Tudor domains are found in many organisms and have been implicated in protein-protein interactions in which methylated protein substrates bind to these domains. Here, we present evidence for the involvement of specific Tudor domains in germline development. Drosophila Tudor, the founder of the Tudor domain family, contains 11 Tudor domains and is a component of polar granules and nuage, electron-dense organelles characteristic of the germline in many organisms, including mammals. In this study, we investigated whether the 11 Tudor domains fulfil specific functions for polar granule assembly,germ cell formation and abdomen formation. We find that even a small number of non-overlapping Tudor domains or a substantial reduction in overall Tudor protein is sufficient for abdomen development. In stark contrast, we find a requirement for specific Tudor domains in germ cell formation, Tudor localization and polar granule architecture. Combining genetic analysis with structural modeling of specific Tudor domains, we propose that these domains serve as `docking platforms' for polar granule assembly.

Tudor (Tud) domains were initially identified as common protein motifs found in the Drosophila Tud protein and in other proteins from a wide variety of organisms and different kingdoms, including fungi, plants and animals (Maurer-Stroh et al.,2003; Ponting,1997; Talbot et al.,1998). Tud domains are related to plant Agenet, Chromo PWWP and MBT domains, which together form the Tud domain `Royal Family'(Maurer-Stroh et al., 2003). Tud domain proteins have been shown to interact with other proteins and efficient binding requires methylated arginine and lysine residues in the target protein (Brahms et al.,2001; Côté and Richard, 2005; Huyen et al.,2004; Kim et al.,2006; Sprangers et al.,2003). The Tud domain of the Survival Motor Neuron (Smn) protein binds directly to spliceosomal Sm proteins during spliceosome assembly(Brahms et al., 2001; Bühler et al., 1999; Selenko et al., 2001; Sprangers et al., 2003). Several Tud domain proteins have been shown to interact with modified histones. In particular, 53BP1 has tandem Tud domains that bind histone H3 on methylated Lys79 and this may be a molecular device for the recognition of DNA double-strand breaks during checkpoint responses(Huyen et al., 2004). Subsequently, Tud domains of several proteins were shown to bind to histones H3 and H4 (Huang et al., 2006; Kim et al., 2006). The recently identified structure of the N-terminal domain of the Fragile X Mental Retardation Protein (Fmrp) revealed two repeats of a Tud domain, and one of these domains was shown to interact with methylated lysine and with an Fmrp nuclear-interacting protein, 82-FIP (Ramos et al., 2006). Structural analysis of Tud domains from different proteins revealed that these domains can either fold into a single barrel-like structure composed of five β strands(Selenko et al., 2001) or form an intertwined structure consisting of two Tud domains(Huang et al., 2006).

tud was the first member of the posterior group of genes identified in Drosophila. The hallmark of this group of maternal effect genes is their dual role in abdomen development and germ cell formation(Boswell and Mahowald, 1985; Thomson and Lasko, 2004; Thomson and Lasko, 2005). Germ cells are formed in a specialized embryonic cytoplasm, called germ plasm,which contains characteristic electron-dense organelles, the polar granules(Illmensee and Mahowald, 1974; Mahowald, 1968). The Tud protein is a component of polar granules(Amikura et al., 2001; Bardsley et al., 1993), and they are severely reduced in number and size in strong tud mutants(Amikura et al., 2001; Boswell and Mahowald, 1985; Thomson and Lasko, 2004). Based on genetic interactions and its protein localization pattern in other mutants affecting germ plasm, tud acts downstream of oskarand vasa in germ plasm assembly(Bardsley et al., 1993; Ephrussi and Lehmann, 1992). Recently, Tud protein was shown in vitro to interact with Valois, which is a component of the methylosome in Drosophila(Anne and Mechler, 2005),suggesting that Tud, like other proteins in the family, may bind to methylated substrates.

Phenotypical analysis of tud mutants revealed abdomen-patterning defects, suggesting that tud is involved not only in germline specification but also in abdomen formation(Boswell and Mahowald, 1985). However abdomen defects are not seen in all of the RNA null mutant embryos(Thomson and Lasko, 2004),demonstrating that tud is not absolutely required for formation of the abdomen. A likely reason for abdomen development defects is the reduced localization of nanos (nos) RNA(Thomson and Lasko, 2004; Wang et al., 1994) and the decreased amount of Nos protein (Gavis and Lehmann, 1994) in tud mutant embryos.

Drosophila Tud protein contains 11 Tud domains(Talbot et al., 1998) and,until now, their function in germ cell specification or abdomen formation remained unknown. Slow progress on understanding Tud was in part due to the large size of the protein, which consists of 2515 amino acids(Golumbeski et al., 1991). As a result of an extensive screen designed to find mutants with germ cell formation defects, we obtained 15 new tud alleles. Characterization of these alleles, as well as the analysis of transgenic lines expressing Tud versions lacking different Tud domains, provided the first evidence for the involvement of specific Tud domains in germline development and in the maintenance of polar granule architecture. On the basis of the structural analysis of Tud domains, we propose that the germline specification and architecture of polar granules are dependent on specific protein-protein interactions between these domains and other polar granule components.

Fly stocks

New tud mutant lines were obtained from a 2R maternal-effect screen using the germ-cell marker faf-lacZ and had the following genotype: w, P[w+, faflacZ];P[w+, FRT(42B)], tudmutant/CyO, P[w+, hs-hid] (Moore et al.,1998). A detailed description of the screen will be published elsewhere (V. Barbosa and R.L., unpublished). Previously isolated tudlines were as follows: tud1, bw, sp/CyO,P[w+, hs-hid] (E. Wieschaus and C. Nüsslein-Volhard, unpublished); tud4, cn,bw/CyO, P[w+, hs-hid](Boswell and Mahowald, 1985). Unless noted, embryos or ovaries from females were transheterozygous for a given tud allele and a tud deletion[Df(2R)PurP133]. For genetic markers used, see Lindsley and Zimm (Lindsley and Zimm,1992). Flies were raised using standard cornmeal-molasses medium at 25°C.

Allele sequencing

Genomic DNA from flies transheterozygous for a given tud allele and a deletion of the tud genomic region[Df(2R)PurP133] was prepared using the DNeasy Tissue Kit(Qiagen). The ∼9 kb tud genomic region was amplified using multiple independent PCR reactions with Pfu Turbo DNA polymerase according to the manufacturer's protocol (Stratagene). Sequencing was performed at the Rockefeller University DNA Sequencing Resource Center and Genewiz (North Brunswick, NJ).

Transgene construction

cDNA containing the complete 7548-nucleotide tud coding sequence(CDS) (from the Berkeley Drosophila Genome Project) and mini-tudconstructs were cloned into pP{CaSpeR-2} with the nos promoter and 5′UTR for germline expression (Wang and Lehmann, 1991), an HA tag at the N terminus and a K103 ′UTR for mRNA stability (Serano et al., 1994). For the generation of a mini-tud Δ1 construct, a NruI-Bsu36I fragment of tud CDS was excised and the Bsu36I-protruding end of the CDS remainder was filled in and blunt-ligated with the NruI end. This procedure resulted in a deletion from Glu1545 to Pro2443 and the expression of a 1616 amino acid protein. Mini-tud Δ2 was generated by PCR of the segment corresponding to the 368 Tud C-terminal amino acids and subsequent cloning of the resultant PCR fragment into a transformation vector. The mini-tudΔ2 sequence was verified and contained no PCR-generated errors. Mini-tud Δ3 was generated by removing a StuI-NruI segment of tud CDS, which resulted in an 1271 amino acid protein with an 1244 amino acid deletion from Leu301 to Arg1544. Transgenic flies were generated using standard procedures(Spradling, 1986).

