QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 19, Issue 1, 378-393, January 2008

This Article Abstract Full Text (PDF) Supplemental Materials All Versions of this Article: E07-01-0008v1 E07-01-0008v2 19/1/378    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Stevens, T. L. Articles by Peifer, M. Search for Related Content PubMed PubMed Citation Articles by Stevens, T. L. Articles by Peifer, M.

Using Bcr-Abl to Examine Mechanisms by Which Abl Kinase Regulates Morphogenesis in Drosophila

Traci L. Stevens^*,, Edward M. Rogers, Laura M. Koontz^,, Donald T. Fox^,, Catarina C.F. Homem, Stephanie H. Nowotarski, Nicholas B. Artabazon^*, and Mark Peifer^,

Department of Biology and Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3280; and *Department of Biology, Randolph-Macon College, Ashland, VA 23005

Submitted January 8, 2007; Revised October 5, 2007; Accepted October 12, 2007
Monitoring Editor: Jean Schwarzbauer

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Signaling by the nonreceptor tyrosine kinase Abelson (Abl) plays key roles in normal development, whereas its inappropriate activation helps trigger the development of several forms of leukemia. Abl is best known for its roles in axon guidance, but Abl and its relatives also help regulate embryonic morphogenesis in epithelial tissues. Here, we explore the role of regulation of Abl kinase activity during development. We first compare the subcellular localization of Abl protein and of active Abl, by using a phosphospecific antibody, providing a catalog of places where Abl is activated. Next, we explore the consequences for morphogenesis of overexpressing wild-type Abl or expressing the activated form found in leukemia, Bcr-Abl. We find dose-dependent effects of elevating Abl activity on morphogenetic movements such as head involution and dorsal closure, on cell shape changes, on cell protrusive behavior, and on the organization of the actin cytoskeleton. Most of the effects of Abl activation parallel those caused by reduction in function of its target Enabled. Abl activation leads to changes in Enabled phosphorylation and localization, suggesting a mechanism of action. These data provide new insight into how regulated Abl activity helps direct normal development and into possible biological functions of Bcr-Abl.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
A significant insight into cancer was provided by chronic myelogenous leukemia (CML). The identification of a characteristic chromosome translocation, the Philadelphia chromosome (Ph) (Nowell and Hungerford, 1961; Rowley, 1973), ultimately led to discovery of the c-abl gene, and the revelation that it encodes a nonreceptor tyrosine kinase. This culminated in development of an Abelson (Abl) kinase-specific inhibitor, imatinib, leading to major advances in treatment of CML (for review, see Deininger et al., 2005). In parallel, scientists pursued normal functions of Abl family kinases in cultured cells and during animal development, revealing key roles in regulating many cellular processes, including cytoskeletal function. However, many questions remain about the normal roles of Abl and how the translocation alters its activity and thus alters cell behavior.

Abl normally exists in an inactive conformation, characterized by intramolecular interactions between the kinase, Src homology (SH)2 and SH3 domains, and locked into place by insertion of the myristoylated N terminus into a binding pocket on the kinase domain (Nagar et al., 2003). The mutant kinase found in CML is constitutively active, due to at least two events (for review, see Advani and Pendergast, 2002). The fusion protein retains most abl coding sequences, but deletes the myristoylation site from the extreme N terminus of Abl, thus "unlocking" the closed, inactive conformation. Sequences from Bcr also contribute to fusion protein activity. Different forms of Ph+ leukemia are associated with slightly different translocations; all have the same region of Abl, but they differ in the amount of Bcr. The shortest fusion protein, p185, is usually associated with acute lymphocytic leukemia, whereas CML patients typically express a longer fusion protein, p210. Most studies agree that the N-terminal Bcr dimerization domain, present in both fusion proteins, disrupts the autoinhibited "closed" conformation (Smith et al., 2003) and promotes transautophosphorylation and kinase activation.

Abl has a long C-terminal tail that mediates cytoskeletal interactions. Near the C terminus is a site that binds and bundles actin (Van Etten et al., 1994; Hantschel et al., 2005). This region is retained in the Bcr-Abl fusion protein, which localizes to the cytoplasm and colocalizes with actin. Cytoskeletal interactions are thought to localize Abl to sites of action. However, Abl binding to actin inhibits the kinase activity of Abl (Woodring et al., 2001), raising the possibility that Abl docks at actin in an inactive form for action in the neighborhood. The mammalian Abl paralogue, Arg, has a second actin-binding site (Wang et al., 2001) and also carries a microtubule-binding site, allowing it to cross-link actin and microtubules (Miller et al., 2004).

Studies in cultured mammalian cells and model animals both suggest that Abl plays a key role at the interface between signal transduction and cytoskeletal regulation. In cultured cells, Abl regulates cytoskeletal and adhesive responses to extracellular stimuli (for review, see Hernandez et al., 2004). For example, Abl is activated by platelet-derived growth factor stimulation, and is required for the cell ruffling and chemotaxis that occur in response (Plattner et al., 1999; 2004). Loss of both Abl and Arg or expression of kinase-dead Abl leads to increased cell motility, whereas overexpression of Abl inhibits migration (Kain and Klemke, 2001). This may suggest a role for Abl in modulating cell–matrix adhesion (for review, see Hernandez et al., 2004). Like Abl, Bcr-Abl also influences cell behavior and the cytoskeleton. In vivo, Bcr-Abl reduces adhesion to bone marrow stroma (Gordon et al., 1987), whereas in vitro it has complex, cell-type–dependent effects on adhesion to fibronectin and cell migration (Salesse and Verfaillie, 2002). Bcr-Abl also has other cytoplasmic functions (for review, see Advani and Pendergast, 2002), allowing it to stimulate proliferation and inhibit apoptosis, rendering cells independent of cytokines. Localization of Bcr-Abl exclusively to the cytoplasm suggests that these cytoskeletal and signaling events are critical.

Parallel studies of Abl function in whole animals also revealed roles connecting signaling and the cytoskeleton. In Drosophila, loss-of-function mutations (Henkemeyer et al., 1987) lead to defects in axon outgrowth in the central and peripheral nervous systems (CNS and PNS, respectively; for review, see Moresco and Koleske, 2003). Abl acts downstream of several axon guidance receptors, including the Robo family and the receptor tyrosine phosphatase Lar, to regulate both the actin and microtubule cytoskeletons in growth cones, thus mediating growth cone guidance. Drosophila Abl also plays important roles in epithelial tissues (Baum and Perrimon, 2001; Grevengoed et al., 2001), regulating the actin cytoskeleton and thus morphogenetic movements in follicle cells and the embryonic epidermis. In the epidermis, Abl works with DE-cadherin, regulating cytoskeletal responses required for cell shape changes and migration.

