The role of segment polarity genes during early oogenesis in Drosophila

The segment polarity gene hedgehog (hh) encodes a cell-cell signaling molecule required for specifying cell identity and regulating cell proliferation in a variety of processes during Drosophila development (reviewed in Ingham, 1995). Hh family proteins have been shown to act over short and long distances (reviewed in Perrimon, 1995; Johnson and Tabin, 1995; Ingham, 1995). In the former case, long-range effects of hh signaling are mediated by the local activation of genes encoding other signaling molecules, namely wingless (wg), a member of the Wnt family of secreted molecules, and decapentaplegic (dpp), a TGF-β homolog. Here we analyze the mechanism of hh signaling in the adult Drosophila ovary.

In the Drosophila embryonic ectoderm, hh is co-expressed with the transcription factor engrailed (en) in a stripe of cells at the posterior of each segment (Tabata et al., 1992; Lee et al., 1992). en is required for hh expression in the same cell (Tabata et al., 1992; Lee et al., 1992), and hh activity is required to maintain the expression of wg in the adjacent anterior cells (Hidalgo and Ingham, 1990). wg signaling, in turn, maintains en and hh expression in posterior cells (Martinez-Arias et al., 1988; DiNardo and Heemskerk, 1990; Mohler and Vani, 1992; Tabata et al., 1992; Lee et al., 1992), a process that requires the β-catenin homolog Armadillo (Arm), which accumulates in the cytoplasm of cells actively transducing the wg signal (reviewed in Kirkpatrick and Peifer, 1995). The stable expression of wg and hh signals in adjacent stripes of cells flanking the parasegment boundary resulting from these interactions is critical for the patterning of the entire embryonic segment (reviewed in Perrimon, 1994).

The maintenance of wg expression by hh involves interaction with another segment polarity gene patched (ptc), which encodes a novel transmembrane protein (Hooper and Scott, 1989; Nakano et al., 1989) and regulates the activity of a zincfinger protein encoded by cubitus interruptus (ci) (Ingham et al., 1991; Forbes et al., 1993). In early embryogenesis, ptc and ci are co-expressed in all non-hh/en-expressing cells (Motzney and Holmgren, 1995); later, during germ-band extension, ptc expression becomes restricted to cells flanking the hh/en domain (Hooper and Scott, 1989; Nakano et al., 1989; Hidalgo and Ingham, 1990). ci activity is required for the expression of ptc and wg. However, ptc represses ci activity, thereby down regulating its own expression as well as that of wg and any other genes similarly regulated by ci (Forbes et al., 1993). hh antagonizes this activity of ptc (Ingham et al., 1991), allowing wg and ptc to be expressed in cells receiving the hh signal. In the embryonic ectoderm, the extent of the hh signal is limited to cells in contact with those expressing hh (reviewed in Perrimon 1994).

Similar interactions between the segment polarity genes are involved in patterning imaginal discs. In wing and leg imaginal discs, hh and en are co-expressed in the posterior compartment (Tabata and Kornberg, 1994), while ptc and ci are expressed in the anterior compartment with ptc expression being elevated in a stripe 5-10 cells wide along the compartment border (Philips et al., 1990; Tabata et al., 1992). The effects of hh on patterning and proliferation throughout these discs are mediated by the activation of other signaling molecules along the compartment border. In wing imaginal discs, hh activates the expression of dpp and, in leg imaginal discs, the effects of hh are mediated via the activation of both dpp and wg (reviewed in Perrimon, 1995; Ingham, 1995). An analogous situation exists in the eye imaginal disc where hh is expressed posterior of the morphogenetic furrow. hh signaling represses ptc activity in anterior cells and thereby results in the activation of dpp in the morphogenetic furrow (reviewed in Heberlein and Moses, 1995). The interactions between hh, ptc and ci may reflect the fact that these genes form part of a regulatory module that has been widely conserved through evolution. In vertebrate embryos, expression of ci and ptc homologs surrounds hh-expressing cells (Roberts et al., 1995).

We have recently shown that hh is expressed in a group of specialized somatic cells at the extreme anterior of the adult Drosophila ovary and that hh activity is required for egg chamber formation (Forbes et al., 1996). The adult Drosophila ovary consists of approximately 16 ovarioles. At the anterior of each ovariole is a structure called the germarium within which the progeny of the germ-line and somatic stem cells assemble into new egg chambers (Koch and King, 1966; Lin and Spradling, 1995; Margolis and Spradling, 1995; also see Fig. 1A). The germarium has three recognizable regions. Region 1 contains the 2 –3 germ-line stem cells and 1 –2 dividing germ-line cysts. At the tip of region 1 is a stack of 6-9 non-dividing somatic cells called the terminal filament which are closely associated with another group of cells within the germarium called cap cells (Forbes et al., 1996). Both the basal cells of the terminal filament and the cap cells are in close proximity to the germ-line stem cells (Fig. 1A). The cystoblasts produced by the germ-line stem cells divide four times to give rise to 16-cell germ-line cysts that pass through region 2a, which is lined with non-dividing somatic inner sheath cells (Margolis and Spradling, 1995). After passing the somatic stem cells located at the region 2a-2b boundary, the germ-line cysts become encapsulated by somatic cells invaginating from the germarium wall. These somatic cells continue to divide and form a follicle cell monolayer around the cyst (Margolis and Spradling, 1995). The most mature cyst in the germarium and its surrounding follicle cells form an egg chamber which constitutes region 3. Anterior of this chamber, somatic cells interdigitate to make an interfollicular stalk of 5 –8 cells that separates the egg chamber from the rest of the germarium. The rest of the ovariole consists of 5 –7 progressively maturing egg chambers, each separated by an interfollicular stalk.

