Drosophila bunched integrates opposing DPP and EGF signals to set the operculum boundary

The Drosophila TGFβ family member DECAPENTAPLEGIC (DPP) (Padgett et al., 1987) is a potent inducer of developmental fate (Raftery and Sutherland, 1999). In some tissues, DPP functions as a morphogen, inducing multiple fates in a concentration-dependent fashion (reviewed in Neumann and Cohen, 1997). In other tissues, such as the longitudinal wing veins, DPP acts locally to specify a distinct cell fate with a sharp boundary (de Celis, 1997). In either situation, it is not known how the boundary to a DPP-induced fate is organized. Here we show that DPP patterns a domain of the ovarian follicle cells that will give rise to the operculum of the eggshell. We further show that the bunched (bun) gene (Dobens et al., 1997; Kania et al., 1995), also known as shortsighted (Treisman et al., 1995), is required both to limit the size of the operculum and to form the collar at the operculum boundary. Enhancer trap reporters in bunched are negatively regulated by DPP and positively regulated by EGF receptor (EGF-R), indicating that bunched integrates opposing DPP and EGF signals to set the boundary to a DPP-induced fate.

bunched encodes a protein similar to TSC-22, a TGFβ early response gene in mouse osteoblasts (Dobens et al., 1997; Shibanuma et al., 1992; Treisman et al., 1995). TSC-22 family members have a leucine zipper similar to bZIP transcription factors (Hamil and Hall, 1994; Shibanuma et al., 1992). In lieu of the basic domain of bZIP proteins, members of the TSC-22 gene family share a conserved sequence called the TSC domain (Dobens et al., 1997), which may function as a DNA-binding domain (Ohta et al., 1996; Seidel et al., 1997). Human TSC-22 can bind an essential region of the C-type natriuretic peptide gene promoter (Ohta et al., 1996), suggesting that TSC-22-like proteins directly regulate gene expression.

Initial studies implicated bunched in both DPP and EGF genetic pathways. Analysis of bunched function in eye development revealed genetic interactions with dpp, as well as with anterior open, which encodes a transcription factor regulated by receptor tyrosine kinases (Treisman et al., 1995). However, bunched mutant phenotypes in the eye (Treisman et al., 1995) and in the embryo (Kania et al., 1995; Perrimon et al., 1996) did not reveal specific roles for the gene. During oogenesis, bunched is a target gene for EGF signaling (Dobens et al., 1997). bunched shows patterned expression in the follicle cells (FC) (Dobens et al., 1997), which surround the developing oocyte and secrete the eggshell (reviewed by Spradling, 1993). bunched expression in the FC requires gurken (grk) (Dobens et al., 1997), a TGF-α-like gene that specifies the dorsal-ventral axis of the egg chamber (Neuman-Silberberg and Schüpbach, 1993; Schüpbach, 1987). Consistent with this, bunched is required in dorsal appendage formation (Dobens et al., 1997). The follicle cells are an excellent system to examine the role of bunched in interactions between DPP and EGF.

EGF signaling induces dorsal FC fates, which include the operculum and dorsal appendage structures of the eggshell (reviewed in Perrimon and Perkins, 1997; Ray and Schüpbach, 1996). The ligand GRK is secreted by the oocyte (Neuman-Silberberg and Schüpbach, 1993) and acts through the EGF-R in the FC (Price et al., 1989; Schejter and Shilo, 1989; Schüpbach, 1987). GRK patterns the dorsal-anterior FC during stage 10 of oogenesis, when these FC overlie the oocyte (Ray and Schüpbach, 1996; Spradling, 1993). Subsequently, additional EGF ligands are expressed or activated in the dorsal-anterior FC, and contribute to formation of the dorsal appendages (Sapir et al., 1998; Wasserman and Freeman, 1998). However, it is not known how the dorsal anterior domain is subdivided along the anterior-posterior axis to create the operculum and dorsal appendage fates.

Recently, it has been suggested that DPP patterns the anterior FC (Deng and Bownes, 1997; Twombly et al., 1996). The size and placement of the operculum and dorsal appendages are quite sensitive to altered levels of DPP signal (Twombly et al., 1996). DPP is expressed in anterior nurse cell-associated FC from stage 8 onward, although these cells do not contribute to the final eggshell. The operculum is formed by anterior dorsal columnar FC after they migrate centripetally to enclose the anterior end of the oocyte (Spradling, 1993).

