Reciprocal roles for bowl and lines in specifying the peripodial epithelium and the disc proper of the Drosophila wing primordium

The Drosophila wing imaginal disc is an excellent model with which to investigate the fundamental mechanisms that generate a complex structure from an undifferentiated field of cells. One of the earliest events in wing development is its subdivision into two apposing surfaces, known as the peripodial epithelium (PE) and the disc proper (DP)(Cohen, 1993; Fristrom and Fristrom, 1993; Milner, 1984). Despite the central role of this symmetry-breaking event in setting up two fields of cells with distinct morphologies and developmental potentials, the mechanisms that generate this major asymmetry remain poorly understood.

The formation of the wing appendage depends on the establishment of the anteroposterior (AP), dorsoventral (DV) and proximodistal (PD) axes within the DP (reviewed by Blair, 1995; Dahmann and Basler, 1999; Klein, 2001; Lawrence et al., 1996; Mann and Morata, 2000). The AP and DV subdivisions are established by the transcription factors engrailed (en) and apterous (ap)(Blair et al., 1994; Diaz-Benjumea and Cohen, 1993; Lawrence and Morata, 1976; Morata and Lawrence, 1975),which program their respective compartment to induce expression of the Decapentaplegic (Dpp) and Wingless (Wg) morphogens adjacent to the AP and DV compartment boundaries (Basler and Struhl,1994; Diaz-Benjumea and Cohen,1995; Fleming et al.,1997; Panin et al.,1997; Tabata and Kornberg,1994; Williams et al.,1994; Zecca et al.,1995). Dpp and Wg, in turn, induce expression of target genes to pattern the AP and DV axes in a concentration-dependent manner(Lecuit et al., 1996; Nellen et al., 1996; Neumann and Cohen, 1996; Zecca et al., 1996). During the second instar, the wing undergoes a third major subdivision along the PD axis to form the wing distally and the notum proximally; this subdivision is mediated by the secreted Wg ligand and the Epidermal Growth Factor (EGF)receptor ligand, Vein. Wg is expressed in a wedge of ventroanterior cells(Couso et al., 1993; Klein and Arias, 1998; Ng et al., 1996; Williams et al., 1993), and induces the expression of the pdm1 homolog nubbin(nub) to initiate wing formation(Cifuentes and Garcia-Bellido,1997; Ng et al.,1995; Ng et al.,1996). Vein emanates from a patch of proximal cells to induce expression of pannier and members of the Iroquois-C gene complex to specify the medial and lateral parts of the notum(Simcox et al., 1996; Wang et al., 2000; Zecca and Struhl, 2002a; Zecca and Struhl, 2002b). As development proceeds, the wing field is further subdivided into the pouch distally, and the hinge proximally. The transcription factors vestigial (vg) and homothorax (hth), teashirt (tsh) and zfh2 are expressed in discrete subregions of the wing pouch and hinge, respectively, and control the identity of these subregions. (Mann and Morata,2000; Kim et al.,1996; Williams et al.,1991; Azpiazu and Morata,2000; Casares and Mann,2000; Whitworth and Russell,2003; Wu and Cohen,2002).

The contribution of the PE to wing growth and patterning has received far less attention. The PE has been implicated in disc eversion and fusion of adjacent discs to form a continuous adult cuticle during metamorphosis(Agnes et al., 1999; Fristrom and Fristrom, 1993; Milner et al., 1984; Pastor-Pareja et al., 2004; Usui and Simpson, 2000; Zeitlinger and Bohmann, 1999). Selective ablation of the PE resulted in smaller and malformed wings(Gibson and Schubiger, 2000),and inhibition of specific signaling pathways in the eye and wing PE resulted in patterning abnormalities and a reduction in disc size(Cho et al., 2000; Gibson and Schubiger, 2000; Pallavi and Shashidhara,2003), suggesting earlier roles for the PE in signaling to the DP to control its growth and patterning (reviewed by Gibson and Schubiger, 2001). Lineage analysis have suggested that peripodial cells stream laterally to populate the DP (McClure and Schubiger,2005; Pallavi and Shashidhara,2003). Despite these intriguing results, little is known about how the PE and the DP are specified or how they interact with each other.

