JNK Signaling in DC
DC is driven by dynamic signal-dependent changes in cell shape, mobility, and adhesion. To investigate whether some or all of these functions might be directed by JNK signaling, we used time-lapse confocal microscopy to observe the DC process in embryos that carry loss-of-function mutations in different components of the JNK signaling pathway, including null alleles or hypomorphic mutations in genes encoding Jun, JNK, and JNKKK (Fig. 1, and Supplementary Videos 1–5, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). Changes in cellular morphology during dorsal closure were visualized in embryos expressing a moesin–green fluorescent protein (GFP) fusion protein that labels the cortical F-actin cytoskeleton (sGMCA transgene, Kiehart et al., 2000).
Figure 1. A–E: A sampling of frames from videos depicting embryogenesis at the dorsal closure (DC) stage of wild-type (genotypically a paternal rescue of a jun germline clone; A), jun (germline clone, glc; B), slpr (C), bsk (D), and scb (E) embryos. Each sampling represents roughly 3 hr from each video except scb, which shows 5 hr. The asterisks in A denote each canthus progressing in DC. The white arrows show the lack of zipping in C,D,E. Each frame is artificially color coded as follows: green, epithelium; blue, amnioserosa; red, internal structures. The uncolored raw data are available online in video format (see Supplementary Videos 1–5). For comparing signal intensities between structures and for identifying structures with cellular resolution, refer to the online videos. Note that the slpr video has a high green fluorescent protein (GFP) background fluorescence due to maternal contribution of the GFP balancer used in the parent stock and because the embryo contains only one copy of the sGMCA transgene. Scale bar = 10 μm.
Download figure to PowerPoint
In wild-type embryos, the actin cable is assembled apically in LE cells before the onset of epithelial stretching. As the cable matures, it begins to contract, thereby transforming the LE front from a scalloped to a straightened appearance (Fig. 1A, Supplementary Video 1, and Kiehart et al., 2000). Once cell stretching initiates (due to contributions of both actin cable and amnioserosal cell contraction; Kiehart et al., 2000; Hutson et al., 2003), the contralateral cell fronts are gradually “zipped” together at the two “canthi” located at the anterior and posterior ends of the dorsal hole (Fig. 1A, asterisks). Zipping results in fusion of the two epithelial cell sheets and exerts a force that contributes to hole closure (Hutson et al., 2003).
Videos of embryos that are homozygous for null alleles of jun or slpr and, thus, deficient for the Jun transcription factor or the relevant upstream JNKK kinase, respectively, show severe defects in DC. The different mutant embryos, as well as embryos lacking both the zygotic and the maternal Jun component (jun germ line clones, GLC, Fig. 1B), display common phenotypic defects (Fig. 1B,C, and Supplementary Videos 2, 3; note that the background fluorescence is higher in the slpr movie than in the others, see legend for Fig. 1 for details). Consistent with previous reports describing JNK signaling-deficient embryos (Ricos et al., 1999; Jasper et al., 2001; Kaltschmidt et al., 2002), the actin cable fails to mature into a pronounced structure in jun mutants (Supplementary Fig. S1). The LE front remains scalloped at a stage of DC when it has become prominent and taut in the wild-type embryos (compare Supplementary Video 1 with Videos 2, 3, and see Fig. S1). Thus, based on previous reports and our data, it is reasonable to hypothesize that JNK signaling is required for efficient formation of the leading edge actin cable. To lend further support to this hypothesis, we asked whether a stronger actin cable can be observed as a result of increased JNK signaling. As such an analysis of actin cable strength in a JNK gain-of-function genotype has not performed to date, we made time-lapse videos of embryos in which the activity of the JNK pathway is elevated. These experiments were performed on puckered mutants, which provide a well-established and thoroughly characterized JNK gain-of-function genotype (Martin-Blanco et al., 1998; Zeitlinger and Bohmann, 1999; Wang et al., 2003). These embryos lack the Drosophila JNK phosphatase and, therefore, have elevated levels of JNK activity. As shown by the analysis of puc mutant movies, increased JNK signaling results in an apparently more contractile cable (Fig. 2A, Supplementary Video 6), with epidermal bunching toward nodes at the LE. To quantify this effect, we measured the speed with which the actin cable retracts in puc mutant and wild-type embryos after being severed with a laser-directed cut (Fig. 2B, Supplementary Video 7). Quantitative analysis of movies showing the retraction demonstrates that increased JNK signaling correlates with higher tension and contractility of the actin cable during DC (Fig. 2C,D). Taken together, these data confirm that signaling strength correlates with cable thickness and cable tension. The combination of gain- and loss-of-function data strongly suggests that one of the contributions of JNK signaling to the DC process is the assembly of the actin cable in the LE. This developmental effect might involve the transcriptional regulation of genes such as profilin, which we have identified previously as JNK responsive (Jasper et al., 2001). It is likely that JNK signaling has other functions as well, which would account for the pleiotropic phenotype of the respective mutant embryos.
