Pointed, an ETS domain transcription factor, negatively regulates the EGF receptor pathway in Drosophila oogenesis

During Drosophila oogenesis, cell-cell signaling occurs between the germ-line-derived oocyte and the somatically-derived follicle cells that surround it. The Gurken(Grk)/Torpedo(Top) pathway mediates cell-cell signaling required for follicle cell fate decisions necessary to generate the dorsal/ventral (D/V) as well as the anterior/posterior (A/P) axes of the egg chamber and the future embryo (Schüpbach, 1987; Gonzalez-Reyes et al., 1995; Roth et al., 1995). Top encodes the Drosophila epidermal growth factor receptor (EGF receptor/Egfr/DER; Price et al., 1989; Schejter and Shilo, 1989) while grk encodes a transforming growth factor-alpha (TGF-α) homolog (Neuman-Silberberg and Schüpbach, 1993). Grk, acting in the oocyte, is thought to bind to the EGF receptor on adjacent follicle cells and activate the EGF receptor signaling pathway in these cells (reviewed in Anderson, 1995).

At least two follicle cell subpopulations are required for the establishment of these axes: posterior and dorsal follicle cells. The EGF receptor pathway is required for the determination of both of these subgroups; during early oogenesis, the EGF receptor and Notch pathways are required for proper posterior follicle cell fate determination (Gonzalez-Reyes et al., 1995; Roth et al., 1995; Ruohola et al., 1991), while during midoogenesis, activated EGF receptor induces dorsal follicle cell fate (Neuman-Silberberg and Schüpbach, 1993; Roth and Schüpbach, 1994; for review, see Rongo and Lehmann, 1996). Furthermore, two different cell types are observed among the dorsal follicle cells: dorsolateral appendage-producing cells and dorsal midline cells (Spradling, 1993).

How activation of the EGF receptor pathway generates different follicle cell fates is not understood. Do common or unique downstream effectors and target genes determine whether activation of the EGF receptor pathway results in posterior versus dorsal-anterior follicle cells or appendageproducing versus midline cells in the dorsal region?

Some of the effectors and target genes of the EGF receptor pathway have already been identified. Ras 1, Raf and MAP kinase function downstream of the EGF receptor in follicle cells and affect dorsal follicle cell fate decisions during oogenesis (Schnorr and Berg, 1996; Brand and Perrimon, 1994; Lu et al., 1994). In addition, expression of both rhomboid (rho) and nudel in follicle cells is regulated by the EGF receptor signaling pathway (Ruohola-Baker et al., 1993; Neuman-Silberberg and Schüpbach, 1994; Hong and Hashimoto, 1995). Rho, a putative transmembrane protein, appears to be part of a positive feedback loop that strengthens EGF receptor signaling in the dorsal-anterior follicle cells, and overexpression of rho is sufficient to dorsalize egg chambers and embryos (Ruohola-Baker et al., 1993). Nudel, a putative serine protease, is required for activation of the embryonic D/V pathway and its expression in dorsal follicle cells appears reduced in response to the EGF receptor pathway (Hong and Hashimoto, 1995). During embryogenesis, the EGF receptor also induces expression of Argos, an EGF-motif containing protein, which can inhibit activation of the EGF receptor in vitro (Schweitzer et al., 1995; Golembo et al., 1996).

To identify downstream components of the EGF receptor pathway that are required for follicle cell fate decisions in oogenesis, we have investigated the role of the ETS transcription factor, Pointed (Pnt). pnt encodes two proteins, PntP1 and PntP2, which each contain an ETS domain that mediates sequence-specific DNA binding through a novel helix-turn-helix structure (Klämbt, 1993; Scholz et al., 1993; Kodandapani et al., 1996). In addition, PntP2 contains a PNT domain, a motif conserved by a subset of ETS proteins, whose function is, as yet, unclear. In vitro studies have demonstrated that PntP2 transactivation activity can be stimulated by activated MAPK, while PntP1 appears constitutively active (O’Neill et al., 1994).

