aubergine encodes a Drosophila polar granule component required for pole cell formation and related to eIF2C

One way in which daughter cells may adopt different cell fates is through the asymmetric localization of cytoplasmic determinants before cell division. A prominent example of this phenomenon is found in early Drosophila development, where determinants localized to different regions of the oocyte specify the fates of cells formed at those regions in the developing embryo. During oogenesis a specialized cytoplasm, called pole plasm, is assembled at the posterior of the oocyte (for a review, see Rongo and Lehmann, 1996). Late in oogenesis, electron-dense organelles called polar granules can be found in the pole plasm, and these persist in the embryonic pole cells, which are germline precursor cells (for a review, see Mahowald, 2001). Pole plasm is necessary for induction of abdominal segmentation and germline formation during embryogenesis, and is sufficient to induce both features when transplanted ectopically to the anterior of the embryo (Illmensee and Mahowald, 1974; Frohnhöfer et al., 1986).

Mutations in several genes affect the assembly of pole plasm and thus germline formation and abdominal patterning. Among these, oskar (osk) has a central role. Both osk mRNA and the protein it encodes are localized in the pole plasm, and Osk protein is a component of polar granules (Kim-Ha et al., 1991; Ephrussi et al., 1991; Smith et al., 1992; Ephrussi and Lehmann, 1992; Breitwieser et al., 1996). The level of other pole plasm components and the number of pole cells are sensitive to osk gene dose, and misexpression of Osk protein induces ectopic pole plasm, pole cells and abdominal body patterning (Ephrussi and Lehmann, 1992; Smith et al., 1992). The distribution of Osk protein is normally spatially and temporally restricted through the combinatorial action of mRNA localization, translational regulation and the subsequent anchoring of Osk protein at the cortex of the oocyte. osk mRNA localizes to the posterior pole at around stage 8 of oogenesis (Kim-Ha et al., 1991; Ephrussi et al., 1991). Before localization, the mRNA is translationally repressed by Bruno (Bru; Aret – FlyBase) protein (Kim-Ha et al., 1995; Webster et al., 1997). Additional factors, including Bicaudal C (BicC; Saffman et al., 1998), Apontic (Lie and Macdonald, 1999) and possibly a factor named p50 (Gunkel et al., 1998) also contribute to repression of osk translation.

In contrast, products of the vasa (vas), staufen (stau) and aubergine (aub) genes are required for the accumulation of high levels of Osk protein at the posterior pole (Kim-Ha et al., 1995; Markussen et al., 1995; Rongo et al., 1995; Wilson et al., 1996). The vas and stau genes have been previously identified. vas encodes an RNA helicase related to the translation initiation factor eIF-4A (Hay et al., 1988b; Lasko and Ashburner, 1988; Liang et al., 1994). Stau is a double-stranded RNA-binding protein (St Johnston et al., 1992). Both Vas and Stau proteins have been shown to localize to the posterior pole of the oocyte, and Vas, like Osk, is a polar granule component (Hay et al., 1988a; Hay et al., 1988b; St Johnston et al., 1991). The identities and distribution of these gene products are consistent with models in which they interact directly with osk mRNA to promote its translation.

We now report that Aub is a cytoplasmic protein related to the translation initiation factor eIF2C. Using a fully functional Aub protein tagged with GFP, we show that Aub accumulates in pole plasm in a manner dependent on vas and is itself a polar granule component. Several lines of evidence strongly suggest that the posterior localization of Aub is not required for its role in translational activation. Instead, the posterior concentration is likely to be required for two additional functions of Aub. The first is a previously defined role in posterior localization of nanos (nos) mRNA (Wilson et al., 1996), whose restricted expression is required for correct abdominal patterning (Gavis and Lehmann, 1994). The second is in pole cell formation, as we demonstrate that Aub is required for this process independently of its function in activation of osk mRNA translation.

