The p27^cip/kip ortholog dacapo maintains theDrosophila oocyte in prophase of meiosis I

During metazoan development, cells alter their cell cycles in response to specific developmental cues. Often these signals result in a change in proliferative state, prompting a cell either to divide in response to a growth signal or to exit the cell cycle in order to begin the process of differentiation. Occasionally, however, cells stray from the canonical mitotic cycle, in which DNA synthesis is followed by chromosome segregation, in order to fulfill a specific biological function. Two common examples of this phenomenon occur during Drosophila oogenesis when cells within a germline cyst undergo either the reductional divisions of meiosis and produce a functional gamete or enter the endocycle and develop as a highly polyploid nurse cell.

Ovarian cyst formation begins when one, out of a group of two to three germline stem cells, divides asymmetrically to produce a cystoblast. The cystoblast undergoes a series of four synchronous mitotic divisions in which cytokinesis is incomplete (de Cuevas et al., 1997; McKearin,1997). Individual cells within the cyst, referred to as cystocytes, are connected in an invariant pattern by actin-rich intercellular bridges called ring canals (Fig. 1A). After all 16 cystocytes complete premeiotic S phase, a meiotic gradient develops with the two cells with four ring canals having the most meiotic features as assayed by the presence of mature synaptonemal complexes and recombination nodules(Chandley, 1966;Carpenter, 1975;Carpenter, 1981;Schmekel et al., 1993). Ultimately, the cell cycles of the future nurse cells and oocyte diverge dramatically. The single pro-oocyte arrests in prophase of meiosis I for most of the growth phase of oogenesis, while the 15 nurse cells go on to complete 10-12 endocycles to become highly polyploid. During these endocycles the nurse cells cycle asynchronously.

How one of the 16 cystocytes is selected to differentiate as the oocyte has long been an issue of interest. It has been known for some time that one of the two cells with four ring canals always develops as the oocyte (reviewed byBüning, 1994). This observation indicates that the invariant pattern of cystocyte connections is crucial for later cell fate decisions within the cyst (reviewed byde Cuevas et al., 1997;McKearin, 1997). Studies over the last few years suggest that oocyte differentiation is a two-step process. The first step entails the establishment of an asymmetry between the pro-oocyte and the pro-nurse cells. Exactly when and how this asymmetry is established is unknown, although it may occur as early as the cystoblast division through the unequal distribution of the fusome, a germline-specific organelle (Lin and Spradling,1995; de Cuevas and Spradling,1998). The second step involves the directional transport of cellular components, including specific mRNAs and proteins, to the pro-oocyte. Drugs that destabilize microtubules eliminate the accumulation of specific mRNAs in the pro-oocyte and lead to the production of cysts that contain 16 nurse cells and no oocyte (Koch and Spitzer, 1985; Theurkauf et al., 1993). Similarly, recessive mutations in the genesBicaudal D (BicD) and egalitarian (egl)prevent the differential accumulation of specific mRNAs and proteins in the pro-oocyte and result in cysts with 16 nurse cells(Suter et al., 1989;Schupbach and Wieschaus, 1991;Suter and Steward, 1991;Mach and Lehmann, 1997). Although it is clear that microtubule-based polarized transport is a critical component of oocyte differentiation, exactly why meiosis proceeds in the oocyte while the adjacent nurse cells enter the endocycle remains undefined.

During meiosis, the integrity of the genome must be maintained through recombination and two rounds of cell division. Animals in which the oocyte develops within the context of a cyst, such as Drosophila, are faced with an additional set of challenges. Intercellular bridges physically connect the Drosophila oocyte to nurse cells that are undergoing repeated rounds of DNA replication. However, to produce a functional gamete, it is essential that the oocyte does not fire a single DNA replication origin. How does the germline cyst maintain this extreme cell cycle dichotomy? One potential mechanism involves the spatial regulation of cyclinE-Cdk2 activity(Lilly and Spradling, 1996). In Drosophila, cyclinE-Cdk2 activity is required for S phase(Knoblich et al., 1994).dacapo (dap) is a vital gene that encodes ap21^CIP/p27^KIP1/p57^KIP2-like cki that specifically inhibits the activity of cyclinE-Cdk2 complexes(de Nooij et al., 1996;Lane et al., 1996). Throughout much of the growth phase of Drosophila oogenesis, the levels of the cki Dap oscillate in the 15-polyploid nurse cells but remain persistently high in the single oocyte (de Nooij et al.,2000). We reasoned that the differential regulation of Dap might provide a mechanism to maintain the oocyte in prophase of meiosis I, while allowing the endocycle to proceed in the adjacent nurse cells.

