Bällchen is required for self-renewal of germline stem cells in Drosophila melanogaster

Tissue specific stem cell populations are actively maintained as a source for distinct cell types required for growth, development and regeneration in the adult organism (Kohlmaier and Edgar, 2008; Weissman et al., 2001). To suppress their differentiation, stem cells require extracellular signalling cues which derive from a distinct cellular microenvironment called the stem cell niche (Losick et al., 2011). In the female ovary of Drosophila melanogaster, germline stem cells (GSCs) generate asymmetric cell fates with each cell division (Li and Xie, 2005; Morrison and Spradling, 2008). One daughter cell maintains the stem cell properties of the parent GSC, whereas the other daughter cell differentiates into a cystoblast. The cystoblast then develops into a germline cyst that includes a single cell that later becomes the oocyte. The germline cyst is surrounded by follicular epithelial cells, which are derived from follicle stem cells (FSCs). Similarly, the GSCs of the male germline also self-renew and differentiate into gonialblasts, which then give rise to sperm.

The stem cell niche of the ovary contains cap cells that secrete the decapentaplegic (DPP) ligand that activates the BMP pathway in GSCs. The resultant BMP signalling leads to phosphorylation of the transcription factor Mothers against Dpp (MAD), which represses the expression of the differentiation factor bag of marbles (BAM) (Chen and McKearin, 2003; Song et al., 2004). In the male germline, the Jak/Stat signalling pathway plays a major role for GSC self-renewal (Kiger et al., 2001). However, its cell autonomous function in GSCs is restricted to orienting the division plane of GSCs relative to the niche cells (Leatherman and Dinardo, 2010). The function of Jak/Stat signalling for GSC self-renewal primarily derives from activating the secretion of the BMP ligand glass bottom boat (GBB) in male somatic stem cells, which in turn activates the BMP pathway in GSCs (Kawase et al., 2004).

The progenitors of GSCs, which are called primordial germ cells (PGCs), also utilise BMP signalling through DPP to repress BAM in the larval ovary (Gilboa and Lehmann, 2004). Maintenance of FSCs in the ovary not only requires DPP signalling but also hedgehog and wingless pathway activity (for a review, see Kirilly and Xie, 2007). Thus, BMP signalling contributes both directly and indirectly to stem cell maintenance in various stem cell populations, but it acts in conjunction with other different external signals to suppress stem cell differentiation. However, common stem cell autonomous components that maintain stem cell properties and prevent differentiation have remained unknown.

Here we report that the protein kinase encoded by the Drosophila melanogaster gene bällchen (ball, also known as Nhk-1) has a common role in maintaining self-renewal of stem cell populations. ball protein (BALL) is orthologous to the Vaccinia-related Kinases (VRKs) of vertebrates and most closely related to VRK-1 (Aihara et al., 2004). VRKs are found in all metazoan species ranging from worms to humans. VRKs of different species were found to phosphorylate the Barrier-to-Autointegration Factor (BAF) (Bengtsson and Wilson, 2006; Gorjánácz et al., 2007; Lancaster et al., 2007; Nichols et al., 2006), which is involved in the assembly of the nuclear lamina in Caenorhabditis elegans (Gorjánácz et al., 2007) and the organisation of chromatin in the nucleus (Margalit et al., 2007). Moreover, hypomorphic mutations in Drosophila ball cause aberrant chromatin organisation in the oocyte nucleus and an altered pattern of histone modifications (Ivanovska et al., 2005). The analysis of ball null mutants revealed defects in proliferating tissues of the larvae, including the brain and imaginal discs (Cullen et al., 2005).

We have used systemic null mutants and mosaic analyses to characterise the function of BALL in both progenitor cells and niche-dependent stem cells. We found that BALL is required to maintain self-renewal of stem cells, which suggests that the previously described defects in proliferating tissues of ball mutant animals is caused by the premature or unscheduled differentiation of progenitor cells rather than a general function of BALL for cellular proliferation.

In order to assess the function of ball in proliferating tissue and in stem cells, we generated a null allele of ball (ball²; supplementary material Fig. S1). ball² homozygotes (hereafter referred to as ball² mutants) die during the pupal stage, confirming previous results described for other ball null alleles (Cullen et al., 2005). ball² mutants already show severe morphological defects by the end of larval development which include considerably reduced gonads in both sexes, the absence of imaginal discs and severely diminished larval brains. This mutant phenotype is solely due to the ball² mutation, as a genomic ball transgene rescued the mutants to produce viable and fertile adults (supplementary material Fig. S1).

