Mutual repression between msh and Iro-C is an essential component of the boundary between body wall and wing in Drosophila

During development, the allocation of cell populations to different developmental fates relies largely on the establishment of developmental boundaries. These boundaries separate adjacent cell populations and provide cells at both sides of the boundary with positional information that governs the future patterning of the tissue. In Drosophila, developmental boundaries have been best characterized in the mesothorax and wings. These parts of the fly body develop from a pair of epithelial sacs, the imaginal wing discs. At metamorphosis, each imaginal disc undergoes final differentiation to give rise (from dorsal to ventral) to heminotum, dorsal wing hinge, dorsal wing blade, ventral wing blade, ventral wing hinge and mesopleura. The pair of heminota and mesopleurae form the mesothoracic body wall, while the remaining structures form the pair of wings.

In the wing disc, the best characterized developmental boundaries are associated with borders of compartments defined by cell-lineage restrictions(García-Bellido et al.,1973) (reviewed by Irvine and Rauskolb, 2001; Mann and Morata, 2000). The earliest boundary subdivides the disc into anterior (A) and posterior compartments (P). It is established by the expression of the selector genes engrailed and invected in the P compartment. It has been proposed that the selector genes confer identity to the cells of a compartment and a differential affinity that prevents cells from apposing compartments to intermingle(García-Bellido and Santamaría, 1972). The end result is a straight boundary that separates both compartments and which, after final differentiation, does not correspond with any morphological feature. A dorsoventral (DV)compartmental subdivision, orthogonal to the AP one, is established by the expression of the gene apterous (ap) in the D compartment. In the wing, this boundary runs along the wing margin and separates the dorsal and ventral wing blades. Compartment borders give rise to specialized cells that are sources of signaling molecules that organize both cell proliferation and patterning of the entire disc (for reviews, see Brook et al., 1996; Teleman et al., 2001; Vincent and Briscoe,2001).

Other developmental boundaries are not associated with cell lineage restrictions (reviewed by Irvine and Rauskolb, 2001; Mann and Morata, 2000; Tepass et al.,2002). This implies that cells can cross the boundary and change fates. An example is the boundary that separates the presumptive notum from the dorsal wing hinge territory (Diez del Corral et al., 1999). A selector-like role has been attributed to the genes araucan, caupolican and mirror(Cavodeassi et al., 1999; Cavodeassi et al., 2000; Wang et al., 2000), the three members of the Iroquois Complex (Iro-C)(Gómez-Skarmeta et al.,1996; McNeill et al.,1997). These genes, which encode related homeodomain proteins conserved from worms to vertebrates (reviewed by Cavodeassi et al., 2001), start to be expressed in the presumptive notal region during the second instar. This expression is essential for notum specification, as clones of Iro-C^– cells induced early within this territory acquire the identity of the dorsal wing hinge (Diez del Corral et al., 1999). Thus, the early domain of expression of Iro-C defines the extent of the notum territory. Moreover, Iro-C genes appear to endow cells with special affinity characteristics, so that cells expressing them assort with each other, rather than with Iro-C non-expressing cells(Diez del Corral et al., 1999; Zecca and Struhl, 2002b). This specific affinity might help maintain the relatively straight and sharp notum/dorsal hinge border in the wing imaginal disc(Zecca and Struhl, 2002b). Similarly to the AP and DV boundaries, the notum/dorsal hinge boundary appears to be a source of positional information, but the molecules involved have not been identified (Diez del Corral et al.,1999).

Some progress has been made in understanding how the border of the Iro-C domain that defines the notum/dorsal hinge boundary is established. It requires the participation of two of the signaling systems that subdivide the early disc in the proximodistal axis (reviewed by Klein, 2001). Thus, on the one hand, signaling by the tyrosine kinase EGF receptor turns on the genes of the Iro-C (Wang et al., 2000; Zecca and Struhl, 2002a; Zecca and Struhl, 2002b). On the other hand, signaling mediated by the BMP2/4 homolog Dpp, which during the early/mid second larval instar is active only in the more distal territories of the disc, represses there the Iro-C and sets the distal border of the Iro-C domain (Cavodeassi et al.,2002). However, at later stages, Dpp signaling occurs in the notum territory and the border of Iro-C expression becomes refractory to it. This suggested that additional agents might control the notal-hinge boundary(Cavodeassi et al., 2002).

