Modulation of Morphogenesis by Egfr during Dorsal Closure in Drosophila

During Drosophila embryogenesis the process of dorsal closure (DC) results in continuity of the embryonic epidermis, and DC is well recognized as a model system for the analysis of epithelial morphogenesis as well as wound healing. During DC the flanking lateral epidermal sheets stretch, align, and fuse along the dorsal midline, thereby sealing a hole in the epidermis occupied by an extra-embryonic tissue known as the amnioserosa (AS). Successful DC requires the regulation of cell shape change via actomyosin contractility in both the epidermis and the AS, and this involves bidirectional communication between these two tissues. We previously demonstrated that transcriptional regulation of myosin from the zipper (zip) locus in both the epidermis and the AS involves the expression of Ack family tyrosine kinases in the AS in conjunction with Dpp secreted from the epidermis. A major function of Ack in other species, however, involves the negative regulation of Egfr. We have, therefore, asked what role Egfr might play in the regulation of DC. Our studies demonstrate that Egfr is required to negatively regulate epidermal expression of dpp during DC. Interestingly, we also find that Egfr signaling in the AS is required to repress zip expression in both the AS and the epidermis, and this may be generally restrictive to the progression of morphogenesis in these tissues. Consistent with this theme of restricting morphogenesis, it has previously been shown that programmed cell death of the AS is essential for proper DC, and we show that Egfr signaling also functions to inhibit or delay AS programmed cell death. Finally, we present evidence that Ack regulates zip expression by promoting the endocytosis of Egfr in the AS. We propose that the general role of Egfr signaling during DC is that of a braking mechanism on the overall progression of DC.


Introduction
Dorsal closure (DC) is a developmental event occurring in the Drosophila embryo between stages 11 and 16, commencing immediately after germband retraction (reviewed in [1]). Upon the completion of germband retraction, a large dorsal opening is evident in the epidermis. The amnioserosa (AS), an extraembryonic tissue composed of a single layer of large flat epithelial cells, spans the opening. The two opposing lateral epidermal flanks elongate in the dorsal-ventral (D-V) axis and move dorsal ward to seal the dorsal hole. The dorsal-most epidermal (DME) cells from one side of the embryo meet their counterpart DME cells at the dorsal midline. The epidermal sealing process occurs in a zipperlike manner, progressing simultaneously from both the anterior and posterior ends of the dorsal opening and finishing at the center of the dorsal midline. During DC, the AS contracts and its cells become more cuboidal in shape; the AS also actively extrudes approximately 10% of its cells with the effect of increasing the rate of DC [2][3][4]. Upon the completion of DC, the entire AS degenerates by programmed cell death [2].
The DME cells experience a range of morphogenetic events during DC. These include elongation in the D-V axis, formation of actin-based membrane extensions, and adhesion with their partners from the other side of the embryo. DC is a popular model system to study cell shape change in epithelial morphogenesis and multiple signaling proteins have been characterized in this context. In addition to signaling pathways and proteins, there are a number of mechanical forces driving DC [4](reviewed in [5]). These include a supracellular actomyosin cable that is assembled at the leading edge (LE) of the DME cells to form a contractile ''purse string''. This contractile apparatus constricts the DME cells in the anterior-posterior axis and thus contributes to their stretching in the D-V axis and movement towards the dorsal midline. Actinbased filopodia and lamellipodia also project from the leading edge of the DME cells, and these are thought to contribute to the alignment and adhesion of opposing DME cells as DC concludes [6,7]. Finally, as was demonstrated by elegant laser microdissection experiments, AS constriction not only removes this tissue as an impediment to movement of the epidermis, but also pulls the DME cells dorsal ward [4].
Among the numerous signaling proteins known to regulate DC is Dpp, a member of the transforming growth factor-b superfamily of cytokines. Dpp expression in DME cells is required for morphogenesis of both the amnioserosa and the epidermis during DC. This requirement for Dpp expression in the DME cells is associated, at least in part, with the regulation of the expression of zipper (zip), which encodes non-muscle myosin II heavy chain [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. We previously demonstrated that two members of the Ack family of nonreceptor tyrosine kinases, Ack and PR2, co-operate with Dpp to regulate myosin levels in the AS and epidermis during DC. This work led us to propose the existence of a diffusible signal that is generated by the AS and is regulated by Ack and PR2 in conjunction with Dpp [26]. The zip product is required for cell shape change in both the AS and the epidermis; zip expression, which is regulated by the Ack/PR2/Dpp signaling network, may ultimately coordinate the overall progression of DC [27].
