent on the transcriptional activator DFos during wound healing. Interestingly, still 18 h after wounding, the mutant cells are separated from the edge of the wound comparably to that observed with wild type at 3 h after wounding. This suggests that the epidermal sheet has been unable to spread under the wound and that the process is blocked at the initiation stage. Since DFos, together with DJun, is the target of the DJNK pathway, and forms the transcription factor AP-1, it is likely that the DJNK pathway is switched on by an integument injury (Ramet, 2002).

To find out if DFos mutation has a cell autonomous effect, the UAS/GAL4 system was used to express a dominant negative form of DFos in the pannier (pnr) expression domain. In the pnr-Gal4 line, Gal4 protein is expressed in a large dorsal band of adult epidermis. A continuous wound was done to overlap this dorsal epidermal expression domain and the dorso-lateral and ventral epidermal domain. puc-lacZ expression was then assayed 12 h after wounding. In control flies, puc is expressed at the wounded epidermis independent of the location. When DFos^bZip dominant negative form of DFos is expressed in the dorsal band, X-Gal staining shows a clear, albeit not total, reduction of puc expression at the expected places. In contrast, puc expression is induced normally outside of the pnr expression domain. This demonstrates that the puc gene induction is under the control of the DFos transcriptional factor in a cell-autonomous manner similar to that observed during dorsal and thorax closure (Ramet, 2002).

To ascertain the importance of the DJNK pathway in wound healing, the phenotype of kay mutant was investigated during the course of wound healing. As expected, the wounded epidermis from kay deficient flies fails to recover. The epithelial cells at the edge of the wound also fail to undergo any evident cell shape change or show any cytoplasmic protrusive extensions. Even at 18 h after injury, the wound is not repaired. Therefore, the transcriptional activator DFos appears necessary for a normal epithelial repair in adult Drosophila. Interestingly, over a period of 6 days, wounded mutant flies did not suffer any higher mortality compared to wounded wild type flies, suggesting that epithelial repair is not crucial for early survival (Ramet, 2002).

An essential function of AP-1 heterodimers in Drosophila development

Fos and Jun proteins homo- or hetero-dimerize to form functional AP-1 transcription factors. Drosophila mutants lacking either Jun or Fos display indistinguishable dorsal open phenotypes, indicating an essential function of both Jun and Fos for embryonic dorsal closure. Experiments were carried out to determine the basis for this dual requirement. By combining mutant alleles and transgenes expressing Fos and Jun variants with altered dimerization preferences, fly lines were generated in which only specifically defined dimer variants could form. Phenotypic analysis of these mutants reveals that homodimers of Fos or of Jun cannot replace the function of the heterodimeric complex. This defect is not explained by the lower stability of homodimers as compared to heterodimers, because 'pseudo-homodimers' which are as stable as native Jun-Fos heterodimers, cannot substitute for native Jun-Fos function. It is concluded that Jun and Fos play complementary roles and that both are required for signal transduction and gene activation during dorsal closure (Ciapponi, 2002).

To compare the role of Jun-Fos heterodimers and homodimers, two types of 'zipper swap mutants' were generated. The FJF mutant represents a version of D-Fos, in which the leucine zipper was precisely replaced with the corresponding domain of D-Jun. The complementary construct, dubbed JFJ, is a D-Jun mutant carrying the D-Fos leucine zipper. This design was chosen so that FJF would be able to form 'pseudo-homodimers' with wild-type Fos, which should have the same stability as Fos-Jun heterodimers. Conversely, FJF should dimerize with Jun only weakly with the affinity of a Jun homodimer. To confirm the expected dimerization characteristics of the chimeric proteins, they were analyzed in a GST pull-down assay. In vitro translated and ³⁵S-labeled FJF or JFJ proteins were incubated in various combinations with bacterially expressed Jun or Fos GST fusion proteins, or with GST alone as a negative control. Retention of radiolabeled JFJ and FJF proteins by GST proteins, which were immobilized on Sepharose beads, was visualized by autoradiography. The results of this experiment indicate that both homo- and hetero-dimeric complexes can form in vitro, with Fos-Fos homodimers being significantly less stable than Jun-Jun homodimers or Jun-Fos heterodimers. It is worth noting that dimerization occurred in the absence of AP-1 binding sites, and might be further stabilized when the dimeric complexes bind to DNA (Ciapponi, 2002).