Western blot analyses

Ovary extracts from wild-type[w,faf-lacZ;tud+/Df(2R)PurP133] and tud mutants[w,faf-lacZ;tudmutant/Df(2R)PurP133] were loaded onto 7.5% SDS polyacrylamide gels. Proteins were transferred on PVDF membrane (Millipore) as previously described(Bardsley et al., 1993) and probed with TUD-A63 antibody generated against Tud N-terminal amino acids 1-554 (Thomson and Lasko,2004). Tud specific bands were detected with an enhanced chemiluminescence protocol (Amersham PharmaciaBiotech). Following Tud protein detection, the membranes were re-probed with anti-Dynein antibody (Chemicon International) to detect Dynein as a loading control. TUD-A63 antibody and anti-Dynein antibody were used (1:2000). For quantitative western blot analysis, Tud and Dynein specific bands were quantified using ImageJ software(http://rsb.info.nih.gov/ij/)and then Tud signals were normalized to the Dynein signals. Average percentage of normalized Tud signals relative to the wild-type ±standard error was then calculated and recorded. Western blot analyses of tud-HA transgene lines expressing either full-length Tud or different mini-Tud versions in a tud1/Df(2R)PurP133background were performed as described above for tud mutants. However, for the detection of Tud bands, the specific anti-HA tag antibody[clone 3F10] (Roche Applied Science) was used (1:1500). For each tudmutant and transgene, several ovary extracts were independently prepared and subjected to western blot analyses.

In situ hybridization and immunohistochemistry

These procedures have been described previously(Lehmann and Tautz, 1994; Navarro et al., 2004; Stein et al., 2002). nos,pgc and gcl antisense RNA probes were generated with a DIG RNA labeling kit (Roche). For whole-mount immunostaining, the following antibodies were used: rabbit anti-Vasa (1:2000)(Stein et al., 2002); rat anti-Vasa (1:1000) (Thomson and Lasko,2004); rabbit anti-Tud (TUD-A63; 1:800)(Thomson and Lasko, 2004);anti-HA tag antibody (clone 3F10; 1:800) (Roche Applied Science).

Cuticle analysis

Cuticle preparations were made essentially as previously described(Wieschaus and Nüsslein-Volhard,1998). For a description of the wild-type cuticle pattern, see Lohs-Schardin et al. (Lohs-Schardin et al., 1979). For quantification of the abdomen phenotype, ∼100 or more cuticles were scored for each tud mutant or transgenic line(see Results).

Electron microscopy

To determine the morphology of polar granules, 30- to 55-minute-old embryos were prepared for electron microscopy. Embryos were dechorionated with sodium hypochlorite and then fixed in heptane saturated with 12.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) for 20 minutes at room temperature. After fixation, vitelline membranes were removed manually. Then the embryos were fixed in 5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) overnight at 4°C and postfixed with 1% osmium tetroxide followed by 1% uranyl acetate. The embryos were dehydrated through a graded series of ethanol and embedded in LX112 resin (LADD Research Industries, Burlington VT). Ultrathin (80 nm)sections were cut on a Reichert Ultracut UCT, stained with uranyl acetate followed by lead citrate, and viewed on a JEOL 1200EX transmission electron microscope at 80 kV. For EM analysis, multiple fields from each of three ultrathin sections per embryo separated by 1 μm were photographed at 10 000× magnification. For each genotype, several embryos were sectioned and analyzed. EM images of germ plasm from mutant embryos were compared with those from wild type. In control experiments, no polar granules were detected in the anterior periplasm of wild-type and mutant embryos.

Homology modeling

Modeling of the domains 1, 7 and 10 of Tud protein, and of the tud4, tudA36 and tudB42 mutants, was carried out using SWISS-MODEL (Schwede et al.,2003) in a semi-automated fashion. The sequences of the domains from the Tud protein, as originally defined, were initially aligned with the program ClustalX (Thompson et al.,1997) and standard parameters. The alignment was then optimized manually (in the last loop of the protein only). The sequences alignment and the coordinates of the Smn Tud structure were fed into the SWISS-MODEL server. The structural quality of the resulting models was evaluated with the program WHATIF (Vriend, 1990). The models of Tud domains 1, 7 and 10 of the Tud protein have a total final energy in the range of -1500 Kj/mol (±10%) and an OLDQUA WHATIF score of∼2 (±10%), in the range expected from similar studies(Amodeo et al., 2001). Visual inspection of the models with the program Molmol(Koradi et al., 1996) showed that they had the expected structural features. The models of the mutant domains 1 and 10 showed energies and structural quality similar to the ones of the wild type (less than 10% difference), and were, as far as the backbone is concerned, essentially identical (RMSD <0.25Å) indicating that the mutant side chains (even the bulky tryptophan) can be easily accommodated during the modeling procedure. Graphic representation of the structure of the domains, superimpositions of the models and RMSD calculation were performed with the program Molmol. The domain 7 mutant has a three-amino-acid deletion that would remove part of the fifth strand of the domain. The sequence identity between domains is not sufficient to predict accurately the structure of this part of the protein and a model of the mutant domain 7 may be unreliable.

Identification of tudor alleles with mutations in single Tudor domains

In an EMS mutagenesis screen for mutations with defects in germ cell formation, we identified 15 new tud alleles (see Materials and methods). These and the existing tud alleles allowed us to analyze the relationship between Tud protein domains and Tud function in embryonic patterning and germ cell formation. Because the sequence of the original tud mutants had not been determined, we analyzed existing alleles, tud1 (E. Wieschaus and C. Nüsslein-Volhard,unpublished) and tud4(Boswell and Mahowald, 1985),together with the new alleles. To establish which mutants expressed Tud protein, we carried out western blot analysis using an antibody generated against the Tud N-terminal amino acids 1-554(Thomson and Lasko, 2004). Of the 17 mutants analyzed, Tud protein was detected in only five mutants: the original tud4 mutant, and the new tudA7, tudA36, tudB42 and tudB45 mutants(Fig. 1). All other mutants had no detectable Tud protein, including tud1, tudA1, tudA10, tudA14, tudA47, tudC3, tudC8, tudC11, tudC30, tudC53, tudC67 and tudC73 (data not shown), suggesting that these mutations either carried deletions that prevented gene expression altogether, or that stop codon mutations severely truncated the protein and decreased the stability of the protein or RNA. Tud proteins from tud4, tudA7, tudA36 and tudB42 migrated similarly to the wild-type protein(Fig. 1B), while the tudB45 mutant protein migrated slightly faster than the wild-type protein, suggesting a truncated but stable protein(Fig. 1B).

Fig. 1.

Identification of tud missense mutations. (A)Missense and nonsense mutations are placed at their corresponding locations within Tud protein sequence. Tud domains are indicated by blue squares.(B) Western blot of ovary extracts from wild type (wt) and tudmutants probed with TUD-A63 antibody and anti-Dynein antibody to detect Dynein as a loading control. (C) Tud proteins and Dynein were quantified by western blot with TUD-A63 antibody and anti-Dynein antibody, respectively (see Materials and methods). Tud signals were normalized to the Dynein signals and the average percentage relative to the wild-type ±s.e.m. is recorded. For each table entry, 3-4 measurements (n) were used to calculate the average percentage.

Fig. 1.