Pnt is required for proper photoreceptor cell determination during Drosophila eye development where it acts as a positive effector, downstream of the Sevenless receptor tyrosine kinase (RTK) pathway (Brunner et al., 1994; O’Neill et al., 1994). Pnt is also required for glial cell differentiation in Drosophila embryos, but its regulation is unknown (Klaes et al., 1994).

We show here that pnt is also required in Drosophila for proper follicle cell fate in the dorsal-anterior region of the egg chamber during oogenesis. Expression of pnt in both dorsal and posterior follicle cells is dependent on grk activity, indicating that pnt acts downstream of the EGF receptor pathway. In addition, overexpression of pnt generates a phenotype similar to that produced by loss of the EGF receptor. Therefore, whereas previous analyses in the eye have demonstrated a role for Pnt as a positive effector (Brunner et al., 1994; O’Neill et al., 1994), our data indicates a novel antagonistic function for pnt in oogenesis; in response to activation of the EGF receptor pathway, pnt is expressed and negatively regulates the EGF receptor pathway, possibly by integrating information from a second pathway.

Drosophila melanogaster stocks were raised on standard corn meal/yeast/agar media at 25°C. The following fly strains were used: Oregon R, l(3)6D1 (69D enhancer trap line) (Bier et al., 1989); fs(1) K10 (Wieschaus et al., 1978); to p^QY1, grkD^C29 (Schüpbach, 1987); grk^2B6 (Neuman-Silberberg and Schüpbach, 1993); pnt^Δ88(Scholz et al., 1993), pnt^Δ33, pnt^Δ78 (O’Neill et al., 1994); UASpntP1, UASpntP2 (Klaes et al., 1994); UASlacZ (Brand and Perrimon, 1993); KL32 (Clark et al., 1994); D-raf^F179, GAL 4 line 55B (Brand and Perrimon, 1994); GAL 4 line c324a (on the second chromosome) was a kind gift from L. Manseau; E3-26 (P[y⁺tubulin FRTlacZ, w⁺]) (Harrison and Perrimon, 1993; D. Harrison, personal communication); 82-w⁺ (FRT(82B), w⁺) and w⁻, hsFLP1; FRT(82B) (Xu and Perrimon, 1993); FRT(82B), ovo^D1 (Chou et al., 1993; Bloomington Stock Center). Crosses producing flies ectopically expressing or overexpressing UASpntP1 were carried out at 18°C.

In situ hybridization and X-GAL staining procedures were performed as described previously (Ruohola et al., 1991) with the following modifications. Ovaries were hand dissected in phosphate-buffered saline. For in situ hybridizations, ovaries were dissected fully into ovarioles before probe was added and all solutions were treated with 0.1% DEPC. In addition, ovaries were not stored for longer than one week at −20°C before use. pointed transcript-specific DNA probes were generated from EcoRI-BamHI digests of the plasmids pRSET P1 and pRSET P2 (O’Neill 1994 thesis, UC Berkeley). The probes correspond to nucleotides #1050-1733(P1) and #779-1746(P2) according to the numbering used in Klämbt (1993). For microscopic examination of chorion structures, eggs were mounted in lactic acid:Hoyers (1:1) as described previously (Wieschaus and Nüsslein-Volhard, 1986).

To obtain FLP-FRT-generated mosaics, recombinants between pnt^Δ88, pnt^Δ33, pnt^Δ78 and w; P[ry⁺, hs-neo, FRT]82B, P[ry⁺, w⁺]90E, hereafter referred to as FRT(82B)w⁺, were generated to produce FRT(82B), pnt/TM3, Sb. FRT(82B), pnt/TM3, Sb were crossed with w⁻, FLP1; FRT(82B)w⁺ and third instar larval progeny were heat shocked for 30 minutes in a 39°C water bath on two consecutive days to induce FLP1 activity. Female offspring of the genotype w, FLP1/w; FRT(82B), pnt/FRT(82B) were fed yeast for two days and eggs laid thereafter were analyzed.

