A translation-independent role of oskar RNA in early Drosophila oogenesis

In many organisms, determination of body axis relies on molecular asymmetries established during oogenesis or early embryogenesis. In Drosophila melanogaster, maternally provided mRNAs and proteins are transported into the growing oocyte and stored until they are required at later stages of development.

During Drosophila oogenesis, a germline stem cell located at the anterior tip of the germarium divides to generate a new stem cell and a sibling cystoblast (reviewed by Gilboa and Lehmann, 2004; Huynh and St Johnston, 2004). The cystoblast undergoes four rounds of divisions with incomplete cytokinesis, giving rise to a 16-cell cyst consisting of 15 nurse cells and an oocyte interconnected by cytoplasmic bridges called ring canals. The nurse cells provide the oocyte with most of the mRNAs and proteins required for its development, and for the development of the future embryo until the onset of zygotic transcription (reviewed by Lawrence, 1992; Spradling, 1993). One of these mRNAs encodes the posterior determinant Oskar(Ephrussi et al., 1991; Kim-Ha et al., 1991). During early oogenesis, oskar RNA is exported from the nurse cells into the oocyte cytoplasm, where the RNA accumulates as a translationally silent transcript. During mid-oogenesis (stage 8), oskar mRNA is transported towards the posterior pole, leading to its asymmetric localization(Ephrussi et al., 1991; Kim-Ha et al., 1991). oskar mRNA is exclusively translated at the posterior pole, where it initiates assembly of the pole plasm(Kim-Ha et al., 1991; Markussen, 1995; Rongo et al., 1995).

oskar mRNA produces two Oskar isoforms, Long Oskar and Short Oskar, generated by the use of two alternative start codons, called M1 and M2(Markussen et al., 1995; Rongo et al., 1995). Together,the two Oskar proteins induce posterior pole plasm assembly and localization by recruiting the additional factors necessary for abdomen and germline formation in the future embryo (Ephrussi and Lehmann, 1992). Embryos from classical oskar mutant mothers fail to form posterior structures and lack germ cells(Lehmann and Nüsslein-Volhard,1986), because they fail to recruit Vasa protein and nanos (nos) mRNA to the posterior pole(Ephrussi et al., 1991; Hay et al., 1990; Lasko and Ashburner, 1990). Vasa is a highly conserved component of the germline, and is required for abdomen and germline formation in Drosophila(Schüpbach and Wieschaus,1986). Nanos, the abdominal determinant, acts by repressing translation of maternal hunchback mRNA in the posterior region of the embryo, allowing posterior activation of gap genes and, thus, formation of posterior structures (Gavis and Lehmann,1992; Wang and Lehmann,1991).

Classical oskar mutants were isolated in screens for maternal effect genes required for anteroposterior patterning of the embryo(Lehmann and Nüsslein-Volhard,1986). These classical alleles all express a significant amount of oskar mRNA, but lack functional Oskar proteins and thus produce embryos lacking germ cells and abdomen(Ephrussi et al., 1991; Kim-Ha et al., 1991). Here, we describe two new oskar mutant alleles showing a strong reduction or complete absence of oskar mRNA, respectively. Intriguingly, these new oskar alleles cause a different, stronger oskar phenotype:the early arrest of oogenesis leading to a complete failure in egg production. Using a rescue approach with a number of transgenes unable to produce Oskar protein, we show that the oskar mRNA transcript, but not the protein,is required for early oskar function. In particular, its 3′UTR is sufficient to overcome the early oogenesis arrest, thus revealing an unexpected function for oskar mRNA.

