Fat and Wingless signaling oppositely regulate epithelial cell-cell adhesion and distal wing development in Drosophila

Development of characteristic size and shape of an organ demands strict spatial regulation of its tissue growth and cell-cell adhesion. Thus,signaling pathways recruited for growth and pattern of an organ primordium also impact on its cell-cell adhesion(Vleminckx and Kemler, 1999). Imaginal discs, the primordia of the adult appendages of Drosophila,offer elegant models with which to investigate the cellular mechanisms that determine the developmental outcome of the signaling pathways regulating organ growth and pattern (Day and Lawrence,2000). Evolutionarily conserved Wg/Wnt signaling pathway(Logan and Nusse, 2004; Peifer and Polakis, 2000)plays a crucial role in the patterning of Drosophila wing(Martinez Arias, 2003), and its role in the regulation of its growth is now well established(Giraldez and Cohen, 2003; Johnston and Sanders, 2003; Neumann and Cohen, 1996). The wing imaginal disc epithelium is also postulated to develop a gradient of cell affinities (Ripoll et al.,1988) and perturbations in Wg signaling affect its cell-cell adhesion (Blair, 1994; Chen and Struhl, 1999; Hoffmans and Basler, 2004). Besides growth and pattern, Wg signaling may thus regulate cell-cell adhesion in wing imaginal disc epithelium.

The Wg/Wnt ligand signals through its receptor Frizzled (Fz), the seven-pass trans-membrane receptor that acts through Dishevelled (Dsh) to antagonize the activity of GSK3/Shaggy, a component of the β-catenin/Arm degradation complex, thereby leading to stabilization of β-catenin/Arm in the cytoplasm. β-Catenin/Arm then translocates to the nucleus where it interacts with the transcription factor TCF/Lef1, known as Pangolin(previously DTcf) in Drosophila, and activates nuclear targets (for a review, see Logan and Nusse,2004). β-Catenin/Arm is also a component of the epithelial cell-cell adhesion complex at adherens junction (AJ). DE-Cadherin (DE-Cad; Shg– FlyBase), the Drosophila homolog of vertebrate E-Cadherin,assembles the AJ complex, as a sub-apical ring around the neck of the epithelial cells and mediates cell-cell adhesion by Ca²⁺ dependent homophilic interactions (Oda et al.,1994). Cytoplasmic domain of DE/E-Cad binds to β-catenin/Arm which in turn associates with the actin cytoskeleton through α-catenin(Gumbiner, 2005; Nelson and Nusse, 2004). Stabilization of the cytoplasmic pool of β-catenin/Arm in response to Wg signaling may influence cell-cell adhesion by facilitating its binding to DE-Cad at AJs (Gumbiner, 2000; Nelson and Nusse, 2004). In mammalian cultured cells, for example, Wnt signaling upregulates the levels of junctional E-Cad (Hinck et al.,1994). However, enhanced junctional recruitment ofβ-catenin/Arm may not be the primary mode of Wg-mediated regulation of cell-cell adhesion as the levels of Cadherin expression, rather than ofβ-catenin, appear to be the rate-limiting factor for AJ complex formation(Gumbiner, 2000). Wg signaling may affect cell-cell adhesion by regulating DE/E-Cad, which itself is a target of Wg. Transcriptional regulation of DE-Cad by Wg signaling enhances cell-cell adhesion in Drosophila cell lines(Yanagawa et al., 1997). By contrast, in mammalian hair follicles(Jamora et al., 2003) or in mouse brain (Shimamura et al.,1994), Wnt signaling represses E-Cad transcription. More strikingly, during zebrafish gastrulation, Wnt11 modulates E-Cad-mediated cell cohesion by regulating its endocytosis by Rab5c(Ulrich et al., 2005). Wnt/Wg signaling thus modulates cell-cell adhesion by regulating DE/E-Cad in a variety of ways.