There is considerable evidence that a major function of Ack is the negative regulation of Egfr, and this is thought to occur through the regulation of Egfr by endocytosis and/or ubiquitination [28][29][30][31][32]. Egfr may, therefore, play a key role in the Ack/PR2/ Dpp regulatory pathway during DC. The Egfr pathway is used repeatedly throughout Drosophila development and appears to regulate a myriad of processes including cell proliferation, cell differentiation, apoptosis, cell motility and adhesion (reviewed in [33,34]). While it has long been recognized that Egfr has multiple roles in regulating morphogenesis, including germband retraction, its role in DC has not been specifically addressed [35,36].
The results presented here demonstrate that Egfr is required in both the AS and epidermis for proper DC. In addition, we demonstrate that the function of Egfr in the AS involves the transcriptional repression of zip in both the AS and the DME cells, and we suggest that this repression involves the regulation of the same diffusible signal previously proposed to be regulated through Ack and PR2. Consistent with this, we present results suggesting that Ack negatively regulates Egfr in the AS by targeting it for endocytosis. Similar to Ack, we suggest that Egfr regulates zip expression in parallel to Dpp signaling, but we find that Egfr also has a strong negative effect on epidermal Dpp transcription. Finally, we confirm that Egfr signaling has an additional role in the AS as an inhibitor of apoptosis. The various roles identified for Egfr signaling during DC are consistent with negative regulation of morphogenesis, and we propose that Egfr acts as a brake to adjust the rate of closure in response to endocytic regulation.

Egfr is Required for Normal DC
To address the potential function of Egfr in DC we chose three alleles previously shown to disrupt embryogenesis: Egfr f2 , a severe loss-of-function allele, Egfr 2C82 , a moderate loss-of-function allele, and Egfr 1F26 , a conditional allele [35][36][37]. Two previously uncharacterized embryonic lethal alleles, Egfr 1a15 and Egfr H25 , were also used in the course of this study [38]. Phenotypic analysis of the latter two alleles indicated that Egfr 1a15 is a severe loss-offunction allele (equivalent to Egfr f2 ) and that Egfr H25 is also a strong loss-of-function allele, but is slightly less severe than either Egfr f2 or Egfr 1a15 (data not shown). Cuticle preparations of embryos heteroallelic for Egfr f2 and Egfr 2C82 showed severe defects in morphogenesis. Greater than 95% of embryos exhibited the previously described ''faint little ball'' phenotype [36,37], being ''curled'' up, with the posterior end of the embryo in close proximity to the head, indicating a defect in germband retraction (Fig. 1B). In less severely curled embryos it was possible to observe holes in the dorsal surface that typically extended anteriorly into the head (Fig. 1C). In general, severe Egfr mutants were associated with a terminal phenotype that included severe defects in head development as well as a complete failure in germ band retraction -both of which effectively precluded any analysis of DC. We, therefore, sought approaches that would allow us to observe DC defects in individuals with impaired Egfr function. The first of the two approaches involved temperature shift experiments using the conditional allele Egfr 1F26 [35]. Embryos collected at the permissive temperature of 18uC were aged for various periods of time before being shifted to the restrictive temperature of 29uC. Embryos transferred to 29uC at any stage prior to stage 10 showed severe cuticle defects similar to those seen with other strong loss-offunction Egfr alleles (data not shown). Shifting slightly older (approximately stage 11) Egfr 1F26 mutant embryos to 29uC, however, resulted in a less severe mutant phenotype comprising a smaller head hole, distinct from other defects in the dorsal surface, and a less severe defect in germband retraction (Fig. 1D). In addition, these embryos consistently displayed creases or ''puckers'' in the dorsal surface, which together with the mild germband retraction defect, gave them a bowed appearance. Approximately 52% of Egfr 1F26 embryos were scored as having this bowed phenotype in a temperature shift experiment where a 2.5hour collection of embryos from the Egfr 1F26 stock was aged at 18uC for 12 hours and then shifted to 29uC (corresponding to late stage 10/stage 11 at the time of the temperature shift). Of these bowed embryos, half exhibited an additional phenotype of a hole or scab in the dorsal surface (Fig. 1D). Few defects were seen in the dorsal surface of Egfr 1F26 embryos when they were shifted to 29uC at stage 12 or later (data not shown).
