|Home | About | Journals | Submit | Contact Us | Français|
A protein complex consisting of Mago Nashi and Tsunagi/Y14 is required to establish the major body axes and for the localization of primordial germ cell determinants during Drosophila melanogaster oogenesis. The Mago Nashi:Tsunagi/Y14 heterodimer also serves as the core of the exon junction complex (EJC), a multiprotein complex assembled on spliced mRNAs. In previous studies, reduced function alleles of mago nashi and tsunagi/Y14 were used to characterize the roles of the genes in oogenesis. Here, we investigated mago nashi and tsunagi/Y14 using null alleles and clonal analysis. Germline clones lacking mago nashi function divide but fail to differentiate. The mago nashi null germline stem cells produce clones over a period of at least 11 days, suggesting that mago nashi is not necessary for stem cell self-renewal. However, germline stem cells lacking tsunagi/Y14 function are indistinguishable from wild type. Additionally, in tsunagi/Y14 null germline cysts, centrosomes and oocyte-specific components fail to concentrate within a single cell and oocyte fate is not restricted to a single cell. Together, our results suggest not only that mago nashi is required for germline stem cell differentiation but that surprisingly mago nashi functions independently of tsunagi/Y14 in this process. On the other hand, Tsunagi/Y14 is essential for restricting oocyte fate to a single cell and may function with mago nashi in this process.
The divergence and restriction of cell fates in the initial phases of oogenesis in Drosophila melanogaster illustrate a fundamental puzzle of development, the diversification of genetically identical cells. At the onset of oogenesis, germline stem cell division produces genetically identical daughter cells (Gilboa and Lehmann, 2004a; Lin, 1997). One becomes a replacement stem cell. The other becomes a cystoblast, undergoing four synchronous rounds of mitosis accompanied by incomplete cytokinesis to produce a cyst of 16 germline cells, interconnected by cytoplasmic bridges or ring canals (de Cuevas et al., 1997; King, 1970; Spradling, 1993) (Fig. 1A,B). Although two of the 16 cells, those with four ring canals, initially exhibit oocyte traits, the oocyte fate becomes restricted to a single cell and the cyst is enveloped by a monolayer of somatically derived follicle cells, thereby becoming an egg chamber that differentiates into an egg.
Sub-cellular localization of molecular determinants and intercellular signaling are two non-mutually exclusive mechanisms for initiating diversification of genetically identical cells (Horvitz and Herskowitz, 1992). Mago Nashi (Mago) and Tsunagi/Y14 (Tsu) are two proteins known to function in localizing molecular determinants (Hachet and Ephrussi, 2001; Mohr et al., 2001; Newmark and Boswell, 1994). Both proteins are evolutionarily highly conserved proteins (Hachet and Ephrussi, 2001; Micklem et al., 1997; Mohr et al., 2001; Newmark and Boswell, 1994). Comparative sequence analysis reveals that Mago is devoid of known structural motifs while Tsu contains just one, a canonical RNA binding motif (Kataoka et al., 2000). Molecular experiments indicate that the two proteins form heterodimeric complexes both in vitro and in vivo (Tange et al., 2004). The crystal structure of the heterodimeric complexes, formed in the absence of RNA, shows that the RNA binding motif is utilized for Tsu/Mago protein:protein interaction and is not readily available for binding RNA (Fribourg et al., 2003; Lau et al., 2003; Shi and Xu, 2003).
Immunohistochemical staining of Drosophila ovaries shows that, as expected for components of a complex, Tsu and Mago have essentially identical distributions in vivo (Hachet and Ephrussi, 2001; Mohr et al., 2001) and co-localize within both germline and follicle cell nuclei throughout oogenesis. The proteins also co-localize within the oocyte’s posterior-pole cytoplasm at two distinct times: stages 1-5 (S1-S5) and stages 8-9 (S8-S9). In addition, as is expected of components of a complex, tsu and mago mutant ovaries have similar phenotypes suggesting that the genes affect the same or similar processes (Mohr et al., 2001).
Co-localization during S1-S5 coincides with reciprocal intercellular signaling between the oocyte and follicle cells bordering the oocyte nucleus (Micklem et al., 1997; Mohr et al., 2001; Newmark et al., 1997). The oocyte-to-follicle cell signal determines posterior follicle cell fate (González-Reyes et al., 1995; González-Reyes and St. Johnston, 1994; Roth et al., 1995). The return, posterior follicle cell-to-oocyte, signal triggers a reorganization of the oocyte’s microtubule network that is essential for establishing the oocyte’s anterior-posterior axis and for anterior migration of the oocyte nucleus. In homozygous tsu6 and tsu7 or hemizygous mago1 oocytes, posterior follicle cell fate is properly specified but reorganization of the oocyte’s microtubule network is defective, suggesting that the Tsu/Mago complex is involved in interpreting the return intercellular signal (Micklem et al., 1997; Mohr et al., 2001; Newmark et al., 1997).
