A role for extra macrochaetae downstream of Notch in follicle cell differentiation

Stem cells in adult tissues exhibit the potential to differentiate into a limited variety of cell types. There is considerable interest in understanding the mechanisms of their maintenance and differentiation. A specific example of adult somatic stem cell differentiation is found in the Drosophilaovary, which possesses both germline and somatic stem cells. The unit of Drosophila oogenesis is the ovariole(Spradling, 1993)(Fig. 1), made up of a number of egg chambers, each of which consists of 16 syncytial germline cells (15 nurse cells and the oocyte) surrounded by a monolayer of follicle cells. The egg chambers in each ovariole grow and are pushed posteriorly by the formation of new egg chambers, so that each ovariole contains multiple egg chambers arrayed from youngest to oldest until the very oldest egg chamber becomes an egg (Fig. 1). Two types of stem cells, germline and somatic, reside in a structure called the germarium at the anterior tip of the ovariole. The germline stem cells produce germline cystoblasts, each of which undergoes four division cycles with incomplete cytokinesis to form a 16-cell cyst. As each cyst ceases division, it encounters somatic stem cells and their progeny and is enveloped, or`packaged', by a layer of somatic follicle cells. Newly packaged cysts exit the posterior end of the germarium in a process called budding. Intercyst cells, which separate the newly budded egg chamber from the germarium, undergo intercalation movements to form a three- to seven-cell structure called the stalk. At each terminus of each stalk lies a pair of polar cells, which organizes the axial polarity of the egg chamber(Frydman and Spradling, 2001; Grammont and Irvine, 2002; Torres et al., 2003) and induces differentiation of its neighbors during later stages of oogenesis(Grammont and Irvine, 2002; McGregor et al., 2002; Silver and Montell, 2001; Xi et al., 2003).

Although much is known about the maintenance and differentiation of the ovarian germline stem cells (GSC) (Bhat and Schedl, 1997; de Cuevas and Spradling, 1998; Ohlstein and McKearin, 1997; Xie and Spradling, 1998), the somatic stem cells (SSC) are less well understood. Two to three SSC are located in a characteristic position within the germarium. Like germline stem cells, the SSC are thought to reside in a microenvironment, known as a niche, which provides signals that allow their maintenance and prevent their differentiation(Spradling et al., 2001; Xie and Spradling, 2000). These signals probably come from the cap and terminal filament cells, which are found at the anterior tip of the ovary, several cells distant from the SSC. These cells are thought to maintain the SSC by secreting Wingless (Wg)(Song and Xie, 2003). Hedgehog(Hh) is also involved in maintaining SSCs(Zhang and Kalderon,2001).

Although the progeny of the somatic stem cells ultimately receive a complex series of cues that cause them to differentiate into multiple spatially organized cell types, many of these differentiation events occur relatively late in egg chamber development. The early differentiation of follicle cells within the germarium is rather simple and involves the choice between two cell lineages: epithelial follicle cells (here termed `main body cells') that contact the developing cyst; and cells that comprise the precursors of the polar and stalk cells (here termed `polar/stalk precursors')(Larkin et al., 1996; Tworoger et al., 1999). Mutants that eliminate the formation of the polar and stalk cells, which results in the formation of fused egg chambers, have been identified(Grammont and Irvine, 2001; Lopez-Schier and St Johnston,2001); one mutant, eyes absent (eya),autonomously changes the fate of the main body cells into polar cells(Bai and Montell, 2002). However, it is not clear which signals limit EYA expression to main body cells.