Hh protein can only be detected in cells at the most anterior tip of the germarium; yet hh activity is required for the proliferation and specification of somatic cells during germ-line cyst encapsulation in region 2b, a distance of 2-5 cells away from the terminal filament and cap cells (Forbes et al., 1996). In order to gain a further understanding of how hh signaling is mediated in the ovary, we have analyzed the expression and function of genes with which hh is known to interact in other developmental processes. We have found that, while neither wg or dpp mediate the effects of hh on somatic cells in the germarium, the antagonistic interaction of hh with ptc appears to be conserved in the adult Drosophila ovary.

The following fly lines were used to analyze the expression patterns of these genes in the ovary. The wglacZ (provided by Dr Norbert Perrimon), dpp-lacZ (Karpen and Spradling, 1992), ci-lacZ (Eaton and Kornberg, 1990) and ptc-lacZ H84 (supplied by Dr Cory Goodman) enhancer trap lines were all generated by inserting a lacZ-bearing recombinant P element into the corresponding loci. The wglacZ construct has been shown by Ingham et al. (1991) to be expressed in the same pattern as the wg RNA. The dpp-lacZ strain was l(2)10638 (Karpen and Spradling, 1992), which carries a PZ insertion 218 bp 5′ to the dpp transcript A site and expresses lacZ in a pattern identical to dpp in embryonic and imaginal discs (Twombley et al., 1996). The ci-lacZ was the ci-D^plac allele of the locus whose lacZ expression in embryos and disks is identical to that of ci RNA (Eaton and Kornberg, 1990). The ptc-lacZ H84 flies was from Dr Cory Goodman via Dr Nipam Patel. It has an embryonic staining pattern identical to that of ptc RNA (Forbes, 1992). In addition, flies carrying a chimeric gene containing a 2.5 kb ptc 5′ upstream sequence fused to the E. coli lacZ gene (Forbes, 1992) were used to analyze the ovarian expression pattern. This chimeric construct shows identical expression patterns as ptc RNA in embryos (Forbes, 1992). We also examined the expression of a ptc-lacZ strain, fs(2)02465 (Karpen and Spradling, 1992), which is a weak ptc allele that displays reduced fertility and a weak ptc wing phenotype (A. S. and Z. F., unpublished results). The above three lines show indistinguishable patterns of ovarian lacZ expression.

Flies containing the following heat-shock constructs, with the fulllength coding region of each gene under the regulation of the hsp70 promoter, were used to express these genes ectopically in the ovary: hs-hhMII (Ingham, 1993), hs-ptc (Ingham et al., 1991), hs-dpp (Twombly et al., 1996), hs-wg (Noordermeer et al., 1992) and hs-en (Heemskerk et al., 1991).

FRT 42 ptc^S2/Cyo and y^FL122; FRT 42 hsCD2 y⁺/Cyo stocks kindly provided by Dr A. Tomlinson were used to make ptc⁻ clones as described in Jiang and Struhl (1995).

All stocks were maintained at 25 °C on standard cornmeal/yeast/ agar media.

To express genes ectopically under the control of the hsp70 promoter, flies were cultured in vials of food with dry yeast and tissue paper. The tubes were immersed in a 37 °C water bath for 1 hour at every 12-hour intervals over a period of 3 days. Ovaries were then dissected and stained at different time points after heat shock as specified in Results.

Ovaries were dissected and fixed as described by Lin and Spradling (1993) using a protocol based on the embryonic protocol described by Patel et al. (1989).

The following antibodies were used to stain ovaries: The anti-en monoclonal antibody mAb49 provided by Dr Nipam Patel (Patel et al., 1989) was used at a dilution of 1:1; anti-lacZ rabbit polyclonal antibody from Kappal was used at a final dilution of 1:200; anti-Drosophila-α-spectrin rabbit polyclonal antibody supplied by Dr Dan Kiehart (Pesacreta et al., 1989) was used at a dilution of 1:200, antiarm N2 rabbit polyclonal antibody from Ms Sandra Oslic and Dr Mark Peifer (Riggleman et al., 1990) was used at a dilution of 1:200.

An anti-ptc monoclonal antibody was raised against the first 72 amino acids of the protein (P. W. I. and Wendy Norris, unpublished results). These antibodies recognizes a single band that corresponds to the molecular weight of the Ptc protein on western blot. It stains the expected Ptc localization pattern in wild-type embryos and the staining is absent in ptc null mutant embryos. In our experiments, this antibody was used at a final dilution of 1:1.

