Two signalling pathways specify localised expression of the Broad-Complex in Drosophila eggshell patterning and morphogenesis

Pattern formation and morphogenesis are two key areas of developmental biology. The Drosophila follicle cells, which produce a polarised eggshell, provide a good model to study the relationship between pattern formation and morphogenesis. During oogenesis, the somatically derived follicle cells interact with the germline cells, and this is critical in the establishment of both the anterior/posterior (AP) and dorsal/ventral (DV) axes (Schüpbach, 1987; González-Reyes et al., 1995; Roth et al., 1995). The cell-cell interactions also cause the follicle cells to be divided into several subgroups, which have different morphology, function, and position along the two major axes. Additionally, during stages 9-12 of oogenesis (staging of oogenesis is based on Spradling, 1993), a series of follicle cell migrations occur. A group of 6-10 follicle cells at the anterior tip of egg chambers migrate through the nurse cell cluster to the oocyte border during stage 9. At the same stage, the majority of the follicle cells move towards the posterior of the egg chamber to form a columnar epithelium covering the oocyte, while the remaining follicle cells stretch to cover the nurse cell cluster. During stage 10b, the anterior columnar cells migrate centripetally between the oocyte/nurse cell border to cover the anterior end of the oocyte. Finally, two groups of columnar cells from the dorsal-anterior region migrate anteriorly to produce a pair of dorsal appendages (filaments). These are bilaterally located in a dorsal-anterior position in the mature egg, marking both the DV and AP polarity of the eggshell (reviewed by Spradling, 1993).

The DV polarity and dorsal appendages are induced by a signalling pathway initiated in the stage-8 oocyte. Gurken (Grk), a transforming growth factor-alpha (TGF-α) homologue, is the germline signal and its transcript is localised in the dorsal-anterior region of the oocyte during stages 8 to 10 of oogenesis (Neuman-Silberberg and Schüpbach, 1993). The Grk signal is thought to be received by Torpedo (Top), the Drosophila epidermal growth factor receptor homologue (EGF-R/DER), in the adjacent follicle cells (Price et al., 1989). The binding of Grk and Top/DER activates a receptor tyrosine kinase signalling pathway in these follicle cells, and induces them to adopt a dorsal fate (Neuman-Silberberg and Schüpbach, 1993; Roth and Schüpbach, 1994). These follicle cells are further divided into two different cell types: the dorsal midline cells and the appendage-producing lateral-dorsal cells (Spradling, 1993).

Many genes have been found to be involved in the Grk-DER signalling pathway in the differentiation of follicle cell fate along the DV axis. cappuccino, spire, fs(1)K10, cornichon, orb and squid are required for the anterior-dorsal localisation of grk transcripts in the oocyte (Manseau and Schüpbach, 1989; Neuman-Silberberg and Schüpbach, 1993; Christenson and McKearin, 1994; Roth and Schüpbach, 1994; Roth et al., 1995). Drosophila homologues of Ras1, Raf and MEK have been shown to transmit the signal in the follicle cells and are required for dorsal follicle cell fate determination in oogenesis (Schnorr and Berg, 1996; Brand and Perrimon, 1994; Lu et al., 1994). In addition, Rhomboid (Rho), a transmembrane protein, is expressed in a dynamic pattern in the dorsal-anterior follicle cells and probably acts to intensify Grk-DER signalling (Ruohola-Baker et al., 1993). Moreover, two effectors of the Grk-DER signalling pathway have been identified. Pointed (Pnt), an ETS domain transcription factor, is a downstream target of the EGF receptor signalling pathway and is expressed in the dorsal midline follicle cells during oogenesis (Morimoto et al., 1996). CF2, a zinc-finger transcription factor, has been found to be expressed in the ventral follicle cells and is thought to be a suppresser of Grk-DER signalling on the ventral side (Hsu et al., 1996). Both CF2 and Pnt are transcription factors expressed in subsets of follicle cells along the DV axis and are thought to define the cell fate in these cells. However, the transcription factor directly involved in dorsal appendage formation has not been identified.