Here we define an anterior domain of columnar FC that directly responds to DPP just before centripetal migration begins. Reduced bunched function causes an expansion of this domain and of the operculum. Thus, bunched is required to limit the operculum fate. We further demonstrate that DPP activates operculum gene expression indirectly by inactivating bunched. The pattern of bunched expression is due to repression by DPP and activation by EGF receptor; bunched expression forms a boundary at the centripetal migrating FC. We conclude that bunched expression is an integrated readout of both DPP and GRK signals, and establishes the boundary to a DPP-induced fate.

All genes are described in Flybase (The FlyBase Consortium, 1999). bun⁴²³⁰ and bun^r143 were previously described as strong and weak alleles of shs, respectively (Treisman et al., 1995). All bunched mutations are due to insertion of a P element carrying a lacZ reporter; only bun^r143 does not express β-galactosidase in the ovary. The enhancer trap line A359 (l(3)A359.1M3) has been described (Dobens and Raftery, 1998; Grossniklaus et al., 1989). Transgenic strains used were: UAS-λtop1 (Queenan et al., 1997); actin>CD2>GAL4 (Pignoni and Zipursky, 1997); Tub1>f⁺>dpp (Zecca et al., 1995); hs-FLP1 (Golic, 1991); hs-FLP2 (from Allan Spradling); Sb hs-FLP, hs-N-MYC and FRT40A (Xu and Rubin, 1993); pUmMad (Sekelsky et al., 1995); hs-GAL4 (from Norbert Perrimon); UAS-dpp (Staehling-Hampton and Hoffmann, 1994).

The UAS-BUN-1 and UAS-BUN-2 constructs were made by cloning the full-length BUN-1 and BUN-2 cDNAs into the EcoRI site of pUAST (Brand and Perrimon, 1993). BUN-X was generated by site-directed mutagenesis of the BUN-1 cDNA, using the Sculptor in vitro mutagenesis kit (Amersham) and the oligonucleotide GACCAGATCCATTGGCTGCACGATTTCGTTATCAATCGC. The amino acid sequence VAIDNKIEQAMD (aa 54-65) was changed to VAIDNEIVQPMD, confirmed by DNA sequencing. An EcoRI fragment containing the full-length mutant cDNA was cloned into pUAST. Transgenic lines were generated as described (Rubin and Spradling, 1982).

Homozygous clones were created using heat-shock-induced, FLP/FRT-mediated recombination and visualized by loss of the N-MYC epitope as described (Xu and Rubin, 1993). Mad¹²FRT40A, Mad¹²bun^r143FRT40A, or Mad¹²bun⁴²³⁰FRT40A was used as indicated. In both these and the flip-out clone experiments, clones were rarely recovered in the anteriormost columnar follicle cells. Females were heat-shocked for 1 hour at 37°C; ovaries were dissected 3 days later.

Clones of cells expressing DPP from the Tubα1 promoter were generated using the genotype hs-FLP; Tub>f⁺>dpp/pUmMad; A359. Clones of cells expressing GAL4 from the Actin5C promoter were generated in the genotypes: (1) hsFLP1/actin5C>CD2>GAL4; UAS-BUN-1/UAS-GFP A359.1M3, (2) UAS-BUN-X/actin5C>CD2> GAL4; hs-FLP2; UAS-GFP A359.1M3, (3) hsFLP1/actin5C>CD2>M GAL4; UAS-dpp/bun⁴²³⁰; UAS-GFP or (4) UAS-λtop1/actin5C> CD2>GAL4; bun⁴²³⁰; UAS-GFP/Sb hs-FLP. Females were heat shocked for 30-60 minutes at 37°C; ovaries were dissected 2 days later.

Dissected ovaries were fixed in 4% paraformaldehyde (Polysciences, Inc.) in PBS for 5-15 minutes. β-galactosidase activity was detected as described (Simon et al., 1985). For immunofluorescence, disrupted ovaries were washed in PSN (Xu and Rubin, 1993) and incubated overnight in mouse anti-MYC (1:200 dilution; Oncogene Science) and rabbit anti-β-galactosidase (1:10,000, pre-adsorbed to fixed tissue; Cappel). Secondary antibodies were from Jackson Labs. Images were collected using a Leica TCS-NT confocal microscope and figures were prepared using Photoshop.