uas-gfp-aub was injected by standard methods (Spradling and Rubin, 1982) using w¹¹¹⁸ flies as recipients. The transgene is derived from the UASp vector of Rorth (Rorth, 1998) and was made by inserting mGFP6 (Siemering et al., 1996) fused in frame to the start codon of an aub cDNA (LD23087 of BDGP; Rubin et al., 2000). Transgenic lines carrying the aub genomic rescue constructs MR-1 and MR-3 (Schmidt et al., 1999) were provided by Ulrich Schäfer. The osk-bcd transgene used was the short osk-bcd version (Breitwieser et al., 1996) and was jumped onto the third chromosome from a stock provided by Anne Ephrussi. vas mutants used were vas^AS/vas^PD23 transheterozygotes. Similar results were also obtained with a vas^PD23/vas^O11 combination. vas-gfp (Breitwieser et al., 1996) flies were provided by Anne Ephrussi. stau mutants were homozygous stau^D3. aub mutants were aub^N11/aub^HN2. The P insertion in region 31E (P31E) used for RFLP analysis came from the stock w¹¹¹⁸; P{w^+mC=NM}31E P{ry^+t7.2=neoFRT}40A (Xu and Rubin, 1993). This stock also carries a ry^+t7.2 insertion at position 40A, which was ignored. The P insertion in 33A (P33A) was believed to be P{lacW}crol^s2346 (Spradling et al., 1995; Spradling et al., 1999). However, subsequent analysis led us to conclude that our flies contained a different insertion that is unexpectedly complemented by Df(2)Prl (Lindsley and Zimm, 1992) and probably lies proximal to 33A. Nevertheless, this line enabled us to generate RFLPs and successfully locate the aub gene.

Recombinants were generated between a Sp-marked aub^N11 chromosome and the P33A line. A single recombinant line was selected, isogenized and crossed to an isogenized P31E stock to generate recombinants between the P insertions. Genomic DNA was prepared according to Protocol 47 (Ashburner, 1989) from the Sp aub^N11 P33A and aub⁺ P31E parental lines and aub⁺ recombinants lacking both P elements. Approximately 5 μg of DNA from each parental line was digested with numerous restriction enzymes, electrophoresed on agarose gels, and Southern blotted. Probes were made by random priming of subcloned BDGP P1 DNA or by transcription using viral promoters flanking P1 inserts, when present. Enzyme/probe combinations that identified polymorphisms between the parental chromosomes were then used to test digested DNA from the recombinant flies. The most informative polymorphisms flanked the closest recombination event distal to the aub^N11 mutation and were generated from P1 DS04172 (Kimmerly et al., 1996), which was partially restriction mapped. While this mapping was being carried out, sequence information from the overlapping P1 DS01831 became available and was used to identify candidate genes in the region, including sting.

The entire aub-coding region was amplified by PCR (Expand™ High Fidelity kit, Boehringer Mannheim; primers: sense 5′-cactctagagcatttgtgacgagtagatcgtaaga-3′ and antisense 5′-cacggtaccgttaacgcatttaatgcaagtcatagt-3′) from genomic DNA obtained from single aub^N11/CyO, aub^N11/aub^HN2, and aub^N11/aub^K86 male flies (Turelli and Hoffmann, 1995). The product from the aub^N11 allele was recognized to contain a small deletion, which enabled identification of the correct PCR product for each of the other mutant chromosomes in trans. The PCR products were cloned into pBluescript SK(+) via XbaI and Asp718 sites in the primers. Two clones from independent PCR reactions were sequenced for each allele using pairs of primers approximately 1 kb apart and between them covering the entire coding region. Nucleotide differences from the wild-type aub sequence (GenBank Accession Number, X94613) were distinguished from PCR errors by their appearance in both of the independent clones. GenBank accession numbers for the coding regions of exons with nucleotide differences from the wild-type are as follows: aub^HN2 exon 9, AF334408; aub^K86 exon 4, AF334409; aub^K86 exon 6, AF334410; aub^N11 exon 4, AF334411; and aub^N11 exon 9, AF334412.

