Pelota-interacting G protein Hbs1 is required for spermatogenesis in Drosophila

Hbs1, which is homologous to the GTPase eRF3, is a small G protein implicated in mRNA quality control. It interacts with a translation-release factor 1-like protein Dom34/Pelota to direct decay of mRNAs with ribosomal stalls. Although both proteins are evolutionarily conserved in eukaryotes, the biological function of Hbs1 in multicellular organisms is yet to be characterized. In Drosophila, pelota is essential for the progression through meiosis during spermatogenesis and germline stem cell maintenance. Here we show that homozygous Hbs1 mutant flies are viable, female-fertile, but male-sterile, which is due to defects in meiosis and spermatid individualization, phenotypes that are also observed in pelota hypomorphic mutants. In contrast, Hbs1 mutants have no obvious defects in germline stem cell maintenance. We show that Hbs1 genetically interacts with pelota during spermatid individualization. Furthermore, Pelota with a point mutation on the putative Hbs1-binding site cannot substitute the wild type protein for normal spermatogenesis. These data suggest that Pelota forms a complex with Hbs1 to regulate multiple processes during spermatogenesis. Our results reveal a specific requirement of Hbs1 in male gametogenesis in Drosophila and indicate an essential role for the RNA surveillance complex Pelota-Hbs1 in spermatogenesis, a function that could be conserved in mammals.

of the stop codon at A site of ribosome and hydrolysis of the peptidyl-tRNA bond for peptide release 16 . eRF1 binds to eRF3 through its C-terminal domain and this interaction increases the affinity of eRF3 for GTP 17 . Similarly, Pelo-Hbs1 interaction is able to augment the affinity constant of Hbs1 to GTP 14 . Structure of the Dom34-Hbs1 complex confirms the analogy of the overall similarity to that of eRF1-eRF3, but also reveals that the N-terminal region of Pelo may function differently [18][19][20] . The binding of Hbs1 to Dom34 leads to a conformational change with a tRNA-like structure formation by the middle domain of Dom34 and positioning a conserved basic region expected for translational termination. Indeed, Dom34-Hbs1 shows maximum kinetic efficiency on the ribosome complexes that contain truncated mRNAs 21 and promotes dissociation of stalled elongation complexes and release of peptidyl-tRNA in an A site codon-independent manner 10 . Additionally, the Dom34/Hbs1/Rli1 complex mediates the dissociation of the stalled ribosomes into 40 S and 60 S subunits. Taken together, the Pelo-Hbs1 complex regulates gene expression at translational level by facilitating the recycle of ribosomes and removal of aberrant mRNAs when elongation stalls occur.
The Pelo-Hbs1 complex seems to be conserved throughout evolution. Although Hbs1 is absent in archaea, its function is performed by elongation factor aEF1a 22,23 . In mammals, Pelo-Hbs1 also promotes dissociation of elongation complexes, but only with an ABC-type ATPase ABCE1 in the complex 12,24,25 . Also, mammalian Pelo seems to have additional roles outside of mRNA quality control and the pleiotropic function of Pelo might be also reflected by the observation that pelo null mice die as early as embryonic day 7.5 12,26 . Conditional knock-out of Pelo at postnatal stages revealed that Pelo is essential for the maintenance of spermatogonial stem cells by regulating PI3K/AKT pathway and the downstream FOXO1 activity 27 . In Drosophila, mutation in pelo also causes loss of ovarian GSCs, in part as a result of compromised BMP signaling pathway activation 28 . These observations indicate an evolutionarily conserved function of Pelo in GSCs and gonad development. We have previously shown that Pelo-Hbs1 is required for transposon silencing in the female germline 29 . However, the biological function of Hbs1 and Pelo-Hbs1 complex beyond female germline remains to be explored in multicellular organisms.
Here in this study, we investigated the biological function of Drosophila Hbs1 by generating and analyzing Hbs1 mutants. We find that Hbs1 mutants are male sterile due to defects in meiosis and spermatid individualization during spermatogenesis. We show that Pelo is also required for spermatid individualization in addition to its known role in early meiosis during spermatogenesis, revealing that Pelo functions at multiple stages during male germline development. By interaction analysis, we show that Hbs1 and Pelo genetically and physically interact with each other. We thus demonstrate an essential role for the Pelo-Hbs1 complex in spermatogenesis, likely through translational regulation.