Using the small bunched cDNA, a large cDNA was isolated from an ovarian library (Stroumbakis et al., 1994). This cDNA is the same size as the large cDNA isolated from eye disks (Treisman et al., 1995), suggesting that it encodes BUN-2. Antisense RNAs specific to each isoform were produced from subcloned regions: nucleotides 2960-3938 for the large cDNA and nucleotides 0-962 for the small cDNA. Whole-mount in situ hybridization was performed as described (Wasserman and Freeman, 1998).

DPP is expressed in the anterior follicle cells and perturbed DPP signaling leads to anterior eggshell defects (Twombly et al., 1996). In particular, two dorsal anterior structures, the operculum and the dorsal appendages (Fig. 1A), are sensitive to altered DPP levels (Twombly et al., 1996). Here, we illustrate the effects of a global increase in DPP levels, using heat-shock-GAL4 to drive UAS-dpp expression throughout the FC (Fig. 1B). The resultant eggs showed expanded opercula and reduced dorsal appendages (Fig. 1A,B). In other respects, the eggshells were normal. At the extreme anterior, normal micropyles were formed. The mutant opercula generally had a normal organization of large cell imprints surrounded by a raised structure, the collar (c, Fig. 1B). Significantly, expansion of the operculum always occurred over the dorsal side of the egg, indicating that dorsal-ventral patterning was unperturbed. A simple hypothesis, consistent with DPP’s role in axial patterning of other tissues, is that DPP patterns the anterior FC (Twombly et al., 1996).

Dorsal-ventral patterning of the follicle cells occurs during stage 10 of oogenesis, when the anterior columnar FC overlay the oocyte nucleus. Thus, we expected that DPP-dependent patterning of anterior FC occurred at about the same stage. We examined DPP expression using both in situ hybridization to dpp transcripts and immunofluorescence confocal microscopy to detect β-galactosidase from the dpp-lacZ enhancer trap, dpp10638. At the onset of stage 10, DPP is expressed in the nurse cell FC (NCFC, Fig. 1C-H). Slightly later, during stages 10B and 11, dpp transcripts accumulate in FC at the leading edge of centripetal migration (Fig. 1D,E). At the end of migration, a few dpp-expressing cells surround the micropyle at the anterior edge of the forming operculum (data not shown, and Twombly et al., 1996). The dpp enhancer trap faithfully reproduces this pattern of expression (Fig. 1F-H). These observations are consistent with a model in which a gradient of DPP defines the number of cells recruited to the operculum fate.

To obtain early markers for follicle cell fates specified by DPP, we sought lacZ enhancer trap insertions expressed adjacent to the DPP-expressing cells. l(3)A359.1M3 (Grossniklaus et al., 1989), hereafter called A359, is expressed in an anterior band of columnar FC (Fig. 2B-D), directly adjacent to, but not overlapping, the DPP-expressing cells (Fig. 2A,B and data not shown). Shortly after the onset of A359 expression, both dpp- and A359-expressing cells migrate centripetally to cover the anterior portion of the oocyte (Figs 1E,F and 2B,C, respectively). At later stages, high levels of A359 expression persist in the centripetal migrating FC while basal levels were detected in other FC groups (Fig. 1D).

To determine whether DPP signaling is necessary for high-level A359 expression, we generated clones of follicle cells unable to respond to DPP. Mad encodes a Smad protein essential for DPP signal transduction (reviewed in Raftery and Sutherland, 1999), so we generated clones homozygous for the null allele Mad¹² (Xu and Rubin, 1993). Numerous, small Mad¹² clones were recovered throughout the columnar FC, identified by the absence of the N-MYC marker. Five clones fell within the domain where A359 is normally expressed; in these clones, A359 was expressed only at basal levels (Fig. 2F). Thus, DPP signal transduction through MAD is necessary for the high level of A359 expression in anterior columnar FC.