We explored the function of BALL in developing male and female gonads. Growth of larval testes relies on asymmetric, niche-supported divisions of GSCs, whereas growth of larval ovaries relies on symmetric divisions of primordial germ cells (PGCs) (Dansereau and Lasko, 2008). Both larval cell types are derived from embryonic PGCs, which reside in the primitive gonads of embryos. In order to establish whether the initial number of PGCs was affected in ball² mutant embryos, we counted their number in embryonic gonads. ball² mutants contained on average 10.2 PGCs (SD = 1.8, n = 25 gonads) which was not significantly different from the number of PGCs observed in wild type control embryos (11.0 PGCs; SD = 1.4, n = 27 gonads). Therefore, the smaller size of larval gonads in ball² mutants is not caused by a reduced number of embryonic PGCs.

We next asked when the size reduction of the male gonads occurs during larval development. In early larval testes, about 8–12 PGCs adopt GSC fate after their recruitment to the somatic hub cells (Fig. 1A). GSCs then divide and give rise to self-renewed GSCs and gonialblasts, respectively. The gonialblasts undergo four incomplete cell divisions and form a 16-cell germline cyst. GSCs and differentiating cysts can be distinguished by their position in the developing testis and by the expression of the adducin-related protein, HTS (Fig. 1A). During mid larval development of wild type testis (48 h after larval hatching, ALH), HTS localises to a spectrosome in GSCs and a branched fusome in cysts, respectively (Fig. 1B,C). In ball² mutant testes 48 h ALH, differentiating germline cysts were formed, but the number of GSCs was reduced (Fig. 1D,E). By the end of larval development (96 h ALH), GSCs were completely depleted from ball² mutant testes (n = 21 testes). The depletion of GSCs in mutant testes was not caused by apoptosis as shown by staining for activated Caspase 3 (supplementary material Fig. S2), suggesting that ball is required for the maintenance of GSCs. At late larval development, ball mutant testes contained cysts at progressive stages of proliferation, including mature 16 cell cysts as in the case of wild type larval testes. Based on these observations, we cannot conclude that all the ball mutant cysts would eventually complete their differentiation program and give rise to 16 cell cysts. However, the presence of germline cysts in ball² mutant testes, which are formed by the proliferation of gonialblasts indicates that BALL is not essential for cell cycle progression and suggests that its primary function is to maintain the self-renewing capability of GSCs.

We next asked whether the absence of BALL has the same effect in the female germline. Larval ovary development depends on symmetric divisions of PGCs (Fig. 1H) (Dansereau and Lasko, 2008). Wild type ovaries during midstage of larval development contain PGCs with spectrosomes, but no differentiating cysts with HTS stained fusomes (Fig. 1I,J). In ball² mutant ovaries, however, we found fused structures in VASA expressing cells, which resembled fusomes of germline cysts (Fig. 1K,L). This result suggests that ball mutant PGCs differentiated prematurely. ball² mutant ovaries also contained fewer PGCs (26.7±3.6, n = 20 ovaries) than wild type ovaries (35.2±6.6, n = 23 ovaries; P = 6*10⁻⁶). The reduced number of PGCs does not result from apoptosis as staining for activated Caspase 3 showed that these differentiating germline cells were not apoptotic at mid larval development (supplementary material Fig. S2). Therefore, as observed with testes, ball² mutant germline cells in ovaries were able to proliferate. However, they proliferated at a slightly reduced rate compared to wild type. This reduction could either be due to a direct function of BALL in modulating cell cycle progression or reflect a difference in the cell cycle length between symmetrically dividing wild type PGCs and prematurely differentiating ball² mutant germline cells. At later stages of larval development, massive apoptosis in ball mutant PGCs was observed (Fig. 1M,N). This observation is reminiscent of previous studies on the function of nanos and pumilio, showing that mutant PGCs enter apoptosis after their premature differentiation (Gilboa and Lehmann, 2004; Wang and Lin, 2004).