Polyclonal antibodies were raised in a rat by Josman, LLC, against the C-terminal region of Aub protein (amino acids 562-866) encoded by the BglII-XhoI fragment of EST LD23087 (Rubin et al., 2000) and expressed in inclusion bodies in Escherichia coli using the pET21a vector (Novagen). Crude ovary extracts were prepared by homogenizing hand-dissected ovaries from wild-type or aub^N11/aub^HN2 transheterozygous females in SDS-loading buffer and boiling for 5 minutes. Approximately one ovary pair equivalent was loaded on each lane of an 8% SDS-polyacrylamide gel. Proteins from the gel were transferred to nitrocellulose by submerged electroblotting at 20V for 45 minutes in transfer buffer (25 mM Tris base, 200 mM glycine, 20% methanol). The membrane was allowed to air dry. For detection, we used the Tropix Western-Light kit according to the manufacturer’s instructions, except that the incubation time with the primary antibody (at 1:1000) was increased to 2 hours and the number of washes was increased to a total of eight.

Ovaries were hand-dissected in PBS and disrupted with flame-drawn Pasteur pipettes. For in situ hybridization, ovaries were fixed and processed as described (Kim-Ha et al., 1991), using an RNA probe prepared from the full-length aub cDNA and labeled with digoxigenin-11-UTP (Boehringer Mannheim). For antibody staining, ovaries were fixed by rotating for 20 minutes in PBS + 6% formaldehyde under heptane, and embryos were fixed as described (Smith et al., 1992). Secondary antibodies coupled to Cy5 (Jackson Immunoresearch) or AlexaFluor™ 488 (Molecular Probes) were used as suggested by the manufacturer. Rabbit anti-Osk antibodies were used at 1:3000 final concentration. Microscopy was performed with a Leica SP-TCS using Leica Confocal Software version 2.00. Identical laser power and photomultiplier settings were used when direct comparisons of protein levels were made.

To investigate the molecular function of aub, we devised a strategy to positionally clone the locus. Previously, aub had been mapped by meiotic recombination to genetic position 39 on the second chromosome (Schüpbach and Wieschaus, 1991). We subsequently mapped aub between P elements reported to be inserted in polytene regions 31E and 33A (see Materials and Methods). We used RFLP mapping as well as clones and sequence information from the Berkeley Drosophila Genome Project (BDGP; Kimmerly et al., 1996; Adams et al., 2000) to narrow the location of the aub gene to a small region at the boundary of polytene divisions 32C and 32D (Fig. 1A, Materials and Methods).

This region contains sting (Schmidt et al., 1999), a member of an ancient gene family that includes the gene for the eukaryotic translation initiation factor eIF2C (Zou et al., 1998). With support from three lines of evidence, we were able to identify aub and conclude that aub and sting are the same gene. First, in each of three aub mutant chromosomes, we identified mutations predicted to truncate the Aub/Sting protein (Fig. 1B). Second, antibodies raised against a bacterially expressed C-terminal portion of Aub/Sting recognize a band of approximately 105 kDa in extracts from wild-type ovaries but not ovaries from aub mutants (Fig. 1C). A band corresponding to a similar molecular weight is produced by in vitro transcription and translation of an aub/sting cDNA (data not shown). Third, two genomic DNA fragments (Fig. 1A; Schmidt et al., 1999) which include the aub/sting gene rescue aub mutants: aub⁻ mothers carrying a single copy of either transgene are fertile and produce viable offspring. Following tradition, we use the original name, aubergine, for this locus.

The Drosophila genome contains four other members of the eIF2C-like gene family. One of these is piwi, a close chromosomal neighbor of aub (Fig. 1A) that acts in germline stem cell maintenance (Cox et al., 1998). Two additional members, CG7439 and dAGO1, are reported in the genome annotation (Adams et al., 2000). The latter is the closest known relative of eIF2C in flies and is presumably the Drosophila eIF2C homolog. We identified the fifth family member, corresponding to the genomic sequence AE003107 (Adams et al., 2000) and EST clot 2083 (Rubin et al., 2000), by tBLASTn searches of the BDGP databases using parts of Aub protein as the query sequence.

