poirot, a new regulatory gene of Drosophila oskar acts at the level of the short Oskar protein isoform

The embryonic polarity of many organisms is defined by maternally provided determinants that are localised in the oocyte during egg development. Localisation and the subsequent site-specific translation of mRNAs provides one of the mechanisms that restrict the biological functions of determinants to specific cytoplasmic regions within the oocytes. This type of regulation has been studied most extensively in Drosophila melanogaster, as the anteroposterior and the dorsoventral embryonic body axes, and germline differentiation are governed by maternally provided localised determinants.

Post-transcriptional regulation of the localised maternal determinants is achieved by different molecular mechanisms. For example, the anterior morphogen bicoid (bcd) is translated only after fertilisation and its translation is regulated by cytoplasmic polyadenylation, a common mechanism of maternal mRNA regulation (Salles et al., 1994). Translation of the posterior morphogen nanos (nos), however, is independent of polyadenylation, and is subject to a more complex regulatory process. Unlocalised bulk cytoplasmic nos mRNA is repressed, but localisation of the nos mRNA by components of the posterior localised germ plasm activates its translation by preventing the interaction of nos mRNA with translational repressors (Dahanukar et al., 1999; Dahanukar and Wharton, 1996; Smibert et al., 1996). Gurken, the key organiser of the dorsoventral polarity of the developing egg and embryo, is also controlled by both positive and negative regulators. According to a recent model reviewed by Cooperstock and Lipshitz (Cooperstock and Lipshitz, 2001), grk mRNA is bound by cytoplasmic translation repressor molecules, but in the anterodorsal corner of the oocyte, where grk is functional, positive translational regulators relieve the effect of repressors. Finally, the post-transcriptional regulation of oskar (osk), the key component of the germ plasm assembly, is an even more complex process. Besides repression-derepression events that are similar to the grk and nos regulation, the OSK protein itself has a positive self-regulatory function. Additionally, it has been shown that the osk mRNA is subject to cytoplasmic polyadenylation (Chang et al., 1999). The complexity of its regulation makes osk one of the most attractive subjects for the analysis of gene regulation in Drosophila melanogaster.

osk mRNA is transcribed exclusively in the nuclei of nurse cells and then transported into the developing oocyte (Ephrussi et al., 1991; Ephrussi and Lehmann, 1992; Smith et al., 1992). Initially, during oocyte development, the osk transcript is uniformly distributed (stages 1-6), then it is transiently localised (stage 7) to the anterior pole, and subsequently (after stage 8) to the posterior pole, where OSK protein is first translated. Here, it defines the place of germ cell and abdomen formation. (Ephrussi et al., 1991; Kim-Ha et al., 1991). It has also been shown that, although osk mRNA is concentrated at the posterior pole, a significant amount of the osk transcript remains unlocalised (Bergsten and Gavis, 1999). This clearly indicates that mRNA localisation alone is not sufficient for the precise restriction of osk function to the most posterior part of the oocyte; instead, the translation and post-translational regulation ensures the wild-type spatial restriction of osk activity.

Translational regulation of osk mRNA involves the activity of repressors, derepressors, translational activators and the OSK protein itself. It has been demonstrated that unlocalised osk is translationally repressed. Three RNA-binding proteins Bruno (Bru), apontic (apt) and Bicaudal-C (Bic-C) have been identified, which play a role in this repression (Kim-Ha et al., 1995; Webster et al., 1997; Lie and Macdonald, 1999; Saffman et al., 1998). Translational activation of posteriorly localised osk mRNA is achieved by derepressors that can override the effect of negative regulators. The 5′ region of osk mRNA contains a derepressor element, whose deletion prevents translation of posteriorly localised osk mRNA (Gunkel et al., 1998). By in vitro RNA-binding assays, two protein species p50 and p68 were identified that specifically bind to this derepressor element (Gunkel et al., 1998). Structural and functional analysis of the dsRNA-binding protein staufen (stau) revealed that its dsRBD5 domain is involved in the derepression of osk mRNA at the posterior pole (Micklem et al., 2000). Several other positive trans-regulators of the osk translation have also been identified. Orb, which is homologous to the Xenopus cytoplasmic polyadenylation element binding protein (CPEB) activates osk mRNA translation, most probably by promoting polyadenylation (Chang et al., 1999). aubergine (aub) enhances osk translation through an interaction with 3′UTR and sequences upstream of 3′UTR of osk mRNA (Wilson et al., 1996). Translation of osk mRNA also requires the DEAD box RNA helicase, Vasa (VAS), which has been shown to interact with BRU; thus, VAS may activate osk mRNA translation by blocking Bru function. However, it has been shown that VAS is required for the translation of osk mRNA in the absence of BRU repression, which suggests that vas overcomes not only Bru-mediated repression (Rongo et al., 1995; Webster et al., 1997; Markussen et al., 1995).

