During the development of the Drosophila embryonic epidermis, the secreted Wingless protein initially spreads symmetrically from its source. At later stages, Wingless becomes asymmetrically distributed in a Hedgehog-dependent manner, to control the patterning of the embryonic epidermis. When Wingless is misexpressed in engrailed cells in hedgehog heterozygous mutant embryos, larvae show a dominant phenotype consisting of patches of naked cuticle in denticle belts. This dose-sensitive phenotype is a direct consequence of a change in Wg protein distribution. We used this phenotype to carry out a screen for identifying genes regulating Wingless distribution or transport in the embryonic epidermis. Using a third chromosome deficiency collection, we found several genomic regions that showed a dominant interaction. After using a secondary screen to test for mutants and smaller deficiencies, we identified three interacting genes: dally, notum, and brahma. We confirmed that dally, as well as its homolog dally-like, and notum affect Wingless distribution in the embryonic epidermis, directly or indirectly. Thus, our assay can be used effectively to screen for genes regulating Wingless distribution or transport.

---

SECRETED signaling molecules play an essential role in patterning developing metazoans. Work from recent years shows that the signaling activity of these molecules is tightly regulated (FREEMAN 2000). Among various levels of regulation, controlling the distribution of ligands in a field of cells is necessary to establish stable gradients that robustly pattern developing tissues (VINCENT and DUBOIS 2002). The distribution of ligands could conceivably be influenced by many factors, including the concentration of receptors, the composition of the extracellular matrix, or the rate of recycling and degradation of the ligand following internalization. The best-documented case is the regulation of ligand distribution by receptors. For example, in Drosophila, Hedgehog (Hh) range of action is regulated by its receptor Patched: if the Hh-binding domain of Patched is mutated, the spread of Hh is extended in the wing imaginal disc (CHEN and STRUHL 1996). Another case is the morphogen Decapentaplegic, whose concentration gradient is regulated by the amount of its receptor Thickveins (LECUIT and COHEN 1998; TANIMOTO et al. 2000). In addition to canonical receptors, low-affinity receptors such as heparan sulfate proteoglycans are likely to affect ligand distribution. For instance, in wing discs, mutants that impede the synthesis of heparan sulfate change Hh distribution (BELLAICHE et al. 1998; THE et al. 1999; TAKEI et al. 2003; HAN et al. 2004a,b).

Gradient formation also depends upon the mechanism by which signaling ligands move in the plane of epithelia. It is not yet clear what these mechanisms are, but two main modes of transport have been proposed (reviewed by VINCENT and DUBOIS 2002; GONZALEZ-GAITAN 2003): (1) facilitated diffusion, where ligands diffuse within the extracellular space, and this diffusion is modulated by cell-surface receptors, and (2) transcytosis, where ligands are transported along the plane of the epithelium by repeated cycles of endocytosis and recycling to the cell surface.

In this study, we searched for genes that influence the distribution or transport of the signaling ligand Wingless (Wg), the homolog of vertebrate Wnt-1 in Drosophila. In the embryonic epidermis, Wg acts at a short range to regulate the activity of target genes, whereas it behaves as a long-range morphogen in the wing imaginal disc (reviewed by SETO and BELLEN 2004). In the wing, Wg forms a stable gradient with a source localized at the boundary between ventral and dorsal compartments and triggers the expression of target genes at different distances from its source in a concentration-dependent manner (ZECCA et al. 1996; NEUMANN and COHEN 1997).

Both facilitated diffusion and transcytosis have been proposed to explain how Wg moves from cell to cell. In the embryonic and imaginal disc epithelia, Wg protein is detected in the basolateral extracellular space as well as in bright intracellular dots localized to the apical side of the cells (PFEIFFER et al. 2000; STRIGINI and COHEN 2000; GRECO et al. 2001; SIMMONDS et al. 2001). These punctate structures are thought to coincide with both exocytic and endocytic vesicles (VAN DEN HEUVEL et al. 1989; GONZALEZ et al. 1991; PFEIFFER et al. 2000, 2002), and have been proposed to be involved in the transport of Wg from cell to cell in embryos (transcytosis model) (BEJSOVEC and WIESCHAUS 1995; MOLINE et al. 1999). In the wing, some of these structures colocalize with vesicles termed argosomes, which have also been suggested to move from one cell to the next through the endocytic compartment (GRECO et al. 2001). On the other hand, other studies have proposed that these vesicles represent Wg secretion rather than Wg movement (PFEIFFER et al. 2000; STRIGINI and COHEN 2000). A second model proposed for Wg transport is diffusion through the extracellular space. In support of this model, STRIGINI and COHEN (2000) found that most of the Wg gradient is extracellular in the wing disc and that the speed of Wg movement is compatible with diffusion within the extracellular space. Wg tightly associates with the cell surface (REICHSMAN et al. 1996), so it is unlikely to diffuse freely, but it could spread through facilitated diffusion by interacting with cell surface receptors. Consistent with this, one of the factors that influences the shape of the Wg gradient in the wing disc is the receptor Frizzled2 (Fz2) (CADIGAN et al. 1998). When Fz2 is overexpressed, Wg appears to be stabilized at the cell surface. The amount of detectable Wg protein increases and this correlates with a broadening of Wg-signaling range. Recently, another factor found to affect the shape of the Wg gradient in wing discs is Notum, a hydrolase that is thought to modify the heparan sulfate chains of the glypicans Dally and Dally-like (GERLITZ and BASLER 2002; GIRALDEZ et al. 2002). When Notum is overexpressed in the disc, the spread of Wg is reduced (GIRALDEZ et al. 2002). Apart from transcytosis and facilitated diffusion, a third mode of transport for Wg has been documented in the embryonic epidermis: cells expressing Wg are able to move away from the source while transporting active Wg protein in secretory vesicles (PFEIFFER et al. 2000).

Here we focus our analysis of Wg distribution and transport to the embryonic epidermis. An interesting aspect of the embryonic Wg gradient is that it becomes asymmetric early in embryogenesis (GONZALEZ et al. 1991). In the abdomen, Wg is expressed in a single row of cells in each embryonic segment, immediately anterior to two to three rows of cells expressing engrailed (en) and hh (Figure 1A). Wg, en, and hh maintain each other's expression initially and then act as a bipartite organizer to control intrasegmental patterning (reviewed in HATINI and DINARDO 2001; SANSON 2001). Early on (at stage 9/10), Wg spreads symmetrically from its source over a few cell diameters. This stage coincides with the time when en expression in the adjacent cells requires Wg signaling for its maintenance. Later on (from stage 11 onward), the Wg gradient becomes asymmetric (Figure 1A). Wg is undetectable posterior to the wg-expressing cells, except occasionally in some cells immediately adjacent to the source (SANSON et al. 1999). By contrast, Wg is detectable over a few cell diameters in the anterior direction. This asymmetric distribution of Wg protein translates into an asymmetric range in Wg signaling. At midembryogenesis, Wg signaling represses the expression of shavenbaby, a gene required for denticle formation in the embryonic epidermis (PAYRE et al. 1999). Thus, the secretion of smooth or naked cuticle provides a convenient readout of Wg-signaling activity (LAWRENCE et al. 1996). Matching the asymmetric distribution of Wg protein, naked cuticle is secreted four to five cell diameters anterior to the Wg source, but only one cell diameter posterior to it, which corresponds to the anterior-most row of Engrailed-expressing cells (DOUGAN and DINARDO 1992; O'KEEFE et al. 1997; SANSON et al. 1999) (Figure 1A).

View larger version (51K): In this window In a new window Download PPT slide   FIGURE 1.— Shape of the Wingless embryonic gradient in different genetic contexts. The diagrams depict the distribution of Wg protein at stage 12 in different genetic contexts. For clarity, Wingless protein distribution has been drawn above the cells, but in the epithelium the Wingless protein is detected both in the extracellular space and in bright intracellular dots, which are mostly apical (see Introduction). On the same diagrams the type of cuticle secreted by each cell at the end of embryogenesis is represented. The corresponding cuticle pattern is shown below. (A) In wild-type embryos, the Wg gradient is asymmetric after stage 11. Wg is expressed in one row of cells in each segment (green cell in diagram). At stage 12, Wg protein can be detected up to four to five cells anterior to the Wg source, but only occasionally in the first row of En cells (in red). As a consequence of this asymmetric gradient, naked cuticle is later on secreted up to four to five cell diameters anterior to the Wg source, but only in the adjoining En cells in the posterior direction. At the time of cuticle deposition (from stage 15 onward), the En stripe is, on average, two cells wide, with the second cell row secreting the first row of each denticle belt in abdominal segments A2–A7. In these segments, each belt is made of six rows of cells making denticles. (B) When Wg is expressed in the En cells, the gradient remains asymmetric. More En cells are specified, and they express a high level of Wg protein. As a consequence, row 1, which should be specified by the posterior-most En cell, disappears completely or partially in all abdominal denticle belts (arrowheads). However, immediately posterior to the En domain, the denticle rows are specified normally, and no Wg protein can be detected there. (C) Expressing Wg in the En cells allows us to remove hh without losing En expression. This reveals the role of Hh in making the Wg gradient asymmetric: in enGal4/UASwg[hh–/hh–] embryos, Wg protein is detected posterior as well as anterior to the En domain, and the gradient is symmetric in shape. Consequently, naked cuticle is made over four to five cell diameters posterior as well as anterior to the En domain. Instead of denticle belts, a thin stripe of denticles similar to row 6 denticles is secreted. (D) An intermediate phenotype is seen in enGal4/UASwg[hh–/+] embryos: small regions of naked cuticle are found within denticle belts, often replacing denticle rows 2–5 (arrowheads).

 
One mechanism that contributes to the asymmetric shape of the Wg gradient is the clearing of the receptors Fz and Fz2 from the En domain through downregulation of their transcription (LECOURTOIS et al. 2001). Presumably, the decrease in the concentration of Fz and Fz2 destabilizes Wg, perhaps by making it accessible to proteases (CADIGAN et al. 1998). However, although this is enough to restrict Wg range initially, an additional mechanism, which involves accelerated degradation of Wg protein posterior to the Wg source, is required later on (DUBOIS et al. 2001). Epidermal growth factor receptor signaling is required for this increase in Wg turnover, although the molecular mechanism is not known (DUBOIS et al. 2001).