Two different antibodies directed against the Ci protein were used. Anti-Ci monoclonal 2A1, provided by Dr Robert Holmgren, was raised against an E. coli-expressed Ci protein missing the first 114 amino acid residues (Motzney and Holmgren, 1995). It recognizes an epitope outside the Zn-finger domain, immunoprecipitates Ci from embryonic extracts and stains only Ci proteins in the embryos as indicated by the absence of staining in ci− embryos. In our experiments, it was used at a final dilution of 1:1. An anti-Ci rabbit polyclonal antibody was raised against a fusion protein extending from residue 23 to 485, which includes most of the zinc-finger domain (P.W.I. and Wendy Norris, unpublished results). This antibody recognizes the Ci protein as well as a 45-50×10³M_r protein on western blots that does not correspond to Ci. In our experiments, this antibody was used at 1:100 dilution. Both 2A1 and the polyclonal antibodies stained somatic cells in the ovariole in the same pattern (see Results). However, the polyclonal antibody also stained large fibrous structures running through the cytoplasm of germ-line cells, which may be due to the cross reaction of this antiserum with the 45-50×10³M_r protein detected on the western blot. Although Ci is a putative transcriptional factor, none of the known anti-Ci antibodies shows nuclear staining in embryos or disks (e.g. Motzney and Holmgren, 1995; P. W. I. and Wendy Norris, unpublished results).

A mixture of three anti-CD2 monoclonal antibodies supplied by Dr Mike Puklavec, MRC Tissue Culture Laboratory, Oxford, were used to mark the ptc⁺ clones. Tissue supernatant from cell lines OX34, OX53 and OX55 were mixed at a 1:1:1 ratio and the mixture was used at a final dilution of 1:20. TRITC-conjugated AffilPure^™ donkey antirabbit secondary antibodies from Jackson ImmunoResearch Laboratories were used at 1:200 dilution.

Staining of DNA using its specific dye, DAPI, and activity staining of β-galactosidase with X-gal as a substrate were carried out as described in Lin and Spradling (1993). All stained samples were mounted in 50:50 PBS:glycerol and examined under a Zeiss Axiophot microscope equipped with Nomarski and epifluorescence optics. To mount fluorescently labeled samples, 2% of the antiquenching agent DABCO was added to the mounting medium. X-gal-stained ovarioles were photographed using 160T Ektachrome slide film; DAPI and antibody-stained ovarioles were photographed with 400T Ektachrome slide film or using a Star I CCD camera (Photometrics Ltd.). Images were processed using the Photoshop^™ program.

Mitotic clones lacking ptc activity were generated in females of the genotype y/y⁺; FL122/+; FRT42 hsCD2/FRT 42 ptc^S2, in which a heat-shock-inducible FLP recombinase (FL122, Golic and Lindquist, 1989) drives recombination between the chromatid bearing a mutant ptc allele and one marked with a heat-shock-inducible mammalian CD2 gene (Jiang and Struhl, 1995). Females of the above genotype were produced by standard crosses. Margolis and Spradling (1995) had previously reported that, in their flipase-mediated mitotic clone induction experiments, a 1-hour heat shock caused 30% of ovarioles to produce persistent clones derived from somatic stem cells. To increase the number of such persistent clones, we induced flipase activity repeatedly over a 15-hour period. Adult flies of 1 –3 days old were put in culture vials which were then immersed in a 37 °C water bath for 1 hour. This was repeated two more times at intervals of 4 hours. The flies were then kept in yeasted vials for 8 days prior to dissection of their ovaries to allow transient clones not derived from stem cells to complete oogenesis and leave the ovary. 4 hours before dissection, flies were heat shocked again for 1 hour at 37 °C to activate the CD2 marker. Ovaries were then dissected, fixed and stained with anti-CD2 antibody and DAPI as described above.

hh is co-expressed with en in a complementary pattern to ptc and ci in the embryonic segment and imaginal discs (Philips et al., 1990; Tabata et al., 1992; Ingham and Feitz, 1995). To study the regulatory relationship among these genes in the adult ovary, we first examined the ovarian expression of these genes. The ovarian expression of en was analyzed by staining wild-type ovaries with an anti-en monoclonal antibody (Patel et al., 1989). en protein is detected in the nuclei of terminal filament and cap cells (Figs 1B, 6B, 8A). Double staining of wild-type ovaries with both anti-en and anti-hh antibodies showed that both proteins are expressed in the same group of somatic cells at the tip of the germarium (data not shown). However, while hh expression is down regulated in the most distal terminal filament cells as flies age (Forbes et al., 1996), en continues to be strongly expressed in these cells. Ectopic expression of en via a hs-en construct has no effect on hh expression in the ovary and ectopic expression of hh does not alter the pattern of en expression (data not shown). This indicates that, in contrast to the early embryonic ectoderm, neither of these genes can induce the activation of the other outside its normal domain of expression in the adult ovary.