As osk translation depends on posterior localisation of osk mRNA, genes that are involved in osk mRNA localisation also indirectly regulate osk translation (Cooperstock and Lipshitz, 2001). Investigation of osk mis-sense and nonsense mutations has revealed that the OSK protein is required to maintain osk mRNA at the posterior pole (i.e. a positive-feedback mechanism exists) (Ephrussi et al., 1991; Kim-Ha et al., 1991; Markussen et al., 1995; Rongo et al., 1995). osk mRNA is translated into long OSK (71 kDa) and short OSK (55 kDa) proteins, of which the short isoform undergoes phosphorylation, which produces a 57 kDa protein (Markussen et al., 1995; Rongo et al., 1995). Phenotypic analysis of in vitro constructed osk transgenes, which express either short or long OSK proteins, has demonstrated that both isoforms are effective in maintaining osk mRNA at the posterior pole, but only the short OSK is able to organise germ plasm assembly (Breitwieser et al., 1996; Markussen et al., 1995). Several other osk regulatory genes, which have a role in mRNA localisation, translational repression, derepression and activation, have been identified; however, no factors that play a role in anchoring OSK protein to the posterior pole have yet been described.

If osk regulation occurs correctly, the function of the OSK protein is restricted to the posterior pole, where it recruits components of the germ plasm. Absence of OSK protein from the posterior pole leads to posterior phenotypes, when embryos develop without abdomens and germ cells. Furthermore, osk gene dose experiments have shown that osk is a limiting factor in determining the number of germ cells and NOS protein activity (Ephrussi and Lehmann, 1992; Gavis and Lehmann, 1994; Smith et al., 1992). The two related osk phenotypes (germ cell-less and abdomenless) have different sensitivity to the germ plasm concentration. Moderate level of the localised OSK protein results in a failure of germ cell formation, while abdomen formation is intact (Erdélyi et al., 1995; Jankovics et al., 2001). Thus, the maternal effect germ cell-less, which is also referred to as the grandchildless (gs) phenotype, provides a very sensitive genetic selection system by which osk regulatory genes can easily be identified.

In this paper, we report on the isolation and characterisation of a novel osk regulatory gene poirot (prt). We show that the lack of prt activity results in delocalisation of OSK protein from the posterior pole but has no effect on osk mRNA. Our results show that the prt mutant allele acts in an isoform-specific manner, only the short OSK isoform is affected while the long OSK isoform remains at the wild-type level.

Flies were kept on standard yeast-cornmeal medium. Crosses were performed at 25°C unless otherwise stated. Oregon-R flies were used as wild type. Df(2R)XTE58, Df(2R)XTED1, Df(2R)l4 and Df(2R)JP1 and T(2;3)Ta^L mutant chromosomes are described by Underwood and co-workers (Underwood et al., 1990). Other mutations and balancer chromosomes are described by Lindsley and Zimm (Lindsley and Zimm, 1992). Seven hundred and fifty homozygous viable hobo H[pHlw2] (Smith et al., 1993) insertion lines were screened in total for grandchildless phenotypes from an unpublished mutant collection from the laboratory of István Kiss (kindly provided by István Kiss’s laboratory). Homozygous females from each mutant line were crossed with Oregon-R, wild-type males and germ cell-less phenotypes of their adult progeny were recognised by hand dissection. prt^gs allele was identified as an incomplete penetrant grandchildless mutation.

H[pHlw2] hobo insertion was remobilised from prt gene by the hobo transposase bearing P[ry⁺ HBL1] transgene (Calvi et al., 1991; Smith et al., 1993). y w/y w; prt^gs/CyO females were crossed to CyO P[ry⁺ HBL1]/Bc Elp males. Individual y w/Y; prt^gs/CyO [ry⁺ HBL1] males and y w/y w; prt^gs/CyO P[ry⁺ HBL1] females were crossed to w/w; SM6b/Sco and w/Y; SM6b/Sco males, respectively. Next, SM6b balanced revertant (prt^gsR) stocks were established from white eyed y w/Y; prt^gsR/SM6b progeny. Revertants were selected by the loss of the grandchildless phenotype.