The Notch (N) pathway is involved at multiple steps in the differentiation of follicle cells (Gonzalez-Reyes and St Johnston, 1998), including at mid-oogenesis, when main body follicle cells normally cease proliferation, differentiate and undergo several cycles of genomic endoreplication. Notch mutant follicle cells fail to differentiate and, instead of endoreplicating, they overproliferate(Deng et al., 2001; Lopez-Schier and St Johnston,2001). In addition, Notch appears to be required for the differentiation of the polar cells themselves(Grammont and Irvine, 2001). One important ligand for Notch signaling in the ovary is Delta, which is expressed in the germline at very low levels in the germarium(Deng et al., 2001; Lopez-Schier and St Johnston,2001) and at high levels at stage 6, when the main body cells differentiate (Bender et al.,1993). Delta signals through the Notch receptor, resulting in a presenilin-dependent cleavage with release of the intracellular domain(N-intra). N-intra binds to Suppressor of Hairless [Su(H)], the Drosophila equivalent of CBF1(Lopez-Schier and St Johnston,2001), and, together with Mastermind, activates transcription of downstream targets. In neural development, major downstream transcriptional targets of Notch are members of the Enhancer of Split family, which suppress transcription of Achaete and Scute (Van Doren et al., 1991). However, the Enhancer of Split complex does not appear to be required for follicle cell differentiation(Deng et al., 2001; Lopez-Schier and St Johnston,2001), and it is unclear what the functional downstream targets of Notch are in these cells. Achaete and Scute are members of a tissue-specific class of basic helix-loop-helix (bHLH) genes, which act with Daughterless, a ubiquitously distributed bHLH, to suppress transcription of Delta, and possibly other targets. Daughterless is involved in Notch signaling in the follicle cells (Cummings and Cronmiller,1994; Smith et al.,2002), but its bHLH partners have not been identified.

Extra macrochaetae (EMC) is a helix-loop-helix protein that is unable to activate transcription because it lacks a DNA-binding basic domain(Ellis et al., 1990; Garrell and Modolell, 1990). Instead, it inhibits the transcriptional activity of bHLH proteins by binding to them and sequestering them (Martinez et al., 1993; Van Doren et al.,1991; Van Doren et al.,1992). EMC is a transcriptional target of Notch in the wing(Baonza et al., 2000) and in the eye (Baonza and Freeman,2001). In addition to its functions in Notch signaling, EMC is required for cell proliferation (Baonza and Garcia-Bellido, 1999). The mammalian counterparts of EMC, the Id(Inhibitor of DNA binding, and Inhibitor of Differentiation) family, are involved in myriad differentiation events, primarily as inhibitors of differentiation. Id proteins interact with the cell cycle machinery(Lasorella et al., 2001; Ruzinova and Benezra, 2003; Zebedee and Hara, 2001) and are upregulated in a number of metastatic cancers(Duncan et al., 1992). Here we identify a role for EMC in follicle cell differentiation, and identify it as an effector of Notch signaling in the ovary.

Fly culture and crosses were performed according to standard procedures. Egg chamber stages are according to Spradling(Spradling, 1993). Wet yeast was added 1 day before all dissections. For the EP screen, flies from the Rørth EP collection (Rørth et al., 1998) were obtained from Exelixis (CA) or from the Szeged Drosophila Stock Centre(http://gen.bio.u-szeged.hu/gen),crossed either to C306-Gal4; PZ80/CyO or to PZ80/CyO;HS-Gal4 and grown at room temperature (or, if inviable at room temperature, at 18°C). PZ80(Karpen and Spradling, 1992)is an insertion into the non-essential Fas3 gene(Patel et al., 1987). C306-Gal4 (C306) drives expression in the germarium and in egg chambers (Manseau et al.,1997). Non-balancer F1 males and females were placed on wet yeast,and either heat shocked in a 37°C water bath for 1 hour(HS-Gal4), then placed at 25°C for 2 days, or placed at 29°C(C306) for at least 5 days with dry yeast. Three females were dissected from each line, fixed and stained for β-galactosidase activity(see below), examined under a dissecting microscope and scored (score = number of examples observed of fused egg chambers or of egg chambers with extra polar cells). EP candidates scoring ≥3 were crossed again to whichever Gal4 driver was not used in the first cross, and five to seven females were dissected and stained from each cross. Candidates scoring ≥1 on the second cross were crossed again to both drivers and to C306; A101/TM3[A101 (Bier et al.,1989) is an enhancer trap in the gene neuralized] and re-examined, and those that scored ≥3 on at least three out of five total tests were retained.

UAS-emc 5.3 and emc^P5C (an enhancer trap in emc) were obtained from A. Garcia-Bellido(Baonza et al., 2000; Ellis et al., 1990) and were either crossed to C306 in parallel with EP3620 (UAS-emc 5.3) or dissected as 2-day-old homozygotes(emc^P5C).