In both Drosophila embryos and imaginal discs, ptc and ci expression are closely associated with that of hh, and these genes are thought to be components of the hh signaling pathway (reviewed in Perrimon, 1995). To analyze the transcriptional regulation of ptc in the adult ovary, the expression of two ptclacZ reporters, an enhancer trap line and a promoter construct (see Materials and Methods), were analyzed in adult ovaries by X-gal staining. These reporter constructs have previously been shown to be expressed in the same pattern as the endogenous ptc gene as detected by RNA in situ hybridization experiments (Forbes, 1992). In the adult ovary, both constructs express lacZ in somatic cells in regions 1 and 2 of the germarium that constitute the germarium wall, within about 10 cells from the hh signaling source (Fig. 2A). lacZ is also expressed at high levels in the ovary sheath (data not shown).

To analyze the distribution of the Ptc protein, a monoclonal antibody raised against the first 72 amino acids of the Ptc protein was used to stain wild-type ovaries. Ptc is detected throughout the ovary in somatic and germ-line cells (Fig. 2B-F) as well as in the epithelial sheath (Fig. 2C,D). In these sheath cells, Ptc levels are considerably elevated around the germarium, particularly near the terminal filament.

The distribution of Ptc protein in the rest of the ovariole is less restricted than ptclacZ expression. While the ptclacZ lines are expressed almost exclusively in regions 1 and 2 of the germarium, Ptc protein is detected in somatic cells from the germarium to stage 13 egg chambers, although levels of protein are considerably higher in the germarium and recently budded egg chambers (Fig. 2B-F). In region 2 of the germarium, Ptc protein is present in the invaginating somatic cells where it accumulates in irregular aggregates (Fig. 2C,D). Similar aggregates are also seen in somatic cells in region 1 (Fig. 2B,D) and protein is occasionally detected in terminal filament cells (Fig. 2D). In region 3, protein is detected in the membranes of the follicle cells surrounding the stage 1 egg chamber and is also present in the cytoplasm (Fig. 2B-D). The distribution is rather uneven with protein aggregating between some follicle cells as well as between some follicle cells and the underlying germ-line cyst.

Ptc protein can also be detected in the germ line in the early stages of oogenesis (Fig. 2C,D). In region 1, perinuclear staining is occasionally observed in cystoblasts. However, in regions 2 and 3, the Ptc protein more typically accumulates in a branching structure in the center of germ-line cysts. Double staining of ovaries with anti-spectrin (Pesacreta et al., 1989) and anti-Ptc antibody reveals that this Ptc staining colocalizes to a large extent with the degenerating fusome, a cytoplasmic structure dense with vesicular material and staining strongly with anti-spectrin antibody (Lin et al., 1994; Fig. 2C,D). Even when the fusome is no longer detectable with anti-spectrin antibody, aggregates of Ptc protein can still be seen within the germ-line cyst (Fig. 2C,D). In postgermarial stage 2 egg chambers, this staining is replaced by perinuclear staining in nurse cells (Fig. 2E). This staining is strongest in the posterior-most nurse cells, which are nearest the oocyte, and therefore, likely to contain the largest amount of fusome breakdown products (Lin et al., 1994). In stage 3 egg chambers, Ptc protein level becomes reduced in the germ line.

In the post-germarial part of the ovariole (vitellarium), Ptc protein is detectable in stalk cells and the follicle cells of all egg chambers, with its level reduced in stage 5 and older egg chambers as compared to stage 1 and newly budded chambers (Fig. 2B). In these older egg chambers, the distribution of the Ptc protein also becomes more even and large aggregates are not observed. A low level of protein is present in all follicle cells (Fig. 2F).

The transcription of the ci gene in the ovary was examined using a cilacZ enhancer trap construct which is expressed in the same pattern as the endogenous gene in embryos and discs (Eaton and Kornberg, 1990; see Materials and Methods). The cilacZ reporter is expressed in somatic cells in the germarium and in the follicle cells of egg chambers up to stage 8-9 (Fig. 3A). In stages 3 to 6 egg chambers, follicle cells stain evenly and strongly. However, in stage 7-9 chambers, cilacZ staining becomes restricted only to posterior follicle cells (Fig. 3A).

To analyze the distribution of Ci protein, two different polyclonal antibodies were used to stain wild-type ovaries (see Materials and Methods); both revealed a protein distribution similar to the expression pattern of the cilacZ line. Ci protein is detected in somatic cells throughout the germarium and in all follicle cells of egg chambers up to stage 6 (Fig. 3B,C). After this stage, the protein level declines throughout the follicle cell layer although posterior follicle cells retain the Ci protein until stage 8 –9. Throughout the ovariole, Ci protein is excluded from the germ line (Fig. 3B,C).

The above staining also revealed cytoplasmic localization of the Ci protein. Although Ci is a zinc-finger protein and a putative transcription factor, it is present at high levels in the cytoplasm and appears to be excluded from the nucleus, as it was in embryonic and disk cells (e.g. Motzney and Holmgren, 1995; P. W. I. and Wendy Norris, unpublished results). This exclusion is evident from invaginating somatic cells in region 2 of the germarium to follicle cells in stage 1 and older egg chambers (Fig. 3B,C).