Embryonic cuticle preparations were made as described by Wieshaus and Nüsslein-Volhard (Wieshaus and Nüsslein-Volhard, 1986). Eggs from mutant females were collected, dechorionated with 50% Chlorox bleach, washed, mounted in Hoyer’s medium lactic acid 1:1 mixture and cleared for 24 hours at 60°C.

The C-terminal coding sequences of prt were cloned into pGEX-4T-1 and Glutatione S-transferase (GST)-fusion proteins were produced in E. coli strain BL21 using standard induction conditions. GST-PRT proteins were purified by Glutatione Sepharose 4B, according to the manufacturer’s instructions (Amersham Pharmacia Biotech). Polyclonal antibody was raised in rabbit against bacterially expressed and purified GST-PRT.

Hand dissected ovaries were sonicated in 10 volumes of extraction buffer (6 mM Tris-Cl, pH=6.8, 6.4% glycerol, 2% SDS, 100 mM DTT and Bromophenol Blue). Extract equivalent of approx. one pair of ovaries was loaded per lane on 10% SDS-PAGE gels. Proteins were transferred to PVDF membrane (Amersham) at ∼7 V/cm for 12-18 hours in cold transfer buffer (20% methanol, 25 mM Tris-Cl, 192 mM Glycine). BioRad Kaleidoscope Prestained Standards were used as molecular weight markers. Blots were blocked using 5% dry milk in TTBS, and incubated with anti-OSK antibody diluted 1:1000, anti-PRT antibody diluted 1:3000, anti-γ-tubulin (Sigma) antibody diluted 1:1000 and in TTBS 5% dry milk. The membrane was washed in dH₂O, TTBS and incubated with peroxidase-conjugated goat anti-rabbit secondary antibody in 5% dry milk in TTBS (Jakson Immuno Research Laboratories). Signals were detected with enhanced chemiluminescence.

Flanking sequences of the prt^gs hobo insertion were amplified by adapter-PCR technique. Genomic DNA was cleaved with AatII restriction enzyme and ligated to a synthetic AatII adapter composed of 5′-ACCAGCTAAACGCAACCCTAAGACGT-3′ (AD1) and 5′-CTTAGGGTTGCGTTTAGCTGGT-3′ (AD2) synthetic oligomers. DNA fragments were amplified with hobo element-specific lacZ-F1 (5′-GGATAGGTTACGTTGGTGTAG-3′) and AD1 primers. One tenth of the PCR products was reamplified using a pair of nested primers: lacZ-F2 (5′-TCGCACTCCAGCCAGCTTTCCGG-3′) and AD1. PCR products were separated by agarose gel electrophoresis, purified from the gel and subcloned into a pBR322 cloning vector. The insert was sequenced with a ABI Prism 310 Sequencer (Perkin Elmer). ORF finder and Blast programs (BDGP BLAST service) were used to analyse the sequences. The UASp-prt construct was generated by inserting the 2.2 kb HL01519 EST of BDGP (Rubin et al., 2000) into a pUASp vector (Rorth, 1998) and transforming into the Drosophila germline, according to the standard protocol (Rubin and Spradling, 1982). The UASp-prt-gfp construct was made by inserting the coding region of prt cDNA (HL01519) in frame to GFP in β-globin 2×GFP vector (J. Knoblich, personal communication) and prt-gfp cloned into Drosophila pUASp vector. Stable w/w; UASp-prt/TM3 transformant lines were established and crossed to nosGAL4-VP16 (Van Doren et al., 1998) germline-specific drivers at the prt^gs homozygous background.

Total RNA was Trizol extracted (Gibco BRL). cDNA was synthesised on 3 μg of total RNA template with M-MuLV Reverse Transcriptase (Fermentas) and random hexanucleotid primers (Amersham). Two primers complementary to the first and fourth prt exons (PRT1F, 5′-TAATACCTGCTGCTGTTACCCGCA-3′; PRT4R, 5′-GGCCCCCTAAGGTTGCTCACTATT-3′) were used in PCR amplifications and a control PCR was performed using rp49 primers (rp49 forward, 5′-GCATACAGGCCCAAGATCCGT-3′; rp49 reverse, 5′-CAATCTCCTTGCGCTTCTTG-3′) in the same reaction. The chimaeric prt-mini-white splice product was amplified by PRT1F and mini-white 3. exon specific (white3R: 5′-GTGTGCTGACATTTGCTGA-3′) primers. RT-PCR products were separated by agarose gel electrophoresis. Results were evaluated by using the Gel Base program. RT-PCR products were direct sequenced with ABI Prism 310 Sequencer (Perkin Elmer).

