The halteres and wings of Drosophila are homologous thoracic appendages, which share common positional information provided by signaling pathways. The activity in the haltere discs of the Ultrabithorax(Ubx) Hox gene establishes the differences between these structures,their different size being an obvious one. We show here that Ubxregulates the activity of the Decapentaplegic (Dpp) signaling pathway at different levels, and that this regulation is instrumental in establishing the size difference. Ubx downregulates dpp transcription and reduces Dpp diffusion by repressing the expression of master of thick veins and division abnormally delayed and by increasing the levels of thick veins, one of the Dpp receptors. Our results suggest that modulation in Dpp expression and spread accounts, in part, for the different size of halteres and wings.
Pattern formation in animals requires the concerted activity of selector genes and signaling pathways. A particular class of selector genes is formed by the Hox genes, which specify different structures along the anterior-posterior axis of metazoans(McGinnis and Krumlauf, 1992). In Drosophila, mutations in these genes frequently transform one structure into another keeping the coordinates provided by underlying positional information, established in part by the activity of signaling pathways. Mutations in the Ultrabithorax (Ubx) Hox gene illustrate this assertion. Wings and halteres are homologous structures located in the second and third thoracic segments, respectively. These appendages greatly differ in size and pattern, and derive from imaginal discs,the wing and haltere discs, which also differ in size but bear a similar morphology. Ubx, which is expressed in the haltere disc but not in the wing disc, determines the difference between these two structures:mutations in Ubx transform halteres into wings whereas Ubxectopic expression changes wings into halteres(Lewis, 1963; Lewis, 1978; Cabrera et al., 1985; White and Akam, 1985; White and Wilcox, 1985). These transformations respect positional cues(Morata and García-Bellido,1976) dictated by signaling pathways.
In Drosophila, one of the best-studied signaling pathways is that of the decapentaplegic (dpp) gene (homologous to the TGF-β in vertebrates). This pathway has been analyzed extensively in pattern formation of the imaginal discs, particularly the wing disc (reviewed by Tabata, 2001). This disc is subdivided early in development into an anterior (A) and a posterior (P)compartment (García-Bellido et al.,1973). The protein encoded by the hedgehog (hh)gene, synthesized in the posterior compartment, activates dpptranscription in anterior cells close to the anteroposterior (A/P) border(Posakony et al., 1991; Basler and Struhl, 1994; Capdevila and Guerrero, 1994; Tabata and Kornberg, 1994). The Dpp ligand diffuses into both A and P compartments, generating a gradient of protein concentration (Entchev et al.,2000; Teleman and Cohen,2000). Dpp behaves as a morphogen, translating the protein concentration gradient into the restricted and overlapping expression of genes like spalt (sal) and optomotor-blind (omb)(Lecuit et al., 1996; Nellen et al., 1996). Among these, the sal gene is repressed by Ubx in the haltere pouch(Weatherbee et al., 1998; Barrio et al., 1999; Galant et al., 2002),indicating that the outcome of Dpp signaling is modified by Ubx in the haltere disc.
Dpp activity can be monitored with an antibody that recognizes the phosphorylated form of Mothers against dpp (Mad)(Persson et al., 1998; Tanimoto et al., 2000), a receptor-regulated Smad that transduces the dpp signal(Newfeld et al., 1997). The analysis of this and other Dpp pathway elements has revealed their different contribution to the formation of the Dpp ligand and activity gradients. Thus,the expression in the wing disc of one type I Dpp receptor, thick veins (tkv), is not uniform, and this unequal distribution modulates Dpp signaling along the A/P axis (Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000). Similarly, the spread and activity of Dpp depends on the presence of cell-surface molecules like those encoded by the division abnormally delayed (dally) and dally-like protein (dlp)genes (Jackson et al., 1997; Nakato et al., 1995; Fujise et al., 2003; Belenkaya et al., 2004). All these elements establish the fine tuning of Dpp activity, which is crucial in determining the form and size of Drosophila wings(Spencer et al., 1982; Capdevila and Guerrero, 1994; Zecca et al., 1995; Lecuit et al., 1996; Nellen et al., 1996; Tsuneizumi et al., 1997;Haerry et al., 1998; Lecuit and Cohen,1998; Campbell and Tomlinson,1999; Jazwinska et al.,1999; Minami et al.,1999; Martín-Castellanos and Edgar,2002; Martín et al.,2004). Therefore, the shape and size of adult derivatives can be established by adjusting the Dpp input in imaginal cells.
Ubx mutations increase the size of the halteres, transforming them into wings (Lewis, 1963)whereas dpp mutations reduce the size of the halteres(Spencer et al., 1982). Changes in the Dpp pathway affect wing size (reviewed by Day and Lawrence, 2000), and recent evidence indicates that cell proliferation in the wing disc is induced by reading different Dpp activity levels(Rogulja and Irvine, 2005). Although the Ubx and dpp effects (in halteres and wings)could be unrelated, the homologous nature of both appendages suggests that Ubx may fix haltere size by modifying the Dpp pathway. We have explored this idea and compared Dpp distribution and activity in the wing and haltere discs. We show that Ubx downregulates dppexpression, alters Dpp activity and reduces Dpp spread, and that the latter is achieved mainly by controlling the expression of tkv and dally. Our results have implications in the way Ubxestablishes the different size of halteres and wings.