In terms of genetic regulation, Hh, which is secreted by the En cells, is the top regulator in making the embryonic Wg gradient asymmetric (SANSON et al. 1999). This was shown by expressing wg in the En domain—using the UAS/Gal4 system—in a hh mutant background. In enGal4/UASwg [hh^–/hh^–] embryos, the expression of en and wg is maintained because of the artificial regulatory loop between the two genes, which bypasses the need for Hh signaling. In enGal4/UASwg embryos, row 1 of each denticle belt is missing but the shape of the gradient remains asymmetric (Figure 1B). By contrast, in enGal4/UASwg [hh^–/hh^–] embryos, Wg is detected in a symmetric gradient, with Wg spreading posterior as well as anterior to the source (SANSON et al. 1999) (Figure 1C). This is reflected in the cuticle pattern: naked cuticle is secreted four to five cell diameters on both sides of the source (Figure 1C). The same phenotype is seen when cubitus interruptus, a downstream component of the Hh pathway, is removed in enGal4/UASwg embryos (SANSON et al. 1999). This shows that Hh signaling controls the expression of one or several genes that regulate Wg distribution.

The requirement for Hh signaling is dose sensitive: when one copy (instead of two) of Hh is removed in enGal4/UASwg embryos, patches of naked cuticle are found within denticle belts (Figure 1D). We used this dominant phenotype in a F₁ screen to find genes giving a similar phenotype, and thus potentially regulating Wg distribution. We conducted a deficiency screen on the third chromosome to identify such genes. Two successive screens led us to the identification of three genes that affect Wg distribution in the embryonic epidermis: dally, dally-like, and notum. An additional role for dally-like in Hh signaling, found while analyzing these results, has been described elsewhere (DESBORDES and SANSON 2003).

Fly stocks:

The Bloomington Drosophila Stock Center (BDSC) provided the third chromosome deficiency kit (see list of stocks in the APPENDIX) as well as Df(3L)Scf-R11, Df(3R)l26c, Df(3R)ry75, Df(3R)urd, and Df(3R)lc4a. The Umea Stock Center (USC) provided Df(3L)th102; Ken Cadigan, Df(3L)Dfz2; Daniel Kalderon, Df(3L)A27; and Daniel St Johnston, Df(3L)XS-543. We obtained the following mutant stocks from the BDSC: l(3)06464, l(3)01629, arf^72A, brm², pip¹, kni¹⁰, Dl³, and hh^AC. Dfz2^C1 was provided by Gary Struhl, dally¹⁰ and dally^E385 by Steve Kerridge, Psn^B3 and Psn^C4 by Mark Fortini, pip² and kni⁸ by USC, brm^T362 and brm^T485 by Jessica Treisman, and not³ and not⁵ by Steve Cohen.

The transgenic strains used were enGal4 (Andrea Brand, University of Cambridge, UK), UASwg (LAWRENCE et al. 1996), and UASGFP (Nick Brown, University of Cambridge, UK). We combined these lines to generate the following genotypes: w; enGal4/+; UASwg/+ (abbreviated as enGal4/UASwg); w; enGal4/+; UASwg.hh^AC/hh^AC (abbreviated as enGal4/UASwg[hh^–/hh^–]); and w; enGal4/+; UASwg.hh^AC/+ (abbreviated as enGal4/UASwg[hh^–/+]).

To generate germline clones, we used the stocks y w hsflp; Dr/TM3 (BDSC), w; FRT80B ovo^D1/TM3Sb (from D. St Johnston), and w; FRT80B not³/TM6B (from S. Cohen). y w hsflp; ; FRT80B ovo^D1/ FRT80B not³ larvae were heat-shocked 1 hr at 37° between 72 and 96 hr of development, and another heat shock was performed 24 hr later. The females derived from these larvae were crossed with either w; FRT80B not³/TM3actin-GFP males to analyze the cuticle of the embryos or w; enGal4.UASGFP; FRT80B not³/TM3hunchback-lacZ males to analyze embryos by immunostainings. The marked balancers TM3actin-GFP (TM3GFP) and TM3hunchback-lacZ (TM3LacZ) were supplied by the BDSC.

Screening procedure:

Each deficiency stock from the third chromosome kit was crossed to the enGal4 stock to generate engal4/+; deficiency/+ F₁ males. We used dominant markers present on the balancer chromosome in the deficiency stock to identify unambiguously the F₁ males carrying the deficiency (in several cases we had to rebalance the stock or use different dominant markers to facilitate the identification). For a few deficiencies from the kit, the F₁ engal4/+; deficiency/+ males did not survive or were sterile. About 50 UASwg virgins were then crossed to 10–30 F₁ engal4/+; deficiency/+ males (see scheme in Figure 2A), and embryos were collected overnight at 25° on grape juice agar plates. The temperature was controlled carefully since the UAS/Gal4 system is sensitive to temperature. Note that we had to use F₁ engal4/+; deficiency/+ males instead of F₁ engal4/+; deficiency/+ females in the screen, because the enGal4; UASwg embryos derived from engal4/+ females were deformed and presented many gaps of naked cuticle in the denticle belts. The reason for this phenotype is unclear (enGal4 might be weakly expressed maternally), but it made it impossible to consistently score additional gaps from interacting deficiencies.

View larger version (38K): In this window In a new window Download PPT slide   FIGURE 2.— Screen for dominant phenotypes in the enGal4/UASwg-sensitized background. (A) To produce enGal4/UASwg embryos heterozygous for a given deficiency chromosome, two successive crosses were performed: first, enGal4 flies were crossed to flies carrying a deficiency over a balancer chromosome with a dominant visible mutation. In the next cross, F1 males that do not exhibit the dominant visible mutation, and therefore are heterozygous for enGal4 and the deficiency chromosome, were selected and crossed to virgin UASwg females. Half of the progeny from this cross do not hatch at the end of embryogenesis: these are the embryos carrying enGal4/UASwg, which have a deformed head and cannot get out of the egg shell. All the dead embryos are collected and mounted ventral side up for cuticle analysis. (B) Ventral side of wild-type larvae just after hatching (head is up). The eight abdominal segments are indicated. Row 1 is clearly seen in A2–A8. (C) Close-up of denticle belt in wild-type A3 segment. Row 1 is indicated by arrowheads. (D) Close-up of denticle belt in the A3 segment of enGal4/UASwg embryos. Row 1 is clearly missing, which allows us to distinguish unambiguously the dead embryos of the enGal4/UASwg genotype from occasional wild-type dead embryos. (D') Dominant phenotype found in enGal4/UASwg[hh–/+] embryos. A region of ectopic naked cuticle is shown in the denticle belt of segment A3 (arrowhead). We score this phenotype as a gap in the screen. (E) The denticle belt in segment A1 is thinner than that in other abdominal segments. (E') For segment A1, we scored a gap when the belt was completely breached by ectopic naked cuticle, as shown here in a enGal4/UASwg[hh–/+] embryo (arrowhead).

 
Since enGal4/UASwg embryos die at the end of embryogenesis, the collected embryos were aged at least 30 hr so that only nonviable embryos remain unhatched. Dead embryos were dechorionated using bleach and their vitelline membrane was removed by hand in Hoyer's on a slide. Embryos were aligned ventral side up in a mixture of Hoyer's and lactic acid (1:1). Cuticle preparations were baked overnight at 55°.

Gaps were scored by visualizing the cuticles with phase-contrast microscopy. For abdominal belts A2–A7, we scored a gap when a patch of smooth cuticle replaced most rows of denticles in the belt (Figure 2, D vs. D'). Because A1 is thinner than the other abdominal denticle belts, we scored only complete breaches of this belt (Figure 2, E vs. E'). We eliminated segment A8 from our analysis, because it was too often folded and difficult to visualize.

To assess the significance of our results, we compared the number of gaps obtained for each deficiency with the number of gaps obtained for wild-type chromosomes. We scored five gaps for 354 enGal4/UASwg embryos examined, when chromosomes from the wild-type yw⁶⁷ or canton^S strains were tested in place of a deficiency chromosome in the screening scheme (Table 1). We used this value as the expected frequency of gaps and employed a standard ² test to compare this expected frequency with the frequency of gaps observed for each deficiency. When the P-value was <0.05, a deficiency was considered not interacting (APPENDIX).

View this table: In this window In a new window   TABLE 1 Testing wild-type and hedgehog mutant chromosomes in the screen

 

RNA interference:

The technique of RNA interference (RNAi) was carried out following the protocol described in DESBORDES and SANSON (2003). Briefly, 300-nucleotide-long sequences from the plasmids pBS(KS)-dlp (BAEG et al. 2001) and pBS(KS)-dally (NAKATO et al. 1995) were amplified using primers pairs containing a T7 promoter sequence (5'-TAATACGACTCACTATAGG-3') at the 5'-end. The PCR products were used as templates for T7 transcription reactions with the Ribomax large-scale RNA production kit (Roche). Following phenol/chloroform extraction and ethanol precipitation, the double-stranded RNA (dsRNA) was resuspended in Spradling injection buffer and injected at a concentration of 8 µg/µl in blastoderm embryos.

In situ hybridizations of whole-mount embryos:

Embryos were fixed and hybridized with a digoxygenin single-stranded RNA probe as described by (JOWETT 1997), except that no proteinase K treatment was performed. The wingless cDNA was a gift from J. P. Vincent.

Antibody stainings:

Immunofluorescence was performed on fixed embryos according to standard protocols. Primary antibodies used were mouse anti-Wingless 4D4 (1:10) (from the Developmental Studies Hybridoma Bank) and rabbit anti-GFP (1:300) (Abcam). Secondary antibodies were Alexa fluorescent conjugates from Molecular Probes (Eugene, OR; Alexa 488 and Alexa 594). The GFP antibody staining was used to label Engrailed-expressing cells in embryos carrying the enGal4.UASGFP transgene combination (see Figure 5, E–E'', and Figure 6, C–D'). To visualize the Wg gradient, five to seven confocal sections were collected every 0.3 µm, starting from the apical surface of the cells, and projected. Sections were collected on a Bio-Rad (Richmond, CA) MRC1024 confocal microscope.

