Notch signalling coordinates tissue growth and wing fate specification in Drosophila

Growth and fate specification of a given organ are two regulated processes that have to be tightly coupled to generate a correctly shaped and sized structure (Lecuit and Le Goff,2007). Uncoupling these two processes has disastrous consequences in development and disease. In the past few years, much has been learnt about the regulation of growth and fate specification. However, very little is known about how these processes are coupled. The Drosophila wing imaginal disc provides a well-studied model system for analyzing at a genetic, cellular and molecular level these processes during development(Cohen, 1993).

The wing primordium contains the progenitors of both the adult body wall and the wing (Bryant, 1975). The developmental decision between wing and body wall is made early in development and is defined by the opposing activities of two secreted signalling molecules, Wingless (Wg) and the EGFR ligand Vein (Vn), in the most ventral and dorsal sides of the wing primordium, respectively(Ng et al., 1996; Wang et al., 2000; Zecca and Struhl, 2002). Genetic manipulations that increase Notch activity or activate Notch at ectopic sites result in increased growth, and also in cancer(Ferres-Marco et al., 2006; Moberg et al., 2005; Radtke and Raj, 2003; Thompson et al., 2005; Vaccari and Bilder, 2005). Here, we present evidence that growth of the early wing primordium mediated by the activity of Notch modulates the cellular response to Wingless and facilitates wing fate specification.

UAS-N^dsRNA(Presente et al., 2002); UAS-notum (Giraldez et al.,2002); UAS-Ser™ and UAS-Dl™(Herranz et al., 2006); UAS-cycD UAS-cdk4 (Datar et al.,2000; Meyer et al.,2000); UAS-dmyc(Johnston et al., 1999); UAS-hippo (Harvey et al.,2003; Wu et al.,2003); UAS-PTEN(Goberdhan et al., 1999). Other stocks are described in FlyBase(http://flybase.org/).

Mouse anti-Wg (4D4, Developmental Hybridoma Bank); mouse anti-Nubbin(Ng et al., 1995); rat anti-Hth (Wu and Cohen, 1999). Other antibodies used are commercially available.

Flies were allowed to lay eggs for 12 hours. Larval size was used to select the same developmental stage in all genotypes analyzed. The number of BrdU-labeled cells was counted in wild-type, sd-gal4; UAS-Ser™and sd-gal4; UAS-Ser™; UAS-CycE early second instar wing discs(48 hours after egg laying) that had been raised in the same conditions and dissected simultaneously. BrdU staining was performed as described by Milan et al. (Milan et al., 1996). Using Image J software, the area of these discs was also measured. The area of the sd-gal4; UAS-Ser™ and sd-gal4; UAS-Ser™;UAS-CycE discs were always compared with wild-type discs raised in the same conditions. Final area measurements were normalized to the wild-type ones and are presented in arbitrary units. Using Microsoft-Excel Software, the average size and standard deviation of wing discs were calculated, and t-test analysis was carried out.

We used the Gal4/UAS system (Brand and Perrimon, 1993) combined with the thermo-sensitive version of Gal80 [Gal80^ts (McGuire et al.,2004)], a repressor of Gal4 protein activity, to precisely control, in time and space, gene expression. Adult flies carrying a Gal4 driver, the Gal80^ts construct and an UAS-transgene were allowed to lay eggs over a period of 24 hours at 18°C. The progeny were then raised at 18°C to maintain the Gal4/UAS system in a switched-off state but transferred to 29°C for different periods of time during larval development to induce Gal4/UAS-dependent gene expression.

During the second larval stage, the antagonistic activities of Wg and Vn specify wing versus body wall fate. Notch activity has been proposed to have a role in this process because the loss of Notch during this developmental stage leads to a failure in the induction of wing fate with a concomitant duplication of body wall structures (Couso and Martinez Arias, 1994). We decided to further analyze the role of Notch in this process, not only in the adult fly, but also in the developing wing primordium, by using the corresponding wing and body wall molecular markers.

