Gbb/Bmp signaling is essential for maintaining germline stem cells and for repressing bam transcription in the Drosophilatestis

In adult tissues that are subject to continuous cell turnover, stem cells are responsible for maintaining tissue homeostasis. A stem cell usually undergoes a stereotyped asymmetric cell division: giving rise to one parent stem cell that continues to undergo asymmetric cell divisions and a daughter that commits to differentiation. Stem cell behavior has been shown to be controlled by specialized regulatory microenvironments or `niches' in several systems (Watt and Hogan, 2000; Spradling et al., 2001; Lin, 2002). However, niche signals in many stem cell systems are still poorly defined. Knowledge of niche signals is also important for expanding stem cells in in vitro cultures for cell replacement therapy. Therefore, the identification of niche signals in different stem cell systems is important not only for a better understanding of how stem cell function is controlled but also for using stem cells in future regenerative medicine.

The Drosophila testis has become one of the premier stem cell systems to study molecular mechanisms governing stem cell self-renewal,differentiation and proliferation (Kiger et al., 2000; Tran et al.,2000; Kiger et al.,2001; Tulina and Matunis,2001). In the testis, there are two types of stem cells, GSCs and somatic stem cells (also known as cyst progenitor cells), which are responsible for producing differentiated germ cells and somatic cyst cells that encapsulate differentiated germ cells, respectively(Fig. 1A). Seven to nine GSCs can be reliably identified by their attachment to hub cells (a group of tightly packed somatic cells) and existence of a spectrosome(Fig. 1A,B). The spectrosome is a spherical fusome that is unique to GSCs and their early progeny, also known as gonialblasts. The fusome is rich in cytoskeletal proteins such as Hu li tai shao (Hts) and α-Spectrin (Lin et al., 1994; de Cuevas and Spradling, 1997). Once a GSC divides, one daughter cell that is in contact with the hub cells retains stem cell identity, whereas the other daughter cell that is not in contact with the hub cells initiates differentiation and becomes a gonialblast. The gonialblast then undergoes four rounds of synchronous cell division to generate a 16-cell germline cluster in which individual germ cells are connected by ring canals and a branched fusome. During the course of germ cell development from a gonialblast to a 16-cell cyst, a pair of somatic cyst cells surrounds the gonialblast, or a developing germ cell cluster, and control proper germ cell proliferation and differentiation (Matunis et al., 1997).

Recent studies indicate that the hub cells and somatic stem cells could function as a niche for GSCs in the testis(Kiger et al., 2000; Tran et al., 2000; Kiger et al., 2001; Tulina and Matunis, 2001). The hub cells are known to express Unpaired (Upd; Os - FlyBase), a secreted ligand that activates the JAK-STAT signaling pathway in GSCs and promotes their self-renewal (Kiger et al.,2001; Tulina and Matunis,2001). GSCs mutant for upd and Dstat(Stat92E - FlyBase) are lost rapidly, whereas overexpression of upd prevents proper differentiation of gonialblasts, resulting in the accumulation of undifferentiated germ cells. Moreover, somatic cyst cells are required for the early stage of spermatogonial differentiation(Kiger et al., 2000; Tran et al., 2000). Loss of function of Egfr and raf in somatic cyst cells disrupts differentiation of gonialblasts, resulting in the accumulation of undifferentiated germ cells in the testis. Somatic cyst cells are also important to ensure that a gonialblast undergoes exactly four rounds of mitosis to generate a 16-cell cluster through a Tgfβ-like signaling pathway (Matunis et al.,1997). Furthermore, bag of marbles (bam) and benign gonial cell neoplasm (bgcn) act autonomously in the germline to restrict the proliferation of amplifying germ cells(Gonczy et al., 1997). However, the connection between the Tgfβ signal from somatic cyst cells and bam/bgcn in the germline remains unclear.

The Drosophila ovary represents another attractive stem cell system in which stem cells and their niche cells can be reliably identified(Xie and Spradling, 2001; Lin, 2002). Germline stem cells have first been demonstrated to be located in the niche, consisting of terminal filament/cap cells and inner sheath cells(Xie and Spradling, 2001; Lin, 2002). fs(1)Yband piwi are expressed in the terminal filament/cap cells and are essential for GSC maintenance (King and Lin, 1999; Cox et al.,1998; Cox et al.,2000; King et al.,2001). decapentaplegic (dpp), a Drosophila Bmp member, is expressed in somatic cells such as cap cells, and is essential for GSC maintenance and division in the Drosophila ovary (Xie and Spradling, 1998; Xie and Spradling, 2000). Interestingly, another Bmp member, glass bottom boat (gbb, also known as 60A), is highly expressed in the male (Wharton et al.,1991; Doctor et al.,1992), but its role in spermatogenesis has not been investigated. Here we show that both gbb and dpp, are expressed in the somatic cells of the testis and act cooperatively on GSCs to control their maintenance. In addition, gbb signaling is essential for repressing bam expression in GSCs in Drosophila.