The Drosophila tumor suppressor Fat (Ft)(Bryant et al., 1988) also regulates epithelial cell-cell adhesion in imaginal discs. Ft is an enormous transmembrane protein with 34 extracellular Cadherin repeats(Mahoney et al., 1991) and a prototype of a class of Cadherins called `Fat-like Cadherins'(Tanoue and Takeichi, 2005). In the absence of Ft, imaginal discs display overproliferation and altered cell-cell adhesion (Agrawal et al.,1995; Fanto et al.,2003; Garoia et al.,2000; Mahoney et al.,1991; Shashidhara et al.,1999). Ft is localized in the apicolateral membrane, above the sub-apical AJ (Ma et al.,2003) and displays Ca²⁺-dependent heterophilic binding(Matakatsu and Blair, 2004)with another Fat-like Cadherin, Dachsous (Ds)(Clark et al., 1995). Ft also regulates cell-cell adhesion both in cell culture assays(Matakatsu and Blair, 2004)and in vivo (Ma et al., 2003; Strutt and Strutt, 2002), and is also known to be involved in an intercellular signaling pathway where its partners ds and four-jointed (fj) modulate its activity during PD patterning (Cho and Irvine, 2004; Rodriguez,2004; Strutt et al.,2004) and cell-cell adhesion(Strutt and Strutt, 2002). Finally, the Fj-Ft-Ds intercellular signaling is also implicated in the regulation of planar cell polarity (PCP), the orientation of epithelial cells orthogonal to their apicobasal axis (Casal et al., 2002; Ma et al.,2003; Rawls et al.,2002; Strutt and Strutt,2002; Yang et al.,2002). Ft, therefore, links tissue growth with cell-cell adhesion and PCP. However, the genetic pathways and cellular mechanisms of Ft-dependent regulation of growth and cell-cell adhesion have yet to be uncovered (for a review, see Tanoue and Takeichi,2005).

Several indirect observations suggest convergence of Ft and Wg signaling in the regulation of wing development. For example, ft genetically interacts with Wg signaling (Greaves et al., 1999), while both Ft (Cho and Irvine, 2004) and its partner Ds(Rodriguez, 2004) regulate Wg expression/signaling in the proximal wing. Ft and Wg signaling may thus coordinate tissue growth with cell-cell adhesion during wing development. Here, we show that Wg signaling positively regulates expression of DE-Cad and sets up a PD gradient of cell-cell adhesion and cell shape in the distal wing imaginal disc epithelium. Ft, by contrast, represses DE-Cad and regulates cell-cell adhesion and cell shape in distal wing, although it does not co-localize with DE-Cad. Ft also intersects Wg signaling by downregulating the cytoplasmic pool of β-catenin/Arm,downstream of its ligand Wg. Finally, Ft and Wg signaling exert opposing regulations on Distalless (Dll), Vestigial (Vg)(Neumann and Cohen, 1997; Zecca et al., 1996) and DE-Cad expression, thereby linking tissue growth with cell-cell adhesion during Drosophila wing development.

Following stocks were used: ft^fd(ft⁸)(Mahoney et al., 1991), ft⁴²² (Rawls et al.,2002), UAS-arm^S10, UAS-arm^S2(Pai et al., 1997), UAS-dsh (Neumann and Cohen, 1996), UAS-DE-Cad(Greaves et al., 1999), UAS-tcfΔN (van de Wetering et al., 1997), UAS-ft(Matakatsu and Blair, 2004), UAS-GPI::DFz2 (Cadigan et al.,1998), N23-Gal4(Prasad et al., 2003), vg-Gal4 (Simmonds et al.,1995). Q-vg-lacZ (Kim et al., 1996), BE-vg-lacZ(Williams et al., 1994) and fz^H15fz2^C1(Chen and Struhl, 1999). Other stocks are described in FlyBase.

Loss-of-function clones were generated by flp/FRT system(Xu and Rubin, 1993) and gain-of-function clones were generated using flip-out technique(Ito et al., 1997). Ectopic expression was induced by UAS/Gal4 system(Brand and Perrimon, 1993). All fly stocks were grown at 24±1°C. For experiments in the vg¹ genetic background, cultures were maintained at 22-23°C for a consistent phenotype. vg¹ is a strong hypomorphic allele (Williams et al.,1990). We noticed that at 25°C or above vg¹ discs show a mild expression of Wg in the DV boundary. At a lower temperature, however, Wg expression in the DV boundary was completely abolished. Unless specified, to generate clones, larval cultures of the desired genotypes were given heat shock (37°C) for 1 hour during 48-72 hours interval after egg laying. For experiments with N^tsmutants, larval culture was grown at 18°C. Culture was shifted to 30°C for 48 hours prior to dissection. Adult wings were mounted in DPX and images were acquired under Leica DMRA.

The antibodies used were anti-β-galactosidase (Sigma), anti-Arm,anti-Wg (DSHB), anti-DE-Cad (Oda et al.,1994), anti-Dll (Panganiban et al., 1995) and anti-Vg(Williams et al., 1993). Secondary antibodies (Alexa 488, Alexa 555, Alexa 633) were from Molecular Probes.