Results
Hbs1 is required for male fertility in Drosophila. ORF of Hbs1 (CG1898) in Drosophila melanogaster encodes a polypeptide of 670 amino acids, with a predicated molecular weight of approximately 74 kDa. Hbs1 is a highly conserved protein in eukaryotic kingdoms, with the full length of fly counterpart sharing a similarity of 48%, 37%, 48%, 45% and 45% with the homologs in yeast, C. elegans, Xenopus, chicken and human, respectively (Fig. S1). Drosophila homologue of Hbs1 contains two conserved domains, a 240 amino acid N-terminal domain of unknown function and a 430 amino acid C-terminal domain bearing four GTP binding sites. Hbs1 is located at the cytological position 62B on the left arm of chromosome 3. To study its biological function, we conducted P element-mediated excision from a P-element insertion allele Hbs1 EY09557 to generate Hbs1 mutants. Via PCR screens, we identified three deletion or insertion alleles among 400 excised lines. One allele, named Hbs1 1 , has a 2307 bp deletion (spans from 1863069 to 1865375 on chromosome 3 L) that covers the entire coding region of Hbs1 (Fig. 1A), and is therefore a null allele. Two other alleles, Hbs1 48 and Hbs1 172 , has a 420 bp and a 15 bp insertion, respectively, left by the original P-element (Fig. 1A). RT-PCR analysis revealed that the Hbs1 1 null allele led to the absence of Hbs1 expression, while the Hbs1 48 and Hbs1 172 alleles showed down-regulation of Hbs1 expression (Fig. 1B). All Hbs1 mutants were homozygous viable. Moreover, Hbs1 1 homozygous females were fertile and produced normal offspring, ruling out a possible maternal function of Hbs1 for viability. However, for all three alleles, homozygous mutant males were sterile (Fig. 1F). Mutant testes looked morphologically normal. However, there was no sperm found in seminal vesicles (SV) and mutant SVs were much smaller than that in control flies. (Fig. 1C,D), suggesting that no mature sperm is formed in the absence of Hbs1. This sterility was efficiently rescued by a transgene containing the genomic region of Hbs1 (Fig. 1E,F), demonstrating that Hbs1 mutation is solely responsible for the male-sterile phenotype.

Hbs1 mutants are defective at meiosis and spermatid individualization during spermatogenesis.
We next investigated the underlying mechanisms of spermatogenesis failure caused by the loss of Hbs1. Meiotic figures with one dark nebenkern and one light nucleus of similar size, which are the results of correct meiotic divisions, could be observed in the 64 connected spermatids of squashed wild-type testis by phase contrast microscopy ( Fig. 2A), as previously described 1 . Correct meiotic divisions depend on the accurate chromosome segregation and cytokinesis. Arrest of cytokinesis during the first or second division in meiosis results in spermatids with two or four normal size nuclei associated with an abnormally large nebenkern. Nuclear size is proportional to its chromosome content in the onion-stage spermatids 1 . Unsuccessful chromosome segregation results in spermatids with abnormal numbers of nuclei of abnormal size, even with micronuclei 30,31 . Failure in both cytokinesis and chromosome segregation together produces spermatids with large nebenkern associated with multiple nuclei of different sizes 32,33 . Spermatogonia and spermatocytes from Hbs1 1 testis showed comparable sizes to the wild type counterpart but abnormal nebenkern to nucleus ratio of 1:2, 1:3 and 1:4 ( Fig. 2B,C, Table 1). Irregular size of nebenkerns and nuclei in the spermatids was frequently observed in mutant testes (Fig. 2B,C,F, Table 1), suggesting the presence of a defect in Hbs1 1 germline cysts to properly undergo cytokinesis and progress through meiosis. The spindle checkpoint is not stringent in spermatocytes and causes only a small delay in meiotic progression 31,33-35 , which is revealed by the presence of only 22.1% spermatids containing nuclei of www.nature.com/scientificreports www.nature.com/scientificreports/  cysts, heterogeneous mixture of ratios of nebenkern to nucleus (from 1:1, 1:2, 1:3 to 1:4) and abnormal large nebenkern(arrow heads) with small or micronuclei(arrows) were frequently observed, which was evidence of failure of cytokinesis and chromosome mis-segregation (See Table 1 for the quantitative data). (D) The meiotic figure appeared normal in Hbs1 48−/− mutant cysts. (E) Normal onion stage spermatids were observed in the Hbs1 genomic rescued testis. (F) There was greater percentage of total irregular spermatids in the Hbs1 1−/− testes. (G,H) The wildtype testis (G-G') exhibited fully formed and well-focused spindle poles, and the condensed chromosomes were either at the metaphase plate (arrowhead) or beginning to separate (arrow). In Hbs1 1−/− testes, the spindle microtubules failed to converge into the poles and were splayed outward, and the chromosome could not be correctly separated (arrow).α-tubulin staining is in red, and DAPI is in blue. Scale bars = 10 μm.