The following fly stocks used in this study were described either in FlyBase or as otherwise specified: tkv⁸ and sax⁴ (Brummel et al.,1994); punt¹⁰⁴⁶⁰ and punt¹³⁵; Med²⁶(Das et al., 1998); Dad-lacZ (Tsuneizumi et al.,1997); dpp^hr4 and dpp^hr56; gbb⁴, gbb^D4 and gbb^D20; bam-GFP(Chen and McKearin, 2003a); vasa-GFP (Nakamura et al.,2001); c587-gal4, hs-gal4 and nanos-gal4VP16(Van Doran et al., 1998); UAS-dpp, UASp-bamGFP(Chen and McKearin, 2003a) and UAS-gbb (Khalsa et al.,1998); hsFLP; FRT_82Barm-lacZ; FRT_40Aarm-lacZ; FRT_G13arm-lacZ. Most stocks were cultured at room temperature. To maximize their mutant effects, dpp,gbb and punt mutant adult females were cultured at 29°C for 2-7 days.

Clones of mutant GSCs were generated by Flp-mediated mitotic recombination,as described previously (Xie and Spradling, 1998). To generate the stocks for stem cell clonal analysis, 2-day-old adult males carrying an armadillo-lacZ transgene in trans to the mutant-bearing chromosome were generated using standard genetic crosses and then heat-shocked at 37°C for 3 consecutive days with two one-hour heat-shock treatments daily separated by 8-12 hours. The males were transferred to fresh food every day at room temperature, and the testes were removed 2 days, 1 week and 2 weeks after the last heat-shock treatment,and then processed for antibody staining.

To construct the stocks for overexpressing dpp or gbb,nanos-gal4VP16 virgins were crossed with UAS-dpp and UAS-gbb males, respectively. The males that carried nanos-gal4VP16 and UAS-dpp or UAS-gbb were cultured at room temperature, or at 29°C, for one week. For examining the expression of bam-GFP in the testes overexpressing dpp or gbb, the bam-GFP/CyO; nanos-gal4VP16 virgins were used in the crosses.

To determine loss of marked mutant GSC clones, GSCs were marked in 1- to 2-day-old males of the appropriate genotype. Subsequently, testes were isolated from some of the males 2 days, 1 and 2 weeks later, and stained with anti-Hts and anti-β-Gal antibodies. The percentage of testes containing one or more marked GSCs was determined by counts of 55-227 testes at each time point.

To measure stem cell loss in gbb mutant testes, the testes with different numbers of GSCs were determined based on anti-Hts and anti-Fas3 antibody staining of gbb⁴/gbb^D4 or gbb⁴/gbb^D20 testes of different ages and different treatments. yw males carrying no gbb mutations served as a control. The 2-day-old control and gbb mutant males were cultured at different temperatures after they eclosed at 18°C. Values are expressed as the average GSC number per testis, and/or the percentage of testes carrying no GSCs.

To examine bam-GFP expression in dpp, gbb and punt mutant testes, we generated males with the following genotypes at 18°C: bam-GFP gbb⁴/gbb^D4, bam-GFP gbb⁴/gbb^D20, bam-GFP dpp^hr56/dpp^hr4 or punt¹⁰⁴⁶⁰/punt¹³⁵; bam-GFP. bam-GFP males carrying no mutations for gbb, dpp or punt served as a control. All the control and mutant males were cultured at 29°C for 4 days before their testes were isolated, stained with antibodies and compared for bam-GFP expression at identical conditions.

The TUNEL cell death assay was performed on punt mutant testes(Intergen Company).

The following antisera were used: polyclonal anti-Vasa antibody (1:2000)(Liang et al., 1994);monoclonal anti-Hts antibody (1:3); polyclonal anti-β-Gal antibody(1:1000; Cappel); monoclonal anti-β-Gal antibody (1:100; Promega);polyclonal anti-GFP antibody (1:200; Molecular Probes); polyclonal anti-pMad antibody (1:200) (Tanimoto et al.,2000). The immunostaining protocol used in this study was described previously (Song et al.,2002). All micrographs were taken using a Leica SPII confocal microscope.