View larger version (88K): In this window In a new window Download PPT slide   FIGURE 5.— The symmetrical Wingless gradient in enGal4/UASwg[dlp RNAi] embryos is a consequence of loss of Hedgehog signaling. (A–D) wg trancription revealed by in situ hybridization in late stage 11 embryos. Embryos are voluntarily oriented with the head to the right, to keep the anterior side of each segment to the left, as in the close-up views and the diagrams. (A–A'') In wild-type embryos, wg is transcribed in a single row of cells in each segment, just anterior to the En-expressing cells. (B–B'') In enGal4/UASwg embryos, wg is transcribed ectopically in the En domain, which occupies two to three rows of cells in each segment. The endogenous expression of wg is visible as a row of cells immediately anterior to the domain of ectopic wg expression (arrowheads in B'). It is possible to distinguish unambiguously ectopic from endogenous wg expression without double label, because endogenous wg expression is restricted at this stage to the ventral cells, whereas ectopic wg expression in the En domain spans the circumference of the embryo. (C–C'') In a hh null mutant background, ectopic wg expression is maintained in enGal4/UASwg embryos, because the regulatory loop created between En and Wg bypasses the need for Hh signaling. However, outside of the En domain, endogenous wg expression still requires Hh for its maintenance and is consequently lost in these embryos (no ventral stripe of endogenous wg expression is seen). (D–D'') Endogenous wg expression is also lost in enGal4/UASwg[dlp RNAi] embryos, suggesting that Hh signaling has failed in these embryos. If Hh signaling is normal, wg endogenous expression should be maintained. Note that wg ectopic transcription in the Engrailed domain is not as robust as in enGal4/UASwg[hh–/hh–] embryos. (E–E''') Wg protein distribution in enGal4/UASwg[dlp RNAi] embryos. The En domain is labeled in red and Wg protein in green in the merge (E''). Wg protein is detected posterior, as well as anterior to the En domain, in dots (arrowheads). This shows that the Wg gradient is symmetric in enGal4/UASwg[dlp RNAi] embryos (E'''), contrary to enGal4/UASwg embryos where the Wg gradient is asymmetric (see Figure 1 and also SANSON et al. 1999; DUBOIS et al. 2001).

 

View larger version (123K): In this window In a new window Download PPT slide   FIGURE 6.— Embryos mutant for notum show ectopic naked cuticle and a broader Wg gradient. (A) not5 embryo showing excess naked cuticle in segments A1, A2, A3, A5, and A6. Note that the denticle belt in segment A5 has been completely replaced by naked cuticle (arrowhead). (B–B'') Embryos derived from not3 germline clones; i.e., embryos both maternally and zygotically mutant for notum exhibit a range of cuticle defects. Some embryos are ball shaped (B) or U shaped (B'), while others have an incompletely retracted posterior end (B''). In addition, most embryos for which the cuticle pattern can be analyzed show a narrowing of the ventral denticle belts, with patches of naked cuticle in the belts or denticle belts missing (arrowhead in B''). (C and D) Stage 12 embryos derived from females bearing not3 germlines clones and not3/TM3LacZ males were stained for Wg (C and D and green signal in C' and D'). These embryos also carry enGal4.UASGFP to label the engrailed domain using an antibody against GFP (red signal in C' and D'). LacZ-positive embryos are paternally rescued and show a Wg gradient indistinguishable from wild-type embryos (C and C'). LacZ-negative embryos are maternally and zygotically mutant for notum and show a Wg gradient that is broader and weaker (D and D').

 

Screen design:

In stage 12 wild-type or enGal4/UASwg embryos, the Wg gradient is asymmetric, and as a consequence, naked or smooth cuticle is secreted anterior to the Wg source, whereas a belt of denticles is secreted posterior to it in each abdominal segment at the end of embryogenesis (Figure 1, A and B; see also Figure 4, A and B). The only difference between the cuticle phenotype of wild-type and enGal4/UASwg embryos is the loss of the first row of denticles in the latter embryos. The patches of naked cuticle found within the denticle belts of enGal4/UASwg [hh^–/+] embryos correspond to regions of the epithelium where Wg diffuses symmetrically (Figure 1D) (SANSON et al. 1999). We call these patches "gaps" and refer to the dominant phenotype as a "gap" phenotype. We reasoned that genes controlling the shape of the Wg gradient downstream of Hh signaling should exhibit a similar dominant phenotype. On the basis of this hypothesis, we screened the Bloomington collection of large deficiencies on the third chromosome for those that give a gap phenotype in the enGal4/UASwg background. As a control, we also tested in this sensitized background two wild-type chromosomes (yw⁶⁷ and canton^S) and three different hh mutations: a hypomorph (hh⁵), a null allele (hh^AC), and a deficiency removing the hh locus (Df(3R)hh).

View larger version (116K): In this window In a new window Download PPT slide   FIGURE 4.— dally-like or dally RNAi in enGal4/UASwg embryos produces ectopic naked cuticle posterior to the Wingless source. (A) Wild-type larva and (A') close-up of abdominal denticle belt. Denticle rows 1–6 are visible. (B) enGal4/UASwg larva: row 1 is completely or partially missing in denticle belts of abdominal segments A2–A7. (B') Close-up showing a belt where row 1 is missing completely. Rows 2–6 are intact. (C) enGal4/UASwg[hh–/hh–] embryo: an identical expanse of naked cuticle is found posterior and anterior to the Wg source. (C') Close-up showing that rows 2–5 have been replaced by naked cuticle. (D) enGal4/UASwg[dlp RNAi] embryo and close-up (D') showing a very similar phenotype to enGal4/UASwg[hh–/hh–] embryos. (E and E') enGal4/UASwg[dally RNAi] embryo shows an intermediate phenotype with ectopic naked cuticle in some of the denticle belts.

 
The crossing scheme for the screen is given in Figure 2A. First, enGal4 flies were crossed with flies carrying the chromosome to be tested. Then F₁ males carrying one copy of enGal4 and one copy of a deficiency chromosome (or a wild-type or mutant chromosome) were crossed with UASwg females to generate enGal4/UASwg [deficiency/+] embryos. The genotype enGal4/UASwg is embryonic lethal (the embryos do not hatch due to a deformed head) and we therefore collected all the dead embryos from this cross for cuticle analysis. Dead enGal4/UASwg embryos are easily distinguished from the occasional dead wild-type embryos by the fact that row 1 is missing from abdominal denticle belt A2–A7 (Figure 2, C vs. D). In this crossing scheme, half of the enGal4/UASwg embryos are heterozygous for a deficiency chromosome (or a wild-type or mutant chromosome), while the other half is wild type (Figure 2A).

We then scored the number of enGal4/UASwg embryos that showed patches of naked cuticle in the abdominal denticle belts A1–A7 (Figure 2, D' and E'), using strict criteria (see MATERIALS AND METHODS). We found a very low frequency of gaps when we tested wild-type chromosomes in our screen: yw⁶⁷ gave 1.9% of embryos with gaps (n = 106) and canton^S, 1.2% (n = 248) (Table 1). This gives an average of 1.4% of embryos with gaps (n = 354) for the wild-type control. By contrast, all hh alleles tested gave a high proportion of embryos exhibiting gaps in the denticle belts: 20% (n = 130) for hh⁵, 28.2% (n = 174) for hh^AC, and 36.5% (n = 138) for Df(3R)hh (Table 1). As expected, the hypomorphic allele, hh⁵, gives a lower proportion of embryos with the dominant phenotype compared to the stronger mutant hh^AC and the deficiency Df(3R)hh. Since only half of the enGal4/UASwg embryos scored with our scheme are actually heterozygous for the hh mutations tested (Figure 2A), these results mean that 60% of embryos of the enGal4/UASwg [hh^–/+] genotype exhibit a dominant phenotype.

Primary screen:

We then screened the collection of deficiencies on the third chromosome for those that showed the dominant phenotype at a frequency significantly higher than that of the wild-type control. We scored gaps in enGal4/UASwg embryos derived from males heterozygous for a given deficiency and used a ² test to assess the significance of our results (see MATERIALS AND METHODS). A total of 59 of the tested deficiencies did not show a frequency of embryos with gaps significantly different from the wild type (P > 0.05). For these deficiencies, the percentage of embryos exhibiting the dominant phenotype varies from 0 to 3.3% (APPENDIX). Seven deficiencies interacted weakly in the screen (0.001 < P <0.05), with percentages of embryos with gaps varying from 4.1 to 7.1% (APPENDIX). These are deficiencies Df(3L)h-i22, Df(3L)Scf-R6, Df(3L)brm11, Df(3L)VW3, Df(3L)rdgC-co2, Df(3R)ry615, and Df(3R)Dl-BX12, shown in red in Figure 3. Two deficiencies, Df(3L)ri-79c and Df(3R)awd-KRB, exhibited a frequency of embryos with gaps close to the frequency observed in enGal4/UASwg embryos derived from males hh^–/+ (34.2 and 36%, respectively; see APPENDIX). These deficiencies are shown in blue in Figure 3.

View larger version (42K): In this window In a new window Download PPT slide   FIGURE 3.— Map of the deficiencies on chromosome 3 tested in the primary screen. Deficiencies from the Bloomington deficiency kit are positioned on a drawing of polytene chromosome 3 (adapted from LINDSLEY and ZIMM 1992) according to cytological positions given in FlyBase (http://flybase.bio.indiana.edu/). Cytological positions are indicated under the drawing. 3L and 3R indicate left arm and right arm of chromosome 3, respectively. Deficiencies that did not interact are indicated in black. The deficiency that uncovered the hh locus is shown in green. Deficiencies that interacted strongly (P < 0.001) are shown in blue and those that interacted weakly (0.001 < P < 0.05) are shown in red.

 
For Df(3L)ri-79c and Df(3R)awd-KRB, we tested the possibility that the dominant phenotype was independent of the enGal4/UASwg background. Of the enGal4/UASwg embryos, 34.2% (n = 111) derived from Df(3L)ri-79c/+ males exhibited gaps in the denticle belts, mainly in the A4 segment. By crossing Df(3L)ri-79c/+ males with the yw⁶⁷ wild-type strain, we found that 25% (n = 60) of the progeny have the same dominant phenotype, thus showing that the dominant phenotype arises independently of the enGal4/UASwg-sensitized background. Since Df(3L)ri-79c removes the gene knirps (kni), we tested a null mutation in this gene, kni⁸, and found that 30% of enGal4/UASwg embryos derived from kni⁸/+ males have gaps in segment A4. This shows that the dominant phenotype detected for Df(3L)ri-79c is due to kni. The excess naked cuticle found in kni^–/+ embryos is unlikely to be a consequence of a change in Wg distribution, since kni is a gap gene required in the segmentation cascade early in embryogenesis (for a review, see ST JOHNSTON and NUSSLEIN-VOLHARD 1992). Rather, a perturbation of the transcription pattern of the segmentation genes in segment A4 is likely to cause the dominant phenotype. In support of this, a compression of the expression pattern of the pair-rule gene fushi-tarazu has been observed in segments A4 and A5 in kni^–/+ embryos (CARROLL and SCOTT 1986).