Whether the zipper swap mutants could replace the function of endogenous Jun or Fos proteins during embryogenesis and rescue the respective mutants when expressed as transgenes was tested. The following mutant alleles were used. Animals homozygous for the jun² null allele express no D-Jun protein and are embryonic lethal. They can be rescued to adulthood by expression of a D-jun transgene. fos mutant alleles are designated kayak. kay¹ is a null allele causing a phenotype that is indistinguishable from that of jun² mutants. The dorsal closure phenotype of kay¹ homozygotes can be rescued by a D-fos transgene; however, the animals do not survive to adulthood due to the loss of one or more essential genes in addition to D-fos on the kay¹ chromosome. The kay² allele, while solely affecting D-fos, only represents a partial loss of function mutation. Occasionally, kay² homozygotes survive to adulthood and show a characteristic thorax cleft phenotype. kay¹/kay² transheterozygotes are strictly lethal. D-jun and D-fos mutant flies provide a background for in vivo complementation assays in which engineered forms of these proteins can be functionally tested in the developing organism (Ciapponi, 2002).

The overexpression of the wild-type D-Fos protein from a transgene in a jun homozygous mutant embryo or of wild-type D-Jun in a kay¹ mutant background is not sufficient to rescue the DC mutant phenotype. This indicates that neither D-Fos nor D-Jun homodimers alone are sufficient to direct the dorsal closure process, even when expressed at elevated levels (Ciapponi, 2002).

If it were the higher stability of the Fos-Jun heterodimer as compared to the two respective homodimers that was required to provide sufficient AP-1 function for the completion of DC, then 'pseudo-homodimers' of Jun and JFJ or of Fos and FJF held together by the Fos-Jun zipper interaction might be expected to rescue the dorsal open phenotype of kay or jun mutants, respectively. To test this possibility, Drosophila stocks carrying JFJ or FJF transgenes under the control of the heat shock promoter were recombined with the kay¹ or with the jun² mutant allele, respectively. In the animals of the hs FJF, jun² and the hs JFJ, kay¹ genotypes only stable Fos-FJF or Jun-JFJ 'pseudo-homodimers' but no Fos-Jun heterodimers can form. In both cases, no rescue could be observed, i.e. no viable flies of the observed genotype could be recovered, nor could either mutant carry out DC. This result indicates that Jun or Fos homodimers are not sufficient for DC to occur properly, even when the homodimer is held together by a more stable hetero-leucine zipper interaction (Ciapponi, 2002).

Next, flies were generated in which the only possible heterodimers are FJF-Jun or JFJ-Fos, respectively. Essentially, these are heterodimers that are nevertheless held together by the weak homotypic interaction between either two Jun or two Fos leucine zippers. These animals carry the hs FJF transgene in a homozygous kay¹ background or the JFJ transgene in flies that are homozygous for the jun² allele. Significantly, FJF and JFJ can rescue the mutant DC phenotypes and the lethality of fos and jun mutants, respectively, in both these combinations. Expression of the hs FJF transgene at least partially suppresses the completely penetrant DC phenotype of kay¹ mutants. Moreover, the strict lethality of kay¹/kay² transheterozygotes can be rescued to adulthood by the hs FJF transgene. Thus, in the different kay mutant backgrounds, FJF expression has the same rescuing potential as transgenic expression of a wild-type Fos protein. The similarity also extends to the adult phenotype of the kay¹/kay² flies that are rescued by FJF or by wild-type D-Fos expression. In both cases adults show a notum cleft phenotype, reminiscent of occasional homozygous escapers of the hypomorphic kay² stock. In line with this result, the JFJ transgene when expressed in a jun² null allele background significantly reverts the dorsal open phenotype (Ciapponi, 2002).

Several conclusions can be drawn from these experiments: (1) the data indicate that both Fos and Jun make non-redundant contributions to the regulation of dorsal closure independent of their leucine zippers, since stabilized homodimers cannot rescue Fos or Jun loss-of-function mutations, whereas destabilized heterodimers can do this. What could the complementary functions of Fos and Jun be? Recent results have indicated that both Fos and Jun represent primary recipients of JNK signaling which is essential for DC and can serve as substrates for the Drosophila JNK homolog, Basket. Both have transcription activation domains. Thus, the functional differences and the basis for the cooperation between Fos and Jun might be more specific. Either transcription factor may contribute distinct contacts to the initiation machinery or mediate separate contacts to other DNA-bound transcription factors in the assembly of regulatory complexes on target gene promoters and enhancers (Ciapponi, 2002).