Our results suggest that the primary function of BALL is to maintain the undifferentiated state of male and female germline progenitor cells. To discriminate between a cell autonomous function of BALL in stem cells and a systemic non-autonomous function of BALL, we generated genetic mosaics in adult ovaries. GSCs and early germline cysts of adult ovaries are located at the anterior tip of the ovariole within a stem cell niche formed by somatic cap cells and terminal filament cells (Fig. 2A). In wild type, GSC divisions typically result in a self-renewed GSC and a cystoblast (CB). The CB divides four times into a 16-cell germline cyst in which cells remain interconnected due to incomplete cytokinesis. The most posterior cell in the cyst adopts an oocyte fate, whereas the remaining 15 cells differentiate into nurse cells. GSCs and CBs both carry a spherical spectrosome, whereas differentiating cysts contain a branched fusome. GSCs remain in direct contact with the stem cell niche with their spectrosome always oriented towards the niche and thus, they can be unambiguously distinguished from CBs (Fig. 2A). In order to generate genetic mosaics, we used the FLP/FRT system (Xu and Rubin, 1993) to functionally identify ball² mutant cell clones that derive from GSCs and CBs by the absence of an ovo^D1 expressing transgene. ovo^D1 expression induces apoptosis in germline cysts (Perrimon, 1998). Thus, only cysts that derive from a ball² mutant GSC or CB lack the ovo^D1 expressing transgene and hence, can develop into eggs. We generated ball² mutant germline clones by heatshock-induced flipase expression (Xu and Rubin, 1993) and monitored the number of eggs deposited at various time points after heatshock treatment (AHT).

The results show that ball² mutant eggs were first deposited six days AHT (Fig. 2B). This indicates that ball² mutant germline clones can differentiate into mature eggs. However, ball² mutant egg deposition vanished after three days (Fig. 2B). Germline specific expression of ball, achieved by combining a UASp-ball-EGFP transgene with a nos-GAL4 driver transgene, restored the capacity of ball² mutant GSCs to sustain continuous egg production (Fig. 2B). In conclusion, these results show that germline autonomous expression of ball is sufficient to allow continuous egg production and, therefore, also to maintain self-renewing GSCs.

The lack of continuous egg production from ball mutant germline clones could be explained by a defect in germline differentiation which, however, was masked by the perdurance of BALL protein after clone induction. In order to directly address the function of ball for GSC self-renewal and germline differentiation, we generated genetic mosaics identifying ball² mutant cell clones by the absence of GFP expression (Fig. 2C). In three replicate time course experiments, we found 0.32±0.017 ball² mutant GSCs per germarium (n = 102) at two days AHT. This number decreased to 0.13±0.021 (n = 94 germaria) and 0.034±0.039 (n = 125 germaria) three and four days AHT, respectively. To distinguish the loss of ball² mutant GSCs from the regular turnover of GSCs, we counted the number of wild type ‘twin clone’ GSCs in the same set of germaria. These GSCs are initially generated at the same frequency as mutant GSCs in genetic mosaics and can be identified by their increased GFP fluorescence intensity (Fig. 2D). The number of twin clone GSCs remained constant over the time course (0.38±0.049, 0.39±0.11 and 0.39±0.11 after two, three and four days, respectively), showing that ball² mutant GSCs were not lost due to regular turnover. Furthermore, ball² mutant cells did not undergo cell death during the time period as shown by staining for activated Caspase 3 (supplementary material Fig. S3). These results show that ball² mutant GSCs had lost their capacity to self-renew and left the niche.

To analyse the differentiation of ball² mutant germline clones, we stained for the ORB protein, which becomes progressively enriched in the oocyte during cyst differentiation (Christerson and McKearin, 1994) (Fig. 3A–C). The distribution of ORB was similar in ball² mutant and wild type cysts (Fig. 3D–F), indicating that oocyte specification was normal. However, we noted that in about half of the ball² mutant germline cysts a single nurse cell degenerated and that all oocytes displayed the characteristic defect in chromatin organisation that has been previously reported (Ivanovska et al., 2005) (supplementary material Fig. S4). Nevertheless, the overall differentiation of ball mutant cysts continued normally beyond the mitotic stages (supplementary material Fig. S4). This finding is consistent with the formation of mature eggs that were observed with the germline clone experiments (see above).

Using anti-BALL antibodies, we did not detect residual BALL protein in differentiating ball mutant germline cysts (data not shown). Germline cyst formation requires four cell divisions during which residual BALL protein present in a ball mutant CB would be diluted. Residual BALL protein in GSCs would similarly be diluted by cell divisions, i.e. ball mutant GSCs were lost within 4 days, which allows about four divisions of GSCs. Although we can formally not rule out residual BALL protein below the detection limit in differentiating cysts, the data strongly support that the requirement of BALL for GSC self-renewal is at least by far more critical than for germline differentiation.