Identification of tud missense mutations. (A)Missense and nonsense mutations are placed at their corresponding locations within Tud protein sequence. Tud domains are indicated by blue squares.(B) Western blot of ovary extracts from wild type (wt) and tudmutants probed with TUD-A63 antibody and anti-Dynein antibody to detect Dynein as a loading control. (C) Tud proteins and Dynein were quantified by western blot with TUD-A63 antibody and anti-Dynein antibody, respectively (see Materials and methods). Tud signals were normalized to the Dynein signals and the average percentage relative to the wild-type ±s.e.m. is recorded. For each table entry, 3-4 measurements (n) were used to calculate the average percentage.

Given the large size of the tud gene (∼9000 nucleotides, 2515 amino acids), we reasoned to sequence only those mutants that would be most informative in correlating a particular amino acid change to Tud function. We therefore sequenced the five alleles (tud4, tudA7, tudA36, tudB42 and tudB45) that expressed protein and the original tud1 mutant, as an example of a mutant that showed no detectable protein expression. tud1has a stop codon (Lys1036UAG) mutation, which effectively abolishes protein expression, presumably because of the instability of the Tud fragment or because of degradation of tud mRNA via nonsense-mediated RNA decay. Thus, in contrast to earlier reports, our study suggests that tud1 is a strong loss-of-function allele. Sequence analysis of the tud alleles, in which stable protein was detected,revealed point mutations in tudA36 and tudB42 that affect the equivalent arginine in Tud domains 1 and 10, respectively (Arg91Trp in tudA36 and Arg2228Cys in tudB42; Fig. 1A). tud4 has an in-frame deletion that removes three amino acids from Tud domain 7 (Asp1708, Ile1709, Lys1710). Finally, tudB45 and tudA7 cause C-terminal truncations as a result of stop codons. An UAG nonsense mutation(Gln2148UAG) in tudB45 situated between Tud domains 9 and 10 causes a truncation of 368 amino acids, whereas tudA7is truncated just 32 codons upstream of the tud natural termination codon (Fig. 1A). Quantitative analysis revealed that Tud mutant proteins are in general less stable than the wild-type protein (Fig. 1C). In particular, the deletion and truncation alleles showed a significant (up to 10-fold) decrease in protein levels.

Mutations in Tudor domains may directly affect binding of polar granule components

Tud domains are found in many proteins involved in gene regulation(Maurer-Stroh et al., 2003; Ponting, 1997). The first high resolution structure of the Tud domain from the Smn protein, showed that its five β strands create a pocket of aromatic amino acids that interact with methylated arginines of protein partners(Selenko et al., 2001; Sprangers et al., 2003). Our finding that individual Tud domains in Tudor have very specific effects on Tud function motivated us to analyze the potential structural consequences of mutations in these domains. In tudA36 and tudB42, an arginine is mutated at identical positions in Tud domain 1 and 10, respectively (Fig. 2A,B). Although the same arginine is conserved in many Tud domains from different organisms (Talbot et al.,1998), the well-studied Tud domain of the Smn protein has a proline residue at this position (Fig. 2C), suggesting that this residue is not critical for the overall structure of the domain. We modeled the sequence of Tud domains 1, 7 and 10 onto the known structure of the Smn Tud domain. In the models, the arginine side chain is exposed to the solvent and is in close proximity to the binding pocket (Fig. 2). Next, we modeled the mutant domains, where tryptophan and cystein substitute for arginine in Tud domain 1 and 10, respectively. The models of the mutant domains do not show a significant change of the overall structure when compared with the wild type (Fig. 2A′,B′), indicating that the different side chains can be easily accommodated by the structure. The proximity of the arginine to the predicted binding pocket for methylated ligands suggests that this residue may control the access of specific Tud protein ligands to the binding cavity. Interestingly, in the Smn Tud domain, glutamate 134, which strongly affects the recognition of protein targets, is located adjacent to the binding pocket,albeit in a different position to the arginines in Tud domains 1 and 10(Fig. 2C)(Selenko et al., 2001), and may play an analogous function. Our models therefore predict that the arginine mutations specifically affect the recognition of Tudinteracting proteins,rather than the stability of the Tud domain. Conversely, the modeling of Tud domain 7 with the three-aminoacid, in-frame deletion observed in tud4 (Fig. 1A) suggests that this mutation causes a significant structural effect on the domain. Because the tud4 mutant exhibits a rather weak phenotype (see below) and produces a stable protein(Fig. 1B), we propose that this Tud domain is structured independently of the other Tud domains.

Fig. 2.

Mutations in Tud domains may affect interactions with polar granule components. (A-C) Structures of (A) Tud domain 10 and (B) Tud domain 1 were predicted based on (C) the known structure of the Tud domain from the Smn protein (Selenko et al.,2001; Sprangers et al.,2003). (A′) Predicted structure of Tud domain 10 with an Arg2228Cys change. (B′) Predicted structure of Tud domain 1 with an Arg91Trp change. Arg, Cys and Trp residues in Tud domains 1 and 10, and the corresponding Pro residue in the Smn Tud domain, are shown in magenta. Glu134 of the Smn Tud domain plays a crucial role in protein-protein interactions(Selenko et al., 2001) and is indicated in green. The cluster of aromatic amino acids, which form a binding pocket for the Smn Tud domain interacting partners, is shown in yellow.

Fig. 2.

Mutations in Tud domains may affect interactions with polar granule components. (A-C) Structures of (A) Tud domain 10 and (B) Tud domain 1 were predicted based on (C) the known structure of the Tud domain from the Smn protein (Selenko et al.,2001; Sprangers et al.,2003). (A′) Predicted structure of Tud domain 10 with an Arg2228Cys change. (B′) Predicted structure of Tud domain 1 with an Arg91Trp change. Arg, Cys and Trp residues in Tud domains 1 and 10, and the corresponding Pro residue in the Smn Tud domain, are shown in magenta. Glu134 of the Smn Tud domain plays a crucial role in protein-protein interactions(Selenko et al., 2001) and is indicated in green. The cluster of aromatic amino acids, which form a binding pocket for the Smn Tud domain interacting partners, is shown in yellow.

Non-overlapping segments of Tudor protein are sufficient for abdomen development

Sequence analysis of tud mutants demonstrates that mutations within a single Tud domain can cause a mutant phenotype. This suggests either that single Tud domains act in concert to provide full function, or,alternatively, that specific Tud domains may have specific functions. Tud is required for both abdomen and germ cell formation(Boswell and Mahowald, 1985). Tud function in abdominal development is mediated via its role in nosRNA localization and translation (Gavis and Lehmann, 1994; Wang et al., 1994). Because tud is a strict maternal effect gene,we will refer to embryos derived from females mutant for a particular allelic combination as `mutant embryos'. In strong tud mutant embryos, nos RNA localization to the posterior pole is reduced when compared with the wild type, and Nos protein synthesis is decreased(Gavis and Lehmann, 1994; Thomson and Lasko, 2004; Wang et al., 1994). However,in contrast to other genes that affect germ plasm assembly, such as oskar or vasa, Tud protein is not absolutely necessary for nos RNA localization and translation, as females carrying a tud null mutation produce embryos with some nos RNA localization, and 15% of these embryos develop into normally segmented larvae(Thomson and Lasko, 2004). By contrast, Tud function is absolutely required for germ cell formation, as embryos from females mutant for any of the strong alleles lack germ cells(Boswell and Mahowald, 1985; Thomson and Lasko, 2004).