Beginning with S9 egg chambers, oskar (osk) mRNA accumulates within the oocyte’s posterior pole cytoplasm, co-localizing with Tsu and Mago proteins (Hachet and Ephrussi, 2001; Micklem et al., 1997; Mohr et al., 2001; Newmark et al., 1997). The accumulation of osk mRNA is necessary to create the specialized posterior cytoplasm that is essential for the determination of primordial germ cells (Ephrussi et al., 1991; Ephrussi and Lehmann, 1992). Although the localization of osk mRNA is dependent on wild-type Tsu and Mago function, the localization of Tsu and Mago precedes and is independent of osk mRNA localization (Mohr et al., 2001; Newmark et al., 1997).
Biochemical studies in cell-based and in vitro systems have identified Tsu and Mago as components of a protein complex, the exon junction complex (EJC), that binds mRNAs 20-24 base pairs upstream of splice junctions after splicing is complete (Tange et al., 2004). EJCs remain bound to mRNAs exported from the nucleus to the cytoplasm and are thought to function in nuclear export, nonsense mediated decay, translational regulation and RNA localization (Tange et al., 2004; Wilkinson, 2005). These studies show that the core of the EJC consists of a heterotetramer, consisting of Mago, Tsu/Y14, eukaryotic initiation factor 4AIII (eIF4AIII) and Barentsz/MLN51. Three dimensional structural analysis of the core components bound to RNA has revealed the various protein-protein interactions within the complex and the interactions of these components with RNA (Andersen et al., 2006; Bono et al., 2006; Stroupe et al., 2006). Stable association of the EJC with mRNA is ATP-dependent and is maintained by inhibition of the eIF4AIII ATPase activity by Mago-Tsu/Y14 (Ballut et al., 2005). Even though Mago and Tsu appear to be constant components of the otherwise dynamic EJC, the relationship of the EJC to the complex involved in sub-cellular localization has not been established.
Although Mago and Tsu have similar characteristics and function together in specific developmental processes, their phenotypes are not identical. During S9-S10 of oogenesis, the transforming growth factor-α-like protein (TGFα/Gurken) and TGFα mRNA are often reduced or undetectable in homozygous tsu mutant oocytes. However, hemizygous mago mutant oocytes are indistinguishable from wild type (Micklem et al., 1997; Mohr et al., 2001; Newmark et al., 1997), suggesting that the two proteins may function independently in some developmental processes.
Several lines of evidence suggested to us that Mago and Tsu might function in the initial phases of oogenesis. Both Mago and Tsu are detected in all cells of the germarium. Previously studied mutations most likely retain residual functions: tsu6 and tsu7 are hypomorphic by genetic tests and mago1 function is temperature sensitive (Boswell et al., 1991; Mohr et al., 2001). Thus, it is plausible that very low levels of Mago and Tsu/Y14 are required during the initial phases of oogenesis. Consequently, we reasoned that tsu and mago mutations, without residual functional activity, would be necessary to investigate the roles of Mago and Tsu/Y14 in very early oogenesis. To investigate whether mago and tsu are necessary for the initial phases of oogenesis, we generated germline clones using null alleles of mago and tsu. The phenotypic defects observed for mago mutations indicate that mago is required for an early oogenic event, germline stem cell differentiation, but tsu is not, suggesting that mago may act independently of tsu/Y14 during germline stem cell differentiation. Germline clones that are tsu null reveal that the wild-type function of the gene is necessary to restrict oocyte fate to a single cell.
Standard methods were used for all crosses and culturing. Stages of oogenesis are according to Spradling (1993). All lines were obtained from the Bloomington Stock Center (http://flybase.bio.indiana.edu), unless otherwise noted. Germline and follicle cell clones were induced as described by Chou and Perrimon (1992); alternatively, follicle cell clones were induced utilizing C587-Gal4 (generous gift from A. Spradling) (Kai and Spradling, 2003). The bam-GFP reporter was a generous gift from D. McKearin (Chen and McKearin, 2003b). The null allele of tsu, tsuΔ3-5, was generated by excision of the tsu1 P-element insertion (Hachet and Ephrussi, 2001; Mohr et al., 2001; Robertson et al., 1988). All mutant phenotypes observed in germline cells homozygous for either magoSHL-1 or for tsuΔ3-5 are rescued employing a p[w+; mago+] and a p[w+; tsu+] transgene, respectively (Mohr et al., 2001; Newmark et al., 1997).
The following lines were utilized to construct germline clones:
Ovary dissections and immunohistochemistry were performed as described in Mohr et al. (2001). The following antibodies were obtained from the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City, IA) and used at the delutions indicated: α-3A9, 1:100 (α-Spectrin) (Byers et al., 1987); and α-Orb 6H4, 1:100 (Lantz et al., 1994). The following antibodies were used at the delutions indicated: mouse monoclonal α-C(3)G, 1:200 (generous gift from S. Page and S. Hawley) (Page and Hawley, 2001); α-Egl, 1:2000 (generous gift from R. Lehmann) (Mach and Lehmann, 1997); α-CP309, 1:500 (generous gift from Y. Zheng)(Kawaguchi and Zheng, 2004); and α-TJ, 1:3000 (generous gift from D. Godt) (Li et al., 2003). The mouse monoclonal α-Tsu antibody (Tsu208) was used at 1:500. Secondary antibodies, Alexa 488 and 594 (Invitrogen) were used at 1:1000. Rat anti-c-Myc antibody (Serotec) was used at 1:1000. Fluorescent images were taken with a TCS SP2 confocal microscope (Leica, Exton, PA) or with an Axioskop 2 plus microscope (Zeiss, Thornwood, NY).