(2) The results further indicate that homotypic interactions, mediated by two Jun or two Fos leucine zipper domains (such as between FJF and Jun) are in principle stable enough to assemble AP-1 dimers in the animal. Therefore, it is possible that in biological situations other than DC, Jun or Fos might act independently and that target genes exist that can be regulated by Jun or Fos homodimers. Indirect evidence indicates that Fos may have functions that are Jun-independent, possibly as a homodimer (Ciapponi, 2002).

Highwire restrains synaptic growth by attenuating a MAP kinase signal

Highwire is an extremely large, evolutionarily conserved E3 ubiquitin ligase that negatively regulates synaptic growth at the Drosophila NMJ. Highwire has been proposed to restrain synaptic growth by downregulating a synaptogenic signal. This study identifies such a downstream signaling pathway. A screen for suppressors of the highwire synaptic overgrowth phenotype yielded mutations in wallenda, a MAP kinase kinase kinase (MAPKKK) homologous to vertebrate DLK and LZK. wallenda is both necessary for highwire synaptic overgrowth and sufficient to promote synaptic overgrowth, and synaptic levels of Wallenda protein are controlled by Highwire and ubiquitin hydrolases. highwire synaptic overgrowth requires the MAP kinase JNK and the transcription factor Fos. These results suggest that Highwire controls structural plasticity of the synapse by regulating gene expression through a MAP kinase signaling pathway. In addition to controlling synaptic growth, Highwire promotes synaptic function through a separate pathway that does not require Wallenda (Collins, 2006).

JNK signaling affects many cellular processes, often by regulating transcription factor activity that leads to changes in gene expression. A common downstream effector of JNK-mediated changes in gene expression is the AP-1 complex of Fos and Jun transcription factors, which can regulate synaptic growth at the Drosophila NMJ. To investigate whether Drosophila Fos or Jun (known as D-fos and D-jun, respectively) are required for highwire-dependent synaptic overgrowth, each was inhibited by expressing dominant-negative transgenes that contain the DNA binding and dimerization domains of Fos and Jun but lack the transcriptional activation domains. Expression of these dominant-negative transgenes in postmitotic neurons allowed circumvention of early embryonic requirements for D-fos and D-jun (Collins, 2006).

When Fos^DN and Jun^DN are neuronally expressed in a wild-type background, there is a modest trend toward inhibition of synaptic growth. When expressed in a highwire mutant background, the Fos^DN transgene confers dramatic suppression of the highwire synaptic phenotype, reducing bouton number and branching (42%) and increasing the intensity of staining for synaptic vesicle markers at the synapse. The reduction in highwire-dependent synaptic overgrowth is much greater than the reduction of growth in a wild-type background. In contrast, Jun^DN does not suppress the highwire phenotype. This suggests the existence of a pathway that is separate from AP-1, consistent with results in Drosophila demonstrating that D-Fos can act independently of D-Jun. The requirement for D-Fos in highwire synaptic overgrowth suggests that the highwire phenotype involves changes in gene expression rather than exclusively local changes to the synapse (Collins, 2006).

If Fos^DN acts downstream of Wallenda to inhibit synaptic overgrowth, it should also suppress the synaptic overgrowth caused by overexpressing wallenda. Indeed, when Fos^DN was coexpressed with UAS-wnd in neurons, Fos^DN could suppress the wallenda gain-of-function phenotype, leading to a 38% reduction in synaptic bouton number, a 52% reduction in synaptic branching, a 54% increase in bouton size, and a 3.8-fold increase in the intensity of staining of synaptic vesicle markers. This is consistent with D-Fos acting downstream of Wallenda to promote synaptic growth. Therefore, the synaptic overgrowth phenotypes caused by loss of highwire and by overexpression of wallenda are similar in their requirements for the transcription factor D-Fos (Collins, 2006).

Current models suggest that Highwire functions as an E3 ubiquitin ligase to downregulate a signaling pathway that promotes synaptic growth. This study identified a MAPKKK, Wallenda, whose protein levels are controlled by Highwire and the activity of ubiquitin hydrolases. Wallenda is both necessary for highwire-dependent synaptic overgrowth and sufficient to promote synaptic growth. Downstream of Wallenda, the MAP kinase JNK and transcription factor Fos are required for highwire-dependent synaptic overgrowth. It is proposed that Highwire restrains synaptic growth by downregulating the MAPKKK Wallenda, thereby inhibiting signaling through the JNK MAP kinase and the Fos transcription factor. In the absence of highwire, this signaling pathway is overactive, leading to changes in gene expression that result in excessive synaptic growth (Collins, 2006).