The msh (muscle-segment homeobox) gene (Dr –FlyBase) is a member of the conserved Msx family. It encodes a homeodomain transcription factor with an Engrailed-type repressor motif(D'Alessio and Frasch, 1996). In the embryonic mesoderm, msh specifies subsets of cardiac and muscle precursors and participates in cross-repressive interactions with other genes (Jagla et al., 2002). msh is also a neural-identity gene that is expressed in the dorsal-most region of the embryonic neuroectoderm(D'Alessio and Frasch, 1996; Isshiki et al., 1997). In the wing imaginal disc, msh imparts a dorsal identity to the dorsal bristles of the wing margin and to wing veins(Milán et al.,2001).

Here, we report that msh also participates in establishing/maintaining the notum/dorsal hinge boundary. From the second instar stage onwards, msh and Iro-C are expressed in adjacent domains of the imaginal wing disc, that of msh being distal to that of Iro-C. This situation essentially persists in the third instar, where msh is strongly expressed in the presumptive hinge region and Iro-C is expressed in the adjacent territory, the lateral notum. Loss- and gain-of-function analyses indicate that msh represses Iro-C in most of the presumptive dorsal hinge, and Iro-C prevents high level expression of msh in the notum,while it allows low expression of msh in this territory. mshis also necessary for proper development of the hinge and for the patterning of the notum. Moreover, the confrontation of cells expressing msh and Iro-C at the notum/hinge boundary appears to favor the correct growth of the notum and hinge territories.

Drosophila stocks used were: msh^Δ68 and UAS-msh(Isshiki et al., 1997), and Df(3)iro^DFM3 and UAS-ara(Gómez-Skarmeta et al.,1996). UAS-mshi was generated according to Nagel et al.(Nagel et al., 2002). The oligonucleotides used to amplify an 880 bp fragment of the first exon of the msh gene were: AAGTCGACGGATCCCAAGCGTGTGACGAACGAGCGC and AAGGTACCTCTAGAGCTGGGCGTGGAACTCGTGGAGGC. Underlined sequences correspond to restriction sites for SalI, BamHI, KpnI and XbaI that helped in the procedure. Other alleles and transgenes are described in FlyBase(http://flybase.org).

Mitotic recombination clones homozygous for the alleles msh^Δ68, Df(3L)iro^DFM3 and Chip^e55 were induced by the FLP-FRT technique (Xu and Rubin,1993) by incubating larvae for 1 hour at 37°C. Larvae were prepared as follows: flies FRT82B msh^Δ68 arm-lacZ/TM6B were crossed with either f ^36a hs-FLP; FRT82B P(f ⁺)M(3)^w124/TM6B for clones to be examined in adults, or hs-FLP; FRT82 ubi-GFP M(3)^w124/TM6B for clones to be analyzed in imaginal discs; flies hs-FLP; mwh Df(3L)iro^DFM3FRT80B/TM6B were crossed with hs-FLP; ubi-GFP FRT80B; and flies FRT42B Chip^e55/Cyo were crossed with hs-FLP; FRT42B ubi-GFP/Cyo flies.

Overexpression clones were obtained by crossing either a UAS-araor UAS-msh line with y w hs-FLP¹²²; act-FRT y+ FRT Gal4 UAS-GFP/SM5 Tb (Ito et al.,1997) flies. Clones were induced by incubation of larvae at 37°C for 8 minutes (UAS-msh) or 15 minutes (UAS-ara). Df(3L)iro^DFM3 clones in individuals overexpressing UAS-mshi were obtained by crossing flies hs-FLP; UAS-mshi/Cyo;Df(3L)iro^DFM3 FRT80B/TM6B with flies ap-GAL4; ubi-GFP FRT80B/TM6B. Clones were induced (37°C, 1 hour) at 60±12 hours AEL (after egg laying) and the flies were raised at 29°C. Cuticles were prepared by boiling in KOH and mounted in ethanol/lactic acid (5:6).