Bleach dechorionated embryos and dissected ovaries were collected in PBS. Ovaries were fixed for 10 minutes in a mixture of 1 volume fixation buffer (100 mM KH₂PO₄ pH=6.8, 450 mM KCl, 150 mM NaCl, 20 mM MgCl₂), 3 volumes dH₂O and 2 volumes 16% formaldehyde. Embryos were fixed for 12 minutes in a mixture composed of one volume of fixation buffer (0.1 M PIPES, 2 mM MgSO₄, 1 mM EGTA pH=6.8), 1/10 volume of 37% formaldehyde and one volume of heptane. The aqueous phase was removed and one volume methanol was added. Ovaries and devitellinized embryos were washed in PBT for 2×20 minutes and blocked for 1 hour in blocking solution (PBS plus 0.1% BSA, 0.1% Triton X-100, 5% normal goat serum and 0.02% NaN₃). Ovaries and embryos were then incubated overnight with rabbit anti-OSK primary antibody (1:500 dilution) or rabbit anti-STAU (1:2000 dilution) and rat anti-VAS antibody (1:750 dilution) in blocking solution. This was followed by a 4×30 minute wash in PBT and an hour wash in blocking solution. The final incubation was with a fluorescein-conjugated anti-rabbit secondary antibody diluted 1:200 (Jakson Immuno Research Laboratories) or anti-rat peroxidase-conjugated secondary antibody 1:100 (Amersham) in blocking solution for 2 hours, following which, ovaries and embryos were washed for 4×30 minutes in PBT. The VAS signal was visualised by DAB detection. Preparations were mounted in 80% glycerol and 4% n-propyl-gallate. The ovaries and embryos were analysed with Zeiss Axioscope II microscope, Axiocam CCD camera and Zeiss LSM 410.

Digoxigenin-labelled DNA probe was synthesised using a Dig DNA labelling kit (Boehringer Mannheim). The osk DNA probe corresponds to the 2.1 kb SacI fragment of osk cDNA (Ephrussi et al., 1991). Hybridisation was carried out as described in by Ephrussi and co-workers (Ephrussi, 1991). Hybridisation signals were detected using the Dig Detection kit (Boehringer Mannheim). Hybridisation was analysed with interference contrast microscopy (Zeiss Axioscope II).

Seven hundred and fifty viable autosomal hobo insertion lines were screened and a single line (prt) was found, of which the females, homozygous for the hobo insertion, exhibited the grandchildless (gs) phenotype. prt^gs/prt^gs females, when mated with wild-type males, gave rise to 70% agametic adult progeny. This incomplete penetrant gs mutation frequently resulted in mosaic gonads. In 10% of the progeny, mosaic pairs of gonads were found when one ovary or testis was agametic, while the other gonad was full of developing egg primordia (data not shown). This observation suggests that the mutation interferes with the number but not the function of adult germ cells, and the reduction of germ cells happened at an early development phase. Indeed, we observed dramatically reduced or completely missing germ cells in 70% of embryos from homozygous prt^gs females (Fig. 1A,B). Comparing the penetrance of the embryonic and adult phenotypes, we concluded that the germ cell-less phenotype observed in adulthood is a sole consequence of the failure in the earliest steps of embryonic germ cell development. This was further confirmed by the stronger phenotype observed at 18°C. Five percent (n=357) of the embryos originated from prt^gs/prt^gs females and wild-type males showed variable but characteristic abdomenless phenotypes (Fig. 1C,D) coupled with a slightly elevated, 85% adult gs phenotype at 18°C. This indicates that the origin of the embryonic germ cell-less phenotype is mutual with that of the abdomen defect and this is a consequence of the misfunction of the early (maternally) acting posterior gene hierarchy.

However, when prt^gs/prt^gs females were mated with prt^gs/prt^gs males, the resulting maternal and zygotic mutant progeny exhibited a more complex mutant phenotype. Abdominal defects characteristic to posterior group mutants were coupled with variable embryonic defects manifested in cuticle holes that were most frequently observed in the head region (data not shown). By crossing to wild-type males, the paternal rescue of zygotic cuticle defects allows us to investigate only the maternal effect of prt mutations. We also observed that egg-laying capacity of the prt^gs homozygous females was only 33% and 30% of wild-type females at 25 and 18°C, respectively, indicating that besides its zygotic function prt has another, early function during oogenesis (Table 1). In this paper we concentrate on the prt gene function in the germ cell formation.