If DPP patterns anterior FC fates, it should be sufficient to induce A359 expression. We found that ectopic A359 expression was induced by DPP overexpression in stage 10B or later egg chambers, when DPP was expressed transiently with a heat-shock promoter (Dobens and Raftery, 1998). Constitutive expression of DPP from the tubulin promoter in flip-out clones (Zecca et al., 1995) led to disruptions of the FC epithelium by stage 10B; in some cases, ectopic A359 expression was present (data not shown, and Fig. 2G).

To directly compare A359 expression with DPP responses, we examined nuclear MAD using a transgenic MYC-tagged MAD (MYC-MAD) construct that rescues Mad mutants (Sekelsky et al., 1995). Consistent with studies of other tissues (Newfeld et al., 1997), nuclear MYC-MAD was undetectable in wild-type egg chambers. However, heat-shock-induced DPP expression reliably caused detectable, but transient, nuclear localization of MYC-MAD at stage 10. In contrast, constitutive DPP expression from flip-out clones was rarely associated with detectable nuclear MYC-MAD by stage 10. One stage 10B egg chamber was recovered that showed ectopic MYC-MAD nuclear localization; all cells with MYC-MAD detectable in the nucleus showed ectopic expression of A359 (Fig. 2G). In cells that expressed low levels of A359, low levels of MYC-MAD were present in the nucleus (gray arrow, 2G); whereas cells with high levels of A359 had higher nuclear levels of MYC-MAD (white arrow, 2G). In sum, DPP signaling through MAD is both necessary and sufficient to induce A359 expression.

Previously, we reported that bunched is required in the dorsal anterior follicle cells for proper dorsal appendage formation (Dobens et al., 1997). During the course of clonal analyses, we noted that induction of bunched mutant clones was associated with increased fragility of the FC epithelium and the resultant eggshell (data not shown). Here we demonstrate that bunched is required in the columnar FC to repress the operculum fate.

To further investigate the requirements for bunched, we examined eggs laid by females homozygous for a weak bunched allele, bunr¹⁴³ (Treisman et al., 1995). These females are sterile, both due to low fecundity and to maternal effect lethality. The eggs had minor defects in the dorsal appendages (94%, n=96; Fig. 3B), consistent with previous results (Dobens et al., 1997). The most striking defects were altered opercula (85%; Fig. 3B), which were expanded posteriorly and ventrally. Every eggshell with an enlarged operculum also was missing the collar structure at the operculum edge. These phenotypes indicate that bunched limits the size of the operculum and is required to form the collar at the operculum boundary.

To determine the origin of the eggshell defects, we examined the expression of A359 in egg chambers from bun^r143 homozygous females. In wild-type females, A359 expression was absent at stage 10A (Fig. 3C). Normal A359 expression began during stage 10B, just prior to centripetal migration (Fig. 3E). In contrast, bun^r143 homozygous FC exhibited precocious A359 expression before stage 10A (Fig. 3D). By stage 10B, mutant egg chambers showed an expanded domain of strong A359 expression (Fig. 3F). Thus, decreased bunched function resulted in expansion of both the operculum of the eggshell and A359 enhancer trap expression in the FC. We conclude that bunched acts before or during stage 10 to repress operculum formation.

The sensitivity of the ventral operculum boundary to reduced bunched function was a surprise. We previously reported that bunched mRNA accumulates at high levels in dorsal anterior FC, with low levels throughout the egg chamber (Dobens et al., 1997). β-galactosidase activity staining revealed only the dorsal-anterior expression from several enhancer traps in the gene (Dobens et al., 1997). However, sensitive immunofluorescence staining for β-galactosidase protein indicated a low level of expression in most of the FC (Fig. 4B). The only populations that lacked detectable expression were the anterior band of columnar FC at stage 10, and the centripetal migrating FC at later stages. These populations include the cells that will make the operculum.

To more precisely map bunched expression, we examined the expression of each of the two bunched mRNAs. The previously reported expression pattern is a composite for both transcripts (Dobens et al., 1997). Alternative 5′ exons give rise to a small and a large transcript, which encode two protein isoforms (Treisman et al., 1995). Both isoforms contain the leucine zipper and TSC domain (highlighted in Fig. 5B) and the small isoform, BUN-1, is similar in size to vertebrate TSC-22. A short probe specific to each transcript was used for in situ hybridization to ovaries (Wasserman and Freeman, 1998). Each transcript was strongly expressed in the nurse cells from stages 8-12, consistent with the maternal effect phenotype (Perrimon et al., 1996).