To examine EMC or EYA expression in FLP-OUT clones, EP3620 or UAS-NΔ34A[(Doherty et al., 1996)obtained from Ken Irvine] were crossed to hsFLP; ayGal4,P[w⁺; UAS-mCD8GFP]/TM3 [ayGal4(Ito et al., 1997) is a transgene in which Gal4 is expressed in clones of cells when FLP is induced by heat shock; expression of UAS-mCD8GFP(Lee and Luo, 1999) marks the clones; both were obtained from the Bloomington Stock Center(http://fly.bio.indiana.edu)]. Non-balancer females were heat-shocked in a water bath for 1 hour at 37°C,then placed at 25°C or 29°C with wet yeast for 2 or 3 days before dissecting.

For analysis of loss-of-function phenotypes, mutant clones were generated using the FLP/FRT system (Xu and Rubin,1993) and hs-FLP; adult females were heat shocked in a 37°C water bath twice in 1 day (4 to 6 hours between heat-shocks), and then placed on wet yeast at 25°C for 1 to 7 days. Clones were marked using Ubi-nlsGFP (emc and N)(Davis et al., 1995) or using anti-EYA antibody (eya^54C2)(Bonini et al., 1993). N^55E11, FRT18A/FM7(Grammont and Irvine, 2001)and emc¹, FRT80B/TM6b [a strong hypomorphic allele of emc (Ellis et al.,1990)] were obtained from Ken Irvine and Bloomington,respectively. eya^54C2, FRT40A was described previously(Bai and Montell, 2002). emc^AP6, FRT2A/TM6b [emc^AP6 is a null allele obtained from Pascale Heitzler(Ellis, 1994; Heitzler et al., 1996)] and emc^EP3620, FRT2A/TM6b were generated by standard methods. Clones of emc¹, FRT80B were generated in parallel with clones of N^55E11, FRT18A, and with clones of a wild-type FRT80B chromosome. They were analyzed in a PZ80-containing background 3 days after heat shock, since they exhibited no phenotype 2 days after heat shock. We were unable to observe clones of emc¹mutant cells more than 3 days after heat shock. Clones of emc^AP6, FRT2A, and of emc^EP3620, FRT2Awere analyzed in parallel with clones of a wild-type FRT2A chromosome and were found as many as 7 days after heat shock.

In order to examine EMC overexpression using Upd-Gal4, a driver expressed only in polar cells and their precursors(Bai and Montell, 2002)(obtained from Doug Harrison), Upd-Gal4; UAS-mCD8GFP/CyO was crossed to w; PZ80/CyO; EP3620/TM6b or UAS-LacZ or w; PZ80/CyO;UAS-mCD8GFP. Non-balancer flies were placed at 29°C for 3 to 5 days before dissection. This was modified for Upd-Gal4 temperature shifts:females and several males of the appropriate genotypes in a vial were fed with wet yeast overnight at 18°C, incubated in a 33°C water bath for 6 hours, then returned to 18°C with fresh wet yeast for 24 hours before dissecting, fixing and staining with anti-βGal antibody and 4,6-diamidino-2-phenylindole (DAPI). Each egg chamber from stages 4 through 9 was scored for the number of polar cells in the expected anterior and posterior polar cell groups.

Ovary dissections were performed in Grace's medium containing 10% fetal calf serum. For the EP screen, whole ovaries were fixed in a 96-well plate for 10 minutes in 4% formaldehyde (Polysciences, PA) in 0.1 mol/l potassium phosphate buffer. Fifty percent of the lines were then washed with NP-40 wash(50 mmol/l Tris pH 7.4, 150 mmol/l NaCl, 0.5% NP-40, 1 mg/ml bovine serum albumin) and stained overnight at 4°C with 1:1 7G10 anti-Fas3 antibody[Developmental Studies Hybridoma Bank (DSHB), University of Iowa, IA] and DAPI as described below. The remaining 50% of the lines were washed once in PBT(0.1 mol/l PBS + 0.5% Triton-X-100), then incubated overnight at 37°C with 0.2% X-GAL in staining solution (10 mmol/l phosphate buffer pH 7.2, 150 mmol/l NaCl, 1 mmol/l MgCl₂, 3 mmol/l K₄[FeII(CN)₆],3 mmol/l K₃[FeIII(CN)₆], 0.3% Triton-X-100). They were then incubated with 1:400 DAPI from 1 mg/ml stock (Sigma, MO) in PBT for 40 minutes at room temperature, and rinsed twice with PBT. Samples were mounted in 50% glycerol/PBT or examined in a depression slide.