MATERIALS AND METHODS
The dppd12 and dppd5 mutations remove regulatory regions of the dpp gene(St Johnston et al., 1990). Ubx6.28 is a null allele of Ubx(Beachy et al., 1985), the TM2 balancer carries the Ubx130 null mutation(Lewis, 1952), and the Df109 deletion eliminates the Ubx gene(Lewis, 1978). The bx3 and pbx mutations eliminate Ubxexpression in the anterior and posterior compartments of the haltere disc,respectively, transforming them into the corresponding ones of the wing disc(Lewis, 1963; García-Bellido et al.,1973; Cabrera et al.,1985; White and Wilcox,1985). The CbxTwt mutation ectopically expresses Ubx in the wing disc(Bender et al., 1983). DfC1-h1 (Szidonya and Reuter,1988) and Df tkv2(Szidonya and Reuter, 1988; Terracol and Lengyel, 1994)are deficiencies that uncover the tkv gene. The following reporter insertions or constructs were used: dpp-lacZBS3.0(Blackman et al., 1991), dpp-lacZ10638 (Twombly et al., 1996), hh-lacZ(Lee et al., 1992) dally-lacZ (Nakato et al., 1995), tkv-lacZ(Tanimoto et al., 2000), mtv-lacZ (Funakoshi et al., 2001) and omb-lacZ(Grimm and Pflugfelder, 1996). The Gal4/UAS method (Brand and Perrimon,1993) was used with the following Gal4 lines and UAS constructs: dpp-Gal4 (Morimura et al.,1996), en-Gal4(Tabata et al., 1995), ptc-Gal4 (Hinz et al.,1994), ap-Gal4(Calleja et al., 1996), MS1096-Gal4 (Capdevila and Guerrero, 1994), UAS-dpp(Capdevila and Guerrero,1994), UAS-Dpp-GFP (Entchev et al., 2000), UAS-tkv(Lecuit and Cohen, 1998),UAS-tkvQ253D (Nellen et al., 1996), UAS-tkvDN (Haerry et al.,1998), UAS-dally (Tsuda et al.,1999), UAS-mtv (T. Tabata, unpublished), UAS-Ubx(Castelli-Gair et al., 1994),UAS-dsRNA>Ubx (Monier et al.,2005) and UAS-GFP (Ito et al.,1997). The tub- Gal80ts/Gal4 system(McGuire et al., 2003) was used to temporally control the induction of transgenes with the Gal4/UAS method. To this aim, larvae were transferred from 17°C to 29°C during the second or third larval instars.
We used the FLP/FRT system (Xu and Rubin, 1993) to induce Ubx mutant clones in the haltere disc with the FRT82B Ubx6.28 chromosome(Weatherbee et al., 1998), the MARCM method (Lee and Luo,1999) to induce clones that lose Ubx and activate tkv in the haltere disc, and the combination of FLP/FRT and Gal4/UAS methods (Pignoni and Zipursky,1997; Ito et al.,1997) to induce Ubx-expressing clones in the wing disc;in all the cases the clones were induced during the larval period. The genotypes of the larvae where the clones were induced are as follows.
Ubx- clones: y hs-flp122; FRT82B Ubx6.28/FRT82B Ubi-GFP, y hs-flp122; FRT82B Ubx6.28 hh-lacZ/FRT82B Ubi-GFP and ptc-Gal4/UAS-flp; FRT82B Ubi-GFP/FRT82B Ubx6.28
Ubx- clones, dpp-lac-Z: y hs-flp122; dpp-lacZBS3.0/+; FRT82B Ubx6.28/FRT82B Ubi-GFP
Ubx- clones, omb-lac-Z: y hs-flp122/omb-lacZ; FRT82B Ubx6.28/FRT82B Ubi-GFP
Ubx- clones, tkv-lacZ: y hs-flp122; tkv-lacZ/+; FRT82B Ubx6.28/FRT82B Ubi-GFP
Ubx- clones, mtv-lacZ: y hs-flp122; mtv-lacZ/+; FRT82B Ubx6.28/FRT82B Ubi-GFP
Ubx- tkv+ clones, omb-lacZ: y hs-flp122 tub-Gal4 UAS-GFP/omb-lacZ;UAS-tkv FRT82B Ubx6.28/FRT82B tub-Gal80
Ubx+ clones, dally-lacZ: y hs-flp122; act5C>y+>Gal4 UAS-GFP/UAS-Ubx; dally-lacZ/+.
