Regulation of Wingless and Vestigial expression in wing and haltere discs of Drosophila

In the fruitfly Drosophila melanogaster, wings and halteres are the dorsal appendages of the second and third thoracic segments, respectively. In the third thoracic segment, wing development is suppressed by the homeotic selector gene Ultrabithorax (Ubx) in order to mediate haltere development (Lewis,1978). Loss of Ubx function from developing haltere discs induces haltere-to-wing transformations, whereas ectopic expression ofUbx in developing wing discs leads to wing-to-haltere transformations(Lewis, 1978;Cabrera et al., 1985;White and Akam, 1985). The differential development of wings and halteres thus constitutes a good genetic system with which to study cell fate determination. They also give insight into the evolutionary trend that has established the differences between fore and hind wings in insects, wings and legs in birds and fore- and hindlimbs in mammals.

Growth and patterning during fly wing development are mediated by signaling from the dorsoventral (DV) organizer. Interactions between dorsal and ventral cells of the wing pouch set up the organizer by activating Notch (N) at the DV boundary (Diaz-Benjumea and Cohen,1993; Diaz-Benjumea and Cohen,1995; Williams et al.,1994; Irvine and Wieschaus,1994; Kim et al.,1995; de Celis et al.,1996b). N, in turn, activates Wingless (Wg), Cut (Ct) and Vestigial (Vg) at the DV boundary (Couso et al., 1995; Kim et al.,1995; Rulifson and Blair,1995; Kim et al.,1996; Neumann and Cohen,1996). Wg is known to diffuse to non-DV cells from the DV boundary to act as a morphogen (Zecca et al.,1996; Neumann and Cohen,1997). High levels of Wg are required for activating Achaete (Ac),whereas moderate levels are sufficient to activate Distal-less (Dll) and low levels to activate Vg (Neumann and Cohen,1997). Thus, Vg is expressed in both DV and non-DV cells. It has been shown that two different promoters regulate Vg expression in DV and non-DV cells (Kim et al.,1996). They are vg-boundary enhancer (vg-BE) andvg-quadrant enhancer (vg-QE).

Previously, we have shown that Ubx downregulates DV signaling to specify haltere fate (Shashidhara et al.,1999). In haltere imaginal discs, Wg and Ct are expressed only in the anterior compartment (Weatherbee et al., 1998; Shashidhara et al.,1999). However, none of the three targets of Wg (i.e. Ac, Dll andvg-QE) is expressed in the haltere disc(Gorfinkiel et al., 1997;Weatherbee et al., 1998;Shashidhara et al., 1999). As expression of Wg itself is robust in the anterior DV boundary of haltere discs, downregulation of its targets, in this compartment at least, could be due to the repression of event(s) downstream of Wg, such as transduction of Wg signaling from the DV boundary. Consistently, although overexpression of Ubx in the wing disc DV boundary results in loss of Wg only in DV boundary cells of the posterior compartment, it causes loss of vg-QE in non-DV cells of both the anterior and posterior compartments(Shashidhara et al., 1999). We show that Ubx functions at multiple levels to repress Vg in non-DV cells,including enhanced degradation of Arm in the haltere pouch. Repression of Vg at multiple levels appears to be crucial for Ubx-mediated specification of the haltere fate. Overexpression of Vg in haltere discs overrides Ubx function and thereby induces haltere-to-wing homeotic transformations.

