Three RNA Binding Proteins Form a Complex to Promote Differentiation of Germline Stem Cell Lineage in Drosophila

In regenerative tissues, one of the strategies to protect stem cells from genetic aberrations, potentially caused by frequent cell division, is to transiently expand the stem cell daughters before further differentiation. However, failure to exit the transit amplification may lead to overgrowth, and the molecular mechanism governing this regulation remains vague. In a Drosophila mutagenesis screen for factors involved in the regulation of germline stem cell (GSC) lineage, we isolated a mutation in the gene CG32364, which encodes a putative RNA-binding protein (RBP) and is designated as tumorous testis (tut). In tut mutant, spermatogonia fail to differentiate and over-amplify, a phenotype similar to that in mei-P26 mutant. Mei-P26 is a TRIM-NHL tumor suppressor homolog required for the differentiation of GSC lineage. We found that Tut binds preferentially a long isoform of mei-P26 3′UTR, and is essential for the translational repression of mei-P26 reporter. Bam and Bgcn are both RBPs that have also been shown to repress mei-P26 expression. Our genetic analyses indicate that tut, bam, or bgcn is required to repress mei-P26 and to promote the differentiation of GSCs. Biochemically, we demonstrate that Tut, Bam, and Bgcn can form a physical complex in which Bam holds Tut on its N-terminus and Bgcn on its C-terminus. Our in vivo and in vitro evidence illustrate that Tut acts with Bam, Bgcn to accurately coordinate proliferation and differentiation in Drosophila germline stem cell lineage.


Introduction
Adult stem cells divide to replenish differentiated, dead, or damaged cells in regenerative tissues. To produce sufficient number of differentiated progeny for tissue homeostasis and to avoid the accumulation of oncogenic mutations derived from frequent cell divisions, stem cell daughters undergo multiple rounds of transit-amplifying (TA) divisions prior to terminal differentiation [1,2,3,4]. However, failure to stop TA divisions and enter programmed differentiation may contribute to tumorigenesis in adult stem cell lineages [2,5,6,7].
Drosophila spermatogenesis is a highly stereotyped and accessible system to study the control mechanisms of accurate TA divisions in adult stem cell lineage. At the apical tip of testis, germline stem cells (GSCs) and somatic cyst stem cells form a rosette surrounding the hub ( Figure 1A). GSC divides asymmetrically to generate a daughter cell adjacent to the hub remaining as a GSC while the other one away from the hub differentiating as a gonialblast (GB). As the founder giving rise to a clonal production of gametes, GB in turn undergoes four rounds of TA divisions to form a cluster of 16 interconnected spermatogonial cells which develop in synchrony thereafter. After the four TA divisions, spermatogonia switch to the meiotic/spermatocyte program and increase 25-fold in cell size [8,9,10]. The dramatic differences in morphology, the availability of molecular markers to distinguish germ cells at different stages, and especially the accurate number of TA divisions make Drosophila spermatogonial proliferation a perfect model to look for deviations upon genetic manipulations.
In the past decades, many intrinsic and extrinsic factors regulating TA divisions have been found in Drosophila spermatogenesis [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. Among them, bag of marbles (bam) is at the center of the picture not only because it was the first player identified in this process, also because in germ cells Bam protein accumulation signals the stop of TA division and/or the start of further differentiation [14,26,27]. Ectopic expression of Bam protein in GSCs leads to the premature differentiation of all stem cells [28,29,30].
Benign gonial cell neoplasm (Bgcn) is an 'intimate' partner of Bam, given that they have the same mutant phenotype (i.e., spermatogonial over-amplification) [14] and they are present in the same protein complex to confer translational repression in both male and female germ cells [15,31,32,33]. In cultured Drosophila S2 cells, Bam and Bgcn repress the expression of a reporter coupled with the 39 untranslated region (39UTR) of DE-Cadherin [33]. In female germline, Bam-Bgcn complex antagonizes Nanos (Nos) expression via nos 39UTR [34]. While in male germ cells, this complex binds mei-P26 39UTR directly to repress Mei-P26, whose initial expression in early TA cells is required for Bam accumulation [15]. Thus, a negative feedback loop is formed between Mei-P26 and Bam to ensure proper accumulation of Bam and accurate TA divisions. However, overexpression of Mei-P26 in late TA cells did not resemble bam or bgcn mutant phenotype [15]. Identification of more genes involved in this process will unravel the regulatory network governing the switch from TA division to meiotic differentiation.
From a mutagenesis screen combined with germline clonal analysis, we isolated a mutant showing spermatogonial overproliferation, a phenotype similar to that of mei-P26, bam, or bgcn. This mutation disrupts CG32364 which we named tumorous testis (tut). We found by genetic and biochemical methods that Tut, Bam, and Bgcn act in a complex to accurately coordinate TA division and differentiation of germline stem cell daughters.