In situ hybridization
Immunohistochemistry was carried out as previously described(Sánchez-Herrero, 1991; Estrada and Sánchez-Herrero,2001). The antibodies used were: mouse and rabbit anti-β-galactosidase (Cappel), mouse Mab4D9 anti-En(Patel et al., 1989), rat anti-Tkv (Teleman and Cohen,2000), rabbit anti-P-Mad(Tanimoto et al., 2000; Persson et al., 1998) [a gift of F. A. Martín and G. Morata, Centro de Biologia Molecular, Severo Ochoa (C.S.I.C.-U.A.M.), Madrid, Spain], and mouse anti-Ubx(White and Wilcox, 1984). Secondary antibodies were coupled to Red-X, Texas Red, FITC and Cy5 fluorochromes (Jackson ImmunoResearch).
Adult cuticle analysis
Flies were kept in a mixture of ethanol:glycerol (3:1) until needed. Flies were then macerated in 10% KOH at 60°C for 10 minutes, dissected, washed with water, dehydrated with ethanol and finally mounted in Euparal for inspection under a compound microscope.
Measurements in the imaginal discs
We calculated the width in haltere and wing pouches with the Measure Tool of Adobe Photoshop 8.0 using the position of the dorsoventral boundary as a reference line for these measurements.
The intensity of the Dpp-GFP dots was calculated with the MetaMorph Offline program. The final profiles of the intensities were obtained following the same procedure in bx3/MKRS wing and haltere discs and in bx3/TM2 haltere discs. We first calculated the average value, along the A/P axis, of the GFP intensity in three different sections of a disc. This gives a mean value for the disc. We repeated this in three different discs of each type, thereby obtaining three mean values in each case. The final profile for either wing, haltere or bx3/TM2 haltere discs was obtained by plotting along the A/P axis the average of those three mean values obtained for each type of disc. The fixation and staining for all the discs was done simultaneously and under the same conditions. All pictures were processed under identical conditions.
The expression of dpp in the haltere disc is downregulated by Ubx
To monitor dpp transcription in the wing and haltere discs we hybridized them with a dpp RNA probe. dpp is expressed in both discs in anterior cells close to or abutting the A/P boundary (hereafter named `anterior A/P Boundary cells' or `AB cells'), but in the haltere disc the dpp stripe is weaker and less straight(Mohit et al., 2006)(Fig. 1A). To better define this expression we have used a dpp-lacZ reporter construct and a dpp-lacZ insertion that mimic dpptranscription in the imaginal discs(Blackman et al., 1991; Twombly et al., 1996; Weatherbee et al., 1998). Both give comparable results. In the wing pouch, the dpp-lacZ stripe(dpp-lacZ10638) is about 8.1 cells wide(n=16) whereas in the haltere pouch it is about 6.0 cells(n=13; Fig. 1B-D′). Anterior Ubx mitotic recombination clones abutting the A/P boundary increase both the intensity of the signal and the width of the stripe (Fig. 1E,E′). To characterize how Ubx regulates dpp signal in A and P compartments, we used the bithorax(bx) and postbithorax (pbx) mutations (see Materials and methods). In bx (bx3/TM2)haltere discs, the width of the dpp-lacZ stripe is about 7.9 cells(n=10; Fig. 1F,F′) approaching that of the wing disc (8.1 cells),whereas in pbx (pbx/TM2) haltere discs the average width is 6.4 cells (n=8; Fig. 1G,G′), slightly higher than in the wild type (6 cells). Therefore, the reduction in dpp transcription by Ubx depends mainly on the Ubx activity in the anterior compartment of the haltere disc.
Ubx controls the response to the dpp signal by retarding Dpp diffusion
To study whether Ubx governs the response to Dpp signaling, we monitored the expression of omb, a target of the dpppathway, with an omb-lacZ insertion(Grimm and Pflugfelder, 1996; Lecuit et al., 1996; Nellen et al., 1996). As in the wing disc (Fig. 2A-A″), omb is expressed in both compartments of the haltere pouch (Weatherbee et al.,1998) (Fig. 2B-B″). In strong bx and pbx mutants omb expression in A and P compartments of the haltere disc resembles that of the corresponding compartments of the wing disc(Fig. 2C-D″). Because in bx mutants the dpp expression is like that of the wing disc(Fig. 1F) but ombexpression in the P compartment is not(Fig. 2C), and because in pbx mutants dpp transcription in the haltere disc is only slightly increased (Fig. 1G),but omb signal is clearly extended(Fig. 2D), Ubxprobably regulates the response to the Dpp signal. This conclusion is reinforced by the analysis of omb transcription in Ubxmutant clones: clones located outside the omb expression domain do not activate omb transcription. When the clones encompass the border of omb expression, the omb signal is extended further anteriorly or posteriorly. Notably, in some cases there is ectopic omb transcription outside the clone, indicating a non-cell-autonomous effect of Ubx loss on omb expression(Fig. 2E-F″). Taken together, these results indicate that Ubx represses ombactivation in cells that receive a low amount of Dpp, but that a certain level of Dpp is required to activate omb even in the absence of Ubx.