Imaginal discs were fixed and stained as in Xu and Rubin(Xu and Rubin, 1993). Antibodies were: rat anti-Ara, which reacts with Ara and Caup proteins(Diez del Corral et al.,1999), rabbit anti-Msh(McDonald et al., 1998)(provided by C. Doe), mouse anti-Wg, rabbit anti-Tsh, mouse anti-Nub (provided by M. S. Cohen and D.S.H.B.), rat anti-Zfh2(Whitworth and Russell, 2003)and rabbit anti-Sc (Skeath and Carroll,1991).

In the third instar wing disc, msh mRNA accumulates in the dorsal compartment of the wing blade region(Milán et al., 2001),and maximally in the presumptive dorsal hinge, a territory adjacent to the presumptive notum (D'Alessio and Frasch,1996). We have compared the distribution of Msh with that of the Iro-C proteins Ara and Caup, which are co-expressed in the notum region of the disc (Gómez-Skarmeta et al.,1996). In late second/early third instar wing discs, the region of Msh accumulation abuts proximally to the Ara/Caup domain and a relatively sharp border separates their respective domains(Fig. 1A). During the third instar, Msh continues to accumulate to high levels in this region and, to low levels, in the dorsal compartment of the wing pouch and in the posterior notum territory, where it overlaps with the expression of Ara/Caup(Fig. 1B,C; for nomenclature of regions of the disc, see Fig. 1C). The high accumulation of Msh does not extend into the wing pouch, as it stops adjacent to the innermost of the two circles of Wg expression (del Álamo Rodríguez et al., 2002). Similar to the second stage disc,the high accumulations of Msh in the hinge and of Ara/Caup in the notum are contiguous, but do not overlap (Fig. 1B,C). This was verified with a z-axis optical section(Fig. 1C, right). Contiguous but non overlapping expressions also occurred in the ventral hinge/mesopleura region of the disc (Fig. 1C,left).

The expression of msh in the dorsal hinge and in the posterior notum (Fig. 1C) suggested that msh may function in the development of these territories. Null conditions for msh are embryonic lethal(Isshiki et al., 1997). Hence,we generated mitotic recombination clones homozygous for the null msh^Δ68 allele and examined their phenotype in adults. The Minute technique was used to obtain clones that comprised relatively large territories of the fly(Morata and Ripoll, 1975). Clone induction at 0-24 hours after egg-lying (AEL) was generally lethal, but at 24-48 hours AEL it was compatible with viability. In the wing disc derivatives, clone associated-defects were seen at the wing hinge, the posterior scutum and the scutellum (Fig. 2K). At the hinge, the most frequently observed anomalies were malformations ranging from small defects such as an outheld wing (not shown),to the partial loss of proximal hinge structures, or to even the complete loss of most hinge structures, namely, sclerites and the proximal dorsal radius(Fig. 2E, compare with D). In the most extreme cases, the wing was fused to the scutellum and scutum(Fig. 2B, compare with 2A) or it was displaced posteriorly (Fig. 2C). Generally, the tegula was not affected. Clone induction between 48-72 hours AEL yielded similar phenotypes, and, in addition, 19% of flies with visible clones displayed ectopic tissue carrying macro- and microchaetae, which indicated its notum identity. The ectopic notal tissue appeared dorsal to the hinge and contiguous to it(Fig. 2F,G), consistent with a transformation of hinge towards notum (see below). Taken together, these data indicated that msh is necessary for the correct formation of the wing hinge and that its absence can lead to formation of ectopic notal tissue.

At the notum, when msh^Δ68M⁺ clones were induced at 24-48 hours AEL, the most frequent anomalies were a reduction of the scutellum(Fig. 2C) and the appearance of depigmented, naked and corrugated cuticle in the lateral posterior scutum adjacent to the allula and the hinge (not shown). In addition, the msh^– clones frequently developed extra macrochaetae(Fig. 2H-J), mostly in the dorsocentral (DC) region of the notum and in the scutellum, and occasionally also induced nearby wild-type cells to differentiate as chaetae, suggesting cell non-autonomous effects (Fig. 2J). The clones also suppressed extant chaetae(Fig. 2G-I), the anterior and posterior supraalars being the most affected, and interfered with the correct formation of the scuto-scutellar suture(Fig. 2J). Interestingly, in clones that comprised the lateral anterior notum, a region where mshexpression has not been detected in the third instar disc, the anterior and posterior notopleural and the presutural macrochaetae were missing in 70, 10 and 15% of cases, respectively (20 heminota examined).