Recombinant chromosomes and combinations of GAL4 drivers, UAS lines,different mutations and/or markers were generated by standard genetic techniques. The GAL4-UAS system (Brand and Perrimon, 1993) was used for targeted misexpression of gene products. The FLP-FRT method (Xu and Rubin, 1993) was used for generating mitotic clones ofarm and vg. P[FRT]18A arm^H8.6 has been reported previously (Neumann and Cohen,1997); we recombined vg¹ to P[FRT]42 πMyc. Clones were generated with the help of hsFLP using either arm-lacZ orUbi-GFP as clonal markers. The original second chromosomevg-quadrant enhancer-lacZ [vg-QE(Kim et al., 1996)] was mobilized to obtain first and third chromosome insertions by crossing to a genetic source of transposase. We selected new insertions that showed original expression patterns in all stages of wing development. UAS lines used in this study were UAS-Flu-ΔArm(Zecca et al., 1996),UAS-arm^S2 and UAS-arm^S10 [both of which are Myc tagged (Pai et al.,1997)], UAS-DN-TCF/pan (van der Wetering et al., 1997),UAS-Dsh (Neumann and Cohen,1996), UAS-APC/CBD(Bhandari and Shashidhara,2001), UAS-N^intra(Fortini et al., 1993),UAS-Ubx (Castelli-Gair et al.,1994), UAS-Vg (Kim et al., 1996), and UAS-Wg(Lawrence et al., 1995). GAL4 strains used were dpp-GAL4(Morimura et al., 1996),en-GAL4 (A. Brand, personal communication to FlyBase, 30 June 1997), omb-GAL4 (M. Calleja, personal communication to FlyBase, 16 October 1996) and vg-GAL4(Simmonds et al., 1995).N23-GAL4 was used to express genes of interest in non-DV cells of wing and haltere discs. This GAL4 line was identified in the lab in an enhancer-trap screen (Shashidhara et al.,1999). Although its activation in non-DV cells is dependent on N signaling in the DV boundary, it is not dependent on Wg or Vg (R.B. and L.S.S., unpublished).

X-gal and immunohistochemical staining was performed essentially as described by Ghysen and O'Kane (Ghysen and O'Kane, 1989) and Patel et al.(Patel et al., 1989),respectively. The primary antibodies used were, monoclonal anti-Arm(Riggleman et al., 1990),anti-Ct (Blochlinger et al.,1993), anti-Salm (de Celis et al., 19666a), anti-Wg(Brook and Cohen, 1996) and anti-β-galactosidase (Sigma) and polyclonal anti-Vg(Williams et al., 1991),anti-Arm (Ruel et al., 1999)and anti-β-galactosidase (Sigma). Monoclonal anti-Arm and anti-Wg antibodies were obtained from the Development Studies Hybridoma Bank,University of Iowa, USA. Confocal microscopy was carried out on Meridian Ultima. The adult appendages were processed for microscopy as described previously (Shashidhara et al.,1999).

We designed several experiments to test the working model of Wg and Vg regulation (which is essentially based on studies on wing imaginal discs) in haltere discs. However, information on certain aspects of Wg and Vg regulation in wing discs is limited. For example, autoregulation of Vg in non-DV cells is not well understood. Understanding these events was a pre-requisite to interpret the results related to the mechanism of Ubx-mediated repression of Wg and Vg in haltere discs.

A mutant version of TCF/pan protein, which lacks the N-terminal Arm interaction domain, functions as a dominant negative for both TCF/pan and Arm(van der Wetering et al., 1997). We overexpressed DN-TCF/pan usingvg-GAL4 to downregulate Wg signaling in the DV boundary. We observed loss of Vg in both DV and non-DV cells when Wg signaling is downregulated (monitored by anti-Vg antibody, vg-BE andvg-QE staining; Fig. 1D-F). This is contrary to the earlier reports that Wg activity is not required for the expression of Vg at the DV boundary(Neumann and Cohen, 1996). We further tested the cell-autonomy of this phenomenon by generating mitotic clones of arm. As loss-of-function clones of null alleles ofarm are lethal, we used arm^H8.6, a temperature-sensitive hypomorphic allele(Neumann and Cohen, 1997). We monitored Vg expression in small arm^H8.6 clones, survival of which were confirmed by DAPI staining. Clonal loss of Arm at the DV boundary resulted in cell-autonomous loss of Vg expression(Fig. 1G), confirming a role for Wg in the maintenance of Vg expression in DV cells. It has been reported previously that ectopic expression of Vg or Scalloped (Sd; a co-factor of Vg in the nucleus) causes ectopic Wg expression(Go et al., 1998;Klein and Martinez-Arias,1998; Klein and Martinez-Arias 1999; Liu et al.,2000). Thus, Wg and Vg may interact to maintain each other's expression in the wing disc DV boundary.