For staining of Actin filaments, fixed imaginal discs were incubated in a 1:1000 dilution of PBS/Phalloidin-Rhodamine (Sigma) for 15 minutes and washed in PBS. Stained discs were mounted in Vectashield mounting medium with DAPI(Vector Labs). Images were acquired under Leica-SP2 Confocal microscope and processed using Leica Confocal software and Adobe Photoshop.

Cells of the dorsoventral (DV) boundary in the wing imaginal disc synthesize Wg (arrow in Fig. 1A). The DV boundary marks the distal end of the growing appendage, while the future hinge region, displaying Wg expression in two concentric rings (arrowheads in Fig. 1A), marks the proximal wing(Cho and Irvine, 2004; Neumann and Cohen, 1996). The lacZ reporter of the quadrant enhancer of vestigial(vg), Q-vg-lacZ (Kim et al., 1996) marks the entire distal wing [i.e. the presumptive wing blade (pouch) (Fig. 1A)].

In optical sections of the imaginal disc epithelium, AJs are visualized in the XY (Fig. 1B) or XZ planes(Fig. 1C) based on immunolocalization of DE-Cad and β-catenin/Arm besides binding with fluorochrome conjugated Phalloidin to F-actin. Both β-catenin/Arm(Collins and Treisman, 2000; Jiang and Struhl, 1998; Prasad et al., 2003) and DE-Cad display characteristic upregulation across the DV boundary along the PD axis of wing imaginal disc (Fig. 1D-D′′). Optical sections along the XY plane reveal higher levels of DE-Cad localization and narrower apical circumferences in the AJs of cells flanking the DV boundary when compared with those of the more proximally located cells (Fig. 1E) (see Fristrom and Fristrom, 1993). Optical sections along the XZ plane further confirmed upregulation of DE-Cad (Fig. 1F) and F-actin (Fig. 1F′) in the AJs of cells flanking the DV boundary (blue in Fig. 1F′′). Thus,along the PD axis of the wing disc, cell shapes and DE-Cad levels are graded.

We further sought to test if the PD gradient of cell shape and DE-Cad levels are linked to Wg signaling. Somatic clones displaying constitutive Wg signaling [induced by overexpression of Dsh(Yanagawa et al., 1997) or of a degradation resistant variant of β-catenin/Arm, Arm^S10(Pai et al., 1997) (not shown here)] induced cell-autonomous upregulation in the levels of DE-Cad and apical cell constrictions (Fig. 2A,A′ and inset in Fig. 2A). Somatic clones expressing secreted Wg, however, are expected to induce non-cell-autonomous effects. Indeed, these clones induced non-cell autonomous and graded upregulation in the levels of DE-Cad in the AJs(Fig, 2B,B′) and changes in apical cell shapes (Fig. 2C,C′). In the presumptive hinge region(Fig. 2D,D′), Wg overexpression produced a more striking pattern of non-cell autonomous changes in cell shapes: cells neighboring the Wg-expressing cells appeared to organize as whorls around the former and displayed epithelial misfolding(Fig. 2D,D′).

Furthermore, expression of GPI-anchored DFz2 receptor GPI-DFz2, which compromises Wg signaling (Cadigan et al.,1998), obliterated the characteristic PD gradient in the levels of DE-Cad (Fig. 2E,E′) and F-actin in the AJs (Fig. 2F-G′). Finally, loss of Wg expression in the DV boundary of wing imaginal disc of N^ts mutants grown at a restricted temperature (Fig. 2H)(Diaz-Benjumea and Cohen,1995) also abolished the PD gradient of DE-Cad(Fig. 2I) and apical cell shapes (Fig. 2I′, see Fig. 1D′′,E). To further test if apical cell constrictions are linked to elevated levels of DE-Cad in AJs (Fig. 1E, 2A-D′), we expressed DE-Cad in somatic clones. These clones were apically constricted (not shown here), consistent with the role of DE-Cad/E-Cad in remodeling cell shape and tissue architecture (Pilot and Lecuit,2005). These results thus link Wg signaling to the PD gradient in the levels of DE-Cad and apical cell shapes in the wing imaginal discs.