The regulation of the MAPKKK Wallenda is conserved in Drosophila and C. elegans (Nakata, 2005). In both organisms, the synaptic phenotype of highwire/rpm-1 requires the Wallenda/DLK-1 MAPKKK and downstream MAPK signaling. However, the downstream MAPK pathways diverge: in C. elegans, the rpm-1 phenotype requires a p38 MAP kinase (Nakata, 2005), while the highwire phenotype requires JNK signaling. This suggests that regulation of the specific MAPKKK Wallenda/DLK-1, rather than a particular downstream MAP kinase pathway, is a fundamental activity of Highwire and its orthologs (Collins, 2006).

Since Highwire functions as an E3 ubiquitin ligase to restrain synaptic growth, Wallenda is a compelling candidate target for the following reasons: (1) wallenda functions downstream of highwire and is essential for the synaptic overgrowth in highwire mutants; (2) increasing the levels of Wallenda by overexpression is sufficient to confer synaptic overgrowth; (3) Highwire regulates Wallenda protein levels through a posttranscriptional and most likely posttranslational mechanism. Each of the points above is conserved in C. elegans (Nakata, 2005 ). (4) Wallenda protein levels are regulated by ubiquitination in vivo, since inhibiting ubiquitination by overexpressing ubiquitin hydrolases increases the levels of Wallenda protein. (5) The RING domain of the C. elegans homolog rpm-1 can interact with the Wallenda homolog DLK-1 (Nakata, 2005) and stimulate its ubiquitination when both are overexpressed in 293T cells (Collins, 2006).

Targeting a MAPKKK, which sits at the top of a MAP kinase signaling pathway, is an attractive mechanism for spatially and temporally controlling a synaptogenic signal without affecting downstream components shared by multiple MAPK signaling cascades. Restraining MAP kinase signaling is essential for controlling diverse cellular processes, including cell proliferation, differentiation, and apoptosis. The targeting of MAPKKKs by specific ubiquitin ligases may be a powerful and general mechanism for regulating MAP kinase signals (Collins, 2006).

While Wallenda is an essential mediator of the highwire mutant phenotypes in both Drosophila and C. elegans, an endogenous synaptic function for Wallenda has not yet been identified in either organism: the wallenda mutants have surprisingly normal synapse morphology and function. This may be due to another pathway that compensates for the loss of wallenda function. Such redundancy would obscure the role of wallenda. A second possibility is that wallenda functions in an aspect of synaptic growth that is not detected or required under laboratory culture conditions. For instance, wallenda could promote synaptic growth as part of a structural plasticity program that responds to unknown experience-dependent stimuli. A third possibility is that Wallenda does not normally function at synapses, but its upregulation in highwire mutants causes a neomorphic phenotype. In this scenario, the regulation of Wallenda by Highwire is required for normal synaptic development, but endogenous Wallenda would not itself regulate the synapse. The neuropil and synaptic localization of Wallenda and the vertebrate homolog DLK (Hirai, 2005) is, however, consistent with a synaptic function (Collins, 2006).

As an activator of MAP kinase signaling, Wallenda and its homologs might also control other processes beyond the synapse. Functional studies in vertebrates suggest that DLK and JNK signaling regulate neuronal migration and axon outgrowth in the developing cortex (Hirai, 2002). Outside of the nervous system, DLK influences keratinocyte differentiation, and LZK is highly expressed in the pancreas, liver, and placenta. In Drosophila, wallenda mutants are female sterile. It is predicted that the regulation of DLK and LZK is conserved from worms and flies to vertebrates. Therefore, the vertebrate homologs of Highwire might regulate some of these neuronal and/or extraneuronal developmental processes (Collins, 2006 and references therein).

Highwire is a large, multidomain protein that, in addition to acting as an E3 ubiquitin ligase, has been shown to inhibit adenylate cyclase, influence TSC signaling and pteridine biosynthesis, and interact with the myc oncogene and the co-SMAD Medea. It is remarkable that throughout millions of years of evolution, members of the Highwire family have retained an exceptionally large size and complex domain structure. An attractive explanation for this conservation is that this molecule could serve as an intersection point for multiple signaling pathways, integrating MAP kinase and other signals during neural development (Collins, 2006).