All Drosophila strains used in this study that have been previously reported are described in FlyBase. osk^A87 was isolated on a ru st e ca chromosome in an EMS mutagenesis screen using a sensitized genetic background (D. St J., unpublished). osk¹⁸⁷ was identified in a P element mutagenesis screen(Erdélyi et al., 1995). Although the mutations were isolated in EMS and P element mutagenesis,molecular analysis reveals that the osk^A87 and osk¹⁸⁷ mutations were induced by a ZAM(Leblanc et al., 1997)retrotransposon and an I element (Crozatier et al., 1988), respectively. Inverse PCR on a circularized SspI digest of genomic DNA purified from osk¹⁸⁷homozygous flies was performed by amplification with primers U8(5′-TTGCGCCTGC TCTTTCGCCTTC-3′) and L5(5′-TTTGTTCCCATTCGCCCACCAT-3′). The amplification product was cloned using a pCRScript kit (Stratagene) and an I element was identified in the sequence, which was obtained by sequencing using vector primers. osk^A87/Df(3R)p^XT103 genomic DNA was digested with SnaBI. Inverse PCR using L5 and oskSeq1 primers(5′-CGAAAAGCACCGTAAGTCTC-3′) was performed on the circularized template. The PCR product was cloned into Topo T/A plasmid (Invitrogen). Sequence analysis of the fragment revealed the insertion of a ZAM retrotransposon (Accession Number, AJ000387) in reverse orientation with respect to oskar transcription. Using ZAM element and the oskar-specific primers ZAM5′rev(5′-GTATGCGTTGTTCTGTCTGAG-3′) and circ osk 3′(5′-TAACTGCAGTTGGTCTTTTCATCCGTT-3′), the 5′ end of the ZAM element and the flanking oskar sequence were amplified. The insertion site of the ZAM element was precisely determined by cloning into Topo T/A vector and sequencing the PCR product.

Southern analysis of the osk^A87 mutant DNA revealed a band whose size precisely reflected the insertion of a ZAM retrotransposon(data not shown). Northern blot analyses were performed using poly(A)⁺ selected RNA and radiolabeled antisense oskar and rp49 RNA probes (Roche RNA labeling kit), according to standard protocols.

For RT-PCR analysis, total RNA from the abdomen of three to six females was extracted using the Absolutely RNA RT-PCR Miniprep kit (Stratagene) including a DNAse digest. Oligo-dT primed cDNA was synthesized with Superscript II(Invitrogen) according to standard procedures. To show the absence of osk transcript in osk^A87(Fig. 1C), RT-PCR was performed on cDNAs from osk^A87/Df(3R)p^XT103 and nos^A10/Df(3R)p^XT103 females with primers A87RT (5′-TTGCTGAGCCACGCCCAGAA-3′) and Bi_osk_control(5′-ACATTGGGAATGGTCAGCAG GAAATC-3′) for 40 cycles (annealing temperature 55°C). Primers bcd_up(5′-AACGAGCAAGAAGACGACGCTACAGTCTTG-3′) and bcd_rt(5′-GCGAATAGCGTATTGCAGGGAAAGTATAGA-3′) were used as positive control.

Quantitative real-time RT-PCR (Fig. 4) was performed on cDNAs from pCogGal4:VP16/w¹¹¹⁸; UAS osk-K10/+; osk^A87,NanosGal4:VP16/Df(3R)p^XT103,pCogGal4:VP16/UAS oskWT;; osk^A87,NanosGal4:VP16Df(3R)p^XT103, and pCogGal4:VP16/UAS osk3′UTR;; osk^A87,NanosGal4:VP16/Df(3R)^pXT103 females using SYBR Green1 chemistry (Molecular Probes) on an ABI PRISM 7900HT real-time PCR apparatus. oskK10 mRNA was amplified using primers oskK10_for(5′-CTCCTGTCTAATCAACGAAAGG-3′) and K10_rev(5′-TTGACCATGGGTTTAGGTATAATG-3′), and primers A87RT and circ osk 3′ were used to amplify total osk mRNA. For normalization, bcd mRNA was amplified using primers bcd_up(5′-AACGAGCAAGAAGACGACGCTACAGTCTTG-3′) and bcd_rt(5′-GCGAATAGCGTATTGCAGGGAAAGTATAGA-3′). Amplification efficiencies were comparable as determined using serial 10-fold dilutions of an initial osk PCR product as substrate.