To gain insight into the function of msh in the hinge, we examined the effect of msh^Δ68M⁺ clones on the expression of genes known to be required for development of this territory. homothorax (hth), teashirt (tsh), zfh-2 and wingless(wg) are expressed at high levels in the presumptive hinge(Azpiazu and Morata, 2000; Casares and Mann, 2000; Klein and Martínez-Arias,1998; Whitworth and Russell,2003) (Fig. 1C and Fig. 3A,B,E). Their characteristic patterns of expression were not overtly modified(Fig. 3A,B,F and data not shown), which suggested that msh was dispensable for the specification of the dorsal hinge territory. Using as landmarks the highly resolved pattern of scute (sc; Fig. 3C)(Cubas et al., 1991; Skeath and Carroll, 1991), we observed a shortening of the distance between the sc proneural cluster at dorsal radius, in the hinge, and the anterior postalar cluster, in the lateral notum (Fig. 3C,D). We also observed the apparent fusion of the distal (d) and proximal (p) sc clusters of the tegula region(Fig. 3C,D). These findings suggested a decrease in the size of the intervening mutant territory, an interpretation further supported by the absence of the fold of the disc that separates the notum and hinge regions (Fig. 1C) when these regions are mutant for msh(Fig. 3A,B, blue arrowheads). By contrast, the fold that separates the hinge from the wing pouch was unaffected by the msh clones (Fig. 3A,B, yellow arrowheads). These results, together with the adult phenotype of the msh^Δ68M⁺ clones, indicate that msh, although dispensable for the specification of the dorsal hinge territory, is required for the proper growth and patterning/differentiation of this territory.

Consistently, overexpression of a UAS-msh transgene in clones did not impose the high levels of tsh expression characteristic of the third instar dorsal hinge into other territories of the disc (not shown). However, UAS-msh driven by ap-Gal4 did promote expression of zfh-2 in the dorsal compartment of the wing pouch and, to low levels,in the notum (Fig. 3G). The resulting pharate individuals showed reduced and crumpled wings, and a reduced notum; however, no hinge-like sensory organs or other structures could be discerned in the latter. Thus, although msh overexpression imposes some hinge-like properties to the prospective wing pouch and notum tissues(zfh-2 expression), it seems unable to force a complete transformation of these tissues into hinge.

The msh^Δ68 M⁺clones derepressed the ara and caup genes of the Iro-C in the hinge territory (Fig. 4A,B). This upregulation was also observed in non-Minute msh^Δ68 clones induced at either 24/48 hours AEL (Fig. 4F) or 48/72 hours AEL (Fig. 4C). In all cases, derepression was cell autonomous, and did not extend into the wing pouch [the latter defined by the expression of Nubbin(Ng et al., 1995)(Fig. 4A,B) or in a triangular area that included the prospective tegula(Fig. 4B, asterisk; Fig. 4F). Similar ara/caup derepression was obtained by overexpressing an msh-interferring RNA (UAS-mshi; ap-Gal4 driver, Fig. 4D). As Iro-C imparts notum identity (Diez del Corral et al.,1999; Wang et al.,2000; Zecca and Struhl,2002b), its ectopic expression in the hinge may account for the notum structures that develop in msh^Δ68 M⁺ clones(Fig. 2F,G). Moreover, forced expression of ara in the dorsal hinge interferes with its proper formation (R. Diez del Corral, PhD thesis, Universidad Autónoma de Madrid, 1998). Thus, the ectopic expression of ara/caup may contribute to the disappearance of the hinge structures observed in msh clones (Fig. 2E).