The ovarian expression of ptc and ci suggests that these genes may interact with hh during oogenesis. In the embryonic ectoderm and eye and appendage imaginal discs, ectopic expression of hh results in the ectopic expression of ptc. To test whether hh interacts with ptc in the same fashion in the adult ovary, we analyzed the effect of ectopic ovarian expression of hh on ptc expression using a promoter construct line (Forbes, 1992). ptclacZ flies, containing a single copy of the hs-hh construct, were exposed to the standard heat-shock regime and stained for lacZ activity 1 day after heat shock (see Materials and Methods). Ectopic expression of hh results in an elevated and expanded ptclacZ expression (Fig. 4). Although the heat-shocked hs-hh ovariole and the control ovariole as shown in Fig. 4 were both stained overnight to ensure the detection of the complete expression pattern, the blue staining in the germarium of the hs-hh ovariole reached the saturation level as shown in Fig. 4B within approximately 10 minutes while it took over 1 hour for the control germarium to reach the saturated level of staining. Hence, not only the level of expression was increased within the germarium as compared to wild-type control, but high levels of expression are also seen in the follicle cells of previtellogenic egg chambers as well as in the excess somatic cells accumulating between these chambers (Fig. 4). In stage 7-8 chambers, lacZ staining tends to be stronger in the posterior follicle cells, and the staining becomes limited only to a subset of posterior follicle cells in older chambers. Therefore, ectopic expression of hh caused ectopic ptc expression in a pattern similar to that of ci.

We have previously shown that the ectopic expression of hh in the ovary results in the production of an excess of somatic cells in the ovariole (Forbes et al., 1996, also see Fig. 5A). Since ectopic hh expression causes ectopic ptc expression, we examined whether the elevated levels of ptc are responsible for the hs-hh phenotype. The effects of ectopic expression of ptc alone were analyzed using flies containing two copies of a hs-ptc construct (Ingham, 1993). These flies were subjected to the standard heat-shock regime and their ovaries were dissected either immediately or 3 days after the heat shock. At both time points, the dissected ovaries show no detectable defects (Fig. 5B). This suggests that elevated levels of ptc cannot bring about the effects on somatic cells produced by the ectopic expression of hh.

It has been shown in embryos and imaginal disks that ptc represses ci activity and its own transcription (Hidalgo and Ingham, 1990; Forbes et al., 1993). The ectopic expression of ptc caused by the ectopic expression of hh suggests that, in the ovary, as in other systems, hh may function by antagonizing the ptc activity. If this is the case, the loss of ptc from somatic stem cells and their progeny would be expected to mimic the effect of ectopic expression of hh. To test this hypothesis, we generated ovarioles containing ptc⁻ mitotic clones using a heat-inducible FLP technique (see Materials and Methods). In this experiment, FLP recombinase drove recombination between a ptc⁻ chromatid and a ptc⁺ chromatid marked by a heat-inducible copy of the mammalian CD2 gene. ptc⁻ cells generated in this way can be distinguished from their surrounding ptc⁺ cells by their lack of CD2 expression. To generate ptc⁻ clones, FLP recombinase was activated over a 15-hour period by repeated heat shocking (see Materials and Methods). The heat-shocked flies were then allowed to grow at 24°C in yeasted vials for 8 days before their ovaries were dissected. Since it takes about 7 days for a germ-line cyst in region 2a to develop into a mature egg (Lin and Spradling, 1993), clones persisting for 8 days after heat shock should be derived from ptc⁻ somatic stem cells. In our experiments, ptc⁻,CD2⁻ clones were seen in more than 50% of the ovarioles. They frequently extend across several egg chambers and almost always include at least one interfollicular stalk. The ptc⁻ stalks contain many more than the normal number of 5-8 cells (Fig. 5C-F). The extra stalk cells appear to be derived from follicle cells that have budded off the adjacent egg chamber(s). The morphology of the giant stalks (Fig. 5D,F) resembles that seen in ovarioles in which hh is ectopically expressed (Fig. 5A; also see Forbes et al., 1996). When almost all of the somatic cells in the ovariole become ptc⁻, the morphology of such largely ptc⁻ ovarioles becomes indistinguishable from that of an extremely abnormal hs-hh ovariole resulting from ectopic hh expression. This similarity supports the proposal that hh affects somatic cell behavior by repressing the activity of ptc in these cells.

In mosaic ovaries, the ptc⁻ and ptc⁺ cells do not seem to contribute equally to egg chambers and excessive somatic cells lying between egg chambers (interfollicular cells, see Fig. 5C-F). Rather, ptc⁻ cells seem to contribute preferentially to inter-follicular cells while ptc⁺ cells associate with germ-line cysts. This hints that ptc may have a role in the adhesion of somatic cells to germ-line cysts. Cells lacking ptc may be less able to maintain contact with the germ-line cells.