These results suggest that Ubx reduces the transcriptional response of haltere disc cells to the Dpp signal. This may be achieved by limiting the spread of the Dpp product. To validate this assumption, we examined the distribution of a Dpp-GFP fusion protein expressed under the control of the dpp-Gal4 driver(Entchev et al., 2000; Teleman and Cohen, 2000). The GFP signal is more restricted in the haltere disc than in the wing disc (not shown). We aimed to quantify these differences and, for that, we used the ptc-Gal4 driver to drive Dpp-GFP expression because the dpp-Gal4 driver directs more irregular and variable expression in the haltere disc. The extension of Dpp-GFP expression is higher in tub-gal80ts/ptc-Gal4; UAS-Dpp-GFP bx3/MKRS wing discs than in haltere discs of the same genotype (Fig. 3A,B). In tub-gal80ts/ptc-Gal4; UAS-Dpp-GFP bx3/TM2 haltere discs the expression levels and anterior extent of the fusion protein are very similar to those of the wing disc, but the spread in the posterior compartment is only slightly increased with respect to the wild type (Fig. 3C). A summary of the data is shown in Fig. 3D. This shows that the Dpp-GFP signal falls to background levels more abruptly in haltere discs than in the wing discs or in the anterior compartment of bx haltere discs. This difference is more evident in medial regions of both compartments: in these regions there is a graded decrease of the Dpp-GFP signal in the wing disc, whereas the distribution profile is flat in the haltere disc. A plot of the GFP-Dpp dot distribution in three different sections of a single wing or haltere disc is shown in Fig. S1 in the supplementary material. These results,together with our previous observations, strongly suggest that Ubxcounteracts the spread of Dpp in the haltere pouch.
High levels of tkv in the haltere disc are induced by Ubx through regulation of mtv expression and dppactivity
We have just shown that Ubx reduces Dpp diffusion in the haltere disc, thus limiting omb expression. To draw a general conclusion about how Ubx regulates Dpp signaling we looked at the distribution of the phosphorylated form of Mad (P-Mad), a major readout of Dpp activity(Tanimoto et al., 2000). The P-Mad signal in the haltere pouch is narrower than the signal in the wing pouch, confined almost exclusively to the anterior compartment and not reduced in AB cells (Fig. 4A-C′). Hence, and also in contrast with the wing disc(Fig. 4D-D″), high levels of both Hh and Dpp signaling coincide in these cells(Fig. 4E-E″). The low Dpp signaling in central cells of the wing disc is due to the reduced expression of tkv, which is more strongly expressed peripherally and is particularly low in AB cells (Brummel et al., 1994; de Celis,1997; Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000)(Fig. 4F). By contrast,although the expression of tkv in the haltere pouch increases slightly in the periphery, it is uniform in the central region and higher than in the corresponding domain of the wing pouch(Fig. 4G). In Ubx- clones the tkv expression is reduced but for the clones induced in the more lateral domains(Fig. 4H-I″). Conversely,ectopic Ubx expression in medial regions of the wing disc of CbxTwt mutants increases tkv expression(Fig. 4J,J′). As Tkv levels are crucial for Dpp signaling and Dpp diffusion (Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000; Funakoshi et al., 2001), we decided to examine in more detail the regulation of tkv expression by Ubx.
In the wing pouch, the distribution of tkv is regulated by two mechanisms (Lecuit and Cohen,1998; Funakoshi et al.,2001). The first mechanism depends on the activity of master of thick veins (mtv)(Funakoshi et al., 2001). In AB cells, high mtv expression, under control of Hh signaling,strongly reduces the tkv signal; in cells located in a medial position along the A/P axis, moderate mtv expression reduces tkv transcription to a basal level(Funakoshi et al., 2001). An mtv-lacZ reporter insertion is prominently expressed in the AB cells and in two peripheral domains of the wing pouch, and expressed at low levels in the rest of the pouch (Funakoshi et al., 2001) (Fig. 4K). In the haltere disc, only the lateral signal remains(Fig. 4L). This difference is due to Ubx because in bx3/TM2 haltere discs an A/P stripe appears (Fig. 4M) and in Ubx- clones mtv is derepressed (Fig. 4N,N′). Reciprocally, ectopic Ubx expression in the wing disc represses mtv in central and medial domains(Fig. 4O,O′). Therefore,the absence of mtv in AB cells of the haltere pouch can explain their high levels of tkv expression and Dpp signaling. Consistently, in MS1096-Gal4; UAS-mtv/+ larvae, in which mtv is strongly expressed in the dorsal region of the haltere pouch, tkv levels are partially reduced dorsally except in the more lateral domains (Fig. 4P, the wild type is shown in Fig. 4G). Therefore, Ubx repression of mtv in central and medial regions of the haltere pouch contributes to their high tkvtranscription.