Drumstick (Drm), Odd-skipped (Odd), Bowl, and Sister of Odd and Bowl (Sob)share a conserved Cys₂His₂ zinc-finger domain, and play diverse roles in patterning ectodermal structures. Bowl is a putative transcription factor (Wang and Coulter,1996). During embryogenesis, the protein Lines binds to and destabilizes Bowl (Hatini et al.,2005). Drm is a small peptide that binds to Lines and localizes it to the cytoplasm, permitting stabilization of Bowl in restricted domains (see Fig. 1L for the regulatory interactions connecting drm, lines and bowl)(Green et al., 2002; Hatini et al., 2005). During larval development, bowl contributes to leg and eye development(Bras-Pereira et al., 2006; de Celis Ibeas and Bray, 2003; Hao et al., 2003). drm mutants, however, do not display phenotypes in these tissues,suggesting that odd and/or sob, which are related to drm in structure and pattern of gene expression, may sometimes act redundantly with drm(Bras-Pereira et al., 2006; Hao et al., 2003). Although gain-of-function experiments support this hypothesis, loss-of-function and biochemical evidence are still lacking. Lines, however, destabilizes Bowl in all tissues examined (Green et al.,2002; Hatini et al.,2005; Iwaki et al.,2001; Johansen et al.,2003; Bras-Pereira et al.,2006; Hao et al.,2003).

We now report that lines and bowl specify alternative DP and PE fates during early stages of wing development. By blocking the specification of the PE, we were further able to provide definitive evidence that the PE is not required for the establishment of the AP, DV or PD patterning systems. The PE is instead required to support the growth and survival of the DP and the formation of the notum. Overall, our work elucidates a mechanism that generates a major asymmetry across the wing primordium and reveals crucial roles for this asymmetry in wing growth and patterning.

Mitotic clones were induced using the FLP/FRT(Golic, 1991; Xu and Rubin, 1993) and the MARCM techniques (Lee and Luo,2001) at 24-36, 36-48, 48-72, 72-96 and 96-120 hours AEL, which correspond to early first, late first, second, early third and mid third instar. Flies of the genotype y w hs-FLP; FRT42D Ubi-GFP and y w hs-FLP; Ubi-GFP FRT40A (B. Edgar) were used to induce FLP/FRT clones and y w hs-FLP Tub-GAL4 UAS-GFP-6Xmyc-NLS; FRT42D Tub-Gal80 hs-CD2,y⁺ (gift of G. Struhl) to induce MARCM clones. The lines^G2 (Bokor and DiNardo, 1996), lines^2f(Nusslein-Volhard et al.,1984), bowl¹(Wang and Coulter, 1996) and drm³ (Green et al.,2002) alleles were used to generate mutant clones. wg-lacZ (Kassis et al.,1992) and P{lacZ}odd^rk111 were used to map domains of gene expression. UAS-Lines (9.2), UAS-Myc-Lines(8), UAS-Flag-Bowl (28)(Hatini et al., 2005), UAS-Drm (2.1) (Green et al.,2002), UAS-LinesRNAi (VDRC), were expressed in clones using y w hs-FLP; act5C>y+>GAL4 UAS-GFP(Pignoni and Zipursky, 1997),and in the PE using Ubx-GAL4(Pallavi and Shashidhara,2003).

Staining protocols have been described elsewhere(Hatini et al., 2005). Primary antibodies used were: rabbit anti-Vg (Kim et al., 1996), mouse anti-Nub(Ng et al., 1996), mouse anti-Dll (Vachon et al.,1992), mouse anti-Sal (gift of S. Cohen), rabbit anti-Tsh(Wu and Cohen, 2000), rat anti-Zfh2 (gift of M. Lundell), rabbit anti-β-galactosidase (Cappel),mouse anti-Wg (4D4, DSHB) (Brook and Cohen, 1996), rabbit anti-Bowl(de Celis Ibeas and Bray,2003), mouse anti-Ubx (White and Wilcox, 1984), rabbit anti-Hth(Pai et al., 1998), guinea pig anti-Hth (Casares and Mann,1998), rabbit anti-Al (gift of G. Campbell), rabbit anti-activated caspase 3 (Cell Signaling Technology), mouse anti-Arm (7A1, DSHB) and rabbit anti-phospho Histone H3 (Upstate Signaling). Confocal images were scanned using a Zeiss LSM510 in multi-tracking mode. β-Galactosidase activity stains are described elsewhere (Patel et al., 1989).