Somatic clones with altered cell-cell adhesion sort out from their neighbors and display smooth clone borders(Dahmann and Basler, 2000). Indeed, somatic clones displaying gain of Wg signaling owing to Dsh(Fig. 2J) or Arm^S10(Fig. 2K) misexpression sorted out from their neighbors and displayed smooth clone borders, akin to those misexpressing DE-Cad (Fig. 2L). Wg signaling may alter cell-cell adhesion by enhancing recruitment ofβ-catenin/Arm to the AJs (Hinck et al., 1994) and/or by its transcriptional input(Nelson and Nusse, 2004). In many cell types, for example, expression of cadherins rather than the levels of catenins appears to be the rate-limiting step of Catenin-Cadherin complex formation at AJs and cell-cell adhesion(Gumbiner, 2000). Wild typeβ-catenin/Arm (Arm^S2), when overexpressed, does not transduce Wg signaling (Pai et al.,1997). Somatic clones overexpressing Arm^S2 display`wiggly' clone borders (Fig. 2M), unlike those expressing Dsh(Fig. 2J) or Arm^S10(Fig. 2K). Thus, expression ofβ-catenin/Arm alone, without a concomitant enhancement of Wg signaling,fails to alter cell-cell adhesion. Cell-cell adhesion in wing imaginal disc epithelium is therefore likely to be regulated by transcriptional input from Wg signaling.

To test if canonical Wg signaling regulates DE-Cad expression, we examined the response of its lacZ reporter, DE-Cad-lacZ. Cells receiving high threshold of Wg signaling in the wing imaginal discs, as in those flanking the DV boundary, displayed higher levels of DE-Cad-lacZ reporter activity when compared with those further away from the source of Wg expression (Fig. 3A,A′). Furthermore, somatic clones expressing Arm^S10 (Fig. 3B,B′) or Dsh (not shown here) displayed cell-autonomous activation of the DE-Cad-lacZ. Finally, clones expressing the secreted Wg induced non-cell-autonomous activation of DE-Cad-lacZ:i.e. in cells within (Fig. 3C,C′) and surrounding the clones (arrows in Fig. 3C,C′). Together(Figs 1, 2, 3), these results suggest that regulation of DE-Cad by the long-range activity of the Wg morphogen(Neumann and Cohen, 1997; Zecca et al., 1996) sets up the PD gradient of cell-cell adhesion and cell shape in the distal wing.

Somatic clones lacking Ft (ft^-/ft^-), marked by loss of GFP (arrows in Fig. 4A), displayed overgrowth and altered cell-cell adhesion with characteristic circular and smooth clone borders, unlike the `wiggly' borders of their wild type (ft⁺/ft⁺) twins that are marked by brighter GFP (arrowheads in Fig. 4A). Furthermore, cells lacking Ft displayed upregulation of DE-Cad in their AJs (Fig. 4B-B′′) and DE-Cad-lacZ(Fig. 4C,C′). By contrast, when we overexpressed Ft, levels of both DE-Cad(Fig. 4D,E′) or DE-Cad-lacZ (not shown here) were downregulated. Besides, following overexpression of Ft in the posterior wing compartment, cells flanking the DV boundary (arrowheads, Fig. 4E,E′) displayed wider apical circumferences when compared with those of the anterior wing compartment. These results suggest that Ft regulates DE-Cad expression, cell-cell adhesion and apical cell shapes in the distal wing.

Dynamic regulations of Wg expression controls both PD wing growth and pattern (Diaz-Benjumea and Cohen,1995; Klein and Arias,1998; Neumann and Cohen,1996; Rodriguez et al.,2002). Ft (Cho and Irvine,2004) and its partner Ds(Rodriguez, 2004) both regulate proximal wing growth and pattern by modulating Wg expression/signaling. Furthermore, opposing regulation of DE-Cad by Ft and Wg signaling (Figs 2, 3) suggests that these two pathways might intersect during distal wing development. We sought to examine the role of Ft in the regulation of two distal wing markers, namely Dll(Cohen et al., 1989; Diaz-Benjumea and Cohen, 1995)and Vg or its pouch specific enhancer Q-vg-lacZ(Kim et al., 1996; Williams et al., 1994).