Figure 2. Increased JNK signaling augments actin cable tension. A: Successive frames (indicated by roman numerals) from a dorsal closure (DC) video of a puc mutant show contractile nodes in the cable (arrowheads in A). B,C: A 5-μm cut at the leading edge (LE; marked by the arrowhead) is made with a ultraviolet laser in similarly staged puc (B) and wild-type (C) embryos. D: The distances between the cut cable edges in B and C are measured and plotted versus time. E: The initial rate of opening (first 20 sec) was averaged over five trials (graphed as average ± standard deviation). This rate of retraction is greater in puc mutants (0.8 ± 0.2 μm/sec for puc compared with 0.5 ± 0.1 μm/sec for wild-type, P < 0.05), providing an indirect comparison of the actin cable tension. Supplementary Videos of A, B, and C are available online (Videos 6,7). Scale bar = 15 μm.
Download figure to PowerPoint
Next, we wanted to investigate whether the zipping component of DC might also be influenced by JNK signaling. The closure process in the strong loss-of-function conditions described above (jun or slipper null mutants) fails too early for zipping to be reliably observed. Thus, we turned to embryos that are homozygous for the hypomorphic bsk2 allele. bsk2 mutations cause a moderate DC phenotype, with a leading edge actin cable that is stronger than that of an embryo completely deficient for JNK signaling. However, movies of bsk mutant embryos reveal a clear zipping defect. Whereas the canthi in a wild-type embryo move steadily toward the center of the dorsal midline (asterisks in Fig. 1A), such zipping is consistently absent or delayed in bsk2 embryos (arrows in Fig. 1D; see Supplementary Movie 4). Thus, the bsk mutant phenotype indicates that JNK is not only required for actin cable maturation but also for zipping at the canthi.
In the jun or slpr null mutants, the amnioserosa and epithelium rip away from each other causing DC to fail soon after the dorsal rows of epithelial cells have started to stretch (Fig. 1A–C). When the two cell layers have separated, the epithelium falls away ventrally, resulting in the typical final appearance of a “dorsal open” embryo. A similar effect is also seen in hypomorphic bsk mutant embryos, even though it is delayed and occurs after DC has already progressed substantially (Fig. 1D). After the contact between epidermis and amnioserosa is disrupted, the latter contracts and appears to disintegrate (this is only apparent when the amnioserosa remnants stay in the focal plane of observation, such as in Supplementary Videos 3 and 4). The breakdown of the amnioserosa is reminiscent of, and might be mechanistically related to, the removal of this tissue as it normally occurs at the end of DC when contact with the yolk sac is lost. It is unclear at this stage whether such an anoikis-like effect is responsible for the destruction of the amnioserosa after its rupture from the lateral epidermis. However, based on its timing, it seems evident that this is a consequence rather than a cause of failed DC. In summary, at least three phenotypic defects are observed in JNK signaling mutants: lack of proper actin cable formation and contraction, a decrease or absence of zipping, and a loss of attachment between the amnioserosa and the epithelium.
JNK Signaling Regulates βPS Integrin Expression During DC
The marked zipping defect of bsk2 embryos described above might be explained by impaired cell adhesion between the contralateral LE cells at the canthi in the absence of JNK signaling. Potential effectors that could conceivably mediate such changes in cell adhesion are the Drosophila βPS integrin, myospheroid (mys) and the αPS3 integrin scab (scb, Stark et al., 1997). Indeed, mutants for mys have been described previously as dorsal open, a defect that apparently results from a lack of zipping (Brown, 1994; Hutson et al., 2003). Similarly, we find that scb mutants display a specific deficiency in zipping (Fig. 1E and Supplementary Video 5 online), resembling the phenotype of both mys (Brown, 1994; Hutson et al., 2003) and the herein described phenotype of bsk mutants (Fig. 1D). Inspection of similarly staged wild-type, mys, scb, and bsk embryos reveals a marked delay in zipping and canthus progression in the mutants (Fig. 3).