We first blocked the activity of Notch and analyzed the resulting adult flies (Fig. 1). To block Notch,we expressed either dominant-negative forms of the Notch ligands Delta and Serrate [Dl™ and Ser™, intracellular deletions of Delta and Serrate well known to repress Notch activity(Sun and Artavanis-Tsakonas,1996)], or a Notch RNA interference construct[N^dsRNA, known to reduce Notch protein levels and to lead to strong Notch loss-of-function phenotypes in Drosophila tissues(Presente et al., 2002)] (see Fig. S1 in the supplementary material). We used the scalloped-Gal4(sd-Gal4) driver because it is expressed in the wing disc from early larval stages (Fig. 1C). The expression of sd-Gal4 in early wing discs was not affected in conditions of reduced Notch activity (see Fig. S2 in the supplementary material). In adult flies, wings were either vestigial or absent (data not shown), and body wall structures were often duplicated(Fig. 1E,G,J). In the developing wing imaginal disc, expression of the homeodomain protein Homothorax (Hth) and the zinc-finger transcription factor Teashirt (Tsh) was restricted to the presumptive body wall, while the homeodomain protein Nubbin(Nub) was expressed in the presumptive wing territory(Ng et al., 1996; Wu and Cohen, 2002)(Fig. 1A). Wg was expressed in the body wall and wing territories of late third instar discs in a characteristic pattern (Fig. 1A). We then analyzed and compared the expression of these molecular markers in mature wing discs in which Notch activity had been compromised. Nub was absent, and the characteristic expression of Hth, Tsh and Wg in the notum showed a mirror-image duplication(Fig. 1D,F). The characteristic expression pattern of sd-Gal4 in the resulting late third instar wing discs also showed mirror-image duplication (see Fig. S2 in the supplementary material). These results confirm the requirement for Notch in wing fate specification.

We next decided to temporally control the reduction in Notch activity by using the Gal4/Gal80^ts system (see Materials and methods for details). Reducing Notch activity for a period of 24 hours during the second instar stage (in sd-gal4; UAS-Ser™ Gal80^ts larvae)gave rise to wing imaginal discs and adult flies in which wings were absent and body wall structures were duplicated(Fig. 1K-N). These effects were not observed when Notch activity was reduced during the first or third instar stages (Fig. 1L), thereby confirming the requirement of Notch in wing fate specification during the second larval stage, as proposed by Couso and Martinez Arias(Couso and Martinez Arias,1994).