The null allele of tsu, tsuΔ3-5, was characterized by preparing genomic DNA from tsuΔ3-5, bw sp/CyO adult flies and PCR amplification of the tsu genomic region, as described in Mohr et al. (2001). PCR products were sequenced using an ABI Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA, USA) and analyzed as described in Mohr et al. (2001).
The distribution of Mago protein was determined in transgenic flies carrying either a myc-mago+ or GFP-mago+ transgene regulated by the endogenous mago promoter (Newmark et al., 1997). The proteins encoded by the transgenes are expressed in both germline and follicle cells. In germaria derived from females carrying the myc-mago+ transgene, immunohistochemical staining with rat monoclonal anti-c-Myc antibody reveals that cytoplasmic Mago preferentially accumulates in one or both pro-oocytes (Fig. 2A). Using our experimental conditions, anti-c-Myc antibody does not recognize Myc-Mago within nuclei, presumably due to epitope masking. The cytoplasmic accumulation of Mago is indistinguishable from proteins such as Orb, BicD and Egl that accumulate in the oocyte’s cytoplasm and are necessary for oocyte development (Fig. 2A) (Lantz et al., 1994; Mach and Lehmann, 1997; Schüpbach and Wieschaus, 1991; Suter et al., 1989; Suter and Steward, 1991). By region 3, cytoplasmic accumulation of Mago is only in the oocyte where it is localized within the posterior pole (Hachet and Ephrussi, 2001; Mohr et al., 2001). In transgenic flies carrying GFP-mago+, Mago is in the nuclei of both somatic and germline cells of the germarium, including the nuclei of germline stem cells (GSCs) (Fig. 2D).
The distribution of Tsu is essentially identical to Mago protein. Unlike Mago, the apparent concentration of Tsu is lower in region 1 than in regions 2 and 3 (Fig. 2B,C). In cysts of region 2, Tsu accumulates preferentially in the cytoplasm in a manner indistinguishable from Egl (Fig. 2B). Thus, even though the earliest documented requirement for mago+ and tsu+ function is during stages S1-S5 when reciprocal signaling between the oocyte and posterior follicle cells initiates axis formation (Hachet and Ephrussi, 2001; Micklem et al., 1997; Mohr et al., 2001; Newmark et al., 1997), the pattern of differential accumulation of Mago and Tsu/Y14 in early germline cells suggests that their function may be required during the germarial stages of oogenesis.
To assess the requirement for mago+ function in germarial stages of oogenesis, we studied mosaic ovaries homozygous for a null allele of mago (magoSHL-1), a recessive lethal, 202 base pair deletion (Newmark and Boswell, 1994). Theoretical translation of magoSHL-1 gives a frame-shifted peptide containing just the initial 14 amino acids of the Mago protein (Newmark et al., 1997). To generate magoSHL-1 mosaic adults, we induced flip recombinase-mediated, FRT-specific recombination by heat-shocking larvae containing a single copy of the Histone H2A variant-Green Fluorescent Protein (Hist2av-GFP) transgene, inserted on the right arm of the second chromosome, and heterozygous, in trans, for magoSHL-1 (Chou and Perrimon, 1992; Clarkson and Saint, 1999; Golic and Lindquist, 1989; Xu and Rubin, 1993). Mosaic cells are identified at any stage of oogenesis by the absence of GFP fluorescence so that mosaic phenotypes can be compared with non-mosaics (presence of fluorescence) in the same ovary. Initially, we examined germline cyst formation in ovaries stained with the anti-α-Spectrin antibody 3A9 (Byers et al., 1987). The α-Spectrin protein is a component of the fusome, a marker for germline cyst formation (de Cuevas et al., 1996). The fusome, a germline-specific organelle, is spherical (spectrosome) in GSCs and cystoblasts (CBs) and becomes an increasingly branched (Fig. 1C and Fig. 3B), filamentous structure during germline cyst formation (de Cuevas et al., 1997; Deng and Lin, 1997; Lin et al., 1994).
In homozygous magoSHL-1 germline mosaics, as in wild-type germaria, spectrosome-containing, putative GSCs are located in region 1 adjacent to the base of the terminal filament (Fig 3). However, in striking contrast to wild type (Fig. 3A,B), regions 1, 2 and 3 of magoSHL-1 mutant germaria are devoid of branched fusomes and are instead populated with predominantly individual germline cells (40/49) containing a spectrosome-like organelle (Fig. 3C,F, I). The spectrosome-like organelle of these cells becomes larger than that of a GSC. Mutant magoSHL-1 germline cells containing these spectrosome-like organelles become encapsulated by follicle cells and bud off of the germarium, forming pseudo-egg chambers (Fig. 3I). Germline cells homozygous for magoSHL-1 and heterozygous or homozygous for a p[w+; mago+] transgene are indistinguishable from wild-type germline cells, demonstrating that the magoSHL-1 phenotype is a consequence of removing mago+ function from the germline.