For immunofluorescence, the following primary antibodies were used: mouse anti-Fas3 7G10 1:10 (DSHB); rabbit anti-β-galactosidase 1:400 (Cappel;NC); rabbit anti-EMC 1:1000 [a gift from Y. Jan(Baonza et al., 2000)]; mouse anti-EYA 10H6 (Bonini et al.,1993) 1:50 (DSHB); rabbit anti-Phosphohistone H3 1:1000 (Upstate Biotechnology, NY); mouse anti-Cyclin B 1:10 (DSHB). Ovarioles were fixed for 10 minutes at room temperature in 4% formaldehyde in 0.1 mmol/l phosphate buffer, rinsed twice in NP-40 wash, stained in the appropriate antibody diluted in NP-40 wash overnight at 4°C, washed for 1 hour at room temperature in NP-40 wash, and stained with appropriate secondary antibodies diluted in NP-40 wash for 1 hour at room temperature. Secondary antibodies used (Molecular Probes, OR) were Alexa 488 anti-rabbit, Alexa 488 anti-mouse,Alexa 568 anti-rabbit, Alexa 568 anti-mouse, Alexa 647 anti-rabbit and Alexa 647 anti-mouse, each at 1:400. DAPI was added to the secondary incubation at 1:400. After incubation with secondary antibody, samples were rinsed twice in NP-40 wash and mounted in Vectashield (Vector Laboratories, CA). Fluorescent images were captured on an Ultraview spinning disk confocal microscope, a Zeiss Axioplan, or a Zeiss Apotome.

In order to identify genes involved in cell fate decisions in the ovary, we performed a genetic screen using the EP system(Rørth, 1996). Briefly,this screen takes advantage of the ability of Gal4 protein to activate expression of genes located adjacent to upstream activating sequences (UAS sites). The EP is a P-element containing 14 UAS sites upstream of a basal promoter (Rørth, 1996). The Rørth EP collection contained about 2300 stocks, each with a random EP insertion into the genome, and therefore should separately drive expression of a wide variety of genes when crossed to a stock expressing Gal4. We crossed lines from the EP collection to Gal4 lines that can drive target gene expression in follicle cells in the germarium, and examined the ovaries for effects on polar cell formation or egg chamber packaging. EPs 3620 and 0415 produced egg chambers containing multiple germline cysts with high penetrance, indicating that they affected packaging(Fig. 2, compare A with B, and data not shown). These EPs are inserted upstream of the extra macrochaetae (emc) open reading frame and oriented in the correct direction to overexpress emc (Berkeley Drosophila Genome Project). Overexpression of EMC using a UAS-emc transgene(Baonza and Garcia-Bellido,1999) produced an indistinguishable phenotype(Fig. 2C).