In mice the two Abl paralogues complicate analysis. Both single mutant mice are relatively normal—abl mutants have defects in lymphocyte and osteoblast development, whereas arg mutants have behavioral problems correlated with changes in synaptic function (for review, see Hernandez et al., 2004). However, abl; arg double mutants die before E11, and they have defects in neural tube closure (Koleske et al., 1998), probably due to defects in the actin cytoskeleton. A recent brain-specific knockout also revealed roles for Abl and Arg in dendrite branch maintenance (Moresco et al., 2005). Surprisingly, the single Caenorhabditis elegans abl gene is dispensable for viability and morphogenesis (Deng et al., 2004).

Several potential effectors may allow Abl family kinases to modulate cytoskeletal activity, including p190RhoGAP, Crk, Abi proteins, and neural Wiskott Aldrich syndrome protein (N-WASP) (for review, see Hernandez et al., 2004). Abl/Arg also can directly affect actin assembly and bundling. In Drosophila, Enabled (Ena), a member of the Ena/VASP family of actin regulators, is an essential Abl target in both the CNS (Gertler et al., 1990) and epithelia (Grevengoed et al., 2001; 2003). Ena/VASP proteins regulate actin polymerization, promoting production of unbranched actin filaments by acting as anticapping proteins (Bear et al., 2002; Barzik et al., 2005). Abl negatively regulates Ena (Gertler et al., 1990), at least in part by regulating its intracellular localization (Grevengoed et al., 2003). Whereas Abl phosphorylates Ena (Comer et al., 1998), mutations of the phosphorylation sites suggest this is not essential for negative regulation. The mammalian Ena relatives Mena and VASP are also Abl targets (Howe et al., 2002; Tani et al., 2003).

Studies of Bcr-Abl in model organisms offer the opportunity to assess the effects of Abl activation on cell behavior in vivo. Fogerty et al. (1999) generated transgenic Drosophila expressing p185 or p210 Bcr-Abl. In these, Bcr and the N-terminal Abl sequences are derived from the human oncogenes (these regions of Abl are highly similar in fly and human), whereas the C-terminal Abl tail is derived from Drosophila, maximizing the likelihood that the Bcr-Abl proteins would effectively interact with endogenous Drosophila regulators and targets. Each transgene is under the control of a GAL4-activated promotor, allowing it to be introduced into the genome in a silent state. Expression of Bcr-Abl can be activated in a tissue and temporally specific manner by crossing to lines expressing GAL4 in particular tissues or times. Interestingly, each Bcr-Abl isoform can rescue abl mutants, suggesting that their signaling properties are not grossly altered. However, neural-specific expression of these oncogenes in a wild-type background disrupts CNS development, whereas neural-specific overexpression of wild-type Abl does not, suggesting that regulated kinase activity is important (Fogerty et al., 1999). Interestingly, some neuronal Bcr-Abl phenotypes resembled those of ena loss-of-function, and Ena phosphorylation was increased in Bcr-Abl expressing flies, consistent with Ena being a Bcr-Abl target.

We used this model system to extend our analysis of Abl function during morphogenesis, examining in parallel the effects of overexpressing wild-type Abl and misexpressing Bcr-Abl. The embryonic epithelium, which undergoes many types of morphogenetic change, provides an excellent model to study the effects of inappropriate Abl activation on cell adhesion and cytoskeletal regulation, revealing insight into the mechanisms of action of Abl during morphogenesis.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Fly Stocks and Cuticle Analysis
Flies carrying UAS-Bcr-Abl and UAS-Abl transgenes were provided by F. Fogerty (University of Wisconsin, Madison, WI). In all experiments, females carrying the UAS transgene were crossed to males with a GAL4 driver (engrailed [en]-GAL4, e22-GAL4, and arm-GAL4:VP16). Sources of fly stocks are as follows: moesin-green fluorescent protein (GFP), Dan Kiehart (Duke University, Durham, NC); UAS-GFP-actin, Paul Martin (University of Bristol, Bristol, United Kingdom); UAS-PKNG58AeGFP, A. Jacinto (Lisbon, Portugal). Wild-type was Canton S. All other fly stocks were obtained from the Bloomington Stock Center (Department of Biology, Indiana University, Bloomington, IN), and they are described at FlyBase (flybase.bio.indiana.edu). Flies were grown at 25°C. Cuticle preparations were as in Wieschaus and Nüsslein-Volhard (1986).

Immunoprecipitation and Immunoblotting
Live embryos were visually selected at the cellular blastoderm stage in halocarbon oil (series 700; Halocarbon Products, River Edge, NJ) and aged defined periods of time. For immunoblotting, embryos were homogenized in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer (62.5 mM Tris-HCl, pH 8.0, 40% glycerol, 8% β-mercaptoethanol, 0.06% bromophenol blue, and 1.3% SDS) with 0.5 mM sodium orthovanadate and Complete EDTA-free protease inhibitors (Roche Molecular Biochemicals, Lewes, United Kingdom). Samples were analyzed by 6% SDS-PAGE and transferred to nitrocellulose. Membranes were blocked and incubated with primary and secondary antibodies in 5% nonfat dry milk (or 5% bovine serum albumin for phosphospecific antibodies) in Tris-buffered saline/Tween 20. Signal was detected using the ECL kit (GE Healthcare Bio-Sciences, Little Chalfont, Buckinghamshire, United Kingdom). For immunoprecipitations, embryos were homogenized in NET buffer (50 mM Tris, pH 7.5, 400 mM NaCl, 5 mM EDTA, and 1% NP-40) on ice. Homogenate was clarified by centrifugation at 3500 x g for 6 min, and incubated with primary antibodies for 1 h at 4°C. Protein A-Sepharose beads were added, incubated 2 h at 4°C, washed six times with NET buffer and boiled in SDS-PAGE sample buffer. Antibodies used were as follows: mouse monoclonals anti-phosphotyrosine (clone 4G10, 1:250; Upstate Biotechnology, Lake Placid, NY), anti-phosphotyrosine-RC20:HRP0 (1:500; BD Biosciences, San Jose, CA), anti-actin (1:500; Chemicon International, Temecula, CA), anti-armadillo (Arm) (1:250), anti-Pnut (1:30), anti-BicD (1:500), and anti-Ena (1:250; all Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA); rat monoclonals anti--catenin (1:250) and anti-DCAD1 (1:500; both Developmental Studies Hybridoma Bank); rabbit anti-c-Abl (phospho Y412) (1:250; Abcam, Cambridge, MA) and anti-Bcr (N-20) (1:250; Santa Cruz Biotechnology, Santa Cruz, CA); and guinea pig anti-dAbl (1:50; Grevengoed et al., 2001).