The tips of the testes for the males of the appropriate genotype were dissected off from the whole testes in Grace's media, and were dissociated with collagenase II (Sigma) solution at a concentration of 6 mg/ml. After sorting of GFP-positive cells, using Cytomation MoFlo, from testes with GFP-marked hub cells, somatic cyst cells or germ cells, total RNA was prepared using Trizol (Invitrogen). RNA samples were further amplified using the GeneChip Eukaryotic Small Sample Target Labeling Assay Version II(Affymetrix). After RNA amplification, 100 ng of total RNA was reverse transcribed (RT) using SuperScriptIII First-Strand Synthesis System for RT-PCR(Invitrogen). The following primers were used in this study:

• dpp, 5′-AGCCGATGAAGAAGCTCTACG-3′ and 5′-ATGTCGTAGACAAGCACCTGGTA-3′;

• vasa, 5′-ATCGAGGAGGAAATCGAGATGGA-3′ and 5′-GGAAGCTATGCCACTGCTGAATA-3′;

• gbb, 5′-AGATGCAGACCCTGTACATAGAC-3′ and 5′-CTCGTCGTTCAGGTGGTACAGAA-3′; and

• rp49, 5′-GTATCGACAACAGAGTCGGTCGC-3′ and 5′-TTGGTGAGCGGACCGACAGCTGC-3′.

• PCR was performed as follows: 95°C for 4 minutes; 40 cycles of 95°C for 30 seconds, 45°C for 30 seconds and 72°C for 45 seconds; and 72°C for 7 minutes. RT-PCR products were electrophoresed on 2% agarose gel in the presence of ethidium bromide.

To investigate the possible role of dpp and gbb in maintaining male GSCs, we examined GSCs in the testes of temperature-sensitive dpp and gbb mutant males. Two homozygous allelic combinations used in this study, gbb⁴/gbb^D4 and gbb⁴/gbb^D20, were allowed to develop to adulthood at 18°C, and subsequently shifted to 22°C or 25°C for 7 days. In this study, an anti-Hts antibody was used to label spectrosomes and fusomes, whereas a DNA dye, DAPI, was used to stain nuclei. The hub can be reliably identified by molecular markers, such as Fasciclin 3 (Fas3), or by DAPI staining (small, DAPI-bright nuclei in the hub cells tightly packed together), whereas GSCs are identified by the presence of a spectrosome and direct contact with the hub cells. The numbers of GSCs in the testes of different mutants were quantified after the testes were immunostained for Hts and Fas3 to visualize spectrosomes in GSCs and hub cells, respectively. A wild-type testis carried an average of 9.1 GSCs (n=42, Fig. 1B), and these stem cells would persist for 7 days at 25°C or 29°C. One week after being cultured at 22°C, the testes from gbb⁴/gbb^D4 or gbb⁴/gbb^D20 mutants contained 1.3(n=21) or 1.8 (n=22) stem cells, respectively(Fig. 1C). One week after being cultured at 25°C, no GSCs were observed in the mutant testis(Fig. 1D). These results indicate that gbb is essential for maintaining GSCs in the testis.

Two temperature-sensitive allelic combinations, dpp^hr4/dpp^hr56 and dpp^e90/dpp^hr56 were used to investigate the role of dpp in maintaining GSCs in the testis. Similarly, dpp homozygous males were developed to adulthood at 18°C and were then shifted to a restrictive temperature (29°C) for one week. In the Drosophila ovary, both mutant combinations lose their GSCs very rapidly at a restrictive temperature (Xie and Spradling, 1998). Surprisingly, one week after being cultured at the restrictive temperature, the testes from dpp mutants had no significant GSC loss: dpp^hr4/dpp^hr56 and dpp^e90/dpp^hr56 mutant testes had an average of 7.4 (n=44) and 8.8 (n=49) GSCs/testis, respectively(Fig. 1E), which is in contrast with the severe GSC loss phenotype in the dpp mutant ovary and in the gbb mutant testis.

The two allelic dpp combinations used in this study represent very weak dpp mutants. As there is a stringent requirement for dpp during early Drosophila development, it is difficult to examine GSC loss in stronger dpp mutants because they do not survive to adulthood, even at 18°C. It is still possible that the role of dpp in the maintenance of male GSCs can be revealed if gbbsignaling is comprised, as dpp and gbb could use the same receptors and downstream components to transduce their signals(Khalsa et al., 1998). To further study the role of dpp in the regulation of male GSCs, we constructed two mutant strains homozygous for two dpp allelic combinations that were also heterozygous for gbb: dpp^hr4/dpp^hr56 gbb^D4 and dpp^e90/dpp^hr56 gbb^D4. The testes from the heterozygous gbb^D4, which were cultured at 29°C for one week, had a normal GSC number (8.6 GSCs/testis, n=38). The testes from dpp^hr4/dpp^hr56gbb^D4 and dpp^e90/dpp^hr56gbb^D4 had an average of 3.0 (n=13) and 5.7(n=56) GSCs/testis, respectively(Fig. 1F), in comparison with 7.4 and 8.8 GSCs/testis for dpp mutants alone, suggesting that partial removal of gbb function can enhance the dpp-mutant GSC-loss phenotype in the Drosophila testis. These results indicate that dpp and gbb function cooperatively to regulate male GSCs in Drosophila.