To determine the role of individual Tud domains in abdomen and germ cell formation, we characterized the mutant phenotype of our new alleles in detail. In addition, we analyzed several mini-tud transgenes expressing Tud fragments that lack different parts of Tud(Fig. 3). In particular, mini-tud Δ1 produces Tud domains 1-6, minitud Δ2 produces domains 10 and 11, and mini-tud Δ3 produces domain 1 and domains 7-11. As a control, a full-length Tud transgene showed complete rescue of abdomen and germline defects in a tud1 mutant background and co-localized with the polar granule marker Vasa in the germ plasm (data not shown). All tud alleles that lack protein expression by western blot show a phenotype very similar to that described for the tud loss-of-function mutation: larvae have segmentation defects and mutant embryos completely lack germ cells(Table 1; data not shown). By contrast, females mutant for any one of the alleles that produce Tud protein generate embryos that are normally patterned(Fig. 4). Because these mutations affect different parts of the Tud protein, this suggests that any part of Tud may be sufficient to provide nos localization and translation function.

Fig. 3.

Design of tud transgenic constructs and their expression.(A) tud full-length and deletion constructs. Tud domains deleted in the constructs are shown as white squares. (B) Western blot detection of transgenic Tud proteins. Ovary extracts from the wild type, which expresses no transgenes (`Wild-type' and `-HA' lanes), and those from transgenic lines, each of which expresses one copy of the transgene in a tud1/Df(tud) background, were used. Endogenous Tud protein and HA-Tud proteins were detected with TUD-A63 (αTud) and anti-HA (αHA) antibodies, respectively. Dynein bands, detected with anti-Dynein antibody, served as loading controls. The full-length transgene is expressed at about 20% of the wild type level. mini-tud Δ2 and mini-tud Δ3 are expressed at similar levels to endogenous,wild-type Tud. Expression of the mini-tud Δ1 transgene was consistently weak (2-3% of the mini-tud Δ3 amount) in several transgenic lines (lines A-C), probably due to poor protein stability.

Fig. 3.

Design of tud transgenic constructs and their expression.(A) tud full-length and deletion constructs. Tud domains deleted in the constructs are shown as white squares. (B) Western blot detection of transgenic Tud proteins. Ovary extracts from the wild type, which expresses no transgenes (`Wild-type' and `-HA' lanes), and those from transgenic lines, each of which expresses one copy of the transgene in a tud1/Df(tud) background, were used. Endogenous Tud protein and HA-Tud proteins were detected with TUD-A63 (αTud) and anti-HA (αHA) antibodies, respectively. Dynein bands, detected with anti-Dynein antibody, served as loading controls. The full-length transgene is expressed at about 20% of the wild type level. mini-tud Δ2 and mini-tud Δ3 are expressed at similar levels to endogenous,wild-type Tud. Expression of the mini-tud Δ1 transgene was consistently weak (2-3% of the mini-tud Δ3 amount) in several transgenic lines (lines A-C), probably due to poor protein stability.

To further test this idea, we determined whether mini-tudtransgenes are active in abdomen development. All of these transgenes rescued the abdominal phenotype of the tud1 mutant; rescue was very efficient for mini-tud Δ3, where about 80% of all larvae showed a wild-type pattern (Fig. 4G), less substantial rescue was achieved with mini-tudΔ1 and mini-tud Δ2(Table 1, Fig. 4H). mini-tudΔ1 is expressed weakly, which may have attributed to the weak rescue(Fig. 3B). The ability of different mutant alleles and transgenes, which express different Tud regions,to induce abdomen formation suggests functional redundancy of Tud segments in this process. The best example of this redundancy comes from a comparison of the tudB45 mutant and the mini-tud Δ2 transgene. The Δ2 transgene generates a small C-terminal part of Tud,complementary to the deletion caused by the stop codon in tudB45 (Figs 1, 3). However, as shown in Fig. 4F,H and Table 1, both the tudB45 mutant and Δ2 transgene are active in abdomen development.

Consistent with normal abdomen formation, nos RNA localization was not affected in tudA7, tudA36, tudB42 and tudB45(Fig. 4C-F). Similarly, the mini-tud Δ3 transgene was able to rescue noslocalization in a tud1 mutant background (compare Fig. 4B with 4G). One copy of mini-tud Δ2 partially rescued nos localization in tud1 mutants, 83.8% embryos show weak noslocalization (Fig. 4H) and 16.2% exhibit no nos localization (n=37), compared with 50.4% (Fig. 4B) and 47.8% in tud1 control embryos (n=109), respectively. nos RNA localization was markedly reduced in tudC30 and tudC73 (data not shown),all of which produced no detectable protein and exhibited significant abdomen defects. Taken together, these results suggest that abdomen formation may depend on the number of Tud domains expressed and the total amount of Tud protein present, rather than on the function of a specific Tud domain.

Germ cell formation requires individual Tudor domains

In contrast to abdomen formation, Tud function is absolutely required for germ cell formation. Like the Tud null mutation, all strong tudalleles, which fail to produce detectable Tud protein, display a grandchildless phenotype: the progeny of mutant females are sterile. By contrast, embryos from females carrying certain tud alleles that produce protein form some germ cells, and these embryos grow up into fertile adults (Fig. 5A)(Boswell and Mahowald, 1985). These mutant proteins include: Tud4, which carries a small,in-frame deletion in Tud domain 7; TudA36, which has a point mutation in Tud domain 1; and TudA7, which is truncated after the last Tud domain. This suggests that Tud domains 1 and 7 may not be crucial for germ cell formation. Furthermore, the fact that Tud protein levels were reduced about fivefold in tudA7 mutant embryos when compared with wild type suggests that even small amounts of the complete set of Tud domains are sufficient for germ cell formation. However, although germ cells did form and fertile progeny were produced, the number of germ cells per embryo was greatly reduced in these mutants(Table 2)(Boswell and Mahowald,1985).

Table 2.

Germ cell formation defects in tud mutant embryos

Mutants*Percentage of embryos with germ cells (number of embryos scored)Average number of germ cells (range of germ cells/embryo; number of embryos counted)
Wild type 100 (166) 24.2 (18-32; 31) 
A36 63.6 (132) 3.6 (1-10; 30) 
A7 2.4 (211) ND 
Mutants*Percentage of embryos with germ cells (number of embryos scored)Average number of germ cells (range of germ cells/embryo; number of embryos counted)
Wild type 100 (166) 24.2 (18-32; 31) 
A36 63.6 (132) 3.6 (1-10; 30) 
A7 2.4 (211) ND 
*

tud mutant embryos were produced by females transheterozygous for the respective tud mutation and tud deletion(Df(2R)PurP133, see Materials and methods). Wild-type control embryos were from tud+/Df(2R)PurP133mothers.

The number of germ cells was not determined (ND) because very few embryos form germ cells.

Two mutations, tudB42, a point mutation in Tud domain 10, and tudB45, a truncation that deletes Tud domains 10 and 11, produced no germ cells. The tudB42 allele is particularly interesting as it suggests a more specific function for Tud domain 10 in germ cell formation. Several arguments support this conclusion. First, the mutant protein is quite stable (53.8% of the wild-type control; see Fig. 1C), suggesting that protein levels alone cannot account for the lack of germ cells. Second, even two copies of the tudB42 mutant gene or a tudB42/tudB45 mutant combination failed to generate germ cells (Fig. 5A; data not shown). This rules out a dosage effect as was described for tud4 mutants, which produced less germ cells in trans to a gene deletion (Boswell and Mahowald, 1985) (data not shown). Finally, a mutation in the homologous amino acid in Tud domain 1 (tudA36) produced germ cells, suggesting that this particular amino acid change in Tud domain 10 affects the function of the domain (Fig. 2A′) and results in a domain-specific germ cell defect.

Fig. 4.

