Novel interplay between JNK and Egfr signaling in Drosophila dorsal closure

Dorsal closure (DC) is a developmental process in which two contralateral epithelial sheets migrate to seal a large hole in the dorsal ectoderm of the Drosophila embryo. Two signaling pathways act sequentially to orchestrate this dynamic morphogenetic process. First, c-Jun N-terminal kinase (JNK) signaling activity in the dorsal-most leading edge (LE) cells of the epidermis induces expression of decapentaplegic (dpp). Second, Dpp, a secreted TGF-β homolog, triggers cell shape changes in the adjacent, ventrally located lateral epidermis, that guide the morphogenetic movements and cell migration mandatory for DC. Here we uncover a cell non-autonomous requirement for the Epidermal growth factor receptor (Egfr) pathway in the lateral epidermis for sustained dpp expression in the LE. Specifically, we demonstrate that Egfr pathway activity in the lateral epidermis prevents expression of the gene scarface (scaf), encoding a secreted antagonist of JNK signaling. In embryos with compromised Egfr signaling, upregulated Scaf causes reduction of JNK activity in LE cells, thereby impeding completion of DC. Our results identify a new developmental role for Egfr signaling in regulating epithelial plasticity via crosstalk with the JNK pathway.


Author summary
The developmental process of dorsal closure (DC) in Drosophila embryogenesis has provided many fundamental insights into conserved mechanisms controlling epithelial dynamics and cell migration, and therefore serves as a model for studying tissue morphogenesis and wound repair. DC is known to require the coordinated action of two signaling pathways: (i) the c-Jun N-terminal kinase (JNK) pathway, which is activated in the dorsalmost leading edge (LE) cells of the migrating epithelia; and (ii) the Decapentaplegic (Dpp) pathway, induced by the JNK pathway, which signals to neighboring lateral epidermis cells to drive the cellular changes required for DC. Here we uncover a new tier of regulation essential for DC, mediated by the Epidermal growth factor receptor (Egfr) pathway. Specifically, we demonstrate that Egfr signalling in the lateral epidermis suppresses the expression of the gene scarface, which encodes a secreted JNK antagonist.

Introduction
Epithelial sheet fusion and collective cell migration are key processes in normal development, wound healing and pathogenesis [1,2]. One system that has offered fundamental insights into the mechanisms controlling epithelial dynamics and cell migration is the embryonic process of dorsal closure (DC) in Drosophila melanogaster. In this developmental setting, two contralateral epithelial sheets from opposing sides of the embryo migrate and converge at the dorsal midline above the amnioserosa (AS), an extraembryonic epithelium tissue, thereby generating a continuous epidermis that seals a large dorsal hole ( Fig 1A) [3]. Two cell signaling pathways, which act sequentially, drive and coordinate this process. Initially, c-Jun N-terminal kinase (JNK) pathway activity in the dorsal-most leading edge (LE) cells induces expression of the gene decapentaplegic (dpp), encoding the TGF-β/BMP family member Dpp [4][5][6]. Secreted Dpp subsequently triggers signal transduction in adjacent, ventrally located lateral epidermis cells, leading to the cell shape changes that are at the basis of the morphogenetic movements of the migrating epithelia [7,8]. Accordingly, embryos mutant for various components of the JNK signaling cascade or for constituents of the Dpp pathway fail to complete DC morphogenesis and consequently display dorsal-open phenotypes [9][10][11]. Thus, DC is an excellent experimental model system with which to identify and characterize the signaling events regulating complex movements of epithelial layers. Spatiotemporal refinement of the expression of dpp and other JNK pathway target genes requires input by two negative feedback regulators, which are expressed in response to JNK signaling in LE cells [9]. One is Puckered (Puc), a dual specificity phosphatase that acts as an intracellular inhibitor of pathway activity by dephosphorylating JNK [12]. The other proposed JNK pathway antagonist is Scarface (Scaf), a secreted serine protease homologue that possibly acts by modifying the receptor mediating JNK signaling or an unknown extracellular signal [13,14]. Input from these two feedback inhibitors restricts JNK pathway activity in LE cells. It is not known, however, whether the expression of these negative regulators, or of other JNK pathway target genes in LE cells, is controlled only by JNK signaling, or whether other signaling pathways originating from adjacent epithelial cells might also contribute to this regulation.