Immunofluorescence and Microscopy
Embryos were dechorionated in bleach and fixed in 1:1 3.7% formaldehyde in phosphate-buffered saline:heptane for 20 min at room temperature. Vitelline membranes were removed in methanol. For actin visualization, embryos were fixed in equal amounts of 37% formaldehyde and heptane for 5 min at room temperature; vitelline membranes then were removed manually. Embryos were mounted in Aqua PolyMount (Polysciences, Warrington, PA), and digital images were obtained on a confocal laser-scanning microscope (LSM510; Carl Zeiss MicroImaging, Jena, Germany) by using 40x (Plan-NeoFluar; numerical aperture [NA] 1.3) and 63x (Plan-Apochromat; NA 1.4) objectives, and LSM software. Antibodies used were mouse monoclonals anti-Ena (1:200), anti-Rho1 (1:100), anti-Crb (1:25), anti-Dlg (1:100; all Developmental Studies Hybridoma Bank) and anti-phosphotyrosine (4G10) (1:1000; Upstate Biotechnology); rabbit anti-c-Abl (phospho Y412) (1:250) and anti-Bcr (N-20) (1:250); and guinea pig anti-dAbl (1:100). Actin was visualized with Alexa 488 and Alexa 568 phalloidin (Invitrogen, Carlsbad, CA). Secondary antibodies conjugated to cyanine (Cy)3, Cy5, Alexa 488, and Alexa 568 (Invitrogen) were used. Adobe Photoshop 7.0 (Abobe Systems, Mountain View, CA) was used to adjust input levels so the main range of signals spanned the entire output gray scale, and it was used to adjust brightness and contrast. When protein levels were compared, compared images were equally adjusted.

Time-Lapse Microscopy
Embryos were dechorionated in bleach and mounted in halocarbon oil (series 700; Halocarbon Products Corporation) on a gas permeable membrane (petriPERM 50, hydrophobic; Sartorius). A Perkin Elmer-Cetus Ultraview spinning-disk confocal, ORCA-ER digital camera (Hamamatsu), and Metamorph software were used. Images were acquired every 10 s, except video 2, in which images were acquired every 15 s. Image J was used for quantitation of lamellipodial area and filopodial length.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Abl and Bcr-Abl as Tools to Explore the Effects of Abl Activity during Morphogenesis
To explore the mechanisms by which Abl acts during morphogenesis, we activated Abl by overexpression or by expressing constitutively active Bcr-Abl isoforms. We used transgenic lines expressing wild-type Abl or a Bcr-Abl isoform under the control of the GAL4-UAS system. Different GAL4 drivers allow us to express these lines at different levels and in different cells, thus providing a broad range of Abl hyperactivity. To begin our analysis, we examined expression levels in embryos of several independent lines of each construct, to control for effects of insertion site. We used the e22c-GAL4 driver, which drives expression ubiquitously in both the embryonic epidermis and amnioserosa. We examined UAS-Bcr-Abl lines by immunoblotting with anti-Bcr; most lines expressed Bcr-Abl at roughly similar levels (Supplemental Figure 1A; we used lines with high-level expression for further experiments). We examined the UAS-Abl lines using our antibody raised against Drosophila Abl (Grevengoed et al., 2001). Once again, expression levels in most lines were equivalent (Supplemental Figure 1B). Expression of UAS-Abl driven by e22c-GAL4 is roughly 4–9 times the level of endogenous Abl (range of two different lines each assayed twice). We could not directly compare expression levels of UAS-Abl and UAS-Bcr-Abl, because our fly Abl antibody does not recognize Bcr-Abl as well as it recognizes fly Abl (Supplemental Figure 1C; contrast reactivity with anti-phosphotyrosine and anti-Abl), likely because the region used as an epitope for our anti-fly Abl antibody is derived from human Abl in the Bcr-Abl transgene. However, given the uniformity in expression levels of different lines of the same construct, we suspect that transcription of Abl and Bcr-Abl lines driven by the same driver is likely to be similar, although we cannot rule out differences in protein stability.

Both Abl and Bcr-Abl Are Enriched at the Cortex When Overexpressed
One key question is where within the cell Abl acts during morphogenesis. As a first step to assess this, we examined the localization of overexpressed Abl and misexpressed Bcr-Abl. In wild-type embryos, Abl is cortically localized in early embryogenesis, but after germband extension cortical enrichment is reduced, with primarily diffuse cytoplasmic staining during germband retraction or dorsal closure (Bennett and Hoffmann, 1992; Fox and Peifer, 2007). At later stages, contrast enhancement of the signal revealed weak cortical enrichment in cells expressing normal levels of Abl (Figure 1D), but it was subtle at best. However, when we overexpressed wild-type Abl in stripes of cells using the en-GAL4 driver, we saw cortical enrichment in addition to significant cytoplasmic staining (Figure 1, A and B'', arrow). Cortical enrichment increased during dorsal closure (Figure 1C'', arrow). When Abl was expressed at even higher levels, using arm-GAL4:VP16, the degree of cortical enrichment was increased (Figure 1, E–H). Interestingly, kinase activity was not essential for cortical enrichment, because the kinase-dead mutant exhibited a similar localization (Figure 1J''). We also examined the effect of Abl overexpression on the levels of tyrosine (Tyr)-phosphorylated proteins, using anti-phosphotyrosine (PTyr) antibodies. Normally, PTyr is significantly enriched at the cortex in adherens junctions (AJs) (e.g., Cox et al., 1996). Overexpression of wild-type Abl led to a significant increase in the levels of PTyr signal in both the cytoplasm and at the cortex, which increased in parallel with Abl accumulation as development proceeded (Figure 1, B' and C'). This depended on Abl kinase-activity (Figure 1J'). We also examined the localization of the Bcr-Abl fusion proteins, using antibodies against Bcr. Like overexpressed Abl, both p185 Bcr-Abl (Figure 1, L'' and N'', arrows) and p210 Bcr-Abl (Figure 1P'', arrow) accumulated in the cytoplasm but were enriched at the cell cortex. This localization also did not require kinase activity of the fusion-protein (Figure 1R'', arrow). Both activated forms of Bcr-Abl stimulated very high levels of accumulation of PTyr both at the cortex and in the cytoplasm (Figure 1L', N', and P'), and once again this was dependent on kinase activity (Figure 1R'). Because Abl can exist in an autoinhibited conformation predicted to be cytoplasmic, and because Abl overexpression or Bcr-fusion is likely to trigger activation, these data are consistent with the hypothesis that active Abl is enriched at the cell cortex.

View larger version (119K): [in this window] [in a new window]   Figure 1. Abl and Bcr-Abl localization. Embryos, anterior to the left, lateral view unless indicated. Antigens are indicated. (A–D) Wild-type UAS-Abl x en-GAL4. A and B, stage 11. C and D, stage 14. Arrows, cells overexpressing Abl. Arrowheads, cells not overexpressing Abl. (E–H) Wild-type UAS-Abl x arm-GAL4:VP16. E and F, stage 10. G and H, stage 14, ventral views. I and J, UAS-Abl-kinase dead x en-GAL4, stage 13. J, arrow, cortical Abl-kinase dead. (K–N) UAS-p185Bcr-Abl x en-GAL4. K and L, stage 10. M and N, stage 13. O and R, UAS-p210Bcr-Abl x en-GAL4. O and P, stage 12. Q and R, UAS-p210Bcr-Abl-kinase dead x en-GAL4, stage 14, ventral view. L'', N'', P'', R'', arrows, increasing levels of cortical Bcr-Abl. Bars, 20 µm.