The patterning of the eggshell along the AP axis requires a transforming growth factor-beta (TGF-β) family member, Decapentaplegic (Dpp) (Padgett et al., 1987; Twombly et al., 1996). Dpp is a morphogen and can produce a long range signal to organise the AP axis in wing imaginal discs (Lecuit et al., 1996). It is expressed in both the centripetal cells and nurse cell associated follicle cells at stage 10, and is required for the formation of the anterior eggshell. Decreases or increases in the level of Dpp expression during oogenesis shift the AP position of the dorsal appendages (Twombly et al., 1996).

Since both the Dpp and Grk-DER signalling pathways are required for the patterning of the eggshell, how do they cooperate? Is there an effector gene regulated by both signalling pathways? Here we show that the Broad-Complex (BR-C) is downstream of both these signalling pathways and is an intermediate in eggshell patterning.

The BR-C encodes a family of C2H2 zinc-finger proteins (Z1, Z2, Z3, and Z4) which share a common amino terminus (the BR-C ‘core’) domain but differ in zinc-finger DNA binding domains (DiBello et al., 1991; Bayer et al., 1996). The core contains a highly conserved amino-terminal motif, called the BTB or POZ domain, which appears to be involved in protein-protein interaction and is widely distributed among metazoans (DiBello et al., 1991; Bardwell and Treisman, 1994; Zollman et al., 1994). The core is alternatively spliced to link with one of the four zinc-finger domains, generating four classes of proteins, the Z1, Z2, Z3 and Z4 isoforms (DiBello et al., 1991; Bayer et al., 1996).

Genetically, the BR-C locus has three fully complementing functions: br (broad), rbp (reduced bristle number on palpus) and 2Bc, as well as a non-complementing npr (nonpupariating) class (Belyaeva et al., 1980). Additionally, a number of BR-C alleles have been categorised to the 2Bab complementation group. These alleles do not complement br or rbp, but do complement 2Bc mutations (Belyaeva et al., 1980). The npr class is thought to be the null mutation, because alleles in this class fail to complement mutations in each of the three-complementing groups. They are also phenotypically indistinguishable from deletions of the locus.

During metamorphosis, the BR-C belongs to the ‘early genes’ that respond to the steroid hormone, 20-hydroxyecdysone (ecdysone), and co- ordinate the ecdysone response among tissues by regulating the expression of effector genes (Emery et al, 1994; Talbot et al., 1993). It is required for a number of morphogenetic activities during metamorphosis, including wing and leg imaginal disc elongation and eversion (Kiss et al., 1988; Emery et al., 1994).

In the studies presented here we report that the transcription factor BR-C is expressed in the lateral-dorsal follicle cells and is required for dorsal appendage formation. Its expression is induced by the Grk-DER signalling pathway in a dose-dependent manner, and further regulated by the Dpp signalling pathway along the AP axis. The study of the generation and regulation of the bilaterally symmetrical expression pattern of a morphogenetic gene, namely the BR-C, may help us to understand the general mechanisms underlying pattern formation and morphogenesis.

The following Drosophila melanogaster strains were used: Oregon R, fs(1)K10¹ (Wieschaus et al., 1978); top^QY1(Schüpbach, 1987); dpp^hr56, dpp^e87 (Twombly, 1996); Gal4-C532 (Deng et al., 1997); br¹, npr⁶ (Kiss et al., 1988); A47 (Deng and Bownes, unpublished data); UAS-lacZ, UAS-Draf^gof (Brand and Perrimon, 1993, 1994); UAS-dpp (Staehling-Hampton and Hoffman, 1994); UAS-pntP1, UAS-pntP2 (Klaes et al., 1994). All stocks were raised on standard corn meal food at 18°C.

The hybridisation probes used were a 0.5-kb BR-C core domain cDNA fragment from plasmid BSCD5; and Z1, Z2, Z3 and Z4 zinc finger domain fragments from plasmids containing PCR products for each of them respectively (kindly provided by C. Bayer). The probes were labelled with digoxigenin as described by the supplier (Boehringer Mannheim Biochemicals). The temperature sensitive flies incubated at 18°C were shifted to 29°C after eclosion and were fed with yeast for 2 days before dissection. Other flies were incubated at 25°C and yeasted for 2 days before dissection.