Identification of tut as an Intrinsic Factor Restricting Transit-amplification
To search for more factors regulating the proliferation and/or differentiation of fly germline stem cell lineage, we performed a large scale EMS screen, and obtained a male sterile mutant line exhibiting germ cell overgrowth. We mapped the gene responsible for this phenotype to CG32364, and designated it as tumorous testis (tut). tut encodes a protein containing a putative RNA recognition motif (RRM). The EMS-induced point mutation in tut 1 generates a premature stop codon ( Figure 1B and S2D). The tut mRNA transcribed from tut 1 genome is much lower than that from wild-type ( Figure S1B). We also generated a null allele, tut 3 ( Figure 1B), by homologous recombination-based gene targeting (Materials and Methods). tut 1 , tut 3 and tut 1/3 mutant testes all exhibited the same phenotypes including the failure to exit TA division and the spatial/temporal pattern of molecular markers (See below). Compared to the wild-type ( Figure 1C), tut mutant testis was filled with early germ cells brightly stained by the DNA dye ( Figure 1D), but lacked late germ cells such as spermatocytes (compare the spermatocyte marker Hrb98DE-GFP [35,36] in Figure S1C and D) or spermatids (compare Dj-GFP [55,36] in Figure S1E and F). tut mutant testis could be completely rescued by the expression of tut cDNA in germ cells ( Figure 1E), but not in somatic cells ( Figure 1F), indicating that tut functions in germ cells.
Consistently, the over-proliferating germ cells were observed only when tut was knocked down in germ cells ( Figure S1G-H), but not in somatic cells ( Figure S1I).
To determine the cell stage at which tut mutant germ cells were arrested, we chose several well-characterized molecular markers for early germ cells including GSC, GB, and spermatogonia [14,37,38]. The over-proliferating germ cells in tut mutant clones possessed all the characteristics of TA spermatogonia, such as branched fusome passing through ring canal ( Figure 1G), synchronized cell division ( Figure 1H), Bam protein expression ( Figure 1I), and bam transcription revealed by bamP-GFP ( Figure  S1J-K). Additionally, no expansion of GSC and GB was detected in the tut mutant testes by the commonly used markers for these cell types ( Figure S1L-O). Thus, tut mutant germ cells arrested at spermatogonial TA stage and over-proliferated. Taken together, we conclude that tut is intrinsically required in germ cells to ensure proper transit amplification of spermatogonia.
Tut Is Required for the Translational Repression of mei-P26 via Its 39UTR tut mutant phenotype is similar to bam, bgcn, or mei-P26 mutant. Bam-Bgcn complex has been shown to regulate mei-P26 expression by binding to its 39UTR [15]. Since Tut protein contains a predicted RRM, we wonder if Tut binds mei-P26 39UTR directly. We found a longer isoform of mei-P26 39UTR in tut, bam, and bgcn mutant testes ( Figure S2A-B), which also existed at low abundance in wild-type testes ( Figure S2C). To examine the interaction between Tut and mei-P26 39UTR, we performed a series of RNA immunoprecipitations and quantified mei-P26 39UTR by realtime PCR. Interestingly, Tut protein binds the longer isoform of mei-P26 39UTR more efficiently than the shorter one that has been reported to interact with Bam ( Figure 2A) [15].
Yeast 3-hybrid is an easy and efficient assay to detect the physical interaction between protein and RNA [39]. Consistently, Tut binds the long isoform of mei-P26 39UTR at high stringent conditions and binds also the short isoform of mei-P26 39UTR at low stringent conditions ( Figure S2E-F). However, we could not detect any interaction between Bam and long mei-P26 39UTR ( Figure S2E-F). Deletion of the RRM domain abolished the association between Tut and mei-P26 39UTR, further supporting that Tut functions as an RBP and binds to mei-P26 39UTR ( Figure S2F).
To test whether Mei-P26 protein level is changed in tut mutant germ cells, we generated an antibody against Mei-P26. We confirmed the specificity of this antibody by immunostaining mei-P26 mutant testis ( Figure S3H). Mei-P26 protein was detectable at low level in wild-type spermatogonia ( Figure S3G) and was upregulated in tut mutant ( Figure S3I).
To test whether tut is required via mei-P26 39UTR for Bam-Bgcn complex-mediated repression of mei-P26, we generated a reporter containing GFP coding sequence and mei-P26 39UTR region (2 kb downstream of the stop codon), from which both long and short isoforms were detected when the reporter was expressed (detected by 39RACE specific for the reporter). The reporter expression driven by bam-Gal4 was repressed in ,80% of Bampositive cysts in the presence of mei-P26 39UTR ( Figure 2B, n = 60), consistent with the reported pattern using a shorter mei-P26 39UTR in a similar construct [15]. As expected, in the absence of Bam, the GFP reporter was de-repressed in spermatogonial TA cells ( Figure 2C and S3B). In tut mutant testis, though Bam and Bgcn were expressed, the GFP reporter was nonetheless de-repressed ( Figure 2D and Figure S3A,D), indicating that Tut is essential for the translational repression mediated by mei-P26