In order to further understand the role of EMC, we examined its expression pattern (Fig. 3). An enhancer trap insertion into emc, emc^P5C, has previously been used to examine EMC expression in the wing(Baonza et al., 2000; Baonza and Garcia-Bellido,1999). We compared its expression with that of Fas3, a marker of follicle cell differentiation. Fas3 is not expressed in somatic cells in germarium region 1, but it is expressed in undifferentiated follicle cells (in germarium regions 2B and 3, and in early egg chambers), and in early polar and stalk cells (Ruohola et al.,1991). It is reduced in differentiated main body follicle cells and stalk cells but is expressed in polar cells throughout oogenesis. By contrast, EMC had a complex pattern of expression. emc^P5C-lacZ was expressed in the terminal filament and inner sheath cells of the germarium (Fig. 3A, arrows), and in the main body follicle cells of egg chambers(Fig. 3E), but not in the early follicle cells (Fig. 3A,bracket) or in polar or stalk cells (Fig. 3E, bracket). An anti-EMC antibody revealed additional features of EMC expression (Fig. 3B-D,F-H):EMC was expressed in the undifferentiated follicle cells of the germarium(Fig. 3D). Its expression was reduced in follicle cells in the intercyst regions between region 2 and region 3 cysts (Fig. 3C,D, blue arrowheads). It was maintained in main body cells but further reduced in stalk and polar cells of stage-1-4 egg chambers(Fig. 3C,D white and red arrowheads; Fig. 3F,G, arrows). Its expression in stalk cells remained low(Fig. 3F,G, arrowheads), but,in polar cells after stage 4, returned to a level indistinguishable from that exhibited by their neighbors (Fig. 3H, arrowheads). The overall level of expression was somewhat reduced from stage 6 to stage 8, and at stage 9 expression in the oocyte-associated follicle cells was further reduced, although still detectable (Fig. 3I, outside dashed lines, and data not shown). In order to confirm that EP3620overexpressed EMC, we also examined expression of EMC when expression of EP3620 was induced using a FLP-OUT cassette that only expresses Gal4 in GFP-marked clones of cells. As expected, induction of EP3620resulted in elevated expression of EMC(Fig. 3I, inside dashed lines)in a mosaic fashion in the Gal4-expressing cells.

The wild-type EMC expression pattern suggested that EMC might be involved in differentiation in the germarium and in main body cells. In order to examine this, we generated mutant clones with loss-of-function alleles of emc (Fig. 4). Small clones of emc (Fig. 4A,B) displayed elevated Fas3 expression, but control, wild-type clones did not (data not shown). This is similar to the phenotype of Notch mutant clones in main body cells(Fig. 4C). emc¹ clones in main body cells failed to express the polar-cell-specific enhancer trap PZ80(Fig. 4D-G, lines; compare with inset in G), indicating they did not become polar cells but were undifferentiated cells. Undifferentiated Notch mutant cells proliferate beyond the programmed cessation of proliferation at stage 6. In order to determine whether proliferation was still occurring in emcmutant cells, we examined expression of Cyclin B, a marker for cell proliferation, in follicle cells of egg chambers during stages 7 through 9. In wild-type follicle cells at these stages Cyclin B is not present because the cells have ceased cell division. Cyclin B was expressed in emc mutant clones (Fig. 4H-L, arrowheads in I-L: 5 out of 66 mutant clones).

Undifferentiated Notch mutant follicle cells are more densely distributed than their heterozygous neighbors(Lopez-Schier and St. Johnston,2001; Deng et al.,2001), because the extra cell divisions are not accompanied by additional cell growth; thus, the cells become smaller as they divide. Undifferentiated cells mutant for a null allele of emc(emc^AP6) are denser than their neighbors (area of mutant cell/area of wild-type cell = 1.8; n=10)(Fig. 4M,N, compare nuclei inside and outside dashed lines), indicating that these cells are unable to change from proliferation to endoreplication at the proper time.

We examined effects of emc mutations on development of polar and stalk cells. Notch signaling is required for differentiation of all the follicle cells, including main body, polar and stalk cells(Grammont and Irvine, 2001; Lopez-Schier and St Johnston,2001). As with Notch, egg chambers with emcmutant clones lacked polar cells at one end or the other(Fig. 4O,P, arrowheads),demonstrating a requirement for EMC in polar cell specification or differentiation. However, in contrast to Notch, the emcpolar cell phenotype was partially penetrant (31.3%: 21 out of 67 clones).

Notch mutant egg chambers can be fused to their neighbors(Grammont and Irvine, 2001; Lopez-Schier and St Johnston,2001). Likewise, fused emc mutant egg chambers could be found, possessing 32 germline nuclei (n=12). The presence of two oocyte nuclei (Fig. 4, compare Q and R, arrows) suggested that these egg chambers probably resulted from fusion of two distinct cysts rather than from an extra round of germline nuclear division. Fused egg chambers in ovarioles containing emcmutant clones, like those observed in ovarioles containing Notchmutant clones, possessed polar cells only at the termini of the fused cysts(Fig. 4 compare S and T,arrowheads), suggesting that fusion resulted from defects in polar cell differentiation or polar/stalk precursor specification, rather than from defects in stalk cell differentiation or morphogenesis.