www.nature.com/scientificreports www.nature.com/scientificreports/ abnormal size, as shown in Table 1. The meiosis defect was rescued by the transgene carrying the genomic region of Hbs1 (Fig. 2E,F. Table 1), confirming that Hbs1 is required for meiosis during spermatogenesis. In addition, we stained for DNA and α-tubulin to directly examine the spindle formation during meiotic division in wild type and Hbs1 mutant testes. We frequently found that the spindle microtubules failed to converge into the poles and were splayed outward in Hbs1 1−/− testes and consequently the chromosome could not be correctly separated (Fig. 2G,H'). These observations provide an explanation to the initially observed abnormal-sized nuclei phenotype found in Hbs1 1−/− testes.
Interestingly, in Hbs1 48 (Fig. 2D,F) or Hbs1 172 homozygous mutant testes, the abnormal germline cysts found in the null mutants were not observed. But still, Hbs1 48 and Hbs1 172 homozygous testes contained no sperm in the seminal vesicles. Moreover, there were also some cysts with normal meiotic figures in Hbs1 1 testes, yet no sperm was detected in the mutant seminal vesicles, suggesting that Hbs1 is also required during later spermatogenesis.
Individualization is the final stage of spermatogenesis, during which the syncytial membrane is remodeled and a whole cyst of 64 syncytial spermatids is divided into 64 individual sperm ( Fig. 3A) 36,37 . Individualization complex (IC), which consists of 64 actin cones, is required for this process. At the beginning of individualization, the actin cones assemble around the spermatid nuclei towards the basal end of the testis (Fig. 3A,B,D), and then move away from the nuclei along the spermatid axonemes to the apical end of the testis (Fig. 3A

Hbs1 genetically interacts with pelo during spermatogenesis. Since both pelo and Hbs1 mutants in
Drosophila showed defects in meiosis during spermatogenesis, we asked whether Pelo and Hbs1 could function together to regulate spermatogenesis. The existing allele of pelo, namely pelo 1 , is a P-element-insertional allele, which produces a truncated transcript of the gene. Mutant fly homozygous for pelo 1 has an early cell cycle arrest at the first meiotic cell division during spermatogenesis 3 . Hbs1 mutants apparently affect both meiosis and individualization during spermatogenesis, but it remains unknown whether pelo also plays a role in spermiogenesis besides its function in meiosis. We therefore generated several new pelo alleles by imprecise P-element-mediated excision 29 . We classified these alleles from weak to strong according to the severity of phenotypes and molecular lesions, as described in Materials and Methods.
Meiotic defects are generally more severe in pelo mutants than that in Hbs1 mutants. Strong loss-of-function (LOF) pelo mutants like pelo PB60 and pelo 1 showed a total blockage of cell division, and mutant spermatocytes did not enter meiosis. Weak pelo PA13 mutants displayed cell division arrest either before or at the first meiotic division (Fig. S2A-F), as previously reported 3 . Despite the observations that the 2N and 4N spermatids seemed to be able to initiate nuclei elongation in the mutants, there was no IC formation in these pelo mutant testes (Fig. S2G,H). Knocking down pelo by two RNAi lines, however, generated weaker phenotypes such that spermatocytes went through meiosis, but ICs showed disorganization (Fig. 4A,G) similar to hypomorphic Hbs1 mutants (Fig. 3H).