The FRT42B, dap⁴(Lane et al., 1996) stock was a gift of Iswar Hariharan. The P[mini-w; arm—LacZ](Lecuit and Cohen, 1997) stock was provided by Steve Cohen. The cycE⁰¹⁶⁷² mutation was recovered from a single P-element mutagenesis(Karpen and Spradling, 1992). All additional stocks were obtained from the Bloomington Stock Center. ThecycE⁰¹⁶⁷², dap⁴ chromosome was generated by meiotic recombination.

Germline clones were generated using the FLP:FRT system(Chou and Perrimon, 1996). Specifically, y, w, FLP12; FRT42B arm-lacZ females were crossed tow; FRT42B, dap⁴/CyO males. Eggs produced from this cross were heat-shocked for 1 hour at 37°C in a circulating water bath 96 hours after egg deposition. A second identical heat shock was administered 24 hours later. After eclosion y, w, FLP12; FRT 42B arm-LacZ/FRT42B, dap⁴ adult females were collected and heat-shocked at least once for 1 hour at 37°C. Females were allowed to recover for 8 days at 25°C. Clones were negatively marked by the absence of β-galactosidase protein as determined by immunocytochemistry.

Ovaries were dissected, immunologically stained and mounted as described previously (Lin et al., 1992). Dap monoclonal antibody (de Nooij et al.,2000) was provided by Iswar Hariharan. Dap rabbit polyclonal antibody was provided by Christian Lehner. Orb monoclonal antibody(Lantz et al., 1994) was provided by the Developmental Studies Hybridoma Bank. BicD monoclonal antibody(Junyoung and Steward, 2001)was provided by Ruth Steward. Ovaries were fixed in methanol/EGTA for tubulin staining. α-Tubulin monoclonal antibody DM1a (Sigma) was used at 1:500. BrdU staining was essentially as described by Avedisov et al.(Avedisov et al., 2000). C(3)G antibody was generated from 301 amino acid C-terminal fragment in rabbits using standard techniques. The C(3)G antibody was used at 1:3000 dilution.

In order to determine the precise onset of the differential regulation of Dap in the nurse cells versus the oocyte, we stained wild-type ovaries with an antibody against Dap (de Nooij et al.,2000). Dap levels oscillate during the four mitotic cyst divisions in region 1 of the germarium (Fig. 2A). In region 2, a meiotic gradient forms that emanates from the pro-oocytes, the two cells with four ring canals(Carpenter, 1975;Carpenter, 1979;Huynh and St Johnston, 2000;Page and Hawley, 2001). Although there is a clear gradient with respect to meiosis, this is not reflected in the distribution of the Dap protein. In young post-mitotic cysts,in region 2a and early 2b, Dap is at uniformly low levels throughout the cyst. However, just prior to when the nurse cells enter the endocycle, in late region 2b/early region 3 (stage 1), Dap levels rise dramatically throughout the cyst (Fig. 2A,B). Thus,uniformly high levels of Dap are observed in all cyst nuclei in early stage 1 egg chambers, which contain a single oocyte in late prophase of meiosis I and 15 nurse cells that are poised to enter the first endocycle S phase.