The staining and detection procedure have been described (Tautz and Pfeifle, 1989) and modified as follows. Ovaries were dissected in PBS and fixed in 4% paraformaldehyde (Sigma) in PBS for 20 minutes, then washed in PTw (PBS containing 0.1% Tween-20) for 3× 5 minutes to remove the fix. After washing in methanol/EGTA (0.5 M, pH 8) (9:1) for 3× 5 minutes, the ovaries can be stored in methanol at −20°C for several months. The stored ovaries were washed in methanol/EGTA once followed by washing in PTw 3 times. They were then incubated in proteinase K (100 mg/ml) (Sigma) for 60 minutes at room temperature, followed by washing for 5 minutes in PTw containing 2 mg/ml glycine to stop the proteinase K reaction. The post-fixation, hybridisation, washing and detection procedures are similar to those of Tautz and Pfeifle (1989). Anti-DIG-Ap-conjugate was pre-absorbed with postfixed wild-type (Oregon R) ovaries at 4°C for overnight. For examination, ovaries were mounted in a mixture of PBS:glycerol (1:4).

The female progeny of the cross of C532-Gal4 with UAS-lacZ were dissected in PBS and stained at room temperature (25°C) overnight in PBS containing 0.2% X-gal (from Biosynth AG), 5 mM K₄[Fe(III)CN₆], 5 mM K₃[Fe(II)CN₆], 0.3% Triton X-100, rinsed in PBS and mounted in PBS 80% glycerol for microscopic analysis.

C532-Gal4 flies were crossed with UAS-dpp; UAS-pntP1; UAS-pntP2, UAS-Δ Draf respectively. The ovaries of the female progeny of these crosses were dissected in PBS to score the eggshell phenotype and to examine the BR-C expression after fixation and treatment as described in the section on ‘RNA in situ hybridisation’.

We have searched for target genes from P[Gal4] enhancer-trap lines which show reporter gene expression in subsets of follicle cells (Deng et al., 1997). The BR-C was identified as the target gene in one of these lines, which shows reporter gene expression in two patches of columnar cells at stage 10 of oogenesis (Deng and Bownes, unpublished data).

The four BR-C isoforms share a 5′ common core domain but differ in zinc-finger domains (Z1, Z2, Z3, and Z4) (DiBello et al., 1991, Bayer et al., 1996). Using digoxigenin-labelled probes from the core domain and different zinc-finger domains to detect BR-C expression, we find that Z1 is the only zinc-finger isoform which is expressed at levels detectable by whole-mount in situ hybridisation during oogenesis. Its expression pattern mirrors that of the core domain (hereafter called BR-C expression pattern). Expression of the BR-C is first detected in all follicle cells at stage 6 of oogenesis (Fig. 1A). At stage 10a, it is expressed in all columnar cells but absent in the dorsal-anterior most region, facing the oocyte nucleus and the localised Grk signal (Fig. 1B,C). The expression is stronger in the follicle cells close to the anterior-dorsal region of the oocyte, and becomes gradually weaker posteriorly and ventrally, diminishing at the posterior pole (Fig. 1B). At stages 10b and 11, BR-C expression disappears entirely in the ventral and posterior follicle cells, and remains in just two groups of dorsal-anterior follicle cells which lie symmetrically at either side of the dorsal midline (hereafter called the BR-C late expression pattern) (Fig. 1D,E). These follicle cells, which are thought to be the progenitor cells of the dorsal appendages, include about 55-65 cells in each group and are 2 cells away from the dorsal midline and 2-3 cells from the centripetal cells (Fig. 1D,E).

We asked if the late expression of BR-C in the lateral-dorsal follicle cells reflected a function in dorsal appendage formation. In order to answer this, a newly generated BR-C mutant, A47, which has been mapped to the 2Bab complementation group (Deng and Bownes, unpublished data), was crossed with a weak mutant, br¹. The female progeny, br¹/A47, survived until adulthood and produced eggs with reduced dorsal appendages (Fig. 2B), when compared to the long appendages in the wild-type egg (Fig. 2A). Normal embryos developed in these eggs and hatched as first instar larvae. Since progeny from the cross of the BR-C null allele, npr¹, with br¹ can also survive until adulthood (Kiss et al., 1988), the eggs produced by br¹/npr⁶ females were also examined. These eggs exhibited a similar phenotype to the eggs produced by the br¹/A47 mothers, and had reduced dorsal appendages (data not shown). These observations, along with the report that another BR-C mutant, br^de12/br^de12, produces ‘appendageless’ eggs (Huang et al., 1992), indicate that BR-C function is required for dorsal appendage morphogenesis during oogenesis.