The second mechanism to regulate tkv depends on Dpp activity. The high tkv levels in peripheral cells of the wing disc are reduced if Dpp signaling is augmented (Lecuit and Cohen, 1998; Tanimoto et al.,2000). Thus, in MS1096-Gal4;UAS-tkvQ253D/+ larvae tkv-lacZlevels are reduced in the dorsal region(Fig. 4Q). By contrast, no such repression is observed in the haltere pouch(Fig. 4R), indicating that Ubx prevents the repression of tkv mediated by Dpp signaling. Collectively, our results show that Ubx promotes high levels of tkv in the haltere pouch by repressing mtv and by inhibiting the Dpp signaling-mediated downregulation of tkv. However,we cannot exclude a direct effect of Ubx on tkv.
The increased expression of tkv reverts the effect of Ubx loss on Dpp activity
High levels of tkv increase Dpp transduction cell autonomously,but also restrict Dpp spread by trapping the Dpp protein (Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000). Given that Ubx increases tkv expression in the haltere disc, these high levels may retain more Dpp ligand than in the wing disc, thus explaining the restricted Dpp diffusion and activity observed in the haltere pouch. To confirm this, we first induced Ubx- clones in the haltere disc and looked at the P-Mad expression pattern. In clones encompassing the border of high P-Mad staining there is an expansion of the P-Mad signal that,frequently, is observed outside the mutant territory(Fig. 5A-A″). This non-cell-autonomous effect is similar to that described previously for omb (Fig. 2F-F″)and suggests that Dpp extends readily through the Ubx mutant territory inducing high levels of signaling distally to the clone.
These observations suggest that the absence of Ubx elevates Dpp signaling in regions distant from the A/P boundary, and that this effect may be mediated by decreasing Tkv levels. If so, this outcome should be compensated by increasing tkv expression in Ubx mutant clones. To corroborate this inference, we made clones that simultaneously lose Ubx and express tkv, and followed Dpp activity by looking at omb and P-Mad expression. As shown in Fig. 5B-B‴, Ubx- tkv+ clones in medial regions of the haltere disc do not show extension of P-Mad or omb expression, in contrast with the results observed in clones just mutant for Ubx(Fig. 2E-E″ and Fig. 5A-A″). This result confirms that Ubx regulates Dpp spread and signaling mainly by controlling Tkv levels. Strengthening this idea, we note that ectopic Ubx expression in the central region of the wing disc(dpp-Gal4/UAS-Ubx larvae) reduces the extent of P-Mad expression with respect to the wild type, but increases it in AB cells(Fig. 5D, compare with the wild type in Fig. 4A), and that these effects are similar to those obtained when tkv transcription is augmented in these same cells (Fig. 5C).
Ubx controls the expression of dally
The function and distribution of Dpp, like that of Wingless (Wg) or Hh,depends on the presence of a kind of cell-surface molecule named heparin sulphate proteoglycans (reviewed by Lin,2004). Two proteoglycan members in Drosophila are dally and dlp (Nakato et al., 1995; Khare and Baumgartner, 2000; Baeg et al.,2001). Both are implicated in Dpp function(Fujise et al., 2001; Fujise et al., 2003; Jackson et al., 1997; Tsuda et al., 1999; Belenkaya et al., 2004) and in the transport of Dpp along the A/P axis(Belenkaya et al., 2004). dlp is expressed at slightly lower levels in the haltere pouch than in the wing pouch (not shown), and we have not investigated it further. However, a different dally expression in the two discs was patent. In the wing pouch, a dally-lacZ insertion shows high expression in two bands along the dorsoventral (D/V) boundary and in the AB cells, with lower signal in the rest of the pouch(Fujise et al., 2001; Fujise et al., 2003)(Fig. 6A). By contrast, in the haltere disc the expression in AB cells is missing, the D/V signal seems to be confined to the anterior compartment (where wg is expressed) and there are lower levels throughout the pouch(Fig. 6B). When Ubxexpression is reduced in the haltere pouch, the pattern of dallyresembles that of the wing disc (Fig. 6C), and in clones that ectopically express Ubx in the wing disc there is a reduction in dally signal(Fig. 6D,D′). These results show that Ubx is necessary and sufficient to differentiate dally expression in both discs. Previous results demonstrated that the ectopic expression of dally in the wing disc augments Dpp activity (Fujise et al., 2003; Takeo et al., 2005). Similarly, we have observed an increase in the extent of P-Mad signal in the dorsal domain of ap- Gal4/UAS-dally haltere discs(Fig. 6E). We conclude that Ubx may reduce the extent of Dpp activity in the haltere disc by controlling dally expression.
In AB cells of the wing disc the expression of dally is induced by Hh signaling but can be downregulated if Dpp signaling is increased(Fujise et al., 2003). mtv, whose expression in these cells is also induced by Hh(Funakoshi et al., 2001), is also downregulated if Dpp activity is elevated(Fig. 6F). Therefore, we wondered if the high Dpp signaling present in the AB cells of the haltere pouch may contribute to the lack of dally and mtvexpression. If this hypothesis is correct, a reduction in Dpp signaling should activate these two genes in the A/P boundary of the haltere disc. We found no such dally (Fig. 6G)or mtv (Fig. 6H)activation when Dpp activity was decreased (in MS1096-Gal4;UAS-tkvDN/+ larvae). The repression of mtv and dally by Ubx in AB cells, therefore, is not maintained by high Dpp activity.