The differential behavior of Dap in the nurse cells versus the oocyte is first observed in late stage 1 egg chambers. Although high levels of Dap persist in the oocyte, the levels of Dap begin to oscillate in the nurse cells as they asynchronously enter the endocycle(Fig. 2C,D)(de Nooij et al., 2000). Dap levels continue to oscillate in the nurse cells until stage 10 of oogenesis when the nurse cells stop replicating their DNA(Fig. 2F). By contrast, Dap levels remain high in the oocyte until well after the nurse cells exit the endocycle. Dap inhibits DNA replication in both mitotic and endocycling cells(de Nooij et al., 1996;Lane et al., 1996;Calvi et al., 1998). Thus, it is likely that the cycling of Dap protein in the nurse cells allows cyclinE-Cdk2 kinase activity to rise high enough to trigger the endocycle S phases (de Nooij et al., 2000;Edgar and Orr-Weaver, 2001). Consistent with this proposal, S phase in the nurse cells occurs when Dap levels are low (Fig. 2G,H). Therefore, as the nurse cells enter the endocycle in stage 1 there is a clear correlation between the differential regulation of Dap and the distinct cell cycles of the nurse cells and oocyte.

Does the cki Dap prevent the oocyte from entering the endocycle with the nurse cells and thus preserve the prophase I meiotic arrest? To answer this question, we generated homozygous germline clones of the dap null allele dap⁴ (Lane et al., 1996). dap⁴ contains a deletion of the conserved CDK-binding domain and acts as a complete loss-of-function allele(Lane et al., 1996). In greater than 80% of the egg chambers (n>250) that containdap germline clones, the oocyte enters the endocycle and becomes polyploid (Fig. 3). The extent of polyploidy in these oocytes is variable. 53±9% of dap cysts have 16-polyploid nurse cells and no oocyte(Fig. 3B). In this phenotypic class, all cells in the cyst have similar DNA contents, indicating all 16 cells have undergone approximately the same number of endocycles. These data suggest that in greater than 50% of the egg chambers with dapgermline clones, the oocyte enters the endocycle at the same time as the nurse cells. In 28±7% of the germline clones the oocyte is polyploid, but can be distinguished from the adjacent nurse cells by its lower DNA content and posterior position (Fig. 3C). In wild-type egg chambers the oocyte DNA condenses into a compact karyosome in stage 3 of oogenesis (Fig. 3A). In the 10±6% of dap clones in which the oocyte is not obviously polyploid, karyosome formation is often aberrant. For example, the oocyte DNA is present in one or more elongated masses within the oocyte nucleus (Fig. 3D) or in a thin rim near the nuclear envelope (data not shown). The ∼9% of dapclones that contained greater than 16 nurse cells were not included in this analysis. These data indicate that dap regulates entry into or is required for the maintenance of the meiotic cycle in the oocyte.

What accounts for the observed variability of the dap phenotype?Because we examined the clonal progeny from germline stem cell clones that had undergone numerous divisions, we do not believe the observed phenotypic variability results from the perdurance of the Dap protein. In support of this conclusion, immunocytochemistry using an antibody against Dap indicated that clonal mutant egg chambers contain no Dap protein (data not shown). In addition, a similar phenotypic distribution was observed when examining the ovaries from dap⁴ homozygous escapers. Finally,dap germline cysts surrounded by dap⁴(n=88 egg chambers) and dap⁺ (n=104 egg chambers) follicle cells had similarly variable phenotypes, indicating that the genotype of the follicle cells is not the source of variability. Protecting the oocyte from inappropriate entry into the endocycle is critical to the production of a functional gamete. Therefore, we predict that additional factors act in concert with Dap to inhibit DNA replication during meiosis.

Mutations that disrupt cyst polarity and oocyte specification, such asBicD and egl, result in the failure to localize oocyte specific markers and the production of egg chambers with 16 nurse cells and no oocyte (Suter et al., 1989;Suter and Steward, 1991;Christerson and McKearin, 1994;Lantz et al., 1994). We wanted to determine if mutations in dap indirectly influence the oocyte cell cycle by disrupting cyst polarity. In wild-type cysts, the Orb and BicD proteins preferentially accumulate in the cytoplasm of the pro-oocyte,beginning in region 2a of the germarium(Suter and Steward, 1991;Lantz et al., 1994). With few exceptions, the distribution of the Orb(Fig. 4A) and BicD (data not shown) proteins in dap cysts is indistinguishable from wild type in region 2 of the germarium (Fig. 4A). As shown in Fig. 4, Orb protein preferentially accumulates in a single centrally located cystocyte in both wild-type (arrow) and dap (arrowhead)cysts. These data indicate that unlike what is observed in BicD andegl mutants, the polyploidization of dap oocytes is not the consequence of a failure to specify an oocyte or a general disruption in cyst polarity. However, it should be noted that in egl mutants, Dap protein does not preferentially accumulate within a single cell in post-germarial egg chambers, but instead appears to oscillate in all 16-cyst cells as they undergo repeated endocycles (data not shown). Thus, althoughdap is not required for oocyte specification, the specification of the oocyte is required to establish the two modes of Dap regulation within ovarian cysts. These data indicate that oocyte specification is upstream of the effects of Dap on the meiotic cycle.