The ubiquitin ligase activity alone could be responsible for regulating more than one downstream target. Interactions with components of TSC (tuberin/hamartin) and TGF-β signaling pathways suggest that Highwire might target either or both of these pathways. The model that Highwire regulates TGF-β signaling through interaction with the co-SMAD Medea has received considerable attention. Since the TGF-β pathway regulates synaptic growth at the NMJ, it has been proposed that synaptic overgrowth of highwire mutants is caused by overactivity of this pathway. Null alleles of wit, which completely disrupt TGF-β signaling at the NMJ, can partially suppress the highwire phenotypes: they partially suppress the increase in bouton number, but show little or no suppression of the reduced bouton size and the reduced intensity for synaptic vesicle markers. This partial suppression of highwire by wit is consistent with the model that overactive TGF-β signaling contributes to the highwire phenotype. However, the data are also consistent with the alternate model that TGF-β signaling and Highwire act in parallel pathways. An assay for the activity of TGF-β signaling is to stain for phosphorylated-MAD (phospho-MAD), the major transducer of BMP signals in Drosophila, in motoneuron nuclei. No change was detected in the levels of phospho-MAD staining in highwire mutants compared to wild-type. This assay is sensitive to changes in pathway activity—neuronal expression of the constitutively active type I receptor thick veins leads to a 40% increase in phospho-MAD staining. Interestingly, this increase in TGF-β signaling does not lead to excess synaptic growth. Combining a highwire mutant with expression of constitutively active thick veins does cause excess growth, but it does not lead to any further increase in phospho-MAD staining. These data are consistent with highwire and TGF-β signaling acting in parallel pathways (Collins, 2006).

Whether or not Highwire regulates TGF-β signaling, it is likely to target an additional pathway. Highwire not only restrains synaptic growth, but also promotes synaptic function. Synaptic function requires the ubiquitin ligase activity of Highwire and is sensitive to the levels of the ubiquitin hydrolase fat facets. This study demonstrates that this regulation of neurotransmitter release does not require Wallenda. Therefore, Highwire must regulate at least two distinct molecular pathways. If Wallenda is a substrate whose downregulation is essential for restraining synaptic growth, there is likely another substrate for Highwire whose downregulation promotes neurotransmitter release (Collins, 2006).

Downstream of Wallenda, the JNK MAP kinase and Fos transcription factor are required for the highwire synaptic morphology phenotype. Therefore, Highwire attenuates a JNK signaling pathway that presumably controls gene expression to regulate synaptic growth. Previous studies have implicated JNK-dependent transcriptional control in activity-dependent growth of the Drosophila NMJ. However, this previously described pathway is probably distinct from the JNK signal that is controlled by Highwire and activated by Wallenda. The previously described role for JNK requires AP-1, a heterodimer of Fos and Jun transcription factors; inhibiting either D-Fos or D-Jun disrupts this pathway. In contrast, highwire-induced overgrowth requires D-Fos, but not D-Jun. The Wallenda pathway could therefore involve a homodimer of D-Fos or another transcription factor that interacts with Fos. Such D-Jun-independent functions of D-Fos have been described previously in Drosophila. The differential requirement for transcription factors suggests that the output of Wallenda signaling cannot simply be activation of JNK, but instead activation of JNK in a particular spatial or temporal context, such as in the presence of cofactors that influence downstream signaling (Collins, 2006).

In addition to transcription factors, substrates for activated JNK include components of the cytoskeleton. Because the NMJ is distant from the motoneuron nucleus, and because vertebrate DLK colocalizes with tubulin in axonal regions of the brain, it was initially expected that the Highwire/Wallenda/JNK pathway would influence synaptic morphology through local action upon the synaptic cytoskeleton. Instead, a requirement was identified for a transcription factor and presumably changes in gene expression. However, this does not exclude an interaction with the cytoskeleton or local changes at the synapse. It is possible that Highwire regulates the Wallenda signal in the cell body. However, the observation that Wallenda accumulates in the synapse-rich neuropil and at the NMJ when Highwire is absent suggests that Wallenda could become activated at the synapse. This would imply the need for a mechanism to transport the activated JNK signal back to the nucleus. In addition, cell-wide changes in gene expression must then be translated into localized growth at the synapse. Activated Wallenda at the synapse is an attractive candidate to integrate changes in gene expression with regulation of the synaptic cytoskeleton to control synaptic growth (Collins, 2006).

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