Herein, we demonstrate an activity mediated by the Epidermal growth factor receptor (Egfr) pathway in lateral epidermis cells, that is pivotal for JNK function in the adjacent LE cells. Specifically, we identify a positive, cell non-autonomous input by the Egfr pathway, upstream of JNK signaling, into JNK pathway activity. We find that the mechanism underlying this effect involves the repression of scaf expression in lateral epidermis cells. Correspondingly, derepression of scaf in embryos defective in Egfr signaling causes a reduction in JNK activity in nearby LE cells. This leads to impaired expression of the JNK target gene dpp, reduced levels of Dpp effector responses, failure of cell elongation and, consequently, aborted DC resulting in a dorsal-open phenotype. Our results thus identify a novel intercellular crosstalk between the Egfr and JNK signaling pathways, shedding new light on DC regulation and potentially on other related processes involving synchronized cell movements and epithelial fusions [15].

Results
Egfr pathway activity is detectable in the lateral epidermis during dorsal closure Intercellular signaling mediated by Receptor tyrosine kinases (RTKs) is essential for multiple patterning events during Drosophila oogenesis, embryogenesis and adult development [16- 19]. We therefore reasoned that RTK-dependent signaling might also play unknown roles during the process of DC. Hence, we immunostained embryos for active mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/Erk) as readout for pathway activity [20][21][22]. Doubly phosphorylated MAPK/Erk (dpErk) was detectable at the time of DC initiation in the lateral epidermis, a region where RTK activity has not been explored before (Fig 1A  and 1B).
Several lines of evidence indicate that MAPK/Erk is activated in this region specifically in response to Egfr signaling. First, dpErk staining borders on the striped domains of rhomboid expression (determined using a rhomboid-lacZ enhancer trap line; Fig 1B-1D); this gene encodes the serine protease that is both necessary and sufficient to trigger Egfr pathway activity, and whose expression pattern forecasts Egfr pathway activation [22]. Second, activation of MAPK/Erk is significantly reduced in embryos in which pannier (pnr)-Gal4 [23] drives the expression of a dominant-negative form of Egfr (Egfr DN ), thereby blocking Egfr function selectively in the lateral epidermis and the LE cells ( Fig 1E and S1 Fig). Third, a reduction in dpErk staining is also evident in rhomboid mutants, as well as in embryos mutant for spitz (spi), which encodes a member of the TGF-α family of Egfr ligands [24], and for Egfr (Fig 1F-1H), though not in mutants for pvr, the gene encoding Drosophila Platelet-derived growth factor/ Vascular endothelial growth factor receptor (PVR) (Fig 1I). Taken together, these results show that signaling via the Egfr pathway takes place in the lateral epidermis during the time of DC.

Egfr pathway activity is required for proper dorsal closure
To establish the functionality of Egfr signaling in the lateral epidermis, we examined the effects caused by the loss of Egfr-mediated signal transduction on different aspects of the closure process. Analyses of cuticle preparations revealed that embryos expressing dominant-negative forms of Egfr or Ras (Ras DN ) via the ectodermal pnr-Gal4 driver (Fig 2C and 2D, respectively), as well as embryos mutant for rhomboid, spi or various alleles of Egfr (Fig 2E-2H), fail to complete DC and exhibit dorsal-open phenotypes (white arrows; cf. wild-type in Fig 2A), characteristic of mutants in genes encoding components of the JNK pathway [e.g., basket (bsk), encoding Drosophila JNK; Fig 2B] [7,25].
To quantify the proportion of embryos that fail to complete closure due to defective Egfr signaling, we scored the percentage of Ras DN -and Egfr DN -expressing embryos that completed DC by stage 16 (st16; 13-16 hours after egg lay), when the process normally ends (S2B Fig). In order to unambiguously segregate incomplete DC phenotypes from other secondary effects seen in Egfr-deficient embryos, such as head involution defects, we demarcated the LE cells using the puc-lacZ enhancer trap line, which concurrently also enabled the unequivocal identification of the embryonic genotypes scored (S2A and S2B Fig) (see Materials and Methods). We find that by st16, 92% of control embryos completed the course of DC, whereas only 47% and 39% of embryos expressing Ras DN or Egfr DN finalized closure, respectively, confirming that Egfr signaling plays a significant role in DC regulation (S2C Fig).