Haltere size depends on Ubx regulation of dppexpression and spread
A major difference between wings and halteres is their size(Fig. 7A). Although this is mostly due to the different size of wing and haltere cells(Roch and Akam, 2000), wing discs are also bigger than haltere discs, even though cell size in both structures is similar (Roch and Akam,2000). At the end of embryogenesis, wing discs are about twice as big as haltere discs (Morata and García-Bellido, 1976; Madhavan and Schneidermann,1977; Bate and Martínez-Arias, 1991). We have measured the size of wing and haltere pouches in late third instar larvae and found the former to be about 3.5-4 times bigger than the latter. Assuming that the size difference found at the end of the embryogenesis applies equally to all regions of the disc, this implies that the wing pouch acquires around twice as many cells as the haltere pouch during the larval period. We have measured the size of Ubx mutant clones, and that of their twin spots, induced during the larval stages and analyzed in the haltere discs of late third instar larvae,and found that they are of similar size(Fig. 7B): the `Ubxclone area/twin clone area' ratio is 1.06 (n=20). This suggests that a different proliferation dynamic of Ubx-expressing cells is probably not responsible for the smaller size of haltere discs.
Given that Ubx controls the Dpp pathway and this is involved in the control of wing disc growth (Spencer et al., 1982; Capdevila and Guerrero, 1994; Burke and Basler, 1996; Lecuit et al.,1996; Nellen et al.,1996; Haerry et al., 1998; Lecuit and Cohen, 1998; Martín-Castellanos and Edgar,2002; Martín et al.,2004), we reasoned that Ubx may reduce haltere size by regulating this signaling route. We tried to prove this assumption by different experiments. First, we investigated whether forcing the transcription of Dpp pathway elements, the expression of which is downregulated by Ubx, may impinge on haltere size. Overexpression of dpp in its own domain increases the size of the haltere pouch(Fig. 7C). The ectopic expression of dally or mtv also augments haltere size: we have measured the width of A and P compartments in the haltere pouches of en-Gal4 UASGFP/+, en-Gal4 UAS-GFP/UAS-dally and en-Gal4 UASGFP/UAS-mtv larvae (all grown at 29°C), and found that the ectopic expression of dally or mtv increases the P/A width ratio by 21% and 35%, respectively, with respect to the control discs (Fig. 7D-F,U). This difference is probably not due to the effect of dally on Hh and Wg signaling (reviewed by Lin,2004) because wg is not expressed(Weatherbee et al., 1998) and hh is not active(Domínguez et al.,1996; Hepker et al.,1997) in this compartment.
As a second test, we studied whether there was a phenotypic interaction between dpp and Ubx as regards to haltere size. dpphypomorphic mutations reduce the size of the distal part of the halteres (the capitellum) (Spencer et al.,1982) (Fig. 7Hcompare with the wild type in 7G). In a mutant background heterozygous for Ubx, the dpp- vestigial phenotype is partially suppressed (Fig. 7I),suggesting that a reduction in Ubx can make up for the low Dpp levels. Several experiments argue that this interaction relies, at least in part, in the control by Ubx of tkv expression. First, wings are smaller (without apparent change in cell size, as judged by trichome size and density) if Ubx (Fig. 7K, compare with the wild type in 7J) or tkv (Haerry et al., 1998; Lecuit and Cohen,1998; Tanimoto et al.,2000) (Fig. 7L) are present in AB cells. Second, in Ubx/+ adults the capitellum is enlarged (Fig. 7M), and this phenotype is stronger in sibling flies that are also heterozygous for a tkv deficiency (Fig. 7N). Finally, the increase in the size of the posterior or anterior compartments in pbx- or bx-mutant haltere discs(Fig. 7O,Q) is partially reverted when tkv levels are elevated(Fig. 7P,R): the P/A width ratio in pbx/Ubx6.28 haltere discs is reduced by 37% in en-Gal4 UAS-GFP/+; pbx/UAS-tkv Ubx6.28 larvae and the A/P width ratio in bx3/Ubx6.28 haltere discs is reduced by 17% in ptc-Gal4 UASGFP/+; bx3/UAS-tkv Ubx6.28 larvae (Fig. 7U). The latter reduction is modest probably because the ptc-Gal4 driver expresses tkv only in part of the anterior compartment and because dpp expression is wing-like. As a summary,our results suggest that Ubx reduces the size of the haltere, as compared with the wing, in part through the expression of tkv.