Since the BR-C is expressed in an asymmetrical pattern along the DV axis, we asked whether its expression in the lateral-dorsal follicle cells is a consequence of Grk-DER signalling, which directs DV pattern formation during oogenesis. This has been investigated by analysing the BR-C expression pattern in mutants that affect this signalling pathway. fs(1)K10 is required for the localisation of grk RNA to the dorsal-anterior corner of the oocyte at stage 8 (Cheung et al., 1992). In the fs(1)K10¹ mutant egg chambers, grk RNA is mis-localised to form a ring in the anterior oocyte (Roth and Schüpbach, 1994; and data not shown). We observed that this induces a BR-C late expression pattern which is expanded to the ventral follicle cells surrounding the oocyte. However, the lack of dorsal BR-C expression in the follicle cells still exists in this mutant (Fig. 3C). Since the BR-C is required for dorsal appendage formation, it can be deduced that the ventrally expanded domain of BR-C expression is the reason for the expansion of the dorsal appendages to the ventral region of the fs(1)K10¹ egg (Fig. 3D). Conversely, in strong grk⁻mutants, in which the Grk signal is absent, we cannot detect the localised BR-C expression pattern in the lateral-dorsal-anterior follicle cells (data not shown), and no dorsal appendages are generated in these mutants. Nevertheless, early expression of the BR-C in all follicle cells during stage 6 seems to be unaffected in the strong grk⁻ mutant (data not shown). Taken together, these data indicate that the BR-C late expression is downstream of the Grk-DER signalling pathway in DV polarity formation during oogenesis.

We next investigated why BR-C expression is absent in the dorsal most follicle cells in which the highest dose of Grk signal is supposed to be received (the fact that these cells receive a higher level of Grk signal is shown in Fig. 4A,B). If the BR-C responds to the Grk-DER signalling pathway in a dose dependent manner, increases or decreases in the signal should shift the BR-C late expression domain along the DV axis. In order to test this, the BR-C expression pattern was examined in transgenic flies with four extra copies of grk gene in the genome (Neuman-Silberberg and Schüpbach, 1994). It was found that the dorsal gap between the two groups of BRC expressing cells is enlarged to about 8 cells wide (Fig. 3E), in comparison with the 4-cell-wide gap in the wild type (Fig. 3A). Consequently, the dorsal gap between the two dorsal appendages is widened in the mature egg (Fig. 3F). The expansion of the dorsal gap is apparently due to the shift of the BR-C expressing cells ventrally, since the number of cells in each group is approximately the same as in the wild type (data not shown). In contrast, a decrease in Grk-DER signalling in a top^QY1 mutant (Schüpbach, 1987) results in BR-C expression in the dorsal most follicle cells (Fig. 3G). The two groups of BR-C expressing cells shift dorsally and fuse to form one group along the dorsal midline (Fig. 3G). This fusion leads to the fusion of the dorsal appendages in the dorsal most region (Fig. 3H). Taken together, these data indicate that the absence of BRC expression in the dorsal most region is due to the high dose of Grk-DER signalling. They suggest that if the signal is higher than a certain level, it represses rather than activates BR-C expression. In the fs(1)K10₁ egg chambers, the Grk signal is stronger in the dorsal corner than in the rest of the anterior ring (Roth and Schüpbach, 1994; and data not shown), and this could explain why BR-C expression is still absent in the dorsal most follicle cells when it expands ventrally (Fig. 3C). In order to further test this dose-dependency on the Grk-DER signalling pathway, BR-C expression and the eggshell phenotype were examined in egg chambers that have increased levels of a constitutively activated Draf, a serine/threonine kinase downstream of Grk and DER in the signalling pathway (Brand and Perrimon, 1994). We found that when Draf is highly expressed in most of the follicle cells at stages 9 and 10 (a Gal4 driver C532, crossed with a UAS linked gain-of-function Draf gene, UAS-Draf^gof; Brand and Perrimon, 1994; Deng et al., 1997), about 90% of the mature eggs have no dorsal appendages, and the BR-C late expression pattern could not be detected in the majority of the stage 10b-11 egg chambers (data not shown).