In the absence of Dap the majority of oocytes do not undergo the two meiotic divisions but instead enter the endocycle and replicate their DNA. Next, we wanted to examine if Dap is required for the initiation and/or the maintenance of meiosis? In order to determine if dap oocytes enter meiosis, we assayed meiotic progression using an antibody directed against the synaptonemal complex (SC) component C(3)G(Page and Hawley, 2001). In wild-type cysts, up to four cells construct SC with the two pro-oocytes entering pachytene (Carpenter,1975; Carpenter,1979; Huynh and St Johnston,2000; Page and Hawley,2001) (Fig. 4B). By late region 2b, the meiotic gradient sharpens such that C(3)G staining is primarily concentrated in the oocyte. dap oocytes progress to pachytene as assayed by the production of continuous SC along bivalents(Fig. 4C). These data indicate that dap is not necessary to initiate the meiotic cycle nor is it necessary for progression to pachytene. Therefore, the first disruption to the meiotic cycle observed in dap oocytes is when they inappropriately enter the endocycle in stage 1. These data are consistent with Dap acting to inhibit DNA replication specifically in the oocyte as the adjacent nurse cells enter the endocycle.

In dap germline cysts there is a delay in the restriction of SC to a single cell such that cysts in region 2b and 3 occasionally have 2 or more cells with strong C(3)G staining (data not shown). We believe it is unlikely that the retention of SC in the adjacent nurse cells is responsible for the subsequent polyploidization of the oocyte observed in dap mutants. The spindle mutants, which activate the meiotic checkpoint(Ghabrial et al., 1998), show a very similar delay in the restriction of SC to the oocyte(Huynh and St Johnston, 2000). Importantly, mutations in the spindle genes do not result in the polyploidization of the oocyte. We believe the delay in the restriction of SC to a single cell may represent an earlier function of dap, perhaps during the mitotic cyst divisions or in the regulation of premeiotic S phase.

In stage 1 egg chambers the cell-cycle environment within the cyst changes dramatically as the nurse cells asynchronously enter the S phase of the first endocycle. In wild-type stage 1 egg chambers, the oocyte remains safely arrested in prophase of meiosis I. However, in the majority of dapmutant cysts, the oocyte enters the endocycle with the nurse cells. Our data indicate that inappropriate entry into the endocycle disrupts oocyte differentiation (Fig. 5). Beginning at stage 1, we observe a decrease, relative to similarly aged wild-type egg chambers, in the preferential accumulation of the BicD and Orb proteins in dap oocytes. As oogenesis progresses, there is a strong inverse correlation between the degree of polyploidization in dapoocytes and the preferential accumulation of BicD and Orb. dap cysts with highly polyploid oocytes have little to no preferential accumulation of BicD (Fig. 5C,D) or Orb (data not shown). By contrast, dap cysts in which the oocyte has undergone limited polyploidization frequently have BicD and Orb levels indistinguishable from wild type (Fig. 5E,F). As is observed in other mutants that disrupt oocyte differentiation, egg chambers that contain dap clones rarely develop beyond stage 6(Suter et al., 1989;Clark and McKearin, 1996;Mach and Lehmann, 1997;Huynh et al., 2001). InDrosophila vitellogenesis begins in stage 7 when the oocyte begins to take up large quantities of yolk. The small percentage of dap oocytes that progress far enough to take up yolk, invariably have undergone little to no polyploidization (data not shown). These data demonstrate that dapis required for oocyte differentiation. In addition, they indicate that the loss of oocyte identity observed in dap clones is a direct consequence of the oocyte entering the endocycle.