We have demonstrated that the absence of Ubx in mutant clones affects Dpp activity both cell autonomously and non-cell autonomously(Fig. 2F-F″ and Fig. 5A-A″), and that Ubx hinders Dpp spread. Therefore, local changes in Ubxexpression may have non-cell-autonomous effects on size. To prove this, we reared larvae of the ptc-Gal4 UAS-GFP/+; Df109UAS-dsRNA>Ubx/tub-Gal80ts genotype at 17°C and transferred them to 29°C at the second or early third larval stage. This procedure eliminates Ubx expression in the ptc domain(not shown). In many of the flies that underwent this treatment we observed that the anterior haltere tissue was bigger than that expected to derive from the Ubx-expressing region, sometimes even bigger than a whole anterior haltere compartment (Fig. 7S,S′). A similar effect was observed in flies in which the absence of Ubx is clonally inherited(ptc-Gal4/UAS-flp; FRT Ubx6-28/FRT GFP flies; Fig. 7T). This suggests that the absence of Ubx in anterior border cells increases the growth of the more anterior, Ubx-expressing haltere region.
Ubx distinguishes wings from halteres by regulating the expression of many genes, including those forming part of signaling pathways(Weatherbee et al., 1998; Barrio et al., 1999; Shashidhara et al., 1999; Galant et al., 2002; Mohit et al., 2003; Mohit et al., 2006; Pallavi et al., 2006). We show here that Ubx controls the Dpp signaling pathway at different levels and that this regulation contributes significantly to the different sizes of these two appendages. Similar results have also been recently reported(Crickmore and Mann,2006).
Ubx and the dpp pathway
Ubx controls dpp transcription, Dpp spread and Dpp activity (Fig. 8A). In the wing disc, dpp transcription is activated by Hh signaling(Basler and Struhl, 1994; Capdevila and Guerrero, 1994; Tabata and Kornberg, 1994),and in the haltere pouch Ubx attenuates this activation(Mohit et al., 2006) (this report). The haltere dpp stripe increases slightly in pbxmutations, showing a non-cell-autonomous effect that is perhaps due to an increase in the numbers of cells expressing hh. However, the major control of dpp expression by Ubx is cell autonomous.
Ubx also governs Dpp spread and activity. A mechanism whereby Ubx may limit the spread of Dpp is by reducing dallyexpression. The protein encoded by this gene seems to be required to transmit the Dpp protein from cell to cell (Belankaya et al., 2004), and we have shown that Ubx downregulates dally expression, thus reducing the extent of Dpp activity. Ubx retards Dpp spread also by augmenting tkv expression (mainly at the A/P boundary and not in the periphery)because high Tkv levels retain the Dpp morphogen (Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000)(Fig. 8B,C). In Ubx- clones, the Dpp product can `travel' more readily through the mutant cells, thus extending Dpp signaling not only within the clone but also in more distal cells (Fig. 8D). If we elevate tkv expression in these clones, Dpp spread is checked, preventing the non-cell autonomous and reducing the extent of the cell-autonomous effect on Dpp signaling(Fig. 8E).
The increased tkv transcription and the suppression of dally expression have a double effect. On the one hand, they reduce Dpp spread; on the other hand, they increase Dpp activity in AB cells. In the wing disc, Hh signaling strongly diminishes Dpp signaling in this domain by inducing mtv and dally transcription; this reduction is required for the correct pattern of the central region of the wing and for substantial dally and mtv expression(Tanimoto et al., 2000; Fujise et al., 2001; Funakoshi et al., 2001; Fujise et al., 2003) (this report). By contrast, our results demonstrate that Ubx allows high levels of both Dpp and Hh activity in these cells of the haltere disc, and that repression of mtv and dally is not maintained by this high Dpp signaling. This suggests that a different mechanism for patterning this domain is acting in haltere and wing discs. Finally, the Ubxmodulation of Dpp activity is complex. Whereas Ubx prevents Dpp signaling from downregulating tkv or mtv in the periphery of the haltere pouch, in MS1096; UAS-tkvQD haltere discs dpp-lacZ expression in the dorsal region is completely suppressed (not shown).
Haltere size control by the Ubx product
Several lines of evidence argue that differences in dpp, tkv and probably dally expression, all of them controlled by Ubx,may be instrumental in reducing the size of the haltere disc compared with that of the wing disc: first, Ubx downregulates dpptranscription, and the increased expression of dpp augments the size of haltere discs (see also Mohit et al.,2006); second, Ubx increases tkv expression, and the ectopic Ubx expression, or the elevated tkvtranscription, reduces wing size (Haerry et al., 1998; Lecuit and Cohen, 1998; Tanimoto et al., 2000) (this report); third, the ectopic expression of mtv or dally (both increasing Dpp spread) in the posterior compartment of the haltere disc substantially increases its size; fourth, the reduction of tkvexpression increases the haploinsufficient phenotype of the Ubx locus and, conversely, reduced Ubx levels partially rescue the small halteres of dpp hypomorphic mutations; finally, the increased size of the haltere disc observed in pbx or bx mutations is partially suppressed if Tkv levels are increased.