It has been shown that Sd binds to vg-QE and thus regulates its expression (Halder and Carroll,2001). Since Sd and Vg are known to physically interact(Simmonds et al., 1998;Halder et al., 1998), Vg may regulate its own expression in non-DV cells by modulating Sd function. To test the autoregulation of Vg, we generated mitotic clones of vg and examined the status of vg-QE. vg^– clones grow very slowly and often they are replaced by the neighboring cells(Kim et al., 1996). We monitored vg-QE expression in small vg¹ clones,the survival of which was confirmed by DAPI staining. Clonal loss of Vg resulted in loss of vg-QE expression(Fig. 2A), thus confirming autoregulation of Vg in non-DV cells.

To answer the question of whether Vg autoregulation is dependent on Wg, we overexpressed Vg in non-DV cells in the absence of endogenous Wg. Invg¹ wing discs, both Vg and Wg are absent at the DV boundary (Fig. 2B,C) and novg-QE expression is seen (data not shown). We overexpressed Vg using the N23-GAL4 driver, which is expressed only in non-DV cells of both wild-type (Fig. 2D) andvg¹ (Fig. 2E) wing discs. Overexpression of Vg in non-DV cells ofvg¹ wing discs was enough to rescue vg-QE expression (Fig. 2G) as well as adult wing phenotypes (Fig. 2I). Rescued wing discs did not show any Wg expression in the presumptive DV boundary (Fig. 2G). The absence of Wg is also reflected in the absence of margin bristles in the rescued adult wing blades(Fig. 2I). These results suggest that Vg in non-DV cells is necessary and sufficient to activate its quadrant enhancer.

Although Vg is capable of activating vg-QE in both wild-type andvg¹ backgrounds (Fig. 2G), ectopic expression of Wg or activated Arm does not induce ectopic vg-QE expression (Nagaraj et al., 1999). This is contrary to the non-cell autonomous loss of both Vg and vg-QE by the ectopic expression of DN-TCF/pan at the DV boundary (Fig. 1F), and cell-autonomous loss of Vg in arm^– mitotic clones generated in non-DV cells (Neumann and Cohen, 1997) (Fig. 1G).

Ectopic expression of activated N using dpp-GAL4 resulted in non-cell autonomous activation of Vg in the wing pouch(Fig. 3C). As N specifies DV boundary activity and vg-QE expression is inhibited in N-expressing cells (Klein and Martinez-Arias,1999), cell-autonomous activation of Vg might correspond to the activation of vg-BE and non-cell autonomous component might correspond to vg-QE. As, ectopic Wg also causes non cell-autonomous activation of Vg (Neumann and Cohen,1997), ectopic N might first cell-autonomously activate Wg, which in turn would activate Vg in neighboring cells. Consistent with this, ectopic N^intra-induced cell-autonomous activation of Wg(Fig. 3D) and activated Arm resulted in cell-autonomous activation of Vg in wing discs(Fig. 3E).

We therefore examined the status of vg-QE inarm^– clones. We observed downregulation ofvg-QE expression in arm^– clones(Fig. 3F). Thus, Wg signaling is required, but is not sufficient to activate vg-QE. As Vg alone was sufficient to activate vg-QE, Wg signaling might activate Vg either indirectly or by activating some other enhancer of Vg (see Discussion). Once Vg is activated, it maintains its own expression by autoregulation, which is mediated through its quadrant enhancer.

With the new insights into the mechanism of Wg and Vg expression in wing discs, we studied the mechanism by which Ubx represses their expression in haltere discs. Wing and haltere discs employ similar genetic pathways for pattern formation along the A/P and DV axes(Williams et al., 1993;Williams et al., 1994). However, although Wg is expressed at the anterior DV boundary, Vg is not expressed in non-DV cells of haltere discs(Weatherbee et al., 1998;Shashidhara et al., 1999). Thus, Ubx may repress event(s) downstream of Wg to inhibit Vg expression in non-DV cells.