Expression in the follicle cells was first detected at stage 8 (Fig. 4C). The BUN-1 transcript showed a pattern of expression similar to the pattern of the enhancer trap reporters (Fig. 4B versus D). At stage 10, expression was highest in dorsal FC, but also present in ventral FC (Fig. 4D). Strong nurse cell expression obscured the centripetal migrating FC at this stage. However, by stage 12, it could be seen that BUN-1 expression was strongly reduced in the operculum-forming FC (Fig. 4E,F). The BUN-2 transcript showed a more restricted pattern, with strong expression in the dorsal anterior FC beginning at stage 10, and expression in the operculum-forming FC at stage 12 (data not shown). The combined high level expression of both bunched transcripts in the dorsal anterior FC may explain why the operculum boundary is most sensitive to reduced bunched function on the ventral side. The BUN-1 expression pattern is consistent with a role in repressing the operculum fate.

Given that bunched opposes A359 expression and operculum formation, we wished to determine whether DPP indirectly induces the operculum fate by inactivating bunched. If DPP acts indirectly through bunched, then cells mutant for both Mad and bunched should express A359. Conversely, if DPP signaling directly induces A359 expression, then cells that lack both Mad and bunched should not express A359. For this experiment, we used the weak allele bun^r143, because it is the only allele that does not express lacZ in the FC. Anterior columnar FC homozygous for both Mad¹² and bun^r143 showed normal A359 expression (n=21; Fig. 5A). Thus, DPP induces A359 expression by inactivating bunched.

To determine whether BUNCHED is sufficient to block A359 expression in the anterior FC, we ectopically expressed each BUNCHED isoform. Random clones were induced using a flip-out GAL4 transgene (Pignoni and Zipursky, 1997) for the GAL4/UAS expression system (Brand and Perrimon, 1993). Ectopic expression of the large BUN-2 isoform had no effect on A359 expression (n=13; data not shown). In contrast, ectopic expression of BUN-1 was sufficient to block A359 expression in a cell autonomous manner (n=6; Fig. 5C) and led to disruptions in centripetal migration (data not shown). These data indicate that BUN-1 acts downstream of the DPP signal transducer MAD to repress A359 expression.

Overexpression of a mutant form of BUN-1 lacking the TSC domain (BUN-X) was not sufficient to block A359 expression (n=10; Fig. 5D). We are confident that BUN-X was expressed, because FC that expressed BUN-X were aberrant at later stages of oogenesis. Thus BUNCHED repression of A359 requires an intact TSC domain, consistent with the proposed function of this domain in DNA binding (Ohta et al., 1996; Seidel et al., 1997).

We have shown that genetic reduction of bunched function bypasses the requirement for MAD in anterior A359 expression and that BUN-1 transcript is not detected in the operculum-forming FC. A simple model to account for these data is that DPP promotes the operculum fate by preventing expression of BUN-1. To test this model, we asked whether bunched-lacZ is expressed in clones homozygous for Mad¹². In this experiment, Mad¹² homozygous clones were also homozygous for bun⁴²³⁰, the bunched-lacZ insertion mutation, because the Mad and bunched genes are on the same chromosome arm. In a separate experiment, FC homozygous for the bun⁴²³⁰ mutation alone showed normal bunched-lacZ expression levels (data not shown). In the posterior, Mad¹² clones showed normal bunched-lacZ expression (Fig. 6E). In contrast, mutant clones in the anterior columnar FC showed ectopic expression of bunched-lacZ (n=8; Fig. 6E). The DPP type I receptor TKV is also required to prevent bunched-lacZ expression in this domain (data not shown). Thus, DPP signaling is necessary to prevent bunched expression in the anterior follicle cells.