We thus examined whether an antibody against this phosphorylated tyrosine recognizes activated fly Abl. On immunoblots, our anti-Abl antibody recognizes a doublet at 180 kDa (Figure 2A, xAbl, lane 4). The activated Abl antibody recognizes this doublet (Figure 2A, xPAbl, lane 4, arrowhead), and it also recognizes bands at >200 kDa (Figure 2A, xPAbl, lane 4, arrow) which may be cross-reacting proteins. Overexpression of either wild-type Abl (line UA22) or kinase-dead Abl (line UL6) elevated the intensity of the 180-kDa bands recognized by anti-Abl antibody (Figure 2A, xAbl, lanes 1, 3). Overexpression of wild-type Abl leads to an increase in the signal detected by the activated Abl antibody (Figure 2A, xPAbl, lane 1 vs. 4, arrowhead). In contrast, overexpression of kinase-dead Abl does not lead to a corresponding increase in the signal detected by activated Abl antibody (Figure 2A, xPAbl, lane 3 vs. 4, arrowhead); active-Abl signal remains in the kinase-dead Abl lane due to the endogenous wild-type Abl present in these embryos. The activated Abl antibody also recognizes the Bcr-Abl fusion proteins (Figure 2A, xPAbl, lane 2 asterisk; data not shown). This signal is very strong, consistent with the expectation that a larger fraction of Bcr-Abl should be in the activated state (the region of the fusion protein around Tyr 412/539 is derived from fly Abl, and thus this difference is not due to a difference in cross-reactivity between species). As a separate test of whether the activated Abl antibody recognized activated Abl in Drosophila, we overexpressed Abl or Bcr-Abl in stripes in the embryonic epidermis, and stained the embryos with anti-PAbl (Figure 2, B–D; the signal in the Bcr-Abl panel is turned down relative to the others to avoid saturating the signal in the stripes). Both Abl overexpression and Bcr-Abl expression led to elevated PAbl staining, with the elevation most pronounced after Bcr-Abl overexpression, consistent with its expected high level of activity. These data suggest that activated Abl antibody is a good tool to examine activated Abl in Drosophila, although caution is warranted by the presence of cross-reacting bands.

View larger version (113K): [in this window] [in a new window]   Figure 2. Localization of active Abl. (A) Anti-PAbl(Y412) recognizes activated Abl in Drosophila. Immunoblots of embryo extracts from wild-type embryos or embryos expressing the indicated transgenes with e22c-GAL4, blotted with the antibodies indicated. UAS-Abl was line UA22, p185 was line 17, and UAS-Abl-kinase-dead (KD) was line UL6. Arrowhead, band at position of wild-type Abl. Arrow, presumed cross-reacting band. Asterisk, band at position of Bcr-Abl. Peanut (Pnut) is a loading control. (B–L) Embryos, anterior to the right. All dorsal up except E, F, H, and J, which are ventral views. Staining with anti-PAbl(Y412). B and C, wild-type UAS-Abl x en-GAL4, stages 11 and 13, respectively. D, UAS-p185Bcr-Abl x en-GAL4, stage 13. E–H, wild-type, stages indicated. I and J, wild-type UAS-Abl x en-GAL4, stage 12, shorter and longer exposures. K, UAS-p185Bcr-Abl x en-GAL4, stage 11. L and L', UAS-p210Bcr-Abl x en-GAL4, stage 11, shorter and longer exposures. (M) Levels of phosphotyrosine induced by transgenes. Embryo extracts from wild-type embryos or embryos expressing the indicated transgenes with e22c-GAL4, blotted with the antibodies indicated. Lines as in A, plus UAS-p210 was line 31. Bars, 20 µm.

Expression of Abl or Bcr-Abl Leads to Embryonic Lethality and Disrupts Morphogenesis
Loss of Abl disrupts a number of processes during embryonic morphogenesis (Grevengoed et al., 2001). We hypothesized that regulated Abl kinase activity also plays a key role in regulating these processes. To test this hypothesis, we examined the biological consequences of activating Abl in epithelial cells, by using three GAL4 drivers to express the transgenes in different spatial patterns and/or levels of expression. en-GAL4 drives expression in the posterior-most two to three cells of each epidermal segment, with little or no expression in the amnioserosa. e22c-GAL4 drives expression ubiquitously and at fairly uniform levels in both the embryonic epidermis and amnioserosa. arm-GAL4:VP16 is the strongest driver, and it also leads to ubiquitous expression in the epidermis and amnioserosa. With all three GAL4-drivers, the onset of transgene expression is during the midextended germband stage, with increased accumulation during germband retraction and subsequent stages.

We first examined whether Abl activation affects embryonic viability. When we expressed wild-type Abl ubiquitously, using either e22c-GAL4 or arm-GAL4:VP16, it led to embryonic lethality, although the penetrance of this lethality was lower with e22c-GAL4 (Table 1), suggesting that embryos can tolerate reasonably high levels of wild-type Abl. Embryonic lethality requires kinase activity, because expression of kinase-dead Abl was not embryonic lethal (Table 1). We also expressed the two isoforms of Bcr-Abl by using the same two GAL4-drivers. Expression of either p185 or p210 resulted in highly penetrant embryonic lethality (Table 1), which once again required kinase activity. We also expressed a version of p185 carrying a mutation altering the putative Grb2 binding site in the Bcr portion of Bcr-Abl (Y177F). Expression of this construct was also embryonic lethal (Table 1). In contrast, expression of Abl or Bcr-Abl using the segmentally restricted en-GAL4 driver did not, in general, lead to embryonic lethality (Table 1; the background lethality of different wild-type stocks is 2–10%). These data suggest that widespread unregulated Abl kinase activity is deleterious, whereas embryos can tolerate reasonably high levels of wild-type Abl.

View larger version (151K): [in this window] [in a new window]   Figure 3. Lethality often occurs without severe effects on the CNS. Axons of the embryonic CNS, revealed by anti-BP102 antibody. (A) Wild type. (B–D) Indicated transgenes crossed to e22c-GAL4. Arrows, commissures. Arrowheads, longitudinal axons. Bar, 20 µm.

View larger version (172K): [in this window] [in a new window]   Figure 4. Abl overexpression and Bcr-Abl expression cause defects in morphogenesis. Embryonic cuticles, anterior to the top. (A and B) Wild-type, ventral and dorsal views. Arrow, head skeleton. (C–J) Average cuticle phenotypes of indicated transgenes crossed to indicated drivers; C–I, arrows, defects in head involution: arrows in C–E indicate mild defects in head skeleton, whereas arrows in F–I indicate more severe head holes. F–I, arrowheads, defects in dorsal closure indicated by dorsal holes (G and I) or dorsal scars (F and H). (K and L) Embryos overexpressing wild-type Abl by using e22c-GAL4, which are either wild type (K) or homozygous mutant (L) for ena. An ena mutant cuticle is essentially wild type. Arrow, head hole. Arrowhead, dorsal hole. Bar, 20 µm.