Author Summary
In regenerative tissues, the successive differentiation of stem cell lineage is well controlled and coordinated with proper cell proliferation at each differentiation stage. Disruption of the control mechanism can lead to tumor growth or tissue degeneration. The germline stem cell lineage of Drosophila spermatogenesis provides an ideal research model to unravel the genetic network coordinating proliferation and differentiation. In a genetic screen, we identified a male-sterile mutant whose germ cells are under-differentiated and overproliferating. The responsible gene encodes an RNA-binding protein whose target belongs to a tumor suppressor family. We demonstrate that this and two other RNA-binding proteins form a physical and functional unit to ensure the proper differentiation and accurate proliferation of germline stem cell lineage. 39UTR. Similarly, the GFP reporter was also de-repressed in bgcn mutant testis ( Figure S3C,J), suggesting that Tut, Bam, and Bgcn may act together on mei-P26 regulation as well as in the development of GSC lineage.

Genetic and Physical Interactions between Tut and Bam in Regulating Spermatogonial TA Division
To examine the expression pattern of Tut during spermatogenesis, we first tried to raise an antibody against Tut but failed after several attempts using different strategies. We then sought to add a tag into tut locus by genomic engineering, but the extra sequences introduced into tut locus affected tut function, generating the weak allele tut 4 instead ( Figure 1B and Figure 3G). Then we tagged a genomic fragment containing tut regulatory sequences ( Figure S4A) that was sufficient to rescue tut mutant phenotype ( Figure 3A-B). The expression pattern of this genomic construct resembled that of Bam protein ( Figure 3C-D), though very weak expression was detected in GSCs ( Figure 3C9). We next tested whether or not tut functions in the same pathway as that of bam by genetic assays. tut 4 is a weak allele and its heterozygous testes were indistinguishable from the wild-type ( Figure 3E). bam D86 is a null allele whose heterozygous testes showed ,60% 'tumor' rate ( Figure 3F and Figure S4D). Comparing the severity of spermatogonial accumulation, we found that tut 4 homozygotes contained mostly spermatocytes mixed with a few spermatogonial tumors ( Figure 3G); whereas disrupting a copy of bam in tut 4 background blocked the germline development at spermatogonial stage (Figure 3H-J; 100%, n.50). The genetic interaction between tut and bam was confirmed by different alleles of both genes ( Figure S4B-D).
Expression of either tut or bam did not rescue each other's mutant phenotype ( Figure S4E-F). This prompted us to ask whether Tut and Bam act in the same protein complex. We first tried yeast 2-hybrid assay and found that Tut and Bam could form a complex ( Figure S4G). Then we co-expressed Tut and Bam in cultured Drosophila S2 cells followed by co-immunoprecipitation assay. Myc-Tut and Flag-Bam co-immunoprecipitated with each other in both ways ( Figure 3K-L). As in the testis, Tut and Bam protein expressed in S2 cells were localized in the cytoplasm ( Figure S4H). Taken together, it is likely that Tut and Bam function in the same protein complex to regulate spermatogonial TA division.

Tut, Bam, and Bgcn Are Present in the Same Protein Complex