In wild type wing imaginal discs, Dll(Fig. 5A) and Vg (data not shown) display a characteristic PD gradient in their expression(Neumann and Cohen, 1997; Zecca et al., 1996). In ft mutant wing imaginal discs, expressions of both Dll(Fig. 5B) and Q-vg-lacZ (not shown here) are upregulated and expanded along the PD axis. By contrast, a strong overexpression of Ft in the posterior compartment of wing imaginal disc resulted in downregulation of Dll(Fig. 5C,C′; larvae grown at 26-27°C; also compare with Fig. 8H,H′) and Vg (data not shown). Furthermore, in mosaic wing imaginal disc, Q-vg-lacZ displayed cell autonomous upregulation in cells lacking Ft throughout the PD axis(Fig. 5D,D′). Plot of the intensity of Q-vg-lacZ expression in the mosaic discs revealed an upregulated and flatter gradient of Q-vg-lacZ activity in cells lacking Ft when compared with the wild-type(ft⁺/ft^-) domains in the mosaic wing pouch(Fig. 5E). Ft thus negatively regulates Dll and Vg expression suggesting its role in distal wing development. Furthermore, overexpression of Ft resulted in truncated growth along the PD axis (Fig. 5F,G).

Mutation in ft had earlier been shown to modify the phenotypes induced by gain or loss of β-catenin/Arm(Greaves et al., 1999). We sought to further confirm genetic interaction of ft with Wg signaling by reducing the dosage of ft (ft^-/ft⁺)in genetic backgrounds where Wg signaling is altered. High threshold of Wg signaling is essential for margin bristle formation(Couso et al., 1994; Neumann and Cohen, 1997). Overexpression of Arm^S10 along the DV boundary induced ectopic wing margin bristles (22-40 bristles, n=25)(Fig. 6A). This phenotype was enhanced by a reduction in the dose of ft (45-87 bristles, n=25) (Fig. 6B). Similarly, the ectopic wing margin bristle phenotype induced by overexpression of the Wg receptor Fz2 was also enhanced by a reduction in the dosage of ft (not shown here). Conversely, compromising Wg signaling by overexpression of dTCFΔN (van de Wetering et al., 1997) abolished the wing margin bristles(Fig. 6C, n>100),while overexpression of Dfz2-GPI (Cadigan et al., 1998) induced notching of wing margin (not shown here). These, phenotypes were partially suppressed by a reduction in the ftgene dose (dTCFΔN: Fig. 6D, n>100; Dfz2-GPI: not shown here).

These results suggest that a reduction in Ft activity enhances Wg signaling. To further test the consequences of complete loss of Ft on Wg signaling, we examined the levels of β-catenin/Arm in somatic clones lacking Ft. Upregulation of β-catenin/Arm signifies stabilization of its cytoplasmic pool, a characteristic of cells receiving Wg signaling(Peifer et al., 1994). In the wing imaginal discs, for example, cells flanking the DV boundary show higher levels of cytoplasmic β-catenin/Arm(Fig. 1D′) in response to Wg signaling (Collins and Treisman,2000; Jiang and Struhl,1998; Prasad et al.,2003). Somatic clones lacking Ft displayed upregulated levels ofβ-catenin/Arm (Fig. 6E-F′). Conversely, overexpression of Ft(en-Gal4xUAS-Ft) downregulated β-catenin/Arm levels(Fig. 6G,H). Taken together with the earlier studies of genetic interactions(Greaves et al., 1999), these results show that Ft modulates Wg signaling, presumably by alteringβ-catenin/Arm levels.

Expression of Wg in the DV boundary (distal wing) in wild type (arrowhead in Fig. 7A) and in ftmutant discs (Fig. 7B) was comparable. In the hinge (proximal wing), however, Wg expression was upregulated in ft mutant discs (arrows in Fig. 7B)(Cho and Irvine, 2004). Furthermore, loss of Ft in somatic clones upregulated Wg in the proximal but not in the distal wing (Fig. 7C,C′). Wg expressions in the distal(Diaz-Benjumea and Cohen,1995) and proximal wing domains are maintained by different mechanisms (by a positive auto-regulatory loop in the latter)(Rodriguez et al., 2002). In accordance with this interpretation, we could further show that constitutive gain of Wg signaling upregulates expression of its ligand, Wg in the proximal but not in the distal wing (data not shown). These results suggest that upregulation of Wg expression in the proximal wing of ft mutants could be due to a positive feedback from enhanced Wg signaling: for example,by upregulation in the levels of β-catenin/Arm (see Fig. 6).

Ft is proposed to regulate growth in the entire wing by a broad based mechanism (Cho and Irvine,2004). As Wg synthesis is not upregulated in the DV boundary of overgrown ft mutant wing imaginal discs(Fig. 7) (see also Cho and Irvine, 2004),regulation of distal wing development by Ft could be independent of the activity of the Wg ligand. We sought to examine this possibility.