Abl Activation Affects Cell Shape Changes during Dorsal Closure and Segmental Groove Retraction
These data demonstrate that activated Abl disrupts the end product of morphogenesis, the embryonic cuticle. To get a more detailed mechanistic understanding of how activated Abl affects morphogenesis, we examined the biological and cell biological consequences of Abl activation. We first examined fixed embryos, looking for alterations in the cytoskeleton or cell adhesion that might help explain the defects in morphogenesis. We followed the expression of Bcr-Abl by using an antibody directed against Bcr. To follow Abl overexpression, we used either an antibody against Drosophila Abl or an antibody directed against the activated form of mammalian Abl (Figure 2). We used phalloidin to reveal cell shapes and to mark the actin cytoskeleton.

We examined embryos beginning with the onset of germband retraction, soon after initiation of transgene expression, and continuing through dorsal closure. During wild-type gastrulation, embryos elongate in the anterior-posterior body-axis and narrow in the dorsal-ventral axis, such that the posterior end is now tucked behind its head. During germband retraction the tail end moves back to the posterior (Figure 5A), leaving a sheet of epidermal cells covering the ventral and lateral surfaces of the embryo, and the aminoserosal cells spread over the dorsal surface (Figure 5A, arrowhead). Before the end of germband retraction, there were only subtle effects on morphogenesis. Germband extension was not affected, but it may occur before significant expression of the transgenes.

View larger version (180K): [in this window] [in a new window]   Figure 5. Cellular and cytoskeletal consequences of ubiquitous Abl overexpression and Bcr-Abl expression. Embryos, anterior left and, unless noted, lateral view, stained with phalloidin to visualize F-actin. Transgenes and GAL4 drivers are indicated [e22c-GAL4 (e22), arm-GAL4:VP16 (VP16)]. (A–G) Stage 12. (H–P) Stage 13. (Q–W) Stage 14 (R–V, dorsal views). Arrows, segmental grooves. Arrowheads in A and B, amnioserosa. Arrowheads in C, D, and G, leading edge cells. Arrowheads in H–W, leading edge actin cable. Brackets in L–N, elongation of leading edge cells. Long arrows in H, K, L, and O–Q, normal amnioserosa (H and L) or holes in amnioserosa (K and O–Q). X–CC, ventral views. X and CC, stage 12. Y–BB, stage 14. Arrows, segmental groove. Arrowheads, apical cell surface. Bars, 20 µm.

We next examined dorsal closure. This is one of the most dramatic events of morphogenesis, involving a complex, coordinated series of cell shape changes, migration events, and cytoskeletal alterations. Before dorsal closure, columnar epidermal cells cover the ventral and lateral sides of the embryo (Figure 5, A and C), whereas squamous amnioserosal cells (Figure 5A, arrowhead) cover the dorsal surface. During germband retraction, cells at the leading edge of each lateral epidermal sheet first change shape (Figure 5C, arrowhead) and then elongate along the dorsal-ventral axis. A wave of elongation then spreads ventrally, until all epidermal cells are elongated along this axis (Figure 5L, bracket). As they elongate, leading edge cells assemble an actomyosin cable along their leading edge (Figure 5, H and L, arrowheads), which is anchored at cell–cell junctions, forming a supracellular contractile purse-string (Kiehart et al., 2000; Jacinto et al., 2002). Meanwhile, amnioserosal cells (Figure 5, H and L, long arrows) begin to constrict their apical ends—this occurs slowly in all cells, with a subset of cells initiating rapid constriction and disappearing from the epithelial sheet. The combined forces of elongation/migration of epithelial cells, actin cable contraction and amnioserosal cell apical constriction drive dorsal closure, culminating in the complete encapsulation of the embryo in the epidermis and the internalization of the amnioserosa (Kiehart et al., 2000; Hutson et al., 2003). The final stages of the process involve zipping together of the leading edge (Figure 5T, arrowhead) and precise alignment of cells from the two sides.

As dorsal closure was initiated, effects of Abl activation became more pronounced, altering many but not all of the cytoskeletal and morphogenetic movements occurring during this process. The effects of overexpression of wild-type Abl were largely confined to deepening of segmental grooves (Figure 5, I and W, arrows); these effects were most pronounced in embryos overexpressing Abl at the highest levels (by using arm-GAL4VP16; data not shown). However, wild-type Abl overexpression did not disrupt other events of dorsal closure, including cell shape changes and actin cable formation (Figure 5W and Y vs. Z).

Ubiquitous misexpression of p185 or p210 Bcr-Abl had more severe effects on dorsal closure, with p210 the most severe. The normally uniform dorsal-ventral elongation of epidermal cells (Figure 5L, bracket) was disrupted, with some cells elongating less than their neighbors (Figure 5, M and N, bracket). Moderate activation (p185 or p210 x e22c-GAL4) led to persistence of very deep segmental grooves throughout dorsal closure (Figure 5, J, K, M–O, and U, arrows), in contrast to wild-type (Figure 5, H, L, and T). This is likely due to defects in the cell shape changes of segmental groove cells, but the dramatic groove depth in mutants made direct observation of cell shape impossible. These two defects combined to cause significant defects in the completion of dorsal closure and in alignment of segments at the dorsal midline (Figure 5R vs. S and T vs. V, arrowheads; the actin projections on epidermal cells in Figure 5, U–W are dorsal hairs; because mutants close more slowly, dorsal hairs form before closure is completed). However, despite these defects, cell shape changes in the ventral epidermis were not dramatically disrupted (Figure 5Y vs. AA). Strong activation of Bcr-Abl (p185 x arm-GAL4:VP16, data not shown; p210 x e22-GAL4, Figure 5, H and L vs. N and O) led to even deeper segmental grooves (Figure 5O, arrow) and affected amnioserosal cell shapes (Figure 5O), although the actin cable remained intact (Figure 5, N and O, arrowheads). In these genotypes, dorsal closure often failed completely, likely due in part to disruption of integrity of the amnioserosal epithelium (Figure 5, K and O, long arrows). High-level expression of p210 also led to abnormal apical actin accumulation in epidermal cells (Figure 5X vs. CC, arrowheads). The very strongest defects were seen when p210 was expressed with arm-GAL4:VP16. This led to extreme deepening and persistence of segmental grooves, dorsally and even midventrally (Figure 5Y vs. Q and BB; ventrally, these grooves engulf the forming denticle belts), and to dramatic disruption of the amnioserosa (Figure 5, P and Q, long arrows). Thus, inappropriate Abl activation can disrupt many but not all of the cell shape changes and movements of morphogenesis, and the strength of these defects is graded based on level of expression and of activation.