The aub transcript is expressed at relatively high levels in the germarium, at lower levels during mid-oogenesis, and again at high levels in the nurse cells and oocyte from about stage 6 of oogenesis (Fig. 2A; data not shown). Of greater interest is the distribution of Aub protein. Our antibodies, which detect Aub on western blots (Fig. 1C), fail to detect Aub protein in whole-mount ovaries. In both wild-type and aub mutant ovaries, we detect a similar low level of uniform staining (data not shown). As an alternate approach to determine the distribution of Aub protein, we used the UAS/GAL4 system to express a GFP-tagged version of Aub in vivo. Flies carrying a uas-gfp-aub transgene were crossed to those that express a nos-gal4-vp16 transgene which drives expression of GAL4-VP16 in the female germline (Van Doren et al., 1998).

The temporal expression of the nos-gal4-vp16 driver parallels that of aub (Van Doren et al., 1998; Fig. 2A), and uas-gfp-aub driven by nos-gal4-vp16 is sufficient to fully rescue aub mutant defects, indicating both that the fusion protein is functional and that its distribution includes the pattern required of native Aub protein. GFP-Aub protein can be visualized throughout oogenesis. In the nurse cells, the protein is distributed evenly in the cytoplasm and in concentrated foci surrounding the nuclei (Fig. 2B). These presumptive perinuclear particles remain until around stage 10. In the oocyte, GFP-Aub is at first dispersed evenly in the cytoplasm. Later, beginning at stage 8/9, it becomes strikingly concentrated in the posterior pole plasm, where it largely overlaps the distribution of Osk protein (Fig. 2C-E). This posterior concentration appears to occur by recruitment of the GFP-Aub protein, as the aub mRNA is not itself localized within the oocyte (Fig. 2A).

The localization of Aub to the posterior pole of the oocyte raises the possibility that it is a component of polar granules. Polar granules are large, electron-dense structures found in the posterior pole plasm of the oocyte (for a review, see Mahowald, 2001). In the embryo, these granules remain localized to the posterior pole where they are largely incorporated into the budding pole cells. Several other posteriorly localized proteins, including Tudor (Tud), Vas and Osk, have been shown by immunoelectron microscopy to be components of the polar granules (Hay et al., 1988a; Hay et al., 1988b; Bardsley et al., 1993; Breitwieser et al., 1996). With fluorescence-based detection methods, polar granule components appear in particles, which have been presumed to be the polar granules. Colocalization of Osk and Vas (fused to GFP) in these particles supports this conclusion (Breitwieser et al., 1996).

To determine if Aub is a polar granule component, we used confocal microscopy to characterize the subcellular distribution of GFP-Aub in embryos. Before pole cell formation, GFP-Aub is found throughout the cytoplasm but strongly concentrated at the posterior pole in distinct particles (Fig. 3A), most of which also contain Osk protein (Fig. 3D,G). The particles are predominantly incorporated into the pole buds, which form at the posterior pole of the embryo and initiate pole cell formation (Fig. 3B,E,H). At this time, the particles increase in size and decrease in number, suggesting that individual particles are fusing. Simultaneously, the level of dispersed, non-particulate GFP-Aub increases in the pole cell cytoplasm. Before cellularization, between nuclear division cycles 11 and 13, the particles often cluster in zones at opposite sides of the pole cell nuclei (Fig. 3C, arrowheads). Most of the particles continue to display colocalization of GFP-Aub and Osk (Fig. 3C,F,I). However, some Osk-containing particles now begin to appear in the pole cell nuclei, while GFP-Aub is not detectable in these particles and consistently remains excluded from the nuclei. By the time of cellularization, following nuclear division cycle 14, only a few large cytoplasmic particles persist in each pole cell, with much of the GFP-Aub distributed evenly throughout the cytoplasm (Fig. 4A). Osk continues to colocalize strongly with GFP-Aub in the particles, but most Osk staining is now detected in the nuclear particles and dispersed in the nucleoplasm (Fig. 4B,C). Both cytoplasmic and nuclear particles often appear in striking hollow spherical shapes (donut-like in optical sections), which are characteristic of some polar granule clusters as well as similar particles that are present in nuclei, called nuclear bodies (Mahowald, 1962; Mahowald, 1968; Fig. 4A-C). In comparable colocalization experiments, we determined that Vas-GFP behaves very similarly to Osk: there is always a high level of colocalization in particles, both nuclear and cytoplasmic (Fig. 4D-F and data not shown).