Whole-mount antibody staining and in situ hybridization using fluorescent RNA probes were performed as previously described(Hachet and Ephrussi, 2004; Tomancak et al., 1998). For whole-mount antibody staining, antigens were detected using the following primary antibodies: mouse anti-BicD (a mix of monoclonals 1B11 and 4C2, 1:10 dilution; gift of Beat Suter), rabbit anti-Staufen (pre-adsorbed, 1:2000 final dilution), rat anti-Bruno (1:5000 dilution)(Filardo and Ephrussi, 2003),mouse anti-Orb (a mix of 48H and 6H4 monoclonals, 1:20 dilution; Developmental Studies Hybridoma Bank) and rabbit anti-Par-1 (1:40 dilution)(Tomancak et al., 2000). For fluorescent detection, Rhodamine- or FITC-coupled goat anti-mouse, rabbit or rat secondary antibodies were used (1:500; Jackson Immuno Research Laboratories).

P(mM1 mM2stop) was made in a two-step PCR reaction. The product of a first round of PCR, generated using primers linkerM2mut(5′-AGCGAGAACAACGGTACCATCATCGAG-3′; ATG→GGT) and osk54Hind(5′-AAGAAAGCTTTCAAACATAAAGCTACTCACTCCTACTCACTGATGCTCGATATCGTGATT-3′)was used as primer for a second round of PCR together with M1BssHII(ATG→CGC). The product was then reamplified with the outside primers M1BssHII(5′-TAGGATCCAAGAATATTGGATCACTTTCCTCCAAGCGCGCGCCGCAGTCACA-3′) and osk54Hind, then cloned into pBluescript (pBSNTL) and sequenced. The fragment was then used to replace the wild-type BamHI (oskarpromoter)-HindIII (first intron) fragment of pGem11go6.45, a pGEM11 vector with a 6.45kb XhoI-ApaI genomic DNA fragment encompassing the oskar locus. The mutated oskar gene was then transferred into pCasper4 (Pirrotta,1988) as an XhoI-NotI fragment (pCaspNTL).

To generate P(mM1SLmM2), the hairpin HP7, which blocks scanning by small ribosomal subunits (Kozak,1989a), was cloned as a blunt-ended BamHI-HindIII fragment into the blunt-ended SphI site of pBSNTL (pBSNTL HP7; orientation: reconstituted BamHI site proximal to the oskar transcription start site). The BamHI-XcmI fragment of pGem11go6.45 was then replaced by the corresponding fragment of pBSNTL HP7 and the BstXI site of the resulting plasmid was destroyed by cutting, filling and religation. This resulted in an additional frame shift (CCACTGG instead of CCACCTGG; sequence not canonical owing to fill-in artefact). The mutated oskar transgene was then transferred as an XhoI-NotI fragment into pCasper4,resulting in P(mM1SLmM2).

pUASp oskWT and pUASp oskΔi(1,2,3) were constructed by cloning genomic and cDNA versions of oskar as BamHI/NsiI fragments of pGem g.osk and pGem g/c.osk,respectively, into pUASp Casper (Rorth,1998) digested with BamHI/PstI. pGem g.osk was constructed by subcloning a 6.45 kb XhoI-ApaI fragment of oskar genomic DNA into pGem11Zf (Promega). pGem g/c.osk was constructed by replacing a 2425 bp BssHII-SacII fragment of pGem g.osk containing all of the oskar introns with the equivalent 2024 bp fragment of the oskar cDNA of Blue-osk(Ephrussi et al., 1991).

Trans-heterozygous oskar mutant females (A87/54, A87/84, A87/346,187/54, 187/84 and 187/346) were produced, as well as P(transgene)/SM6B; osk¹⁸⁷/Df(3R)p^XT103 and P(transgene)/SM6B; osk^A87/Df(3R)p^XT103 females [in which P(transgene) represents the various rescue constructs under osk promoter]. Rescue analysis using the UAS yeast inducible promoter was performed in the osk^A87/Df(3R)p^XT103background using pCog-Gal4:VP16 and Nanos-Gal4:VP16 drivers simultaneously.