To further confirm that Bmp signaling is essential for maintaining male GSCs, we studied mutant phenotypes for one of the Bmp downstream components, punt, which encodes a type II serine/threonine kinase receptor for dpp, and also possibly for gbb(Letsou et al., 1995; Ruberte et al., 1995). A punt allelic combination, punt¹⁰⁴⁶⁰/punt¹³⁵, exhibits a temperature-sensitive phenotype: developing to adulthood at 18°C and showing mutant phenotypes at 29°C(Theisen et al., 1996). Interestingly, punt¹⁰⁴⁶⁰/punt¹³⁵ mutant males had normal GSC numbers (8.5 GSCs/testis, n=20) after being cultured at 22°C for a week (Fig. 2A). However, one week after shifting to 29°C, almost all the mutant testes completely lost their GSCs (0.1 GSCs/testis, n=58; Fig. 2B,C), although wild-type testes still maintained normal GSC number under the same conditions (data not shown). To exclude the possibility that Bmp signaling is important for GSC survival, we applied the TUNEL labeling assay to look for dying GSCs in punt mutant testes. During the one-week period at 29°C, no dying GSCs were detected in the punt mutant testes, but some rare dying cyst cells or differentiated germ cells were observed (n=38, Fig. 2D), suggesting that GSC loss is most likely caused by differentiation triggered by the lack of sufficient Bmp signaling. This result further supports the idea that Bmp signaling is essential for maintaining GSCs in the Drosophilatestis.

The GSC loss caused by defective Bmp signaling could be due to direct and/or indirect signaling to GSCs. To investigate whether Bmp signals are directly received by GSCs, we assessed Bmp signaling activities in GSCs by examining the expression of Dad. Dad is a dpp-responsive gene that negatively regulates dpp signaling(Tsuneizumi et al., 1997). Interestingly, Dad-lacZ, which reflects Dad mRNA expression(Tsuneizumi et al., 1997), was expressed in GSCs and gonialblasts, but not in more differentiated spermatogonial cells (Fig. 3A),indicating that Bmp signals function as short-ranged signals, and their activities are restricted to GSCs and gonialblasts. Moreover, it was also expressed in cyst cells at higher levels but generally not in cyst progenitor cells (Fig. 3A).

To further determine whether mutations in gbb affect Dadexpression in GSCs, we examined the expression of Dad-lacZ in the gbb⁴/gbb^D20 mutant background. After gbb mutant males were cultured at 22°C for four days, 97% of mutant GSCs (n=101) did not express Dad-lacZ(Fig. 3B). Interestingly, the gbb mutant testes carrying the Dad-lacZ mutation (4.8 GSCs; n=21) had more GSCs than the gbb mutant testes (2.4 GSCs; n=16). Dad-lacZ is a P element insertion mutation, which augments dpp signaling(Tsuneizumi et al., 1997). We also examined the expression of Dad-lacZ in the dpp^hr4/dpp^hr56 mutant background. Even after the dpp mutant males were cultured at 29°C for 4 days, Dad-lacZ expression was only slightly reduced (data not shown), which is consistent with no obvious GSC loss in the testis of dpp mutants. These results suggest that Dad is primarily a gbb-responsive gene that is also likely to negatively regulate gbb signaling in the Drosophila testis.

To further test whether Dad could inhibit both gbb and dpp signaling, we overexpressed Dad in germ cells using the Gal4-UAS bipartite expression system (Brand and Perrimon, 1993). A germline-specific nanos-gal4VP16driver can express a target gene under the control of UAS promoter specifically in germ cells (Van Doran et al., 1998), whereas a UAS-Dad transgene can be used to produce Dad under a gal4 driver to inhibit dpp signaling(Tsuneizumi et al., 1997). When the UAS-Dad transgene was used to overexpress Dad in germ cells by nanos-gal4VP16, all GSCs were lost in the testes before adulthood (Fig. 3C), indicating that blocking Bmp signaling causes GSC loss or prevents the formation of GSCs. The GSC loss phenotype induced by Dad overexpression mimics that of gbb mutants, suggesting that Dad overexpression probably inhibits not only dpp signaling but also gbb signaling. Thus, Dad-lacZ expression in GSCs may reflect the activities of both dpp and gbb signaling pathways. Together, these results suggest that Bmp signals appear to function as short-range signals to control GSC maintenance through direct signaling to GSCs.