We next investigated what the link is between the high dose Grk-DER signalling and the repression of BR-C expression in the dorsal most follicle cells. An ideal candidate is pointed (pnt), an ETS domain transcription factor, which acts downstream of the Ras signalling pathway in a number of developmental processes (Kl ämbt, 1993; Brunner et al., 1994). During stages 9-10a of oogenesis, both isoforms of pnt (pntP1 and pntP2) are expressed in the dorsal-anterior most columnar cells in response to the Grk-DER signalling pathway (Morimoto et al., 1996). Ectopic pnt expression in all anterior columnar cells resulted in eggs with reduced dorsal appendages (Morimoto et al., 1996). In order to determine if pnt is the repressor of BRC expression, we examined BR-C expression in egg chambers with ectopic pntP2 in most of the follicle cells at stage 9-10 (Fig. 4) (both UAS-pntP1 and UAS-pntP2 were crossed with Gal4-C532, but only C532/UAS-pntP2 survived until adulthood). The reporter gene expression pattern of Gal4C532, crossed to a UAS-lacZ line, is shown in Fig. 4C. There are about 10-15 follicle cells expressing BR-C at both sides of the dorsal midline at stage 10b (Fig. 3I), 4-5 times less than in the wild type. The dorsal appendages produced by these flies are thinner and shorter than those produced by the wild type (Fig. 3J). These observations indicate that pnt is a mediator of high dose Grk-DER signalling in the dorsal-anterior-most follicle cells, acting to repress the expression of BR-C. The remaining BR-C expression in a small number of follicle cells is probably due to the fact that the Gal4 reporter gene is not expressed uniformly in Gal4-C532. As shown in Fig 4C some follicle cells fail to show β-galactosidase staining while neighbouring cells are stained.

An hypothesis for why BR-C expression is also missing in the anterior most columnar cells is that there is probably another repressor produced in these cells. A good candidate is a downstream gene of the Dpp signalling pathway (Padgett et al., 1987), which is required for anterior eggshell patterning (Twombly et al., 1996). If Dpp signalling acts as a repressor of BR-C expression in oogenesis, a decrease in Dpp levels should result in expression of the BR-C in the anterior most columnar cells, where it is absent in the wild-type eggshell. In order to determine if this is the case, a temperature sensitive mutant, dpp^hr56/dpp^e87, was used to examine the BR-C expression pattern. The mutants were raised at 18°C and then shifted to restrictive conditions after the flies eclosed. As predicted, the anterior most columnar cells express the BR-C when the level of DPP expression is decreased in stage-10b mutant egg chambers (Fig. 5A). Since the anterior frontier of the BR-C late expression pattern in the wild type is about 2-3 cells distant from the Dpp source in the centripetal cells, Dpp could act as a morphogen here. Conversely, when Dpp is ectopically expressed in most of the follicle cells at stage 9 (Gal4-C532 crossed with a UAS-dpp) (Fig. 4C), BR-C expression is found over the middle part of the oocyte at a more posterior position than in the wild-type egg chambers (Fig. 5C). As a result, the dorsal appendages are also generated at a more posterior site (Fig. 5D). Additionally, ectopic expression of Dpp also leads to the disruption of the bilaterally symmetrical expression pattern of BR-C. We often observed that three or more groups of the follicle cells expressed the BR-C, which results in the formation of three or more malformed dorsal appendages (Fig. 5D). Taken together, these results indicate that expression of the BR-C is specified by Dpp along the AP axis. It is probably via an indirect route, because ectopic expression of Dpp in all follicle cells does not eliminate BR-C expression.

Our data suggest that the follicle cell and eggshell pattern, which is partly dependent upon the late expression pattern of the BR-C, requires signals from both the germline and the somatic follicle cells. The patterning along the DV axis is specified by Grk-DER signalling from the germline in a dose dependent way, as the dorsal-anterior columnar cells respond differently, according to how much signal they receive. Dpp signalling is required for the patterning of the eggshell along the AP axis. These two signalling pathways, one from the oocyte and the other from the follicle cells, co- ordinately specify patches of follicle cells to express the BR-C in a unique position in respect to both major axes. This, subsequently, directs the differentiation of the dorsal appendages in the correct position on the eggshell. Hence, the interaction of these two signalling pathways can divide the dorsal-anterior columnar follicle cells into at least three separate subgroups: BR-C late expression marks the dorsal appendage forming cells in the lateral-dorsal-anterior region; follicle cells anterior to them form the operculum and other anterior structures; follicle cells dorsal to them form the gap between the two appendages (Fig. 6).