In the Drosophila embryonic ectoderm and imaginal discs, the effects of hh are mediated to a large extent by the local activation of wg and/or dpp via the hh/ptc/ci pathway (reviewed in Perrimon, 1994, 1995). In order to investigate whether wg and/or dpp mediate the effects of hh on somatic cells in the adult ovary, the expression of these two genes was analyzed in wild-type ovaries and in ovaries in which hh was ectopically expressed. To confirm these analyses, the phenotypes produced by the ectopic expression of wg and dpp in the ovary were also examined.

The expression of wg was analyzed using an enhancer trap line that is expressed in the same cells as endogenous wg in embryos (Ingham et al., 1991; see Materials and Methods). X-gal staining and immunofluorescent staining using an anti-β-gal antibody showed that wglacZ is expressed in 1-3 somatic cells near the tip of the germarium, which appear to be the basal terminal filament cells and/or cap cells (Fig. 6C). Double labeling of ovaries from wglacZ flies with anti-en and anti-β-gal antibody showed that wglacZ is expressed in a subset of en-expressing cells (Fig. 6A-C). wglacZ expression revealed by both antibody and X-gal staining varies between ovarioles from a single ovary. While 20-30% of ovarioles contained no X-gal staining cells, one to three X-gal-positive cap cells were seen at the tips of the remaining ovarioles. Among these ovarioles, for those containing only a single wglacZ-expressing cell, this cell tends to be the cap cell furthest from the base of the terminal filament. This variability in staining was not seen with enlacZ or hhlacZ lines in which staining is seen in every ovariole. Therefore, the variability of wglacZ staining is not the result of uneven accessibility to staining reagents or other technical artifact but may reflect the cyclic activation and repression of wg expression in the cap cells.

The expression of wg in cap cells at the tip of the germarium makes it a candidate for a signal mediating the effect of hh on more posterior somatic cells. To test whether the increased proliferation of somatic cells caused by the ectopic expression of hh results from the ectopic activation of wg, wglacZ expression was analyzed in ovarioles from flies containing a single copy of the hs-hh construct exposed to the standard heat-shock regime. Ectopic expression of hh has no effect on wglacZ expression (Fig. 6D,E). This indicates that the effects of hh on somatic cell proliferation are not mediated by wg.

To confirm the above conclusion, we analyzed the effect of ectopic wg expression on ovariole development. Flies containing a single copy of a hs-wg construct were exposed to the standard heat-shock regime and their ovaries were dissected immediately after heat shock. Ectopic expression of wg does not mimic the effects of ectopic hh expression but resulted in a variety of other defects (Fig. 6F). The defects include very abnormal-looking germaria typically with an empty and elongated region 1 and fewer than normal numbers of invaginating somatic cells. In addition, the size of interfollicular stalks is reduced as well as the number of egg chambers produced per ovariole (Fig. 6F). Finally, degeneration of the germ line was observed in a significant number of previtellogenic chambers (Fig. 6F).

As wg is expressed mostly in cap cells and, yet, does not appear to mediate hh signaling, it is possible that wg signaling is involved in the regulation of hh expression in the germarium tip. Due to the lack of appropriate temperature-sensitive alleles of wg, it was not possible to test the requirement for wg in these cells. However, we analyzed the effect of ectopic expression of wg on hh expression. hhlacZ expression is unchanged in ovaries from flies containing a single copy of the hs-wg construct exposed to the standard heat-shock regime (data not shown). This indicates that wg cannot induce the activation of hh outside its normal domain of expression in the adult ovary.

The expression of dpp was analyzed using a dpplacZ line (see Materials and Methods). dpplacZ is not expressed in the germarium or early egg chambers (Fig. 7A). Expression is first detected in the anterior follicle cells of stage 8-9 egg chambers. In stage 10 egg chambers, the expression remains in the follicle cells associated with nurse cells as well as in centripetally migrating follicle cells but not in the majority of follicle cells covering the oocyte (Fig. 7A). In stage 14 chambers, there is a ring of strong staining surrounding the micropyle (data not shown), which confirmed the results of Twombly et al. (1996).

As dpp is not expressed in the germarium or early egg chambers, it is unlikely to be a second signal activated by hh in the wild-type germarium. However, it is possible that under conditions of ectopic hh expression dpp could be ectopically activated. We tested this possibility in dpplacZ flies containing a single copy of the hs-hh construct. When the flies are exposed to the standard heat-shock regime, ectopic hh expression had no effect on dpplacz expression (Fig 7B). This indicates that dpp does not mediate the effects of ectopically expressed hh in the germarium.

The above conclusion was further confirmed by analyzing the effect of ectopically expressing dpp in the ovary. Flies containing 6 copies of a hs-dpp construct were exposed to the standard heat-shock regime and their ovaries were dissected immediately after the heat shock. Ectopic expression of dpp resulted in the formation of fused egg chambers containing several germ-line cysts within a single follicle (Fig. 7C). This defect is opposite to the phenotype produced by ectopic hh expression.