The capacity of msh to repress ara/caup was further demonstrated by overexpressing UAS-msh in clones in the notum. Early induction of overexpression (12-36 hours AEL) invariably repressed ara/caup (Fig. 4E). Late induction (60-84 h AEL) still did so if cells were located at the lateral notum, but in central and medial regions, many cells within clones or even entire clones failed to repress ara/caup (not shown). Expression of UAS-msh in the whole dorsal compartment of the disc (ap-Gal4driver) at 17°C reduced the size of the notum territory(Fig. 3G). At 25°C, it strongly inhibited the development of this territory(Fig. 4G) and individuals died at the pupal stages. We conclude that msh can repress ara/caup in the notum, but in the central/medial regions this capacity gradually decreases as the disc grows. Evidently, this repression requires high concentrations of Msh, such as those provided by the UAS/Gal4 system, because in the wild-type disc, Ara/Caup co-exist with low levels of endogenous Msh (Fig. 1B,C).

We next examined whether ara/caup might repress msh in the notum by using clones that lack essentially all Iro-C function(iro^DFM3 mutation). When induced early, these clones acquire a hinge fate and consequently upregulate tsh(Diez del Corral et al.,1999). Thus, as expected they expressed msh at levels similar to those of the hinge (Fig. 5A). However, iro^DFM3 clones induced late in development do not undergo fate transformations(Diez del Corral et al.,1999). Still these clones, when located within the posterior lateral notum, increased autonomously the accumulation of Msh(Fig. 5B), but were unable to upregulate tsh (not shown). Hence, Iro-C downregulates mshin this region of the disc even in cells that are not transformed into hinge cells. In the anterior lateral notum, late clones do not derepress msh (not shown), suggesting the presence of additional repressors.

The ability of Iro-C to repress msh was further demonstrated with clones overexpressing UAS-ara in the hinge. msh was autonomously repressed (Fig. 5C). This result was verified by driving UAS-ara with ptc-Gal4 or by constitutively activating the EGFR signaling pathway,which upregulates Iro-C in the hinge(Zecca and Struhl, 2002b) (not shown). Taken together, these data support a mutual repression between msh and Iro-C in the notum/hinge region of the wing disc.

As indicated above, cells within the notum that lack Iro-C function switch fate and autonomously develop as dorsal hinge(Diez del Corral et al.,1999). We now find that msh is necessary for proper development of the hinge. Thus, it seemed pertinent to examine the fate of cells that were simultaneously depleted of Iro-C and msh activities. Accordingly, we induced iro^DFM3 clones in discs expressing UAS-mshi (ap-Gal4 driver) and examined the clones in the third instar discs (adults failed to emerge). In the dorsal wing and hinge,clones appeared with normal frequency (Fig. 5D). However, in the notum territory, clones did not survive or were very small, when compared with the twin wild-type clones(Fig. 5D). Thus, in this territory, cells that are neither specified as notum nor can properly develop as hinge are not viable or are outcompeted by the wild-type cells.

It is known that the dorsal selector gene ap positively regulates msh in the dorsal wing blade territory(Milán et al., 2001). We investigated whether ap also regulates msh at the dorsal hinge and notum. Wing discs homozygous for the null ap^UGO35 allele have in general a profoundly altered morphology. Some discs, however, have recognizable territories and, in both early and late third instar, had very little expression of msh in the presumptive hinge and notum (Fig. 5E; not shown). As expected, Iro-C expression was expanded(Fig. 5E). Clones that lack the essential Ap co-factor Chip are equivalent to reducing ap function(Fernández-Fúnez et al.,1998; Morcillo et al.,1997). In these clones, msh expression was eliminated at the dorsal hinge and wing pouch, but not at the ventral hinge, where ap is not expressed (Fig. 5F). Hence, msh is under the positive control of ap in the whole ap domain.

Because in the second instar disc, Dpp signaling confines ara/caupexpression to the notal region of the disc(Cavodeassi et al., 2002), we examined whether this inhibition was mediated by msh. The expression of msh under loss- and gain-of-function conditions for Dpp argued against this possibility (not shown). Wg signaling is most important to form and pattern the hinge and the wing (Couso et al., 1993; Ng et al.,1996; Sharma and Chopra,1976). Again, loss- and gain-of-function conditions for Wg did not prevent expression of msh in the dorsal hinge (not shown), indicating that Wg does not control msh.