The same was found for Df(3R)awd-KRB: 36% (n = 136) of enGal4/UASwg embryos derived from Df(3R)awd-KRB /+ males exhibit gaps in the denticle belts. Curiously, the gaps are found in alternating segments (not shown) (DESBORDES 2004). This dominant phenotype is also found in Df(3R)awd-KRB /+ embryos, showing that it can occur independently of the enGal4/UASwg background. The excess cuticle found in Df(3R)awd-KRB/+ embryos had also been noted by COX et al. (2000), and the molecular basis for this phenotype is unknown. However, because the dominant phenotype is pair-rule, this again suggests an effect on early segmentation rather than on the regulation of Wg distribution.

We therefore decided to concentrate on the six regions defined by the deficiencies interacting weakly in the screen. Although the percentage of embryos with gaps found for these deficiencies is low (4.1–7.1%), the statistical test indicates that the interactions detected have a reasonable chance to be significant (0.001 < P < 0.05). Also, only half of the enGal4/UASwg embryos collected in a sample are actually heterozygous for the deficiency tested (the other half are of genotype enGal4/+; UASwg/+; see Figure 2A). This means that the true percentage of embryos of genotype enGal4/+; UASwg/deficiency showing the dominant phenotype is 8–14% for these deficiencies, which is significantly higher than the proportion of embryos with gaps found for the genotype enGal4/+; UASwg/+ (1.4%) (Table 1). We thus performed a secondary screen by testing smaller deficiencies and point mutations that map to these six regions. Table 2 summarizes the findings from the secondary screen, and the details about levels of interaction and cytological positions are given in the APPENDIX.

View this table: In this window In a new window   TABLE 2 Summary of the interactions detected in the secondary screen

 

Secondary screen:

Region 66:

From the primary screen, two interacting deficiencies overlapped in cytological location 66E: Df(3L)h-i22 (cytological breakpoints 66D10-11 and 66E1-2) and Df(3L)Scf-R6 (66E1-6 and 66F1-6). EnGal4/UASwg embryos derived from Df(3L)h-i22/+ or Df(3L)Scf-R6/+ males show a dominant gap phenotype with a similar frequency of 6.4% (n = 249) and 7.1% (n = 210), respectively. This suggests that the same gene is removed by both deficiencies and that the gene should be found in the region of overlap between the two deficiencies. Moreover, Df(3L)Scf-R11, a deficiency that overlaps with the proximal end of Df(3L)Scf-R6, did not interact, suggesting that the interacting gene is removed by the distal end of Df(3L)Scf-R6 (Table 2 and APPENDIX). We mapped genetically the overlap between Df(3L)h-i22 and Df(3L)Scf-R6 by complementation tests with sequenced P-element insertions in the region. Df(3L)hi22 failed to complement l(3)10631 (66D14-15), l(3)01629 (66DE1-2), and l(3)06464 (66E1-2) but complements l(3)10534 (66E2-3). Df(3L)Scf-R6 complemented l(3)10631, but failed to complement l(3)01629, l(3)06464, and l(3)10534. This suggests that the region of overlap between Df(3L)h-i22 and Df(3L)Scf-R6 falls between l(3)10631 and l(3)10534, which represents an interval of 215 kb on the genome sequence. This interval contains several genes, including dally, one of the two Drosophila homologs of glypicans (the other one is dally-like; see below). Dally was particularly interesting since it had been implicated in the regulation of Wg signaling (LIN and PERRIMON 1999; TSUDA et al. 1999). A RNA probe corresponding to exons 1–6 of dally cDNA was generated and hybridized to polytene chromosomes from Df(3L)h-i22/+ or Df(3L)Scf-R6/+ larvae. On wild-type chromosomes, the in situ hybridization signal maps at cytological location 66E1-3, which corresponds to the cytology given in FlyBase (data not shown) (DESBORDES 2004). No signal was observed on the deficient chromatid of Df(3L)h-i22/+ polytene chromosomes, demonstrating that most dally sequences are missing in this deficiency (data not shown) (DESBORDES 2004). This shows that Df(3L)h-i22 is an amorph for dally, in contrast to a previous report suggesting that Df(3L)h-i22 was a hypomorph for this gene (TSUDA et al. 1999). On Df(3L)Scf-R6/+ polytene chromosomes, the signal was clearly reduced on the deficient chromatid, suggesting that some of the sequences present between exons 1 and 6 of dally are missing in this deficiency (data not shown; see DESBORDES 2004). Df(3L)Scf-R6 was generated by X-ray irradiation of Scf1, which is an inversion (breakpoints 65A12-15; 66F3) (KOPP and DUNCAN 1997). This inversion is still present in Df(3L)Scf-R6 and since one of the breakpoints is very close to where dally maps, the dally gene might be rearranged in this chromosome. Also, the complementation tests (see above) show that dally^P2 (l(3)06464), as well as two P elements inserted distal and proximal to this gene [l(3)01629 and l(3)10534, respectively], do not complement Df(3L)Scf-R6. Together, these observations suggest that Df(3L)Scf-R6 is also a null for dally.

We tested three mutants in dally, dally¹⁰ (gift from Steve Kerridge), dally^E285 (MERABET et al. 2002), and dally^P2 (also named l(3)06464) (NAKATO et al. 1995). We found that two of these alleles, dally¹⁰ and dally^E285, gave a similar percentage of embryos with gaps as the overlapping deficiencies, 10.1% (n = 238) and 6.3% (n = 221), respectively. dally^P2 is a P-element insertion thought to be a hypomorphic allele of dally (TSUDA et al. 1999) and does not interact in the screen. Taken together, these results suggest that dally is responsible for the interaction detected in the screen. This was verified later by RNAi silencing of dally in enGal4/UASwg embryos (see below).

Region 72:

In the primary screen, we found one deficiency interacting in this region, Df(3L)brm11: 5.7% (n = 157). A smaller deficiency overlapping distally, Df(3L)A27, did not interact, whereas a deficiency overlapping proximally, Df(3L)th102, did, with a higher percentage of embryos with gaps: 18.6% (n = 156). This suggested that this deficiency removed not only the first locus found in the primary screen, but also a second interactor. Among other genes, we tested brahma (brm) and notum in the screen. The rationale for testing notum is that it codes for a hydrolase thought to modify the two Drosophila glypicans, Dally and Dlp (GERLITZ and BASLER 2002; GIRALDEZ et al. 2002). One allele of brm, brm², interacted at a similar level as Df(3L)brm11: 7.3% (n = 150). Two weaker alleles, brm^T362 and brm^T485, did not interact. This suggests that brm is responsible for the interactions detected with Df(3L)brm11. We tested two alleles of notum and found that both interact (not³, 17%, n = 112; not⁵, 8.3%, n = 96), suggesting that notum is the second interacting locus removed by Df(3L)th102.

Region 75:

Df(3L)VW3 interacted in the primary screen with 5.2% of embryos with gaps (n = 134). We tested a small deficiency on the distal side, Df(3L)fz2, and found a similar level of interaction: 8.0% (n = 209). This deficiency removes fz2, and we therefore tested the only mutant allele available for this gene, fz2^C1 (CHEN and STRUHL 1999). We did not detect an interaction with this allele. We also investigated pipe, a gene coding for a sulphotransferase, which is removed by Df(3L)VW3, but two presumed null mutant alleles, pip¹ and pip², did not interact.

Region 77:

In region 77, we had found two deficiencies in the primary screen, Df(3L)ri-79c and Df(3L)rdgC-co2. Df(3L)ri-79c interacted strongly (34.2%), but this was caused by a dominant phenotype due to loss of kni and was independent of the enGal4/UASwg background (see above). Df(3L)rdgC-co2 interacted weakly: 5.4% (n = 149). We tested one possible candidate gene removed by this deficiency, Presenilin (Psn), but did not find an interaction with the two tested alleles, Psn^B3 and Psn^C4. We found another deficiency in the region, Df(3L)XS-543, which showed a similar interaction to the primary screen deficiency: 7.6% (n = 144). However, another deficiency tested in the primary screen, Df(3L)XS533, which overlaps with a large part of both Df(3L)XS-543 and Df(3L)rdgC-co2, did not interact, thus confusing the mapping of the interacting locus. Precisely mapped deficiencies will need to be generated in this region to pursue the mapping further.

Region 87:

In region 87, we found one deficiency, Df(3R)ry615, that interacted very weakly in the primary screen (4.1%, n = 169). We then found three smaller deficiencies in the region that showed slightly higher levels of interaction: Df(3R)ry75: 8.3% (n = 120), Df(3R)urd; 8.8% (n = 113), and Df(3R)l26C: 9.2% (n = 129). A fourth deficiency in the area, Df(3R)lc4a, did not interact. Although the same level of interaction suggested a single locus interacting in region 87, the given cytological breakpoints for these four deficiencies did not allow us to isolate a small interacting genomic area, suggesting that either more than one locus interacts or the mapping of the breakpoints for these deficiencies is imprecise.

Region 92:

In region 92, Df(3R) Dl-BX12 was found to interact weakly in the primary screen: 6.0% (n = 116). Since this deficiency removed Delta, we tested one allele in the screen, Dl³, but did not find an interaction. The overlapping proximal and distal deficiencies Df(3R)Cha7 and Df(3R)H-b79 did not interact in the screen, suggesting that the interacting locus is present in the central region of Df(3R) Dl-BX12. We tested one smaller deficiency that removed this central region, Df(3R)Dl-A143, but did not find an interaction.

Further investigation of the role of dally, dally-like, and notum in Wingless distribution:

Of six regions selected after the primary screen, five (cytology 66, 72, 75, 77, and 87; see Table 2) were found to still interact in the secondary screen. This shows that although weak, the majority of the interactions detected in the primary screen were meaningful. For two of these regions (cytology 66 and 72), we identified three genes, dally, brm, and notum, for which at least one mutant allele interacted in our screen. Because brm is a transcriptional regulator, we assumed that it is more likely to regulate the expression of segmentation genes than the distribution of Wg (TAMKUN et al. 1992; COLLINS and TREISMAN 2000). We thus focused on examining the effect of notum and dally on Wg distribution in the embryonic epidermis. Since dally has a close homolog, dally-like, we also analyzed its role in this process.