To explore further the apparent loss of oocyte identity that accompanies inappropriate entry into the endocycle, we examined the distribution of microtubules in dap cysts. The preferential accumulation of Orb and BicD in the oocyte is dependent on a polarized network of microtubules that directs these, and other oocyte-specific factors, from the nurse cells to the oocyte (Theurkauf et al.,1993). The disruption of this network by microtubule depolymerizing agents leads to the production of egg chambers with 16 polyploid nurse cells (Koch and Spitzer,1985; Theurkauf et al.,1993; Huynh and St Johnston,2000). In wild-type cysts, the asymmetric distribution of microtubules within the germline cyst can be visualized as a preferential accumulation of α-tubulin in the single oocyte, which contains the microtubule-organizing center (Theurkauf et al., 1992). In dap cysts, this focus ofα-tubulin staining is present in nonpolyploid oocytes but absent in polyploid oocytes (Fig. 6). These data suggest that entry into the endocycle may disrupt the polarized microtubule network, which in turn blocks oocyte differentiation. However, the exact relationship between entry into the endocycle and the disruption of the microtubule network remains undefined.

In females homozygous for the hypomorphic mutationcycE⁰¹⁶⁷², a fraction of egg chambers contain two cells that have oocyte-like nuclear features, such as low ploidy values, an endobody and a small DNA mass in a very large nucleus(Lilly and Spradling, 1996). Egg chambers that contain two oocyte nuclei have only 14 polyploid nurse cells, indicating that a cell that was destined to develop as a nurse cell has been partially transformed towards the oocyte fate. The extra oocyte nucleus,which can be distinguished from the true oocyte by its presence in a small cell that lacked signs of cytoplasmic oocyte differentiation, almost invariably is the other four-ring canal cell in the cyst(Lilly and Spradling, 1996). Interestingly, these transformed nuclei accumulate persistently high levels of Dap protein in a manner similar to the true oocyte(Fig. 7A,B). Cells with persistently high levels of Dap have low ploidy values, indicating they have either not entered the endocycle or have prematurely exited the cycle. By contrast, in wild-type egg chambers the other four-ring canal cell develops as a highly polyploid posterior nurse cell in which Dap levels oscillate. Thus,mutations in dap result in germline cysts in which all 16 cells enter the endocycle and develop as nurse cells, while a mutation in cycEhas the opposite effect, resulting in two or more cells that have persistently high levels of Dap that cannot enter and/or maintain the endocycle.

To examine the relationship between cycE and dap in the regulation of the cell-cycle program of ovarian cysts, we examined if mutations in dap could dominantly modify thecycE⁰¹⁶⁷² two oocyte phenotype. We found that reducing the dose of the dap gene by half, resulted in ∼2.5 fold suppression of the cycE⁰¹⁶⁷² two oocyte phenotype. In wild-type egg chambers the four posterior nurse cells, which are connected to the oocyte via ring canals, have the highest ploidy values in the cyst(Fig. 7C). IncycE⁰¹⁶⁷² females 37±7% (n>200) of egg chambers contain a cell adjacent to the true oocyte with inappropriately low ploidy values (Fig. 7D,E). When a single copy of the null allele dap⁴ was placed in thecycE⁰¹⁶⁷² background, fewer than 14±7%(n>200) of egg chambers had a posterior nurse cell with a reduced DNA content. These data indicate that whether a cyst cell enters and/or maintains the endocycle is at least partially determined by the balance of CycE and Dap. In addition, they strongly suggest that, as is observed during embryogenesis, the primary target of Dap in the ovary is the CycE/Cdk2 complex.