Abdomen development in tudor domain mutants and mini-tud transgenes. (A-H) Left panels: in situ experiments showing localization of nos RNA in preblastoderm embryos. Right panels: dark-field photographs of larval cuticles. anterior left, dorsal up. (A) Wild type. Diagram of full-length Tud with Tud domains indicated with blue squares is shown at the top. (B) tud1, (C) tudA36, (D) tudB42, (E) tudA7, (F) tudB45, (G)mini-tud Δ3 and (H) mini-tud Δ2 transgenes. Deleted Tud domains are indicated by the white squares.

Fig. 4.

Abdomen development in tudor domain mutants and mini-tud transgenes. (A-H) Left panels: in situ experiments showing localization of nos RNA in preblastoderm embryos. Right panels: dark-field photographs of larval cuticles. anterior left, dorsal up. (A) Wild type. Diagram of full-length Tud with Tud domains indicated with blue squares is shown at the top. (B) tud1, (C) tudA36, (D) tudB42, (E) tudA7, (F) tudB45, (G)mini-tud Δ3 and (H) mini-tud Δ2 transgenes. Deleted Tud domains are indicated by the white squares.

Fig. 5.

The role of individual Tudor domains for germ cell formation.(A) Cellular blastoderm stage embryos from wild-type (wt), tudmutant (A36, A7, B42 and tud1) and tud1 females that express the mini-tud Δ3 transgene; anti-Vasa antibody marks germ cells. (B,C) In situ experiments showing gcl (B) and pgc (C) RNA staining in wild type and different tud mutant embryos generated by females transheterozygous for the respective tud allele and Df(2R)PurP133. For all panels, anterior is to the left and dorsal is up.

Fig. 5.

The role of individual Tudor domains for germ cell formation.(A) Cellular blastoderm stage embryos from wild-type (wt), tudmutant (A36, A7, B42 and tud1) and tud1 females that express the mini-tud Δ3 transgene; anti-Vasa antibody marks germ cells. (B,C) In situ experiments showing gcl (B) and pgc (C) RNA staining in wild type and different tud mutant embryos generated by females transheterozygous for the respective tud allele and Df(2R)PurP133. For all panels, anterior is to the left and dorsal is up.

Our data suggest that germ cell formation requires the function of specific Tud domains. To test this hypothesis more directly, we analyzed the mini-tud Δ3 transgene, which expresses Tud domain 1,and domains 7-11, but lacks five Tud domains and about ∼50% of the entire Tud protein. Embryos from tud1 females carrying this transgene formed germ cells (Fig. 5A). Although at blastoderm stage almost all embryos had some germ cells, only 40% of embryos at stage 11 had germ cells. The average number of germ cells was 3.3, ranging from 1-11 germ cells/embryo. These germ cells were fully functional because they gave rise to gonads capable of making eggs and sperm and, subsequently, to adult progeny (data not shown). The other two transgenes, mini-tud Δ1 and mini-tud Δ2 failed to rescue germ cell formation in the tud1 mutant background (several independent insertion lines were tested for each transgene). mini-tud Δ1 and mini-tud Δ2 transgenes also failed to induce germ cell formation in two other tudmutant backgrounds in which inactive endogenous Tud protein was produced,namely, tudB42 and tudB45 (data not shown). These results demonstrate that only a subset of Tud domains is sufficient to support germ cell formation and that specific domains are absolutely required for germ cell formation.

Tudor domains control localization of germ plasm components

Germ cells form in the germ plasm, which contains RNA and protein components. Localization of these components to the germ plasm has been directly associated with the specializations of the germ plasm, such as polar granules, and the ability of germ plasm to induce germ cells. Two RNAs, germ cell-less (gcl) and polar granule component(pgc) are known to localize to the germ plasm in a Tud-dependent manner (Jongens et al., 1992; Nakamura et al., 1996; Thomson and Lasko, 2004). gcl plays a role in germ cell formation and pgc controls germ cell transcriptional silencing once germ cells have formed(Jongens et al., 1992; Martinho et al., 2004). Localization of these RNAs is severely affected in tud1(Fig. 5B,C)(Jongens et al., 1992; Nakamura et al., 1996) and tud RNA-null mutants (Thomson and Lasko, 2004). Despite their strong defect in the germline development, gcl and pgc RNAs were localized similarly to wild type in tudA36 and tudA7 mutants(Fig. 5B,C). In tudB42 and tudB45 mutants, which form no germ cells, localization of these RNAs was affected, albeit to a lesser extent than in protein and RNA null mutants(Fig. 5B,C)(Thomson and Lasko, 2004). Thus the localization of these RNAs per se is not sufficient to form germ cells.

Fig. 6.

Tudor domains are required for protein localization and for proper architecture of polar granules. (A-D″) Preblastoderm embryos from wild-type (wt), Tud domain mutants tudA36 and tudB42, and tudB45 were co-stained with rabbit anti-Tud (A-D, green channel) and rat anti-Vas antibody (A′-D′, red channel). Overlay images are shown in A ″-D″. (A-A″) Wild type; (B-B″) tudA36;(C-C″) tudB42; (D-D″) tudB45. Anterior is to the left and dorsal is up.(E,F,F′,H-K) Electron micrographs of germ plasm. Polar granules are indicated with arrows; m, mitochondria; MVB,multivesicular body, frequently observed at the egg cortex(Mahowald et al., 1981). (E)Germ plasm of the wild-type embryos shows distinct round or barrellike electron-dense polar granules. (F,F′) Different ultra-thin sections across the same wild-type polar granule demonstrating hollow sphere morphology. (G,G′) A simplified diagram of a polar granule. (H) Polar granules of tudA36 mutant frequently show a string- or rod-like architecture. Polar granule remnants of tudB42 (I) and tudB45 (K) are extremely rare. Scale bars: in E, 500 nm for E,H-K; in F, 500 nm for F,F′.

Fig. 6.

Tudor domains are required for protein localization and for proper architecture of polar granules. (A-D″) Preblastoderm embryos from wild-type (wt), Tud domain mutants tudA36 and tudB42, and tudB45 were co-stained with rabbit anti-Tud (A-D, green channel) and rat anti-Vas antibody (A′-D′, red channel). Overlay images are shown in A ″-D″. (A-A″) Wild type; (B-B″) tudA36;(C-C″) tudB42; (D-D″) tudB45. Anterior is to the left and dorsal is up.(E,F,F′,H-K) Electron micrographs of germ plasm. Polar granules are indicated with arrows; m, mitochondria; MVB,multivesicular body, frequently observed at the egg cortex(Mahowald et al., 1981). (E)Germ plasm of the wild-type embryos shows distinct round or barrellike electron-dense polar granules. (F,F′) Different ultra-thin sections across the same wild-type polar granule demonstrating hollow sphere morphology. (G,G′) A simplified diagram of a polar granule. (H) Polar granules of tudA36 mutant frequently show a string- or rod-like architecture. Polar granule remnants of tudB42 (I) and tudB45 (K) are extremely rare. Scale bars: in E, 500 nm for E,H-K; in F, 500 nm for F,F′.