Dally:

To verify the weak interaction detected with two alleles of dally, we silenced the gene by RNAi in the enGal4/UASwg background. We found that 44% (n = 43) of the enGal4/UASwg embryos injected with dally dsRNA exhibit gaps in the denticle belts, which confirms its interaction in the screen (compare Table 3 with Figure 4E). By contrast, none of the enGal4/UASwg embryos injected with buffer exhibited any gaps (Table 3), and they were indistinguishable from noninjected enGal4/UASwg embryos (Figure 4B). In the secondary and primary screen, deficiencies and mutants in dally interacted with percentages ranging from 6.3 to 10.1% (APPENDIX). Because only half of the embryos examined in the screen are actually mutant or deficient for dally, 12–20% of enGal4/UASwg embryos heterozygous for a dally mutation should exhibit patches of naked cuticle. Knocking down both zygotic and maternal dally mRNAs with RNAi in enGal4/UASwg embryos significantly increases this proportion (44%) as expected if dally is responsible for the interaction. However, we did not find any enGal4/UASwg[dally RNAi] embryos exhibiting the "full" phenotype found in enGal4/UASwg [hh^–/hh^–] embryos, where a symmetric expanse of naked cuticle is produced anterior as well as posterior to the Engrailed domain (Table 3 and Figure 4C) (SANSON et al. 1999). This is expected if dally is not the only factor that regulates Wg gradient formation in embryos.

View this table: In this window In a new window   TABLE 3 RNAi silencing of dally and dally-like in enGal4/UASwg embryos

 
The phenotype of excess naked cuticle seen in enGal4/UASwg[dally RNAi] embryos seems to occur independently of a major role for dally in segmentation. Indeed, dally RNAi in a wild-type background does not cause segmentation phenotypes at a frequency significantly different from buffer-injected controls (DESBORDES and SANSON 2003). This is exemplified by the control in the experiment presented here: half of the embryos injected were of the enGal4/TM3GFP genotype (and thus wild type) and were identified by GFP expression (the other half of the embryos were enGal4/UASwg). Both buffer-injected and dally dsRNA-injected enGal4/TM3GFP embryos showed a similar proportion of segmentation defects, 23 and 21%, respectively (Table 3). These segmentation defects are very weak, consisting of occasional fusions or deletions of denticle belts, and they are a consequence of the injection process (DESBORDES and SANSON 2003).

In conclusion, these experiments confirm that dally is the gene responsible for the interaction detected in the screen and suggest that its loss affects the distribution of Wg protein in wild-type embryos, without having a detectable effect on Wg signaling per se.

Dally-like:

At the time we did this work, no mutations were available in dlp, so we tested a potential interaction in our screen by RNAi silencing in the enGal4/UASwg background. We found that 54% (n = 61) of the enGal4/UASwg[dlp RNAi] embryos showed excess naked cuticle in the abdominal segments (whereas enGal4/UASwg embryos injected with buffer showed no phenotype) (Table 3). Of these 54%, 15% showed gaps in the denticle belts, which was the phenotype looked for in the screen and is similar to the phenotype seen in enGal4/UASwg [hh^–/+] embryos. Moreover, 39% showed the same phenotype as enGal4/UASwg [hh^–/hh^–] embryos, e.g., a symmetric expanse of naked cuticle anterior and posterior to the Engrailed domain (Table 3 and Figure 4, C vs. D). This suggested that the Wg gradient had become symmetric in these embryos. To demonstrate that the cuticle phenotype seen in enGal4/UASwg[dlp RNAi] embryos was due to a change in Wg protein distribution, we stained stage 12 embryos with a Wg antibody (Figure 5). While in enGal4/UASwg embryos, Wg protein is never detected posterior to the En domain (SANSON et al. 1999); in enGal4/UASwg[dlp RNAi] embryos, Wg is detected both anterior and posterior to the En domain (Figure 5, E–E'''). This symmetric distribution is identical to the distribution observed in enGal4/UASwg[hh^–/hh^–] embryos (SANSON et al. 1999). The symmetric distribution in enGal4/UASwg[dlp RNAi] can be explained in two ways: either dlp regulates Wg distribution directly, downstream of Hh signaling, or dlp regulates Hh signaling, and thus affects Wg distribution indirectly, through Hh. We found that the latter hypothesis is correct. In enGal4/UASwg[hh^–/hh^–] embryos, endogenous wg expression is lost, because Hh signaling is absent and the maintenance of wg transcription fails (Figure 5, B' vs. C'). If Dlp were regulating Wg distribution directly, without affecting Hh signaling, wg endogenous expression would be maintained in enGal4/UASwg[dlp RNAi] embryos. We found that this is not the case: wg endogenous expression disappears in enGal4/UASwg[dlp RNAi] embryos as in enGal4/UASwg[hh^–/hh^–] embryos (Figure 5, C and D), suggesting that Dlp is required for either Hh expression or signaling. We demonstrated elsewhere by additional experiments that Dlp is strictly required for Hh signaling in the embryonic epidermis (DESBORDES and SANSON 2003).

We also looked at the effect of dlp RNAi in a wild-type background. In contrast to dally RNAi, dlp RNAi causes a full segment polarity phenotype (DESBORDES and SANSON 2003). This is found as well in the experiment presented here: 82% (n = 71) of the control embryos, enGal4/TM3GFP, exhibit segmentation defects when injected with dlp dsRNA (Table 3). Contrary to what is seen in embryos injected with buffer or dally dsRNA, most of these segmentation defects are actually strong segment polarity phenotypes, identical to the phenotypes seen in wg or hh loss of function (not shown; see DESBORDES and SANSON 2003). On the basis of our study in DESBORDES and SANSON (2003), which shows that Dlp is required for Hh signaling, and the evidence from Figure 5, we conclude that the dramatic change of Wg distribution found in enGal4/UASwg[dlp RNAi] embryos is (at least in part) the indirect consequence of loss of Hh signaling.

It is worth noting that enGal4/UASwg[RNAi dlp] embryos do not have a phenotype strictly identical to enGal4/UASwg[hh^–/hh^–] embryos. Ectopic wg transcription in enGal4/UASwg[RNAi dlp] embryos is less robust than that in enGal4/UASwg[hh^–/hh^–] embryos (Figure 5, C and C' vs. D and D'). This suggests that en maintenance might be impaired in dlp RNAi embryos. Moreover, in enGal4/UASwg[RNAi dlp] embryos, 38% exhibit segmentation defects, whereas the rest exhibit excess naked cuticle (Table 3). Most of these segmentation defects are strong segment polarity phenotypes, suggesting that normal segmentation fails in these embryos, despite the presence of the artificial autoregulatory loop between en and wg. A possibility is that Dlp affects Wg signaling in addition to being required for Hh signaling (see DISCUSSION).

Notum:

The two tested alleles of notum, not³ and not⁵, interact unambiguously in our screen. The interaction with not³ is stronger than that with not⁵ (17.0 vs. 8.3%; see APPENDIX), which is consistent with the characterization of not³ and not⁵ as amorphic and hypomorphic, respectively (GIRALDEZ et al. 2002). We looked at the phenotype of these mutations in a wild-type background. not⁵ is homozygous viable so we were able to look at embryos mutant for the zygotic and maternal contribution of the gene (i.e., embryos derived from not⁵/not⁵ mothers). We found that 85% (n = 85) of the embryos have row 1 missing in most denticle belts and often have a breach of naked cuticle in segment A1. Ten percent of the embryos have stronger phenotypes, with patches of naked cuticle within denticle belts and sometimes complete replacement of denticle belts by naked cuticle (Figure 6A). Similar phenotypes, albeit weaker, were observed in not³/not³ embryos derived from heterozygous mothers (the not³ chromosome is homozygous embryonic lethal). The presence of a wild-type maternal contribution in not³/not³ embryos could account for the weaker phenotype. To test this, we examined embryos from mothers bearing germline clones homozygous for the not³ mutant chromosome. Germline clones bearing females were crossed with not³/TM3GFP males to generate embryos both maternally and zygotically mutant for notum. The GFP-expressing embryos, which have a wild-type copy of notum contributed paternally, hatched and did not show any obvious phenotype. The embryos that do not express GFP, and thus are maternally and zygotically mutants for notum, exhibit a range of severe defects. First, most of the mutants have an abnormal morphology. For example, out of a collection of 76 mutant embryos, 33 (43%) have a posterior end that is not completely retracted, 9 (12%) are U-shaped, and 26 (34%) are shaped like a ball (Figure 6, B–B''). These phenotypes suggest an early requirement of notum in patterning and/or morphogenesis, which we have not attempted to characterize here. We could not analyze reliably the ventral cuticle pattern of U-shaped or ball-shaped embryos, but we analyzed it for the 41 remaining embryos: all except one showed thinner denticle belts in the abdomen, and in addition, 16 embryos showed either gaps of naked cuticle in the belts or missing belts (Figure 6B''). Thus the majority of embryos for which the cuticle pattern can be analyzed showed ectopic naked cuticle in the abdominal segments. These phenotypes are similar, albeit much more severe, than those found for the hypomorphic not⁵ mutant.

To test if the phenotype of ectopic naked cuticle was due to a change in Wg distribution, we stained not³ mutant embryos with a Wg antibody. Females bearing not³ germline clones were crossed with males not³/TM3LacZ, using LacZ to distinguish the paternally rescued embryos from the not³ mutant embryos. Both paternally rescued and not³ mutant embryos were stained together to enable us to compare the intensity of staining for the Wg protein (Figure 6). Strikingly, at stages 12 and 13 Wg staining in notum mutant embryos appears weaker and more diffuse than that in paternally rescued embryos (Figure 6, C and C' vs. D and D'). In paternally rescued embryos, most of the staining is concentrated on a single row of cells immediately anterior to the Engrailed-expressing cells, as in wild-type embryos (Figure 6, C and C', and not shown; see also SANSON et al. 1999). By contrast, in not³ mutant embryos, Wg staining is weaker overall and appears to span two to three cell diameters rather than one (Figure 6, D and D'). This experiment indicates that either wingless is transcribed in a broader domain and less actively or the shape of the Wg gradient is altered in not³ mutant embryos.

A screen for genes required for Wingless gradient formation:

We found six regions interacting as deficiencies in our screen, and from these we isolated the genes brm, dally, and notum as interactors. Since we completed the screen, dally and notum have been found to affect Wg distribution, respectively, in cell culture (LUM et al. 2003; FRANCH-MARRO et al. 2005) and in the wing imaginal disc (GIRALDEZ et al. 2002), thus validating the ability of our screen to find genes regulating the Wg gradient.