To determine if DPP is sufficient to prevent bunched-lacZ expression, we first examined the effect of random small clones of cells that constitutively express dpp. We used a flip-out GAL4 driver (Pignoni and Zipursky, 1997) so that cells expressing UAS-DPP were autonomously marked with UAS-GFP. Ectopic DPP was associated with ectopic repression of bunched-lacZ (n=15). However, we also observed persistent bun-lacZ expression in GFP-positive cells, even if these cells were adjacent to cells with ectopic repression of bun-lacZ (Fig. 6F). It seems likely that the availability of extracellular ligand is not simply determined by induction of gene expression.

To establish that repression of bunched-lacZ is a cell-autonomous response to DPP, we used the flip-out GAL4 system to create clones of cells that express constitutively active TKV receptor (TKV-A; Haerry et al., 1998). UAS-GFP-marked clones expressing TKV-A show cell autonomous repression of bunched-lacZ, regardless of size or position (n=15; Fig. 6G). Together, these data demonstrate that DPP signaling is necessary and sufficient to prevent bunched-lacZ expression in the anterior FC.

We previously reported that the peak of dorsal anterior bunched-lacZ expression is dependent on EGF signaling (Dobens et al., 1997). Another signal appears to maintain bunched-lacZ expression in the ventral posterior FC of late stage 10 egg chambers (LL Dobens and LA Raftery, unpublished data). We wished to determine whether EGF signaling could antagonize DPP signaling to induce ectopic bunched-lacZ expression. Clones of cells expressing constitutively active EGF receptor, λtop (Queenan et al., 1997), were induced using the flip-out GAL4 system. In the anterior FC, λtop expression was sufficient to induce cell-autonomous expression of bun⁴²³⁰ (n=5; Fig. 6H). Thus, elevated EGF-R signal can override the negative regulation by DPP signaling. We conclude that the sharp dorsal-anterior boundary for bunched-lacZ expression is set by the balance of positive signals through EGF-R and negative signals through TKV.

Members of the TGFβ superfamily play prominent roles in patterning both vertebrates and invertebrates. In particular, the bone morphogenetic proteins (BMPs), which include DPP, are required for the development of many tissues (reviewed in Hogan, 1996; Raftery and Sutherland, 1999). In some tissues, BMPs can function as morphogens, for they can induce multiple cell fates across a developmental field (reviewed in Neumann and Cohen, 1997; Whitman, 1998). In other tissues, BMPs appear to specify a distinct fate with a sharp boundary (e.g. in feather tract patterning in chickens, (Noramly and Morgan, 1998) and wing vein specification in flies (de Celis, 1997)). In vertebrates, BMPs often appear to oppose extracellular signals that activate receptor tyrosine kinases (RTKs) (Kretzschmer et al., 1997; Neubuser et al., 1997; Niswander and Martin, 1993). However, in Drosophila, both cooperative (e.g., Szüts et al., 1998) and antagonistic (e.g., Wappner et al., 1997) interactions have been observed. The follicle cells that surround the developing Drosophila oocyte provide a tractable system to examine interactions between BMP and RTK pathways. Here we provide evidence that bunched sets the boundary for the eggshell operculum, and demonstrate that the bunched expression pattern is set through negative regulation by DPP and positive regulation by EGF-R.

To understand how DPP patterns follicle cells, we sought genes that respond to DPP signaling. Enhancer trap reporters at two loci were identified. The enhancer trap A359 is expressed in a narrow stripe of follicle cells adjacent to, but not overlapping with, DPP-expressing cells. Enhancer trap insertions in bunched are specifically repressed in anterior columnar FC and are expressed in the remaining columnar FC. The cells that do not express bunched will subsequently express A359 and then undergo centripetal migration.

Both A359 expression and bunched-lacZ repression are directed by DPP. Clones of anterior columnar FC that lack the DPP signal transducer MAD fail to express A359 and ectopically express bunched-lacZ. Ectopic DPP can drive ectopic A359 and repress bunched-lacZ in the columnar FC. Ectopic DPP ligand can cause bunched-lacZ repression in nearby cells, whereas the activated DPP type I receptor, TKV-A, represses bunched-lacZ cell autonomously. Thus, DPP expression in the nurse cell FC directs a domain of gene expression in the adjacent columnar FC. This domain becomes the centripetal migrating FC, and subsequently gives rise to the operculum on the dorsal side. DPP expression continues in the leading edge of the centripetal migrating FC (LEFC), perhaps acting to maintain the identity of these cells.