We also examined the effects of localized expression of Abl or Bcr-Abl in the posterior cells of each segment, by using en-GAL4. This allowed us to directly compare neighboring cells with and without Abl activation, looking for more subtle changes in actin and cell architecture. Little effect on morphogenesis was seen through germband retraction, even in embryos expressing the strongest p210 lines (Figure 6, A–D). Dorsal closure was essentially wild type in embryos expressing increased wild-type Abl or expressing p185 in en-stripes. As dorsal closure was completed, there was a subtle deepening of segmental grooves dorsally (Figure 6, J' and L'), when they should have disappeared (Figure 6J'). However, there was little effect on cell shape changes (Figure 6N') or the actin cable, and dorsal closure proceeded normally (Figure 6, K' and L', yellow arrowheads, and N'). Expression of p210 triggered deeper, more persistent segmental grooves (Figure 6E' vs. F' and G). In addition, misexpressing leading edge cells were slightly less elongated along the dorsal-ventral axis than their wild-type counterparts, and they had slightly elevated apical actin (Figure 6H' vs. I', arrows), but the actin cable remained unaltered (Figure 6I', arrowhead). As dorsal closure was completed, expression of p210 in en-stripes led to a drastic deepening and persistence of the segmental grooves dorsally (Figure 6M'), with the p210-expressing cells invaginated deep into the embryos (Figure 6O', arrowheads).

View larger version (152K): [in this window] [in a new window]   Figure 6. The cellular and cytoskeletal consequences of localized Abl overexpression and Bcr-Abl expression. Embryos, anterior left, and, unless noted, lateral view. Antigens indicated. Transgenes indicated were driven by en-GAL4. (A–D) Stage 12. Brackets in C and D, en-GAL4 expression domain. (E–I) Stage 13 (G is a ventral view). E–G, blue arrowheads, en-GAL4–expressing cells. Yellow arrowheads, segmental grooves. H and I, arrows, en-GAL4–expressing cells. White arrowheads, leading edge actin cable. (J–O) Stage 14. White arrowheads, segmental grooves. Yellow arrowheads, leading edge actin cable. (N') Red arrowhead, unaltered cell shapes. (O) Red arrowheads, hyperconstricted Bcr-Abl–expressing cells. Bars, 20 µm.

Abl Activation Alters Cell Behavior during Dorsal Closure
These data suggested that Abl activation affects cell shape changes and morphogenetic movements, especially during dorsal closure. To address the effect on Abl activation on cell behavior directly, we examined morphogenesis in living embryos expressing GFP-tagged cytoskeletal proteins. This reveals a wider range of cell behaviors than can be seen in fixed embryos, and it also adds dynamic information. We, initially examined the effects of ubiquitous overexpression of Abl and misexpression of Bcr-Abl, using e22c-GAL4.

We first assessed dorsal closure in living embryos ubiquitously expressing Moe-GFP (F-actin–binding domain of moesin fused to GFP). This reports F-actin localization in live embryos (Figure 7A and Supplemental Video 1; Edwards et al., 1997). It is a superb reporter for cytoskeletal rearrangements, cell shape changes, and other events during complex morphogenetic processes. In wild-type embryos, when dorsal closure begins, epidermal cells have elongated along the dorsal-ventral axis and cover the dorsal and lateral surfaces of the embryo (Figure 7A, 0:00, white arrow), whereas squamous amnioserosal cells (Figure 7A, black arrowheads) cover the dorsal surface. Epidermal leading edge cells assemble the actomyosin cable (Figure 7A, white arrowheads), the contraction of which helps drive closure (Kiehart et al., 2000; Jacinto et al., 2002). Leading edge cells extend filopodia and lamellipodia over the surface of the adjacent amnioserosal cells (Jacinto et al., 2000), whereas amnioserosal cells apically constrict (Figure 7A, black arrowheads), also helping drive closure (Kiehart et al., 2000; Hutson et al., 2003). Apical ends of the amnioserosal cells are covered by filopodia (Figures 7A, 0:00, black arrowhead, 8, A and C, and Supplemental Video 2). As the epithelial sheets approach one another, they zipper together (Figure 7A, black arrows).

View larger version (150K): [in this window] [in a new window]   Figure 7. Effect of Abl overexpression and Bcr-Abl misexpression on morphogenesis and cell behavior. Stills, movies of dorsal closure in living embryos expressing moesin-GFP. Times in hours and minutes. (A) Wild-type. Black arrowheads, amnioserosal cells. White arrowheads, actin cable at leading edge. White arrow, leading edge cells elongated in dorsal-ventral axis. Black arrows, zippering together as two sheets meet. (B) Embryo expressing Bcr-Abl p185(6a1) by using e22c-GAL4. Note delay in closure. Black arrowheads, abnormally shaped amnioserosal cells, often with ectopic lamellipodia (e.g., 0:27). White arrowheads, uneven leading edge. Black arrow, puckering during zippering. White arrow, persistent segmental groove. (C) Embryo expressing Bcr-Abl p210(11a) by using e22c-GAL4. Black arrowhead, abnormally shaped amnioserosal cells. White arrowheads, uneven leading edge. Black arrows, rip in aminoserosa. White arrows, persistent segmental groove. (D) Embryo overexpressing Abl (UA22) by using e22c-GAL4. Black arrowheads, amnioserosal cells. White arrowheads, uneven leading edge. White arrows, persistent segmental grooves. Bar, 20 µm.

View larger version (137K): [in this window] [in a new window]   Figure 8. Abl overexpression and Bcr-Abl misexpression alter filopodia and lamellipodia. (A–E) Stills, movies of amnioserosa in living embryos expressing moesin-GFP. Misexpression driven by e22c-GAL4. A and C, wild-type (WT). Amnioserosal cells surrounded by filopodia (arrows). B, p210(11a). D, p185(6a1). Note reduction of filopodia and replacement with lamellipodia (arrows). E, Wild-type Abl overexpression (UA22). Filopodia are reduced (arrow) without appearance of lamellipodia. (F–H and K and L) Stills, movies of dorsal closure in living embryos expressing actin-GFP (F–H) or GFP-Ena (K and L) and the indicated transgene by using the en-GAL4 driver. Time = minutes:seconds. F, wild-type. Cell projections are highly dynamic lamellipodia with filopodia extensions. G, p185. Lamellipodia remain but filopodia are reduced substantially (arrows). H, overexpressed Abl. All protrusions are reduced in size and number. Most remaining protrusions are short and blunt (arrows). (I) Quantitation of filopodial number, with standard deviations. Filopodia were defined as any protrusion <1.25 µm wide extending beyond the lamellipodium or leading edge, and they were quantitated within en-GAL4 stripes in three to five embryos of each genotype as the leading edges moved from 26.9 to 7.9 µm apart. (J) Quantitation of average lamellipodial area, with standard deviations. Lamellipodia were defined as any projection in the direction of migration >1.3 µm in width and >1.3 µm in length, and area was calculated in ImageJ in frames 1 min apart during the same part of dorsal closure using for filopodial measurements. p values are in text. K, Co-overexpression of GFP-Ena and p185 restores filopodia. L, expression of GFP-Ena alone. Bars, 5 µm.