We next investigated whether the distribution of other germ plasm components, such as Vasa protein and Tud itself, is changed in the mutants. Vasa localization is not affected in tud mutants during early embryogenesis (Fig. 6A′-D′), which places this gene upstream in the germ plasm assembly pathway (Thomson and Lasko,2004). Tud protein localizes to the germ plasm during oogenesis,its localization is maintained during early embryogenesis and the protein is present in germ cells as they form(Bardsley et al., 1993)(Fig. 6A). Tud protein generated by a full-length tud transgene recapitulates this localization pattern (Fig. 7A,D). We analyzed the localization of Tud protein in tudmutants by using an anti-Tud antibody and Tud transgenes that are tagged with an HA epitope. Tud protein is localized to the posterior pole in tudA36 and tud4 mutant embryos, and in embryos carrying the mini-tud Δ3 transgene(Fig. 6B, Fig. 7F; data not shown). Because these three genotypes support significant germ cell formation, we conclude that Tud protein localization is a prerequisite for the function of Tudor in germ cell formation. tudB42, tudA7 and tudB45, and the mini-tud transgenes mini-tud Δ1 and minitudΔ2, showed no Tud protein localization to the posterior pole(Fig. 6C,D, Fig. 7E; data not shown). With the exception of tudA7, which rarely produces germ cells,these mutants and transgenes are defective in germ cell formation. Furthermore, the fact that Tud protein with a point mutation in Tud domain 10(tudB42, Fig. 6C) fails to localize to the posterior pole, whereas a point mutation at the same position in Tud domain 1 does not affect localization(Fig. 6B), suggests a specific role of Tud domain 10 in Tud protein localization and/or transport.

Fig. 7.

Tudor localization to the nuage is not required for germ plasm localization and germ cell formation. (A-F) Ovaries (A-C) and preblastoderm embryos (D-F) expressing full-length tud and mini-tud transgenes were stained with anti-HA antibody to determine transgenic Tud protein distribution during oogenesis and early embryogenesis. Nuage localization is indicated with arrows and germ plasm localization in the oocyte with arrowheads. (A,D) Full-length Tud; (B,E) mini-tudΔ1 protein; (C,F) mini-tud Δ3 protein.

Fig. 7.

Tudor localization to the nuage is not required for germ plasm localization and germ cell formation. (A-F) Ovaries (A-C) and preblastoderm embryos (D-F) expressing full-length tud and mini-tud transgenes were stained with anti-HA antibody to determine transgenic Tud protein distribution during oogenesis and early embryogenesis. Nuage localization is indicated with arrows and germ plasm localization in the oocyte with arrowheads. (A,D) Full-length Tud; (B,E) mini-tudΔ1 protein; (C,F) mini-tud Δ3 protein.

At stage 10 of oogenesis, Tud is localized to the posterior pole and is part of the nuage that surrounds the nuclear envelope of the nurse cells(Bardsley et al., 1993). We confirmed this localization pattern using the more specific HA epitope tag of the full-length tud transgene(Fig. 7A). We next analyzed the localization of Tud protein synthesized by the mini-tud transgenes. All transgenes are expressed during oogenesis, however, only mini-tudΔ1 and mini-tud Δ3 show a specific localization pattern,whereas the two Tud domains, Tud 10 and 11, expressed in mini-tudΔ2 are not sufficient for posterior or nuage localization (data not shown). mini-tud Δ1 protein, containing Tud domains 1-6,localized to the nuage but not to the germ plasm(Fig. 7B,E). By contrast,mini-tud Δ3 protein, which contains Tud domains 1 and 7-11,failed to localize to the nuage during oogenesis but localized well to the germ plasm of oocytes (Fig. 7C)and early embryos (Fig. 7F). Because mini-tud Δ3, but not mini-tud Δ1 protein, is able to support germ cell formation, we conclude that Tud localization to the nuage is not absolutely required for germ cell formation. These results further support the conclusion that specific Tud domains control Tud protein localization.

Tudor domains control polar granule architecture

Tud protein is a component of the polar granules, and controls the size and number of polar granules (Boswell and Mahowald, 1985; Thomson and Lasko, 2004). At the ultrastructural level, polar granules are large electron-dense RNA-protein spheres that are in close association with mitochondria during late oogenesis and early embryogenesis(Mahowald, 1968)(Fig. 6E). Polar granules undergo dynamic changes as germ cells form and contribute to the nuage that is associated with the nuclear envelope of germ cells throughout the life cycle(Mahowald, 1971). Tud protein and RNA null mutants form very few polar granule-like structures, which are considerably smaller than wild type (data not shown)(Amikura et al., 2001; Boswell and Mahowald, 1985; Thomson and Lasko, 2004). Some tud mutants that express protein formed fewer and smaller polar granules (Fig. 6I,K)(Boswell and Mahowald, 1985). In particular, these granules lacked the characteristic `hollow sphere'morphology, where the inner core of the granule is less electron-dense than the periphery (Fig. 6F′). Interestingly, tudA36 (Tud domain 1 mutated) produced a normal number of polar granules, but the granules had a strikingly different morphology (Fig. 6H). Contrary to wild-type granules, which are round or `barrel'-like(Fig. 6E), tudA36 mutant polar granules had a `string' or `rod'-like shape (Fig. 6H). Like wild type, these mutant polar granules associated with mitochondria but were not found in the periplasm of the anterior pole or in other parts of the embryo(data not shown). These findings suggest that specific Tud domains play a direct role in polar granule architecture.

In this study, we analyzed tud mutants and transgenes to address the role of individual Tud domains in the large Tud protein. Our data demonstrate that single Tud domains play distinct and crucial roles in germ cell formation: (1) by controlling Tud protein localization to the germ plasm;and (2) by maintaining the architecture of polar granules.

Analysis of Tud domains 1 and 10, both of which carry a point mutation in the same arginine residue in tudA36 and tudB42, respectively, predicts that this arginine faces the solvent and that mutations in this residue do not affect the overall structure of the domains. Furthermore, the arginine is in close proximity to the cluster of hydrophobic amino acids that in Smn form a binding pocket for interaction with other proteins (Selenko et al., 2001; Sprangers et al., 2003) (see Results). In Smn, target recognition is dependent not only on the hydrophobic cluster but also on E134, a glutamate located nearby (Fig. 2C). Tud-domain proteins can interact with flexible peptides carrying methylated amino acids and it is possible that charged amino acids close to the hydrophobic pocket,like the arginines in Tud domains 1 and 10, and glutamate 134, act as a gateway, contributing to the recognition of specific targets. Recently, a new structure of Tud domains was identified in the protein JMJD2A, which revealed an intertwined folding of two Tud domains(Huang et al., 2006). Other tandem Tud domain structures have been reported(Charier et al., 2004; Huyen et al., 2004; Ramos et al., 2006), and the analysis of sequences from these domains show that the two domains are separated by no more than 20-30 amino acids. As individual Tud domains in Tud are separated by no less than ∼100 amino acids, and because we were able to create functional proteins after the deletion of large parts of Tud protein, we presently do not have any evidence predicting such dual domains in Tudor.

Despite the virtual lack of polar granules in tudB42and tudB45 mutants, we detected substantial (albeit reduced) germ plasmspecific accumulation of pgc RNA, although we confirmed previous results that failed to find pgc RNA localized to the germ plasm of strong tud mutants(Nakamura et al., 1996; Thomson and Lasko, 2004). Because pgc RNA can accumulate in germ plasm that lacks clearly discernable polar granules, we conclude that some localization and anchoring of RNA to the germ plasm can occur independently of complete polar granule assembly and that smaller particles containing germ plasm components may be sufficient to tether RNA. The role of Tud in germ plasm formation may be to assemble these pre-particles into a larger order granule. Because abdomen formation and nos RNA localization were normal in tudB42 and tudB45 mutants, we propose that these `pre-particles' may be sufficient to promote noslocalization and translational derepression at the posterior pole.