We did not find any genes that gave a dominant phenotype as penetrant as hh in our sensitized background. The interactions detected are weak, which suggests that a combination of factors rather than a single one regulate the shape of the Wg gradient in embryos, downstream of Hh signaling. It also seems likely that most factors involved are maternally provided, as is the case for dally and notum. Our screen did not remove any maternal contribution since we tested heterozygous males for a given deficiency chromosome, due to the constraints in the reagents available at the time (see MATERIALS AND METHODS). Hence, the wild-type maternal contribution would weaken the interactions detected for genes contributed maternally and zygotically, such as dally and notum. This can also explain why not all the mutant alleles tested for dally interact: weak modifiers present elsewhere on the third chromosome of mutant alleles might be enough to mask an interaction already diminished by the maternal contribution.

However, although weak, most interactions detected in our primary screen were confirmed in the secondary screen and we believe that it can be applied to the whole genome to find additional factors regulating Wg gradient formation. In light of our results, two changes could be made to improve screen efficiency. First, females heterozygous for a given deficiency, instead of males, could be used to remove half of the maternal contribution in addition to halving the zygotic contribution (note that F₁ UASwg/+; Deficiency/+ females should be used rather than enGal4/+; Deficiency/+ females, since the progeny of the latter display an interfering phenotype; see MATERIALS AND METHODS). This should uncover maternal factors that might have been overlooked in our screen. Second, the use of precisely mapped deficiencies from the new Drosdel collection (RYDER et al. 2004), instead of the current deficiency collection, should improve dramatically the mapping of the interactors. Indeed, the current deficiency kit does not cover the whole genome and many deficiencies either are imprecisely mapped or represent complex aberrations [such as Df(3L)ScfR6], which renders analysis difficult. Also, the current deficiency kit contains some very large deficiencies, which might not interact because they remove modifiers. Another problem with large deficiencies is that some are haplo-insufficient and could not be tested in the screen because heterozygous males did not survive or were sterile.

Another use of the enGal4/UASwg-sensitized background is to test the role of candidate genes in Wg gradient formation. For example, deep orange (dor) and Chlathrin (Chc), two genes required for lysosomal degradation, have already been found to produce ectopic naked cuticle when their zygotic contribution is removed in the enGal4/UASwg background (DUBOIS et al. 2001). This contributed to the demonstration that Wg is degraded at a faster rate posterior to the En domain. It is worth noting that dor and Chc zygotic mutants do not exhibit any cuticle phenotype and thus do not reveal their role in Wg distribution without the sensitized background. Moreover, the maternal contribution of dor and Chc cannot be removed genetically in the maternal germline, since these genes are required for oogenesis. Thus the use of the enGal4/UASwg background provides an assay for genes that could not be tested otherwise in embryonic gradient formation.

Role of Dally-like in Wingless gradient formation:

RNAi silencing of dally or dlp in enGal4/UASwg embryos generated ectopic naked cuticle in both cases, but dlp silencing had a much stronger effect (Figure 4 and Table 3). This strong phenotype is explained at least in part by the strict requirement for Dlp in Hh signaling (DESBORDES and SANSON 2003) (Figure 5). In enGal4/UASwg [dlp RNAi] embryos, the failure of Hh signaling generates a symmetrical gradient of Wg protein and results in a phenotype identical to enGal4/UASwg [hh^–/hh^–] embryos (Figure 5) (SANSON et al. 1999).

Dlp could be regulating Wg distribution as well, but its requirement for Hedgehog signaling makes this difficult to assess. We noted, however, that a significant proportion of enGal4/UASwg [dlp RNAi] embryos shows a segment polarity phenotype (Table 3) and that the stripe of wg transcription in the en domain of these embryos is weak compared to the control enGal4/UASwg [hh^–/hh^–] embryos (Figure 5, C and C' vs. D and D'). The simplest way to explain these phenotypes is if en expression is somewhat compromised in enGal4/UASwg [dlp RNAi] embryos. Surprisingly, this effect would be specific to en, since naked cuticle can still form in the absence of dlp (Figure 4D) (DESBORDES and SANSON 2003; FRANCH-MARRO et al. 2005). It is possible that in contrast to naked cuticle specification, En maintenance requires high levels of Wg. In support of this idea, only the cells immediately adjacent to the Wg-expressing cells are able to maintain En expression at stage 10 (VINCENT and LAWRENCE 1994). Thus, it is possible that Dlp is required to maintain a high concentration of Wg at the surface of the Wg-expressing cells and that this contributes to the maintenance of En expression in adjoining cells. In the absence of Dlp, Wg would be released from the cells, thus compromising En maintenance without affecting naked cuticle specification (this might even promote naked cuticle specification over a longer distance). Consistent with this hypothesis, very recent results show that in the wing disc Dlp has opposite effects on short-range and long-range Wg signaling (KIRKPATRICK et al. 2004; KREUGER et al. 2004; FRANCH-MARRO et al. 2005).

Role of Dally in Wingless gradient formation:

The excess naked cuticle observed in enGal4/UASwg [dally RNAi] embryos could be due to a partial requirement in Hh signaling or, alternatively, to a direct effect on Wg distribution. Both hypotheses are plausible since Dally is required redundantly with Dlp for Hh signaling in wing discs (HAN et al. 2004a) and recent data indicate that Dally can affect Wg distribution (LUM et al. 2003; FRANCH-MARRO et al. 2005). In Drosophila cultured cells, Dally transfection causes accumulation of exogenous Wg at the cell surface (FRANCH-MARRO et al. 2005), whereas dally RNAi results in loss of Wg staining (see note 20 in LUM et al. 2003). Taken together, these observations suggest that Dally is able to retain Wg protein at the cell surface. It may be that, in the embryo, loss of dally releases Wg and allows it to signal at a longer range to make naked cuticle.

This effect of Dally seems to be independent of a requirement for Wg signaling per se, since embryos either depleted of dally mRNA following RNAi (this study and DESBORDES and SANSON 2003) or zygotically and maternally mutant for dally (FRANCH-MARRO et al. 2005) do not exhibit any segment polarity phenotype. However, Dally might cooperate with Dlp to modulate Wg signaling because embryos mutant for both dally and dlp have a stronger segment polarity phenotype than hh mutants (Dlp is strictly required for Hh signaling) and phenocopy the phenotype of wg hh double mutants (FRANCH-MARRO et al. 2005).

Role of Notum in Wingless gradient formation:

Halving the dose of notum in enGal4/UASwg embryos generates ectopic naked cuticle (APPENDIX). Moreover, notum mutants exhibit variable amounts of excess naked cuticle (Figure 6) (see also GIRALDEZ et al. 2002). Consistent with this effect, overexpression of Notum in embryos inhibits the formation of naked cuticle (GIRALDEZ et al. 2002). We found that in embryos maternally and zygotically mutant for notum, the Wg gradient appears broader than in wild-type embryos (Figure 6, D and D'). This broadening is seen clearly for the anterior Wg gradient, whereas it is less easy to observe posterior to the Wg source. This is expected if the fast degradation of Wg occurring posterior to the source (DUBOIS et al. 2001) is operating normally in notum mutants, thus masking changes in the distribution of Wg protein. We suggest that in notum mutants more Wg is available for signaling before it is targeted for degradation, resulting in the formation of ectopic naked cuticle posterior to the Wg source. A broadening of the Wg gradient has also been found for notum mutant clones in the wing disc (GIRALDEZ et al. 2002). In contrast to what has been observed in the wing disc, however, the broader Wg gradient in notum mutant embryos also appears weaker (Figure 6, D and D'). This does not seem to have an impact on signaling, however, since no segment polarity phenotypes were observed in notum mutant embryos, suggesting that engrailed maintenance is normal.

Notum has homologies to pectin acetylesterases and appears to modify the activity of glypicans, in particular Dlp (GERLITZ and BASLER 2002; GIRALDEZ et al. 2002). Recent work indicates that Notum directly or indirectly induces cleavage of the glycosylphosphatidylinositol anchor of Dlp (but not Dally), leading to shedding of this glypican from the cell surface (KREUGER et al. 2004). It has been hypothesized that Wg bound to Dlp could be made unavailable to signaling because of shedding from the cell surface, thus explaining the increase in Wg activity associated with loss of Notum. What is the substrate of Notum in the embryo? In the epidermis, both notum and dlp are upregulated in the cells that respond to Wg (KHARE and BAUMGARTNER 2000; GIRALDEZ et al. 2002). By contrast, dally is downregulated in the same cells, suggesting that Wg signaling might repress dally expression (TSUDA et al. 1999). This could reflect a different or even opposite role of the two glypicans in Wg gradient formation. Since in the wing disc Notum appears to act on Dlp and not Dally, the effects that we observe could be attributed to the regulation of Dlp by Notum. If this is true, this would reveal a role for Dlp in Wg gradient formation in the embryo, in addition to its role in Hh signaling. Interestingly, in addition to changes in the cuticle pattern, loss of notum causes severe defects in the patterning and/or morphogenesis of the early embryo (Figure 6), suggesting that Notum may have substrates other than the glypicans.