A second approach for facilitating the analysis of DC defects in Egfr-deficient embryos, which also permitted characterization of the tissue specificity of Egfr function, involved the inducible expression of a dominant negative version of Egfr, Egfr DN . Egfr DN competes with endogenous Egfr for ligand binding, but lacks the cytoplasmic domain, which contains the tyrosine kinase domain necessary for trans-phosphorylation and receptor activation [39]. As a result, Egfr DN attenuates activation of the Egfr cascade in a cell autonomous manner. We expressed a UAS-Egfr DN transgene in various spatial patterns, starting with general epidermal expression and epidermal stripes using the 69B-Gal4 and ptc-Gal4 drivers, respectively [40,41]. Both patterns of Egfr DN expression resulted in bowed embryos with occasional dorsal holes or scabs, similar to the temperature shift phenotype (Fig. 1E, F). 100% of embryos in which Egfr DN was expressed with ptc-Gal4 were bowed and 10% had a dorsal hole or scab. The dorsal epidermis plays an important role in DC and we impaired Egfr function in this tissue by expressing Egfr DN using the LE-Gal4 driver, which is active only during DC primarily within a subset of cells in the first two rows of dorsal epidermal cells flanking the amnioserosa (Fig. 1G, H, and Movie S1) [13]. This resulted in a bowed embryo phenotype in about a quarter of Egfr DN -expressing embryos (Fig. 1I). Bowed embryos and dorsal holes were also seen when Egfr activity was blocked in the AS using the AS-specific driver Gal4 c381 [42] (Fig. 1J). We conclude that the normal progression of DC requires Egfr signaling in both the epidermis and the AS.
A major route for signaling by Egfr is the Raf-MAPK pathway, the activation of which can be detected using anti-phospho-MAPK antibodies [43,44]. We observed strong phospho-MAPK immunoreactivity in the central AS cells of wild-type embryos, but little staining in cells of the AS periphery or in the dorsal epidermis ( Fig. 1 K, K'). This result suggests that either Egfr is not using the MAPK pathway in these cells or that the pathway is under tight negative control. Consistent with Egfr signaling generating phospho-MAPK in the AS, the anti-phospho-MAPK immunore- Egfr Signaling during Dorsal Closure PLOS ONE | www.plosone.org activity in the AS was absent in embryos in which Egfr DN was expressed with Gal4 c381 (data not shown).
We extended our analysis of Egfr function in DC by live imaging embryos homozygous or heteroallelic for the alleles Egfr f2 [37], Egfr 1a15 and Egfr H25 [38] and carrying a Ubi-DEcadherin-GFP transgene to visualize cell outlines [45]. At least four movies were taken for each genotype and very consistent phenotypes were observed. Prior to the initiation of germband retraction, degradation of the AS commenced in Egfr mutant embryos, which in the most severe cases led to a complete and dramatic loss of the tissue (Compare Movie S2 to Movies S3 and S4, and Fig In some embryos the AS persisted throughout germband retraction and DC, but had noticeably fewer cells than wild-type (Movie S5 and Fig. 2G, H, M-P). In such embryos germband retraction proceeded to a point and then appeared to reverse, with the posterior end of the embryo moving anteriorly. Accompanying this was a bunching of the epidermis characterized by inappropriate adhesion between the dorsal end of non-adjacent segments on the same side of the embryo producing (arrowhead in Fig. 2H). In addition, AS morphogenesis was abnormal, with the tissue constricting perpendicular to the normal anterior-posterior direction (Compare Movie S5 to Movie S6 and Fig. 2I-L to Fig. 2M-P). Finally, heads of Egfr mutant embryos exhibited a precipitous loss of epidermal integrity during embryogenesis with the brain becoming exposed during DC and pushing toward the posterior end of the embryo (Fig. 2G, H, Movie S5).