Flies of the following genotypes were analyzed:

pCog-Gal4:VP16/UAS oskWT;; osk^A87,Nanos-Gal4:VP16/Df(3R)p^XT103;

pCog-Gal4:VP16/UAS oskΔi(1,2,3);; osk^A87,Nanos-Gal4:VP16/Df(3R)p^XT103;

pCog-Gal4:VP16/UAS osk3′UTR;; osk^A87,Nanos-Gal4:VP16/Df(3R)p^XT103;

pCog-Gal4:VP16/w¹¹¹⁸; UAS osk-K10/+; osk^A87,Nanos-Gal4:VP16/Df(3R)p^XT103.

Test females were collected as virgins and mated with Oregon-R(Fig. 3B), or w¹¹¹⁸ males (Fig. 3D). The egg-laying capacity of at least 30 individual test females from each experiment was monitored over four days, in egg-laying blocks on apple-juice agar plates, at 25°C. Values (including the wild-type controls) were normalized to the average of eggs laid per day per Oregon-R or w1118 female. The standard deviation was calculated from normalized values.

Two new oskar (osk) mutants, osk^A87and osk¹⁸⁷, were isolated in independent mutagenesis experiments. When transheterozygous with the oskar nonsense alleles osk⁵⁴, osk⁸⁴ and osk³⁴⁶(Kim-Ha et al., 1991; Lehmann and Nüsslein-Volhard,1986), both osk^A87 and osk¹⁸⁷ produce embryos that display the classical`posterior group phenotype' (lack of an abdomen and germline; data not shown;see also Fig. 3A,B) and thus fail to complement previously known oskar alleles. Molecular analysis revealed the presence of I (Crozatier et al., 1988) and ZAM (Leblanc et al., 1997) transposable elements in the upstream regulatory sequences or 1st exon of oskar in osk¹⁸⁷ and osk^A87, respectively(Fig. 1A). The position of these mutagenic elements suggested that they might affect the production or stability of oskar transcripts. Consistent with this hypothesis,northern analysis, in situ hybridization and RT-PCR of mutant osk¹⁸⁷ egg chambers confirmed that only residual amounts of oskar RNA are present, which are even further reduced in hemizygous animals (Fig. 1B,D;data not shown). In the case of osk^A87/Df(3R)p^XT103 egg chambers, we detected no oskar mRNA either by RT-PCR or by in situ hybridization. osk^A87 therefore is a true RNA null allele of oskar (Fig. 1C,D).

Surprisingly, and in contrast to females carrying the previously known nonsense alleles of oskar, osk⁵⁴, osk⁸⁴ and osk³⁴⁶, which do not express Oskar protein (as judged by western blots) and are thus considered strong loss of function alleles(Markussen et al., 1995; Rongo and Lehmann, 1996),females carrying only an osk^A87 or osk¹⁸⁷ allele[osk^A87/Df(3R)p^XT103,osk¹⁸⁷/Df(3R)p^XT103] fail to lay eggs and are sterile, owing to an early arrest of oogenesis. In both mutants, oocytes are determined and begin to develop, as indicated by the accumulation of BicD(Suter and Steward, 1991), Orb(Lantz et al., 1994), Bruno(Webster et al., 1997) and Par1 (Vaccari and Ephrussi,2002) in the oocyte (Fig. 2D,F,H,J). Only Staufen, a RNA-binding protein(St Johnston, 1992), the localization of which within the oocyte is interdependent with that of oskar mRNA (St Johnston et al.,1991), is not detected in osk^A87/Df(3R)p^XT103 oocytes(Fig. 2B), indicating that Staufen accumulation in the oocyte is mediated by oskar mRNA. Defects in osk^A87/Df(3R)p^XT103 egg chambers first become evident at stage 2, when fragmentation of the normally compact karyosome is observed (Fig. 2L). Mutant egg chambers continue to develop until stage 7, when they begin to degenerate.