In Drosophila, Dpp brings type I receptors Tkv and Sax, and the type II receptor Punt to form receptor complexes, which in turn phosphorylate Mad (Brummel et al., 1994; Nellen et al., 1994; Xie et al., 1994; Letsou et al., 1995; Ruberte et al., 1995; Newfeld et al., 1996; Newfeld et al., 1997). The phosphorylated Mad (pMad) is then associated with Medea (Med) and translocated to the nucleus to function as transcriptional activators for dpp-responsive genes (Das et al.,1998; Wisotzkey et al.,1998). pMad expression has been directly associated with dpp signaling activity in responding cells(Tanimoto et al., 2000). To further determine whether gbb signaling is responsible for pMad expression in GSCs, we examined the pMad expression in wild-type, gbband punt mutant GSCs in the testis. pMad preferentially accumulated in GSCs but was absent from gonialblasts and two-cell germ cell clusters(Fig. 3D), which is in contrast to Dad-lacZ expression in both GSCs and gonialblasts. This difference could be due to the perdurance of lacZ mRNA and/or protein. Alternatively, levels of pMad in gonialblasts are low and undetectable with the existing anti-pMad antibody. In the gbb mutant testes that still maintained some GSCs, pMad expression in the GSCs was severely reduced and below the limits of detection (Fig. 3E). In the testes of punt¹⁰⁴⁶⁰/punt¹³⁵ mutant males cultured at 29°C, pMad levels in GSCs were severely reduced and were sometimes difficult to detect (Fig. 3F). However, pMad expression in late 16-cell germ cell clusters remained high in both the gbb and punt mutant testes (data not shown). Therefore, gbb probably signals through common Bmp receptors, which leads to phosphorylation of Mad and Dad transcription.

To definitely confirm that Bmp signals directly act on GSCs and control their maintenance, we used the FLP-mediated FRT mitotic recombination to generate marked GSC clones mutant for Bmp downstream components (Xie and Spradling et al., 1998; Kiger et al.,2001; Tulina and Matunis,2001). The armadillo-lacZ transgenes that are strongly expressed in all the cells in the tip of the testis were used to mark mutant GSC clones. The marked GSCs were induced in adult testes by heat-shock treatments and identified as lacZ negative, spectrosome-containing germ cells that are in direct contact with the hub cells. The percentages of testes carrying one or more marked GSCs were determined at different time points after clone induction. The rate of loss of GSCs mutant for different Bmp downstream components can be used to determine how each Bmp downstream component contributes to the regulation of GSCs.

GSC clones mutant for punt, tkv, sax, mad and Med were generated as described previously (Xie and Spradling, 1998), and their testes were examined 2 days later. Two days after clone induction, 100% of the testes carried one or more marked wild-type GSCs (Fig. 4A),whereas 2 weeks after clone induction, 63% of the testes still carried one or more marked wild-type GSCs (Fig. 4B, Table 1). Two days after clone induction, over 80% of the testes still carried one or more marked GSCs mutant for tkv, sax, punt, mad or Med(Fig. 4C,E; Table 1). In contrast to wild-type clones, marked GSC clones mutant for punt, tkv, sax, madand Med were lost rapidly 2 weeks after clone induction(Fig. 4D,F; Table 1). For example, none of the testes mutant for punt¹⁰⁴⁶⁰, punt¹³⁵, tkv⁸, mad¹² and Med²⁶ had any GSCs left 2 weeks after clone induction. punt¹⁰⁴⁶⁰ is a moderate allele, while the rest are strong or null alleles(Brummel et al., 1994; Nellen et al., 1994; Xie et al., 1994; Letsou et al., 1995; Ruberte et al., 1995; Das et al., 1998; Wisotzkey et al., 1998). Interestingly, even though sax⁴ is a strong or null sax allele (Brummel et al.,1994), 2 weeks after the clone induction, 6.3% of the testes carried sax mutant GSCs, indicating that sax plays a less important role in maintaining GSCs than tkv, the other type I receptor. Previous studies suggest that gbb preferentially uses sax to transduce its signal, whereas dpp prefers tkv for its signal transduction(Haerry et al., 1998). Our results argue that gbb preferentially uses tkv instead of sax to transduce its signal in male GSCs. Therefore, we conclude that Bmp signals directly act on GSCs and control their maintenance in the Drosophila testis.

In the Drosophila ovary, overexpression of dpp causes accumulation of GSC-like cells, which contain a spectrosome and do not express the cytoplasmic form of Bam protein (BamC)(Xie and Spradling, 1998). In the ovary, bam expression in GSCs is normally transcriptionally repressed, but forced bam expression in GSCs leads them to differentiate (McKearin and Spradling,1990; Ohlstein and McKearin,1997; Chen and McKearin,2003a), suggesting the possibility that one function of dpp is to prevent bam gene expression in GSCs. In wild-type testes, BamC is present in the cytoplasm of two-cell to early 16-cell germline cysts (Kiger et al., 2000). This also raises a similar possibility that Bmp signals directly or indirectly repress bam expression in male GSCs.