While the absence of fusomes in homozygous magoSHL-1 germline mosaics, indicates that oogenesis is blocked prior to germline cyst formation, the presence of putative GSCs implies that mago+ function is not required for GSC maintenance, division or viability. To further investigate whether mago+ function is required for GSC maintenance, division and viability, magoSHL-1 germline mosaics were induced as described above, except that, F1 adult females were heat shocked rather than F1 larvae (Harrison and Perrimon, 1993). Ovaries were dissected 4, 8 and 11 days post heat shock and examined to determine the percentage of germaria with magoSHL-1 GSC clones. Since normally, germline cells require ~6 days to transit the germarium (King, 1970), the presence of mosaic germline cells within the germarium seven days or more after induction of clones is expected only if GSCs lacking functional Mago are viable and divide.
Germline mosaics induced by heat shocking F1 adult females were indistinguishable from similar clones induced in F1 larvae. Homozygous magoSHL-1 GSC clones were identified in germaria 4 days (9%, n=93 germaria), 8 days (22%, n=55) and 11 days (21%, n=170) post heat shock. A similar percentage of GSC clones per germaria is observed when ovaries are dissected from control females 4 days (8%, n=63) and 11 days (14%, n=93) post heat shock. The fact that GSC clones: (a) are observed 7 days or more after heat shock, (b) occur at a frequency indistinguishable from wild type, and (c) are identical phenotypically to those induced in F1 larvae suggests that mago+ function is not required for the viability, maintenance and division of GSCs.
The differentiation of ovarian GSC into a CB involves passage through an intermediate state, the pre-cystoblast (Gilboa et al., 2003; Ohlstein and McKearin, 1997). GSCs are readily identified because they are adjacent to the terminal filament in the anterior of the germarium (the stem cell niche consisting of terminal filament cells, cap cells and inner germarium sheath cells), they contain a spectrosome and, in GSCs, transcription of the bag of marbles (bam) gene is repressed (McKearin and Ohlstein, 1995; Xie and Spradling, 1998; Xie and Spradling, 2000). In contrast, although pre-cystoblast cells contain spectrosomes and silencing of bam transcription is maintained, pre-cystoblasts are no longer in contact with the niche. CBs also contain spectrosomes and are outside the stem cell niche but, in CBs, bam transcription is not silenced. Therefore, we postulate two possible explanations for the magoSHL-1 phenotype (absence of fusomes), either mago+ function is required for CB differentiation and in its absence, development is arrested in a pre-CB state; or alternatively, CB differentiation is not impaired but mago+ function is required for the four rounds of mitosis that produce a germline cyst.
Oocyte differentiation is regulated by extrinsic factors produced by somatic cells of the stem cell niche and germline intrinsic factors. Two germline intrinsic factors, bam and benign gonial cell neoplasm (bgcn), are necessary for CB differentiation (Lavoie et al., 1999; McKearin and Ohlstein, 1995; Mckearin and Spradling, 1990; Ohlstein et al., 2000). Transcription of bam which is subject to stringent, negative regulation by Bone morphogenetic protein (Bmp/Dpp) signaling from the niche cells, is turned off in GSCs and on in CBs (Chen and McKearin, 2003a; Chen and McKearin, 2003b; Song et al., 2004). Germline cells that lack Bam fail to differentiate into CBs but continue to divide, resulting in germ cell hyperplasia. Ectopic expression of Bam within GSCs induces these cells to differentiate into CBs, resulting in loss of the GSCs from the germarium (Ohlstein and McKearin, 1997). Utilizing a bam reporter transgene (bamP-GFP; a transgene with the bam promoter fused directly to GFP), Chen and McKearin (2003b) were able to distinguish between GSC-like cells (GFP−) and differentiating CBs (GFP+).
To investigate whether magoSHL-1 germline clones are blocked in a pre-CB state or if CB differentiation is disrupted due to a mitotic defect, we examined bamP-GFP transgene expression in magoSHL-1 germline clones (Fig. 4A,B). Homozygous magoSHL-1 germline clones were identified either by the absence of Hist2Av-GFP fluorescence or by employing α-3A9 to identify germline cells that contained spectrosome-like organelles and were displaced from the stem cell niche. Regardless of how mosaic germline cells were identified, bamP-GFP transcription was not detectable in germline cells lacking mago function (Fig. 4B). The absence of detectable expression of the bam transcriptional reporter in homozygous magoSHL-1 germline cells suggests that these cells are in a pre-CB state.