Stabilization of cytoplasmic Arm is a key step in the transduction of Wg signaling. Although Arm is present in all cells, cytoplasmic levels of Arm,which transduces Wg signaling, are higher only in cells in which Wg signaling is active (Peifer et al.,1994). For example, cells immediately adjacent to the wing disc DV boundary show higher levels of Arm than do non-DV cells(Fig. 4A,B). In the absence of Wg signaling, cytoplasmic Arm is subjected to Ubiquitin-mediated degradation.

In haltere discs, Arm levels are uniform in the entire pouch(Fig. 4C,D). In particular, we did not observe increased levels of Arm in cells surrounding the DV boundary. This is true for both the anterior compartment (in which Wg is expressed at the DV boundary) and the posterior compartment, which suggests that Ubx interferes with Arm stabilization. Interestingly, cells neighboring Wg-expressing hinge cells showed increased levels of Arm(Fig. 4C,D), similar to those in wing discs (Fig. 4A,B). These observations suggest that Ubx inhibits the stabilization of Arm specifically to downregulate DV signaling during haltere development. This is further supported by the observation that misexpression of Ubx at the wing disc DV boundary causes downregulation of Arm(Fig. 4E). In both anterior and posterior compartments there was a severe reduction in Arm levels, although Wg was suppressed only in the posterior compartment(Shashidhara et al.,1999).

To further test if Arm degradation is enhanced in haltere discs, we used Myc-tagged degradation-resistant and degradation-sensitive forms of Arm[Arm^S10 and Arm^S2(Pai et al., 1997)]. Arm^S10 has an internal deletion of 43-87 residues at the N terminus. This deletion removes residues that are normally phosphorylated by Sgg, thus making it degradation resistant. Arm^S2 expresses normal protein and is susceptible to the degradation machinery. arm-mutant embryos rescued by Arm^S2, secrete normal denticle belts and also have normally patterned naked cuticle (Pai et al., 1997). Thus, similar to endogenous Arm, Arm^S2is stabilized only in Wg signaling cells. Thus, relative levels of Arm^S2 at the DV boundary of wing and haltere discs can be used as an estimate of the relative efficiency of the Arm-degradation machinery. We expressed Arm^S10 and Arm^S2 using theomb-GAL4 driver, and stained wing and haltere discs with anti-Myc and anti-Arm antibodies. We observed uniform levels of the degradation-resistant form of Arm in both wing(Fig. 4F) and haltere discs(Fig. 4G). However,degradation-sensitive Arm^S2 accumulated in the DV boundary of wing discs (Fig. 4H) but not of haltere discs (Fig. 4I). This suggests that Arm is degraded more efficiently at the DV boundary of haltere discs than at the DV boundary of wing discs.

In DV cells of the wing disc, in which Wg signaling is active, Arm degradation is inhibited owing to inhibition of the degradation machinery. Dsh functions immediately downstream of Wg, and inhibits Sgg activity and thereby stabilizes Arm. Overexpression of Dsh at the haltere DV boundary did not induce the stabilization of Arm (Fig. 5A) suggesting that the Ubx-mediated inhibition is downstream of Dsh function. One possibility is that Ubx interferes with Dsh-mediated inhibition of Sgg activity, thus keeping Sgg active and causing degradation of Arm. To test this hypothesis, we overexpressed the human colon cancer geneAPC in wing and haltere discs. In both Drosophila and mammalian cells, it has been shown that APC binds to Arm/β-catenin even when Wg/Wnt is active (Papkoff et al.,1996; Bhandari and Shashidhara,2001). In those cells, APC sequesters Arm/β-catenin, rather than recruiting it to the degradation machinery. For example, overexpression of APC in wing discs sequesters Arm only in DV cells(Bhandari and Shashidhara,2001) (Fig. 5B). In other cells, overexpressed APC participates in the Arm-degradation machinery and hence no change in Arm expression was observed. Thus, the amount of Arm sequestered by overexpressed APC could be an assay for the level of Wg/Wnt activity. As only unphosphorylated Arm is sequestered and not the phosphorylated form (Munemitsu et al.,1996), such an assay could also be used to obtain a relative estimate of Sgg activity. When we overexpressed human APC at the haltere DV boundary using vg-GAL4, we observed increased levels of Arm(owing to sequestration) in the anterior compartment(Fig. 5C) but not in the posterior compartment, indicating that Sgg is inactive in the anterior compartment and active in the posterior compartment. Thus, Ubx-mediated inhibition of Arm stabilization in the anterior compartment is downstream of Sgg.