To assess whether the function of BALL for self-renewal of GSC is specific for the female germline, we examined the function of BALL in GSCs of adult testes. In adult testes, about nine GSCs are in direct contact with the stem cell niche formed by hub cells at their anterior tips (Fig. 4A). Both, GSCs and the differentiating daughter cells, called gonialblasts, carry a spectrosome with a random orientation. GSCs can be unambiguously distinguished from gonialblasts by their position adjacent to the niche. We generated genetic mosaics with ball² mutant cells visualised by the absence of GFP expression (Fig. 4B,C). Instead of a twin-clone analysis, we generated GFP marked wild type cells, since twin clone analysis was not reliable in testes GSCs (Fig. 4D,E). We counted the number of marked GSCs per testis, showing that two days AHT, both ball² mutant GSCs and wild type GSCs were equally frequent, with 2.04±0.15 (n = 93 testes) and 1.9±0.42 (n = 79 testes) GSCs per testis, respectively. After three and four days AHT, however, the frequency of ball² mutant GSCs decreased to 0.95±0.17 (n = 99 testes) and 0.32±0.11 (n = 85 testes), respectively, whereas the wild type control GSCs were maintained at the originally observed high frequency, i.e. 1.85±0.4 (n = 35 testes) and 1.86±0.56 (n = 75 testes) GSCs, respectively. No cell death was found in ball² mutant GSCs between day two and four AHT (n≥50 testes per day). These results establish that BALL has a critical function in maintaining self-renewing GSCs both in the male and female germline.

We next asked whether the function of BALL is limited to stem cells of the germline and thus, we extended our analysis on the somatic cell lineage in ovaries. During egg development, germline cysts become encapsulated by epithelial follicle cells (FCs), which are continuously generated by somatic follicle stem cells (FSCs) in the germarium (Fig. 2A).

Before FCs terminally differentiate, they undergo multiple cell divisions in the follicle epithelium around the egg chambers. To assess the proliferation of wild type and ball mutant FCs, we compared the sizes of corresponding mutant and wild type twin clones (Fig. 5A–D). The average ratio of cell numbers in ball² versus wild type twin clone pairs was 1.06±0.45 (n = 31 pairs), indicating that wild type and ball² mutant FCs proliferated at a similar rate.

Since FCs are continuously displaced from the germarium together with the developing egg chambers, the functional availability of FSCs can be monitored by the appearance of newly generated FCs in the germarium. Two days after induction of clones, 45% of the germaria (n = 33) contained ball² mutant FCs (Fig. 5E,F). However, ten days after induction of clones, no ball mutant FCs were detected (n = 35), but 40% of the germaria (n = 30) contained wild type twin clone FCs and non-mutant control clones were also still present (Fig. 5G,H). Because we did not directly identify FSCs as we did in the case of the GSCs, we could not discriminate whether ball mutant FSCs died, stopped proliferation or have been dislocated from the niche and differentiated. The ball loss of function phenotype in the somatic lineage of the ovary however shares extensive similarity with the germline phenotype in a sense that BALL is dispensable for the proliferation of differentiating cells but required for the continuous supply of cells that maintains tissue homeostasis through self-renewing stem cells.

The stem cell populations in both male and female gonads rely directly or indirectly on BMP signalling. We used the female GSC system, in which the requirement of BMP signalling is most prominent, to address the relevance of BALL to BMP signalling. We assayed two hallmarks of BMP signalling in GSCs by immunostainings, i.e. the phosphorylation of the transcription factor MAD (pMAD) and the repression of the bag of marbles (bam) gene (Song et al., 2004). In wild type as well as in ball² mutant GSCs, pMAD was stained in GSCs (Fig. 6A,B). Consistent with functional BMP signalling and MAD phosphorylation, bam was properly repressed in ball² mutant GSCs (Fig. 6C,D) and bam expression was normally upregulated in differentiating cells of ball² mutant cysts (Fig. 6E,F). In conclusion, these observations indicate that BALL acts independently or downstream of the BMP signalling pathway in female GSCs.

Upregulation of the differentiation factor BAM is critically required for the differentiation of CBs into germline cysts (Chen and McKearin, 2003; Song et al., 2004). bam^Δ86 single mutant GSCs generate daughter cells that fail to differentiate but instead develop stem cell-like tumours in the anterior region of the germarium (Fig. 7A). We generated genetic mosaics with ball² bam^Δ86 double mutant GSCs and asked whether their differentiation into cysts is a prerequisite for the elimination of ball mutant GSCs from the niche. ball² bam^Δ86 double mutant GSCs, were lost from the stem cell niche (Fig. 7B) as described for ball² single mutant GSCs (Fig. 2C). Moreover, in a time course experiment, we could not detect apoptotic ball² bam^Δ86 double mutant GSCs in the niche (supplementary material Fig. S3D). Importantly, niche-detached ball² bam^Δ86 double mutant cells did not differentiate into germline cysts (Fig. 7B). Thus, loss of BALL does not bypass the requirement of BAM for differentiation. These results indicate that ball mutant GSCs are not eliminated from the niche because they ectopically initiate differentiation, but through a differentiation-independent process.