The control of size by Ubx relies on the reduction of dppexpression and Dpp spread. Thus, it is not surprising that it involves non-cell-autonomous effects. We have shown that if Ubx is removed from the central region of the haltere pouch, this domain is transformed into wing, but the remaining haltere tissue is bigger than expected. This occurs with Gal4 lines that drive a dsRNA>Ubx construct or when Ubx mutant clones are induced in the anterior compartment of the haltere disc. These results suggest that Dpp spread is increased within the mutant region so that more Dpp reaches the distal haltere domain. As a result,differences in Dpp activity between adjacent cells extend over a larger domain, and both the region that is transformed into wing and the tissue that remains as haltere, increase their size. The growth control is, in part,non-cell autonomous, but the differentiation is strictly cell autonomous(Morata and García-Bellido,1976; Hart and Bienz,1996; Roch and Akam,2000). A non-cell autonomous role of Ubx on organ size has also been described in the development of the third leg(Stern, 2003). Our observations may explain some previous results: first, in pbx and bx mutants there is a slight increase in the size of the untransformed compartment compared with wild-type flies(González-Gaitán et al.,1990), perhaps because dpp expression is higher. Second,if wing and haltere tissues are confronted in the wing disc of Contrabithorax mutations (which partially transform wing into haltere), the haltere (transformed) tissue is also bigger than expected(González-Gaitán et al.,1990), possibly because dpp expression and spread are increased. Third, by changing the activity of Ubx during development with the temperature-sensitive bx1 mutation, halteres bigger than normal are observed(Sánchez-Herrero and Morata,1983), maybe because the initial growth (wing-like) and the posterior haltere differentiation are relatively uncoupled. However, although we stress the role of the Dpp pathway in regulating the size of the haltere,we are aware that other factors are also likely to contribute to this effect. For instance, wingless is not expressed in the posterior compartment of the haltere disc, and this absence has been correlated with its small size(Weatherbee et al., 1998).
It is unclear how these effects on the Dpp pathway are translated into a reduction in disc size as cell size is similar in both discs(Roch and Akam, 2000). Recently, Rogulja and Irvine (Rogulja and Irvine, 2005) have proposed a model to account for the proliferation in the wing disc. This model proposes that differences in Dpp signaling in the medial and lateral regions of the wing disc induce non-cell-autonomous proliferation for a short time. A similar mechanism may exist in the proliferation of the haltere discs because both discs give rise to homologous structures and rely on similar patterning cues. In the haltere disc, both the lower amount of Dpp synthesized and its lower spread result in a more narrow and sharp Dpp activity gradient. We have shown that a gradient of Dpp-GFP signal is established in medial regions of the wing but not the haltere discs. This will extend the differences in Dpp signaling between adjacent cells further in the wing disc, perhaps allowing for more cells to enter division at early larval stages. Madhavan and Schneidermann(Madhavan and Schneidermann,1977) reported a slight delay in cells of the haltere disc reassuming division after embryogenesis compared with the wing disc, and a somewhat bigger cell-doubling time at the beginning of the second instar. Previous results indicated that the variation in clone size is similar in haltere or wing discs throughout larval stages(Morata and García-Bellido,1976); in this experiment, however, it was assumed that each haltere cell secreted one trichome whereas a later study showed that haltere cells can secrete more than one (Roch and Akam, 2000). Our results indicate that Ubx does not seem to delay cell division autonomously, and that it is necessary to mutate a big region of the haltere disc to observe a clear size difference.
In the grasshopper, the increased expression of dpp in the metathoracic legs has been suggested to account for the larger size of these appendages (Niwa et al.,2000), and we propose that changes in dpp transcription and Dpp spread underlie size differences between halteres and wings. The regulation of morphogen levels and of proteins that limit the movement of the ligand (such as Tkv and Dally) by Hox genes may be a general mechanism used in evolution to differentiate size in homologous organs within a certain animal,or between homologous organs in different species. Because the Dpp pathway also controls pattern, Ubx may differentiate the size and pattern in halteres and wings, coordinating both processes by a single mechanism through the change in the amount and distribution of the Dpp product.
Note added in proof
Similar results to those described here have been recently reported by Makhijani, K., Kalyani, C., Srividya, T. and Shashidhara, L. S. Modulation of Decapentaplegic morphogen gradient during haltere specification in Drosophila. Dev. Biol. (in press).
We thank L. S. Shashidhara and K. Makhijani for sending stocks, RNA probes and communicating results before publication; G. Morata for critically reading the manuscript and for support; and F. A. Martín for discussions and help with the experiments. We also thank T. Tabata for the unpublished UAS-mtv flies; M. Akam, K. Basler, S. Baumgartner, M. Calleja, S. Carroll, S. Cohen, X. Franch-Marro, A. García-Bellido, M. González-Gaitán, I. Guerrero, A. Macías, H. Nakato, G. Marqués, M. Milán, L. Perrin, R. Terracol, R. White and the Bloomington Stock Centre for stocks, probes or antibodies; and R. González for technical assistance. This work was supported by grants from the Dirección General de Investigación Científica y Técnica (N° BMC 2002-00300), the Comunidad Autónoma de Madrid (N° GR/SAL/0147/2004) and an Institutional Grant from the Fundación Ramón Areces. L.d.N. is supported by a fellowship from the Spanish Ministerio de Educación y Ciencia and D.L.G. by a I3P fellowship, co-financed by the European Social Fund, from the C.S.I.C.