Consistent with the cuticular defects, the consequences of loss of Egfr-mediated signal transduction are also apparent at the cellular level. Normally, LE cells of the advancing epidermis elongate along the dorsoventral (D/V) axis upon DC initiation. Subsequently, a similar elongation of lateral epidermis cells is observed in more ventral locations, as reflected by DE-Cadherin immunostaining that outlines cell membranes (Fig 3A and 3A') [26,27]. In contrast, mutations in various constituents of the Egfr pathway, as well as ectodermal Egfr DN or Ras DN expression, prevent epithelial cell elongation to a large extent ( Fig 3C-3F', arrowheads; quantification in S3 Fig). Instead, many cells remain polygonal in shape, as do analogous cells in bsk mutant embryos ( Fig 3B and 3B'; arrowheads) [28,29].
Collectively, our findings identify a requirement for functional Egfr signaling in the process of DC, which is already apparent at the level of the cell shape changes that normally occur during early DC.

Egfr signaling is required for full induction and activity of Dpp signaling
Dpp signaling is known to coordinate the morphogenetic movements during DC [8]. Considering the impaired cell shape changes observed in embryos defective in Egfr signaling (Fig 3), we next assessed the expression of the JNK pathway target dpp in LE cells of rhomboid mutants, as well as in embryos expressing Egfr DN in the ectoderm, finding that it is reduced in both genotypes (black arrowheads in Fig 4D and 4E, respectively; cf. wild-type dpp expression in Fig 4A; S4 Fig) [6,7]. Noteworthy, in both genetic backgrounds JNK-independent dpp expression is unaffected, for example in the visceral mesoderm and lateral ectoderm (white asterisks and arrowheads, respectively, in Fig 4A, 4D and 4E) [5,25]. Similar outcomes were observed when Egfr DN was expressed in stripes, using paired (prd)-Gal4 (S5 Fig). Reciprocally, the dpp In Dpp-responding cells, activation of the Dpp receptor complex brings about the phosphorylation of the SMAD family member Mothers against dpp (Mad) [30]. We find that levels of phosphorylated Mad (pMad) fully mirror the changes in dpp expression levels in the different genetic backgrounds (Fig 4B''-4F''; cf. wild-type in Fig 4A''). To determine which cells are most affected by the loss of Egfr signaling, we utilized the puc-lacZ enhancer trap line to delineate LE cells. We find that pMad is detectable in LacZ-expressing LE cells in embryos expressing Egfr DN and Ras DN under pnr-Gal4 regulation, consistent with residual dpp expression in these cells ( Cumulative effects due to dysfunction of the Egfr signaling pathway at earlier stages of development could contribute to the dorsal open phenotypes observed in embryos defective in Egfr signaling. To specifically assess the impact of this pathway on dorsal closure, independently from earlier contributions, we used a temperature sensitive Egfr allele (Egfr SH2 ) [31]. We conclude that, Egfr signaling is required for the full induction of dpp and for the phosphorylation of the downstream Dpp pathway effector molecule, Mad. The JNK pathway is epistatic to Egfr signaling Given the phenotypic similarities between embryos defective in Egfr and JNK signaling, we next set out to determine the epistatic relationship between the two pathways using pMad staining. First, we assessed the ability of constitutive JNK pathway activation to suppress loss of Egfr signaling in the double combination of Hep Act and Ras DN . When individually expressed, pMad staining is reduced in pnr>Ras DN embryos compared to controls (Fig 5A; cf. Fig 4A''), and broader in those expressing pnr>Hep Act alone ( Fig 5B). Strikingly, the Ras DNmediated reduction in pMad staining is fully suppressed by Hep Act co-expression, with the pattern indistinguishable from that observed in embryos singly expressing Hep Act (Fig 5C; cf. Fig  5B). This result indicates that the JNK pathway acts downstream of the Egfr pathway.