Our data demonstrate that GFP-Aub is found in particles that (1) also contain Osk and are inferred to contain Vas, both known polar granule components; (2) are located, like polar granules, at the posterior pole of the oocyte and embryo and are incorporated into pole cells; (3) increase in size within the pole cells with similar timing to that previously described for polar granules; and (4) frequently appear, as do polar granules, in spherical structures that are donut-shaped in sections. From these results, we conclude that Aub is a polar granule component. However, unlike Osk and Vas, which are largely nuclear after the cellular blastoderm stage, Aub is restricted to the cytoplasmic class of particles, indicating that the nuclear bodies and cytoplasmic polar granules do not have identical compositions (see Discussion).

Because GFP-Aub concentrates at the posterior pole of the oocyte and is found in polar granules, we reasoned that Aub localization should depend on the activity of genes required for polar granule assembly. stau and vas are two such genes with somewhat different roles in this process. Stau acts primarily in the localization and translation of osk mRNA, prerequisites for polar granule formation (Kim-Ha et al., 1991; Ephrussi et al., 1991; Kim-Ha et al., 1995; Micklem et al., 2000). Although Stau is concentrated at the posterior pole of the oocyte, it is not incorporated into the polar granules early in embryogenesis (Breitwieser et al., 1996). By contrast, Vas is a polar granule component and appears to be more directly involved in the formation and integrity of these structures (Schüpbach and Wieschaus, 1986; Hay et al., 1988a; Hay et al., 1988b; Hay et al., 1990; Lasko and Ashburner, 1990; Liang et al., 1994).

Not surprisingly, we find that GFP-Aub posterior localization is almost completely defective in stau mutants (Fig. 5A) and greatly reduced in vas mutants (compare Fig. 5D with 5F), confirming that Aub is localized by the same general mechanism implicated in the localization of other polar granule components. In addition, a second and equally striking defect occurs in only the vas mutants: perinuclear localization of GFP-Aub in the nurse cells is almost entirely eliminated (compare Fig. 5B with 5C). Taken in conjunction with the known perinuclear concentration of Vas protein (Hay et al., 1988a; Liang et al., 1994), this result raises the possibility that vas-dependent localization of Aub initiates in the nurse cells and involves a complex already containing both Vas and Aub proteins (see Discussion).

Our identification of Aub as a polar granule component suggests that it, like other such components, may be required for pole cell formation. Although aub mutants lack pole cells, this could be attributed to their deficiency in synthesis of Osk protein, a prerequisite for pole cell formation. Therefore, we used an osk-bcd transgene to bypass the requirement for aub in osk mRNA translation and assess the role of aub in pole cell formation. In this transgene, the osk 3′UTR is replaced with the bicoid (bcd) localization signal, and the encoded mRNA is localized to the anterior of the oocyte (Ephrussi and Lehmann, 1992). Eggs produced by aub⁺ females carrying an osk-bcd transgene synthesize Osk protein at both poles (Wilson et al., 1996; Fig. 6A,B). In embryos produced by aub mutants carrying the osk-bcd transgene, Osk protein is absent from the posterior pole but efficiently translated at the anterior, reflecting a difference in the requirement for aub for translation of the transgenic and endogenous mRNAs (Wilson et al., 1996; Fig. 6E,F). However, while most of the Osk protein at the anterior pole is taken up in ectopic pole cells in embryos from aub heterozygous mothers (Fig. 6C), Osk protein particles become dispersed over the cortex in the anterior part of embryos from homozygous aub mutant mothers (Fig. 6G) and fail to direct pole cell formation at the anterior pole (Fig. 6H). Thus, aub is required for the development of ectopic pole cells at the anterior of the embryo, and we infer that, like genes for other polar granule components, aub is involved in pole cell formation at the posterior as well.

Although the posterior concentration of Aub in oocytes might seem to contribute to its role in translation of osk mRNA, which occurs predominantly at that site, three observations are at odds with this inference.