Genetic studies show that Bam protein forms a complex with Bgcn protein and that the complex is required for CB differentiation (Ohlstein et al., 2000). Mutant ovaries lacking either bam or bgcn function are filled with spectrosome containing cells that divide. Ovaries containing magoSHL-1 germline clones consist of cells with spectrosome-like structures that are larger than spectrosomes detected in wild-type GSCs and magoSHL-1 mutant cells fail to differentiate. Although the phenotypes of bam or bgcn mutant germline cells are similar to magoSHL-1 germline cells, magoSHL-1 germline cells are distinct in that the cells contain enlarged spectrosome-like structures as opposed to the smaller spectrosomes typically observed in wild-type, bam or bgcn mutant cells. Thus, if germline cells homozygous for magoSHL-1 are blocked in a GSC/pre-CB state, we would expect germline cells double mutant for mago and bgcn to exhibit a magonull phenotype and to be readily distinguishable from bgcn1 mutant germline cells based on spectrosome size. In double mutant magoSHL-1, bgcn1 germline clones, germline cells are indistinguishable from magoSHL-1 germline clones, indicating that the bgcn mutant phenotype requires mago+ (mago is epistatic to bgcn; Fig. 4C). Considered together, the transcriptional silencing of bamP-GFP in magoSHL-1 germline clones and the fact that mago is epistatic to bgcn indicate that CB differentiation requires mago+ within the GSC or pre-CB.
To ascertain whether mago+ function is required for the division or differentiation of another stem cell population, we investigated the role of mago+ in the division and differentiation of somatic stem cells (SSC) that produce the ovarian follicle cells (FCs; Fig. 1A). Two to three SSCs that give rise to the FCs reside near the boundary of R2a and R2b (Margolis and Spradling, 1995; Zhang and Kalderon, 2001). During egg chamber development, FC precursors contact the germline cyst, extend centripetal processes to separate germline cysts from one another, encapsulate the germline cysts and differentiate into distinct populations of cells (epithelial FCs, polar cells and stalk cells) (Horne-Badovinac and Bilder, 2005). When the egg chamber buds off from the germarium, approximately 80 FCs enclose the germline cyst (S1) and these cells divide to form ~650 FCs, ceasing division at the end of S6 (King and Vanoucek, 1960; Margolis and Spradling, 1995).
Homozygous magoSHL-1 SSC clones divide and differentiate, producing S1-S9 egg chambers that are indistinguishable from wild type (Fig. 5). As with magoSHL-1 germline mosaics, ovarian somatic cell mosaics were generated employing FLP/FRT recombination and the Hist2av-GFP transgene to mark mutant clones by the absence of GFP fluorescence. However, a C587-Gal4, UAS-Flp, chromosome was introduced into the flies to express Flp in the ovarian somatic cells and to generate SSC clones (Kai and Spradling, 2003). The Traffic Jam protein, the Drosophila homolog of the retroviral oncoprotein v-Maf8 and vertebrate large MAF transcription factors), is expressed in germarial somatic cells (Li et al., 2003). Thus, we employed α-TJ to mark ovarian somatic cells. The observation that homozygous magoSHL-1 SSC clones develop to at least stage 9 of oogenesis suggests that although mago+ function is required for GSC differentiation, mago is not essential for the divisions and differentiation of SSCs that give rise to the FCs. The result also suggests that the magonull phenotype is not due to the disruption of a function that is generally required within cells but is likely caused by disrupting a function specific for GSC differentiation.
To determine whether tsu+ function is required in early oogenesis, we utilized the recessive lethal mutation tsuΔ3-5, an 880 base pair deletion that was generated by imprecise excision of EP(2)0567, a P-element insertion in the tsunagi 5′-UTR (Hachet and Ephrussi, 2001; Mohr et al., 2001). The deletion extends upstream from the EP(2)0567 insertion site, eliminating the transcriptional promoter and all but the 3′-most, 72 base pairs of the 5′-UTR. Tsu protein is undetectable by immunohistochemistry in cells that are homozygous tsuΔ3-5/ tsuΔ3-5 (Fig. 2C).
As with magoSHL-1, we employed FLP/FRT mediated recombination and the Hist2av-GFP transgene to generate homozygous tsuΔ3-5 germline mosaics that are marked by the absence of GFP fluorescence. Once again, we monitored germline cyst formation by fusome morphogenesis using the anti-α-Spectrin antibody 3A9. We find that in germline clones, spectrosome containing cells are at the base of the terminal filament (putative GSCs) and in the anterior part of region 1 (putative CBs; Fig. 3).
In contrast to magoSHL-1, in tsuΔ3-5 germline clones, we observe branched fusomes in regions 1, 2 and 3 (Fig 3D,G,J). Germline clones that are lacking both mago and tsu function (tsuΔ3-5, magoSHL-1 homozygotes) exhibit a phenotype that is indistinguishable from homozygous magoSHL-1 germline cells, indicating that the tsu mutant phenotype requires mago+ (mago is epistatic to tsu; Fig. 3E,H,K). Thus, unlike mago+ function, tsu+ function is not required prior to germline cyst formation and consequently is not essential for CB differentiation. Furthermore, we find that homozygous tsuΔ3-5 germline clones are blocked shortly after egg chambers form. Unlike wild-type egg chambers, the nuclei of homozygous tsuΔ3-5 germline clones remain polytene, suggesting that they do not develop beyond stage 4 (Spradling, 1993). These results indicate that in addition to its role in axis formation, tsu+ function is essential during early oogenesis but not necessary for GSC differentiation.