We also assessed morphogenesis in fixed embryos by staining with an antibody against phosphotyrosine to reveal cell outlines. Heteroallelic Egfr mutant embryos showed highly variable cell shape change of the DME cells compared to wild-type, suggestive of misregulated actomyosin contractility (Fig. 2Q, R). We suspect that this uneven contractility in the dorsal epidermis underlies the bunching of the segments in Egfr mutant embryos, and consistent with this expression of Egfr DN with LE-Gal4 led to segmental bunching (Fig. 2S).

Egfr Negatively Regulates dpp Expression in the Epidermis During DC
The bowed embryo phenotype associated with reductions in Egfr function is similar to the cuticle phenotype of embryos in which Dpp is ectopically expressed in the dorsal epidermis [17,[46][47][48]. Furthermore, crosstalk between Egfr and Dpp/TGFb signaling in the form of antagonistic or cooperative interactions has been reported for a number of developmental events . We, therefore, next examined dpp expression in embryos having altered Egfr function. We confirmed a previously published observation that ectopic dpp expression extends ventrally along the segmental grooves of Egfr mutant embryos [35], and saw a similar pattern of dpp expression in embryos in which Egfr signaling was attenuated in the epidermis via UAS-Egfr DN expression using 69B-Gal4 or LE-Gal4 drivers ( Fig. 3C-F, arrowheads). The phenocopy of Egfr loss-of-function mutants by UAS-Egfr DN expression was underscored by the fact that widespread epidermal expression of this transgene resulted in a reduction in the separation, from one side of the embryo to the other, between the ventral stripes of dpp expression as previously reported for Egfr and D-raf mutant embryos [35,83]. This decrease in separation allowed the ventral stripe on the other side of the embryo to be seen in a lateral view (arrows in Fig. 3C, E). Given the many studies indicating communication between the AS and epidermis during DC [10,13,25,26,[84][85][86][87][88][89] and the requirement for Egfr in the AS during DC, we looked at dpp expression in embryos in which Egfr DN had been expressed in the AS using Gal4 c381 , but found no effect (data not shown).
To examine the effects of excessive Egfr signaling on dpp expression we used two transgenes, UAS-sSpi and UAS-Egfr-EGFP. UAS-sSpi encodes a secreted, active version of the Egfr ligand, Spitz, which can directly bind to Egfr to activate the Egfr pathway, whereas UAS-Egfr-EGFP encodes a biologically active Egfr tagged with enhanced green fluorescent protein (EGFP) [90,91]. These transgenes were expressed in stripes in the embryo using ptc-Gal4 and effects on dpp expression assessed by in situ hybridization. For both transgenes, the dpp expression stripes in the dorsal and ventrolateral epidermis became fragmented (Fig. 3G, H). Staining with anti-GFP antibodies revealed that remaining patches of dpp expression were in areas where Egfr-EGFP had not been expressed (Fig. 3H, H). We conclude from our loss-and gain-offunction studies that Egfr signaling negatively regulates dpp expression in the epidermis during DC.

Egfr Negatively Regulates zip Expression in the Epidermis and AS During DC
An important target of Dpp regulation during DC is zip, and we evaluated zip expression in Egfr mutant embryos and embryos with tissue-specific attenuation of Egfr signaling. zip shows two major events of transcriptional upregulation that are relevant to the morphogenetic events during DC: first, a burst of expression occurs in the AS during germband retraction and terminates around the beginning of DC; and, second, upregulation occurs in the DME cells beginning during the germband retraction stage and persists throughout DC [12,24,26] (Fig. 4A). Egfr mutant embryos showed excessive accumulation of zip transcripts in the DME cells, in addition to some ectopic zip transcription in the epidermis (Fig. 4B). We occasionally found less severely disrupted Egfr mutant embryos where the AS was intact; in these zip transcripts persisted in the AS during DC, in contrast to wild-type embryos where the AS was devoid of transcripts by this stage (compare Fig. 4C with Fig. 4A). Reduction of Egfr function in the epidermis through expression of UAS-Egfr DN using the LEand 69B-Gal4 drivers also caused excessive epidermal zip expression (data not shown). We previously demonstrated that the tyrosine kinase Ack, a putative negative regulator of Egfr, controls zip levels in the AS [26]; we, therefore, also tested the effect of disrupting Egfr specifically in this tissue by expressing UAS-Egfr DN using the AS-specific Gal4 c381 driver. In Gal4 c381 .UAS-Egfr DN embryos we observed ectopic zip expression throughout the AS during DC as well as elevated zip levels in the head, the latter indicating some degree of cell non-autonomous control of zip by Egfr (Fig. 4D).