Expression of strong p210 lines using the same driver had similar but more severe consequences (Figure 7C and Supplemental Video 5). The leading edge was uneven (Figure 7C, white arrowheads), segmental grooves were deep and persistent (Figure 7C, white arrowheads, the dorsal opening oval-shaped (Figure 7C, 0:51), and amnioserosal cells were abnormal in shape and extended lamellipodia rather than filopodia (Figures 7C, 0:00, black arrowhead; and 8B). One additional defect was seen in embryos misexpressing p210—the amnioserosal cell sheet often ruptured (Figure 7C, 1:21–2:10, black arrows), thus contributing to the dorsal holes in cuticles of this genotype. Because rupture ended dorsal closure, it was impossible to calculate time to closure.

Misexpression of wild-type Abl affected dorsal closure in largely similar ways (Figure 7D and Supplemental Video 6), but the effects were less severe than those of Bcr-Abl. Abl overexpression led to an uneven leading edge (Figure 7D, 0:51, white arrowheads) and deepened segmental grooves (Figure 7D, 2:10 and 2:37, arrows). The shape of the dorsal opening was also more oval (Figure 7D, 0:51), as was seen in embryos expressing Bcr-Abl, and in the later stages of dorsal closure the amnioserosa was in a deeper focal plane than the epidermis; both suggest that zippering together of the two epidermal sheets is slowed. Abl overexpression had distinct effects on the cell protrusions produced by amnioserosal cells from Bcr-Abl; filopodia were reduced but lamellipodial formation was not triggered by Abl overexpression (Figures 7D, black arrowheads; and 8E). Abl overexpression also significantly slowed dorsal closure (average time for Abl-overexpressing embryos to close 128 min [n = 5] vs. 79 min in wild type [n = 5]; p = 0.001). Thus, live analysis revealed dramatic effects of Abl activation on cell behavior during dorsal closure, even in genotypes where closure was successful.

Bcr-Abl Reduces Filopodial Number in Leading Edge Cells
We next zoomed in on behavior of individual cells. During dorsal closure, epidermal cells at the leading edge send out dynamic, actin-based cell processes, which can be visualized by expressing actin-GFP in stripes of epidermal cells (Jacinto et al., 2002). In wild-type embryos, these actin-based cell processes consist of broad lamellipodia from which emerge filopodia, thus, resembling neuronal growth cones (Supplemental Video 7 and Figure 8F, 5:00, arrows). These structures are highly dynamic: new processes constantly emerge, evolve, and retract. We thus assessed whether Bcr-Abl expression affected the cell processes formed by leading edge cells.

When we expressed both Bcr-Abl and actin-GFP by using en-GAL4, we saw a striking change in the nature of actin processes formed. Cells still produced broad lamellipodia, but there were far fewer filopodia extending from them (Supplemental Video 8 and Figure 8G, arrows). To quantitate this, we measured the number and maximum length of each filopodium (Figure 8I; see figure legend for methodology). Wild-type leading edges made an average of 21.9 filopodia per en-stripe (n = 9). In contrast, embryos misexpressing p185 made 6.4 filopodia per en-stripe (n = 14), whereas embryos misexpressing p210 made 6.2 filopodia (n = 12) and 8.8 filopodia (n = 20) per en-stripe in the two lines quantitated. All were significantly different from wild-type (p < 7 x 10^–7). In embryos misexpressing p185, the remaining filopodia were also somewhat shorter than in wild-type (2.14 µm [n = 90] in p185 vs. 2.54 µm [n = 197] in wild-type; p = 0.008). To our surprise, the filopodia in embryos expressing p210 were not shorter than those in wild-type—in fact, in they were somewhat longer (2.91 µm [n = 74] and 2.94 µm [n = 175] in p210 vs. 2.54 µm in wild type). We also compared total area of the lamellipodial cell projections made by wild-type and Bcr-Abl misexpressing embryos (Figure 8J; see figure legend for methodology). No dramatic differences were seen—lamellipodial area in p185-expressing embryos slightly increased (from 36 µm² in wild-type [n = 262] to 42 µm² [n = 478]; Figure 8J) and lamellipodial area in p210-expressing embryos was statistically unchanged or slightly decreased in the two lines we assessed (34 µm² [n = 391] and 29 µm² [n = 634]; Figure 8J).

Overexpression of wild-type Abl had distinct effects from Bcr-Abl. Leading edge cells overexpressing Abl produced fewer protrusions overall (Supplemental Video 9 and Figure 8H). Many of the protrusions produced were short "filopodia" that emerged directly from the cell, rather than from a lamellipodium. Filopodia were substantially reduced in number (11.6 filopodia per en-stripe [n = 16] vs. 21.9 per en-stripe in wild-type; p = 7 x 10^–5; Figure 8I) and in length (2.07 µm [n = 186] in Abl-overexpressers vs. 2.54 µm in wild-type; p = 6 x 10^–5). In addition, total lamellipodial area was substantially reduced from wild-type (Figure 8J; from 36 µm² in wild-type [n = 262] to 11 µm² in Abl overexpressers [n = 390]; p < 3 x 10^–56), contrasting with Bcr-Abl misexpression. Thus, Bcr-Abl and Abl overexpression both affect cell behavior and the actin cytoskeleton during morphogenesis, and their effects on cell behavior are similar but not identical.

Ena Is a Key Target of Activated Abl
These dramatic cell biological effects suggest that deregulated Abl alters cell behavior and are consistent with an effect on actin dynamics. To understand the mechanisms of Abl action, we need to identify its key targets. One known target of Abl is the actin modulator Ena, which Abl negatively regulates (Gertler et al., 1990). In parallel work, we examined the effects of loss of maternal and zygotic Ena function on morphogenesis (Gates et al., 2007). Strikingly, many of the phenotypes induced by Ena depletion were very similar to those caused by Abl activation; both caused defects in head involution, resulted in substantially deepened segmental grooves, slowed dorsal closure, and substantially reduced the numbers of filopodia produced by leading edge cells. These data suggest that down-regulation of Ena activity may be a key part of the mechanism by which deregulated Abl kinase disrupts morphogenesis.

To test whether Ena regulation is important for effects of activated Abl on morphogenesis, we first tested whether reducing Ena levels modifies the consequences of Abl activation. We overexpressed Abl in a background zygotically mutant for ena (reducing but not eliminating Ena, due to its maternal contribution). Reduction of the Ena dose substantially enhanced the cuticle phenotype of Abl overexpression (Figure 4, K vs. L, and Supplemental Table 1), consistent with Ena down-regulation playing an important role in the effects of Abl activation.