The prt^gs hobo insertion was mapped to the 51 D6-12 chromosomal region by in situ hybridisation. Uncovering deficiencies [Df(2R)XTE58, Df(2R)Jp1 and Df(2R)l4] from this region did not complement the prt^gs grandchildless phenotype. We observed that penetrance of prt^gs homo- and hemizygous phenotypes was almost identical (70% and 66%, respectively), indicating that prt^gs is most likely to be a null allele. With other deficiencies [Df(2R)XTED1 and Df(2R)X9] from this region, which did complement the grandchildless phenotype, the prt^gs mutation was mapped to the 51D12-51E5-7 chromosomal region.

To prove that the hobo insertion is responsible for the grandchildless phenotype, we remobilised the insertion from the prt locus. Out of 455 independent lines, five revertants were isolated by the complete loss of the grandchildless phenotype. The revertants have proved that prt^gs is a hobo-induced mutation, and this made the gene accessible for molecular analysis.

In order to clone the prt gene, a flanking genomic DNA sequence of the hobo insert was amplified by an adapter PCR strategy (see Materials and Methods). The resulting 1.5 kb PCR hobo flanking sequence was identical to a part of the 72 kb sequenced P1 phage, DS04940, which maps to the 51D-E region (Hartl et al., 1994). The insertion point was precisely mapped at 34,569 bp in P1 phage sequence, in the first intron of the CG7761 (Adams et al., 2000; Rubin et al., 2000) annotated gene (Kimmerly et al., 1996) that we named poirot (prt). We collected four available partially sequenced CG7761 EST-s, completed the sequence of HL01519 and determined the exon-intron boundaries within the genomic DNA (Fig. 2A).

In order to prove that the prt gene is responsible for the grandchildless phenotype we performed a phenotypic rescue experiment. HL01519 cDNA was cloned into a pUASp vector, transformed into Drosophila and expressed exclusively in the female germline by the nosGal4Vp16-UASp expression system (Rorth, 1998; Van Doren et al., 1998). The UASp-HL01519 transgene completely rescued the phenotype of prt^gs homozygous females when nosGAl4Vp16 was present. In the absence of GAL4Vp16 trans-activation, we observed characteristic grandchildless phenotype. Based on this result, we concluded that the prt^gs mutation is germline dependent and the HL01519 cDNA contains the entire prt function.

Conceptual translation of prt cDNA predicted a 477 amino acid protein that shows extensive homology to human and mouse Sab proteins (Fig. 2B) (Matsushita et al., 1998; Yamadori et al., 1999). Sequence analysis revealed that the 257 amino acid N-terminal part of PRT shows significant homology to the Sab protein family. PRT shares 48% identity and 66% similarity to the human, and 45% identity and 63% similarity to murine Sab proteins, while the C-terminal region did not show any significant homology to known proteins. Sequence comparisons also identified a predicted C. elegans K03E6.7 (38% identity and 56% similarity) and a Drosophila CG14408 (38% identity and 54% similarity) homologue of the 252 amino acid N-terminal part of the PRT protein. The N-terminal region of human Sab contains an unconventional SH3-binding domain, which preferentially binds to the SH3 motif of Bruton’s Tyrosine kinase (Btk) protein (Matsushita et al., 1998). Interestingly, the 32 amino acids long SH3-binding domain of human Sab shows 48% identity and 73% similarity to the corresponding part of PRT, indicating that the two proteins may have conserved functions.