Bgcn and Bam have been demonstrated to form a complex in both female and male germline [15,32] ( Figure S5D). Although genetic interaction between bam and bgcn in germ cell development has been revealed in fly females [40], their interaction in male germline was not known due to the lack of weak allele of bam or bgcn. We generated a weak allele of bgcn named bgcn 2 . Like tut 4 ( Figure 3G), bgcn 2 testis exhibited mild over-proliferation phenotype ( Figure S5E). Removing one copy of bam dramatically enhanced bgcn 2 mutant phenotype ( Figure S5F).
Given the close relationship between Tut and Bam as well as Bam and Bgcn, plus the concurrence of the three proteins in spermatogonia ( Figure 3D, 4A, and S5H), it is conceivable that three of them form a functional unit in spermatogonial TA cells. We then tested the genetic relationship between tut and bgcn. The double heterozygotes of tut 4 and bgcn (20093 = null; QS2 = Cterminal truncation) did not show any spermatogonial tumor growth ( Figure 4B-C; 100%, n.50). Further disrupting the other copy of tut made the majority of germ cells to keep dividing but unable to differentiate beyond spermatogonial stage ( Figure 4D-E and S5G; 100%, n.50; also see Figure 3G for tut 4 homozygous phenotype). Consistently, removing one copy of bgcn dramatically enhanced tut knockdown phenotype ( Figure S5A-C). These data suggest that tut functions with bgcn in controlling spermatogonial TA proliferation.
To determine if Tut, Bam, and Bgcn are present in the same protein complex, we carried out the two-step co-immunoprecipitation assay [34,41,42] by co-expressing TAP-Tut, Flag-Bam, and Flag-Myc-Bgcn in S2 cells. After two rounds of successive immunoprecipitations, Bam and Bgcn were still present in Tut complex ( Figure 4F), suggesting that these three proteins form a trimeric complex rather than exclusive heterodimers such as Bam/ Tut, Bam/Bgcn, or Tut/Bgcn. To examine the existence of this complex in vivo, we used the extracts of fly testes to do the coimmunoprecipitation and again, demonstrate that these three factors are physically associated with each other ( Figure 4G).
Although Tut and Bgcn formed a complex in the presence of Bam ( Figure 5A-B, left panels), we failed to detect physical interaction between Tut and Bgcn in the absence of Bam in coimmunoprecipitation ( Figure 5A-B, right panels) or in yeast 2hybrid assays ( Figure S6A). These observations raised the possibility that Bam brings Tut and Bgcn together to form a complex. Bgcn has been reported to interact with Bam C-terminus [32]. To map which region of Bam associates with Tut, we expressed different fragments of Bam in yeast and tested the interacting activity by yeast 2-hybrid assay. The Fragment containing N-terminal 100 amino acids of Bam was both necessary and sufficient to bind Tut ( Figure S6B). We confirmed this interaction by co-immunoprecipitation using S2 cells ( Figure 5C-F). These data suggest that Bam recruits Tut and Bgcn proteins to form a complex via its N-terminus and C-terminus, respectively.
Given that all three proteins are RNA binding proteins (This study and [15]), we wonder if the formation of Tut-Bam-Bgcn complex is RNA-dependent. We found that the formation of this complex in S2 cells was not disrupted by the treatment of RNaseA ( Figure S6C). Thus, Tut, Bam, and Bgcn form a protein complex in an RNA-independent manner.