To further prove that the oogenesis defects are due to mutations in oskar, we performed genetic rescue experiments using three oskar transgenes, encoding either both of the Oskar isoforms(Markussen et al., 1995; Rongo et al., 1995), or each isoform individually (Fig. 3A). The first transgene, P(osk+)(Markussen et al., 1995),consists of a genomic DNA fragment that encompasses the oskar locus and encodes both Oskar isoforms. P(M1L) contains a mutation in M1,the first translation initiation site in oskar mRNA, and therefore produces only Short Oskar, the isoform responsible for pole plasm formation(Markussen et al., 1995). Conversely, P(M139L) produces only Long Oskar, owing to a mutation in M2, the second translation initiation site in oskar mRNA(Markussen et al., 1995). Long Oskar is essential for the cortical anchoring of oskar RNA and thus for correct localization of the pole plasm, but fails to rescue the abdominal and germ cell defects of the oskar protein null mutants(Markussen et al., 1995; Rongo et al., 1995; Vanzo and Ephrussi, 2002). All three transgenes, P(osk+), P(M1L) and P(M139L),fully complement the egg-less phenotype of osk^A87/Df(3R)p^XT103 and osk¹⁸⁷/Df(3R)p^XT103, indicating that these are indeed oskar alleles(Fig. 3B).

The surprising observation that the osk⁵⁴,osk⁸⁴ and osk³⁴⁶ nonsense mutant alleles(Kim-Ha et al., 1991) rescue the early oogenesis defect of the new oskar mutants (giving rise to the `posterior group phenotype'; see also Fig. 3A,B), suggested that the early function of oskar might be mediated by oskar RNA,rather than by Oskar protein. To rule out the possibility that a truncated,unstable and thus undetectable Oskar peptide that is responsible for rescue of the oogenesis defects of osk^A87 and osk¹⁸⁷ is produced by the nonsense alleles, we constructed two translationally incapacitated, protein null oskar alleles, and tested their ability to rescue the new alleles. The first construct, P(mM1 mM2stop), consists of an oskar gene identical to the osk⁵⁴ nonsense allele, but whose capacity to produce a short peptide initiating from M1 or M2 was abolished by mutation of M1 and M2 to CGC and GGT, respectively (Fig. 3A). The second construct, P(mM1SLmM2), also containing a mutated M1 and M2, was additionally designed to prevent the initiation of translation by blocking scanning small ribosomal subunits, as well as to abolish putative translation of Oskar peptides that might initiate from in-frame methionine codons elsewhere in the oskar transcript. To this end, a sequence predicted to adopt a stable hairpin structure and that has been shown to block translation by stalling scanning ribosomes(Kozak, 1989b) was inserted between mutated M1 and M2, and a frame-shift was introduced downstream of M2(Fig. 3A). Both P(mM1 mM2stop) and P(mM1SLmM2) fully complement the egg-less phenotype of osk^A87/Df(3R)p^XT103 and osk¹⁸⁷/Df(3R)p^XT103, to the same extent as the original P(osk+) transgene(Fig. 3B). However, the embryos produced lack an abdomen, confirming the absence of Oskar protein. These results demonstrate that no feature of Oskar protein is required for early oogenesis, indicating that this function of oskar is mediated by another aspect of the gene.