In Drosophila, the homeodomain gene msh is known to be involved in different processes. Thus, it participates in regional specification of muscle progenitors/founders(Nose et al., 1998); together with vnd and ind, it helps subdivide the embryonic neuroectoderm along the dorsoventral axis (reviewed in Cornell and Ohlen, 2000; Gómez-Skarmeta et al.,2003; Skeath,1999); and it confers dorsal identity to the dorsal bristles of the anterior margin of the wing(Milán et al., 2001). We report additional functions of msh, namely, the formation/maintenance of the subdivision between the territories of the wing disc that will give rise to the notum (dorsal mesothoracic trunk) and the dorsal hinge (appendix), the proper growth of the dorsal hinge, and the patterning of this region and of the notum.

In the developing wing disc, msh is expressed most strongly in the territory of the dorsal hinge, the region between the notum and the dorsal wing blade territories. Removal of msh in clones results in malformations that range from small defects, such as an outheld wing, to partial or even complete loss of most hinge structures. In the latter cases,the hinge may be posteriorly misplaced and ectopically attached to the scutellum. In addition, in a fraction of flies ectopic notum tissue appears contiguous to the extant hinge. Because at least a large part of the hinge tissue is still present, we surmise that the absence of recognizable hinge structures is due to the failure of their proper differentiation. This phenotype correlates well with that observed in third instar wing discs displaying msh^- clones. Indeed, even large clones that remove msh from most of the dorsal hinge territory allow the specification of this territory, as demonstrated by the relatively unmodified characteristic patterns of expression of genes such as wg, zfh-2, hthand tsh, and the presence of recognizable proneural clusters of sc expression. Moreover, the presence in mutant hinges of relatively well resolved clusters of sc expression(Fig. 3D) indicate that the prepatterning of the hinge can proceed to a large extent in the absence of msh. We conclude that msh is largely dispensable for specification of the dorsal hinge territory, but it is required for the final stages of its patterning and differentiation.

Mosaic analyses aimed at studying the patterns of cell proliferation in the wing disc disclosed the presence of the anterior, posterior, dorsal and ventral compartments of the wing with borders that imposed absolute restrictions to cell proliferation(García-Bellido et al.,1976). A border of this type was suggested to exist between the notum and dorsal hinge, as well as between the pleura and the ventral hinge(García-Bellido et al.,1976), but the complex morphology of these regions and the unavailability of appropriate cuticular markers made the proposal uncertain. In fact, analyses performed later in the wing disc, showed that clones could straddle the notum/dorsal hinge boundary, this being defined by the distal border of the Iro-C domain (Diez del Corral et al., 1999). Hence, at this boundary, the descendants of a cell would adopt their developmental fate not according to lineage, but depending on the side of the boundary they were located. The issue thus arose of how the boundary between the notum and the dorsal hinge territories would be established and maintained. Considering that the extent of the notum territory is defined by the expression of the Iro-C(Diez del Corral et al., 1999; Wang et al., 2000), this issue can be largely resolved by explaining how the distal border of the Iro-C domain of expression is defined.

So far, several genetic interactions have been identified that together permit to suggest a mechanism that partially answers this question(Fig. 6). In the second instar disc, the EGFR pathway activates ap and Iro-C(Wang et al., 2000; Zecca and Struhl, 2002a; Zecca and Struhl, 2002b). The distinct but overlapping domains of expression of these genes, the dorsal compartment (ap) and the notum territory (Iro-C), may be defined by differential sensitivity to EGFR signaling(Zecca and Struhl, 2002a) or,alternatively, in the case of Iro-C, by Dpp signaling(Cavodeassi et al., 2002). In these early stages, Dpp signaling is active only in the distal part of the disc, where it represses the Iro-C and sets its distal limit of expression. Hence the antagonistic actions of the EGFR and the Dpp pathways would define the position of the distal limit of the Iro-C domain, and therefore the position of the notum/hinge subdivision.