To mark induced clones, a tubulin promoter FRTlacZ construct inserted around 84 (E3-26, Harrison and Perrimon, 1993 and D. Harrison pers. comm.) was recombined with the FRT(82B)w⁺ chromosome. FRT(82B), tubulinFRTlacZ(84)/TM3, Sb males were crossed to w⁻, FLP1; FRT(82B)w⁺ females to produce w⁻, FLP1/Y; FRT(82B), tubulinFRTlacZ(85D)/ FRT(82B)w⁺ males. These males were crossed with FRT(82B), pnt^Δ88, w⁺/TM3, Sb females and mosaic clones were induced in their progeny as described above. w⁻, FLP1/w; FRT(82B), pnt^Δ88, w⁺/FRT(82B), tubulinFRTlacZ(84) females were distin-guished by their dark red eye phenotype [FRT(82B), pnt^Δ88, w⁺] and by X-GAL staining of their ovaries. As the tubulin FRTlacZ construct E3-26 contains a FRT, we checked the resultant w, FLP1/w; FRT(82B), tubulinFRTlacZ(84)/TM3, Sb progeny of the above cross for mosaic patches in their ovaries. Out of 100 dissected females, no mosaic patches were found in these control animals.

Germ-line clones of pnt^Δ88 were generated using FRT(82B), ovo^D1 (Chou et al., 1993). w; FRT(82B), pnt^Δ88, w⁺/TM3, Sb females were crossed with w, FLP1/Y; TM3, Sb/+ to obtain w, FLP1/w; FRT(82B), pnt^Δ88, w⁺/TM3, Sb females, which were crossed to w; FRT(82B), ovo^D1/TM3, Sb males, and mosaic clones were induced in the progeny as before. Females of the genotype FRT(82B), pnt^Δ88, w⁺/FRT(82B), ovo^D1 and either w, FLP1/w or w/w were selected and their ability to lay eggs was scored.

The pointed locus encodes two proteins, PntP1 and PntP2, which share a region homologous to the ETS family of transcription factors (Klämbt, 1993). The P1 and P2 transcripts also contain mutually exclusive exons that allow the generation of transcript-specific probes to distinguish the expression of P1 and P2 RNA during oogenesis by in situ hybridization (see Materials and Methods for details).

In wild-type Drosophila ovaries, the P1 transcript is expressed in the germ line and in posterior and dorsal-anterior follicle cells. Weak expression of P1 RNA is detected in the germarium (data not shown). Accumulation of the transcript in the oocyte is detected around stage 5-6 (Fig. 1A). At stage 6-7, P1 is expressed in the follicle cells that surround the oocyte at a time when the oocyte nucleus is still located at the posterior of the oocyte (Fig. 1B). By stage 8, expression is restricted to posterior follicle cells (Fig. 1C,D). During stage 9-10, P1 is detected in dorsal-anterior (Fig. 1E, ‘early pattern’) and posterior follicle cells (comparable to Fig. 1D). During later stages, both of these expression patterns refine into two areas; two dorsal patches and two posterior half-circles (Fig. 1F, ‘late pattern’).

Similar to P1, P2 expression is detected in the germ line and follicle cells in wild-type ovaries. P2 is present in the germarium and in the oocyte from stages 2-8 (Fig. 2A). During stage 8, the P2 transcript is localized at the anterior of the oocyte (Fig. 2A). Follicle cell expression is observed in stage 9-10 egg chambers in a pattern resembling the early P1 expression pattern in dorsal-anterior follicle cells (Fig. 2B,C, ‘early pattern’). During later stages, P2 expression is restricted to two groups of anterior dorsolateral follicle cells that flank the oocyte nucleus (Fig. 2D,E, ‘late pattern’).