First, in examination of more than 20 stage 8/9 egg chambers, we discovered that the posterior accumulation of GFP-Aub is slightly delayed from the appearance of Osk protein in the same region (Fig. 7A,B). In the oocytes, we often detected significant Osk protein when very little GFP-Aub concentration at the posterior had occurred. In contrast, we never observed an egg chamber that had localized GFP-Aub but only low levels of Osk protein. Thus, it appears that Aub is not concentrated at the posterior before the onset of osk translation.

Second, we do not detect any concentration of Aub in the anterodorsal region of the oocyte where gurken (grk) mRNA is localized (Fig. 7C) even though aub has also been implicated in the activation of grk translation (Wilson et al., 1996).

Finally, it is possible to greatly reduce the posterior concentration of GFP-Aub and still initiate translation of Osk at normal levels. This situation occurs in vas mutants. Although previous reports have shown that vas mutant ovaries accumulate reduced levels of Osk protein (Markussen et al., 1995; Rongo et al., 1995), we find that Osk initially appears at normal levels in vas mutants. The previous analyses relied on western blots of total ovarian protein, and so were unable to reveal any temporal specificity to the reduction. We used confocal microscopy to compare Osk protein levels in ovaries and embryos from vas mutant and heterozygous females. Up to stage 11 of oogenesis, before deposition of the vitelline membrane interferes with whole-mount antibody staining, vas mutants accumulate normal amounts of Osk (compare Fig. 5E with 5G). By contrast, early embryos from vas mutant mothers display dramatic reductions in Osk protein (compare Fig. 7D with 7E). Thus, Osk protein levels are affected in vas mutants only in late stages of oogenesis, either through reduced translation or stability. The fact that normal levels of Osk appear in stage 9-11 vas mutant oocytes despite greatly lowered levels of GFP-Aub posterior accumulation, together with our other observations, strongly suggest that the posterior concentration is not required for aub-dependent activation of osk translation.

Thus, the uniform low level of Aub found throughout the ooplasm appears to be sufficient for its action in osk and grk translation at the posterior pole and anterodorsal corner of the oocyte, respectively. The higher levels of Aub resulting from posterior recruitment presumably reflect other roles for Aub. These include two roles Aub shares with Vas: the posterior localization of nos mRNA (Wilson et al., 1996) and pole cell formation.

Previous analysis of aub implicated it in two processes: the translational activation of a narrowly restricted subset of ovarian mRNAs, including osk and grk, and the subsequent localization of nos mRNA (Wilson et al., 1996). The latter function could be indirect, with Aub activating translation of a factor directly involved in nos mRNA localization, or Aub might itself play a more direct role in the process. Our demonstration that Aub is related to the translation initiation factor eIF2C lends further support for a direct role of Aub in translational activation. Evidence that Aub is a component of polar granules and is required for pole cell formation adds an additional function and also suggests that its later role in localization of nos mRNA may be relatively direct.

The central and C-terminal portions of Aub contain two conserved regions, designated the PAZ and Piwi domains (Cerutti et al., 2000), which are encoded by a group of genes from organisms as diverse as plants, fungi and metazoans (including vertebrates). Recently, several of these genes have been characterized genetically and have been found to play essential roles in development. Both argonaute (ago1) and pinhead/zwille are required for maintenance of the axillary shoot meristem in Arabidopsis thaliana (Bohmert et al., 1998; Moussian et al., 1998; Lynn et al., 1999). In Drosophila, piwi has a demonstrated role in germline stem cell maintenance (Cox et al., 1998; Cox et al., 2000). Similarly, two Caenorhabditis elegans genes closely related to aub and piwi, prg-1 and prg-2, are also likely to be involved in germline proliferation (Cox et al., 1998). Other genes in the eIF2C/piwi family are implicated in mediating double-stranded RNA interference (RNAi) in C. elegans (rde-1; Tabara et al., 1999; Grishok et al., 2000) or the potentially related phenomena of post transcriptional gene silencing (PTGS) in Arabidopsis (ago1; Fagard et al., 2000) and quelling in Neurospora (qde-2; Catalanotto et al., 2000). The roles for ago1 in both PTGS and a cell fate decision reveal that a single gene in the family can carry out two functions, but it is not known if these functions are mechanistically distinct.