We now find that at approximately the time Iro-C starts to be expressed in the more proximal part of the disc, i.e. that which will become the notum,expression of msh, by means of ap, is turned on in the adjacent dorsal hinge territory (see also Milán et al., 2001). These essentially complementary patterns of expression are maintained, with some qualifications, in the third instar disc. Loss- and gain-of-function experiments show that msh prevents ara/caup from being expressed in the hinge; and ara/caup restrain msh from being expressed in the notum at the high levels typical of the hinge (although it is expressed at a low level in part of the notum). This mutual repression also occurs late during development.

How relevant is this mutual repression for the establishment of the notum/dorsal hinge territorial subdivision? As indicated above, in ∼19% of flies with msh clones, the removal of Msh from the hinge induces extra notum tissue. In the remaining cases, this removal does not substantially affect the identity of the hinge territory. Thus, the mutual repression between msh and Iro-C is crucial for the notum/hinge territorial subdivision in only a small but substantial fraction of the discs. This indicates that additional agents, probably expressed in the hinge,participate in effecting the subdivision. By contrast, notum cells that lose Iro-C always change their fate to hinge cells and, consequently, depending on position, they modify the notum/hinge subdivision or create an ectopic notum/hinge boundary (Diez del Corral et al., 1999). Hence, the relevance of the msh/Iro-C mutual repression to define/maintain that subdivision relies mainly on its defining/maintaining the border of the Iro-C domain, and thereby preventing the expression of hinge genes within the notum territory. Thus, a `pronotum'gene (Iro-C) and a `hinge differentiation' gene (msh), despite their different positions within the genetic hierarchies that govern the development of their respective domains, cross-regulate each other and participate in the early definition of their respective territories. Our current data do not permit us to distinguish between the possibilities that the mutual repression between msh and Iro-C is instrumental in establishing this territorial border, or, alternatively, that it stabilizes a previous border defined by the antagonistic actions of EGFR and Dpp on the Iro-C.

The relevance of the mutual repression between Iro-C and msh is also manifested by their respective overexpression. Ectopic Iro-C products in the hinge impair the proper differentiation of hinge structures (R. Diez del Corral, PhD thesis, Universidad Autónoma de Madrid, 1998). High levels of Msh in the notum turn on a hinge-specific marker like zfh-2(Fig. 3G) and are detrimental for notum development (Fig. 4G).

In the third instar disc, the distal border of the Iro-C domain is no longer straight and displays a pronounced `bay' where ara/caup are downregulated (Fig. 1C, blue arrow, red channel). This roughly coincides with the area of highest expression of msh in the lateral notum. msh is probably responsible for this downregulation of ara/caup, as the `bay'disappears in msh clones (Fig. 4F). Moreover, the abutting domains of msh and Iro-C in the ventral hinge and pleura, respectively(Fig. 1C), suggest that a similar mutual repression may occur there to establish the subdivision between these neighboring regions. Finally, the removal of msh does not activate Iro-C in the anterior part of the hinge territory(Fig. 4B), suggesting again that agents other than msh and Dpp(Cavodeassi et al., 2002) help maintain Iro-C expression confined to the notum territory.

Iro-C^– clones located within the medial notum not only undergo an autonomous transformation to dorsal hinge. They also become surrounded by a fold similar to that which separates the notum and hinge territories, and they modify the expression of several markers in the surrounding wild-type tissue in a way consistent with a transformation of this tissue towards lateral notum (Diez del Corral et al., 1999). These nonautonomous effects suggest that signals emerge from the Iro-C^– clones, and that these signals alter the fate of the aposed notum tissue. Hence, it was inferred that, in the wild-type disc, signaling would take place across the hinge/notum boundary and this would help pattern at least the lateral notum(Diez del Corral et al.,1999). This is reminiscent of the DV and AP compartment boundaries, where signaling mediated by the diffusible molecules Wg, and Hh and Dpp, respectively, are key to stimulating the growth and pattern of the wing disc (for reviews, see Brook et al.,1996; Teleman et al.,2001; Vincent and Briscoe,2001). However, in the hinge/notum boundary, the signaling agents have not been identified. They could be either diffusible molecules or cell-bound molecules that mediate this cell to cell communication.