Similar defects in wing fate specification were obtained when the activity of Wg protein (by overexpression of the Wg antagonist Notum) or its signalling pathway (by overexpression of the antagonist of the Wg pathway, the kinase Shaggy/Gsk3) were compromised in the sd domain(Fig. 1H-J; data not shown)(Giraldez et al., 2002; Morata and Lawrence, 1977; Sharma and Chopra, 1976), or when a temperature-sensitive mutant allele of wg was used to block its function for a period of 24 hours during the second instar(Couso and Martinez Arias,1994). Thus, Notch might either control the expression or activity of Wg, or collaborate with the Wg pathway during wing fate specification. Consistent with this, reducing the amount of Wg protein, in a wgheterozygous background, increased the frequency of duplicated nota(Fig. 2L). However, only in the case of collaboration between Notch and Wg pathways might Notch be required in a cell-autonomous manner. Therefore, to choose between these two alternatives,we decided to block the Notch or Wg signalling pathways in a subset of cells within the presumptive wing primodium and examine the expression of Nub. Unfortunately, classical clonal analysis cannot be used to address this issue,as clones of cells lacking Notch or Wg activity cannot be recovered in the wing primordium, probably because of impaired cell proliferation or viability(de Celis and Garcia Bellido,1994; Giraldez and Cohen,2003; Johnston and Sanders,2003). For this reason, we decided to use the Gal4/UAS system to compromise Notch activity in discrete territories within the wing primordium. Blocking Notch activity in a stripe along the anteroposterior compartment boundary (in patched-Gal4; UAS-N^dsRNA larvae) or throughout the dorsal compartment (in apterous-Gal4;UAS-N^dsRNA larvae) did not result in the loss of Nub expression (Fig. 2A,C). By contrast, blocking the response to Wg by the overexpression of Axin or Shaggy/Gsk3 [two antagonists of the Wg pathway(Logan and Nusse, 2004)] in the same domains induced the loss of Nub expression in the anterior(Fig. 2B; data not shown) or dorsal (Fig. 2D) compartments. These results indicate that Wg signalling is required in a cell-autonomous manner to induce wing fate specification, as previously shown(Ng et al., 1996). By contrast, the requirement of Notch signalling in this process is not cell-autonomous and it might be mediated by the activity of Wg. Indeed,epistatic analysis confirmed this hypothesis. The expression of Wg ligand or the activation of the Wg pathway in wing discs in which Notch activity had been compromised rescued Nub expression(Fig. 2E,F) and adult wing specification (Fig. 2G). By contrast, activation of the Notch pathway in wing discs in which Wg activity had been compromised did not rescue the expression of this protein(Fig. 2H,I), nor adult wing specification (data not shown).

The most plausible explanation for this epistatic relationship is that Notch activity positively regulates Wg expression. However, the following results indicate that this might not be the case. Early in wing development,Wg is expressed in a ventral wedge of cells located in the anterior compartment (Fig. 2J)(Couso et al., 1993). Blocking Notch activity in these cells throughout development (in patched-Gal4;UAS-N^dsRNA larvae) did not result in the loss of Nub expression (Fig. 2A), and Wg expression was not affected in early wing discs in which Notch activity had been compromised (in sd>Ser™ larvae, Fig. 2K). Wg is required to repress Vn expression in the most dorsal region of the early disc(Wang et al., 2000). Vn antagonizes wing development and specifies body wall fate. An alternative mechanism by which Notch might affect Wg activity would be by interfering with the ability of Wg to repress Vn expression. Indeed, we noted that reducing the amount of Vn protein, in a vn heterozygous background, decreased the frequency of duplicated nota caused by compromised Notch activity(Fig. 2L). However, Vn expression was not affected by blocking Notch activity (compare Fig. 2J and 2K). Taken together, these results indicate that Notch activity is required upstream of Wg in the process of wing fate specification, but that Notch does not regulate the relative expression patterns of Wg and Vn.

Notch is thought to promote growth in the early wing imaginal disc(de Celis and Garcia Bellido,1994), and growth induced by Notch is required for specification of the eye within the Drosophila eye-antenna primordium(Kenyon et al., 2003). We therefore hypothesized that the requirement of Notch in wing fate specification is because of its control of tissue growth.

We first measured the size and analyzed the proliferation dynamics of early second instar wing discs after blocking Notch activity. Early second instar wing discs expressing the dominant-negative form of Serrate (Ser™) in the sd-Gal4 domain were on average 34% smaller than were wild-type primordia raised in the same conditions(Fig. 3I). The average wing disc sizes, in arbitrary units, were 1±0.3 (wild type) and 0.67±0.14 (sd>Ser™; number of scored discs: wild type, n=22; sd>Ser™ n=40; P<10^-6). The number of cycling cells, monitored by BrdU incorporation, was also reduced. The number of BrdU-positive cells in wild type and sd>Ser™ wing discs was 14±4 and 6±1,respectively (wild type, n=7; sd>Ser™, n=11).