Tut Is Required for Bam to Drive Germline Stem Cell Differentiation
Bam has been shown to promote differentiation when overexpressed in GSCs [29,30] or in TA spermatogonia [26]. We wondered if Tut is required for Bam pro-differentiation function, and compared the consequences of Bam over-expression in GSCs (by the combination of nos-Gal4 and UASp-bam-GFP) in tut mutant v.s. wild-type background. As expected, ectopic expression and tj-Gal4/UAS-GFP-tut; tut 1/3 (F) stained with DNA dye DAPI. (G-I) Immunofluorescence images of the apical part of the testes containing tut mutant clones. tut 3 clones were marked by the absence of LacZ (G9, H) or GFP (I). (G) A clone of over-proliferating germ cells with branched fusome (1B1) running through ring canal (Zip-GFP). G0 shows the high magnification view of the boxed region in G9. (H-H9) A testis stained for clone marker LacZ and S phase marker BrdU. Red arrowhead points to the tut 3 mutant clone with all cells entering S phase in synchrony. (I-I9) A testis containing tut 3 mutant clones was stained for Bam, GFP, and DNA (DAPI). Bam was expressed both in wild-type spermatogonia (yellow arrowhead) and tut 3 mutant cells (red arrowhead). Scale bars: 200 mm (C-F) and 25 mm (G-I). See also Figure S1. doi:10.1371/journal.pgen.1004797.g001 Figure 2. Tut is required to repress mei-P26 expression via mei-P26 39UTR. (A) Physical interaction of Tut protein and mei-P26 39UTR detected by RNA immunoprecipitation (RIP). TAP-Tut or TAP-GFP and mei-P26 39UTR were co-expressed in S2 cells. IgG beads were used to enrich TAP-Tut or TAP-GFP followed by TEV digestion to release Tut or GFP and bound RNA. RNA was extracted and reverse transcribed. The quantity of Actin5C mRNA, of Bam in GSCs eliminated all germ cells ( Figure 6A-B) [30]. However, in tut mutant background, Bam over-expression in GSCs just resembled tut mutant phenotype ( Figure 6C-D), suggesting that bam requires tut to promote GSC differentiation.
In wild-type testis, GSCs are the first tier of germ cells surrounding the hub where Bam is repressed [8,9]. When Bam-GFP was over-expressed in tut mutant background, it was present in the germ cells next to the hub ( Figure 6D). To determine whether this tier of Bam-GFP positive germ cells were indeed GSCs, we immunostained these cells for fusome marker alphaspectrin and GSC marker Stat92E. Fusome morphology changes from spherical in GSCs/GBs to branched in spermatogonia/ spermatocytes ( Figure 1A). We found that these hub-adjacent Bam-GFP-positive cells contained dot-shaped fusome ( Figure 6E, yellow arrowhead) and expressed Stat92E ( Figure 6F). Moreover, these Bam-GFP positive GSC-like cells maintained cell division, which was revealed by BrdU incorporation (Figure 6G), and such dividing activity was detected even 10 days after eclosure ( Figure 6H). Thus, we demonstrate that Tut is required for Bam to drive GSC differentiation. Similarly, Bam also failed to drive GSC differentiation in bgcn mutant background ( Figure S7), further supporting that Tut, Bam, and Bgcn form a functional unit to promote differentiation in GSC lineage.