We then analyzed potential genetic interactions between Hbs1 and pelo during spermatid individualization. We found that pelo-RNAi in the germline caused defects in the organization of ICs (Fig. 4B,H) and there were only about a number of 9.4 relatively normal-sized ICs in these testes (Fig. 4G). However, IC number declined significantly to an average of 5.3 when one copy of Hbs1 was further removed (Fig. 4G). In the same animals, IC genotype total spermatid number number of different spermatid types  (Fig. 2C insert). "Irregular" means there are abnormal size of nuclei (Fig. 2B insert). "Other irregular" are irregular spermatids that are not included the former 6 kinds, such as one regular nebenkern with 3 nuclei of different size, abnormal size nebenkerns, and so on. The number of "total irregular" is the sum of all the abnormal spermatids, including the spermatids with the abnormal size nuclei (the "irregular") and the spermatids with abnormal ratio of nebenkerns to nuclei (1:2, 1:4 and others irregular www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ disorganization and nuclei scattering phenotype was more severe compared to that in animals with pelo knockdown alone (Fig. 4C,H). Furthermore, there were about 10.6 ICs per testis in Hbs1 transheterozygotes (Hbs1 1 over Hbs1 172 ), and about 12.6 ICs per testis in pelo PB60 heterozygotes (pelo PB60 −/+), but IC number dropped significantly to an average of 4.2 when one copy of pelo was removed, and both the IC and nuclei were severely disorganized (Fig. 4D-H). These observations indicate that pelo and Hbs1 genetically interact with each other to promote spermatid individualization during spermatogenesis.

Hbs1 co-localizes with Pelo in germ cell cytoplasm and they physically interact with each other.
Next, we asked whether Pelo and Hbs1 could form a protein complex in Drosophila. When Pelo tagged with RFP and Hbs1 tagged with GFP at their C terminal regions were co-expressed in Drosophila S2 cells, they predominantly co-localized in the cytoplasm (Fig. 5A-C). Interestingly, unlike Hbs1-GFP, a significant fraction of Pelo-RFP was also localized to the nucleus (Fig. 5A). We also constructed UAS-pelo-FLAG and UAS-Hbs1-myc transgenes, and co-expressed them in the male germline. Both protein products were found to co-localize with each other in the cytoplasm of germ cells (Fig. 5D-F and insets D'-F'). Additionally, Hbs1 and Pelo exhibited strong interaction with each other in a yeast-two-hybrid assay (Fig. 5G). In summary, these data suggest that Hbs1 and Pelo form a protein complex, and along with the evidence for a genetic interaction during spermiogenesis, they likely function in a same protein complex to regulate late spermatogenesis in Drosophila.

The putative Hbs1-binding motif is important for Pelo function in vivo.
To test the functional importance of the interaction between Pelo and Hbs1 during spermatogenesis, we asked whether specifically disrupting the interaction between Pelo and Hbs1 could disrupt their function during spermatogenesis. Structural and biochemical data have shown that the conserved PGF motif in the central domain of Dom34 interacts with a conserved RDF motif in the GTPase domain of Hbs1, and the PGF motif is required for both the interaction of Pelo with Hbs1 and the no-go-decay activity 8,18 . We therefore used a transgene of Pelo carrying Pro210 to Ala (P210A) mutation that is supposed to disrupt the Pelo-Hbs1 complex function possibly by altering the allostatic configuration of the interaction interface 29 , and tested its ability in rescuing spermatogenesis defects caused by the loss of Pelo using a germline-specific driver nosGAL4 (Fig. S3A) 38 . As described earlier, in wild type testes, the nuclei bundles and ICs were detected by DAPI and phalloidin staining, respectively ( Fig. 3A-C). In strong pelo mutants (pelo 1/PB60 ), cell cycle was arrested at early stages of meiosis and there was no elongated spermatid nuclei bundles and therefore no IC formed in these animals (Fig. 6A-A' ,E). As a control, the spermatogenesis defects and the sterility were fully rescued by a wild type transgene Pelo (WT) (Fig. 6B-B' , D,E). However, the defects largely remained unrescued by Pelo (P210A), as ICs could not be formed, although spermatid nuclei seemed to get elongated to some extent ( Fig. 6C-C' ,D,E). Interestingly, Pelo (P210A) was able to fully rescue the GSC loss phenotype in the ovary (data not shown), which is consistent with the observation that Hbs1 is not required for GSC maintenance, as shown below. Pelo (P210A) partially rescued the GSC loss phenotype in the testis while the wild type Pelo could fully rescue it (Fig. S3), indicating the possible participation of other co-factors that bind to Pelo to maintain the GSCs in the testis. These data suggest that the putative Hbs1-binding motif is required for Pelo function during late spermatogenesis, further supporting the notion that Pelo and Hbs1 from a complex to regulate late spermatogenesis.

Hbs1 is not required for ovarian/testicular GSC function.