The expression of the P1 and P2 transcripts specifically in posterior and dorsal-anterior follicle cells (Figs 1, 2) suggested that expression of pnt may be regulated by the signaling cascade that determines these follicle cells’ fates, namely, the EGF receptor pathway. As activation of the EGF receptor pathway is dependent upon a signal from the oocyte (Schüpbach, 1987; Neuman-Silberberg and Schüpbach, 1993), we first examined whether P1 expression in follicle cells is induced by the oocyte. In some mutants that cause female sterility, compound egg chambers are generated in which an oocyte can be found at both the anterior and posterior end of the egg chamber (King, 1970; Neuman-Silberberg and Schüpbach, 1993). In situ hybridization of compound egg chambers produced by top^QY1 mutants with a P1-specific probe indicated that P1 is specifically expressed in follicle cells at both ends of the egg chamber, adjacent to the oocytes (Fig. 1G). These data show that anterior follicle cells can express P1 if the oocyte is in close proximity and suggests that signaling from the oocyte induces P1 expression in posterior follicle cells in wild-type egg chambers. To determine whether grk-mediated signaling from the oocyte is required for P1 expression in posterior follicle cells, P1 expression was examined in grk^2B6/grk^DC29 null mutant egg chambers. Whereas control egg chambers exhibit strong posterior follicle cell staining (Fig. 1D), no P1 RNA is detected in posterior follicle cells in grk^2B6/grk^DC29 egg chambers (Fig. 1H). In addition, P1 expression in dorsal follicle cells was not observed in these mutant egg chambers (data not shown), however germ-line expression was unaltered. These results demonstrate that P1 RNA expression in the follicle cells is dependent upon grk and suggest that PntP1 expression in posterior and dorsal follicle cells is controlled by the EGF receptor signaling pathway.

To test whether P2 expression is also dependent on the Grk/Top signaling pathway, P2 RNA expression was examined in grk^2B6/grk^DC29 egg chambers by in situ hybridization with a P2 transcript-specific probe. In contrast to the dorsal-anterior follicle cell expression observed in wild-type egg chambers (Fig. 2B-E), P2 is not detected in follicle cells in these grk null mutant egg chambers (Fig. 2F,G). Germ-line expression of P2, however, was unaffected by the loss of grk activity (compare Fig. 2D to F). Mutations in genes involved in the Grk/Top signaling pathway, such as top and fs(1)K10 (Wieschaus et al., 1978) also affect pnt expression. In non-compound egg chambers generated by top^QY1 mutants, the region of P2 expression is reduced, while in fs(1)K10 mutants, P2 expression is expanded (data not shown). These alterations in pnt expression correlate with the changes in dorsal follicle cell fate previously observed in top or fs(1)K10 mutants (Schüpbach, 1987; Wieschaus et al., 1978). These results indicate that P1 and P2 expression in posterior and dorsal follicle cells is dependent on Grk/Top signaling.

The induction of pnt expression by the Grk/Top pathway suggested a functional requirement for pnt in follicle cell determination. To investigate this possibility, the phenotype of pnt null mutants in oogenesis was examined. As null alleles of pnt cause embryonic lethality (Mayer and Nüsslein-Volhard, 1988), mosaic animals were generated using the FLP/FRT recombination system (Golic, 1991) and the allele, pnt^Δ88, which is null for both P1 and P2 (Scholz et al., 1993; Brunner et al., 1994). The fate of the follicle cells was examined by analyzing the outer layer of the egg shell, the chorion, which is secreted by follicle cells and therefore serves as a ‘footprint’ for these cells. The dorsal appendages, located at the dorsal-anterior end of the egg shell, are part of the chorion and serve as a marker for the fate of dorsal-anterior follicle cells. Whereas two dorsolateral appendages are detected in control eggs (Fig. 3A), the eggs laid by mosaic animals exhibit a single broad appendage. This mutant appendage is four times wider than a single wild-type appendage, suggesting that the phenotype does not result from two appendages fusing together, but rather cells in the middle region taking on an appendage-producing cell fate (Fig. 3B). Importantly, a similar phenotype (comparable to Fig. 3B) was detected when either P1(pnt^Δ33) or P2(pnt^Δ78) specific mutant alleles were used, suggesting that both P1 and P2 are required in the same process during oogenesis. pnt^Δ88, the allele null for both P1 and P2 (Scholz et al., 1993) was used in subsequent experiments.