Several lines of evidence indicate that bunched antagonizes operculum patterning. The operculum was expanded in eggs from females homozygous for a weak bunched allele. Consistent with this, egg chambers from these females had precocious and expanded expression of A359 in the follicle cells. BUN-1 expression was strongly reduced in the operculum-forming FC, and ectopic expression of BUN-1 was sufficient to block A359 expression in the anterior columnar FC. In addition, eggshells from bunched mutant females lacked the collar at the operculum boundary. These results suggest that the level of BUNCHED activity is critical to set the operculum boundary and to elaborate cell fates at this boundary.

Our previous work established that bunched is important for proper dorsal appendage formation (Dobens et al., 1997). Large dorsal anterior clones of bunched mutant FC result in dorsal appendages that are short and broadened with split ends. However, we found no evidence that bunched has an inductive activity in dorsal appendage formation. Rather, our data are consistent with a role for bunched as a repressor of the operculum fate. Perhaps ectopic activation of genes involved in operculum formation leads to aberrant dorsal appendage formation in bunched mutant clones.

Given that DPP promotes the operculum fate and bunched opposes it, we could genetically order their contributions. Anterior FC that are disabled for DPP signaling and defective in bunched function show normal levels of A359 expression. Thus, DPP indirectly promotes operculum formation by negative regulation of a negative regulator, BUNCHED.

Reduction of BUNCHED activity did not result in A359 expression throughout the columnar follicle cells (Figs 3D,F, 5A). This may be due to the use of a weak bunched allele for this analysis; low levels of BUNCHED may be sufficient to repress A359 in posterior FC. Alternatively, an additional signal may collaborate with DPP to direct A359 expression in the centripetal migrating FC. Isolation of null bunched alleles will resolve these possibilities.

A role for bunched as an antagonist of operculum patterning correlates well with the strict exclusion of bunched-lacZ from the centripetal migrating follicle cells (CMFC), the anteriormost population of columnar FC. The dorsal CMFC undergo changes in shape that lead to the unique cell imprints in the operculum. Expression of bunched-lacZ occurs throughout the posterior FC, strongest over the dorsal anterior, FC. This expression pattern reflects the pattern of BUN-1 mRNA accumulation, which encodes the isoform that can block A359 expression. We conclude from these data that the dorsal anterior boundary of BUN-1 mRNA expression defines the future boundary of the operculum.

Previously, we demonstrated that GRK signaling through the EGF-R is necessary for normal bunched-lacZ expression in the dorsal anterior FC (Dobens et al., 1997). Here we show that ectopic expression of activated EGF receptor is sufficient to induce ectopic bunched-lacZ in the centripetal migrating FC. Conversely, DPP signaling is both necessary and sufficient to repress bunched-lacZ in columnar FC. Thus the dorsal anterior boundary of bunched-lacZ expression is set by a balance of positive EGF and negative DPP signals. DPP also sets the anterior boundary for Broad-Complex expression; however, the regulation of this gene by EGF signaling is more complex (Deng and Bownes, 1997).

In summary, we propose a model (Fig. 7) where the boundary for the operculum is set by the boundary of BUNCHED activity, which is positioned by opposing activity of DPP and EGF signals in the dorsal FC. Dorsal anterior FC are exposed to high levels of EGF ligands, GRK (Neuman-Silberberg and Schüpbach, 1993), SPITZ (Sapir et al., 1998; Wasserman and Freeman, 1998) and VEIN (Wasserman and Freeman, 1998), and thus have elevated bunched expression. High anterior DPP signaling represses bunched expression. The close apposition of these signals in the dorsal anterior FC creates a sharp boundary of bunched expression. BUN-1 functions to repress A359 and define the boundary to centripetal migrating FC fates, including the operculum. Our data indicate that the ventral operculum boundary is also set by bunched; however, another signal appears to promote ventral bunched expression at late stages (LL Dobens and LA Raftery, unpublished data). The normal operculum border is defined by the eggshell collar. This structure is lost as bunched activity is lowered, suggesting that the boundary of bunched expression may serve to further organize cell fates at the operculum boundary.