In order to find out the consequence of the hobo insertion on the prt transcription, we performed RT-PCR analyses. Complementary primers to the first, second and fourth exons of the prt gene were used to amplify two different regions of the prt RNA sequence between exons 1 and 4 and 2 and 4. The 1747 bp PCR product, corresponding to the exons 1-4, was successfully amplified from RNA template of wild-type females, while we did not detect a PCR fragment from the prt^gs homozygous mutant females (Fig. 3A, lanes b,c). The RT-PCR reaction with exon 2- and exon 4-specific primers also detected no mRNA in the mutant (data not shown). We obtained the same negative results by using RNA templates from both maternally and zygotically mutant prt^gs embryos (Fig. 3A, lane f). By contrast, RT-PCR analysis of embryos that were only maternally mutant, but fertilised by wild-type male, revealed the presence of a robust zygotic transcript from the paternally inherited wild-type prt allele (Fig. 3A, lanes d,e). These data indicated either the complete loss of the prt mRNA from the prt insertion alelle or the presence of a new, elongated fusion transcript from the mutant chromosome. To exclude one of the above possibilities, we tried to identify possible aberrant splice forms, which would include exons of both the prt gene and hobo transposon. As the orientation of mini-white marker gene in the inserted hobo element was the same as that of prt, we analysed possible chimaeric splice forms between the prt and mini-white genes. Using prt and mini-white specific primers, we were able to amplify a 882 bp long RT-PCR product on RNA template purified from prt^gs homozygous females (Fig. 3B). We sequenced the PCR product and found that the aberrant splice product contained the first exon of prt and the second, third and fourth exon of the mini-white gene. As this chimaeric mRNA contained the complete coding sequence of the mini-white gene, the active White protein product might be translated from this aberrant splice form.

To identify the distribution of PRT protein in the developing oocyte, we raised a polyclonal rabbit antibody against the PRT protein. Western blot of wild-type ovary protein extracts showed that the anti-PRT antiserum recognised a 53-55 kDa doublet that was absent from the prt^gs ovary extract (Fig. 3C). Computer analysis of the PRT protein sequence revealed several potential phosphorylation sites, indicating that the 55 kDa PRT protein species could be the result of a post-translational modification. However, the anti-PRT antiserum failed to detect PRT protein in whole-mount ovaries. In both wild-type and prt^gs mutant ovaries, we obtained a low level uniform staining (data not shown). As an alternative approach to visualise the distribution of PRT protein during oogenesis, we used the UASp/nosGal-Vp16 system (Rorth, 1998; Van Doren et al., 1998) to express a GFP-tagged version of PRT in the Drosophila germline. Several independent PRT-GFP transgenic lines were established and assayed for GFP expression after nosGal4VP16 induction. All these lines showed similar GFP expression pattern and rescued the prt^gs germ cell-less phenotype (Fig. 3D). In early egg chambers, (stages 2-4) PRT-GFP is concentrated in large cytoplasmic aggregates around the nuclei of nurse cells. During stages 4 to 7, this particulate appearance of PRT-GFP is detectable both in nurse cells and oocytes. From stage 8, when the microtubule network of oocytes is reorganised, the PRT-GFP aggregates can be found exclusively in the subcortical region. During stages 9-10, the nurse cells and the oocytes contain similar amounts of PRT-GFP protein. Formerly, it has been reported that the Exuperantia, Me31B, ORB and CUP proteins show similar distribution during egg development to PRT-GFP (Wilsch-Brauninger et al., 1997; Nakamura et al., 2001; Mansfield et al., 2002; Keyes and Spradling, 1997). Thus, we next asked whether PRT was colocalised with any of the above proteins; however, it was not (data not shown).

As prt^gs resulted in both germ cell- and abdomenless phenotypes, which are characteristic to mutations of the posterior group genes, we investigated the mRNA and protein distribution of osk, the key element of the posterior gene hierarchy. Using RNA in situ hybridisation, we did not observe significantly reduced posterior localised osk mRNA in prt^gs/prt^gs mutant ovaries (Fig. 4A,B). Consistently, an RNA-binding protein, Staufen, which colocalises with osk mRNA and thus is used as an osk mRNA distribution marker (St Johnston et al., 1991) was found at the posterior pole in prt^gs mutant oocytes and in wild types (Fig. 4C,D). However, we found a decreased OSK protein level when we performed anti-OSK antibody staining on developing oocytes and on whole-mount embryos originated from homozygous prt^gs females (Table 2 and Fig. 4E,F,J,K). Furthermore, confocal analysis of anti-OSK stained mutant ovaries revealed delocalised OSK protein in 15% of stage 10 oocytes (Fig. 4G-I). While OSK protein become concentrated at the posterior pole, the PRT-GFP aggregates concentrated at the subcortical region at stage 10 oocytes and showed colocalization with OSK protein at the posterior pole in wild type (Fig. 4L-N).