For germ cell formation, specific Tud domains are essential and it is likely that these individual domains interact with specific partner proteins. Similar to Smn protein and other Tud domain proteins, these partners are likely to be methylated. Indeed, two germline proteins, Valois and Capsuléen, are components of the Drosophila methylosome and required for germ cell formation (Anne and Mechler, 2005; Cavey et al.,2005; Schüpbach and Wieschaus, 1986). In particular, Capsuléen is a homolog of the mammalian PRMT5 methyltransferase that has been recently implicated in germline specification in the mouse(Ancelin et al., 2006). Anne and Mechler identified a particular region in Valois that interacts with Tud in vitro and their analysis suggests that the interaction of Tud with the methylosome may tether Tud to the posterior pole, possibly via specific methylated binding partners (Anne and Mechler, 2005). Our analysis of transgenes lacking different Tud domains showed that mini-tud Δ3 is sufficient for germ cell formation and abdomen segmentation. This transgene construct lacks the Tud segment that was responsible for the strong interaction with Valois protein in vitro (Anne and Mechler, 2005). The ability of mini-tud Δ3 to induce germ cell formation indicates that the Tud-Valois interaction may not be absolutely necessary for germ cell formation. However, this interaction may be required for efficient germline development, as mini-tud Δ3 could not generate a normal number of germ cells (see Results). Alternatively, the weak binding detected between Valois and other Tud fragments that overlap with regions present in mini-tud Δ3 (Anne and Mechler, 2005) may be sufficient for the formation of some germ cells.

Tud protein localizes to both the nuage, an electron-dense material associated with nurse cell nuclei, and the germ plasm(Bardsley et al., 1993)(Fig. 7A,D). Besides Tud, three other proteins, Vasa, Aubergine and Valois are found in both the nuage and the germ plasm (Anne and Mechler,2005; Harris and Macdonald,2001; Hay et al.,1988; Liang et al.,1994), and it has been suggested that the nuage forms a precursor stage of germ plasm assembly during oogenesis. This notion is supported by the finding that Vasa localization to both the nuage and the germ plasm was equally affected in vasa mutants(Liang et al., 1994). However,our analysis of mini-tud transgenes shows that nuage localization is not necessary for Tud localization to the germ plasm or for germ cell formation. Thus, Tud localization to the germ plasm and its function in germ cell formation can be uncoupled from its association with the nuage during oogenesis. These results are consistent with findings by Snee and Macdonald(Snee and Macdonald, 2004),who showed using in vivo imaging that posteriorly localized Aubergine is not transported to the germ plasm as a protein associated with nuage particles. Thus, the role of the perinuclear nuage and the function of Tud in this organelle remain to be elucidated.

Note added in proof

Recently Gonsalvez et al. (Gonsalvez et al., 2006) showed that absence of the human PMRT5 homolog Dart5 [referred to as capsuléen by Anne and Mechler(Anne and Mechler, 2005)] in Drosophila results in the loss of symmetric arginine dimethylation on spliceosomal Sm proteins and produces a phenotype resembling that of tudor mutants.

We thank T. Thomson and P. Lasko for sharing unpublished data and reagents. We thank the Bloomington Stock Center for providing fly stocks. We are indebted to V. Barbosa for organizing the 2R maternal-effect screen and to other members of the screen team: E. Arkova, F. Kimm, P. Kunwar, T. Marty, A. Renault, H. Sano and H. Zinszner. We thank F. Fraternali for useful discussions on the modeling procedure, and L. Cummins, J. Jimenez and F. Macaluso at Albert Einstein College of Medicine for their help with electron microscopy. We also thank the members of the Lehmann laboratory for critically reading the manuscript, especially R. Cinalli. A.L.A. was supported by a postdoctoral fellowship from the American Cancer Society. R.L. is an HHMI investigator.