To understand the role of bowl in wing development, we examined the dynamic pattern of Bowl distribution relative to Wg. During the second instar, Wg accumulates in a wedge of ventro-anterior cells in the DP(Fig. 1B, arrow)(Couso et al., 1993; Ng et al., 1996; Williams et al., 1993). At this stage, Bowl accumulated only in peripodial cells(Fig. 1A), and not in DP cells(Fig. 1B). At the early to mid third instar, Wg was upregulated along the DV compartment boundary(Fig. 1D,F, arrows). Bowl,however, was detected in the PE (Fig. 1C,E) and in the lateral margins of the notum (arrowheads in Fig. 1D,F). At the completion of larval development, Bowl was upregulated along the lateral margins of the notum (arrowheads in Fig. 1H)and the ventro-anterior margin of the PE (asterisks in Fig. 1G,H). However, Bowl was downregulated in the medial region of the PE and absent from its posterior margin. After pupariation, Bowl was detected at high levels in both the ventro-anterior hinge (asterisk in Fig. 1I) and the lateral margins of the notum (data not shown). An odd-lacZ reporter that is expressed similarly to Bowl(Fig. 1J) was detected in the axillary region and in a subset of proximal hinge derivatives in ventral nuclei in adult wings (Fig. 1K), suggesting that a subset of odd-lacZ- and Bowl-expressing cells are fated to contribute to the ventral hinge.

drm is expressed in restricted domains, where it inhibits Lines,in part, by localizing it to the cytoplasm(Hatini et al., 2005). To probe the pattern of Lines activity in the wing imaginal disc, we induced expression of a `weak' Myc-Lines transgene that minimally affects wing development (UAS-Myc-Lines 8) in `FLP-out' cell clones using a combination of the FLP/FRT and the GAL4/UAS techniques(Pignoni and Zipursky, 1997). Myc-Lines accumulated in the cytoplasm in clones in the lateral margins and the medial region of the PE (Fig. 2A,B, respectively). However, Myc-Lines accumulated in nuclei in the DP (Fig. 2C). Thus, the distribution of Lines and Bowl in the PE and the DP was reciprocal. In the PE where Bowl accumulates, Lines was cytoplasmic and inactive. Reciprocally, in the DP where Bowl was absent, Lines was nuclear and active.

We also expressed a Flag-Bowl transgene in Flp-out clones to determine whether Bowl was destabilized in the DP. Indeed, Flag-Bowl accumulated only in cell clones that were generated in the medial and ventro-anterior margin of the PE where Bowl is normally detected(Fig. 2D-D′; arrow and arrowhead in D′, respectively). However, Flag-Bowl was not detected in DP clones, suggesting that endogenous Lines destabilized Bowl in this region. To test this idea, we examined Bowl accumulation in lines mutant clones and in drm-expressing clones and detected a stabilization of Bowl in these clones (Fig. 2E,F, respectively). Reciprocally, Bowl was destabilized in lines-expressing cell clones generated in the PE(Fig. 2G). However, Bowl accumulation was unaffected in drm mutant clones generated in the PE(data not shown), consistent with the proposal that odd and/or sob may sometimes act redundantly with drm to stabilize Bowl(Bras-Pereira et al., 2006; Hao et al., 2003). We were unable to test sob function as sob mutant alleles are not available. However, Bowl accumulation was unaffected in odd mutant clones. We also attempted to induce homozygous clones for the drm(P2)deficiency to generate drm odd sob triple mutant clones but failed to recover these clones. Therefore, additional studies will be necessary to define the multiple inputs necessary to stabilize Bowl in the PE.