As Wg itself is repressed in the posterior compartment, it was expected that Sgg would be active in that compartment. We further examined whether overexpression of Dsh in the posterior compartment was capable of inhibiting Sgg activity. We co-expressed Dsh and APC and monitored the sequestration of Arm, with the assumption that, if Dsh inhibits Sgg activity, overexpressed APC would be able to sequester Arm in the posterior compartment. Indeed,overexpression of Dsh and APC together resulted in the sequestration of Arm in the posterior compartment at levels similar to those in the anterior compartment (Fig. 5D).

Although levels of Arm were much lower in haltere discs than in wing discs,it is possible that available amounts of Arm are sufficient to transduce Wg signaling. We used Wg-autoregulation as a test for Arm function at the DV boundary of haltere discs. It has been shown that Wg is autoregulated and Arm is necessary for this process (Hooper,1994; Yoffe et al.,1995). For example, ectopic activation of Arm function in leg discs induces ectopic Wg expression(Bhandari and Shashidhara,2001). We observed repression of Wg at the DV boundary when we overexpressed DN-TCF/pan in wing discs using vg-GAL4(Fig. 6A). However, we did not observe any such loss of Wg at the haltere DV boundary(Fig. 6B), nor was there any change in the size of haltere pouch. These results suggest that Arm function is indeed downregulated at the haltere DV boundary.

We then examined how, in the absence of autoregulation, Wg expression is maintained at the anterior haltere DV boundary. Invg¹/vg¹ haltere discs, in which Vg is not expressed at the DV boundary (Fig. 6C), Wg expression is completely absent(Fig. 6D). Thus, the maintenance of Wg expression at the anterior haltere DV boundary, even when Arm function, and thereby Wg autoregulation is inhibited by Ubx, could be attributed to Vg function. This raises the question of why Wg is not expressed in the posterior compartment, in spite of robust expression of Vg. This is particularly intriguing because in wing discs it has been observed that ectopic Vg is capable of activating Wg even in the absence of N signaling(Klein and Martinez-Arias,1999). We further tested the ability of DV boundary cells in the posterior compartment to express Wg by ectopic expression or overexpression of activated N (UAS-N^intra), Dsh and activated Arm using thevg-GAL4 driver. None of these positive regulators induced Wg expression in the posterior compartment. It is likely that Ubx (probably with a posterior-specific co-factor) directly inhibits Wg expression.

In haltere discs, in which Wg is not expressed in the posterior compartment, Vg is still expressed all along the DV boundary, suggesting that Vg is independent of Wg function in the posterior compartment. In the anterior compartment also, Vg might not be dependent on Wg as Arm function is downregulated by Ubx. Indeed, expression of DN-TCF/pan at the haltere DV boundary did not affect Vg expression in haltere discs(Fig. 6E). The possibility that DN-TCF/pan did not downregulate Vg in haltere discs owing to its late expression (we used vg-GAL4) is ruled out because in wing discs it downregulated Vg in both DV and non-DV cells(Fig. 1D-F). This suggests that Vg expression at the DV boundary of haltere discs is independent of Wg function.

Overexpression of N, Wg or activated Arm (both Flu-ΔArm and Arm^S10) at the haltere DV boundary using thevg-GAL4 driver did not induce activation of Vg in non-DV cells (monitored by both anti-Vg antibody and vg-QE staining) in haltere discs, nor did they induce any adult haltere phenotypes (data not shown). This suggests that Ubx inhibits additional events downstream of DV signaling.