Discussion
In the development of regenerative tissues, the successive differentiation of stem cell lineage is well controlled and coordinated with proper cell proliferation at each differentiation stage. This is clearly exemplified in Drosophila spermatogenesis which provides a nice research system for us to address related questions. RBPs play major roles in germline development but the molecular mechanisms how they exert their function remain largely unclear.

Tut, Bam, and Bgcn Act in the Same Complex to Regulate Spermatogonial TA Division in Drosophila
From a large-scale genetic screen, we identified tut as an intrinsic factor restricting spermatogonial proliferation. Using cultured S2 cells or fly testes, we demonstrated that Tut, Bam, and Bgcn formed a protein complex ( Figure 4F-G). Mutations disrupting any of the three components block differentiation and lead to severe spermatogonial over-proliferation ( Figure 1D and [14]). Furthermore, either Tut or Bgcn is required for Bam to drive GSC differentiation in testis ( Figure 6 and Figure S7), suggesting that they could function as a complex in GSCs as well. Given the genetic and physical associations observed, we conclude that Tut, Bam, and Bgcn form a complex in spermatogonia to ensure precise TA divisions.
TA divisions are very sensitive to bam dosage and Bam protein level is under intricate control [26,43]. Why is Bam protein level so critical for the timely transition from TA divisions to meiotic differentiation? In the Tut-Bam-Bgcn complex, Tut and Bgcn do not interact with each other unless Bam protein is present ( Figure 5 and S6A-B). Bam acts as a nexus to bring Tut and Bgcn together, in the manner that the N-terminus of Bam interacts with Tut and the C-terminus with Bgcn. Bam expression is dynamic, first detected in 2,4-cell of TA spermatogonia and peaked at 8cell stage. Thus, Bam level determines the quantity of Tut-Bam-Bgcn complex. This may explain why Bam protein level serves as an 'index' for the spermatogonia to respond as when to stop TA divisions and start differentiation.
Regarding how mei-P26, the downstream target of the complex, is regulated, there is a discrepancy between the study by Insco et al. [15] and our current one. They found that Bam bound to the segments contained in the 549 nt of mei-P26 39UTR, which we designated as the short form ( Figure S2A-C). However, we could not detect any interaction between Bam and mei-P26 39UTR long form by Y3H ( Figure S2E-F). This could be due to the different assays we employed. It has been demonstrated that correct RNA folding is essential for protein-RNA interaction in Y3H [39]. Nonetheless, we both demonstrated that Bam and Bgcn are present in a complex required for the repression of mei-P26.