Deficiencies and mutations tested in the primary and secondary screens

Deficiency Cytology % enGal4/UASwg embryos with gaps Interaction Primary screen Df(3L)emc-E12 61A; 61D3  2.7 (n = 111) No Df(3L)Ar14-8 61C5-8; 62A8  1.6 (n = 126) No Df(3L)Aprt-1 62A10-B1; 62D2-5  1.7 (n = 275) No Df(3L)R-G7 62B9; 62E7  1.5 (n = 136) No Df(3L)HR119 63C2;63F7  3.3 (n = 121) No Df(3L)GN24 63F4-7;64C13-15 1.2 (n = 86) No Df(3L)ZN47 64C;65C 1.0 (n = 98) No Df(3L)XDI98 65A2;65E1  0.9 (n = 109) No Df(3L)pbl-X1 65F3;66B10  3.0 (n = 101) No Df(3L)66C-G28 66B8-9;66C9-10   0 (n = 123) No Df(3L)h-i22 66D1011;66E1-2  6.4 (n = 249) Yes (P < 0.01) Df(3L)Scf-R6 66E1-6;66F1-6  7.1 (n = 210) Yes (P < 0.01) Df(3L)Rdl-2 66F5;66F5  3.3 (n = 153) No Df(3L)29A6 66F5;67B1 2.2 (n = 89) No Df(3L)AC1 67A2;67D11-13   0 (n = 74) No Df(3L)lxd6 67E1-2;68C1-2   0 (n = 109) No Df(3L)vin2 67F2-3;68D6  2.7 (n = 113) No Df(3L)vin5 68A2-3;69A1-3   0 (n = 92) No Df(3L)vin7 68C8-11;69B4-5   0 (n = 92) No Df(3L)iro-2 69B1-5;69D1-6  0.8 (n = 124) No In(3LR)C190 69F3-4;70C3-4 + 89;89   0 (n = 83) No Df(3L)fz-GF3b 70C2;70D5 1.2 (n = 85) No Df(3L)fz-M21 70D2-3;71E4-5 2.1 (n = 94) No Df(3L)BK10 71C;71F 1.2 (n = 81) No Df(3L)brm11 71F1-4;72D1-10  5.7 (n = 157) Yes (P < 0.02) Df(3L)st-f13 72C1-D1;73A3-4  1.6 (n = 127) No Df(3L)81k19 73A3;74F   0 (n = 88) No Df(3L)W10 75A6-7;75C1-2  1.7 (n = 119) No Df(3L)Cat 75B8;75F1   0 (n = 127) No Df(3L)VW3 76A3;76B2  5.2 (n = 134) Yes (P < 0.02) Df(3L)kto2 76B1-2;76D5  2.4 (n = 83) No Df(3L)XS-533 76B4;77B  2.3 (n = 130) No Df(3L)rdgC-co2 77A1;77D1  5.4 (n = 149) Yes (P < 0.02) Df(3L)ri-79c 77B-C;77F-78A 34.2 (n = 111) Yes (P < 0.001) Df(3L)ME107 77F3;78C8-9  0.9 (n = 107) No Df(3L)Pc-2q 78C5-6;78E3-79A1  2.0 (n = 153) No Df(3L)Ten-m-AL29 79C1-3;79E3-8  2.9 (n = 104) No Df(3L)Delta1AK 79F;80A 2.1 (n = 96) No Df(3R)ME15 81F3-6;82F5-7  1.9 (n = 104) No Df(3R)6-7 82D3-8;82F3-6 1.2 (n = 81) No Df(3R)3-4 82F3-4;82F10-11 1.3 (n = 76) No Df(3R)Tpl10, 83C1-2;84B1-2   0 (n = 77) No Df(3R)Scr 84A1-2;84B1-2   0 (n = 78) No Df(3R)p712 84D4-6;85B6   0 (n = 145) No Df(3R)p-XT103 85A2;85C1-2   0 (n = 102) No Df(3R)by10 85D8-12;85E7-F1  0.8 (n = 122) No Df(3R)by62 85D11-14;85F6   0 (n = 91) No Df(3R)M-Kx1 86C1;87B1-5   0 (n = 79) No Df(3R)T-32 86E2-4; 87C6-7 1.2 (n = 82) No Df(3R)ry615 87B11-13;87E8-11  4.1 (n = 169) Yes (P < 0.05) Df(3R)ea 88E7-13;89A1  2.1 (n = 141) No Df(3R)P115 89B7-8;89E7-8;20  2.8 (n = 145) No Df(3R)DG2 89E1-F4;91B1-2  1.9 (n = 104) No Df(3R)Cha7 90F1-4;91F5 2.0 (n = 98) No Df(3R)Dl-BX12 91F1-2;92D3-6  6.0 (n = 116) Yes (P < 0.01) Df(3R)H-B79 92B3;92F13  2.7 (n = 113) No Df(3R)e-R1 93B6-7;93D2  2.9 (n = 103) No Df(3R)e-N19 93B2-13; 94A3-8  1.6 (n = 127) No Df(3R)mbc-30 95A5-7;95C10-11  2.0 (n = 101) No Df(3R)mbc-R1 95A5-7;95D6-11 2.5 (n = 79) No Df(3R)crb-F89-4 95D7-11;95F15   0 (n = 105) No Df(3R)crb87-5 95F7;96A17-18 1.1 (n = 88) No Df(3R)Tl-P 97A;98A1-2   0 (n = 95) No Df(3R)D605 97E3;98A5  0.9 (n = 115) No Df(3R)3450 98E3;99A6-8   0 (n = 109) No Df(3R)Dr-rv1 99A1-2;99B6-11  1.4 (n = 72) No Df(3R)awd-KRB 100D1;100D3-4 36.0 (n = 136) Yes (P < 0.001) Df(3R)faf-BP 100D;100F5  0.9 (n = 107) No Secondary screen Region 66  Df(3L)Scf-R11 66E3-4; 66F1-2  1.8 (n = 172) No  dally10 66E1-3 10.1 (P < 0238) Yes (P < 0.001)  dallyE385 66E1-3  6.3 (P < 0221) Yes (P < 0.05)  l(3)06464(dallyP2) 66E1-2 2.5 (n = 80) No  l(3)01629 66E1-2  1.1 (n = 347) No Region 72  Df(3L)A27 70E; 72A3-4 1.0 (n = 99) No  Df(3L)th102 72B1-72D12 18.6 (n = 156) Yes (P < 0.001)  brm2 72C1  7.3 (n = 150) Yes (P < 0.001)  brmT362 72C1  1.9 (n = 103) No  brmT485 72C1   0 (n = 139) No  arf72A 72C1  2.0 (n = 108) No  Not3 72D1 17.0 (n = 112) Yes (P < 0.001)  Not5 72D1 8.3 (n = 96) Yes (P < 0.001) Region 75  Df(3L)Dfz2 75F10-11;76A1-5  8.0 (n = 209) Yes (P < 0.001)  Dfz2C1 75F9-10   0 (n = 92) No  pip1 76A3   0 (n = 99) No  pip2 76A3  3.2 (n = 124) No Region 77  Df(3L)XS-543 76B; 77A  7.6 (n = 144) Yes (P < 0.001)  psnB3 77B9-C1  1.5 (n = 135) No  psnC4 77B9-C1  1.0 (n = 111) No Region 87  Df(3R)l26c 87E1; 87F11  9.2 (n = 129) Yes (P < 0.001)  Df(3R)ry75 87D2; 87D14  8.3 (n = 120) Yes (P < 0.001)  Df(3R)urd 87F1-15  8.8 (n = 113) Yes (P < 0.001)  Df(3R)lc4a 87E5-4; E11  1.7 (n = 113) No Region 92  Df(3R)Dl-A143 91F13; 92A2-3  3.3 (n = 151) No  Dl3 92A1-2  3.3 (n = 123) No

The cytology of each deficiency and gene is given according to FlyBase (http://flybase.bio.indiana.edu/). The percentage of enGal4/UASwg embryos presenting ectopic naked cuticle (gaps) is given for each chromosome tested. A ² was performed for each result, using the number of embryos with gaps found with wild-type chromosomes as the expected frequency (see MATERIALS AND METHODS). A mutant chromosome was considered interacting when the P-value was <0.05. Interacting deficiencies or mutants are underlined. n, the total number of embryos of genotype enGal4/UASwg examined.

We thank Clive McKimmie for his early contribution to the screen, Margit Pál for the in situ hybridizations on polytene chromosomes, Daniel St. Johnston and Jean-Paul Vincent for their comments on the manuscript, and Francois Balloux for advice on the statistics. We also thank K. Cadigan, S. Cohen, M. Fortini, D. Kalderon, S. Kerridge, D. St Johnston, G. Struhl, J. Treisman, the Hybridoma Bank, the Bloomington and Umea stock centers, and FlyBase for fly strains, reagents, and valuable information. This work was supported by a Career Development Award from the Wellcome Trust (054525/Z/98) to B.S. B.S. was also supported by the Cambridge Newton Trust and S.D. by the Cambridge European Trust.

¹ Present address: Laboratory of Stem Cell and Tumor Biology, Neurosurgery and Developmental Biology, Memorial Sloan Kettering Cancer Center, 1275 York Ave., New York, NY 10021.

BAEG, G. H., X. LIN, N. KHARE, S. BAUMGARTNER and N. PERRIMON, 2001 Heparan sulfate proteoglycans are critical for the organization of the extracellular distribution of Wingless. Development 128: 87–94.[Abstract]

BEJSOVEC, A., and E. WIESCHAUS, 1995 Signaling activities of the Drosophila wingless gene are separately mutable and appear to be transduced at the cell surface. Genetics 139: 309–320.[Abstract]

BELLAICHE, Y., I. THE and N. PERRIMON, 1998 Tout-velu is a Drosophila homologue of the putative tumour suppressor EXT-1 and is needed for Hh diffusion. Nature 394: 85–88.[CrossRef][Medline]

CADIGAN, K. M., M. P. FISH, E. J. RULIFSON and R. NUSSE, 1998 Wingless repression of Drosophila frizzled 2 expression shapes the Wingless morphogen gradient in the wing. Cell 93: 767–777.[CrossRef][Medline]

CARROLL, S. B., and M. P. SCOTT, 1986 Zygotically active genes that affect the spatial expression of the fushi tarazu segmentation gene during early Drosophila embryogenesis. Cell 45: 113–126.[CrossRef][Medline]

CHEN, C. M., and G. STRUHL, 1999 Wingless transduction by the Frizzled and Frizzled2 proteins of Drosophila. Development 126: 5441–5452.[Abstract]

CHEN, Y., and G. STRUHL, 1996 Dual roles for patched in sequestering and transducing Hedgehog. Cell 87: 553–563.[CrossRef][Medline]

COLLINS, R. T., and J. E. TREISMAN, 2000 Osa-containing Brahma chromatin remodeling complexes are required for the repression of wingless target genes. Genes Dev. 14: 3140–3152.[Abstract/Free Full Text]

COX, R. T., D. G. MCEWEN, D. L. MYSTER, R. J. DURONIO, J. LOUREIRO et al., 2000 A screen for mutations that suppress the phenotype of Drosophila armadillo, the ß-catenin homolog. Genetics 155: 1725–1740.[Abstract/Free Full Text]

DESBORDES, S., 2004 The role of the Drosophila glypicans in Wingless and Hedgehog signalling during embryogenesis. Ph.D. Thesis, University of Cambridge, Cambridge, UK.