To further examine this issue we also undertook a reciprocal approach, by testing whether loss of JNK signaling suppresses constitutive activation of the Egfr pathway. Specifically, the pMad pattern expands in embryos expressing Ras V12 alone (Fig 5D; cf. Fig 4A'') and narrows in embryos expressing a dominant-negative form of Bsk (Bsk DN ) by itself ( Fig 5E). Markedly, pMad staining in combined pnr>Bsk DN ; Ras V12 embryos resembles that seen in embryos expressing Bsk DN alone (Fig 5F; cf. Fig 5E; S8 Fig). Taken together, these results demonstrate that the JNK cascade is epistatic to the Egfr pathway. Consistent with this conclusion, we find that MAPK/Erk activation in response to Egfr signaling is unaffected in bsk mutants as well as in embryos expressing Hep Act (S9 Fig). Thus, our data identify a positive input by the Egfr pathway, acting upstream of JNK signaling, into the expression of a central JNK pathway target gene, dpp, and consequently into the phosphorylation of the key Dpp pathway effector, Mad.
EGFR signaling is required for repression of scarface, a secreted inhibitor of the JNK pathway, in the lateral epidermis The rhomboid and dpErk patterns (Fig 1D) indicate that the Egfr pathway is active throughout the dorsal ectoderm. To better understand the hierarchal link between the Egfr and JNK pathways, we co-stained embryos expressing puc-lacZ, a target of JNK signaling in the LE cells [12], for dpErk and LacZ. Surprisingly, dpErk staining does not overlap with the puc-LacZ pattern, and is evident only in the adjacent lateral epidermis (Fig 6A-6B'). The finding that dpErk is excluded from the LE demonstrates that the JNK and Egfr pathways are active in distinct cell types within the dorsal ectoderm, and therefore implies that the Egfr pathway influences JNK signaling cell non-autonomously.
We hypothesized that the indirect input by Egfr signaling in lateral epidermis cells into JNK activity in neighboring LE cells occurs either via induction of a secreted JNK agonist or by the suppression of a secreted JNK antagonist. Among known JNK pathway elements, the secreted JNK antagonist Scarface (Scaf) appeared to be an attractive candidate to perform this intermediary role. scaf is normally expressed in LE cells under JNK regulation (Fig 6C) [13,14]. However, reduced Egfr activity, in rhomboid and spi mutants as well as in embryos expressing Egfr DN or Ras DN , results in ventral expansion of scaf expression in a striped configuration ( Fig  6D-6G; cf. Fig 6C). By contrast, embryos expressing Ras V12 under pnr-Gal4 regulation show a notable decrease in scaf expression even in LE cells (Fig 6H).
These data strongly suggest that Egfr signaling normally suppresses scaf expression, and provide a plausible explanation for the positive, non-autonomous effect of the Egfr pathway on JNK signaling in the LE. According to this scenario, in wild-type embryos Egfr signaling inhibits scaf expression in lateral epidermis cells, thereby blocking its secretion and antagonistic effect on JNK signaling in neighboring LE cells. Under Egfr loss-of-function conditions, scaf is up-regulated in lateral epidermis cells, leading to reduced JNK activity in LE cells. Accordingly, scaf expression in the LE itself is reduced relative to normal embryos (Fig 6D-6G'; cf. Fig 6C), and thus responds to the loss of Egfr signaling like other JNK pathway targets, such as dpp.
To establish more directly how scaf deregulation influences JNK pathway output, we assessed dpp transcription and Mad phosphorylation in embryos over-expressing scaf or in scaf mutants. We find that scaf over-expression in the ectoderm obstructs dpp expression ( Fig  7A; cf. wild-type in Fig 4A) and reduces Dpp pathway activity as reflected by immunostaining for pMad (Fig 7C; cf. wild-type in Fig 4A''), in a manner resembling Egfr loss-of-function backgrounds in which scaf is derepressed (Fig 4D-4E''). Conversely, both dpp expression and the pMad domain expand in scaf mutants (Fig 7B and 7D). These observations support a role for Scaf as a suppressor of Dpp signaling in the dorsal ectoderm.