To determine which cells require pnt function, marked mosaic clones were generated (for details see Materials and Methods). Briefly, the wild-type chromosome was marked with the lacZ gene controlled by a β-tubulin promoter (Harrison and Perrimon, 1993) which directs expression in both follicle cells and the germ line. The chromosome bearing the pnt^Δ88 null allele does not carry the lacZ marker, thus pnt^Δ88 homozygous mutant cells can be detected due to a lack of β-galactosidase (β-gal) activity (Fig. 4B-D; Table 1B). X-gal staining of ovaries from potentially mosaic females indicated that eggs lacking a mosaic patch had wild-type dorsal appendages (Fig. 4A; Table 1B, n=740). In addition, a complete lack of pnt in the germ line had no effect on dorsal appendage formation (Fig. 4B, n=48). However, the mutant broad appendage phenotype was detected in all egg chambers that had a β-gal-negative patch of follicle cells in the dorsal-anterior region (Fig. 4C,D; Table 1B, n=17/17). Lastly, in eggs in which the entire follicle cell layer was mutant for pnt^Δ88, no stronger phenotypes in the dorsal-anterior region were detected (Fig. 4C) indicating that the broad appendage is the strongest phenotype generated by the lack of pnt.

The appendage phenotype generated by the loss of pnt indicated that pnt affects patterning of the dorsal-anterior region of the chorion. Mutations in grk and top also affect the dorsal region of the egg shell (Schüpbach, 1987). However, these mutants affect the entire dorsal region whereas loss of pnt specifically affects the midline region (Figs 3B, 4C-D). pnt may also have a function in posterior follicle cells since loss of pnt function in these cells leads to mislocalization of an A/P axis marker, Kinesin-β-galactosidase (Clark et al., 1994; K. T. and H. R.-B., unpublished observations).

As the Grk/Top pathway affects the D/V axis of the embryo as well as the egg shell (Schüpbach, 1987), we next tested whether loss of pnt in the follicle cells also alters embryonic patterning. Eggs exhibiting the broad appendage phenotype were isolated and allowed to develop. The majority of these eggs hatched and developed into fertile adults. Therefore loss of pnt in the dorsal region does not lead to severe grk- or top- like D/V axis defects in the embryo. This was expected as the loss of pnt only affects a subgroup of the dorsal follicle cells affected by mutations in grk or top. However, since the percentage of eggs without early embryonic development was higher than that detected in the control group (data not shown), we cannot rule out the possibility that the loss of pnt in follicle cells may cause some embryonic defects.

Previous studies indicated that pnt is not required in the germ line during oogenesis; mosaic females with a germ line homozygous for the pnt^9J31 or pnt^B74 mutation did not exhibit any apparent defects in oogenesis (Mayer and Nüsslein-Volhard, 1988). As the possibility that pnt^9J31and pnt^8B74 are not null mutations could not be ruled out, the function of pnt in the germ line was reinvestigated. Using the mutant allele, pnt^Δ88, that is null for both P1 and P2 (Scholz et al., 1993), we generated germ-line clones by both ovo^D1 (Chou et al., 1993) and FLP/FRT-methods (Golic, 1991). Similar to the results found with pnt^9J31 and pnt^8B74, eggs from mosaic females with a germ line homozygous for the pnt^Δ88 mutation did not exhibit any defects in the dorsal appendages or any discernible embryonic phenotype (Fig. 4B; Table 1). Thus, in the germ line, pnt does not appear to play an important role in oogenesis. These observations indicate that the phenotype detected in the pnt^Δ88 mosaics is due to a lack of pnt in follicle cells. The following analyses therefore focus on the function of pnt in follicle cells. Furthermore, within the follicle cell layer, Pnt function appears to be required in midline follicle cells for proper formation of that region. In marked mosaic clones, the presence of cells wild type for pnt in the midline region but not in the dorsolateral region correlated with wild-type midline structures (Table 1B). This observation indicates that pnt expression is required in the midline region to elicit proper dorsal follicle cell patterning and suggests that the early pattern of pnt expression with high concentration in the midline region (Figs 1E, 2B,C) relates functionally to the above observed phenotypes.