To determine whether bam transcription is repressed in male GSCs,we used a bam-GFP transgene (a bam promoter fused to the GFP gene) to examine its transcription (Chen and McKearin, 2003a). Interestingly, bam was transcribed predominantly in the differentiated germ cells but not in GSCs and gonialblasts in the Drosophila testis(Fig. 5A), which is consistent with the BamC expression pattern. Our result that Dad is expressed only in GSCs and gonialblasts supports the idea that Bmp signaling suppresses bam expression in GSCs and gonialblasts. If bam repression in GSCs is mediated by Bmp signaling, we would predict that bamexpression in GSCs defective for Bmp signaling would be upregulated. To test this hypothesis, we generated dpp, gbb or punt homozygous mutant males that also carried the bam-GFP transgene to monitor bam expression. As predicted, bam-GFP was not obviously upregulated in dpp^hr56/dpp^hr4 mutant GSCs just like in wild-type ones (data not shown), consistent with the fact that the dpp mutations have little effect on the maintenance of male GSCs. Interestingly, bam-GFP expression was elevated in the gbb⁴/gbb^D4 or gbb⁴/gbb^D20 mutant GSCs(Fig. 5B), indicating that gbb signaling is essential for repressing bam transcription in GSCs. Furthermore, at 22°C, bam expression was undetectable in the punt mutant GSCs, but it was elevated in punt mutant GSCs at 29°C (Fig. 5C). To further confirm this observation, we generated marked Med mutant GSCs that carried the bam-GFP transgene. Consistently, 66% of 3-day-old marked lacZ-negative Med mutant GSCs expressed bam-GFP, while neighboring unmarked lacZ-positive wild-type GSCs failed to express it (Fig. 5D). These results demonstrate that Bmp signaling is required to suppress bam transcription in GSCs in the Drosophilatestis.

To further investigate whether forced bam expression causes GSC loss in males, we used two different means to ectopically express bamin male GSCs: hs-bam (the bam gene under the control of a hsp70 promoter) and nanos-gal4VP16 driven UASp-bamGFP expression (Ohlstein and McKearin, 1997; Chen and McKearin, 2003a). Two days after heat-shock treatments (4 hours per day for 3 days), GSCs in all the testes without hs-bam remained normal with an average of 9.3 GSCs per testis (n=26), and a week later they retained 7.6 GSCs per testis (n=30). By contrast, 2 days after heat-shock treatments, GSCs in most of the testes carrying hs-bam was reduced to an average of 3.9 GSCs per testis(n=31) (Fig. 5E),whereas 1 week later, about 70% of the testes completely lost their GSCs with an average of 2.0 GSCs (n=33)(Fig. 5F). Similarly, nanos-gal4-driven germ cell-specific bam expression caused GSC loss in testes before adulthood, whereas forced bam expression in somatic cyst cells had no effect on GSC maintenance (data not shown),supporting the idea that bam overexpression triggers GSC loss in a germ cell-specific manner. The UASp-bam we used here has been used previously to induce GSC differentiation in the Drosophila ovary(Chen and McKearin, 2003a). These results indicate that bam misexpression can cause GSC loss in males like in females, and further suggest that GSC loss caused by defective Bmp signaling could be, at least in part, caused by elevated bamexpression in GSCs.

In the Drosophila ovary, dpp overexpression completely blocks germ cell differentiation, resulting in the formation of GSC-like tumors (Xie and Spradling,1998). To determine whether dpp or gbboverexpression can also prevent germ cell differentiation in the testis, we overexpressed dpp or gbb using the nanos-gal4VP16driver. In the testes overexpressing dpp, the hub appeared to be bigger with more cells, and there were slightly more single germ cells with a spectrosome around the hub cells (Fig. 6A), whereas the testes overexpressing gbb appeared to be normal (Fig. 6B). In the dpp-overexpressing testes, the gonialblasts could still differentiate and divide but failed to stop after four rounds of cell division for normal gonialblasts, resulting in the formation of the spermatogonial clusters with more than 16 germ cells (data not shown). These results suggest that overexpression of either dpp or gbb does not block gonialblast differentiation. These observations suggest that the contribution of dpp signaling to the regulation of the GSC lineage differentiation is different in males and in females. It seems that dppoverexpression directly or indirectly influences hub cell formation during early development as the nanos-gal4VP16 driver is expressed in germ cells during early gonadal development. Extra single germ cells in the dpp-overexpressing testes are probably a consequence of more hub cells, as the bigger hub could potentially produces more Upd molecules, which are known to influence germ cell differentiation.