To investigate the role of lines in wing development, we next overexpressed a lines transgene that strongly destabilizes Bowl(UAS-Lines 9.2) in the PE where Lines is repressed, or removed lines function in the DP where Lines is active. First, we broadly expressed lines with Ubx-GAL4(Pallavi and Shashidhara,2003) to broadly destabilize Bowl in the PE using the GAL4/UAS technique (Brand and Perrimon,1993). We found that the flattened morphology of wild-type wing discs (Fig. 3A) was replaced with a spherical morphology (Fig. 3B), in which the thin PE (Fig. 3A′,C,D′-D‴) was replaced with the thickened pseudostratified CE that forms the DP (Fig. 3B′,E′-E‴). Coincidentally, the expression of the peripodial markers Bowl, Ubx and Eya was lost in these discs(Fig. 4B,C; data not shown). Conversely, Lines was broadly nuclear (Fig. 4D), indicating that the PE (where Lines is enriched in the cytoplasm) was replaced with the DP (where Lines is enriched in nuclei). The Ubx>Lines wing discs were also roughly 10-30% the size of wild-type wing discs (compare Fig. 3A with 3B; experimental discs, 39 mm², n=7, s.d.=18 mm²; wild type, 147 mm², n=7, s.d.=21 mm²). These discs survived beyond the larval stages and elongated during metamorphosis (see Fig. S1 in the supplementary material) but failed to contribute to adult structures. We obtained similar phenotypes by driving expression of UAS-Lines with C311-GAL4 and Tsh-GAL4 (see Fig. S2 in the supplementary material).

Organ growth depends largely on a balance between cell proliferation and cell death. Therefore, to determine the reason for the reduction in wing size,we examined the expression of Phospho-Histone H3 (PH3) and activated caspase 3(Ice) in Ubx>Lines discs to detect dividing and apoptotic cells, respectively (Ryoo et al.,2002). We detected similar levels of proliferating cells(Fig. 3G), but elevated levels of apoptotic cells in these discs (Fig. 3F) compared with age-matched controls (not shown). We therefore propose that the PE supports wing growth by promoting cell survival in the DP.

We considered the possibility that the reduction in wing size resulted from the loss of positional identities along the wing PD axis. We therefore analyzed Vg (Kim et al., 1996; Williams et al., 1991), Nub(Cifuentes and Garcia-Bellido,1997; Ng et al.,1995; Ng et al.,1996) and Zfh2 expression(Whitworth and Russell, 2003)to determine whether the pouch, the hinge or the notum were affected. In wild type, Vg is restricted to the distal pouch, whereas Zfh2 is restricted to the hinge (Fig. 4E). Nub is detected in the pouch and in part of the dorsal hinge(Fig. 3D,D″,D‴). In experimental discs, Nub (Fig. 3E,E″-E‴), Vg and Zfh2(Fig. 4F,J) were detected in both layers of the disc epithelium, indicating that the PE was replaced with a mirror image duplication of the DP. Only a small region near the disc stalk did not express Zfh2 (arrowhead in Fig. 4F), indicating that the notum, which is Zfh2 negative, was nearly lost. The reduction in notal growth could partially account for the reduction in overall wing size.

We also considered the possibility that the reduction in wing growth resulted from the disruption of the AP or the DV patterning systems. We therefore examined the expression of the Wg(Fig. 4G-H) and Dpp (see Fig. S3 in the supplementary material) morphogens and their respective transcriptional targets Dll (Fig. 4I,J) and Sal (Fig. 4K,L). In wild type, Wg is detected in the pouch along the DV compartment boundary, in the hinge in two concentric rings, and in the notum in a band that transverses the AP axis(Fig. 4G). Wg and Dpp, in turn,respectively induce expression of Dll (Fig. 4I) and Sal (Fig. 4K) in broad domains within the pouch(Lecuit et al., 1996; Nellen et al., 1996; Neumann and Cohen, 1996; Zecca et al., 1996). In experimental discs, Wg was detected in both layers of the disc epithelium, in ring-like structures that encircled the pouch (short arrow) and the hinge(long arrows) (Fig. 4H). dpp was expressed orthogonally to Wg (see Fig. S3 in the supplementary material). Dll and Sal were also detected in both layers of these discs (Fig. 4J,L,respectively), indicating that the Wg and Dpp morphogens were able to signal broadly to induce their targets. We thus conclude that at least part of the AP, DV and PD axes are established independently of the PE. These findings imply that the PE supports wing growth in parallel to these systems, and plays a crucial role in promoting the growth of the notum.