The lack of pnt expression in grk null mutant egg chambers (Figs 1H, 2F,G) and the phenotype associated with null alleles of pnt (Figs 3B, 4C,D) suggest that pnt acts downstream of the Grk/Top pathway and affects dorsal follicle cell patterning. To further examine the role of pnt in dorsal follicle cells, we overexpressed it in subsets of follicle cells by using the GAL4/UAS system (UAS=upstream activation sequence for GAL 4; Brand and Perrimon, 1993). Two different GAL4 insertion lines, 55B (Brand and Perrimon, 1994) and c324a (a gift from L. Manseau) were used (see Materials and Methods). The GAL4 expression pattern for these lines was determined by crossing them to a line carrying a UAS-lacZ gene and staining their progeny for β-gal activity. 55B activates β-gal expression in an approximately 8-cell-wide band of follicle cells at the anterior of the oocyte in stage 8-9 egg chambers (Brand and Perrimon, 1994). Line c324a activates β-gal expression in a patchy manner in anterior follicle cells with higher (Fig. 5I; 50% of the egg chambers) or equivalent expression in dorsal versus ventral follicle cells around stage 9-10. Thus, the 55B line will induce expression of a UAS-target gene in the anterior follicle cells at the ventral and dorsal sides, and c324a will induce expression either in dorsal (50% of the egg chambers) or in dorsal- and ventral-anterior follicle cells. To create GAL4-responsive pnt P1 and P2 genes, the coding sequences for pntP1 and pntP2 were cloned downstream of a tandem array of five optimized GAL4-binding sites (a generous gift from C. Klämbt; Klaes et al., 1994).

When either 55B or c324a was crossed to flies carrying UASpntP2, the female progeny laid wild-type eggs which developed normally (data not shown). However, when 55B or c324a was crossed to flies carrying the UASpntP1 construct, the females were sterile or semisterile, respectively, and produced eggs that had dorsal defects as judged by their chorion phenotypes. Instead of two wild-type dorsal appendages (Fig. 5A), no appendages or small ‘nubs’ of appendage material were detected (Fig. 5B, big arrowheads). In addition, ‘crumpled’ chorion material was consistently observed (Fig. 5B, small arrowheads). However, no defects were detected on the ventral side of the egg shell. Ectopic expression or overexpression of P1 but not P2 can therefore cause a loss of dorsal appendages, a phenotype associated with loss of Grk/Top signaling. In contrast, the loss-of-function studies (Figs 3, 4) indicated a functional requirement for both pntP1 and pntP2 for proper dorsal patterning. As ectopic or overexpression of pntP1 suffices to affect dorsal cell fates, whereas expression of pntP2 does not, the results together suggest that pntP1 may act in parallel or downstream of pntP2. For instance, the expression of pntP1 may be regulated by pntP2 which may in turn require posttranslational modifications such as phosphorylation for full activity (O’Neill et al., 1994).

The loss of dorsal appendages caused by ectopic expression or overexpression of P1 suggests that pnt acts in an opposite manner to the Grk/Top signal with respect to the chorion. This chorion defect, however, is a terminal phenotype which could be generated either by a transformation of appendage-producing cells into midline follicle cells or by loss of both follicle cell types. To analyze the follicle cell defect that leads to the chorion defect following overexpression of P1, a P element enhancer trap line, 69D (Fig. 5D) was used as a cell fate marker. 69D is downstream of the Top pathway and expression of this marker is expanded in dorsalized egg chambers overexpressing Rho (Ruohola-Baker et al., 1993), or a constitutively active form of Raf (D-raf^F179; data not shown) and in fs(1)K10 mutant egg chambers (Fig. 5H), whereas expression is reduced in a top^QY1mutant background (Fig. 5G). Importantly, this line marks both the dorsal midline and the dorsolateral follicle cells during stages 9-14 in oogenesis Therefore, if overexpression of pntP1 transforms dorsolateral, appendage-producing cells to midline cells, no change in 69D pattern is expected. However, if both cell types are lost, loss of 69D marker should be observed.