The scaf over-expression and loss-of-function phenotypes closely parallel those observed for Egfr loss-of-function (Fig 4D-4E'') and RTK constitutive activation (Fig 4F and 4F''), respectively. To establish a regulatory link between Egfr signaling and scaf, we stained embryos, singly or doubly mutant for rhomboid and scaf, for pMad. The pMad domain broadens in scaf mutants ( Fig 7D) and narrows considerably in rhomboid mutant embryos (Fig 7E). Importantly, embryos doubly mutant for rhomboid and scaf show an expanded pMad pattern (Fig 7F), similarly to embryos mutant for scaf alone (Fig 7D). These results indicate that Egfr signaling acts as an upstream negative regulator of scaf in the lateral epidermis cells. Thus, in the absence of scaf, JNK activity and consequent Dpp pathway activity is robust whether or not the Egfr pathway is functional. Collectively, our data indicate that Egfr-mediated suppression of scaf in lateral epidermis cells is required for full JNK signaling activity in LE cells, accounting for how Egfr activity affects JNK signaling in nearby cells and explaining why the Egfr pathway is required for the completion of the DC process (Fig 7G and 7H).
We next addressed the mechanism by which Egfr signaling impacts on scaf expression. scaf could represent an exceptional example of a gene that is directly repressed by the Egfr pathway. Still, previous studies indicated that, at least in flies, RTK signaling pathways predominantly activate gene expression, often by downregulating negative transcriptional regulators such as the ETS transcription factor Yan (also known as Anterior open) [32]. Therefore, Egfr signaling could be affecting scaf expression indirectly by inducing an intermediary repressor of this gene. To begin testing this idea, we expressed a non-phosphorylatable derivative of Yan, Yan Act , that is insensitive to attenuation by Egfr-mediated signaling [33] and, hence, should dominantly block the induction of such a repressor. We find that pnr>Yan Act embryos display dorsal open phenotypes (Fig 8A and 8A'; cf. Fig 2A), loss of dpp expression and reduced pMad staining (Fig 8B and 8C; cf. Fig 4A-4A"). Remarkably, scaf is derepressed in these embryos (Fig 8D; cf. Fig 6C), consistent with the idea that Yan normally represses a scaf repressor.
S. Noselli and colleagues recently demonstrated that scaf is under complex transcriptional control, involving several activators as well as at least two repressors: Engrailed (En) and Abdominal-A (Abd-A) [34]. They showed that in embryos mutant for en, scaf is derepressed only in LE cells, perhaps due to the combined repressor activity of Abd-A. Nevertheless, when they expressed a form of En that was converted into an activator, scaf was ectopically expressed  in the lateral epidermis [34]. This raised the question whether the Egfr-Yan axis induces En, which in turn directly represses scaf. Indeed, we find that En is dominantly repressed in the in the lateral epidermis to prevent expression of the JNK antagonist, scaf, thus supporting maximal JNK activity in LE cells. (H) When Egfr signaling is defective, deregulated Scaf subsequently attenuates functional JNK signaling in LE cells, thus hindering the process of DC. Bold text and arrows/bars indicate normal levels of gene expression and regulation, whereas gray fonts designate abnormally lower levels of expression and regulation, respectively. https://doi.org/10.1371/journal.pgen.1006860.g007 lateral epidermis of embryos expressing pnr>Yan ACT , specifically in the pnr domain (Fig 8F; cf. 8E). Thus, a transcriptional regulatory cascade could explain how the Egfr pathway represses scaf (Fig 8G): by downregulating Yan, Egfr signaling induces expression of En (and perhaps other repressors) which, in turn, silences scaf expression. In embryos in which Egfr pathway activity is blocked, En is not induced and scaf is derepressed in the lateral epidermis, thereby hindering JNK signaling in LE cells.