By western analysis, we also detected reduced OSK protein level in ovaries dissected from prt^gs homozygous females (Fig. 5). However, the decrease in the OSK protein level was exclusively due to reduction of the short OSK isoform, while the 71 kDa OSK isoform was virtually unaffected. In addition, we detected an alteration in the post-translational modification of the short OSK isoform (Fig. 5). The 55 kDa unphosphorylated short OSK isoform had almost disappeared and reduced amount of 57 kDa phosphorylated short OSK isoform could be detected. These results strongly suggest that the prt gene acts at the short OSK protein level in the posterior gene hierarchy.

The biochemical function of Sab, the human homologue of the prt gene has been revealed by in vitro and in vivo assays. The Sab protein preferentially binds to the Bruton’s tyrosine kinase protein and negatively regulates its function (Matsushita et al., 1998; Yamadori et al., 1999). Because, theoretically, the absence of a negative regulator can be suppressed by decreasing the level of the regulated component, we tested whether hypomorphic Btk alleles can suppress prt^gs mutation. Recombinant chromosomes were generated from hypomorphic Btk29A mutations (k00206, k05610 and ficPL) (Roulier et al., 1998; Baba et al., 1999), and null pr t mutation bearing chromosomes and double mutant females were tested for the grandchildless phenotype. We observed that the original 70% penetrant prt^gs homozygous phenotype was completely suppressed in Btk29A^ficPL prt^gs/Btk29A^ficPL prt^gsdouble mutant females, while inBtk29A^k05610 prt^gs/Btk29A^ficPLprt^gsand Btk29A^k00206 prt^gs/Btk29A prt^gs mutant females some residual gs phenotype, 4.4% and 4.0%, respectively, were measured. Thus, we concluded that Prt negatively regulates Drosophila Btk, because the prt null mutant phenotype is manifested only if Btk activity is present. To exclude the possibility that Drosophila Btk29A has, itself, a loss-of-function grandchildless phenotype, we repeated the Btk29A^k00206 and Btk29A^k05610 germline mosaic analyses published by Roulier (Roulier, 1998). Both Btk29A^k00206 and Btk29A^k05610 homozygous germline clones resulted in the earlier published loss-of-function, dumpless and maternal effect head defect phenotypes. We also found some normal looking eggs that were fertilised and developed to adulthood. Twenty such adult individuals were dissected in both experiments but no grandchildless gonad was found, indicating that Btk29A activity is not required for germline formation.

In this paper, we have identified the prt gene, which has an important role in the proper formation of embryonic germ cells in Drosophila. Null mutant of prt produces embryos that display a significant reduction in the number of germ cells, show a reduction of short OSK protein levels and abnormal OSK distribution. prt^gs and the BTK29A mutations were found to interact genetically. The PRT protein is found in large protein aggregates that are localised subcortically in the oocytes.

Posteriorly anchored OSK protein is a key component of the germ plasm and subsequent embryonic germ cell formation. Although, the anchoring process is an important aspect of osk regulation very little is known about its mechanism. Genetic analysis of the anchoring mechanism is especially difficult, as OSK anchoring is interdependent on the presence of both the mRNA and the protein at the posterior region (Glotzer et al., 1997; Markussen et al., 1995; Rongo et al., 1995). Therefore, mutations of regulatory genes that interfere with OSK protein localisation often also result in mRNA delocalisation, making it impossible to separate these two processes. An example of the above is provided by the TmII mutation, which abolishes the localisation of both osk mRNA and protein (Erdelyi et al., 1995; Tetzlaff et al., 1996). Therefore, the actin-binding TmII protein and the actin cytoskeleton may be directly connected, not only to the mRNA but also the OSK protein, at the posterior pole. Interestingly, in prt^gs mutants, the OSK protein delocalisation does not seem to be coupled with osk mRNA localisation. While osk mRNA was localised normally at the posterior pole in all stages of oogenesis, some OSK protein was detached from the subcortical region in prt^gs/prt^gs mutant ovaries. This delocalisation can reduce OSK concentration significantly at the posterior pole. The decreased OSK level leads to germ plasm reduction and the formation of fewer pole cells, which explain the prt mutant phenotype. As in early stages of oogenesis, OSK localisation seems completely normal and only becomes abnormal gradually during the subsequent stages, prt is very unlikely to interfere with OSK translation, rather it plays a role in OSK protein anchoring at the posterior pole or increases the stability of the anchored OSK protein.