All constructs that rescue the oogenesis arrest phenotype of the oskar RNA null alleles (see above) contain the endogenous oskar promoter, the coding region, the introns and the oskar3′UTR. We therefore wanted to address which region of the oskarlocus is required for its function in early oogenesis. To exclude the promoter region, we made use of a set of transgenes in which the oskarpromoter was replaced by the yeast UAS promoter, thus placing the oskar gene under Gal4 transcriptional control(Brand and Perrimon, 1993). We first analyzed the UAS oskWT transgene which only differs from the P(osk+) transgene by the replacement of the oskar promoter region by the UAS promoter (Fig. 3C). When driven by a combination of pCogGal4:VP16 and NosGal4:VP16 drivers, UAS oskWT rescues the early oogenesis arrest of osk^A87/Df(3R)p^XT103 females, as well as the posterior group phenotype of the progeny(Fig. 3D; data not shown). This rules out an involvement of the promoter region in oskar rescue activity and demonstrates that these drivers are sufficient to drive expression of a rescuing transgene. To assess a possible role of oskar intronic sequences in early oskar function, we tested the rescue ability of a UAS oskar transgene called UAS oskΔi(1,2,3), in which the three oskarintrons were deleted (Fig. 3C). osk^A87/Df(3R)p^XT103females expressing oskΔi(1,2,3) produce a normal number of eggs (Fig. 3D),indicating that the oskar introns are not essential for early oskar function.

We then tested whether the oskar-coding region or the 3′UTR are required to rescue the early oogenesis phenotype. Expression of a UAS osk-K10 transgene in which the oskar 3′UTR was replaced by that of K10 (Fig. 3C) (Riechmann et al., 2002) under the control of the same Gal4 drivers successfully used above, is unable to rescue the early oogenesis defect of the new oskar alleles, even when the flies are raised at 29°C in order to increase the expression level of the transgene(Fig. 3D and not shown). To confirm the functionality of this transgene, we tested whether osk^A87/osk⁺ heterozygous flies overexpressing UAS osk-K10 produce delocalized Oskar activity during embryogenesis. Indeed, 83% of embryos laid by such females are bicaudal, showing that the UAS osk-K10 transgene is functional. Furthermore, real-time PCR on cDNA of ovaries from osk^A87/Df(3R)p^XT103females expressing the UAS osk-K10 transgene under control of both Gal4 drivers confirmed that the UAS osk-K10 transgene is expressed at early stages of oogenesis (before stage 7 when oogenesis arrests; Fig. 4C; see 4A and 4B for schematic and exact genotypes). The transcript is, however, present at lower levels than the rescuing oskWT transcript, probably because of the degeneration of the ovaries (note that in wild-type background, the transcript levels of UAS oskWT and UAS oskK10 are similar; data not shown). Nevertheless,antibody staining showed that the UAS osk-K10 RNA is translated well before oogenesis arrests (Fig. 4D). The fact that UAS osk-K10 fails to rescue early oogenesis confirms our observation that Oskar protein does not provide early oskar function. In addition, it indicates that the oskar3′UTR might provide the early oogenesis function of oskar.

We, thus, directly tested the capacity of the oskar 3′UTR to rescue the osk^A87/Df(3R)p^XT103 early oogenesis arrest. Remarkably, expression of the oskar 3′UTR alone from the UAS osk3′UTR transgene(Fig. 3C)(Filardo and Ephrussi, 2003)driven by the combination pCogGal4:VP16 and NosGal4:VP16 is sufficient to rescue the early oogenesis and egg-less phenotypes(Fig. 3D). As expected, the 3′UTR also rescues the karyosome defect(Fig. 5D), but does not rescue the late oskar phenotype of the resulting embryos, which display the`posterior group phenotype' (data not shown). This demonstrates that the function of oskar during early oogenesis is mediated by oskar RNA, independent of Oskar protein, and demonstrates that the oskar 3′UTR is sufficient to perform this function.

We also analyzed the localization of Staufen in the ovaries of osk^A87/Df(3R)p^XT103 females expressing either the UAS osk3′UTR or the UAS osk-K10transgene. In the case of UAS osk-K10, no Staufen protein was detected in the oocyte, indicating that Staufen fails to associate with osk-K10 RNA (Fig. 5B). By contrast, expression of the oskar 3′UTR alone was sufficient to restore Staufen accumulation in the oocyte, showing that Staufen associates with oskar RNA though its 3′UTR(Fig. 5C). However, during late oogenesis, Staufen and oskar 3′UTR RNA fail to localize at the posterior of osk^A87/Df(3R)p^XT103 oocytes[compare Fig. 5F,H with 5E,G (UAS osk WT)]. These observations confirm our previous results showing that the oskar 3′UTR is not sufficient for oskar RNA localization at the posterior of the oocyte(Hachet and Ephrussi, 2004). It therefore appears that the function supplied by the oskar 3′UTR for oocyte progression past early oogenesis is independent of the assembly of the oskar RNP localization complex that mediates its transport to the oocyte posterior in midoogenesis.