Although the genetic pathways in which these family members act have been identified, the direct activity through which any plays a role in development or RNAi has not been determined to date. The only gene product in the family for which a specific biochemical activity has been demonstrated is the translation initiation factor eIF2C (formerly Co-eIF-2A) (Zou et al., 1998). eIF2C purified from rabbit reticulocytes has two related activities that affect the ternary complex, which is composed of initiator methionine tRNA, GTP and eIF-2. The ternary complex binds the 40S ribosomal subunit to allow scanning for AUG codons in mRNA (for a review, see Hinnebusch, 2000). Purified eIF2C stimulates formation of the ternary complex from components present at physiological levels, and it stabilizes the complex against dissociation in the presence of natural mRNAs (Roy et al., 1981; Bagchi et al., 1985; Chakravarty et al., 1985; Roy et al., 1988). The known activity of eIF2C offers a trivial explanation of the roles of other family members in multiple processes: each could simply enhance the translation of a protein required for that particular process. This seems unlikely, in large part because Piwi has been shown to be nuclear in both the germline and somatic cells in which its activity is required (Cox et al., 2000). Furthermore, we argue here that Aub may play a direct role in nos mRNA localization. Nevertheless, the conserved domains that define this protein family should be expected to perform similar functions in the different proteins, and so some similarity to eIF2C function among other family members would not be surprising.

The Drosophila genome contains five members of the eIF2C/piwi family. One of these, dAGO1, is most closely related to eIF2C and is likely to be the Drosophila eIF2C homolog. Our identification of another family member as aub, a gene already implicated in translation, suggests that Aub protein may have biochemical activities similar to those of eIF2C. While eIF2C is predicted to play a global role in translation, mutants of aub only detectably affect the translation of a limited number of transcripts, revealing specificity in Aub function. We propose two broad classes of models for the role of Aub in message-specific translational activation. First, Aub may simply elevate the level of eIF2C-like activity in the ovary, leading to enhanced translation of mRNAs that depend most on this initiation factor. Differential dependence would presumably reflect some feature of mRNA structure, such as folding or AUG sequence context, or association of the mRNA with other proteins or regulatory factors. A variation of this model invokes association of Aub with dependent mRNAs, either directly through an RNA-binding activity or indirectly via binding to other factors, effectively elevating the local concentration of the eIF2C-like activity. Second, Aub may perform an activity distinct from that of eIF2C but still be involved in some aspect of translation. As with the first model, the activity could be concentrated on dependent mRNAs though direct or indirect binding. We have shown that a GFP-tagged form of Aub is not concentrated at the site of grk mRNA localization, and we have argued that posterior concentration of Aub is not required for osk mRNA translation. Thus, if Aub does associate with dependent mRNAs as part of its role in translation, the interaction is likely to be transient.

Efforts to define the feature of transcripts that makes them dependent on Aub are constrained by the limited number of such mRNAs which have been identified: at present only osk and grk. For osk mRNA, aub dependence is conferred in part by the 3′ UTR (Wilson et al., 1996), a region known to bind multiple proteins and mediate both mRNA localization and translational repression (Kim-Ha et al., 1993; Kim-Ha et al., 1995; Gunkel et al., 1998; Lie and Macdonald, 1999). Therefore, Aub-dependence may be conferred by some aspect of translational repression or mRNA localization. The features of grk mRNA that confer aub-dependence are unknown.

osk mRNA is repressed by Bru and BicC, probably in concert with other factors (Kim-Ha et al., 1995; Webster et al., 1997; Saffman et al., 1998; Gunkel et al., 1998; Lie and Macdonald, 1999). Bru also binds to grk (Kim-Ha et al., 1995), and this interaction has been suggested to mediate repression (Nilson and Schupbach, 1999). If repression by Bru is indeed a shared feature of the osk and grk mRNAs, then Aub might inhibit or override this process. However, elimination of Bru-mediated repression by mutation of the BREs, the Bru-binding sites in the osk mRNA 3′ UTR, does not abrogate the requirement for aub in translation of posteriorly localized osk mRNA (Wilson et al., 1996). Thus, Aub does not simply alleviate Bru-dependent repression at the posterior pole of the oocyte (although it may contribute to this process).