Previous studies demonstrate that pelo is required for GSC self-renewal and oocyte development during oogenesis 28 . We therefore asked whether Hbs1 is also required for the self-renewal of GSCs. Interestingly, no defect was observed in either GSC maintenance or germline cyst differentiation in Hbs1 mutant females (data not shown). Similar to its role in ovarian GSCs, we found that pelo was also important for the maintenance of testicular GSCs (Fig. S4A-F). However, Hbs1 mutants showed normal GSC number and proper early cyst differentiation (Fig. S4G, J-M), even in aged males (40 days old). Additionally, the gradual GSC loss phenotype caused by pelo mutation was not further enhanced by the loss of Hbs1 (Fig. S4H-J). These data together suggest that Hbs1 is not required for the maintenance of GSCs, implying that the stem cell function of Pelo is independent of Hbs1.

Discussion
Hbs1 regulates spermatogenesis in collaboration with Pelo. Here we demonstrate a novel and specific function of Hbs1 in spermatogenesis in Drosophila. Loss of function of Hbs1 causes defects in meiotic cell division and spermatid differentiation during spermatogenesis, leading to male infertility with no apparent defect in viability, and this revealed a male-specific germline function of Hbs1.
Our results also suggest that Hbs1 may collaborate with Pelo to regulate multiple processes during spermatogenesis in Drosophila, which is supported by several observations. First, Hbs1 physically interacts with Pelo away with disorganized actin cones scattering along the testicular tube (F,H). (I,J) ICs could not be detected in Hbs1 1−/− mutant testis, and spermatid nuclei were randomly scattered along the testicular tubes (indicated by hollow arrowheads). (K) The number of ICs in Hbs1 48−/− and Hbs1 1−/− mutant testis was less than that in wild type control testes. ICs number was restored in the flies with the genomic rescue of Hbs1. Mean ± SEM. n = 17 for control and genomic rescue testes, n = 20 for Hbs1 48 www.nature.com/scientificreports www.nature.com/scientificreports/ in a yeast-two-hybrid assay. Hbs1 also co-localizes with Pelo in the cytoplasm of S2 cells and germline cells. Second, Hbs1 and Pelo genetically interact with each other during spermatid individualization, as phenotypes caused by partial loss of function in one can be enhanced by the loss of function of the other. The phenotypes produced by Hbs1 or pelo mutations could not be rescued by the UAS-pelo and UAS-Hbs1 respectively (data not shown), which excludes the possibility that their function can compensate for each other. Finally, the putative Hbs1-binding motif-mutant version of Pelo cannot substitute the wild type Pelo for spermatogenesis, further supporting the idea that Pelo and Hbs1 form a protein complex to function during spermatogenesis. Our results, together with the previously reported results that yeast as well as mammalian homologues of Hbs1 and Pelo interact with each other 6,14,15,24 , collectively support that the physical interaction between Hbs1 and Pelo are evolutionary conserved from yeast to mammals.
Pelo-Hbs1 complex may regulate translation during spermatogenesis. Hbs1-Pelo complex likely regulates spermatogenesis through translational regulation. Hbs1 belongs to a protein family known to be involved in several steps of translational processes, including translational elongation, release, and decay www.nature.com/scientificreports www.nature.com/scientificreports/ of nonfunctional mRNAs 39,40 . Also, increased expression of Hbs1 in yeast could suppress the growth defect in strains deficient for SSB1/2, which encode molecular chaperones that are associated with nascent polypeptides and function to facilitate transportation of newly synthesized peptides to the cytosol 5 . In addition, yeast Hbs1 www.nature.com/scientificreports www.nature.com/scientificreports/ mutants show defective protein synthesis upon limiting translation initiation 6 . Consistently, in partnership with Dom34/Pelo, Hbs1 has been demonstrated to be involved in no-go and no-stop mRNA decay 41,42 . In this study, expression of a tagged Hbs1 in the male gonad reveals that Hbs1 is predominantly localized in the cytoplasm of www.nature.com/scientificreports www.nature.com/scientificreports/ germline cells, consistent with a role of Hbs1 in translational regulation. Dom34/Pelo and Hbs1 have been shown to have a role in degrading aberrant 18S rRNAs. Nonfunctional rRNAs and mRNAs to be removed by no-go decay are both localized within P bodies, places where mRNA decay is conducted 43 . Therefore, Pelo-Hbs1 could remove both nonfunctional rRNAs and aberrant mRNAs for proper spermatogenesis. In mammalian cells, the Pelo-Hbs1complex along with ABCE1 is important for recycling vacant 80S ribosomes 12 , it is therefore possible that Pelo-Hbs1 are important for ribosome recycling to ensure proper spermatogenesis.