In this paper we show that the generation of bilateral morphological structures, is directed by a symmetrical expression of a transcription factor which directs the subsequent morphogenesis of these cells. The localised spatial pattern is established by a TGF-α signalling pathway in a dose dependent manner and is further regulated by a TGF-β signalling pathway from different sources. The mechanisms described here, using the eggshell as a model, could be used to explain the generation of other bilaterally symmetrical structures.

BR-C expression in the lateral-dorsal follicle cells is induced by a medium concentration of Grk-DER signalling. Higher concentrations of Grk signal seems to induce another transcription factor Pnt, which is likely to be a repressor of BRC in the dorsal midline cells. It has been reported that the ectopic expression of pntP1 but, not pntP2, results in reduced appendage materials (Morimoto et al., 1996). This is different from our observations. The reason for this is probably that the Gal4 driver we used has a higher level of Gal4 expression in columnar cells. This may signify that pntP1 and pntP2 are functionally redundant, but work at different concentrations.

Is the BR-C under the control of ecdysone during oogenesis, as it is during metamorphosis? The answer could be yes, because expression of the BR-C observed in all follicle cells at stage 6 is not induced by the Grk-DER signalling pathway, and expression of the ecdysone receptor (EcR) is detected in all follicle cells during the same stage (Deng, Mauchline and Bownes, unpublished data). This coincidence might imply that the earliest BR-C expression pattern is mediated by ecdysone. However, the BR-C late expression pattern in the dorsal-lateral follicle cells is unlikely to be induced by ecdysone, since the mis-localisation of Grk signal in the ventral-anterior oocyte is sufficient to induce the ectopic expression of BR-C in the ventral follicle cells in egg chambers mutant for fs(1)K10.

The DV polarities of both the embryo and eggshell are initiated by the dorsal-anteriorly localised Grk. The downstream genes, Ras1, Draf and Mek, which are involved in transmitting the signal, are involved in the formation of both these DV polarities. However, the defects caused by mutation or mis-expression of these signal-transmitting genes are more severe in the eggshell patterning than in the embryonic patterning when compared with the grk and top/DER mutants. This led to the hypothesis that another signalling pathway may be involved in the transmission of the DV signal (Schnorr and Berg, 1996).

In addition, the existence of some mutants which only affect embryonic DV pattern formation indicates that there is a branch of the DV patterning pathway used only for eggshell polarity. We suggest that the BR-C, which only affects the eggshell pattern when it is mutated during oogenesis, is an effector gene for this branch. Another transcription factor, Pnt, which is expressed in the dorsal-midline follicle cells could also be part of this branch, because no embryonic defects have been reported when it is mis-expressed in oogenesis. Additionally, a ventrally expressed transcription factor, CF2, has been shown to be involved in both the eggshell and embryonic DV pattern formation (Hsu et al., 1996). Therefore, there should be a transcriptional repressor of CF2 in the dorsal follicle cells downstream of the Grk-DER signalling pathway. This transcriptional repressor must be involved in the establishment of DV polarities of both the embryo and eggshell, and remains to be identified.

The dorsal-anteriorly located appendages also mark the AP polarity of the eggshell. Their position is induced by the Grk-DER signalling pathway, thus the AP position of the appendages in the eggshell is originally established by the Grk signal. Our data suggest that this position is further refined by Dpp signalling, which is initiated in the nearby centripetal cells. Dpp signalling is directly involved in the specification of the anterior-most columnar cells and anterior eggshell structure, the operculum (Twombly, 1996). However, the regulation of the AP position of BR-C late expression and the subsequent AP position of the dorsal appendages seems to be indirectly controlled by Dpp signalling. Cells have to choose between the columnar cell fate and appendage producing fate. When Dpp levels decrease, the number of anterior columnar cells decreases, and this leads to the BR-C being expressed in a more anterior position when compared with the wild type. Conversely, the ectopic expression of Dpp in most of the follicle cells at stage 9 leads to the expansion of the anterior-follicle cell fate. Consequently, BR-C expressing cells are pushed towards the posterior. However, the posterior follicle cells seem to be predetermined, probably by the Grk-DER signalling pathway, during the early stages of oogenesis when the AP axis is initiated (González-Reyes et al., 1995; Roth, et al., 1995), and this may define the posterior frontier of the appendage-producing cells, which are marked by BR-C expression. Therefore, BR-C expressing cells are squeezed to form a curved line over the oocyte at a more posterior position than in the wild-type egg chambers.