Overexpression of pnt, generated by GAL4 lines 55B or c324a, leads to a severe reduction in the number of follicle cells expressing the 69D marker (Fig. 5E,F). Using the 55B line, the extreme phenotype detected in 34% of the egg chambers (n=35 egg chambers) was a loss of 69D expression in all dorsal-anterior follicle cells except a row of cells between the nurse cells and the oocyte (Fig. 5E,F). The remaining ring of 69D-positive follicle cells is thought to be independent of the D/V patterning system (Neuman-Silberberg and Schüpbach, 1994) suggesting that the extreme phenotype generated by overexpression of pntP1 is a total loss of dorsal follicle cell fate. An intermediate phenotype was detected in 29% of the egg chambers in which a milder reduction of the 69D pattern was observed (data not shown). These results demonstrate that ectopic or overexpression of pntP1 results in an overall reduction in the number of both dorsolateral and dorsal midline follicle cells, a phenotype similar to that generated by the loss of grk or top (Schüpbach, 1987).

Preliminary results also indicate that overexpression of pntP1 reduces expression of rhomboid, another dorsal follicle cell marker (data not shown). As rhomboid is required downstream of the EGF receptor for dorsal follicle cell determination (Ruohola et al., 1993; Neuman-Silberberg and Schüpbach, 1994), the results with both the 69D and rhomboid markers indicate that overproduction of pnt results in a down regulation of components of the EGF receptor pathway. These results suggest that Pnt antagonizes EGF receptor signaling by affecting components of this pathway.

We have examined the expression pattern and functional requirements for Pnt, an ETS transcription factor, during Drosophila oogenesis. We show here that pnt is required for proper dorsal follicle cell fate. Loss of pointed results in a transformation of dorsal midline follicle cells to appendageproducing follicle cells. Overexpresssion of pntP1 results in a loss of dorsal-anterior follicle cells, a phenotype similar to that generated by loss of EGF receptor pathway. In addition, expression of pointed is induced by the EGF receptor pathway and overproduction of pointed down-regulates components of the EGF receptor pathway. These results indicate a novel antagonistic function for pointed; in response to activation of the EGF receptor pathway, pointed is expressed and negatively regulates the signal mediated by the EGF receptor pathway in dorsal follicle cells. As such, pointed may identify an inhibitory branch of the EGF receptor pathway or may integrate inhibitory information from another pathway.

During mid-oogenesis, Grk induces dorsal follicle cell fate (Neuman-Silberberg and Schüpbach, 1993, 1994). In response to Grk/Top signaling, dorsolateral follicle cells secrete the dorsal appendage chorion structures, whereas the dorsal midline cells do not. Thus, the dorsal region of the egg shell exhibits a pattern arbitrarily designated 2-1-2 where 2 designates the appendage regions and 1 designates the dorsal midline (Fig. 6). Although grk is required to establish the entire dorsal region of the egg, it is currently unclear how these different dorsal cell fates are generated.

One possibility is that a simple concentration gradient of a morphogen induces these different dorsal follicle cell fates. For example, a gradient of nuclear localization of the morphogen Dorsal is sufficient to specify the expression of different sets of zygotic genes and, thus, different cell fates (Roth et al., 1989; Rushlow et al., 1989; Steward, 1989). During midoogenesis, grk is localized at the dorsal-anterior of the oocyte and appears most concentrated at the dorsal midline in wild-type flies (Roth and Schüpbach, 1994). A simple concentration gradient model predicts that a low concentration of Gurken would induce the appendage-forming cell fate and a high concentration would induce the midline cell fate. The model also predicts that overactivation of the EGF receptor pathway would induce the midline fate. However, two sets of experiments indicate that overactivation of the EGF receptor pathway can result in the appendage-producing, not midline, cell fate. First, the most extreme phenotype generated by overexpression of grk is an excess of dorsal appendage material over the dorsal side (Fig. 1D in Neuman-Silberberg and Schüpbach, 1994). Second, flies bearing a gain-of-function mutation in Raf(D-raf^F179) lay eggs that exhibit a phenotype in which more cells take on an appendage-producing cell fate, including cells at the dorsal midline (Brand and Perrimon, 1994; this laboratory). It is worth pointing out that while the milder phenotype in both of these cases is expansion of the midline region, the extreme phenotype is the transformation of the midline cells to the appendage-producing cells. Therefore a simple morphogen concentration gradient model does not explain how wild-type patterning is generated in the dorsal region of the egg.