Despite our finding to show that Hbs1 and Pelo function together during late spermatogenesis, our results also indicate Hbs1-independent functions of Pelo in maintaining both male and female GSCs and in early spermatogenesis. Hbs1 is dispensable for GSC maintenance and mutants show relatively weaker phenotype in male meiosis compared to that of Pelo mutants 3,28 . This is similar to the situation in yeast, in which Dom34/Pelo mutants display stronger and more profound phenotypes, such as defective sporulation and decreased polyribosomes, whereas Hbs1 mutants show no obvious phenotype 4,6 . Moreover, Hbs1 seems to play a minor role in no-go decay compared to the critical requirement of Dom34/Pelo 7 . It is possible that there are other redundant G proteins that can functionally substitute Hbs1 or interact with Pelo at different stages together with Hbs1 during spermatogenesis, and Pelo may be a key player in the complex, while Hbs1 may only have a supportive role.
Translational regulation during spermatogenesis: multiple levels of control. Translational regulation plays crucial roles in biological processes in which transcription is largely absent and developmental events are obligatory to be carried out. It is known that transcription is nearly absent at postmeiotic stage during spermatogenesis in Drosophila, while a short period of transcription exists postmeiotically in mammals 1,44,45 . So translational regulation is critical in regulating gene expression during the postmeiotic stage, where germ cells undergo dramatic remodeling while transcription is shut down in spermiogensis [46][47][48] . In those circumstances, many mRNAs are produced and translationally repressed in spermatocytes until the encoded proteins are required for later spermatogenesis, such as don juan and don juan like, both of which are initially transcribed in primary spermatocytes but translationally repressed until spermatid elongation 47,49 . Translational activation is also required for the sequential gene expression in later development in spermatogenesis. Boule, Drosophila orthologue of the vertebrate Y-linked Deleted in Azoospermia-like (DAZl) promotes the efficient translation of twine for meiotic entry 50 . The general translation system, which includes translation initiation, elongation, and termination, also contributes to the expression of different genes at different stages of germline development. eIF4G and eIF4G2, paralogues of the translation initiation complex component eIF4G, are required for meiotic progression and spermatid differentiation in Drosophila by controlling temporal expression of core cell cycle regulators, implicating that alternate translation initiation machinery may be responsible for gene expression and germline development in spermatogenesis [51][52][53] . Our results here suggest another level of translational regulation during spermatogenesis mediated by RNA surveillance factors. The Drosophila Pelo-Hbs1 complex, which has been demonstrated to be required for no-go decay in cultured Drosophila cells and transposon silencing in female germline 13,29 , could be involved in degrading poisonous/defective mRNAs such as transposon transcripts, rRNAs and/or facilitating ribosome recycling that would otherwise lead to defects in meiosis and spermatid differentiation. Conditional depletion of Pelo in postnatal mice causes progressive spermatogonial stem cell loss and sterility, although later spermatogenesis appears normal before germline exhaustion 27 . It remains possible that the perdurance of gene product follow gene knock-out in stem cells may be sufficient to allow later spermatogenesis. Therefore it would be worthwhile to re-evaluate the potential function of Pelo-Hbs1 complex in mammalian gametogenesis by germline depletion of Hbs1.

Materials and Methods
Drosophila strains. All flies were raised at 25 °C and cultured on standard media, unless otherwise stated.