In addition to producing differentiation defective CBs, bam^Δ86 mutant GSCs become “super-competitive stem cells”, i.e. they displace wild type GSCs from the stem cell niche (Fig. 7A) through cellular competition (Jin et al., 2008; Rhiner et al., 2009). The fact that this increased competiveness of bam^Δ86 mutant GSCs is abrogated by the loss of ball in ball² bam^Δ86 double mutant GSCs suggests that BALL is a factor that mediates cellular competition and that ball mutant GSCs are lost from the stem cell niche due to reduced competitiveness and independent of their ability to differentiate.

We noticed that ball² bam^Δ86 double mutant cells did not form stem cell tumours after leaving the niche but eventually degenerated (Fig. 7B). This finding is different from ball² bam^Δ86 double mutant GSCs that were not degenerating while they resided in the niche (supplementary material Fig. S3D), i.e. in an environment that promotes the undifferentiated state of GSCs. Similarly to ball² bam^Δ86 double mutant germline cells outside the niche, larval PGCs lack BAM expression and ball² mutant PGCs undergo apoptosis (Fig. 1M,N). Therefore, we asked whether there is a general synergistic requirement of BAM and BALL for the survival of differentiating germline cells. Ectopic expression of bam from a transgene was sufficient to induce differentiation of wild type PGCs (Fig. 7C–E), but did not maintain the survival of differentiating ball² mutant PGCs (Fig. 7D–F). These observations argue against an interdependent requirement of BAM and BALL for the survival of differentiating germline cells.

Loss of BALL-dependent phosphorylation of BAF was identified as the primary cause for the chromatin organization defects in ball mutant oocytes (Lancaster et al., 2007). However, we did not observe severe chromatin organization defects in ball² mutant GSCs or cells other than the oocyte (supplementary material Fig. S5). This observation suggests that BALL is not generally involved in global chromatin organization, but might affect specific aspects of it in specific cells or organs. While reinvestigating the subcellular localization of BALL, we found that BALL not only covers the entire chromatin (Aihara et al., 2004), but is also highly enriched in the nucleolus (Fig. 8A–E). The nucleolus is assembled around the chromatin region that harbours the rDNA repeats and can be visualized with antibodies against Fibrillarin that marks the granular zone of the nucleolus. Compared to GSCs, the size of the nucleolus becomes significantly decreased during the differentiation of germline cysts, which suggests a reorganization of the nucleolar content, and it increases again when postmitotic germline cyst cells start to polyploidize (Neumüller et al., 2008). To see whether this change of nucleolar organization includes a reorganization of the chromatin that carries the rDNA genes, we performed DNA-FISH experiments in combination with high-resolution STED microscopy. Comparing GSCs to differentiating germline cells revealed that the rDNA genes in the nucleolus become significantly compacted during differentiation (Fig. 8G,H).

The volume of nucleoli, as assayed by Fibrillarin staining, continuously decreased with the distance of germline cells from the stem cell niche, which reflects the differentiation process of wild type germline cysts (Fig. 8I). bam^Δ86 mutant tumourous germline cells maintained a GSC-like nucleolar size irrespective of the distance to the stem cell niche (Fig. 8I,J), consistent with the fact that these cells proliferate like GSCs outside the niche. Hence, we used the bam^Δ86 mutant cells to address whether BALL is required for a GSC-like organization of the nucleolus. In ball² bam^Δ86 double mutant mosaics, the nucleoli of about half of the mutant cells were only slightly smaller (53%, n = 110 cells) than the nucleoli of bam^Δ86 mutant cells (Fig. 8I). This observation suggests that the mechanisms that induce the expansion of nucleoli in GSCs are still functional in the absence of BALL. Maintenance of this nucleolar expansion is however dependent on BALL activity, since in the other half of the mutant cells the nucleoli were either fragmented or completely disintegrated (47%, n = 110 cells) (Fig. 8K). Interestingly, we did not observe degeneration of nucleoli in those ball² bam^Δ86 double mutant cells that resided in the stem cell niche where self-renewal is supported by BMP signalling. This result suggested that BALL is either directly or indirectly required to maintain the nucleolus of GSC-like tumorous cells that cannot differentiate due to the absence of BAM, but no longer receive the niche derived maintenance signals. To see whether BALL is required for maintenance of expanded nucleoli in cells that normally differentiate, we asked whether the nucleoli of follicle cells, which increase their nucleoli during endoreduplication, are affected by the absence of BALL. Staining of these nucleoli with Fibrillarin antibodies showed that the nucleoli of neither ball² mutant (Fig. 8L,M) nor ball² bam^Δ86 double mutant cells were affected in differentiating follicle cells (Fig. 8N,O). These observations suggest that BALL is not generally required for the generation or maintenance of an expanded nucleolus but has a specific function in tumorous stem cells.