We next tested whether overexpressed cell cycle regulators or growth promoters were able to rescue tissue growth and wing fate specification in conditions of reduced Notch activity. Overexpression of cell cycle regulator Cyclin E [which drives G1-S transition(Neufeld et al., 1998)] in wing discs in which Notch activity had been compromised was able to restore the size of the wing primordia (Fig. 3I). The average size of sd>Ser™,CycE discs(1.13±0.36, n=21) was significantly bigger than that of sd>Ser™ discs (0.67±0.14, n=40, P<10^-8) grown in the same conditions. Similarly, the number of proliferating cells was also restored (an average of 13±3 BrdU-positive cells, n=7 discs). Interestingly, Nub expression(Fig. 3A,C) and adult wing specification (Fig. 3B,D,J)were restored in these conditions (in sd>Ser™,CycE larvae and flies). Cyclin E did not show this capacity in the absence of Wg activity(Fig. 3E,F). Consistent with this, the size of the wing discs was not reduced after blocking Wg activity(the average wing disc sizes were 1±0.3 for wild type and 1.04±0.2 for sd>notum; P=0.77; n=8 and 5 discs, respectively). Adult fate specification was also rescued when the cell cycle regulator String [previously known as Dcdc25, which drives G2-M transition (Neufeld et al.,1998)] was expressed in conditions of reduced Notch activity(Fig. 3J). It is interesting to note that, in late third instar wing discs, overexpression of CycE and String has been reported to drive G1-S and G2-M transitions without causing any increase in tissue size (Neufeld et al.,1998). We therefore wondered whether the overexpression of these cell cycle regulators was able to induce tissue growth in second instar discs. Interestingly, the average size of sd>CycE (1.7±0.4, n=39) and sd>Stg (1.13±0.22, n=33) early second instar wing discs was significantly bigger than that of wild-type(1±0.22, n=47) discs grown in the same conditions(P<10^-13 and P=0.01, respectively; see Fig. S3 in the supplementary material). These results suggest that the ability of CycE and String to rescue wing fate specification is a consequence of increased tissue growth. Alternatively, the increased cell cycling caused by these cell cycle regulators might interfere with the ability of the cells to transduce Notch signalling. We therefore analyzed the ability of Ser™ to block Notch signalling in the presence of high levels of CycE or String. Similarly, we analyzed the ability of a dominantly active form of the Notch receptor [N^intra (Struhl and Adachi, 1998)] to activate the expression of Notch target genes in the presence of high levels of CycE or String. As shown in Fig. S4 (see supplementary material), the activity of the Notch pathway is not affected by the overexpression of these cell cycle regulators.

The growth promoters Cyclin D and Cdk4(Datar et al., 2000), or dMyc(Dm - FlyBase) (Johnston et al.,1999), were also able to rescue wing fate specification under conditions of reduced Notch activity (Fig. 3J). Conversely, expression of the growth repressor PTEN(Goberdhan et al., 1999)increased the frequency of duplicated nota under conditions of reduced Notch activity (Fig. 3J), and blocking growth by means of overexpression of the tumor suppressor gene hippo (Harvey et al.,2003; Wu et al.,2003) in the early wing primordium induced failure in wing fate specification with a concomitant duplication of body wall structures(Fig. 3G,H). Taken together,these results indicate that growth induced by Notch activity is essential for the Wg-dependent induction of wing fate.