During wing development, Wg and Vg maintain their mutual expression in the DV boundary (Prasad et al.,2003). In vg¹ wing imaginal discs, synthesis of the Wg ligand in the DV boundary is abolished (arrowhead in Fig. 8A), besides those of Q-vg-lacZ (Fig. 8A)(Prasad et al., 2003) or Dll(Fig. 8E). These distal wing markers are activated in vg¹ wing imaginal discs under conditions that circumvent the loss of the endogenous Wg ligand activity: for example, by expressing the Wg ligand from a transgene(Fig. 8B) or by overexpression of Dsh (not shown here). Expression of Q-vg-lacZ or Dll in the vg¹ wing imaginal disc thus signifies activation of the Wg signaling pathway, independent of the activity of endogenous Wg ligand.

Loss of Ft in vg¹ discs in somatic clones showed cell-autonomous activation of Q-vg-lacZ; expression of Wg in these clones, however, remained extinguished(Fig. 8C,C′). In ft^fd vg¹ double mutant discs,Q-vg-lacZ (Fig. 8D′) and Dll (Fig. 8F) were activated, while Wg in the distal wing remained extinguished (arrow, Fig. 8D-F). Levels of expression of Q-vg-lacZ(Fig. 8D′) and Dll(Fig. 8F) appeared milder in ft^fd vg¹ double mutants than in their respective wild-type counterparts (see Fig. 1A and Fig. 8H′). Furthermore, between these two markers, Q-vg-lacZ (Fig. 8D′) displayed stronger level of derepression than that of Dll (Fig. 8F) in ft^fd vg¹ double mutant discs.

We carried out additional tests in genetic backgrounds that abolished/undermined the activities of the Wg ligand in the distal wing. Notch(N) regulates Wg expression in the DV boundary(Diaz-Benjumea and Cohen,1995). At a restrictive temperature, Wg expression in the DV boundary (Fig. 8G,G′) of N^ts mutant wing imaginal disc is extinguished(Fig. 2F); in addition,expression of Vg or its boundary and quadrant enhancers is also silenced in N^ts mutants (not shown here). N^ts/Y; ft^fd/ft⁴²² double mutant larvae grown at restrictive temperature, however, showed distal wing growth and activation of Q-vg-lacZ (Fig. 8G′) but not that of Wg (arrow, Fig. 8G). In wild-type wing disc, expression of Dll is seen to be nearly symmetrical on either side of the AP boundary (Fig. 8H,H′)and flanking the DV boundary (Fig. 5A). Expression of the GPI-DFz2 in the wild-type discs downregulates Wg ligand activity (Cadigan et al., 1998) and reduced the level of expression of Dll(Fig. 8I,I′). Expression of Vg was also repressed following expression of GPI-DFz2 (data not shown). In the ft^fd mutant wing discs, however, expression of GPI-Dfz2 largely failed to repress Dll(Fig. 8J,J′) or Vg (data not shown). Taken together, these results show that Ft intersects Wg signaling downstream of its ligand and upstream of the cytoplasmic stabilization ofβ-catenin/Arm.

We further sought to test if Fz and Fz2 receptors are required to mediate Ft-dependent growth and target gene regulation. In Drosophila, Fz and Fz2 act as redundant receptors of Wg(Bhanot et al., 1999; Chen and Struhl, 1999). In the complete absence of Fz activity (Fz null), which is abolished when both fz and fz2 are mutated, cells of the wing pouch fail to grow and survive (Fig. 8K)(Chen and Struhl, 1999). However, when induced late during larval development, these Fz-null mutant clones displayed modest survival but failed to express Dll(Chen and Struhl, 1999). Activation of the downstream components of Wg signaling, however, rescues the Fz-null embryos from growth and pattern defects(Chen and Struhl, 1999). We argued that if Ft signaling were Fz independent, then Fz-null clones would survive when induced in ft mutant discs. However, Fz null clones induced in ft mutant discs during early larval stages failed to survive (Fig. 8K,L). However,Fz-null clones induced 24 hours before dissection survived but appeared undergrown by comparison with their wild-type twins and also failed to activate Dll expression (Fig. 8M,M′, n=10). These results suggest that Ft intersects Wg signaling either upstream or parallel to its receptors, Fz/Fz2,reminiscent of its interaction with the non-canonical Wg signaling pathway for the regulation of PCP (Ma et al.,2003; Strutt and Strutt,2002; Yang et al.,2002).