Now, we find that the imaginal disc territories flanking the notum/hinge border are reduced in size when they are mutant for msh(Fig. 3). We do not know whether this effect is due to decreased cell proliferation, increased cell death or both, and whether it mostly affects the hinge or the lateral notum. However, it is clear that by removing msh and allowing Iro-C to be expressed in the hinge, the msh clones suppress the confrontation of proper hinge cells with notum cells. It is tempting to speculate that this could affect the net growth of the territory by removing positional values(García-Bellido et al.,1994) and/or by suppressing or making ineffective the postulated signaling associated with the hinge/notum border. Consistently, a reduced size of the notum plus hinge region (and a simplification of the patterning) is also observed in discs overexpressing UAS-ara in the dorsal compartment (Diez del Corral et al.,1999) (E.V.-C. and J.M., unpublished), a condition that removes most msh expression from the hinge. The failure of Iro-C^– clones within the notum territory to grow and survive when they are also depleted of Msh (Fig. 5D) might result from the absence of proper signaling across a boundary where wild-type notum cells confront Iro-C^–msh^– cells. Considering that the activity of the EGFR signaling pathway is necessary for notum cell proliferation(Díaz-Benjumea and García-Bellido, 1990; Simcox et al., 1996; Wang et al., 2000), it would be of interest to examine whether this pathway is involved in, or is modulated by, the presence of the notum/hinge boundary.

In Drosophila, the Iro-C genes and msh respectively participate in the DV subdivision of the eye(Cavodeassi et al., 1999; McNeill et al., 1997) and of the neuroectoderm (reviewed by Cornell and Ohlen, 2000; Gómez-Skarmeta et al.,2003; Skeath,1999). In vertebrates, although to our knowledge no instance of mutual repression between homologs of Iro-C and msh has been described, members of each family participate in establishing borders by repression with other genes in the spinal cord, the brain (reviewed by Gómez-Skarmeta et al.,2003) and between rhombomeres(Lecaudey et al., 2004). Clearly, both genes are used frequently to subdivide territories and establish alternative differentiation pathways at each side of the border that separates them.

Throughout the third instar, msh is expressed at relatively low levels in the posterior notum territory. Here, removal of msh most often results in impaired growth of the scutellum, absence of the scutellum/scutum suture and alterations of the bristle pattern. Interestingly,the lateral/anterior notum macrochaetae are often missing, even though they arise in a region apparently devoid of msh expression. This suggests that either msh is expressed there at very low but functional levels,or that the suppression of macrochaetae results from non-autonomous effects of the absence of Msh from neighboring territories. It should be noted that non-autonomous macrochaetae suppression is also associated with Iro-C^– clones that cause notum to hinge transformations(Diez del Corral et al.,1999). This has suggested that modification of the putative signaling across the notum/hinge boundary interferes with macrochaetae patterning at the notum. It is possible that the msh clones might also interfere, as indicated above, with signaling from this border. If so,the presence of clusters of sc expression at the anterior lateral notum within large msh clones(Fig. 3D) suggest that this interference might occur at a stage later than the emergence of the proneural clusters.

The absence of msh function does not modify the expression of Iro-C in the lateral notum or the characteristic patterns of expression of eyg and hth (E.V.-C., unpublished), genes that are high in the hierarchy that control notum development(Aldaz et al., 2003; Aldaz et al., 2005). But it removes the scutum/scutellar suture and promote development of extra bristles in the dorsocentral and scutellar regions. Again, these are phenotypes suggestive of an interference with the late patterning and differentiation of these structures.

We are grateful to S. Campuzano, J. Culí, J. L. Gómez-Skarmeta, M. Ruiz-Gómez and colleagues of the J.M. laboratory for advice on the work and constructive criticism of the manuscript; to P. Martín for invaluable help in constructing the mshi stocks; and to M. Milán, A. Nose and C. Doe for providing reagents and stocks. Predoctoral fellowship from Ministerio de Educación y Ciencia to E.V.-C. is acknowledged. Work was supported by a grant from Dirección General de Investigación Científica y Técnica (BMC2002-411) and an institutional grant from Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa.