Boundary Formation in Drosophila Wing: Notch Activity Attenuated by the POU Protein Nubbin

See allHide authors and affiliations

Science  17 Jul 1998:
Vol. 281, Issue 5375, pp. 409-413
DOI: 10.1126/science.281.5375.409


Cell interactions mediated by Notch-family receptors have been implicated in the specification of tissue boundaries in vertebrate and insect development. Although Notch ligands are often widely expressed, tightly localized activation of Notch is critical for the formation of sharp boundaries. Evidence is presented here that the POU domain protein Nubbin contributes to the formation of a sharp dorsoventral boundary in the Drosophila wing. Nubbin represses Notch-dependent target genes and sets a threshold for Notch activity that defines the spatial domain of boundary-specific gene expression.

Spatially localized activation of Notch is required for specification of the dorsoventral (DV) boundary of the Drosophila wing (1–5). Notch signaling has also been implicated in establishing tissue boundaries in somite formation, in neurogenesis, and at the DV boundary of the vertebrate limb (6–8). The tight localization of Notch activity in these systems contrasts with the broad distribution of Notch ligands. The problem of spatially limiting Notch activation is partially solved through activity of fringe genes (9), which modulate the sensitivity of Notch for its ligands and contribute to spatially limiting Notch activity (8, 10). Certain features of the abnormal wings in flies mutant for the nubbin gene suggested a possible role for Nubbin protein in spatially limiting Notch activity at the DV boundary of the wing (11, 12). Thenubbin gene encodes a POU domain protein that is expressed in the developing wing primordium (11) (Fig. 1A).

Figure 1

Ectopic activation of Notch target genes in nubbin mutant wings. (A) Nubbin protein expression (brown) in a mature third-instar wing imaginal disc visualized by histochemical staining with mouse anti-Nubbin (27). (B) Cuticle preparation of a wild-type wing margin (detail of a region in the anterior compartment). The dorsal and ventral surfaces of the wing were peeled apart and flattened so the margin is viewed from the edge. The dorsal surface to the top. Orderly rows of wing-margin bristles line the dorsal and ventral sides of the DV boundary of the wing (arrow). (C) Cuticle of anubbin 1 wing margin prepared as in (B). Bristles are found at a distance from the DV boundary (arrow). The DV identity of bristles is ambiguous in the nubbin 1 wing margin, possibly because clones fail to respect the DV lineage restriction in nubbin 1 wings (12). (D) Wild-type and (E)nubbin 1 mutant wing discs labeled for Nubbin protein (red) and for a wingless-lacZ reporter gene [green, visualized with anti–β-Gal. (F) Wild-type and (G) nubbin 1 mutant wing discs labeled for a cut-lacZ reporter gene.

The row of sensory bristles that makes up the wing margin is disorganized in nubbin mutant wings (11), suggesting a defect in Wingless or Notch activity. In preparations where the wing margin is viewed edge on, this disorganization reflects a broadening of the region where bristles form (Fig. 1, B and C). Margin bristles are normally specified in cells very close to the DV boundary, reflecting a requirement for high levels of Wingless signaling activity (13). The broadening of the margin suggests that Wingless might be ectopically expressed innubbin mutant wing discs. Wingless is normally expressed in a stripe of two to three cells straddling the DV boundary (Fig. 1D). Innubbin mutant discs this stripe is widened considerably (Fig. 1E). Expression of the Notch targets vestigial andcut is similarly expanded at the DV boundary innubbin mutants (Fig. 1, F and G) (14).

To determine whether the effect on bristle specification is a direct consequence of removing nubbin activity, we generated clones of nubbin mutant cells in a wild-type background. Ectopic wing margin bristles were found in nubbin mutant clones located near the endogenous wing margin (Fig. 2A) (11, 15). Thenubbin mutant clones showed ectopic expression ofneuralized-lacZ, a molecular marker for precursors of the sensory neurons that innervate the bristles (Fig. 2B). Ectopic bristle precursors were usually confined to the clone, but in rare instances they arose in adjacent wild-type cells (14). Specification of wing margin sense organs and induction of neuralized-lacZare known to depend on localized expression of Wingless at the DV boundary (13). Clones of cells that are simultaneously mutant for wingless and nubbin do not show ectopic neuralized-lacZ staining (Fig. 2C), suggesting that ectopic bristle specification in nubbin mutant cells is due to ectopic Wingless activity. The nubbin mutant clones misexpressed wingless and vestigial (Fig. 2, D and F). The largely autonomous effect of nubbin mutant clones on bristle specification may be due to the relatively low levels of ectopic Wg protein expressed in nubbin mutant clones (14). Together with the results on cut expression (Fig. 1), these observations suggest that Notch target genes are transcriptionally up-regulated in nubbin mutant cells near the DV boundary.

Figure 2

Cell-autonomous effects of nubbinmutant clones. (A) nubbin 1 mutant clone in the adult wing (15). Of 26 clones examined, 20 showed ectopic bristles distant from the wing margin (arrow). We observed one adult clone that showed ectopic bristle differentiation immediately outside of the clone border (14). All clones caused buckling of the wing surface, caused by formation of some ectopic wing-vein material and possibly mild overgrowth. Consequently, the field of wing hairs is not all in the same focal plane. Thenubbin 1 mutant cells (outlined) are marked by loss of a forked + transgene on the wild-type chromosome. The forked marker labels bristles and wing hairs. (B to F) nubbin 1mutant clones in the wing disc were visualized by the absence of Nubbin protein (red). Expression of the lacZ reporter genes,neuralized-lacZ, wingless-lacZ, andvestigial-lacZ, was visualized with anti–β-Gal (green). Overlap of the two signals appears yellow. Clones were induced in mid–second-instar larvae except as indicated. All discs are shown with dorsal to the top and anterior to the left. (B)neuralized-lacZ in a wing disc carrying anubbin 1 mutant clone (nub ). Ectopic expression ofneuralized-lacZ (arrow) was observed in 7 out of 10nubbin mutant clones examined. Clones were induced in first-instar larvae to get large clones that cross the DV boundary. Two of 10 examples showedneuralized-lacZ expression in wild-type cells adjacent to the clone (14). (C) neuralized-lacZ expression in a wing disc carrying a wingless CX4 nubbin 1 double-mutant clone (nub wg ) (15). The clone is ventral anterior and abuts the row of neuralized-lacZ–expressing cells. The anterior compartment of this disc is slightly distorted so that the row of neuralized-lacZ–expressing cells bends more sharply than usual. neuralized-lacZ was not misexpressed in 7 out of 7 clones examined (arrow). (D) wingless-lacZ expression in a wing disc carrying a nubbin 1 mutant clone.wingless-lacZ was ectopically expressed in cells lacking Nubbin (arrow) in 13 out of 13 clones examined. We also found that 15 out of 17 clones examined misexpress Wingless protein (14). Ectopicwingless expression was also observed in clones that do not touch the DV boundary. (E) wingless-lacZ expression in a wing disc carrying a nubbin 1 Su(H) double-mutant clone (nub Su(H) ;15). wingless-lacZ is not expressed in 11 out of 11 nubbin 1 Su(H) double-mutant clones examined. (F) Expression of the vestigial DV boundary enhancer in a wing disc carrying a nubbin 1 mutant clone. Ectopic expression of vestigial-lacZ in cells lacking Nubbin (arrow) was observed in 7 out of 11 clones examined.

To test whether ectopic activation of these genes in nubbinmutant clones directly depends on Notch signaling activity, we generated clones of cells that were simultaneously mutant fornubbin and Suppressor of Hairless[Su(H)]. Su(H) encodes a DNA-binding protein that mediates transcriptional activation of Notch target genes (16). Su(H) is autonomously required for the expression of wingless, vestigial, andcut at the DV boundary (4) and binds directly to the vestigial DV boundary enhancer (17). Clones of cells mutant for both nubbin and Su(H) do not ectopically activate wingless (Fig. 2E), demonstrating that ectopic expression of wingless in nubbin mutant cells depends on activity of the Notch pathway. To confirm that Nubbin acts downstream of Notch, we tested whether overexpression of Nubbin could suppress the effects of a ligand-independent form of Notch (Notch[intra]) (18). Expression of Notch[intra] causes ectopic Wingless expression in the wing disc (1) (Fig. 3A). When Nubbin is coexpressed with Notch[intra], ectopic Wingless expression is strongly reduced (Fig. 3B). Together, these observations suggest that Nubbin may act as a direct repressor of Notch-dependent target gene expression (19). These findings argue that the effects of Nubbin are unlikely to be mediated by indirect effects on expression of Notch ligands (20).

Figure 3

Nubbin acts downstream of Notch. Wing imaginal discs were simultaneously labeled for Notch protein (red), Nubbin protein (green), and Wingless protein (blue) (19). Wg expression is shown separately below (in black and white). (A) Disc of genotype dpp-GAL4/UAS Notch(intra). Notch(intra) expression was driven by dpp-GAL4 in a stripe along the AP compartment boundary (perpendicular to the endogenous stripe of Wg expression along the DV boundary). Notch(intra) induced a high level of ectopic Wg expression (arrow; compare with the narrower endogenous Wg domain). The normal Nubbin expression domain is shown in green. (B) Disc of genotype UAS-Nub/+; dpp-GAL4/UAS Notch(intra). Both Nubbin and Notch(intra) are expressed in the dpp-GAL4 stripe. Ectopic Wg expression is prevented in the cells where a high level of Nubbin is expressed (strong green label, arrow) together with notch(intra). A small area of ectopic Wg expression is apparent near the DV boundary where Nubbin is not overexpressed (arrowhead).

Ectopic expression of wingless and vestigial innubbin mutant clones indicates that the Notch signaling pathway is active in cells at a considerable distance from the wing margin. Notch is activated by Delta and Serrate. The broad distribution of both ligands in the developing wing disc poses a problem in limiting high-level activation of the Notch pathway to cells near the DV boundary. This is partly solved by modulating the sensitivity of Notch for Serrate and Delta through Fringe activity (9, 10). Serrate is expressed only in dorsal cells at the time when boundary-specific expression of wingless andvestigial is induced; nonetheless, Serrate activates Notch in ventral cells. Dorsal cells, which express both Fringe and Serrate, are refractory to Serrate at this stage (1, 2, 5). Delta is expressed both dorsally and ventrally, but appears to activate Notch mainly in dorsal cells (3, 5, 21). Although Fringe modulates the sensitivity of Notch to Serrate and Delta, our findings indicate that Fringe is not sufficient to limit high-level Notch activity to cells near the wing margin. In the absence of Nubbin, Notch targets are activated in cells at a distance from the boundary. We can estimate the range over which Notch activity is sufficient to induce boundary-specific genes by examining wherenubbin 1 mutant clones induce ectopic expression of wingless or vestigial. Clones located more than 10 cell diameters from the DV boundary do not induce target genes (as seen in 20 out of 20 examples) (14).

These observations suggest that Notch is activated in a broad region centered on the DV boundary and that Nubbin antagonizes the ability of Notch to induce its target genes. If so, overexpression of Nubbin should interfere with endogenous expression of Notch targets at the DV boundary. Large Nubbin-expressing clones (22) cell-autonomously reduce Wingless expression when they cross the DV boundary (Fig. 4). Thus Nubbin appears to act as a repressor that competes with a Notch-dependent activation signal to determine the amount of target gene expression.

Figure 4

Nubbin overexpression represses Wingless expression at the DV boundary. (A) Portion of a wing imaginal disc with a clone of cells overexpressing Nubbin under Gal4 control (22). The clone expresses GAL4 under control of the Act5C promoter (act >> Gal4) and directs expression of both β-Gal (UAS-lacZ) and Nubbin (UAS-Nub). The disc was labeled with anti–β-Gal to mark the clone (red) and with anti Wingless (green). Large Nubbin-expressing clones reduce Wingless expression (nine out of nine clones examined). Small Nubbin-expressing clones, induced later in development, show a weaker effect (14). (B) Same disc as in (A), but showing Wingless expression alone. The clone is outlined (white).