Amikura, R., Hanyu, K., Kashikawa, M. and Kobayashi, S.(
2001
). Tudor protein is essential for the localization of mitochondrial RNAs in polar granules of Drosophila embryos.
Mech. Dev
.
107
,
97
-104.
Amodeo, P., Fraternali, F., Lesk, A. M. and Pastore, A.(
2001
). Modularity and homology: modelling of the titin type I modules and their interfaces.
J. Mol. Biol.
311
,
283
-296.
Ancelin, K., Lange, U. C., Hajkova, P., Schneider, R.,Bannister, A. J., Kouzarides, T. and Surani, M. A. (
2006
). Blimp1 associates with Prmt5 and directs histone arginine methylation in mouse germ cells.
Nat. Cell Biol
.
8
,
623
-630.
Anne, J. and Mechler, B. M. (
2005
). Valois, a component of the nuage and pole plasm, is involved in assembly of these structures, and binds to Tudor and the methyltransferase Capsuléen.
Development
132
,
2167
-2177.
Bardsley, A., McDonald, K. and Boswell, R. E.(
1993
). Distribution of tudor protein in the Drosophila embryo suggests separation of functions based on site of localization.
Development
119
,
207
-219.
Boswell, R. E. and Mahowald, A. P. (
1985
). tudor, a gene required for assembly of the germ plasm in Drosophila melanogaster.
Cell
43
,
97
-104.
Brahms, H., Meheus, L., de Brabandere, V., Fischer, U. and Lührmann, R. (
2001
). Symmetrical dimethylation of arginine residues in spliceosomal Sm protein B/B' and the Sm-like protein LSm4, and their interaction with the SMN protein.
RNA
7
,
1531
-1542.
Bühler, D., Raker, V., Lührmann, R. and Fischer,U. (
1999
). Essential role for the tudor domain of SMN in spliceosomal U snRNP assembly: implications for spinal muscular atrophy.
Hum. Mol. Genet
.
8
,
2351
-2357.
Cavey, M., Hijal, S., Zhang, X. and Suter, B.(
2005
). Drosophila valois encodes a divergent WD protein that is required for Vasa localization and Oskar protein accumulation.
Development
132
,
459
-468.
Charier, G., Couprie, J., Alpha-Bazin, B., Meyer, V.,Quéméneur, E., Guérois, R., Callebaut, I., Gilquin, B. and Zinn-Justin, S. (
2004
). The Tudor tandem of 53BP1: a new structural motif involved in DNA and RG-rich peptide binding.
Structure
12
,
1551
-1562.
Côté, J. and Richard, S. (
2005
). Tudor domains bind symmetrical dimethylated arginines.
J. Biol. Chem
.
280
,
28476
-28483.
Ephrussi, A. and Lehmann, R. (
1992
). Induction of germ cell formation by oskar.
Nature
358
,
387
-392.
Gavis, E. R. and Lehmann, R. (
1994
). Translational regulation of nanos by RNA localization.
Nature
369
,
315
-318.
Golumbeski, G. S., Bardsley, A., Tax, F. and Boswell, R. E.(
1991
). tudor, a posterior-group gene of Drosophila melanogaster,encodes a novel protein and an mRNA localized during mid-oogenesis.
Genes Dev
.
5
,
2060
-2070.
Gonsalvez, G. B., Rajendra, T. K., Tian, L. and Matera, A. G. (
2006
). The Smprotein methyl transferase, dart5, is essential for germ-cell specification and maintenance.
Curr. Biol.
16
,
1077
-1089.
Harris, A. N. and Macdonald, P. M. (
2001
). Aubergine encodes a Drosophila polar granule component required for pole cell formation and related to eIF2C.
Development
128
,
2823
-2832.
Hay, B., Jan, L. Y. and Jan, Y. N. (
1988
). A protein component of Drosophila polar granules is encoded by vasa and has extensive sequence similarity to ATP-dependent helicases.
Cell
55
,
577
-587.
Huang, Y., Fang, J., Bedford, M. T., Zhang, Y. and Xu, R. M.(
2006
). Recognition of histone H3 lysine-4 methylation by the double tudor domain of JMJD2A.
Science
312
,
748
-751.
Huyen, Y., Zgheib, O., Ditullio, R. A., Jr, Gorgoulis, V. G.,Zacharatos, P., Petty, T. J., Sheston, E. A., Mellert, H. S., Stavridi, E. S. and Halazonetis, T. D. (
2004
). Methylated lysine 79 of histone H3 targets 53BP1 to DNA double-strand breaks.
Nature
432
,
406
-411.
Illmensee, K. and Mahowald, A. P. (
1974
). Transplantation of posterior polar plasm in Drosophila. Induction of germ cells at the anterior pole of the egg.
Proc. Natl. Acad. Sci. USA
71
,
1016
-1020.
Jongens, T. A., Hay, B., Jan, L. Y. and Jan, Y. N.(
1992
). The germ cell-less gene product: a posteriorly localized component necessary for germ cell development in Drosophila.
Cell
70
,
569
-584.
Kim, J., Daniel, J., Espejo, A., Lake, A., Krishna, M., Xia, L.,Zhang, Y. and Bedford, M. T. (
2006
). Tudor, MBT and chromo domains gauge the degree of lysine methylation.
EMBO Rep
.
7
,
397
-403.
Koradi, R., Billeter, M. and Wüthrich, K.(
1996
). MOLMOL: a program for display and analysis of macromolecular structures.
J. Mol. Graph
.
14
,
51
-55.
Lehmann, R. and Tautz, D. (
1994
). In situ hybridization to RNA.
Methods Cell Biol
.
44
,
575
-598.
Liang, L., Diehl-Jones, W. and Lasko, P.(
1994
). Localization of vasa protein to the Drosophila pole plasm is independent of its RNA-binding and helicase activities.
Development
120
,
1201
-1211.
Lindsley, D. L. and Zimm, G. G. (
1992
).
The Genome of Drosophila melanogaster
. San Diego:Academic Press.
Lohs-Schardin, M., Cremer, C. and Nüsslein-Volhard, C.(
1979
). A fate map for the larval epidermis of Drosophila melanogaster: localized cuticle defects following irradiation of the blastoderm with an ultraviolet laser microbeam.
Dev. Biol
.
73
,
239
-255.
Mahowald, A. P. (
1968
). Polar granules of Drosophila. II. Ultrastructural changes during early embryogenesis.
J. Exp. Zool
.
167
,
237
-261.
Mahowald, A. P. (
1971
). Polar granules of Drosophila. 3. The continuity of polar granules during the life cycle of Drosophila.
J. Exp. Zool
.
176
,
329
-343.
Mahowald, A. P., Allis, C. D. and Caulton, J. H.(
1981
). Rapid appearance of multivesicular bodies in the cortex of Drosophila eggs at ovulation.
Dev. Biol
.
86
,
505
-509.
Martinho, R. G., Kunwar, P. S., Casanova, J. and Lehmann, R.(
2004
). A noncoding RNA is required for the repression of RNApolII-dependent transcription in primordial germ cells.
Curr. Biol
.
14
,
159
-165.
Maurer-Stroh, S., Dickens, N. J., Hughes-Davies, L., Kouzarides,T., Eisenhaber, F. and Ponting, C. P. (
2003
). The Tudor domain `Royal Family': Tudor, plant Agenet, Chromo, PWWP and MBT domains.
Trends Biochem. Sci
.
28
,
69
-74.
Moore, L. A., Broihier, H. T., Van Doren, M., Lunsford, L. B. and Lehmann, R. (
1998
). Identification of genes controlling germ cell migration and embryonic gonad formation in Drosophila.
Development
125
,
667
-678.
Nakamura, A., Amikura, R., Mukai, M., Kobayashi, S. and Lasko,P. F. (
1996
). Requirement for a noncoding RNA in Drosophila polar granules for germ cell establishment.
Science
274
,
2075
-2079.
Navarro, C., Puthalakath, H., Adams, J. M., Strasser, A. and Lehmann, R. (
2004
). Egalitarian binds dynein light chain to establish oocyte polarity and maintain oocyte fate.
Nat. Cell Biol.
6
,
427
-435.
Ponting, C. P. (
1997
). Tudor domains in proteins that interact with RNA.
Trends Biochem. Sci
.
22
,
51
-52.
Ramos, A., Hollingworth, D., Adinolfi, S., Castets, M., Kelly,G., Frenkiel, T. A., Bardoni, B. and Pastore, A. (
2006
). The structure of the N-terminal domain of the fragile X mental retardation protein: a platform for protein-protein interaction.
Structure
14
,
21
-31.
Schüpbach, T. and Wieschaus, E. (
1986
). Germline autonomy of maternal-effect mutations altering the embryonic body pattern of Drosophila.
Dev. Biol
.
113
,
443
-448.
Schwede, T., Kopp, J., Guex, N. and Peitsch, M. C.(
2003
). SWISS-MODEL: an automated protein homology-modeling server.
Nucleic Acids Res
.
31
,
3381
-3385.
Selenko, P., Sprangers, R., Stier, G., Bühler, D., Fischer,U. and Sattler, M. (
2001
). SMN tudor domain structure and its interaction with the Sm proteins.
Nat. Struct. Biol
.
8
,
27
-31.
Serano, T. L., Cheung, H. K., Frank, L. H. and Cohen, R. S.(
1994
). P element transformation vectors for studying Drosophila melanogaster oogenesis and early embryogenesis.
Gene
138
,
181
-186.
Snee, M. J. and Macdonald, P. M. (
2004
). Live imaging of nuage and polar granules: evidence against a precursor-product relationship and a novel role for Oskar in stabilization of polar granule components.
J. Cell Sci.
117
,
2109
-2120.
Spradling, A. C. (
1986
). P element-mediated transformation. In
Drosophila: A Practical Approach
(ed. D. B. Roberts), pp.
175
-197. Oxford: IRL Press.
Sprangers, R., Groves, M. R., Sinning, I. and Sattler, M.(
2003
). High-resolution X-ray and NMR structures of the SMN Tudor domain: conformational variation in the binding site for symmetrically dimethylated arginine residues.
J. Mol. Biol.
327
,
507
-520.
Stein, J. A., Broihier, H. T., Moore, L. A. and Lehmann, R.(
2002
). Slow as molasses is required for polarized membrane growth and germ cell migration in Drosophila.
Development
129
,
3925
-3934.
Talbot, K., Miguel-Aliaga, I., Mohaghegh, P., Ponting, C. P. and Davies, K. E. (
1998
). Characterization of a gene encoding survival motor neuron (SMN)-related protein, a constituent of the spliceosome complex.
Hum. Mol. Genet
.
7
,
2149
-2156.
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. and Higgins, D. G. (
1997
). The CLUSTAL_X windows interface:flexible strategies for multiple sequence alignment aided by quality analysis tools.
Nucleic Acids Res.
25
,
4876
-4882.
Thomson, T. and Lasko, P. (
2004
). Drosophila tudor is essential for polar granule assembly and pole cell specification, but not for posterior patterning.
Genesis
40
,
164
-170.
Thomson, T. and Lasko, P. (
2005
). Tudor and its domains: germ cell formation from a Tudor perspective.
Cell Res
.
15
,
281
-291.
Vriend, G. (
1990
). WHAT IF: a molecular modeling and drug design program.
J. Mol. Graph
.
8
,
52
-56.
Wang, C. and Lehmann, R. (
1991
). Nanos is the localized posterior determinant in Drosophila.
Cell
66
,
637
-647.
Wang, C., Dickinson, L. K. and Lehmann, R.(
1994
). Genetics of nanos localization in Drosophila.
Dev. Dyn
.
199
,
103
-115.
Wieschaus, E. and Nüsslein-Volhard, C.(
1998
). Looking at embryos. In
Drosophila: A Practical Approach
(ed. D. B. Roberts), pp.
179
-214. Oxford: IRL Press.