We also induced expression of lines in FLP-out clones to determine whether lines eliminates or transforms the PE into DP. If lines transforms the PE into DP, then it should be possible to recover large lines-expressing clones in wing rudiments that lack PE. If, however, lines overexpression eliminates the PE then it should be possible to recover wing rudiments in which the lines-expressing clones were lost together with the PE. Ectopic expression of linesduring the early first instar gave rise to two general classes of wing abnormalities. In the most severe class, the PE was replaced with a mirror image duplication of the pseudostratified CE, which correlated with the expression of the DP markers Nub and Tsh in both layers of the wing disc(z-section in Fig. 5D). These discs were spherical and as small as the Ubx>Lines discs described above. The clones that gave rise to this phenotype formed large confluent patches that occupied roughly 56% of the entire wing surface (n=7, s.d.=17%). As wing rudiments lacking large patches of lines-expressing clones were not recovered(n=58 rudimentary discs examined), we were forced to conclude that the presumptive peripodial cells survived in experimental disc and assumed an inappropriate DP fate. In the second class, the discs formed secondary wing fields (Fig. 5E,F; arrows in`merged' images point to secondary wings; small arrowheads in red channel point to the two wing margins). The notum appeared intact in some discs(Fig. 5E) but reduced or missing in others (Fig. 5F,asterisk in red channel indicates the loss of Wg expression in the notum). The clones that gave rise to this phenotype adopted DP fate (z-section in Fig. 5F), and occupied ∼26%of the surface area of the entire wing (n=9, s.d.=5%) and roughly 50%of the surface area of the secondary wing fields. To further clarify the origin of clones that permitted induction of secondary wings, we expressed UAS-Lines with Ptc-GAL4 in both the DP and the PE (marked by apposing arrowheads) along the AP compartment boundary. The Ptc>Lines-expressing cells in the PE adopted DP fate and permitted the induction of secondary wing (Fig. 5H). We conclude that smaller patches of lines-expressing clones generated in the PE adopted DP fate, and subsequently acquired the competence to respond to wing inducing signals. Clones that did not disrupt wing development localized to the DP even when they formed large confluent patches (Fig. 5C). Taken together, we interpret these results as evidence that lines can reprogram the PE to adopt DP fate during early stages of wing development.

To investigate the contribution of bowl to wing development, we induced marked bowl mutant clones and examined clone recovery relative to wild-type twin spots generated by the same mitotic recombination event. Control clones and their twin clones survived in both the PE and the DP(Fig. 6A; PE nuclei are spread out; DP nuclei are densely packed). Most of the bowl mutant clones generated at the early first instar survived in the DP. However, only 50% of clones survived in the PE mostly near the disc stalk(Fig. 6F, arrow indicates a stalk clone) indicating that the bowl mutant clones were either dying or sorting out from the PE at early stages. bowl mutant clones generated at the second instar survived in the PE(Fig. 6G) and adopted PE fate,suggesting that the maintenance of PE fate depends on additional mechanisms.

We also induced expression of bowlRNAi transgene in FLP-out clones at first larval instar to determine if reduced levels of bowlaffected the survival or the fate of the PE. Clonal expression of bowlRNAi gave rise to two general classes of wing abnormalities that phenocopied the phenotypes induced by ectopic expression of lines. Large patches of clones led to the formation of small and spherical discs in which the PE was replaced with DP. Smaller clones permitted the induction of secondary wings (data not shown). The bowlRNAi transgene can also knockdown the expression level of three other genes, odd, sob and not1, indicating that bowl, either alone or together with odd or sob or both genes, specifies PE fate and inhibits DP fate.

We also examined the recovery of lines mutant clones induced at the first instar. lines mutant and wild-type twin spots survived in the PE (Fig. 6B). However,mostly wild-type twin spots survived in the DP(Fig. 6C), and were much smaller than respective wild-type twin spots. Similar to the linesmutant clones, drm-expressing clones generated at the first instar survived in the PE but not in the DP (Fig. 6E). We therefore conclude that at early stages of wing development bowl promotes cell survival in the PE, whereas lines promotes cell survival in the DP. The quantitative analysis is presented in Fig. 6G.