Our study shows that the VRK-1 kinase BALL is required for self-renewal of germline stem cells in Drosophila, including the symmetrically amplifying PGCs of larvae and both male and female GSCs. These stem cells are actively maintained undifferentiated and they require BMP signalling for self-renewal that emanates from their cellular niche environments (Losick et al., 2011). In ball² mutant female GSCs, where the requirement of BMP signalling for self-renewal is most pronounced, known targets of BMP signalling are regulated as in wild type GSCs. This indicates that BALL participates neither in the transmission nor the regulation of BMP signalling, and that it is needed to maintain stem cell character in a cell autonomous manner, irrespective of the tissue-specific maintenance signals that emanate from the niches.

The loss of self-renewing stem cells could be caused by the induction of ectopic differentiation in these cells. We blocked the differentiation pathway in ball mutant female GSCs by removing also the central differentiation factor BAM. The results suggest that ball mutant GSCs are not eliminated from the stem cell niche because they initiate germline differentiation but that the GSCs differentiate because they lost the capacity for self-renewal.

It is unclear by which mechanism BALL mediates the ability for GSC self-renewal. In ovaries, GSCs and FSCs undergo a regular turnover and are continuously replaced in the niche either by their own daughter cells or by symmetric divisions of the neighbouring stem cells (Nystul and Spradling, 2007; Xie and Spradling, 2000). The replacement of GSCs involves competition between stem cells. Cells lacking BAM for instance, successfully displace less competitive wild type stem cells in the niche (Jin et al., 2008). However, if BALL is additionally removed from bam mutant cells, they appear to loose their competitive advantage. The molecular basis of stem cell competiveness is still poorly understood. However, it has been shown that overexpression of the Drosophila dMyc transcription factor diminutive enhances the competitiveness of GSCs (Rhiner et al., 2009) and causes significantly enlarged nucleoli and increased rRNA expression in epithelial cells (Grewal et al., 2005). These observations suggest a correlation between ribosome biogenesis and GSC competitiveness. Downregulation of ribosome biogenesis appears in fact to be directly required for germ cell differentiation, since BAM activates the Mei-P26 protein which downregulates the expression of dMyc. Furthermore, when overexpressed from a transgene, dMyc abrogates the tumour growth phenotype and the size increase of nucleoli in bam mutant cells (Neumüller et al., 2008). Additional support for the proposal that increased ribosome biogenesis in stem cells is crucial for their competitiveness and for maintaining their undifferentiated state derives from studies on wicked. Wicked is an essential component of the U3 snoRNP pre-rRNA processing, which is required to maintain the self-renewal of GSC (Fichelson et al., 2009). Our findings that BALL is enriched in stem cell nucleoli and required for the structural integrity of nucleoli in tumourous GSCs provide a plausible link between ribosome biogenesis and BALL-dependent competitiveness of GSCs.

Once displaced from the niche, ball² mutant female GSCs differentiate according to their germline fate with only minor defects. We did not address whether the differentiation of ball mutant cells is fully completed like in the respective wild type lineages in systems other than the adult female germline, but we found, irrespective of the system we looked at, that BALL is not strictly required as proliferation factor. Especially the analysis of dividing follicle cells showed that BALL is not a cell cycle regulator. With two remarkable exceptions, BALL is also not essential for cellular survival. These exceptions, i.e. ball mutant PGCs at late larval stages and ball² bam^Δ86 double mutant germline cells outside the ovarian niche, represent conditions in which differentiation is either not supported by the tissue or not possible due to the lack of a differentiation factor. Therefore, it is tempting to speculate that BALL becomes only essential for cellular survival, when the ball mutant stem cells are unable to ‘escape’ from self-renewal into differentiation. Although ball clearly has multiple functions, e.g. oocyte chromatin organization or modulation of female PGC proliferation rate, the common defect observed in all systems examined so far is a failure in maintaining pools of undifferentiated cells. Since BALL is not an essential proliferation factor, our data suggest that also the additional defects in ball mutant larvae, i.e. lacking imaginal discs and degenerate brains, could be due to premature loss of undifferentiated progenitor or stem cells. Analysis of these systems will eventually show whether BALL is broadly required to maintain the undifferentiated state of cells during development.