A major downstream effector for Egfr is Ras and expression of a dominant negative Ras transgene, RasN17 [92], in the AS was found to be associated with a similar increase in zip transcript levels (Fig. 4E). Given the robust levels of zip transcripts normally seen in the DME cells, we found it difficult to ascertain if knock down of Egfr in the AS affected zip expression in the DME cells, but we suspect that it caused a modest elevation (Fig. 4D). To determine if excessive Egfr signaling in the AS would have the opposite effect on zip expression, we expressed the Egfr-EGFP transgene with Gal4 c381 (Fig. 4F). zip levels in the DME cells were decreased in association with excessive Egfr signaling in the AS and we conclude that Egfr represses zip transcription in a cell nonautonomous manner during DC.

Egfr Inhibits Apoptosis in the AS
The similarity of Egfr mutant embryonic phenotypes to those of the U-shaped group of genes, which are required for maintenance of the AS, has been noted and studies on fixed preparations suggest premature apoptosis in Egfr mutant embryos [35,93,94]. Apoptosis of the AS cells contributes to the forces driving DC, and negative regulation of cell death in the AS could be an important component of the participation of Egfr in DC [2,3]. A negative regulatory role for the Egfr pathway is further supported by the observation that expression of a constitutively active version of Ras, RasV12, in the AS causes the tissue to persist longer than wild-type [95]. Furthermore, Ras appears to negatively regulate apoptosis throughout the embryo, as revealed by acridine orange staining of embryos with global gains or losses of Ras signaling [96]. As described above, AS cells are lost prematurely in Egfr mutant embryos (Fig. 2D, F) suggesting an early onset of apoptosis. To visualize the effects of losses or gains of Egfr signaling on AS apoptosis in live embryos, we used the caspase sensor Apoliner [97,98]. Apoliner consists of a monomeric red fluorescent protein (RFP) tethered to EGFP by a caspase-sensitive linker [97]. Furthermore, the design of the Apoliner construct includes a transmembrane domain that precedes the RFP component while the EGFP component includes a nuclear localization signal (NLS). As a consequence, the two fluorophores co-localize to membranes in live cells lacking caspase activity, but caspase activation in live cells results in separation of the fluorophores, with Apoliner-RFP remaining at membranes while Apoliner-EGFP is translocated to the nucleus due to its NLS. At the beginning of germband retraction in wild-type embryos there was little nuclear EGFP in the AS, indicating minimal caspase activity   Fig. 5 A-A), but there was strong nuclear EGFP in the AS of Egfr mutant embryos at the same stage (Fig. 5 B-B). As DC proceeded nuclear EGFP accumulated in the AS of wild-type embryos ( Fig. 5C-C) and this accumulation could be blocked by expression of the baculovirus caspase inhibitor p35 [99] (Fig. 5D-D). To promote Egfr signaling in the AS, we expressed either sSpi or RasV12 and found that, in both cases, cells showed little nuclear EGFP even late in DC, similar to what was seen with p35 expression (Fig. 5E-F). We conclude that Egfr signaling inhibits caspase activation in the AS. If Egfr impedes apoptosis in the AS, then excessive Egfr signaling might be expected to affect AS morphogenesis. A robust increase in Egfr levels in the AS through expression of Egfr-EGFP using the double driver combination Gal4 NP3312 + GAL4 NP5328 resulted in a failure of the AS to properly complete morphogenesis (compare Fig. 5G to Fig. 5H, I and Movie S7 to Movies S8 and S9). In addition, the AS of these embryos persisted beyond the normal time of AS programmed cell death.