We then examined the events in non-DV cells that might contribute to the suppression of Vg in the haltere pouch. We observed cell-autonomous activation of Vg in non-DV cells when we expressed activated N usingdpp-GAL4 (Fig. 7A). However, unlike in wing discs(Fig. 3D), ectopic N expression failed to activate Wg in haltere discs(Fig. 7B). Furthermore,overexpression of Wg, Dsh or activated Arm directly in non-DV cells usingdpp-GAL4, omb-GAL4 or N23-GAL4 drivers did not activate Vg (monitored by both anti-Vg antibody and vg-QE staining; data shown only for Dpp-GAL4/UAS-activated Arm;Fig. 7C). These results further suggest that in addition to repressing DV signaling, Ubx downregulates event(s) downstream of Arm in both anterior and posterior non-DV cells.

As Vg is not expressed in non-DV cells of haltere discs, we examined the effect of its `ectopic expression' in those cells. In haltere discs theN23-GAL4 driver is expressed only in the posterior compartment(Fig. 7E), in which Wg is not expressed. Ectopic expression of Vg in non-DV cells using N23-GAL4activated its own expression in haltere discs, as seen with vg-QE staining (Fig. 7G). Even in avg¹ background, ectopic expression of Vg in non-DV cells was sufficient to activate vg-QE(Fig. 7H). This suggests that in non-DV cells, Ubx functions downstream of Arm and upstream of Vg-autoregulation.

vg is a pro-wing gene: ectopic expression of Vg induces ectopic wing development (e.g. ectopic Vg induces ectopic wing tissue on T2 legs)(Kim et al., 1996). Interestingly, ectopic Vg in T3 leg discs induces ectopic haltere development(Weatherbee et al., 1998). Activation of vg-QE (Fig. 7G) by ectopic Vg in non-DV cells of haltere discs results in homeotic transformation, albeit only partial. Ubx regulates haltere development by modifying wing-patterning events at multiple levels(Weatherbee et al., 1998;Shashidhara et al., 1999). As haltere discs express several other wing-patterning genes (includingvg at the DV boundary), ectopic expression of Vg might override Ubx function in non-DV cells of haltere discs but not in T3 leg discs. We therefore expressed Vg in haltere discs using several GAL4 lines. We observed a high degree of haltere-to-wing homeotic transformations when Vg was expressed using omb-GAL4(Fig. 8B). In addition, we observed enhanced homeotic transformations when Vg was expressed in aUbx-heterozygous background (Fig. 8D). Ectopic expression or overexpression of Wg, Dsh or activated Arm in haltere discs did not induce homeotic transformation (data not shown). This is consistent with the inability of Wg, Dsh and activated Arm to activatevg-QE or Vg protein expression in the haltere pouch.

All the reported haltere-to-wing homeotic transformations at the cuticle level are associated with the loss of Ubx protein. The only exception is a minor sensory bristle phenotype induced by the overexpression of Ac(Weatherbee et al., 1998). In this context, we tested if ectopic Vg downregulated Ubx levels. Anti-Ubx antibody staining of omb-GAL4/UAS-vg haltere discs did not reveal any reduction in Ubx protein levels (data not shown). Furthermore, we did not observe any wing-margin bristles (Fig. 8B), which is the characteristic phenotype of both null(manifested in heterozygous flies) and hypomorphic alleles (manifested in homozygous flies) of Ubx. We also examined the expression pattern of Salm, which is a direct target of Ubx in the haltere pouch(Weatherbee et al., 1998;Galant et al., 2002). Salm expression remained repressed in the haltere pouch(Fig. 8F), which suggests that Vg-induced phenotypes are caused by a reversal of Ubx function and not due to downregulation of Ubx itself.

Suppression of hind-wing development marks the evolution of dipteran flies from their ancestral four-winged insects. However, Ubx, the master regulatory gene that specifies haltere development in Drosophila, is expressed during lepidopteran hind-wing development(Warren et al., 1994;Weatherbee et al., 1999). It is therefore likely that Ubx functions by repressing a few key genes required for wing development rather than by acting as a global repressor(Weatherbee et al., 1999). Previous reports suggest that Wg and Vg, the two genes that play crucial roles during Drosophila wing development, are targets of Ubx activity during haltere development (Weatherbee et al., 1998; Shashidhara et al.,1999). We have examined the mechanism by which Ubx modifies Wg and Vg expression and thereby downregulates DV signaling.