To examine whether nubbin might directly regulate Notch target genes, we tested Nubbin protein binding to thevestigial boundary enhancer (23). In deoxyribonuclease I footprinting experiments, Nubbin bound to a cluster of four sites between residues 260 and 400, as well as to two weaker sites (Fig. 5) (24). The cluster of Nubbin binding sites is well separated from the single Su(H) binding site required for Notch-dependent activation of this enhancer (residues 100 to 108) (17) (diagram in Fig. 5B). To determine whether Nubbin mediates repression of the boundary enhancer through these binding sites, we compared the expression of a lacZ reporter gene under control of the wild-type enhancer with that under control of a mutant enhancer from which the central two Nubbin binding sites had been deleted. We observed considerable broadening of the stripe of reporter gene expression in the mutant enhancer (Fig. 5C) (24).

Figure 5

Nubbin represses thevestigial boundary enhancer. (A) deoxyribonuclease I footprinting of the vestigial boundary enhancer with bacterially expressed Nubbin protein. The probe was asymmetrically end-labeled by end repair at a unique Age I site. AG lanes: A+G chemical cleavage sequencing reaction. (−) No added protein. Nubbin protein was diluted 1:10, 1:30, 1:90, and 1:120 (left to right). Four strong binding sites are indicated by red brackets. The positions of these sites are indicated as red boxes in (B) and (C). The uppermost binding site is located toward the left end of the enhancer, as depicted in the maps in (B) and (C). Protected sequences (sites from left to right in B): TTATACAAGCCGC, TTATGTAAGTAACC, TTTTGCATGCCCAT, and CCGCCTGGATATTGCGC. POU protein binding sites do not have a simple consensus (28). The asterisks indicate the positions of the restriction sites used to delete binding sites in (C). (B) Expression of a lacZ reporter gene by the modified wild-type vestigial enhancer (24), visualized by X-Gal staining for β-Gal activity. The map shows the positions of the four strong Nubbin binding sites (red boxes). Two weaker Nubbin sites are indicated by pink boxes. The single Su(H) site is indicated by a black box. The Sac II site is indicated by one asterisk, the Sph I site by two asterisks. (C) Two Nubbin binding sites were removed from the boundary enhancer by deleting sequences between the Sac II and Sph I sites (24). The deleted region (Δ) is indicated by brackets. To permit direct comparison, we prepared and processed in parallel samples expressing the wild-type and mutant enhancers.

Nubbin and Notch appear to play opposing roles in the regulation of boundary-specific genes. Nubbin acts as a general repressor ofwingless, vestigial, and other Notch targets in the wing primordium. Activation of the Notch signaling pathway to high levels at the DV boundary provides sufficient stimulation to override repression by Nubbin. The idea that Nubbin sets a threshold level for Notch activity is supported by observations that overexpression of Nubbin can prevent Notch from activating Wingless at the DV boundary.

Tight regulation of Notch signaling is necessary for normal wing development. We have shown that Notch is activated in a broad region of the wing at levels sufficient to induce boundary-specific target genes and that Nubbin appears to limit the effective width of this domain so as to create a sharp boundary. Members of the fringe gene family also contribute to limiting Notch activity to cells near the DV boundary (2, 8). At later stages of wing development a third mechanism comes into play. Wingless limits its own expression to the DV boundary, possibly by modulating Notch activity (25) or by modulating the late expression of Notch ligands (26).

Three distinct mechanisms are used to spatially limit Notch activity in boundary formation in the Drosophila wing. The complex, multilevel control of Notch signaling highlights both the importance of tightly regulating Notch activity to allow precise boundary formation and the difficulty in achieving the necessary precision of regulation. Notch signaling and members of the fringe gene family have been implicated in boundary formation in vertebrate development (7, 8). It will be of interest to determine whether a factor similar to Nubbin functions to restrict Notch signaling in vertebrate boundary determination.

  • * Present address: MPI for Developmental Biology, D-72076 Tubingen, Germany.

  • To whom correspondence should be addressed. E-mail: scohen{at}


View Abstract

Navigate This Article