The levels of Cadherin expression influence the overall strength of cell-cell adhesion (for a review, see Gumbiner, 2005). Linking developmental regulation of Cadherins to signaling pathways, implicated in the control of organ growth and pattern, would ensure that cells within a regional specialization do not migrate and expand in inappropriate places(Gumbiner, 2005; Vleminckx and Kemler, 1999). Here, we have shown that apical cell shapes and cell-cell adhesion are both graded along the PD axis of the developing wing imaginal discs. Furthermore,readout of Wg signaling along the PD axis regulates expression of DE-Cad. As DE-Cad levels affect cell-cell adhesion(Fig. 2) (see Dahmann and Basler, 2000), our results suggest that by regulating DE-Cad expression, Wg signaling integrates cell-cell adhesion with tissue growth and pattern. Regulation of DE-Cad expression could be a prevalent mechanism for coordination of the emerging pattern in an organ primordium with the spatial control of its cell-cell adhesion. For example, DE-Cad levels are also upregulated in cells flanking the stripe of cells, along the AP boundary (Figs 1 and 3) which express the morphogen Decapentaplegic (Dpp) (Lecuit et al.,1996; Nellen et al.,1996); misregulation of Dpp signaling also affects DE-Cadexpression (M.J. and P.S., unpublished). The Ft tumor suppressor, by contrast,negatively regulates DE-Cad expression in the distal wing. This may also explain the inverse correlation between the levels of DE-Cad in AJs and the activity of Ft (Fig. 4). Thus, besides its heterophilic binding with Ds(Matakatsu and Blair, 2004; Strutt and Strutt, 2002), Ft controls cell-cell adhesions at AJs by regulating DE-Cadexpression.

Cell-cell adhesion apart, DE/E-Cad regulation may impact a variety of other cellular processes and developmental mechanisms(Gumbiner, 2005). E-Cad has been shown to mark the sites of actin assembly on cell surface(Kovacs et al., 2002). Cadherin complexes regulate cytoskeletal networks and cell polarity(Harris and Peifer, 2004),while disruption of AJ associated components affect asymmetric cell division(Lu et al., 2001). Fat1, a mammalian homolog of Drosophila Ft, modulates actin dynamics(Moeller et al., 2004; Tanoue and Takeichi, 2004). Interestingly, Ft also regulates orientated cell division (OCD) in imaginal epithelium, which is mirrored by orientation of the spindles of the dividing cells; OCD may also regulate organ shape along the PD axis(Baena-Lopez et al., 2005). Misregulation of DE-Cad may thus affect the cytoskeleton and produce OCD phenotype in ft mutant discs.

In both loss- and gain-of-function (Figs 5, 8 and data not shown) assays,we could show that Ft downregulates Dll and Vg/Q-vg-lacZ(Fig. 5, data not shown) in the distal wing. Although Vg/Q-vg-lacZ and Dll have not been ascertained to be the direct targets of Wg, all available evidence(Klein and Arias, 1999; Neumann and Cohen, 1997; Prasad et al., 2003; Zecca et al., 1996) so far suggests that these targets positively respond to Wg signaling(Schweizer et al., 2003). These results also show that Ft and Wg signaling intersect and control distal wing growth and pattern, presumably through their opposing regulation of a common set of targets, namely, DE-Cad, Vg and Dll. Apart from Wg signaling,Dpp signaling also regulates Q-vg-lacZ(Kim et al., 1996); however,its long-range target, Omb (Lecuit et al.,1996; Nellen et al.,1996) is not upregulated in ft mutant clones (not shown here) suggesting that regulation of distal wing targets by Ft is mediated by its intersection with Wg signaling.

Our results show that Ft negatively regulates Wg signaling. Loss or gain of Ft induces a telltale sign of perturbations in Wg signaling, namely, changes in the cellular pool of β-catenin/Arm(Collins and Treisman, 2000; Jiang and Struhl, 1998; Peifer et al., 1994; Prasad et al., 2003),consistent with its role as a suppressor of Wg signaling in genetic tests(Fig. 6). Our results further reveal intersection of Ft with Wg signaling downstream of the Wg ligand(Fig. 8), while with respect to its receptor, Ft is likely to act either upstream of or parallel to Fz/Fz2(Fig. 8). It is interesting to note here that the role of Ft in PCP regulation has also been suggested to be either parallel to or upstream of the Fz receptor(Ma et al., 2003; Strutt and Strutt, 2002; Yang et al., 2002). We also note that Ft co-localizes with neither Fz(Ma et al., 2003) nor Fz2 (not shown here) and does not mediate their subcellular localization (not shown here), thereby suggesting that Ft interacts with Fz indirectly. Unraveling the genetic and molecular basis of this interaction may explain how Ft straddles both the canonical (growth and cell-cell adhesion) and non-canonical (PCP)(Veeman et al., 2003) Wnt signaling pathways.