Although our data suggest that EGF signals antagonize operculum patterning, EGF signaling is essential for operculum formation. First, grk and EGF-R mutant eggs have no opercula (Schüpbach, 1987). Second, overexpression of activated EGF-R can result in operculum expansion, although interpretation of the specific phenotype is not straightforward (Queenan et al., 1997). Thus, we expect that DPP does not prevent all EGF-R-induced events in the operculum-forming FC. It is likely that EGF signaling is active in cells that lack BUNCHED activity, and that DPP inactivation of BUNCHED modifies the response of these cells to EGF signals. In cultured mammalian cells, RTK signaling can directly antagonize BMP signaling by preventing nuclear accumulation of Smad protein, offering a possible molecular mechanism for these interactions (Kretzschmer et al., 1997).

bunched belongs to the TSC-22 family, a group of proteins that share both a leucine zipper and the adjacent, highly conserved TSC domain (Dobens et al., 1997; Treisman et al., 1995). TSC-22, the founding member of this family (Shibanuma et al., 1992), has a solution structure quite similar to that of bZIP transcription factors (Seidel et al., 1997). TSC-22 can bind to specific sequences in the promoter element from the C-type natriuretic peptide gene; these sequences drive TSC-22-dependent expression of a heterologous reporter (Ohta et al., 1996). Mutational analysis has shown that the TSC-domain is required for DNA binding and reporter gene regulation (Ohta et al., 1996). A recent report shows that TSC-22 can also act as a transcriptional repressor (Kester et al., 1999). Our data indicate that the A359 locus may be a direct target for negative regulation by BUNCHED. The sensitive A359 repression assay also shows that the TSC domain is critical to BUNCHED function: altering these amino acids makes BUN-1 inactive.

The role of bunched in other tissues is poorly understood. Embryos homozygous for bunched mutations die with morphological defects in the peripheral neurons (Kania et al., 1995) and subtle defects in cuticle pattern (L. L. D. and L. A. R., unpublished observations). bunched maternal effect phenotypes are pleiotropic, ranging from very early defects (J. S. P. and L. A. R., unpublished observations) to segmental defects in the embryonic cuticle (Perrimon et al., 1996). During eye development, bunched promotes photoreceptor differentiation, and shows genetic interactions with dpp, wingless, hedgehog and components of the EGF-R genetic pathway (Treisman et al., 1995). It remains to be determined whether bunched has a similar role throughout development, for example as an RTK target gene or a repressor of DPP target genes.

Here we show that bunched antagonizes DPP function in the follicle cells. This was surprising, for mutations in the dpp and bunched genes synergize to severely arrest eye development (Treisman et al., 1995). A mechanistic interpretation of this genetic interaction awaits better understanding of DPP functions during eye development. We note that the bunched eye phenotype is rescued by the BUN-2 transcript (Treisman et al., 1995), whereas BUN-1 is the antagonist of DPP in the FC. The BUN-2 transcript is expressed in the operculum-forming FC, raising the possibilities that this isoform has a distinct role, or that it is subject to post-transcriptional regulation. Further studies of the functions of the two isoforms will be needed to resolve these differences.

We propose here that the boundary to a DPP-induced fate in the follicle cells is set by transcriptional regulation of a downstream transcriptional repressor, BUNCHED. Recently, a similar role has been proposed for the gene brinker in setting threshold gene expression responses to DPP in the Drosophila wing (Campbell and Tomlinson, 1999; Jazwinska et al., 1999; Minami et al., 1999). Thus, DPP induction of gene expression through negative regulation of a negative regulator may be a common theme in development. Regulation of the expression of these key downstream repressors provides a powerful mechanism to modulate responses to DPP signaling.

We thank Yimin Ge for confocal assistance and Chuck Lang for fly food. We thank Vern Twombly, Bill Gelbart and Fotis Kafatos for discussions and encouragement at the beginning of this work, and Kristin White, Iswar Hariharan and anonymous reviewers for valuable comments. We are grateful to Doug Harrison, Norbert Perrimon, Gerald Rubin, Vern Twombly, Lawrence Zipursky and Trudi Schüpbach for fly strains. This work was supported by a CBRC Scholar’s Fellowship (L. L. D.) and grants from Shiseido Company of Japan, Limited to the CBRC (L. A. R.) and the American Cancer Society to L. A. R.