We designed experiments to test the current model of Wg and Vg regulation(which is essentially based on studies on wing imaginal discs) in haltere discs. In wing discs, both Wg and Vg are subjected to an elaborate regulatory circuit, the understanding of which would help us to unravel crucial events during wing development. To examine the Wg and Vg interactions further in DV and non-DV cells, we carried out experiments that are essentially complementary to those reported previously.

The experiments described in this report further suggest that Wg and Vg interact to maintain each other's expression at the DV boundary. We have shown that Vg-mediated activation of Wg is independent of Arm and TCF/pan function,which suggests that Vg may activate Wg either directly or through the N signaling pathway. We have also shown that Vg is capable of specifying wing development, even in the absence of Wg signaling. Overexpression of Vg in avg¹/vg¹ background (in which no Wg or Vg is expressed) was sufficient to rescue wing phenotypes. This is particularly significant because we expressed Vg in this experiment only in non-DV cells. Our results also suggest that Vg cell-autonomously regulates its own expression through its quadrant enhancer. Clonal analysis of armsuggested that Wg is required to activate vg-QE and Arm was not able to activate this enhancer in vg¹ background. Wg signaling might activate Vg either indirectly or by activating some other enhancer of Vg. Once activated, Vg might maintain its expression by autoregulation, which is mediated through its quadrant enhancer(Fig. 8G). This could ensure the maintenance of Vg expression in non-DV cells, once it is activated by Wg signaling. It might also explain how the Wg gradient is translated into uniformly higher levels of Vg in non-DV cells.

However, the above-mentioned model does not reconcile the observation that Vg, and not Wg, is capable of activating vg-QE inSer^– background(Klein and Martinez-Arias,1999). As the vg gene is intact inSer^– background, ectopic expression of Wg usingdpp-GAL4 should have activated one of the enhancers to induce Vg expression, which in turn would activate vg-QE. A model that reconciles all the results would, therefore, include a third component,which may act either in parallel to or downstream of Wg and Vg at the DV boundary (Fig. 8G). The presence of such a signaling molecule downstream of Vg has been previously predicted (Neumann and Cohen,1996). Although there is no direct evidence for the existence of such a molecule, the fact that N23-GAL4 expression in non-DV cells is dependent on N function and independent of Vg and Wg function (R.B. and L.S.S., unpublished observations) suggests such a possibility.

We also studied possible mechanisms by which Ubx regulates expression of Wg and Vg in haltere discs. One important finding was the downregulation of Wg signaling by Ubx at the level of Arm stabilization. We have further shown that Ubx inhibits stabilization of Arm by acting on event(s) downstream of Sgg. Normally, the Arm degradation machinery is very efficient and can degrade even overexpressed Arm. This is evident from the fact that embryos overexpressing Arm (from arm^S2) secrete normal denticle belts(Pai et al., 1997). If a downstream component functions with enhanced efficiency (either by direct enhancement of its expression by Ubx or owing to repression of a positive component of Wg signaling), residual activity of Sgg may be sufficient to cause enhanced degradation of Arm. Thus, enhanced degradation of Arm in haltere discs provided us with a new assay system to identify additional components of Wg signaling. For example, in microarray experiments to identify genes that are differentially expressed in wing and haltere discs, we observed that several transcripts of known (e.g. Casein kinase) and putative (e.g. Ubiquitin ligase) negative regulators of Wg signaling are upregulated in haltere discs (M.P. and L.S.S., unpublished).