lines mutant clones generated at the second instar survived poorly in the DP relative to wild-type twin spots(Fig. 6G). The surviving clones, however, extruded basally from the DP(Fig. 7). Only clones that were generated in the PE, the lateral margins of the notum and the ventro-anterior hinge intermingled freely with their wild-type neighbors(Fig. 6B; data not shown). Occasionally, clones that originated in the DP grew to a large size. We examined the positional identity of these clones with molecular markers. We found that the clones lost expression of the DP specific proteins Nub and Vg(Fig. 7A,B, respectively) and ectopically expressed the PE specific proteins Ubx and Eya(Fig. 7C,D). Ubx is restricted to the posterior compartment, and was detected in the DP only in lines mutant clones generated in the posterior compartment(McClure and Schubiger, 2005; Pallavi and Shashidhara,2003). Tsh and Hth, which are expressed in the PE as well as in the notum and hinge, but are excluded from the pouch, were also ectopically expressed in the clones (Fig. 7E-F). We conclude that lines is necessary to specify DP fate and to inhibit the specification of PE fate at early stages of wing development.

lines could be needed either continuously or transiently in the DP to specify DP fate and to inhibit PE fate. Moreover, lines could play yet another role at later stages of wing development. To address these issues,we examined the behavior of lines mutant clones and drm-expressing clones that were generated at the mid to late third instar following the formation of the PE and the DP. Most of the clones generated in the DP and the posterior lateral margin of the hinge minimized contact with their wild-type neighbors and formed round vesicles that extruded basally (Fig. 8A-A″,B′,C-H). We examined the positional identity of these clones with molecular markers to determine whether the clones assumed an alternative cell fate. The peripodial markers Ubx and Eya were not detected in the clones, indicating the clones did not assume PE fate (data not shown). However, the Tsh, Hth, Wg and Zfh2 proteins, which localize to the hinge and control hinge formation, were ectopically expressed in the clones(Fig. 8C-E; data not shown). Expression of these markers was lower near the AP compartment boundary and increased at a distance from this boundary, suggesting that Dpp signaling antagonized this fate transformation in a graded manner, as previously proposed (Azpiazu and Morata,2000; Casares and Mann,2000). Hinge-specific markers were also induced in linesRNAi clones generated in the pouch, but were either downregulated or lost in linesRNAi clones lacking bowlfunction, indicating that the lines clonal phenotypes were due to the stabilization of Bowl (see Fig. S4 in the supplementary material). nub, which localizes to both the pouch and the distal hinge, was maintained in these clones (Fig. 8F). Reciprocally, the pouch specific proteins Dll and Sal were lost in lines mutant clones generated in the pouch in both distal and proximal positions (Fig. 8G,H,respectively). Overall, these results indicate that lines and bowl are required at early stages to specify the DP and the PE,suggesting that the maintenance of these fates depends on additional mechanisms. However, they reveal a later role for lines in maintaining distal pouch identities and in inhibiting the specification of proximal hinge identities.

The generation of molecular asymmetries across the wing primordium is central to the control of wing growth and patterning. Early during development, the two surfaces of the flattened wing imaginal disc assume distinct DP and PE fates, and hence distinct morphologies and developmental roles. Our data defines, for the first time, a mechanism that establishes this asymmetry. We show that lines and bowl act as a binary switch to specify DP and PE fates, respectively. By blocking the specification of the PE we were further able to explore the earliest requirement of the PE to wing growth and patterning.

The wing PE can be identified molecularly and morphologically as a thin epithelial sheet overlying the thickened DP epithelium(Baena-Lopez et al., 2003; Cho et al., 2000; Gibson and Schubiger, 2000; McClure and Schubiger, 2005; Pallavi and Shashidhara,2003). Our mapping studies show that the distribution of Bowl and Lines correlates with the establishment of this asymmetry (Figs 1, 2). The wing primordium inherits its subdivision into en-expressing cells that form the posterior compartment and adjacent anterior compartment cells from the embryonic epidermis (Cohen et al.,1991; Cohen,1990). Bowl accumulates in the posterior en-expressing cells in the embryonic epidermis (Hatini et al., 2005), suggesting that the wing primordium also inherits the PE/DP subdivisions from pre-existing asymmetries across this tissue.