The new oskar alleles osk^A87 and osk¹⁸⁷ reveal that oskar plays a crucial role during early oogenesis. This early function of oskar is unusual, as it is mediated by oskar RNA, and not by Oskar protein. The low but detectable amount of oskar RNA observed in osk¹⁸⁷mutants indicates that a threshold exists, below which egg chambers fail to develop. Our findings raise the possibility that many proteincoding genes may fulfill additional non-coding functions and thus increase the spectrum of gene function.

What might the function of oskar RNA be during early oogenesis?Oskar protein serves as a scaffold for the assembly of cytoplasmic structures essential for germline development, the polar granules, at the posterior pole of the oocyte and embryo. During early oogenesis, oskar RNA might provide a similar scaffold function for assembly of cytoplasmic complexes essential for the progression of oocyte development. RNAs are associated with proteins in ribonucleoprotein complexes during most of their existence, from their emergence as nascent transcripts in the nucleus, during nucleo-cytoplamic transport, to their final cytoplasmic localization,translation or degradation (Shyu and Wilkinson, 2000). Staufen protein requires oskar RNA for its transport from the nurse cells into the oocyte(Fig. 2B). In addition, we have shown that it is the oskar 3′UTR that mediates accumulation of Staufen protein in the oocyte (Fig. 5). This reveals that the well-known mutual interdependence of Staufen and oskar mRNA in their localization during oogenesis is mediated by interaction of Staufen with the oskar 3′UTR. Of the candidate proteins we examined, only Staufen showed a clearly altered distribution in the oskar RNA null mutant, yet Staufen itself does not appear to play a role during early oogenesis(St Johnston et al., 1991). It is therefore reasonable to assume that oskar RNA acts as a structural partner for the transport into the oocyte of additional, so far unidentified proteins or RNAs essential for its development. In this regard it is interesting to note that both VgT RNA and the non-coding Xlsirts RNA have been shown to mediate anchoring of several RNAs at the vegetal pole of the Xenopus oocyte(Heasman et al., 2001; Kloc and Etkin, 1994; Kloc et al., 2005).

Alternative functions of the oskar 3′UTR are also plausible. In particular, the oskar 3′UTR might bind and sequester a negative regulator that, in its free form (i.e. in an oskar RNA null background), inhibits early oogenesis. One candidate that has been shown to bind to the oskar 3′UTR is the translational regulator Bruno(Kim-Ha et al., 1995). However, overexpression of Bruno - at least in the presence of wild-type levels of oskar mRNA -does not cause a phenotype similar to that of the oskar RNA null mutant (Filardo and Ephrussi, 2003). Thus, to fully understand the mechanism of oogenesis arrest resulting from absence of oskar mRNA, it will be important to identify other proteins and RNAs binding to the oskar3′UTR that are required for egg chamber development.

We thank Elisa Wurmbach for help with QRT-PCR and Beat Suter for anti-BicD antibodies. The monoclonal antibodies orb6H4 and orb4H8 developed by Paul Schedl were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa,Department of Biological Sciences, Iowa City, IA 52242, USA. This work was supported by grant T048393 from the Hungarian National Science Foundation(OTKA) to M.E. A.J. was supported by fellowships from the Swiss National Fonds and EMBO and O.H. by a fellowship from a predoctoral `Allocation de Recherche'from the French government. A.J. and P.Z. also received support from a Human Frontier Science Program grant to A.E.