DESBORDES, S. C., and B. SANSON, 2003 The glypican Dally-like is required for Hedgehog signalling in the embryonic epidermis of Drosophila. Development 130: 6245–6255.[Abstract/Free Full Text]

DOUGAN, S., and S. DINARDO, 1992 Drosophila wingless generates cell type diversity among engrailed expressing cells. Nature 360: 347–350.[CrossRef][Medline]

DUBOIS, L., M. LECOURTOIS, C. ALEXANDRE, E. HIRST and J. P. VINCENT, 2001 Regulated endocytic routing modulates wingless signaling in Drosophila embryos. Cell 105: 613–624.[CrossRef][Medline]

FRANCH-MARRO, X., O. MARCHAND, E. PIDDINI, S. RICARDO, C. ALEXANDRE et al., 2005 Glypicans shunt the Wingless signal between local signalling and further transport. Development 132: 659–666.[Abstract/Free Full Text]

FREEMAN, M., 2000 Feedback control of intercellular signalling in development. Nature 408: 313–319.[CrossRef][Medline]

GERLITZ, O., and K. BASLER, 2002 Wingful, an extracellular feedback inhibitor of Wingless. Genes Dev. 16: 1055–1059.[Abstract/Free Full Text]

GIRALDEZ, A. J., R. R. COPLEY and S. M. COHEN, 2002 HSPG modification by the secreted enzyme Notum shapes the Wingless morphogen gradient. Dev. Cell 2: 667–676.[CrossRef][Medline]

GONZALEZ, F., L. SWALES, A. BEJSOVEC, H. SKAER and A. MARTINEZ ARIAS, 1991 Secretion and movement of wingless protein in the epidermis of the Drosophila embryo. Mech. Dev. 35: 43–54.[CrossRef][Medline]

GONZALEZ-GAITAN, M., 2003 Endocytic trafficking during Drosophila development. Mech. Dev. 120: 1265–1282.[CrossRef][Medline]

GRECO, V., M. HANNUS and S. EATON, 2001 Argosomes: a potential vehicle for the spread of morphogens through epithelia. Cell 106: 633–645.[CrossRef][Medline]

HAN, C., T. Y. BELENKAYA, B. WANG and X. LIN, 2004a Drosophila glypicans control the cell-to-cell movement of Hedgehog by a dynamin-independent process. Development 131: 601–611.[Abstract/Free Full Text]

HAN, C., T. Y. BELENKAYA, M. KHODOUN, M. TAUCHI and X. LIN, 2004b Distinct and collaborative roles of Drosophila EXT family proteins in morphogen signalling and gradient formation. Development 131: 1563–1575.[Abstract/Free Full Text]

HATINI, V., and S. DINARDO, 2001 Divide and conquer: pattern formation in Drosophila embryonic epidermis. Trends Genet 17: 574–579.[CrossRef][Medline]

JOWETT, T., 1997 Tissue in Situ Hybridization: Methods in Animal Development. John Wiley & Sons, New York.

KHARE, N., and S. BAUMGARTNER, 2000 Dally-like protein, a new Drosophila glypican with expression overlapping with wingless. Mech. Dev. 99: 199–202.[CrossRef][Medline]

KIRKPATRICK, C. A., B. D. DIMITROFF, J. M. RAWSON and S. B. SELLECK, 2004 Spatial regulation of Wingless morphogen distribution and signaling by Dally-like protein. Dev. Cell 7: 513–523.[CrossRef][Medline]

KOPP, A., and I. DUNCAN, 1997 Control of cell fate and polarity in the adult abdominal segments of Drosophila by optomotor-blind. Development 124: 3715–3726.[Abstract]

KREUGER, J., L. PEREZ, A. J. GIRALDEZ and S. M. COHEN, 2004 Opposing activities of Dally-like glypican at high and low levels of Wingless morphogen activity. Dev. Cell 7: 503–512.[CrossRef][Medline]

LAWRENCE, P. A., B. SANSON and J. P. VINCENT, 1996 Compartments, wingless and engrailed: patterning the ventral epidermis of Drosophila embryos. Development 122: 4095–4103.[Abstract]

LECOURTOIS, M., C. ALEXANDRE, L. DUBOIS and J. P. VINCENT, 2001 Wingless capture by Frizzled and Frizzled2 in Drosophila embryos. Dev. Biol. 235: 467–475.[CrossRef][Medline]

LECUIT, T., and S. M. COHEN, 1998 Dpp receptor levels contribute to shaping the Dpp morphogen gradient in the Drosophila wing imaginal disc. Development 125: 4901–4907.[Abstract]

LIN, X., and N. PERRIMON, 1999 Dally cooperates with Drosophila Frizzled 2 to transduce Wingless signalling. Nature 400: 281–284.[CrossRef][Medline]

LINDSLEY, D. L., and G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press, New York.

LUM, L., S. YAO, B. MOZER, A. ROVESCALLI, D. VON KESSLER et al., 2003 Identification of Hedgehog pathway components by RNAi in Drosophila cultured cells. Science 299: 2039–2045.[Abstract/Free Full Text]

MERABET, S., F. CATALA, J. PRADEL and Y. GRABA, 2002 A green fluorescent protein reporter genetic screen that identifies modifiers of Hox gene function in the Drosophila embryo. Genetics 162: 189–202.[Abstract/Free Full Text]

MOLINE, M. M., C. SOUTHERN and A. BEJSOVEC, 1999 Directionality of wingless protein transport influences epidermal patterning in the Drosophila embryo. Development 126: 4375–4384.[Abstract]

NAKATO, H., T. A. FUTCH and S. B. SELLECK, 1995 The division abnormally delayed (dally) gene: a putative integral membrane proteoglycan required for cell division patterning during postembryonic development of the nervous system in Drosophila. Development 121: 3687–3702.[Abstract]

NEUMANN, C. J., and S. M. COHEN, 1997 Long-range action of Wingless organizes the dorsal-ventral axis of the Drosophila wing. Development 124: 871–880.[Abstract]

O'KEEFE, L., S. T. DOUGAN, L. GABAY, E. RAZ, B. Z. SHILO et al., 1997 Spitz and Wingless, emanating from distinct borders, cooperate to establish cell fate across the Engrailed domain in the Drosophila epidermis. Development 124: 4837–4845.[Abstract]

PAYRE, F., A. VINCENT and S. CARRENO, 1999 ovo/svb integrates Wingless and DER pathways to control epidermis differentiation. Nature 400: 271–275.[CrossRef][Medline]

PFEIFFER, S., C. ALEXANDRE, M. CALLEJA and J. P. VINCENT, 2000 The progeny of wingless-expressing cells deliver the signal at a distance in Drosophila embryos. Curr. Biol. 10: 321–324.[CrossRef][Medline]

PFEIFFER, S., S. RICARDO, J. B. MANNEVILLE, C. ALEXANDRE and J. P. VINCENT, 2002 Producing cells retain and recycle Wingless in Drosophila embryos. Curr. Biol. 12: 957–962.[CrossRef][Medline]

REICHSMAN, F., L. SMITH and S. CUMBERLEDGE, 1996 Glycosaminoglycans can modulate extracellular localization of the wingless protein and promote signal transduction. J. Cell Biol. 135: 819–827.[Abstract/Free Full Text]

RYDER, E., F. BLOWS, M. ASHBURNER, R. BAUTISTA-LLACER, D. COULSON et al., 2004 The DrosDel collection: a set of P-element insertions for generating custom chromosomal aberrations in Drosophila melanogaster. Genetics 167: 797–813.[Abstract/Free Full Text]

SANSON, B., 2001 Generating patterns from fields of cells. Examples from Drosophila segmentation. EMBO Rep. 2: 1083–1088.[CrossRef][Medline]

SANSON, B., C. ALEXANDRE, N. FASCETTI and J. P. VINCENT, 1999 Engrailed and hedgehog make the range of Wingless asymmetric in Drosophila embryos. Cell 98: 207–216.[CrossRef][Medline]

SETO, E. S., and H. J. BELLEN, 2004 The ins and outs of Wingless signaling. Trends Cell Biol. 14: 45–53.[CrossRef][Medline]

SIMMONDS, A. J., G. DOSSANTOS, I. LIVNE-BAR and H. M. KRAUSE, 2001 Apical localization of wingless transcripts is required for wingless signaling. Cell 105: 197–207.[CrossRef][Medline]

ST JOHNSTON, D., and C. NUSSLEIN-VOLHARD, 1992 The origin of pattern and polarity in the Drosophila embryo. Cell 68: 201–219.[CrossRef][Medline]

STRIGINI, M., and S. M. COHEN, 2000 Wingless gradient formation in the Drosophila wing. Curr. Biol. 10: 293–300.[CrossRef][Medline]

TAKEI, Y., Y. OZAWA, M. SATO, A. WATANABE and T. TABATA, 2003 Three Drosophila EXT genes shape morphogen gradients through synthesis of heparan sulfate proteoglycans. Development 131: 73–82.

TAMKUN, J. W., R. DEURING, M. P. SCOTT, M. KISSINGER, A. M. PATTATUCCI et al., 1992 brahma: a regulator of Drosophila homeotic genes structurally related to the yeast transcriptional activator SNF2/SWI2. Cell 68: 561–572.[CrossRef][Medline]

TANIMOTO, H., S. ITOH, P. TEN DIJKE and T. TABATA, 2000 Hedgehog creates a gradient of DPP activity in Drosophila wing imaginal discs. Mol. Cell 5: 59–71.[CrossRef][Medline]

THE, I., Y. BELLAICHE and N. PERRIMON, 1999 Hedgehog movement is regulated through tout velu-dependent synthesis of a heparan sulfate proteoglycan. Mol. Cell 4: 633–639.[CrossRef][Medline]

TSUDA, M., K. KAMIMURA, H. NAKATO, M. ARCHER, W. STAATZ et al., 1999 The cell-surface proteoglycan Dally regulates Wingless signalling in Drosophila. Nature 400: 276–280.[CrossRef][Medline]

VAN DEN HEUVEL, M., R. NUSSE, P. JOHNSTON and P. A. LAWRENCE, 1989 Distribution of the wingless gene product in Drosophila embryos: a protein involved in cell-cell communication. Cell 59: 739–749.[CrossRef][Medline]

VINCENT, J. P., and L. DUBOIS, 2002 Morphogen transport along epithelia, an integrated trafficking problem. Dev. Cell 3: 615–623.[CrossRef][Medline]

VINCENT, J. P., and P. A. LAWRENCE, 1994 Drosophila wingless sustains engrailed expression only in adjoining cells: evidence from mosaic embryos. Cell 77: 909–915.[CrossRef][Medline]

ZECCA, M., K. BASLER and G. STRUHL, 1996 Direct and long-range action of a wingless morphogen gradient. Cell 87: 833–844.[CrossRef][Medline]

Communicating editor: K. V. ANDERSON

International Access Link

Online ISSN: 1943-2361  Print ISSN: 0016-6731
Copyright 2009 by the Genetics Society of America
phone: 412-268-1812   fax: 412-268-1813   email: genetics-gsa{at}andrew.cmu.edu