lines and odd-skipped genes act as a switch to specify alternative cell fates across fields of cells(Fig. 1L). bowl and lines specify the alternative 1°-3° and 4° cell fate across the dorsal embryonic epidermis(Bokor and DiNardo, 1996; Hatini et al., 2000; Hatini et al., 2005). bowl and lines also specify alternative cell fates in the developing gut, leg and eye imaginal discs(Iwaki et al., 2001; Johansen et al., 2003; Bras-Pereira et al., 2006; de Celis Ibeas and Bray, 2003; Hatini et al., 2005). The asymmetric distributions of Bowl and odd-skipped genes, and the reciprocal distribution of Lines in the wing primordium are also used to specify the alternative PE and DP fates. Indeed, our functional studies show that ectopic lines expression or the inhibition of bowlfunction in the PE transforms PE into DP fate. Reciprocally, the removal of lines function from the DP transforms DP into PE fate. Our data further suggest that lines exerts its function by controlling the stability of the Bowl protein (see Fig. S4 in the supplementary material). Thus, lines and bowl act as a switch to specify alternative DP and PE fates across the wing primordium. The distribution of Lines and Bowl correlates with the subdivision of the wing primordium into a thin squamous and a thickened columnar epithelial sheet. The activation of EGF receptor and Wg signaling in the DP may specify the formation of a columnar epithelial morphology (Baena-Lopez et al.,2003). The pathways that specify the squamous morphology of the PE downstream to bowl remain to be elucidated.

Previous studies that relied on surgical and genetic ablations of the PE,and on inhibition of certain signaling pathways within the PE, suggested important roles for the PE in disc growth and patterning(Baena-Lopez et al., 2003; Cho et al., 2000; Gibson and Schubiger, 2000; McClure and Schubiger, 2005; Pallavi and Shashidhara,2003). We were now able to examine wing development in discs lacking PE. These discs were significantly smaller than wild type(Fig. 3B,E; Fig. 4B-D,F-L), and the notum was dramatically reduced in size relative to the pouch and hinge(Fig. 4F). Progenitor cells that originate in the PE may stream laterally to populate the growing notum(McClure and Schubiger, 2005; Pallavi and Shashidhara,2003), and the loss of this progenitor cell population may account for the severe reduction in notal growth. The reduction in wing growth could have resulted from the disruption of Wg or Dpp signaling activities, as these morphogens control cell survival and cell proliferation in the wing(Burke and Basler, 1996; Gibson and Perrimon, 2005; Giraldez and Cohen, 2003; Johnston and Sanders, 2003; Martin-Castellanos and Edgar,2002; Moreno et al.,2002; Neumann and Cohen,1996; Shen and Dahmann,2005). Indeed, a block to Dpp or Wg signaling results in formation of tiny wing rudiments (Adachi-Yamada et al., 1999; Bryant,1978; Couso et al.,1993; Morata and Lawrence,1977; Sharma and Chopra,1976). However, the expression of Wg and Dpp and their target genes was normal in these discs, indicating that the reduction in wing growth was not a consequence of the loss of wg or dpp expression,or of signaling activities (Fig. 4J,L). Our findings instead suggest that the PE acts in parallel to the AP, DV and PD patterning systems, in part, by promoting cell survival in the DP, and in part by promoting the growth of the notum.

The results presented in this study argue that lines and bowl function as field-specific selector genes to specify the identity and the behavior of the DP and the PE of the wing primordium,respectively. Selector genes control cell fate, cell affinity and the competence to send and respond to patterning signals in a cell autonomous manner (reviewed by Mann and Morata,2000). lines and bowl act cell autonomously to control the fate, affinity and the interaction between the PE and the DP. As a putative transcription factor, Bowl may regulate a developmental program to control the identity and the behavior of the PE. By inhibiting Bowl accumulation, lines may allow the execution of an alternative developmental program in the DP. Our studies define a new system to identify,in a systematic way, these developmental programs.

We thank S. DiNardo for his support and crucial insight at the early stages of this work; S. Bray, S. Cohen, S. Carroll, G. Campbell, M. Lundell, R. Mann,G. Struhl, H. Sun and R. White for providing reagents; and K. Commons, P. Jou,L. Zeng, E. Kula and N. Neuman for critically reading the manuscript. We thank the Bloomington Stock Center, the Vienna Drosophila Research Center(VDRC) and the Developmental Studies Hybridoma Bank for fly stocks and antibodies, respectively. This work was supported by the National Institute of Health grants (GM068069) to V.H. and (T32HD007403) to D.N.