The ru ca e ball² chromosome was generated by imprecise excision of P{EP}ball^EP863 (ball¹). The chromosomes P{neoFRT}82B, P{neoFRT}82B e ball², P{neoFRT}82B e bam^Δ86 and P{neoFRT}82B e bam^Δ86ball² were constructed by meiotic recombination. The transgene P{w^+mC UASp-ball.T:Avic/EGFP = pballE}2.1 was used for BALL-EGFP expression.

Clones with ovo^D1 transgenes were induced in 1–3 (ball):

1. y¹w* P{ry⁺, hs-FLP}1/w*; P{neoFRT}82B P{ovoD1-18}3R/P{neoFRT}82B e ball²

2. y¹w* P{ry⁺, hs-FLP}1/w*; P{pballE}2.1/P{GAL4-nos.NGT}40; P{neoFRT}82B P{ovoD1-18}3R/P{neoFRT}82B e ball²

3. y¹w* P{ry⁺, hs-FLP}1/w*; P{pballE}2.1/wg^Sp1; P{neoFRT}82B P{ovoD1-18}3R/P{neoFRT}82B e ball²

GFP-marked clones were induced in:

1. y¹w* P{ry⁺, hs-FLP}1/w*; P{neoFRT}82B P{Ubi-GFP}83/P{neoFRT}82B e ball² (males: w* X-chromosome replaced for Y chromosome)

2. y¹w* P{ry⁺, hs-FLP}1/w*; P{neoFRT}82B P{Ubi-GFP}83/P{neoFRT}82B e bam^Δ86

3. y¹w* P{ry⁺, hs-FLP}1/w*; P{neoFRT}82B P{Ubi-GFP}83/P{neoFRT}82B e bam^Δ86ball²

4. y¹w* P{ry⁺, hs-FLP}1/Y; P{neoFRT}82B P{Ubi-GFP}83/P{neoFRT}82B

5. y¹w* P{ry⁺, hs-FLP}1/P{bamP-GFP}, w*; P{neoFRT}82B P{arm-lacZ.V}83B/P{neoFRT}82B e ball²

Female flies were used (1) to assay the ball requirement in egg production, (2) for germline specific rescue and (3) to serve as control to show that rescue was not caused by basal expression of the ball transgene. Female and male flies (4) were used to generate marked ball² mutant clones in germaria or testis. Control non-mutant clones were induced in male flies (7). Repression of a bam-GFP reporter construct was assayed in (8). To generate bam mutant and double mutant clones, the loss of function allele bam^Δ86 was used in (5) and (6), respectively.

To obtain ball² mutant larvae, this allele was balanced with TM3, Ser¹, P{ActGFP}JMR2 and newly hatched GFP negative larvae were collected over a two hours interval. Controls and were treated identically. Larvae were subsequently reared at a controlled density before dissection. Induced expression of bam through the P{hs-bam.O}18d integration was accomplished by heat shocks (45 min, 37.5°C) every 24 hours starting from 12 h after larval hatching.

Flies aged 0–2 d were fed on yeast for 1 d at 25°C. Subsequently, three heat shocks (1 h, 38°C) were applied in 12 h intervals by placing flies in empty vials with moist foam stoppers in a water bath. In between heat shocks flies were kept on yeast at 25°C. Following heat shock treatments flies were mated to wild type flies and kept well fed at 25°C for a time course starting with the last heat shock.