In prt^gs mutants, only the short OSK isoform level was found to be reduced, while the long isoform remained at wild-type levels. This may explain the normal osk mRNA distribution, as the long OSK isoform can maintain its own mRNA at the posterior pole as described earlier (Markussen et al., 1995). However, the isoform specificity of prt^gs reveals that the two OSK isoforms may use independent posterior anchoring mechanisms or can be subjected to different post-translational regulation processes during their anchoring. However, genetic and molecular evidence suggests that some residual short OSK must still be present in prt^gs null mutants, because the complete loss of the short OSK isoform would not only result in grandchildless but rather a complete abdomen and germ cell-less phenotype (Markussen et al., 1995). This result also demonstrates that prt function in OSK anchoring must be redundant.

Subcellular localisation of PRT is reconcilable with osk regulatory function, because at the subcortical region their localisation overlaps. This colocalisation potentiates the functional interaction. Some osk translational regulatory proteins, which form a large ribonucleoprotein (RNP) complex, demonstrate a similar expression pattern during their transport to oocytes to that of PRT. Biochemical evidence exists that this RNP contains EXU and at least seven other proteins, in addition to osk and bicoid mRNAs (Wilsch-Brauninger et al., 1997; Wilhelm et al., 2000). We demonstrated that Exu, Orb and Me31 B, three elements of the above RNP complex, as well as CUP, an RNP independent protein with similar localisation pattern (Keyes and Spradling, 1997), are not colocalised with PRT in nurse cells. Consequently PRT aggregates, in spite of their similar subcellular localisation, have an independent transport system from either the EXU or the CUP complexes.

The genetic interaction found between prt and Btk29A indicates that the PRT regulatory function is evolutionary conserved. This is especially interesting, as the Drosophila BTK homologue is not required in germ cell formation. The Drosophila genome contains a single BTK homologue Btk29A, which encodes two protein species. Btk29A has several pleiotropic functions, such as male fertility and ring canal formation. It is expressed in the adult head, the larval immune system, the male and female gonads, and several other tissues (Roulier et al., 1998; Baba et al., 1999; Vincent et al., 1989; Wadsworth et al., 1990). However, our germline clone analysis of the two hypomorhic alleles failed to reveal a role in germ cell formation. Based on the structural conservation of the Sab and PRT proteins, we anticipated that PRT might also exhibit a negative regulatory function similarly to Sab. According to this hypothesis, when PRT, the presumed negative regulator, is absent, an ectopic BTK activity would interfere with the normal function of the posterior gene hierarchy. Being normally suppressed, loss of function of such a negatively regulated gene would not be expected to cause any posterior phenotype. Indeed, the suppression of prt^gs phenotypes by hypomorphic Btk29A alleles indicates that in Drosophila, PRT also negatively regulates BTK29A.

Therefore, we propose that, in wild-type oocytes, PRT inactivates the BTK protein in the subcortical region. In prt^gs mutants, however, unregulated BTK interferes either with the localisation of subcortical cellular components in the oocyte, or it modifies the phosphorylation pattern of the short OSK protein itself. We find that the latter is a less feasible explanation, because in prt^gs mutants we do not find increased levels of phosphorylated short OSK that would be the result of extra kinase activity, instead the level of 57 kDa phosphorylated short OSK isoform is also significantly reduced. We suggest that uncontrolled BTK kinase activity modifies the anchoring capability of short OSK directly or indirectly to the subcortical region, resulting in delocalisation and degradation of both phosphorylated and unphosphorylated short OSK proteins. Because only the phosphorylated short OSK isoform was detected in prt^gs mutants, we suppose that either this protein is more stable or it is better anchored to the posterior pole, compared with the unphosphorylated one. The weakly anchored OSK protein is displaced by the cytoplasmic streaming and loses its pole plasm organising activity. Future research should be directed towards determining the precise biochemical function of PRT and elucidating the anchoring mechanism of OSK protein.

We thank Anne Ephrussi for providing osk cDNA, anti-OSK and anti-VAS and anti-STAU antibodies. We are grateful for Satoru Kobayashi for anti-Me31 B, Allan Spradling for anti-CUP, Christiane Nüsslein-Volhard for anti-Exu antisera. We thank Steven Beckendorf for FRT-Btk, Daisuke Yamamoto for fic^P mutant lines and Juergen Knoblich for β-globin-2xGFP vector. We are also grateful to Selen Muratoglu for molecular technical help, and to Barbara Botos, Mim Bower and the BRC Drosophila Community for critical reading of the manuscript. The work was supported by a grant (T22096) from the Hungarian National Science Foundation (OTKA). prt gene has been named after the famous detective Hercule Poirot.