To investigate the development of homozygous tsuΔ3-5 germline cysts, we characterized the distribution of Orb protein, a homologue of the Cytoplasmic Polyadenylation Element Binding protein (CPEB) that is essential for oocyte differentiation (Christerson and McKearin, 1994; Lantz et al., 1994). In wild-type germaria, Orb protein is first detected in the cytoplasm of all 16 cells of a germline cyst in early region 2a (Fig. 6A,B,C). Subsequently, it accumulates in the cytoplasm of the two pro-oocytes. Accumulation is restricted to the oocyte alone in late region 2a, and is localized within the oocyte’s posterior pole in region 3 (Fig. 6C). Under our experimental conditions, Orb accumulation is observed in 2-3 cysts in region 2 of non-mosaic ovarioles. Orb accumulation in a single cell (the oocyte) is detected in 98% of S1 (n =52) and in 99% of early stage (S2-S6) egg chambers (n = 137).
In tsuΔ3-5 germline mosaics, Orb protein is detectable but accumulation of Orb in mutant cysts is aberrant (Fig. 6D,E,G). Accumulation in region 2 is undetectable in 40% and transient in the remaining 60% of tsuΔ3-5 germaria (n = 33). By stage S1 the majority of egg chambers lack detectable accumulation of Orb (23/33) and Orb accumulation is rarely detected in mosaic egg chambers of the vitellarium (6/106). Transient accumulation is abnormal in that Orb is often observed in reduced amounts and fails to localize within the cell’s posterior pole. Aberrant accumulation indicates that the initiation of oocyte differentiation is defective in tsuΔ3-5 clones and that tsu+ function is required within germarial cysts for continued oocyte development.
The meiotic state of germline cells can be used as a second marker of germline cyst differentiation. Unlike Orb accumulation, changes in the meiotic state of germline cells are seen in the nucleus rather than the cytoplasm of cells and these changes are not microtubule-dependent (Huynh and St Johnston, 2000). Cells in meiotic prophase contain a specific structure, the synaptonemal complex (SC) that is associated with paired meiotic chromosomes (Page and Hawley, 2004). Meiotic cells can be identified immunohistochemically by staining for the SC component, C(3)G protein (Page and Hawley, 2001).
In wild-type germaria, both pro-oocytes enter meiosis in early region 2a (Fig. 7A,B) (Huynh and St Johnston, 2000; Page and Hawley, 2001). Subsequently, the two cells with 3 ring canals enter and then exit meiosis. In region 3, meiosis is restricted to a single cell, the oocyte, and the losing pro-oocyte develops as a nurse cell. Under our experimental conditions, all wild-type control ovarioles (100%, n = 155) have one or more cysts in region 2 with synaptonemal complexes. The average number of such region 2 cysts is 3.5 (n = 47). Usually (42/47), synaptonemal complexes are observed in both pro-oocytes of the most posterior cyst of region 2. Synaptonemal complexes are generally restricted to a single cell, the oocyte, in region 3 cysts (93%, n = 44) and early vitellarial egg chambers (94%, n = 29).
SC formation in homozygous tsuΔ3-5 germline clones deviates significantly from that which we find in wild-type ovarioles (Fig. 7C,D,E). Although entry into meiosis is normal in tsuΔ3-5 mutant cysts (Fig. 7E), in regions 2b and 3, most tsuΔ3-5 clones (84%, n=25) retain synaptonemal complexes in four adjacent cells (Fig. 7E). Therefore, we conclude the following: (a) in tsuΔ3-5 homozygous germline clones four adjacent cells fail to exit meiosis; and (b) the earliest detectable step in the selection of the oocyte, restriction of meiosis to the two pro-oocytes, is defective in cysts devoid of tsu+ function.
We utilized centrosome migration as a third marker of oocyte differentiation (Bolivar et al., 2001; Grieder et al., 2000). Like Orb, Egl and BicD, centrosomes initially accumulate at the anterior of the oocyte and translocate to its posterior within region 2b. In region 3, the S1 oocyte has centrosomes concentrated within the posterior pole. Unlike Orb, Egl and BicD, however, the translocation of centrosomes into the oocyte is not disrupted by microtubule depolymerizing drugs (Bolivar et al., 2001). Thus, centrosome migration also allows an assessment of transport of components to the oocyte along stable microtubules.