One feature that is unquestionably shared by the osk and grk mRNAs is localization to specific domains within the oocyte. Perhaps tenacious binding of localization factors places unusual constraints on the translational apparatus that can only be overcome through the action of Aub. Supporting this possibility is the curious requirement for Aub in translation of posteriorly localized oskBRE⁻ mRNA, as noted above, but not in the precocious translation that occurs prior to localization, when oskBRE⁻ mRNA is present throughout much of the oocyte (Wilson et al., 1996). Thus, posterior localization of osk mRNA correlates with the Aub-dependent phase of its translation. However, anterior localization of osk mRNA, achieved by exchanging the osk 3′ UTR with the bcd mRNA localization signal, obviates the need for Aub. Therefore, mRNA localization in and of itself need not impose the requirement for Aub, although different localization signals may have unique properties that affect the relative translatability of their transcripts.

We have shown that GFP-Aub colocalizes with Osk in polar granules in the pole plasm and pole cells. Moreover, as Osk and Vas are also largely colocalized in polar granules, we infer that the Aub-containing particles also include Vas. GFP-Aub is also highly enriched in the perinuclear zones of the nurse cells, a distribution shared by Vas (Hay et al., 1988a; Hay et al., 1988b; Liang et al., 1994) and Tud, another polar granule component (Bardsley et al., 1993) whose distribution we did not monitor in our experiments. Notably, vas is required for the presence of normal levels of GFP-Aub in both the polar granules and the perinuclear zones. Pole plasm components have been suggested to be transported in particles from the nurse cells to the posterior ooplasm (Mahowald, 1971; Hay et al., 1988a; Liang et al., 1994), and our results are consistent with such a scheme. The defect in GFP-Aub accumulation at the posterior in vas mutants may be a direct result of the initial failure to recruit Aub to the periphery of the nurse cell nuclei, a process that may well involve the assembly of complexes containing both Vas and Aub. In contrast, Stau is not required for the perinuclear localization of either Vas (Lasko and Ashburner, 1990) or GFP-Aub, and thus may only act later in the recruitment of particles to the posterior pole of the oocyte.

In the embryo, polar granules are present in the cytoplasm of pole cells, and nuclear bodies that are similar in appearance form in the nuclei of these cells (Mahowald, 1968; Mahowald et al., 1976). Evidence supporting a structural relationship between the two classes of particle has come from the demonstration that Vas is present in both (Hay et al., 1988a; Hay et al., 1988b). Our work provides additional supporting evidence of such a relationship: further indication of structural similarity comes from our demonstration that Osk and Vas are colocalized in both nuclear and cytoplasmic aggregates, some of which have the peculiar donut-like shape that is characteristic of a subset of polar granules and nuclear bodies. However, our data also reveal that the cytoplasmic and nuclear particles are not identical in composition, as the nuclear bodies are distinguished from the polar granules by the absence of GFP-Aub. Nevertheless, Aub, like Osk and Vas, is involved in pole cell formation, and all are predominantly or exclusively cytoplasmic during that process. It is possible that later aspects of germ cell development require certain polar granule components, such as Osk or Vas, to act in the nucleus. Future experiments in which the selective partitioning of the various polar granule and nuclear body components is perturbed may provide insight into how pole cell fates are specified and maintained or lost during development.

We thank Joanne Connell for the preliminary mapping of aub, and Ulrich Schäfer for transgenic flies, mutants and information provided prior to publication. We are grateful to Anne Ephrussi for providing anti-Osk antibodies, and vas-gfp and osk-bcd flies; Andrea Brand for the mGFP6 construct; Pernile Rorth for UASp; and Mark van Doren and Ruth Lehmann for nos-gal4-vp16 transgenic flies. We thank Eric Arn, Koen Braat and Jennifer Harris for critical comments on the manuscript. This work was supported by NIH Grant GM54409 to P. M. M.