If Ena is a key Abl target, the negative regulation triggered by Abl activation might be alleviated by Ena overexpression. To test this mechanistic hypothesis, we examined whether Ena overexpression would alleviate the effects of activated Abl on cell behavior, by co-overexpressing Bcr-Abl with a GFP-tagged form of Ena (Gates et al., 2007). As described above, Bcr-Abl expression substantially reduced the production of filopodia. GFP-Ena localizes to the ends of filopodia, allowing us to visualize them in living embryos (Gates et al., 2007). We found that co-overexpression of GFP-Ena restored filopodia to leading edge cells expressing p185 Bcr-Abl (Figure 8K; co-overexpressing embryos were selected using a GFP-marked Balancer chromosome); the filopodia resembled those seen in embryos expressing GFP-Ena alone (Figure 8L). These data further support the hypothesis that Ena is a key target of Abl action during morphogenesis.

Abl Activation Disrupts Ena Localization
These data raise the question of the mechanism by which activated Abl regulates the actin regulator Ena. Activated Abl and Ena both localize to AJs, especially at tricellular junctions (Figure 9, A vs. B). When Abl is inactivated in early embryos, Ena accumulates at ectopic locations (Grevengoed et al., 2003). Our genetic interaction experiments described above (Figure 4, K vs. L) suggest that Ena misregulation accounts, at least in part, for the morphogenesis defects seen upon Bcr-Abl misexpression. We thus examined Ena localization in embryos misexpressing wild-type or activated Abl to determine whether alterations in its localization explain some of the cell biological effects we observed. In wild-type embryos, Ena localizes to AJs of epidermal epithelial cells, with enrichment at tricellular junctions where three cells meet (Figure 9, B and D', white arrowheads). During dorsal closure Ena also localizes to prominent dots at AJs of leading edge cells, where they meet the amnioserosa (Figure 9D', arrow), and Ena also accumulates at higher levels at the dorsal and ventral cell interfaces of the single row of epidermal cells that initiates the segmental groove (Figure 9D', blue arrowhead). In the amnioserosa, Ena outlines the apical ends of the cells (Figure 9, G, I, and L, red arrowheads).

View larger version (128K): [in this window] [in a new window]   Figure 9. Abl overexpression and Bcr-Abl misexpression alter Ena localization. Embryos, anterior left and, unless noted, lateral view. Antigens and transgenes indicated. (A–F) Wild-type or en-GAL4 x indicated transgene. A, stage 12. B, C, and F, stage 14. D and E, stage 13. Brackets, en-GAL4 expression domain. White arrowheads, tricellular junctions. Blue arrowheads, segmental groove cells. Blue arrows, ectopic cortical P-Abl and Ena. (G–N) Wild-type or e22c-GAL4 x indicated transgene. G and H, stage 12. G, arrowhead, tricellular junction. H, arrowhead, ectopic cortical Ena. I–N, stage 14 (K–M are dorsal views). Red arrowheads, amnioserosal cell cortex. Blue arrowheads, tricellular junctions or ectopic cortical Ena. Arrows, Ena at leading edge. Bars, 10 µm.

To determine whether the changes we observed were due to an alteration of Ena accumulation or simply a change in its localization, we examined the total levels of Ena in embryos ubiquitously expressing Bcr-Abl. No effects on Ena levels were seen (Figure 10A; data not shown). However, consistent with Ena being a target of Bcr-Abl, tyrosine-phosphorylation of Ena significantly increased (Figure 10B). Interestingly, a second tyrosine-phosphorylated protein of 120 kDa coimmunoprecipitates with Ena in extracts from embryos misexpressing p185 (Figure 10B); the identity of this protein is not known. In our previous work on embryos deficient for maternal and zygotic Abl, we saw reduced recruitment of the AJ proteins Arm and -catenin into junctions and reduced levels of their accumulation (Grevengoed et al., 2001). We thus also examined the levels and Tyr-phosphorylation of Arm in embryos overexpressing or misexpressing Bcr-Abl—no consistent differences were seen (Figure 10C). We also observed no consistent changes in the association of Arm with its junctional partner DE-cadherin, as assessed by coimmunoprecipitation (Figure 10C). In embryos misexpressing Bcr-Abl, levels of DE-cadherin or -catenin were also unchanged (Figure 10A).

View larger version (45K): [in this window] [in a new window]   Figure 10. Abl overexpression and Bcr-Abl expression alter Ena phosphorylation but not levels. In all cases the transgenes were ubiquitously misexpressed by using e22c-GAL4. (A) Embryonic extracts from wild-type embryos or embryos ubiquitously misexpressing the indicated transgene, immunoblotted with the antibodies indicated (Pnut is a loading control). P210K>R is a kinase-dead transgene, and p185Y>F has an alteration in the Grb2-binding site. (B) Ena was immunoprecipitated from wild-type or p185-expressing embryos (e22c-GAL4) and immunoblotted with antibodies to PTyr or Ena. (C) Arm was immunoprecipitated from wild-type, Abl-overexpressing, or p185-expressing embryos (e22c-GAL4) and immunoblotted with antibodies to PTyr, Arm, or DE-cad.

View larger version (147K): [in this window] [in a new window]   Figure 11. Abl activation and Rho. Embryos, anterior to the left, and, unless noted, lateral view. Antigens and transgenes indicated. (A–C) en-GAL4 x the Rho-activity probe-GFP (UAS-PKNG58AeGFP) in wild-type or embryos expressing the indicated transgene. A and B, stage 15. C, stage 13. (D and E) Rho localization, stage 13. (F–I) DE-cad and Ena (F and G) or active Abl (H and I) localization in embryos expressing DN-Rho x en-GAL4. Bars, 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Deregulation of Abl Kinase Activity Disrupts Morphogenesis
Loss-of-function mutations in abl disrupt many morphogenetic events (see Introduction). To further our mechanistic understanding of the roles of Abl, we explored whether deregulated kinase activity disrupts morphogenesis. Inappropriate activation of Abl affects many of the same morphogenetic events disrupted by loss of Abl. Normally, Abl is likely to exist primarily in an inactive form (Nagar et al., 2003). Docking with ligands for the SH2 or SH3 domains may help trigger the active conformation. Embryos are relatively resistant to overexpression of wild-type Abl. We suspect that increasing protein levels are largely accommodated by normal regulatory mechanisms until levels become extremely high. Consistent with this, we found that active Abl has a more restricted localization than total Abl, suggesting that Abl activation is normally restricted to the apical cell cortex and in particular to tricellular junctions, with a pool of inactive Abl in the cytoplasm. Increasing wild-type Abl levels may drive formation of more active Abl, or it may titrate negative regulators.

Misexpression of Bcr-Abl has more drastic consequences on morphogenesis, consistent with the