Discussion
Although extensively investigated, it is not fully understood how complex morphogenetic processes such as DC are controlled, and by which signaling pathways. In this manuscript we report that, in addition to the well-established JNK and Dpp pathways, signaling mediated by the Egfr is also instrumental to DC. We uncover a novel interplay between the Egfr and JNK pathways, specifically by demonstrating that Egfr signaling in the lateral epidermis suppresses the expression of the gene scaf, which encodes a proposed secreted JNK antagonist. Through this regulatory switch the Egfr pathway facilitates JNK signaling in LE cells (Fig 7G and 7H). Egfr signaling thus contributes cell non-autonomously to the expression of the JNK target dpp, to the phosphorylation of the downstream Dpp effector Mad and, consequently, to the synchronized morphogenetic movements orchestrated by Dpp signaling that are essential to successful DC.
It is currently unknown whether Scaf, the protein product of an established JNK pathway target gene [14,[34][35][36], impinges on JNK pathway activity directly or indirectly. Scaf could be negatively regulating JNK signaling directly, by acting on an extracellular signal or on a putative receptor of the pathway [13]. However, it could also be playing a more general role, for example by degrading the extracellular matrix [35] or by establishing correct basement membrane protein localization [14], thus influencing JNK pathway outcomes indirectly. Although our study does not distinguish between these mechanisms, our results showing that Scaf impacts on the expression and activity of Dpp are consistent with the notion that Scaf is an antagonist of the JNK pathway. Rousset et al. reached similar conclusions, based on their findings that scaf loss-of-function mimics JNK over-activity as well as on other data [13]. Further research, however, will be required to conclusively elucidate the molecular mechanism(s) underlying Scaf function in the context of dorsal closure.
Our findings that the Egfr and JNK pathways are linked at the level of a JNK feedback inhibitor exemplify an important emerging theme in cell signaling: that Egfr signaling frequently impacts on the activity of other developmental pathways or master regulators via the induction of genes, whose protein products subsequently modulate the activity of these secondary pathways and/or factors. For instance, in the fly eye imaginal disc, Egfr signaling induces expression of the Delta ligand in photoreceptor cells, and thus positively stimulates Notch signaling in neighbouring cone cells [37,38]. In other cases, Egfr-regulated targets act as negative feedback regulators. For example, the Egfr pathway induces expression of the gene wntD, which encodes an antagonist of the Rel transcription factor, Dorsal. Through this negative feedback regulation, Egfr signaling opposes the nuclear localization of Dorsal, thereby affecting the expression of multiple Dorsal targets along the D/V axis of the embryo [39].
In our analyses, we have focused on the input to DC by Egfr signaling taking place in the lateral epidermis. It is conceivable that Egfr-mediated signal transduction also plays additional regulatory roles during DC. For example, this pathway has been previously implicated in the suppression of zipper, the gene encoding Drosophila non-muscle myosin II heavy chain, in the AS and in a cell non-autonomous manner also in LE cells [40]. Furthermore, our results do not preclude the involvement of additional RTK pathways in this developmental process. As a case in point, signaling by PVR in DC supports proper midline zippering in addition to AS internalization and removal, via the PI3K pathway and independently of JNK signaling [41]. Future studies will uncover the full impact of RTK-mediated signal transduction in DC.
In conclusion, our work illuminates a novel mechanism of signal integration between the Egfr and JNK pathways, linking Egfr signaling to the core regulatory network controlling DC. Our results thus reinforce the idea that different signaling pathways that regulate morphogenesis are interlinked, acting in a coordinated manner. A deeper understanding of the cross-regulation between these pathways, and the elucidation of further roles for Egfr signaling in DC, should facilitate our understanding of how diverse signal transduction pathways intersect to synchronize collective cell behavior, and how this circuitry ultimately leads to precise and coordinated morphogenetic processes.

In situ hybridization and antibody staining
Embryos were dechorionated in bleach and fixed in 8% formaldehyde/PBS/heptane for 20 minutes. Expression patterns of dpp and scaf were visualized by whole-mount in situ hybridization using digoxigenin-UTP labeled antisense RNA probes and anti-digoxigenin antibodies conjugated to alkaline phosphatase (Roche).

Microscopy
Light microscope images were acquired using a Zeiss Axioplan2 microscope and confocal images were taken using a Zeiss LSM710 confocal microscope. Images were processed using Adobe Photoshop software, and the ZEN 2012 blue edition was used to measure LE cell length in embryos compromised in Egfr signaling.