The expression of Wg in the most ventral part of the wing disc specifies the wing field at the same time as restricting Vn expression to the most dorsal part (Fig. 4A). Vn is required to block the responsiveness of body wall cells to Wg. Thus, the relative concentration of the diffusible proteins Wg and Vn(Neumann and Cohen, 1997; Schnepp et al., 1996; Zecca et al., 1996)experienced by disc cells directs their wing versus body wall fate. It is interesting to note that the expression of these two molecules is established long before the wing field is induced in the presumptive wing primordium(Wu and Cohen, 2002). Wg expression starts long before wing field specification takes place, as revealed by the later induction of Nub expression and the reduction in the expression of the body wall cell marker Tsh(Fig. 4D-G). We therefore propose that tissue growth modulates the cellular response to these signalling molecules and controls, in time, wing fate specification. In the early wing primordium, Vn might reach every wing cell, thereby blocking responsiveness to Wg and repressing wing fate specification. Growth induced by Notch activity might pull the sources of Wg and Vn apart and, thus, most ventral cells might not sense sufficient Vn levels, so Wg would be able to induce wing fate. Interestingly, the overexpression of Wg or overactivation of its signalling pathway is able to bypass the requirement of growth in this process(Fig. 2E,F), which indicates that the cells sense the relative levels of Wg and Vn. Once the wing field has been specified, Wg starts to be expressed along the presumptive wing margin,where it exerts a fundamental function in the maintenance of the Notch-dependent organizing center along the DV boundary(Buceta et al., 2007; Couso et al., 1994; Rulifson and Blair, 1995). Note that the organizing activity of Notch at the DV boundary takes place long after the early function of Notch revealed in this work, which is involved in promoting growth and facilitating wing fate specification. As revealed by the expression of the Notch target E(spl)m-β, it is not until late in the second instar that the expression of Notch is restricted to the DV boundary (see Fig. S1 in the supplementary material). During the process of wing fate specification that takes place during second instar, it is uniformly expressed in the whole wing disc (Fig. S1 in the supplementary material). These results imply that growth also facilitates the reiterative use of signalling molecules, such as Wg and Notch, to exert different functions during the development of a multicellular organ like the wing primordium.

At the same time that wing and body wall fate specification takes place in the wing primordium, Vn is involved in the induction of apterousexpression in the dorsal region (Wang et al., 2000; Zecca and Struhl,2002) (see also Fig. S6 in the supplementary material). Consistent with the model proposed above, the activity of Vn, as monitored by the expression of apterous, was modulated by tissue growth (see Figs S5,S6 in the supplementary material). In the absence of Notch activity, even though Vn expression is not affected (Fig. 2K), Vn appears to reach every wing cell, as apterousexpression was expanded ventrally (see Fig. S5 in the supplementary material). Increased levels of Wg expression or growth promoted by CycE appear to re-establish the dorsally restricted range of activity of Vn, as apterous expansion was blocked under these circumstances (see Fig. S5 in the supplementary material).

Growth promoted by Notch has also been shown to be directly involved in the specification of the eye within the Drosophila eye-antenna primordium(Kenyon et al., 2003), a process that also depends upon the opposing activities of two secreted signalling molecules, in this case Dpp and Wg. Thus, Notch coordinates in a very elegant manner both eye and wing primordia tissue growth and eye/wing specification, by modulating the response of the cells to the activities of signalling molecules. These results indicate that the same mechanism might be commonly used in animal development to coordinate tissue growth and fate specification.

The evolution of wings was crucial in the process of adaptation, allowing insects to escape predators or colonize new niches. It has recently been shown that the loss and recovery of wings has occurred during the course of evolution (Whiting et al.,2003). This finding would suggest that wing developmental pathways are conserved in wingless insects and are being re-used. According to our results, we speculate that adaptative changes in animal size could modulate the cellular response to signalling molecules such as Wg, thereby helping to drive some of these extraordinary reversible transitions.

We thank N. Azpiazu, S. Cohen, B. Edgar, D. Hipfner, L. Johnston, M. Furriols, J. P. Vincent, the Bloomington Stock Center and the Developmental Studies Hybridoma Bank for flies and reagents, and I. Becam, S. Cohen, H. Herranz and three anonymous reviewers for comments on the manuscript. N.R. was funded by a predoctoral fellowship from the Ministerio de Educación y Ciencia, Spain, and M.M.'s laboratory was funded by Grants from Dirección General de Investigación Científica y Técnica (BFU2004-00167/BMC and BFU2007-64127/BMC) and Generalitat de Catalunya (2005 SGR 00118), intramural funds and the EMBO Young Investigator Programme.