Evidence that Ack and Endocytosis Negatively Regulate Egfr Levels in the AS
An important route through which Egfr signaling is down regulated is by clathrin-mediated endocytosis (reviewed in [100]). When imaging Egfr-EGFP in the AS for the apoptosis study, we noticed that in addition to localizing cortically in AS cells, much of the protein appeared to be accumulating in vesicles (Fig. 6F). Given the literature demonstrating that Ack family tyrosine kinases promote down regulation of Egfr by endocytosis and subsequent degradation [29][30][31][32], we looked for evidence that AS Ack was controlling zip expression through down regulation of Egfr in this tissue. Over-expression of Ack in the AS during germband retraction causes a dramatic increase in zip levels in this tissue [26](Compare Fig. 6A to Fig. 6B), but co-expression with Egfr-EGFP (but not a control lacZ transgene) restored wild-type zip levels, suggesting that Ack controls zip through down regulation of Egfr (Fig. 6C, D). We subsequently over-expressed Ack in prd stripes in the AS and examined the effect on Egfr distribution by comparison with adjacent amnioserosa cells not over-expressing Ack. AS cells with endogenous levels of Ack showed strong cortical Egfr immunostaining as well as staining in cytoplasmic puncta, some of which were Rab5 positive and therefore early endosomes (Fig. 6E-E). In Ack-over-expressing cells (identified by increased levels of phosphotyrosine [101]) there was a decrease in cortical Egfr staining and an increase in Egfr-positive cytoplasmic puncta, with some of these being Rab5-positive; these cells also showed a general increase in the levels of early endosomes (Fig. 6E-E). Many of the Egfr-positive puncta in these cells were Rab5-negative and we suspect that they may be multivesicular bodies, where endocytosed Egfr is known to accumulate (reviewed in [100]). The increase in early endosomes in Ack-over-expressing cells indicates that Ack over-expression leads to a general increase in endocytosis in AS cells. In support of this is an observation we made when trying to observe the effects of Ack on apoptosis using the Apoliner reporter. Here, with the expression of kinase-dead Ack, which is more effective than wild-type Ack at inducing zip expression [26], a highly punctate distribution of membrane-localized Apoliner-RFP signal was observed while control embryos at the same stage of DC displayed a typical homogeneous distribution (Fig. 6G, H). We interpret this difference as reflecting a general increase in intracellular vesicular traffic, consistent with the effect of Ack overexpression in promoting Egfr endocytosis.

Discussion
We have determined that Egfr is required in both the AS and epidermis for DC to proceed normally and our results suggest that Egfr signaling has a least three distinct roles in DC, all of which act to repress morphogenesis (see model in Fig. 7). Egfr is a negative regulator of dpp expression in the epidermis as loss of Egfr function in either Egfr mutant embryos or as a result of Egfr DN expression leads to ectopic dpp expression. dpp is expressed in two stripes during DC, one composed of the DME cells and the other running along the ventrolateral epidermis, where dpp expression in the DME cells, but not the ventrolateral stripe, is dependent on a JNK MAPK cascade [13][14][15][16][17][18]. Consistent with the notion that Egfr functions as a negative regulator of dpp expression, activation of the Egfr pathway can repress dpp expression in either stripe. The down regulation of dpp expression in both stripes, however, supports the view that Egfr does not reduce dpp transcription by impacting the JNK pathway, in which case we would expect to observe down regulation of dpp expression only in the DME stripe. How might Egfr signaling be regulating dpp expression? Wingless (Wg) is a diffusible signal required for proper dpp expression in both the dorsal and ventrolateral stripes during DC [102,103]. Egfr negatively regulates Wg levels in the eye imaginal disc by transcriptionally regulating phyllopod [104]; we looked for evidence that Egfr might be controlling dpp expression through Wg, but altering Egfr signaling had no discernable effect on phyllopod transcription or Wg distribution in the embryo (X. C., unpublished results). The defects in morphogenesis seen in embryos with impaired Egfr signaling are likely at least in part due to misregulated actomyosin contractility. A recurring theme associated with various circumstances of Egfr impairment is the bowed embryo phenotype, where segments are bunched together at the leading edge of the epidermis during DC. We suggest this is due to uneven actomyosin contractility in the dorsal epidermis associated with excessive zip expression. The loss of epithelial integrity or ''pulling apart'' of the head seen in live imaging of Egfr mutant embryos may similarly be due to misregulated actomyosin contractility and/or excessive apoptosis. Previous work has indicated that a major function for Egfr in the head is inhibition of apoptosis, similar to its role in the AS [74,105,106]. Parallels between the regulation of head involution and DC have been noted and Egfr may function to modulate these two morphogenetic events through similar mechanisms [107].
While it is likely that increased levels of Dpp in Egfr mutant embryos contribute to the elevated zip levels, our results of manipulating Egfr signaling support the interpretation of a separate route for zip regulation that involves signaling from the AS to both the AS and the epidermis. This signaling is not operating through the regulation of dpp expression as impairment of Egfr signaling in the AS does not affect Dpp levels. Thus, we consider this zip regulation a second distinct role for Egfr in DC and we believe this signaling is the same as that regulated by Ack in its control of zip expression. Consistent with this, gains or losses of Ack do not affect the Dpp pathway, supporting the view that Ack operates in parallel to Dpp signaling [26,101]. The Ack/Egfr-regulated signal could be a diffusible ligand (''X'' in Fig. 7) produced in the AS cells that activates a pathway in the AS and DME cells, thereby driving zip expression [26]. Alternatively, Egfr could promote the production of a signal that negatively regulates the pathway required for zip expression. Moreover, the upregulation of zip expression in Egfr signaling deficient embryos does not appear to be due to premature apoptosis of the AS as elevated zip can be seen in Egfr signaling deficient embryos that still have an intact AS. It is possible, however, that the signaling events regulating zip expression occur at a stage prior to the stage at which the AS is lost in Egfr mutants, i.e. before initiation of germband retraction. Preliminary data leads us to propose that Dpp acts in parallel to produce a second diffusible ligand (''Y'' in Fig. 7) that activates a second pathway contributing to zip expression (W. S and X. C., unpublished observations). Thus, the interplay between Egfr and Dpp during DC is complex, involving multiple pathways and bidirectional communication between two tissues, and this complex signaling arrangement may function to ensure the coordinated morphogenesis of the AS and epidermis.
A third major role for Egfr in DC is as a negative regulator of apoptosis in the AS. Enhancement of apoptosis accelerates DC whereas suppression of apoptosis slows it, indicating that apoptosis, similar to actomyosin contractility, provides a force for morphogenesis [3]. Thus, down regulation of Egfr in the AS during DC provides two means to accelerate the process: increased myosin expression and increased cell death. The ''tweaking'' of Egfr function in the AS could constitute an important regulatory mechanism for controlling the rate of closure. We have provided evidence that endocytosis, promoted by Ack, is a route by which Egfr signaling is controlled in the AS cells. Our results suggest that Ack would have a pro-apoptotic role in the AS through promotion of Egfr endocytosis. This in contrast to the Drosophila eye in which Ack has an anti-apoptotic function that is independent of Egfr [108].
A recent study has demonstrated that endocytosis in the AS is required for its correct morphogenesis during DC, but this work focused on the role of endocytosis in removing membrane to promote cell shape change [109]. Our results indicate that another route of action for endocytosis in the AS is in regulation of Egfr signaling. It has been suggested that endocytosis could act as a rheostat in which membrane area is adjusted in response to actomyosin contractility [109]; such a rheostat could also be used to adjust Egfr signaling throughout DC.
Additional avenues for Egfr regulation during DC could be control of ligands binding to Egfr and feedback inhibition [110,111], but we have yet to address these. In summary, we have identified Egfr signaling as an inhibitor of morphogenesis during DC that acts at several distinct levels. Having a single pathway control multiple aspects of this complex process may simplify feedback regulation, ensuring that morphogenesis occurs in a coordinated fashion. In essence, Egfr signaling acts as a brake that can be applied when required to ensure that closure proceeds smoothly and without loss of epidermal integrity. DC shows striking parallels to the healing of induced wounds in the Drosophila embryo, with the two processes using similar cytoskeletal and signaling machineries [112][113][114]. Egfr has recently been shown to be required for healing of induced wounds in the embryo and it will be of interest to determine if it uses similar routes of action in this as we have shown in DC [115].