One of the remarkable aspects of development of an organ primordium is that a stereotypic PCP is achieved even while it passes through dynamic changes in its size and shape (Day and Lawrence,2000; Eaton,2003). The fact that changing organ sizes/shapes does not alter PCP suggests an in-built mechanism to regulate constancy of PCP during animal development (Eaton, 2003). A link between PCP and growth through the activity of Ft has been speculated, as it regulates both (Lawrence,2004). Intersection of Ft and the canonical Wg signaling seen here might provide a mechanism to coordinate PCP and organ growth.

Drosophila wing growth is under dynamic spatial and temporal regulation by Wg signaling (Giraldez and Cohen, 2003; Johnston and Sanders, 2003; Martinez Arias,2003; Neumann and Cohen,1996). Furthermore, different thresholds of Wg signaling impact cell proliferation in their characteristic ways and activate distinct sets of PD markers. Although at a very high threshold, Wg signaling inhibits cell proliferation (Johnston and Sanders,2003; Neumann and Cohen,1996), at a modest threshold it has been shown to stimulate growth(Giraldez and Cohen, 2003; Neumann and Cohen, 1996). We note that loss of Ft fail to activate Wg targets that demand a high threshold of its signaling, e.g. Ac (not shown here), which is required for wing margin specific bristle development (Couso et al., 1994; Neumann and Cohen,1997). Conversely, overexpression of Ft also did not lead to loss of margin bristles (Fig. 5),suggesting that it is not a strong repressor of Wg signaling either. The short-range Wg target, fz3-lacZ(Sato et al., 1999; Sivasankaran et al., 2000),which responds to a high threshold of Wg signaling, is also not upregulated by loss of Ft (not shown here). Dll responds to a higher threshold of Wg signaling than that required for Vg/Q-vg(Neumann and Cohen, 1997; Zecca et al., 1996). Dll and Vg display modest and strong upregulation (Figs 5 and 8), respectively, following loss of Ft. These results suggest that loss of Ft upregulates Wg signaling to only modest thresholds, consistent with the growth-promoting effect of the latter (Neumann and Cohen,1996).

Over-proliferation in ft mutant imaginal discs is induced by perturbation of as yet unidentified disc-intrinsic mechanisms that determine their characteristic final sizes (Bryant et al., 1988; Cho and Irvine,2004). The imaginal discs of ft mutants continue to grow and the extent of their over-proliferation appears to be constrained only by the developmental time available during the extended periods of their larval life (Bryant et al., 1988). By contrast, growth in wild-type imaginal discs is determinate, which ceases after they attain their predetermined sizes even under conditions of unlimited developmental time; for example, on transplantation into wild-type adult host abdomen that can sustain development(Bryant, 1975; Bryant et al., 1988). ft mutant imaginal discs thus acquire unlimited proliferative potential, akin to immortalization, a crucial step during tumorigenesis(Hanahan and Weinberg, 2000). It is significant that the Ft tumor suppressor downregulates Wg/Wnt signaling,a pathway implicated in cancers (Logan and Nusse, 2004; Peifer and Polakis, 2000). Several orthologs of Ft have been identified in vertebrates with diverse functions (Tanoue and Takeichi, 2005). It will thus be interesting to explore if these orthologs of Ft in higher vertebrates also interact with Wnt signaling and thereby behave as tumor suppressors.

We thank L. S. Shashidhara, S. Cohen, S. Carroll, T. Wolf, G. Struhl, J. P. Vincent, Mark Peifer, S. S. Blair, A. Tomlinson and the Bloomington Stock Center for stocks; G. Panganiban H. Oda, S. Cohen and DSHB for antibodies; and Larry Marsh, K. VijayRaghavan, Priya Srivastava, Subhabrata Pal and Hina Patel for discussion and technical help. Confocal facility established from the MPLADS Funds to Mr Arun Shourie is thankfully acknowledged. This work wassupported by the financial assistance from the Department of Biotechnology,New Delhi (Project No. BT/PR416/MED/12/164/2003) and Indo-French Center, New Delhi to P.S. M.J. is a recipient of research assistantship from IIT Kanpur.