Given that arm is a component of the wg transduction machinery, it is interesting that Arm has been shown to accumulate in cells at the tip of the germarium thought to be germline stem cells (Peifer et al., 1993). We analyzed in more detail the expression of Arm in the germarium tip using the anti-Arm polyclonal antibody N2 (Riggleman et al., 1990) and anti-en antibodies. Double staining for Arm and En shows that the Arm-accumulating cells at the tip of the germarium are cap cells (Fig. 8). A small amount of Arm protein can also be detected in some of the more distal cells of the terminal filament. However, in the germarium, Arm is present at lowest level, if any, in the germ-line stem cells. The Arm staining that outlines the germ-line stem cell cluster appears to be due to the staining of Arm in the surrounding somatic cells, since the cell membrane interphasing the two stem cells is barely stained (Fig. 8B).

hh activity is required for the proliferation and specification of somatic cells in region 2 of the germarium, yet Hh protein can only be detected in terminal filament and cap cells, a distance of 2 –5 cells from the site of hh requirement (Forbes et al., 1996). In this paper, we provide evidence that hh directly effects region 2 somatic cells via a signaling pathway which includes ptc and ci, but not wg or dpp.

The relative expression patterns of hh, en, ptc and ci in ovarioles show certain similarities with their expression patterns in imaginal discs, suggesting that the interactions between these genes may be similar in these systems. The coexpression of en and hh in terminal filament and cap cells is analogous to their expression in the posterior compartment of appendage imaginal discs; ci is expressed in the rest of the diploid somatic cells in the ovariole, analogous to the anterior compartment, and ptc is expressed at low levels throughout the ovariole with elevated levels in cells within about 10 cell diameters of the source of the hh signal, analogous to the elevated expression along the compartment border.

The ectopic activation of ptc expression in the ovariole when hh is ectopically expressed suggests that, as in other systems, elevated levels of ptc expression reflect the distance over which hh signaling is effective (reviewed in Ingham, 1995). The expression of ptc at high levels throughout region 1 and 2 of wild-type germaria, therefore, indicates that hh diffusing from the germarium tip affects these cells directly.

Ectopic hh expression results in the increased proliferation of somatic cells in region 2 (Forbes et al., 1996) as well as in ectopic ptc expression. However, it is not the elevated levels of ptc per se that result in the increase in somatic cell proliferation, since the ectopic expression of ptc alone has no effect on somatic cell number. In fact, it is the loss of ptc activity from somatic cells during egg chamber formation that mimics the effect of ectopic hh expression on follicle and stalk cells. ptc⁻ somatic stem cell clones give rise to ovarioles with egg chambers separated by giant stalks that closely resemble hs-hh ovarioles (Forbes et al., 1996). This indicates that hh acts by repressing ptc activity, analogous to its role in embryonic and imaginal disc development. Given the known interactions between ptc, hh and ci in other processes (reviewed in Perrimon, 1994, 1995; Ingham, 1995), it is reasonable to suggest that hh diffusing from the germarium tip antagonizes the negative effect of ptc on ci activity in cells in regions 1 and 2 of the germarium, thereby allowing the expression of ci-dependent genes in these cells, including ptc itself (Fig. 9). In region 2b, ci-dependent genes may mediate the effects of hh on cell proliferation and specification (Forbes et al., 1996). There must, however, be a mechanism for preventing these genes from acting in inner sheath cells. Inner sheath cells express high levels of ptc, indicating that they are responding to the hh signal. However, they do not proliferate or contribute to developing egg chambers. In the embryonic ectoderm, certain factors have been shown to alter the response of cells to hh signaling (Cadigan et al., 1994). The cells on the two sides of the hh domain respond differently to hh signaling. While ptc is expressed in all cells adjacent to the hh domain, only those anterior cells in which sloppy paired has been expressed are competent to express wg. In regions 1 and 2a of the germarium, signals from the underlying unencapsulated germ-line cysts may keep inner sheath cells in a quiescent state and alter their response to the activation of ci.

The accumulation of Ptc in the core region of the fusome in region 2 and 3 germ-line cysts is consistent with its being a transmembrane protein, even though its functional significance in germ-line development remains unknown. The abundant membrane vesicles in the fusome (Lin et al., 1994) are likely to be the site of Ptc accumulation. Future analysis of ptc⁻ germline clones and hts mutant ovaries defective in fusome will shed light on the role of Ptc accumulation in the fusome.

Like the neurogenic genes and the spitz group of genes (reviewed in Ruohla-Baker et al., 1994) hh, ptc and ci can be considered members of a gene cassette that encodes components of a signaling pathway that is used several times during Drosophila development (Fig. 9). However, the target genes of this pathway vary. In the embryonic ectoderm and imaginal discs, wg and dpp are regulated by hh, and these signals, to a large extent, mediate the effects of hh on patterning and proliferation (Basler and Struhl, 1994; Fig. 9). However, neither of these genes appears to be a target of the hh/ptc/ci pathway in the ovary. It is possible that other Wnt homolog(s) and/or other TGF-β family member(s) may be involved in responding to the hh/ptc/ci gene cassette in ovarian somatic cells.