In order to fulfill their specific biological functions, the nurse cells and the oocyte of Drosophila ovarian cysts maintain very different cell cycles throughout much of oogenesis. The oocyte faithfully executes the meiotic cycle in order to produce the genetic material for the egg. By contrast, the biosynthetically active nurse cells enter the endocycle and become highly polyploid. During oogenesis, these apparently incompatible cell cycles must co-exist for several days within the shared cytoplasm of the cyst. The studies presented here address how DNA replication is inhibited in the meiotic oocyte, while simultaneously promoted in the adjacent nurse cells. We demonstrate that this is accomplished through the differential regulation of the cki dap. Specifically, dap is required to maintain the oocyte in prophase of meiosis I during the growth phase of Drosophilaoogenesis. In dap mutants, the majority of oocytes enter the endocycle with the nurse cells. Furthermore, our data indicate that inappropriate entry into the endocycle inhibits oocyte differentiation and promotes the nurse cell developmental pathway.

Our observations suggest a model for how the meiotic cycle and the endocycle are independently maintained within Drosophila ovarian cysts (Fig. 8A). We propose that the presence of high levels of the cki Dap in the oocyte, throughout the time the nurse cells are in the endocycle, persistently inhibits cyclinE-Cdk2 kinase activity and prevents inappropriate DNA replication during meiosis. Without the inhibition of cyclinE-Cdk2 kinase activity provided by high levels of Dap, the majority of dap mutant oocytes abandon the meiotic cycle and enter the endocycle with the nurse cells. Importantly, these data indicate that, as has recently been observed in mice, oocytes in prophase of meiosis I are competent to replicate their DNA(Czolowska and Borsuk, 2000). In the mouse oocyte, the inhibition of DNA replication during prophase of meiosis I may be accomplished through the downregulation of the G1 cyclins and Cdk2 (Moore et al., 1996;Czolowska and Borsuk, 2000). In the Drosophila oocyte, the inhibition of cyclinE-Cdk2 activity by Dap achieves the same aim. In contrast to the oocyte, the nurse cells require a period when Dap levels are low to allow cyclinE-Cdk2 kinase activity to rise high enough to trigger each endocycle S phase. These low points occur during the oscillations of the Dap protein. Our data indicate that it is through the differential regulation of Dap that two apparently incompatible cell cycles are stably maintained within the common cytoplasm of the ovarian cyst.

The regulatory relationship between cyclinE-Cdk2 activity and the Dap ortholog p27 suggest a feedback loop that may account for the long-term stabilization of Dap in post germarial oocytes. In mammalian cells,phosphorylation by cyclinE-Cdk2 targets the p27 protein for destruction by the proteasome (Pagano et al.,1995; Sheaff et al.,1997; Vlach et al.,1997; Montagnoli et al.,1999). Similarly, the Dap protein contains a CDK phosphorylation consensus site (Ser²⁰⁵) and can be phosphorylated by mammalian cyclinE-Cdk2 in vitro (de Nooij et al.,1996). We propose that in early stage 1 egg chambers, the balance of cyclinE-Cdk2 activity and Dap protein is slightly different in the 15-nurse cells versus the single oocyte (Fig. 8B). In the oocyte, the balance is tipped towards the inhibitor Dap, resulting in diminished cyclinE-Cdk2 activity. Lower cyclinE-Cdk2 activity leads to a reduced rate of Dap phosphorylation and proteolysis,thereby increasing the concentration of the Dap protein. The stabilization of the Dap protein ultimately results in the permanent inhibition of cyclinE-Cdk2 activity in the oocyte. In contrast to the oocyte, in stage 1 nurse cells cyclinE-Cdk2 kinase activity reaches high enough levels to trigger the phosphorylation and subsequent destruction of the Dap protein, thus allowing endocycle progression. The above model predicts that additional proteins that are targeted for destruction by cyclinE-Cdk2 phosphorylation should be stabilized in the oocyte but not in the nurse cells. Like p27, the proteolytic destruction of CycE itself is dependent on phosphorylation by the cyclinE-Cdk2 complex (Clurman et al., 1996;Won and Reed, 1996). As predicted by the model, CycE is stabilized in the oocyte and accumulates to high levels as oogenesis progresses (Lilly and Spradling, 1996). This model allows a slight difference in the balance of cyclinE-Cdk2 activity and Dap early in oogenesis to be amplified,resulting in the two cell types of the germline cyst permanently adopting dramatically different cell cycles.