Different Expression Patterns of Tut, Bam, Bgcn, and Mei-P26 Are Associated with Different States of Germline Differentiation
Based on the previous and our current findings of the Tut-Bam-Bgcn complex and its target Mei-P26, we propose a model describing how the dynamic expression patterns of these proteins are associated with germline differentiation ( Figure S8). Under normal conditions, Bgcn is present in all stages of spermatogenic cells whereas the other 3 proteins are not ( Figure S8A). Tut is very weakly expressed in GSC, GB, or early TA cells, thus Mei-P26 cannot be completely repressed. Because Mei-P26 promotes Bam expression [15] and Bam is required for the full expression of Tut ( Figure S3E-F), Tut accumulates and peaks in the late TA cells in which Tut, Bam, and Bgcn can form a complex bound on the 39UTR of mei-P26 to repress its expression ( Figure S8A). The shorter version of mei-P26 39UTR is much more predominant than the longer one under this circumstance ( Figure S2B-C). At the end of TA stage when Tut is degraded and Bam decreases, Mei-P26 gets derepressed and the germ cells enter the meiotic cycle.
In the mutants of tut, bam, or bgcn, the protein complex cannot be formed on the 39UTR of mei-P26 to repress its expression, and mei-P26 39UTR exists as a longer isoform ( Figure S8B). However, derepression of Mei-P26 in TA cells is not sufficient to block the differentiation towards spermatocytes because overexpression of Mei-P26 in late TA cells did not phenocopy tut, bam, or bgcn mutant (this study and [15]). Even if Tut-Bam-Bgcn complex is the 'master switch' for TA cell transition to meiosis, there should be more downstream targets than just mei-P26 mRNA.
Bam is normally not expressed in GSCs whereas Tut and Bgcn are present, but ectopic expression of Bam in GSCs leads to GSC premature differentiation and eventually GSC loss ( [29,30] and Figure 6A-B), indicating that Bam could exert its 'pro-differentiation' function in GSCs. However, in the absence of Tut or Bgcn, ectopic Bam cannot drive GSC to differentiate, further suggesting the coordinated action of the three proteins in germ cell differentiation ( Figure S8C). short (M3US) or long (M3UL) isoform of mei-P26 39UTR was determined by real-time PCR. The Y axis represents the ratio of RIP/Input, which was normalized to 1 for TAP-GFP. Error bar indicates SD. ***, p,0.0001 in t test. n.s., not significant. (B-B0) Genotype: bam-Gal4/Y; UAS-GFP-mei-P26 39UTR (2k)/+. A genomic region of 2 kb in length downstream of mei-P26 stop codon was selected to cover both short and long isoforms (see

Tut May Be a Male-Specific Component of the Translational Repression Complex in Germline
RBPs play central roles in germline development across species [44,45]. Bam-Bgcn complex may act as part of the translational machinery but execute different functions in female and male germline by binding to different RNA targets with different partners. In female germline, Sxl binds to nos 39UTR directly and associates with Bam-Bgcn complex to repress nos translation [46,47]. But Sxl is not expressed in testis [46]. Although tut mutant spermatogonia fail to differentiate and over-proliferate, tut 3 females are fully fertile and their ovarioles were indistinguishable from the wildtype by immunostaining (Figure S1P-Q). Thus, Tut may represent a male-specific partner of Bam-Bgcn complex.
It is plausible that Tut-Bam-Bgcn complex functions as part of the translational repression machinery to inhibit target mRNA translation in fly male germline. First, Tut also binds to mei-P26 39UTR, though the preferred isoform of binding is different from Bam or Bgcn (Figure 2A, S2E-F). Secondly, Tut, Bam, and Bgcn are all expressed in spermatogonia ( Figure S5H), and their mutant testes exhibit the same phenotype (this study and [14]). Thirdly, Tut, Bam, and Bgcn form a protein complex in Drosophila testes and S2 cells ( Figure 4F-G). Fourthly, Bam protein binds the translation initiation factor eIF4A directly, and removing one copy of eIF4A partially suppresses the phenotype of bam mutants in both male and female systems [26,33]. We speculate that Tut-Bam-Bgcn complex binds mei-P26 39UTR to repress the translation of mei-P26 mRNA in TA cells. Noticeably, Dnd1, the putative homolog of Tut in Zebrafish, has been reported to protect mRNA from miRNA-mediated repression by binding to the 39UTR in germline [48]. Furthermore, Tut-Bam-Bgcn complex is likely to target additional mRNAs in spermatogonia.
Characterizing more of these mRNAs will further elucidate the molecular mechanisms how Tut-Bam-Bgcn complex promotes differentiation in GSC lineage.
Fly stocks were maintained under standard culture conditions and all flies were dissected 0-2 days after eclosure unless otherwise indicated. For RNAi experiments, flies were cultured at 25uC for 6 days and transferred to 29uC for another 6 days before dissection.  For germline clonal analysis, flies were heat-shocked in 37uC water bath for 1 hour at late pupal stage and dissected 4-5 days after clone induction. tut 4 homozygous and bam heterozygous phenotype varies at different temperatures, age, or nutritions. For tut 4 , bgcn 2 , and bam/+ related experiments, flies were cultured at 24uC, fed with fresh yeast daily, and dissected within 12 hours after eclosure.