To investigate the distribution of centrosomes in tsuΔ3-5 clones within the germarium, we visualized centrosomes by staining with antibodies against the centrosomal protein CP309, the Drosophila Pericentrin-like protein (Kawaguchi and Zheng, 2004; Martinez-Campos et al., 2004). In wild-type germaria (Fig. 8A), we detect centrosomes concentrated within the oocyte in regions 2b and 3 in almost all germaria (35/36). However, in germline cells homozygous for tsuΔ3-5, centrosomes in regions 2b and 3 are not concentrated within a single cell but are evenly distributed (n=15; see Fig. 8B). Considered together, with the results observed in the meiotic state of germline cells devoid of tsu+ function, the failure to concentrate centrosomes within a single cell suggests that Tsu is a component necessary for restricting oocyte fate to a single cell within the cyst. Moreover, the observations indicate that tsu+ is required for the transport of oocyte-specific components along stable microtubules.
In this paper we present evidence that in addition to previously known roles in axis formation and oskar RNA anchoring/localization during late oogenesis, mago and tsu are also required in essential steps during early oogenesis. Mago is necessary for the entry of the GSC into the oogenic pathway. Unexpectedly, Mago functions independently from Tsu/Y14 to regulate this process. Tsu/Y14 is not required for the differentiation of GSCs. However, Tsu/Y14 is necessary to restrict oocyte fate to a single cell. Thus, our studies reveal a new role for Mago in GSC differentiation and a new role for Tsu/Y14 in oocyte fate specification.
Establishing where Mago functions within the network of pathways regulating GSC differentiation is fundamental to understanding Mago’s mechanism of action. This network includes the Bmp, Nanos/Pumilio and Bam/Bgcn pathways.
Bmp, produced within the somatic cells of the GSC niche, is an extracellular signaling molecule that regulates self-renewal and asymmetric division of GSCs (Wong et al., 2005). A signal transduction pathway within GSCs, activated by Bmp, produces a complex containing two Smad family member proteins. The complex translocates to the nucleus, binds a transcriptional silencer within the bam promoter and prevents transcription of bam (Chen and McKearin, 2003a; Song et al., 2004).
The nanos (nos) and pumilio (pum) genes function intrinsically to regulate GSC self-renewal and asymmetric division (Forbes and Lehmann, 1998; Gilboa and Lehmann, 2004b; Lin and Spradling, 1997; Wang and Lin, 2004). Based on the observation that Pum and Nos form a heterodimer that binds and inhibits translation of hunchback mRNA in embryos (Sonoda and Wharton, 1999), it has been proposed that in GSCs Pum:Nos suppresses the translation of mRNAs encoding proteins essential for cystoblast differentiation.
The bam (encoding a novel protein) and bgcn (encoding a DExH-box RNA-binding protein) genes play key roles in GSC differentiation (McKearin and Ohlstein, 1995; Ohlstein et al., 2000; Ohlstein and McKearin, 1997). While Bmp signaling silences bam transcription in GSCs and pre-cystoblasts, bam transcription is upregulated within cystoblasts (Gilboa et al., 2003; Ohlstein and McKearin, 1997). Within cystoblasts, Bam is synthesized and forms a complex with Bgcn, resulting in differentiation of the cystoblast. Chen and McKearin (2005) and Szakmary et al. (2005) have recently proposed that the Bam:Bgcn complex acts by antagonizing the Pum:Nos complex, thus relieving translational repression and inducing the synthesis of cystoblast differentiation factors.
Although GSCs that are null for mago divide, we did not detect the expression of a bam transcriptional reporter, suggesting that these cells are blocked in a GSC or in a pre-cystoblast state and accounting for the epistatic relationship between mago and bgcn. Within the network of pathways regulating GSC differentiation, there are two candidate sites where Mago may function. First, Mago may negatively regulate the Bmp signal transduction pathway within GSCs or a redundant pathway. In this case, transcriptional silencing of bam would be maintained, even outside of the niche due to stability of the negative regulatory signal. Alternatively, Mago may be required for transcriptional activation of bam. In the absence of Mago, there would be insufficient Bam:Bgcn to displace Pum:Nos and consequently translational repression of cystoblast differentiation factors would continue. Future research will help to distinguish between these alternatives. In addition, since Mago’s function in GSC differentiation is independent of Tsu/Y14, it will be of interest to elucidate Mago’s mechanism of action during this process.
Oocyte selection in Drosophila can be divided into the following three major steps: (a) fusome formation, (b) concentration of determinants first within the pro-oocytes and finally within the oocyte, and (c) oocyte maintenance (Huynh and St Johnston, 2004). Asymmetry during oogenesis is first apparent during fusome formation, when the cystoblast inherits a third of the fusome from the GSC. Subsequent divisions result in the acquisition of a larger amount of the fusome material by the two pro-oocytes than other cells of the cyst. Without functional fusomes, microtubule polarity within a cyst is aberrant and an oocyte is not specified. A polarized microtubule cytoskeleton is critical for establishing the intra-cyst asymmetry that results in the selective concentration of specific RNAs and proteins (e.g., Org, Egl, BicD and Par-1), first within the two pro-oocytes and ultimately within the definitive oocyte. Accumulation of proteins/RNAs is required either for oocyte determination or for maintenance of oocyte fate. Mutations in genes required for oocyte determination disrupt the accumulation of proteins/RNAs within a single cell within the cyst. However, mutations in genes required for oocyte maintenance do not interfere with the initial accumulation of oocyte-specific factors within a single cell but disrupt the translocation of these components from the anterior to the posterior of the oocyte. Prior to our analysis of tsu, only two genes were known to be required for the accumulation of both oocyte determination and maintenance factors within the presumptive oocyte.