The Dpp signalling pathway is widely used during development, including dorsal epidermal determination during embryogenesis, as well as wing, leg and eye-antennal disc patterning. A number of genes downstream of Dpp have been identified in other developmental processes. It has been reported that the Dpp receptors, saxophone (sax) and thick veins (tkv) are also involved in eggshell patterning (Twombly et al., 1996). However, in these sax and tkv mutants the dorsal-ventral positioning of the appendages, rather than the anterior-posterior positions was affected by the dpp mutants. A preliminary search for the target genes in the follicle cells has been undertaken by examining the expression pattern of some known Dpp downstream genes used during wing morphogenesis, including spalt, spalt-related and omb (Lecuit et al., 1996; de Celis et al., 1996). However, they may not be targets of Dpp in follicle cell patterning, since no β-galactosidase expression is detected in oogenesis in enhancer-trap lines containing P[lac-Z] insertions in these genes (Deng and Bownes, unpublished data). It seems likely there are other genes downstream of Dpp in follicle cell patterning, which remain to be identified.

Genetic analysis of the partial ‘loss-of-function’ BR-C mutants indicates that dorsal appendage morphogenesis requires a functional BR-C locus. This has been supported by the examination of BR-C expression in different mutants that affect DV patterning. Additionally, the BR-C is found to be involved in morphogenetic movements during metamorphosis. A few target genes of the BR-C, including Sgs4, Ddc and Fbp2, have been identified in larval and pupal tissues (Bayer et al., 1997).

Other target genes could include actin-binding proteins, because the morphogenetic movements require cell shape changes, which result from the rearrangement of the actin cytoskeleton within the cells. The non-muscle myosin-II gene, zipper, is a good candidate gene, since it has been shown to be involved in morphogenetic movements associated with dorsal appendage formation (Edwards and Kiehart, 1996). Strikingly, zipper has been identified as the Enhancer of broad (E(br)) locus during imaginal disc morphogenesis. E(br)/+, in br¹ flies, interferes with appendage elongation during metamorphosis (Gotwals and Fristrom, 1991; Fristrom and Fristrom, 1993). These observations may suggest a conserved regulatory pathway mediated by the BR-C involved in morphogenetic movements throughout development.

The discovery of the requirement for BR-C function in dorsal appendage formation and the role BR-C plays in linking pattern formation and morphogenesis indicates that the BR-C may provide a useful tool to study eggshell morphogenetic evolution in Drosophilinae. Sturtvant (1921) reported that variable numbers of dorsal filaments are present in different Drosophilinae species. For example, D. melanogaster has two appendages, while D. hydei has four, and other species have 3, 8, or 10 appendages, suggesting that the number and size of dorsal appendages has evolved rapidly. In this paper, we show that the size and number of dorsal appendages is modified when the size and number of groups of BR-C expressing cells is changed (Figs 3J, 5D). Therefore, the modification of the regulation of BR-C expression in the dorsal-anterior follicle cells may be the driving force for this morphogenetic evolution. By studying the mechanisms regulating BR-C expression in a small number of representative Drosophilinae species, some aspects of morphogenetic evolution may be understood at a molecular level.

We thank C. Bayer and J. Fristrom for sending us BR-C cDNAs; T. Schüpbach for anti-Grk antibody; C. Bayer, J. Fristrom, T. Schüpbach, A. Gon.zlez-Reyes, D. St Johnston, V. Twombly, M. Singer, F. M. Hoffmann, K. Matthews, A. Jarman, K. Kaiser and D. Glover for fly stocks; D. Zhao and other colleagues in M. Bownes lab for helpful discussions; T. Schlitt for help in complementation analysis; S.A. Krauss and M. Heck for dark-field microscopy; A. Jarman and two anonymous reviewers for critical reading of the manuscript. W.-M. D. is a Darwin Trust Ph.D. student. This research was supported by the Wellcome Trust.