The co-expression of hh/en and wg in cap cells differs from the embryonic ectoderm where their expression is mutually exclusive, probably reflecting the differences in regulatory interaction between these genes in ovary and embryo. In contrast to the embryonic ectoderm, the ectopic expression of any of these genes in the adult ovary cannot induce the expression of any other outside its normal expression domain (Heemskerk et al., 1991; Noordermeer et al., 1992; Ingham, 1993). However, these experiments do not address the possibility that these genes may be required for the expression of the others within their normal expression domains and, because of the absence of suitable temperature-sensitive alleles and the difficulty in making clones in non-dividing inner sheath cells, it has not been possible to further analyze these interactions.

PKA has been shown to antagonize Hh signal in leg and eye disks (Pan and Rubin, 1995; Li et al., 1995; Jiang and Struhl, 1995; Strutt et al., 1995; Herberlein et al., 1995; Lepage et al., 1995). In the ovary, the PKA activity is only required in the germ line (Lane and Kalderon, 1994, 1995). DCO⁻ clones (which lack the catalytic subunit of PKA) in ovarian somatic cells do not show any detectable abnormalities. In particular, they do not phenocopy the defects of mis-expressing hh or removing the ptc activity (A. F. and H. L., unpublished results). This indicates that PKA is not involved in the hh/ptc/ci regulatory cassette in the ovarian somatic cells (Fig. 9), which reflects the diversification of interactions between these signal transduction components in different tissues.

Cap cells are closely associated with the basal cells of the terminal filament and, like the terminal filament cells, express both en and hh. The expression of wg in some cap cells and the accumulation of elevated levels of arm protein in cap cells, but not terminal filament cells, indicates that cap cells are not simply a continuation of the terminal filament. Differences in the modulation of hh and en expression within the terminal filament also suggests heterogeneity within the terminal filament itself. The expression of a number of enhancer trap lines in different and overlapping subsets of somatic cells in the germarium tip (Lin and Spradling, 1993; H. L. and A. C. S., unpublished observations) further indicates a complexity within this group of cells, which may reflect the multiplicity of their functions. As well as their role in regulating somatic cell proliferation and specification via hh, cap cells and terminal filament cells may be involved in regulating the division of the underlying germ-line stem cells. Partial laser ablation of the terminal filament suggests that a signal repressing germ-line stem cell division is produced by terminal filament cells (Lin and Spradling 1993). Other signals from the somatic cells at the germarium tip may also influence germ-line stem cell asymmetry (H. L. and A. C. S., unpublished observations)

The expression of wglacZ in cap cells is variable, with only 70 –80% of ovarioles expressing the construct in one to three cells at any one time. Although the exact dynamics of lacZ expression may not faithfully reflect that of the Wg protein, this intermittent pattern of expression suggests that wg expression may switch on and off in a cyclic fashion in cap cells. It is possible that the cyclic activation of this signal may in some way be linked to the cell cycle of the underlying germline stem cells. Although we were not able to analyze the effects of removing wg activity from the ovary, the disruption of germarium structure that results from the ectopic expression of wg, particularly the defects in region 1, suggests that wg signaling may have a role in the earliest stages of oogenesis. However, wg signaling may not be directly involved in controlling germ-line stem cell division, since we found that Arm, a component of wg signaling pathway, is present in somatic cap cells but not germ-line stem cells as previously proposed (Peifer et al., 1993). As the accumulation of cytoplasmic Arm is a good indication of the transduction of the wg signal (Kirkpatrick and Peifer, 1995), this suggests that wg is signaling within the cap cells rather than directly to the germ-line stem cells. This possibility is further supported by the study of dishevelled (dsh), another component of the wg signaling pathway (van den Heuvel et al., 1993; Theisen et al., 1994; Couso et al., 1994; Yanagawa et al., 1995). dsh⁻ germ line produces eggs normally (Noordermeer et al., 1994), suggesting that wg/dsh pathway is not directly involved in soma to germ-line signaling. Further analysis on the role of the components of the wg signaling pathway in the ovary may reveal other cassettes of genes used in cell-cell and possibly autocrine signaling during early oogenesis as well as during a number of other stages of fly development.

We thank Vern Twombly and William Gelbart for generously providing dpplacZ and hs-dpp strains and for sharing their unpublished results. We also thank Andrew Tomlinson for FRT-ptc flies, Bob Holmgren and Cindy Motzney for anti-Ci antibody and cilacZ flies, Nipam Patel for anti-en antibodies and ptclacZ H84 flies, Sandra Orsulic and Mark Piefer for the anti-arm antibodies, Dan Kiehart for anti-spectrin antibodies and Michael Puleavec for an anti-CD2 mon-oclonal antibody. This work was supported by the Howard Hughes Medical Institute to A. C. S. and by a grant from National Institutes of Health (1R01HD33760-01) and an American Cancer Society Institutional Research Grant (383-0073) to H. L., who is also a recipient of a Junior Faculty Research Award (JFRA-608) from American Cancer Society and a Basil O’Connor Award (5-FY95-1111) from the March of Dimes Birth Defects Foundation.