Mutations in genes encoding fusome-associated components that are required for restricting oocyte specification to a single cell within the 16-cell cyst are expected to disrupt the accumulation of oocyte determination and maintenance factors within the presumptive oocyte. One such fusome component is the Drosophila spectraplakin protein, Shot, a cytoskeletal linker protein that contains a domain capable of bundling and stabilizing microtubules (GAS2) (Roper and Brown, 2004). Germline cells lacking shot function contain fusomes, but components required to specify and maintain oocyte fate fail to accumulate in a single cell. Consequently, all 16 cells of the cyst develop as nurse cells. Furthermore, the centrosomes, which normally migrate into the oocyte along fusomes, fail to accumulate within a single cell in shot mutant germline cells (Roper and Brown, 2004).
Dynein heavy chain 64C is another fusome-associated protein. Dynein heavy chain 64C (Dhc64C) mutant germline cells contain abnormal fusomes. As observed in shot mutants, oocyte-specific components (e.g., BicD) fail to accumulate in a single cell and centrosome migration is aberrant (Bolivar et al., 2001; McGrail and Hays, 1997). Par-1 (a serine/threonine kinase) is a third component of the fusomes (Cox et al., 2001; Huynh et al., 2001). However, unlike shot and Dhc64C mutant germline cells, in par-1 mutant germline cells, centrosome migration is normal. Oocyte-specific components accumulate in a single cell but fail to translocate from the anterior to the posterior of the oocyte. Thus, fusomes are important for oocyte determination (Shot and Dhc64C) and for maintenance (Par-1) of oocyte fate.
Two proteins not known to be associated with fusomes, BicD and Egl, are also essential for the accumulation of oocyte-specific components within a single cell (Bolivar et al., 2001; de Cuevas and Spradling, 1998; Mach and Lehmann, 1997; Ran et al., 1994; Schüpbach and Wieschaus, 1991). However, mutations in these genes do not disrupt the migration of centrosomes into a single cell, even though oocyte specification fails. Migration of centrosomes into the oocyte occurs in the presence of colchicine, suggesting that migration of centrosomes into the oocyte does not occur along dynamic microtubules. Consistent with this observation, acetylated tubulins (a form of tubulin found in stable microtubules), are found in a population of microtubules associated with fusomes (Roper and Brown, 2004). Based on the presence of acetylated tubulins in microtubules associated with fusomes, the ability of Shot’s GAS2 domain to stabilize microtubules and the shot mutant germline phenotype, Roper and Brown (2004) suggest that centrosomes migrate along stable microtubules.
Utilizing the synaptonemal complex formation, Orb accumulation and centrosome migration, we monitored the ability of wild-type and tsunull germline cysts to select a single cell of the 16-cell cyst as the definitive oocyte. The synaptonemal complex marks progression of individual cells of the cyst through meiosis. In tsu mutant germline cysts, four cells enter meiosis in germarial region 2a and remain in meiosis into region 3, suggesting that selection of the oocyte is abnormal. Microtubule depolymerizing drugs do not interfere with the formation the synaptonemal complex but the SC is not maintained in germarial region 3, indicating its maintenance at this stage is dependent on dynamic microtubules. Given that the synaptonemal complex persists in tsu mutant germline cysts in germarial region 3, we conclude that dynamic microtubules are not disrupted in tsu mutant cells.
Previously, only shot and Dhc64C had been demonstrated to be essential for the polarized transport of all oocyte-specific components. Here, we show that Tsu/Y14 is an additional factor regulating the polarized transport of centrosomes and all other oocyte-specific components. This suggests that, like Shot and Dhc64C, Tsu/Y14 functions upstream of BicD and Egl to restrict oocyte fate to a single cell. Reduced function alleles of mago exhibit phenotypes very similar to those observed in tsunull germline cells (Bennett and Boswell, unpublished), suggesting that Mago and Tsu/Y14 function together to restrict oocyte fate to a single cell. Further studies will be necessary to establish the position of Tsu/Y14 in the pathway relative to Shot and Dhc64C.
The data presented in this paper, considered with previous studies of the role of Tsu/Y14 in the localization/transport of RNAs, provides evidence that Tsu/Y14 is involved in an early step in the polarized transport of oocyte-specific RNAs/proteins. Future studies will determine the position of Tsu/Y14 within the oocyte specification pathway and reveal additional components involved in polarized transport during early oogenesis.
We thank Kathy Sheehan for helpful discussions and critical reading of the manuscript. The research was supported by an NIH training grant (GM7135) to P.E.B. and NIH (GM57989) to R.E.B. A National Science Foundation grant (DBI-0216118) was used to purchase the Leica confocal microscope.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.