Alternatively, the EGF receptor pathway may have the potential to induce the entire dorsal region to take the follicle cell fate 2 and an additional mechanism produces the cell fate 1 at the dorsal midline. pnt could represent a component of an independent pathway that induces the dorsal midline cell fate. However, ectopic expression of P1 in ventral follicle cells had no apparent effect. Also overexpression of pnt results in a loss of the 69D marker in dorsal midline cells (Fig. 5E,F), indicating that overproduction of pnt reduces the number of dorsal follicle cells and that pnt does not directly induce the dorsal midline cell fate.

How then is patterning generated in the dorsal region of the egg? If the effective activity of EGF receptor signaling were reduced at the dorsal midline by an inhibitory factor, a pattern 2, 1, 2 could be generated. Our data suggest that pnt may encode this inhibitory activity. pnt is expressed in dorsal midline follicle cells (‘early pattern’, Figs 1E, 2B,C) and the loss of pnt in this region results in the midline cells taking on an appendage-producing cell fate, 2 (Figs 3B, 4C-D). This result suggests that Pnt reduces the effective activity of the EGF receptor pathway at the dorsal midline (Fig. 6). Loss of pnt may result in the dorsal region receiving a uniform level of the EGF receptor signal, resulting in a loss of patterning in the dorsal region. Preliminary results indicate that overexpression of pntP1 reduces expression of another dorsal follicle cell marker, rhomboid. As rhomboid is required downstream of the EGF receptor for dorsal follicle cell determination (Ruohola et al., 1993; Neuman-Silberberg and Schüpbach, 1994), the results with both the 69D and rhomboid markers argue that, when overproduced, pnt affects dorsal follicle cell determination and not necessarily later stages of differentiation. However, it is currently unclear whether pnt directly or indirectly affects the expression of these components of the EGF receptor pathway. For example, pnt could act indirectly by inducing the expression of a factor that creates a negative feedback loop. It is interesting to note that in C. elegans, Lin-1, an ETS-domain transcription factor defines an inhibitory branch for the EGF receptor pathway (Beitel et al., 1995).

As grk induces expresssion of pnt (Figs 1H, 2F,G), high levels of grk should induce high levels of pnt expression. Our model predicts that when Grk is overexpressed, patterning in the dorsal region should be maintained, as high levels of pnt would negate the effects of high levels of the Grk/Top signal. However, the extreme phenotype generated by overexpression of Grk is dorsal midline cells taking on an appendageproducing cell fate, suggesting that high levels of Pnt alone are insufficient to inhibit high levels of Grk activity. As many ETS proteins require a protein partner to function (for review see Wasylyk et al., 1993) an additional factor may be required for Pnt to function. The expression of this cofactor would be predicted to be limiting and not induced by Grk (Fig. 6II). Alternatively, an as yet unidentified pathway may mediate cell-cell signaling that establishes the dorsal midline cell fate and pnt may function to integrate this pathway with the Top pathway (Fig. 6I).

We thank Drs D. Baker, C. Berg, D. Kimelman, J. Schnorr, members of the Ruohola-Baker laboratory and anonymous reviewers for very helpful comments on the manuscript. We also thank C. Klämbt for helpful comments throughout this study and for the UASpntP1, UASpntP2 and pnt^Δ88 fly strains. Thanks also go to T. Schüpbach for grk null alleles; N. Perrimon for the UASDraf and 55B; D. Harrison and N. Perrimon for tubulinFRTlacZ lines; L. Manseau for providing the c324a GAL4 line prior to publication and K. Holder for the art work. We would also like to thank D. Montell for directing us to the P-element insertion lines which turned out to be pnt. This work was supported in part by grants from the National Institute of Health (T32 GM07270) to K. C. J. and (R01 HD 32464) to H. R.-B., the American Cancer Society (IRG 184B ACS) to A. M. M. and (DB-123) to H. R.-B., the American Heart Association (Established Investigatorship Award 95002990), the March of Dimes Birth Defects Foundation (Basil O’Conner Starter Research Grant 5-FY94-0929) and the Pew Memorial Trust (Pew Scholarship in the Biomedical Sciences) to H. R.-B.