Generation and characterization of new pelo alleles. pelo PB60 was caused by a 107 bp deletion in the second exon, and the predicted protein product is a peptide with 77 amino acids (compared to the full length with 395 amino acids). Flies homozygous for pelo PB60 showed similar GSC loss phenotype during oogenesis and meiosis defect during spermatogenesis to transheterozyotes (pelo PB60 in trans to a deficiency Df(2 L)Exel6024 in which the entire pelo locus is deleted). pelo PB60 homozygous testes also displayed similar severity of meiosis defects to that of pelo 1 homozygous mutants, suggesting that both pelo PB60 and pelo 1 alleles can be considered as strong loss of function or genetic null alleles. pelo PA13 was resulted from a 557 bp insertion left by the original KG6646 P element, and homozygous mutant cysts from testes squash displayed meiosis arrest either before the first or the second meiotic cell division. Based on the severity of defects in female GSC maintenance and male meiosis, we considered pelo PA13 as a hypomorphic allele. www.nature.com/scientificreports www.nature.com/scientificreports/ Generation of transgenic flies and expressing vectors. For the genomic rescue experiment, the genomic region of Hbs1 together with 1X Flag coding region at the C-terminal end was amplified by PCR using KOD polymerase (Takara Bio, Otsu) and cloned into Spe I and Not I sites of pCasPeR 4. The following primers were used: 5′ GGACTAGTCC TGAAGCAAAT CAGAACAGTC 3′ and 5′ TTGCGGCCGCAA CTACTTGTCATC GTCATCCTTGTAATC GCG GAT CTT GGT GAC CAT T 3′. To generate epitope-tagged transgenes, ORFs of both pelo and Hbs1 were first cloned into pENTRY/D-TOPO vector and then into pPWF and pPWM vectors respectively through the Gateway reaction to make UASP-Pelo-FLAG and UASP-Hbs1-Myc (The Drosophila gateway vector collection developed by Terrence Murphy: http://emb.carnegiescience.edu/labs/murphy/Gateway%20 vectors.html). To generate pelo carrying Pro210 to Ala (P210A) mutation, pelo was cloned from Canton-S ovarian cDNA library and ligated into pEasy-blunt Simple vector (TransGen Biotech). Single amino acid substitution was generated by Easy Mutagenesis System (TransGen Biotech) according to the manufacturer's instruction. Full length pelo coding sequence was amplified by PCR with primers 5′-GGTACCATGAAGCTGCTGGGCAAATA-3′ and 5′-TCTAGACTAGTCGCTATCGCTATCTGCC-3′. The PCR fragment was restriction digested with KpnI and XbaI, and subsequently cloned into pUASP. All the plasmid DNA was injected for transformation via standard procedures.

Remobilization
To generate Pelo-RFP and Hbs1-GFP constructs for expression in Drosophila S2 cells, ORFs of Pelo and Hbs1 were cloned into pAWG and pAWR vectors using the Gateway kit. Constructs were transfected into Drosophila S2 cells using the standard calcium phosphate transfection protocol, and the expression pattern of Pelo-RFP and Hbs1-GFP was examined 4 days after the initial transfection.
RNA isolation and qPCR assays. Total RNA from 50-60 testis was extracted by RNAiso Plus reagent(-Takara,9109) and Direct-zol RNA Miniprep kit(Epigenetics, R2050) with DNase treatment. Complementary DNA(cDNA) was synthesized by 5X All-In-One RT MasterMix kit(abm,G490). RT-qPCR was performed in three duplicates using SYBR Select Master Mix(Thermo Fisher,4472908) on ABI PRISM 7500 fast Real-time PCR System(Applied Biosystems). Endogenous Actin5c was used for normalization. Fold changes for mRNA levels were calculated using ΔΔC t method 57 . The primers used are listed in Table S1.
Fertility assay. Fertility assay was performed as previously described 58 . Each male was put together with two w 1118 virgin females in one food vail in twenty duplicates. The flies were raised at 25 °C and separated from each other after 5 days. The offspring were examined within 2 weeks.
Analysis of testis content. Newly enclosed males were collected and manually dissected for testes in Testis Buffer (183 mM KCl, 47 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA, 1 mM PMSA). Testes were subsequently transferred onto a coverslip containing a 2-ul drop of Testis buffer. Testes were then opened using forceps, squashed and examined by phase contrast using a Zeiss Axio microscope.
Yeast-two hybrid. ORF of Pelota was cloned into NdeI and EcoRI sites of pGADT7 AD Vector. The primers used were 5′ CCCATATGAAGCTGCTGGGCAAATACGT 3′, and 5′ CCGAATTCTAGTCGCTATCGCTATCT 3′. ORF of Hbs1 was cloned into Nde I and BamH I sites of pGBKT7 DNA-BD vector, and primers were 5′ CCCATATG TCGCGGCACAGGATAGT 3′ and 5′ CCGGATCCTAGCGGATCTTGGTGACCATT 3′. The yeast two hybrid assay was based on the MATCHMAKER kit (Clontech). pGBKT7-p53 with the Gal4 DNA-BD fused with murine p53 and pGADT7-T with the Gal4 AD fused with SV40 large T-antigen were used as positive control.