In the Drosophila male, loss of bam function results in unrestricted proliferation of spermatogonial cells(Gonczy et al., 1997). In females, dpp overexpression completely prevents BamC accumulation in the germ cells (Xie and Spradling,1998). Possibly, the unrestricted proliferation of spermatogonial cells in the dpp overexpressing testis could result from the suppression of bam expression. To test whether hyperactive Bmp signaling can inhibit bam expression in the testis, we used the nanos-gal4VP16 driver to overexpress dpp, an activated tkv (tkv^*) and gbb in the germ cells of testes. Interestingly, all the germ cells including differentiated germ cell clusters in the testes overexpressing dpp failed to express bam-GFP (Fig. 6C). Similarly, overexpression of tkv^* using nos-gal4VP16 and UAS-tkv^* resulted in complete suppression of bam expression in all the germ cells (data not shown). However, in the testis overexpressing gbb, bam-GFP expression was repressed in some two-cell and four-cell germ cell clusters where it is normally expressed in wild-type testes(Fig. 6D). These results indicate that elevated dpp signaling but not gbb signaling is sufficient to inhibit bam transcription in the germ cells of the testis.

To determine the sources for Gbb and Dpp in the testis, we used RT-PCR to study the presence of gbb and dpp mRNAs in the purified hub cells, somatic cyst cells and germ cells using fluorescent-activated cell sorting (FACS). The hub cells were marked by the upd-gal4 driven UAS-GFP expression (Fig. 7A). The somatic cyst cells and somatic stem cells were marked by the c587-gal4-driven UAS-GFP(Fig. 7B). vasa is a germline-specific gene (Lasko and Ashburner, 1988; Hay et al.,1988). The germ cells were marked by a vasa-GFP transgene(Nakamura et al., 2001)(Fig. 7C). The tips of the testes were isolated and dissociated, and the GFP-positive cells were purified from the dissociated testicular cells by FACS. As a control, vasamRNAs were present in the whole testis and isolated germ cells but were absent in the somatic cyst cells and hub cells(Fig. 7D). Interestingly, gbb and dpp mRNAs were present in the hub cells and the somatic cysts/somatic stem cells but were absent in the germ cells(Fig. 7D). In addition, dpp mRNAs appeared to be less abundant than gbb mRNAs in the testis. These results indicate that both Dpp and Gbb are probably somatic cell-derived Bmp signals that directly regulate GSC maintenance in the testis.

It has been proposed that the hub cells and the somatic stem cells function as a niche for maintaining GSCs in the Drosophila testis. The Upd signal from the hub cells activates the Jak-Stat signaling pathway in GSCs and controls their maintenance in the testis(Kiger et al., 2001; Tulina and Matunis, 2001). The unknown signal regulated by EGFR and Raf in the somatic cysts also appears to be important for the proper differentiation of gonialblasts in the testis(Kiger et al., 2000; Tran et al., 2000). Here, we propose that the Bmp signals from somatic cells, Dpp and Gbb, are essential for maintaining GSCs in the Drosophila testis(Fig. 8). In this study, we also show that Gbb is essential for keeping bam repressed in GSCs and gonialblasts, and that bam misexpression causes GSC loss in the testis. Moreover, this study reveals similarities between Drosophilamales and females with regards to GSC regulation: both male and female GSCs require Bmp signaling for their maintenance and for repressing bamtranscription in GSCs.

dpp signaling has been shown to be essential for controlling GSC maintenance and division in the ovary (Xie and Spradling, 1998). The Bmp signaling pathway mediated by punt and shn in the somatic cyst cells is known to be important for controlling the proliferation of spermatogonial germ cells(Matunis et al., 1997). However, it is not known whether dpp and gbb are involved in the regulation of GSCs in the testis. In this study, we have provided molecular and genetic evidence that suggests both gbb and dpp are expressed in the somatic cells of the testis and work cooperatively to maintain GSCs and to repress bam transcription in GSCs.

In the gbb mutant testis, GSCs are lost very rapidly but gonialblasts still develop into 16-cell cysts, suggesting that gbbfunctions specifically to control GSC maintenance during germ cell development in the testis. Surprisingly, mutations in dpp have very little effect on GSC maintenance, which is in contrast to the role of dpp in the ovary (Xie and Spradling,1998). However, a mutation in one copy of the gbb gene greatly enhances the stem cell loss phenotype of dpp mutants even though heterozygous gbb males have normal GSC number, indicating that dpp and gbb work cooperatively to control GSC maintenance. In the Drosophila testis, dpp plays a less important role than does gbb with regards to GSC regulation, which could be due to much lower dpp expression. In the Drosophila ovary, both dpp and gbb are equally important for maintaining GSCs and repressing bam transcription in GSCs(Xie and Spradling, 1998; Song et al., 2004). Although dpp overexpression in the ovary completely blocks cystoblast differentiation and causes the accumulation of GSC-like germ cells(Xie and Spradling, 1998),overexpression of either dpp or gbb has little effect on differentiation of gonialblasts in the testis. These observations suggest that Bmp signaling is essential for maintaining GSCs in both sexes but gbband dpp contribute differently.