If EMC affected polar cell precursor specification in addition to polar cell differentiation, then we would expect that clones generated in the germarium (i.e. those observed ≥4 days after heat shock) would produce fused egg chambers more frequently than those generated in existing egg chambers (i.e. those observed 1-2 days after heat shock). However, fused egg chambers were observed at the same low frequency in both cases. Thus, current evidence does not support a role for EMC in generating polar/stalk cell precursors.

EMC functions downstream of Notch in some tissues, and emc and Notch mutant clones share some phenotypic characteristics, so we examined whether expression of EMC in the main body cells depended upon Notch. EMC expression was greatly reduced in large Notch clones on the main body at stage 6 (n=36) (Fig. 5A,B). By contrast, EMC expression was normal in mutant clones of STAT, another gene required for follicle cell differentiation (data not shown).

We next examined the relationship between EMC and EYA. EYA is a negative regulator of both polar and stalk cell fate. Forced expression of EYA leads to elimination of both cell fates and thus to fused egg chambers, and loss of eya transforms main body cells into polar cells(Bai and Montell, 2002). EMC and EYA expression are similar, and EMC overexpression in the stalk leads to fused egg chambers, so we tested whether EMC was required for EYA expression. Consistent with this hypothesis, EYA was reduced in clones of emc on the main body (n=23; Fig. 5C,D). As expected, EMC was not reduced in eya clones(n=14; Fig. 5E,F). Since EMC regulates EYA and Notch regulates EMC, we examined EYA expression in Notch mutant cells. As expected, EYA was reduced in Notchmutant cells on the main body (n=23; Fig. 5G,H). We observed that in cells at the border of the clone, EYA was expressed but was cytoplasmic. This suggests local non-autonomous effects and thus that there is communication between follicle cells on the main body. In order to confirm that Notch promotes expression of EMC and EYA in main body follicle cells, we examined their expression in FLP-OUT Gal4 clones in which expression of activated Notch(UAS-NΔ34A) was forced(Fig. 5I-L). As EYA and EMC are expressed at high levels in main body cells until stage 8, we analyzed their expression only in clones of cells covering the oocyte at stage 10. These cells normally express EMC only at low levels and do not express EYA. We found a partially penetrant elevation in the expression of both EMC (11 out of 37 clones) (Fig. 5I,J, arrows and inside dashed lines) and EYA (29 out of 60 clones)(Fig. 5K,L, arrows).

The similarity between the expression patterns of EMC and EYA, the dependence of EYA on EMC in main body cells, and the formation of fused egg chambers when each is overexpressed, suggest that EMC is a regulator of EYA expression in all the follicle cells. Therefore we tested whether forcing EMC expression in the polar cells could induce EYA expression and affect polar cell formation. We generated FLP-OUT Gal4 clones of cells that expressed EMC. The polar cells were rarely targeted and out of several thousand clones observed, we obtained just six that resulted in fused egg chambers or the elimination of a single polar cell pair from one end of the egg chamber(Fig. 6A-D, arrows in Fig. 6A,C, and data not shown).

The low frequency of these effects precluded us from closely examining the effect of EMC overexpression on polar cell development using these drivers. Therefore, we used a Gal4 driver specific to the polar cells, Upd-Gal4 (Bai and Montell,2002; Silver and Montell,2001). When we drove expression of UAS-LacZ (data not shown) or UAS-GFP (Table 1), egg chambers appeared normal and polar cells developed normally. By contrast, when we drove UAS-emc, polar cells were absent from one end or other of about 20% of egg chambers(Table 1; Fig. 6G,H, arrowheads). Thus,expression of EMC can prevent polar cells from forming. In addition, the oocyte was sometimes mispositioned (Table 1; Fig. 6E,F,arrows) and egg chambers were sometimes round in shape at the anterior end(Table 1; compare Fig. 6G,H, arrowheads),phenotypes that are observed when polar cells are not formed(Grammont and Irvine,2002).