EMS Mutagenesis
Isogenized flies bearing FRT and UAS-Flp were fed with EMS overnight. The progeny of EMS-treated flies were crossed to flies carrying FRT-GFP and nos-Gal4. Their male offspring were dissected and stained with DAPI [19,56,57,58]. tut 1 was one of the mutants with germline over-growth phenotype, and was mapped by deficiency screen and candidate gene sequencing.
tut Alleles tut 1 bears a point mutation (3L: G8203128A) that creates a new stop codon. Wild-type Tut protein is 230 amino acids in length, and tut 1 is expected to produce only the N-terminal 174 amino acids. tut 3 and tut 4 alleles were generated by genomic engineering [59] using 3.1 kb upstream from start codon and 3.1 kb downstream from stop codon flanking sequences. Genomic coding region of tut was replaced with attP and loxP, generating the null allele tut 3 . tut genomic region was introduced back into the tut locus of in tut 3 background via attP-attB incorporation. However, after this manipulation, 91 bp (attR and vector sequence) sequences were inserted upstream of tut start codon and 68 bp (loxP and vector sequence) downstream of stop codon, generating the weak allele tut 4 .

tut-Related Transgenic Flies
The w 1118 and p51D stocks were chosen as the hosts for Pelement and attB-attP mediated transgenesis, respectively [60,61]. tutP-tut-GFP contains 1.5 kb sequence upstream of tut start codon, tut genomic region (introns included and stop codon removed) tagged with GFP at its C-terminus, and 2.2 kb sequence downstream of tut stop codon. tutP-tut-TAP and tutP-GFP contain the same regulatory sequences as tutP-tut-GFP [41,62,63]. UAS-GFP-mei-P26 39UTR was generated by cloning GFP coding sequence and mei-P26 39UTR (2 kb downstream of stop codon) to replace the SV40 element in pUAST vector.

Yeast 2-Hybrid Assay
Yeasts were cultured on SD/-Ade/-His/-Leu/-Trp medium supplemented with Aureobasidin A and X-a-Gal (QDO/X/A) to test protein-protein interactions or on SD/-Leu/-Trp medium (DDO) to confirm the transformation of testing plasmid DNA.

Yeast 3-Hybrid Assay
Yeasts were cultured on SD/-His/-Leu/-Ura medium supplemented with X-b-Gal (TDO/X) to test protein-RNA interactions or on SD/-Leu/-Ura medium (DDO) to confirm the transformation of testing plasmid DNA [39].

39RACE (Rapid Amplification of cDNA Ends)
Total RNA was extracted using TRIzol (Ambion, 15596-018) from 10 pairs of w 1118 testes and 39RACE was performed by following the manufacturer's instructions (TAKARA, 6106). Outer primer  and inner primer (59-GTTCTAGTCCTGAACACCCT-39) were used to amplify mei-P26 39UTR. PCR products were loaded into 2% argarose gel and electrophoresed at 100 V for 1.5 h on ice. control, respectively. M3US or M3UL symbolizes the short or the long isoform of mei-P26 39UTR, respectively. TDR is the construct described in D. For higher stringency assay, yeasts were cultured on SD/-His/-Leu/-Ura medium supplemented with Xb-Gal (TDO/X). For lower stringency assay, yeasts were cultured on SD/-Leu/-Ura medium, transferred to filter paper, permeabilized and soaked in solution containing X-b-Gal (DDO/X). (TIF)