All dissections were done in Schneider's cell culture medium (Life Technologies, Paisley, UK) at room temperature (RT) for no longer than 15 min before fixation in 4% (w/v) Paraformaldehyde/PBS/50 mM EDTA, pH 7.0 for 10 min. After fixation tissue was at RT rinsed in PBS, 0.1% (w/w) Triton X-100 (PBTx), extracted with 1% (w/w) Triton X-100 for 30 min and blocked with PBTx, 10% (v/v) goat serum for at least 30 min. Staining was done in blocking solution at 4°C over night (primary antibodies) or at RT for 2 h (secondary antibodies). Primary antibodies were affinity purified rabbit anti BALL (1:400, generated against residues 1–352 of BALL, A.H.), rabbit anti VASA (1:2500, generated against full length Vasa, A.H.), rabbit anti Cleaved Caspase3 Asp175 (Xu et al., 2006) (1:150, Cell Signaling Technologies, Boston, MA), rabbit anti pSMAD1 PS1 (1:500, gift from C. H. Heldin, Uppsala, Sweden), rabbit anti GFP (1:500; Synaptic Systems, Göttingen, Germany), mouse anti HTS 1B1 (1:10, DSHB, Iowa, USA), mouse anti ORB (4H8/6H4 1:1 mix, diluted 1:30, DSHB), mouse anti SHG DCAD2 (1:10), mouse anti Fibrillarin 38F3 (1:1000, Abcam, Cambridge, UK), rat anti BAM-C (1:500) (McKearin and Ohlstein, 1995), chicken anti beta-Galactosidase (1:1000, Abcam). Secondary antibodies against mouse and rabbit IgGs were coupled to Alexa-488, −568, −633 (1:400, Life Technologies). Secondary antibodies against chicken IgY were coupled to Cy5 (1:400, Abcam). After antibody incubations tissue was rinsed twice in PBTx and washed 3 times for 20 min in PBTx at RT. To visualize actin, staining with Phalloidin coupled to Alexa 568 (Life Technologies) was done at 2 U/ml in PBTx for 30 min at RT. For staining DNA, tissue was treated with RNaseA at 2 mg/ml in PBTx for 30 min, followed by staining with 10 µM draq5 (Biostatus, Shepshed, UK) or 1 µg/ml propidium iodide (Life Technologies) in PBTx for 10 min, respectively. Before mounting in Prolong Gold antifade-medium (Life Technologies) tissue was rinsed and washed for 10 min in PBTx. For identification of GFP-marked clones direct GFP fluorescence was assayed. For analysis of the bam-GFP reporter expression germaria were stained with anti GFP primary and Alexa 488 coupled secondary antibodies.

Germaria were fixed 4 min at 37°C in 4% paraformaldehyde, 15 mM PIPES, 80 mM KCl, 20 mM NaCl, 2 mM EDTA, 0.5 mM EGTA, 0.5 mM spermidine, 0.15 mM spermine, 1 mM DTT pH 7.2, rinsed 4 times with 2× SSC, 0.1% Tween-20 (WB), washed 10 min at RT in WB and incubated 30 min at RT with 1 mg/ml RNase A (DNase free, Sigma). After two washes with WB for 10 min at RT, germaria were rinsed and washed once in 2× SSC, 30% formamide (deionized, Sigma), 0.1% Tween-20 (PHB) at RT. After incubation with new PHB at 37°C for 30 min PHB was removed and 60 µl of probe solution was added that contained a 1 µM mix of 5′ biotinylated oligonucleotides in 3× SSC, 30% formamide, 0.1% Tween-20 (HYB). Oligonucleotides were (5′ to 3′): taaagaattttatcaagagt, caaacacctcgtcattaac, aggcagtggttgccgacctc, atattgttcaaaacgtatgt, catatgattttggcaattatatgag, aattaaatcatatacatatgaa, atatttattatatgtataagtg, aatattgaaatattcccatattctc, gtattatagagaatataattaat. Germaria were incubated with the probe solution for 15 min at 37°C, heated for 2 min to 91°C and incubated over night at 37°C. Then 500 µl of PHB were added at 37°C and germaria were washed in PHB 4 times for 10 min at 37°C.

For detection germaria were washed once in EB and twice in WB for 20 min at RT each and then for 20 min in blocking reagent/WB (Roche). Then Streptavidin-dye conjugate SA-Atto647N (ATTO-TEC, Siegen, Germany) diluted 1:25 in 250 µl new blocking solution was added. Incubation was for 2 h at RT in a thermomixer at 900 rpm. Germaria were then rinsed once and washed three times for 20 min at RT in WB. DNA was counterstained with 10 µM draq5 (Biostatus) in WB and samples were mounted in Prolong Gold antifade medium (Invitrogen). Images were acquired on a Leica TCS SP8 STED microscope.

For nucleolar size measurements z-stacks were collected at 0.1 µm z intervals and analysed by a modified Connected Threshold Grower plugin for Image J. Details on the modified plugin are available on request.

We thank Maria Warnefors for help with affinity purification of anti BALL antibodies, Sabine Häder for technical assistance, Anja Schmidt and Alexander Egner for help with the STED microscopy, and Ruth Lehmann and Dennis McKearin for sharing reagents.

This work was supported by the Max-Planck-Society.