Consistent with the above model, removing one copy of dapdominantly inhibits the two oocyte phenotype observed in thecycE⁰¹⁶⁷² hypomorph. In cycE⁰¹⁶⁷²females, Dap protein frequently accumulates to high levels in one or more cyst nuclei in addition to the true oocyte. The cells that inappropriately stabilize Dap have low DNA contents and are connected to the true oocyte through a ring canal. In effect, it appears that in thecycE⁰¹⁶⁷² mutants, the stabilization of Dap and the accompanying inhibition of the endocycle that is normally restricted to the oocyte is allowed to spread outwards to the adjacent nurse cells. Thus,reducing CycE levels favors the stabilization of Dap, which leads to the inappropriate inhibition of the endocycle in cells connected to the true oocyte. Importantly, we demonstrate that this phenotype is suppressed by removing a single copy of the dap gene from thecycE⁰¹⁶⁷² background. Thus, by bringing CycE and Dap levels back into balance, the nurse cells near the oocyte are no longer inappropriately drawn into the feedback loop that inhibits the endocycle.

However, an important question remains. Beyond the specification of the oocyte, what molecular pathway accounts for the proposed difference in the balance of cyclinE-Cdk2 activity and Dap in nurse cells relative to the oocyte in early stage 1 egg chambers? There are at least three possible mechanisms,not mutually exclusive, that may explain this initial asymmetry. Intriguingly,dap mRNA is transported to the oocyte beginning in region 2 of the germarium and remains at high levels in the oocyte throughout oogenesis(de Nooij et al., 2000). Thus,the oocyte may simply be able to translate more Dap protein and therefore keep Dap levels slightly higher in the oocyte. This slight difference may be below the resolution of immunocytochemistry, thus explaining why Dap appears to be evenly distributed in early stage 1 egg chambers. Alternatively, cyclinE-Cdk2 activity may be partially inhibited in stage 1 oocytes via a Dap-independent mechanism. This possibility is supported by the observation that even in the complete absence of Dap, not all oocytes enter the endocycle, indicating that oocytes have an alternative mechanism to inhibit inappropriate DNA replication. Finally, Dap proteolysis may be less efficient in the oocyte for reasons independent of the level of cyclinE-Cdk2 activity, such as a general inhibition of the proteasome. The identification of additional genes that influence the maintenance of the meiotic cycle during oogenesis will help distinguish between the above possibilities.

Studies over the last few years indicate that the differentiation of both the nurse cells and oocyte are strongly influenced by cell-cycle events within the germline cyst (Lilly and Spradling,1996; Ghabrial et al.,1998; Ghabrial and Schupbach,1999; Myster et al.,2000; Page et al.,2000). Our data support the conclusion that cell-cycle regulation and oocyte differentiation are closely coupled. Specifically, we find that inappropriate entry into the endocycle disrupts oocyte differentiation indap germline cysts and results in the presumptive oocyte developing like a nurse cell. The more polyploid the oocyte, the greater the disruption in oocyte differentiation as measured by the oocyte-specific accumulation of the proteins BicD and Orb, as well as yolk uptake. In mutants likedap, where the oocyte is specified but ultimately enters the endocycle, cause and effect can be difficult to determine. Does the oocyte enter the endocycle because of the inability to maintain the oocyte fate or alternatively does entry into the endocycle disrupt the ability to maintain the oocyte identity? Considering the known role of dap in cell-cycle regulation, we believe that the phenotype observed in dap mutants reflects the second scenario. Our data indicate that entry into the endocycle is incompatible with many aspects of the oocyte developmental program and can serve as the primary cause of the loss of oocyte identity. The identification of additional genes that influence the maintenance of both the nurse cell and oocyte identities will help clarify the exact role of cell-cycle programming in the nurse cell/oocyte fate decision of Drosophila ovarian cysts.

We thank Iswar Hariharan, Christian Lehner, Steve Cohen, Scott Hawley,Dennis McKearin and Ruth Steward for providing stocks and antibodies. We thank Nicole Greider, Maggie de Cuevas, Brian Calvi, Andrew Golden and Lori Pile for critical reading of the manuscript.