Our results suggest that Wg and Vg regulation in haltere discs is different from that of wing discs. We have observed that Wg is not autoregulated in haltere discs. In addition, Vg expression at the haltere DV boundary is independent of Wg function. However, in both wing and haltere discs, Wg expression at the DV boundary is dependent on Vg. Wg expression at the anterior DV boundary of haltere discs could be redundant because overexpression of DN-TCF at the haltere DV boundary shows no phenotype. However, Vg at the DV boundary appears to have an independent function.vg¹ flies exhibit much smaller halteres than do wild-type flies (Williams et al., 1991). As Wg function (and expression in the posterior compartment) is already repressed in haltere discs, reduction in haltere size invg¹ flies suggests Wg-independent long-range effects of Vg from the DV boundary. This could be one of the reasons why Ubx does not affect Vg expression at the DV boundary but represses Vg expression in non-DV cells. In wing discs too, Vg may have such a function on cells at a distance(Neumann and Cohen, 1996).

One way to test the requirement of Ubx in DV and non-DV cells directly is by removing Ubx only from the haltere DV boundary or from non-DV cells. We have previously reported that clonal removal of Ubx solely from the haltere DV boundary does not induce cuticle phenotype in the capitellum (Shashidhara et al.,1999). However, we could not ascertain the effect onvg-QE because of haploinsufficiency,Ubx^–-heterozygous haltere discs themselves show activation of lacZ in the entire haltere pouch (data not shown). The activation of vg-QE in Ubx^–/+haltere discs could be a result of reduced Ubx function at the DV boundary, or in non-DV cells, or in both. We had previously shown that misexpression of Ubx at the wing disc DV boundary causes non-cell-autonomous reduction invg-QE expression (Shashidhara et al., 1999). Our current results suggest that Ubx represses additional event(s) in non-DV cells to downregulate Vg expression. This is consistent with the recent report on cell-autonomous repression ofvg-QE by ectopic Ubx in wing discs(Galant et al., 2002). We propose that Ubx inhibits the activation of Vg in non-DV cells at three different levels (Fig. 8G): (1)Wg in the posterior compartment; (2) event(s) downstream of Sgg that inhibit the stabilization of Arm; and (3) additional event(s) downstream of Arm in non-DV cells. In wing discs, as discussed above, Wg and a hitherto unknown DV component may function together to activate Vg in non-DV cells. As Vg-autoregulation is not inhibited in haltere discs, it is possible that Ubx represses Vg activation in non-DV cells by interfering with the Wg-mediated activation of Vg and/or by repressing the activity of the unknown DV-signal molecule in the haltere.

We have also provided evidence that repression of Vg in non-DV cells by Ubx is crucial for haltere development. Overexpression of Vg in haltere discs causes haltere-to-wing transformations. This is particularly significant considering the fact that haltere-to-wing homeotic transformations are always associated with loss of Ubx, by direct removal of Ubx, by activation of its repressors (e.g. polycomb proteins) or by suppression of its activators (e.g. trithorax proteins). Mitotic clones ofUbx^– alleles in the haltere capitellum normally`sort out' and often remain as an undifferentiated mass of cells(Morata and Garcia-Bellido,1976; Shashidhara et al.,1999). This is attributed to differential cell-adhesion properties of transformed (Ubx^–) and non-transformed(Ubx⁺) cells. No such sorting out of wing-like trichomes was observed in halteres overexpressing Vg. This implies that cells surrounding the wing-like trichomes are also transformed, at least at the level of cell-adhesion properties. This is consistent with our earlier observations that removal of Ubx from the DV boundary or over-growth caused by mutations in the tumor-suppressor gene fat confers wing-like cell-adhesion properties to capitellum cells(Shashidhara et al., 1999). As DV signaling is closely associated with the activation of Vg in non-DV cells and Vg is primarily a growth-promoting gene, it is likely that the cell-sorting behaviour of Ubx^– clones is linked to their changed growth properties.

We thank S. Artavanis-Tsakonas, R. Barrio, K. Basler, S. Carroll, S. Cohen,R. Nagaraj, J. Woodgett, the Bloomington Stock Center, and the Developmental Studies Hybridoma Bank for fly stocks and antibodies. We thank members of the laboratory for technical help, advice and discussions, and the anonymous referees for suggestions to improve the manuscript. This work was supported by a grant to L.S.S. from the Department of Science and Technology (Government of India).