Figure 3. Zipping is decreased in JNK and integrin mutants. A–H: Phosphotyrosine stains of wild-type (A,E), bsk (B,F), mys (C,G), and scb (D,H) embryos at a similar late stage of dorsal closure (DC) reveals zipping at the canthi in wild-type and a deficiency in zipping in bsk, mys, and scb mutants. Arrows point to areas where the epithelium and amnioserosa are beginning to separate from each other.
Download figure to PowerPoint
Of interest, by analyzing the genome-wide transcriptional response to JNK signaling, we previously had obtained evidence suggesting that mys was up-regulated by ectopic JNK activation in the embryo, suggesting that mys might be a JNK target gene (Jasper et al., 2001). Thus, based on the phenotypic similarities between bsk, mys, and scb and our genomic evidence, we hypothesized that JNK signaling might stimulate integrin expression to mediate efficient zipping at the canthi. To test this idea, we first examined whether expression of mys and/or scb might be regulated by JNK. To this end, JNK was ectopically activated by expression of Hepact (a constitutively activated form of JNKK) in the embryo using the engrailed-GAL4 driver (Figs. 4, 5). In such embryos, increased expression of Mys protein is detected at the cell surface throughout the embryonic segment expressing Hepact (Fig. 4B,D,F,H, the engrailed domain is marked by coexpression of GFP). In embryos expressing GFP alone (Fig. 4C,E,G), such a cortical stain of Mys protein is not observed (compare Fig. 4E and F, which are at the same cortical level as C and D, respectively). Expression of Hepact in the engrailed domain also induces ectopic expression of scb mRNA by in situ hybridization (Fig. 5A,B). Based on these in situ hybridization and immunofluorescence data, we conclude that JNK signaling can regulate the levels of αPS3 and βPS integrins in epithelial cells.
Figure 4. JNK signaling regulates Mys expression. A–H: Confocal images of embryos expressing (C,E,G) engrailed-driven GFP alone or (A,B,D,F,H) GFP and Hepact stained for green fluorescent protein (GFP; green), phalloidin (gray), and Mys protein (red). Mys protein is increased at the cell surface of the Hepact domain. C,E,G and D,F,H are at the same plane of section showing the cortical actin cytoskeleton, as stained by phalloidin. I–T: Confocal images of wild-type (I–K, close-ups in O–Q) and bsk (L–N, close-ups in R–T) mutant embryos stained for F-actin (green) and Mys protein (red). Mys protein staining is decreased specifically at the leading edge (LE) of bsk mutant embryos. The phalloidin stain showing the LE and cortical cytoskeleton is at the same plane of section as the Mys stain. Scale bars = 15 μm in N, 5 μm in D,T.
Download figure to PowerPoint
Figure 5. JNK signaling regulates scb expression. A,B: In situ hybridizations for scb are shown in embryos expressing engrailed-GAL4 alone (A) or engrailed-driven Hepact (B) just after dorsal closure but before cuticle secretion. Ectopic scb RNA is detected in the stripes of Hepact expression. C: Reverse transcriptase-polymerase chain reaction (RT-PCR) -based quantification of scb RNA in dorsal closure (DC) stage embryos expressing either dominant-negative Bsk (BskDN) or Hepact under the control of arm-Gal4 in comparison to flies expressing the driver alone (wt). Simultaneous amplification (duplex PCR) of rp49 was performed as an internal control.
Download figure to PowerPoint
Next, we examined whether JNK is necessary to direct integrin expression in relevant areas during DC. To do this, we monitored Mys protein expression in wild-type and bsk mutant embryos. Wild-type, DC-stage embryos display Mys protein in the amnioserosa, the LE and at muscle attachment sites (Fig. 4I–K,O–Q). Similarly staged bsk mutants, however, exhibit a decrease in Mys protein specifically in the LE (compare Fig. 4O–Q with R–T). Mys protein levels in the amnioserosa and at muscle attachment sites remain unchanged in bsk mutants (Fig. 4L–N). Therefore, in bsk mutants, JNK signaling appears to be required for inducing expression of Mys only in LE cells, suggesting that JNK controls integrin levels specifically in DC. Similar experiments for Scb could not be performed, as a specific antibody is not available. Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) experiments show, however, that scb mRNA levels are decreased in embryos expressing a dominant-negative form of Bsk and increased when Hepact is expressed (Fig. 5C), indicating that scb is regulated by JNK signaling in a manner that is similar to Mys.