We expected that these phenotypes would occur if EMC affected EYA expression in polar cells, and, indeed, egg chambers in which polar cells were missing from one end exhibited EYA expression in all the cells at that end(Fig. 6H, arrowhead). In some cases, cells expressed both EYA and GFP, indicating that EMC was driving expression of EYA in those cells (Fig. 6I,K, arrowheads). Thus, EMC can force expression of EYA in the polar cells, and this affects their fate and function.

In addition to loss of polar cells, extra polar cells were sometimes formed when EMC was expressed either with Upd-Gal4(Table 1) or other drivers such as HS-gal4 (data not shown). In wild-type ovarioles, EMC is reduced in immature polar cells from stage 1 through stage 4 but is expressed in mature polar cells at stage 5 and above (see Fig. 3). This coincides with stages in which a reduction of polar cell number occurs in wild-type ovarioles: up to four or five polar cell precursors are observed in polar cell groups belonging to stage-2 and stage-3 egg chambers, but two to three of these precursor cells die upon maturation into polar cells at stage 4 or 5(Besse and Pret, 2003). We hypothesized that the ability of EMC to induce loss or gain of polar cells was a consequence of differing roles of EMC at different times during polar cell development. Specifically, we thought that EMC might be a negative regulator of polar cells in stage 1 through stage 4, and a positive regulator subsequent to this. In order to test this hypothesis, we took advantage of the temperature-dependence of Gal4 activity. We shifted flies expressing UAS-emc under the control of Upd-Gal4 to high (Gal4-active)temperature for 6 hours in order to express EMC in immature polar cells, and then down to low (Gal4-inactive) temperature for 24 hours to allow these cells to mature. We then determined in which egg chambers polar cells were lost, and in which egg chambers extra polar cells were formed. Based on the known durations of various stages (Spradling,1993), we can estimate at what stage the egg chambers expressed EMC, and thus determine at what stages polar cell development was inhibited or enhanced by EMC. Egg chambers that were in stage 2 or 3 during EMC expression,and were in stage 4-8 when they were analyzed, showed more frequent loss of polar cells (Fig. 7A). By contrast, extra polar cells were observed most frequently in egg chambers that were at stage 9 at analysis (Fig. 7B), which were at stage 5 during EMC expression. Thus, UAS-emc inhibits polar cell development at stages 2 and 3, but promotes it at stage 5.

In this work, we demonstrated a number of similarities in the phenotypic effects of Notch and EMC in egg chamber development, and we conclude that EMC is a downstream effector of Notch in the differentiation of follicle cells. In other tissues, genes of the Enhancer of split complex are major downstream effectors of Notch signaling, but they do not appear to be required for differentiation of the follicle cells(Lopez-Schier and St Johnston,2001). By contrast, EMC is expressed in most somatic cells of the ovariole and has a loss-of-function phenotype similar to that of Notch. Like Notch, EMC is required for the formation of polar cells, and we observe egg chambers lacking polar cells at one end when emc clones are generated. emc mutant cells also fail to downregulate Fas3 expression at stage 6, and the cells fail to enter endoreplication at the proper time,indicating that emc is required for follicle cell differentiation. In addition, EMC expression in the main body follicle cells is dependent on Notch and can be induced by forced expression of activated Notch. Taken together,this leads us to propose that EMC is an effector of Notch activity in the follicle cells.

Notch and EMC are both involved in separation of adjacent egg chambers. Follicles mutant for Notch or emc include egg chambers containing multiple germline cysts and lacking intervening polar and stalk cells. Recent work has shown that Notch signaling is not required for early packaging (Torres et al.,2003), but its role in specifying polar/stalk precursors, which might occur subsequent to packaging, has not been addressed specifically. However, Notch is required autonomously for polar cell differentiation(Grammont and Irvine, 2001; Lopez-Schier and St Johnston,2001), and at least one pair of differentiated polar cells is required to induce a stalk. Therefore, the packaging defects observed in Notch mutants may be due solely to either the failure of polar cells to differentiate or to a particular role in specifying polar/stalk precursors. The fused egg chambers we observed in emc mutants could also be due to a role in one or both processes. If emc were involved separately in both processes, then we would expect to see a dramatic increase in the frequency of fused egg chambers when clones were generated in the germarium,where precursors are formed. Since we did not see such an increase, we do not posit a role for emc in polar/stalk precursor formation; however,without a better understanding of polar/stalk precursors, this cannot be confirmed.