Cuticle Preparations
Cuticles were prepared as described but with the fixation step removed [116]. At least 100 embryos were examined in each experiment.

Immunohistochemistry and RNA in Situ Hybridization
Fixing and antibody staining of embryos were done as previously described [117,118]. The following primary antibodies were used: mouse anti-phosphotyrosine (Cell Signaling)(1:1000), Figure 7. Model for Egfr acting as a brake on DC. Egfr negatively regulates the production and/or secretion of a diffusible signal ''X'' in the AS (AS) and is itself negatively regulated by Ack through endocytosis. ''X'' signals into both the AS and the DME cells where it activates a pathway promoting transcription of myosin from the zip locus. Previous work from our group and others, and unpublished results from our group, suggest that Dpp from the DME cells diffuses to the AS where it regulates production of a second diffusible signal ''Y'' providing a parallel input into zip transcription. Myosin produced through the cooperation of the two pathways then drives morphogenesis of the AS and DME cells. Egfr additionally regulates this signaling network by negatively regulating dpp transcription in the epidermis, including the DME cells. Egfr further regulates AS morphogenesis by inhibiting apoptosis in this tissue. doi:10.1371/journal.pone.0060180.g007 rabbit anti-GFP (1:500)(Sigma), mouse anti-GFP (1:500)(Sigma), goat anti-Egfr (Santa Cruz)(1:5), rabbit anti-phospho-MAPK (Cell Signaling)(1:50) and rabbit anti-Rab5 (Abcam)(1:1000). All secondary antibodies were from Vector Laboratories and used at a 1:200 dilution. In situ mRNA hybridizations using digoxigeninlabeled RNA probes and FISH were performed as described [119,120]. cDNAs for in situ hybridization probes were obtained from the Canadian Drosophila Microarray Centre. Fluorescentlystained embryos were examined on either a Zeiss LSM 410 laserscanning confocal microscope or a Quorum spinning disk confocal microscope, and digoxigenin-labeled embryos imaged using a Zeiss Axioplan 2 microscope. Images were processed in Adobe Photoshop. The genotypes of all fluorescently-stained embryos were established by tracking balancer chromosomes bearing GFP reporters.

Live Imaging of Embryos
Chromosomes carrying Ubi-DEcadherin-GFP in combination with the alleles Egfr f2 , Egfr 1a15 , and Egfr H25 were recovered by meiotic recombination. Similarly, the tub-Apoliner insertion, which expresses Apoliner under the control of the tubulin 1 a promoter (described in [97]), was recombined with Egfr f2 . All recombinant chromosomes were maintained over so-called ''GFP-balancer'' chromosomes obtained from the Bloomington Drosophila Stock Center (either CyO, P{w[+mC] = GAL4-Kr.C}DC3, P{w[+mC] = UAS-GFP.S65T}DC7 or CyO, P{w[+mC] = GAL4twi.G}2.2, P{UAS-2xEGFP}AH2.2). For live imaging experiments Egfr mutant embryos were unambiguously identified as those lacking GFP expression derived from the GFP-balancer. Since the onset of the Egfr mutant phenotype preceded the timing of GFP expression associated with either GFP-balancer stock, the selection of embryos for live imaging was random and identification of mutant embryos was achieved subsequent to image acquisition. The early onset of the Egfr mutant phenotype also preceded the time at which all available amnioserosa specific GAL4 drivers could induce reporter gene expression, and for this reason our analysis of caspase activity in Egfr mutant embryos required using tub-Apoliner carried by the maternal parent.
Embryos were prepared for live imaging using the hanging drop protocol, which eliminates effects of compression on the mounted embryo [121]. Time-lapse confocal microscopy was performed using a 20X Plan Apo VC objective on a Nikon Eclipse 90 i microscope with a Nikon D-eclipse C1 scan head. Images were saved as animated projections using the Nikon EZ-C1 3.70 software and further processed using ImageJ (NIH).

Supporting Information
Movie S1 Time-lapse showing restricted expression pattern of LE-Gal4 driver, revealed using a UAS-GFP-NLS reporter. Cell outlines were revealed through expression of a Ubi-DEcadherin-GFP transgene.