emc is probably not the only effector downstream of Notchin the ovariole, since emc mutant clones exhibit some differences from Notch mutant clones. In general the effects of emc were less penetrant than those of Notch. For example, emc mutant cells, like Notch mutant cells, exhibited persistent Phosphohistone H3 (data not shown) and Cyclin B labeling but at low frequency and at low levels. In addition, although emc is involved in polar cell specification or differentiation, it is only partially required for this,since normal polar cells could be observed even in clones of a null emc allele. Thus, there are likely to be additional downstream effectors of Notch in the ovary.

Molecular epistasis indicates that Notch signals through EMC to induce or maintain EYA expression in the main body follicle cells. EYA expression is lost in large Notch mutant follicle cell clones and can be induced by forced expression of activated Notch in the follicle cells. EYA is involved in inhibiting polar and stalk cell fate, so one might expect that Notchor emc loss-of-function mutants would make extra polar cells. This is not the case, however; it seems likely that the reason Notch and emc mutant cells do not become polar cells is that they fail to differentiate. EYA does not appear to be required for differentiation, but rather for main body cell fate, since eya mutant cells differentiate into polar cells (Bai and Montell,2002). Thus, the Notch pathway must branch downstream of EMC, with one pathway leading to EYA expression and repression of polar cell fate, and a separate pathway leading to differentiation(Fig. 8).

EMC can have both positive and negative effects on the number of polar cells. Although this might initially seem mysterious, it can be explained by the dynamic expression of EMC in ovaries. EMC is expressed in the germarium,but it is reduced in polar/stalk precursors. Its expression remains low in stalk cells but, in polar cells, returns to the same level as that of their neighbors around the time of differentiation. Our temperature-shift experiments show that forced expression of EMC in immature polar cells can lead to expression of EYA, which is the presumed cause of loss of polar cells. By contrast, forced expression of EMC in maturing polar cells appears to lead to potentiation of polar cell number, i.e. polar cell number per group is not reduced from four to two, as in wild-type, but remains at four polar cells per group. This is probably due to a role for EMC in polar cell differentiation,because loss-of-function emc clones can result in loss of polar cells.

Three lines of evidence suggest that EMC may be a key regulator of EYA expression. First, EMC and EYA expression are similar in multiple respects. Both are upregulated in follicle cells in region 2B of the germarium, and in main body follicle cells from stages 2 through 6. Both are downregulated in the polar/stalk lineage from the germarium through stage 3, and in the oocyte-associated follicle cells at stages 7 through 9. Second, EMC is required for EYA expression in the main body. Third, EMC and EYA produce similar overexpression phenotypes, including fused egg chambers and the loss of polar cells, and, when EMC is overexpressed in the polar cells, EYA expression is induced. Taken together, these data suggest that the expression of EYA in the follicle cells is largely regulated by EMC. EMC is a helix-loop-helix protein that lacks the basic DNA-binding domain of the bHLH transcription factors. It normally opposes the activity of bHLH transcription factors by sequestering them in non-productive complexes. Thus, the dependence of EYA on EMC is likely to be indirect. Presumably, EMC inhibits a bHLH protein that inhibits expression of EYA. The identity of this protein remains an interesting subject for further study.

We thank Tina Bridges and Tanika Harris for help with the latter part of the screen; Ken Irvine, Antonio Garcia-Bellido, James Posakony, Doug Harrison,Pascale Heitzler, Exelixis and the Szeged and Bloomington stock centers for fly stocks; Yuh-Nung Jan for anti-EMC antibody; the Developmental Studies Hybridoma Bank for antibodies; members of the laboratory for valuable discussions; and Michelle Starz-Gaiano and Jocelyn McDonald for comments on the manuscript. This work was supported by NIH grant R01-GM 46425.