Even though gbb has been shown to work synergistically with dpp potentially through the use of common Bmp receptors in patterning wing imaginal discs (Khalsa et al.,1998), it is not known whether gbb signaling directly contributes to the production of pMad. This study suggests that gbbsignals through previously defined dpp receptors to regulate the phosphorylation of Mad. We show that pMad in gbb mutant GSCs is severely reduced just like in punt mutant GSCs. Dad has been established as a dpp-responsive gene in other developmental processes(Tsuneizumi et al., 1997). In this study, we show that Dad-lacZ expression in GSCs and gonialblasts is beyond detection in the gbb mutant testis. Interestingly, partial removal of Dad function can also partially suppress the stem cell loss phenotype of gbb mutants, suggesting that Dadnegatively regulates gbb signaling. However, Dad-lacZexpression is only slightly reduced in dpp mutant GSCs and gonialblasts. These results indicate that Dad is primarily a gbb responsive gene in the Drosophila testis. They also argue that gbb indeed signals through common dpp receptors,promotes Mad phosphorylation and activates Dad transcription in GSCs in the same way as dpp does.

Dpp can function as a long-range gradient, which elicits different responses at different concentrations (reviewed by Podos and Fugerson, 1999). In the tip of the Drosophila testis, Gbb and Dpp appear to function as short-range signals and their signaling activities are restricted to GSCs and gonialblasts based on expression of Dad-lacZ and pMad. gbband dpp mRNAs appear to be expressed in both the hub cells and the somatic cyst cells. In the ovary, Bmp signals also appear to function as short-range signals (Kai and Spradling,2003; Song et al.,2004). Gbb and Dpp must be produced and/or activated around the hub cells and the somatic stem cells in order for them to signal locally to GSCs and gonialblasts. It would be very interesting to see whether Gbb and Dpp are localized and/or activated around the hub cells.

Stem cells must remain undifferentiated and continue self-renewal at every cell division. The relationship between the undifferentiated state and self-renewal remains to be defined. Even though several signals have been identified for stem cells in different systems, there is little known about their direct target genes in stem cells, which could help us to understand how these signals are translated into the self-renewal or undifferentiated state of GSCs. In order for a stem cell to maintain its identity, it at least requires the repression of the genes that are important for differentiation of stem cell daughters. In this study, we show that Bmp signals from the niche cells are involved in repressing bam transcription in GSCs in the testis.

In the present study, we demonstrate that Bmp signaling mediated by Dpp and Gbb is essential for maintaining GSCs in the testis. bam is known to be both necessary and sufficient for the differentiation of a cystoblast in the Drosophila ovary (McKearin and Spradling, 1990; Ohlstein and McKearin, 1997). In the bam mutant testis, GSCs are well maintained, and gonialblasts still differentiate but overproliferate into clusters with more than 16 germ cells(Gonczy et al., 1997),suggesting that bam is not necessary for the initial differentiation of gonialblasts. In this study, we show that forced expression of bamin GSCs causes them to be lost, which may be due to differentiation and/or cell death. Normally, bam transcription is absent in GSCs, suggesting that an active mechanism exists to repress bam expression in GSCs. The mechanism appears to be mediated by Bmp signals that originate from the surrounding somatic cells - the niche cells. In the testis, the GSCs mutant for gbb, punt and Med have elevated bamtranscription. dpp overexpression leads to bamtranscriptional repression in all the germ cells of the testis. These results demonstrate that Bmp signaling is essential for keeping bam repressed in GSCs. In the Drosophila ovary, Bmp signaling appears to directly repress bam expression in GSCs(Chen and McKearin, 2003b; Song et al., 2004). Whether the repression of bam transcription in GSCs mediated by Bmp signaling in the testis is direct remains to be determined. This study indicates that niche signals maintain the undifferentiated or self-renewal state of stem cells, at least in part, by repressing the expression of the genes that are important for the differentiation of their progeny.

Upd is another known signal for GSCs in the Drosophila testis, and activates the Jak-Stat signal transduction cascade in GSCs to maintain their stem cell identity (Kiger et al.,2001; Tulina and Matunis,2001). upd overexpression disrupts normal differentiation of gonialblasts, leading to the accumulation of stem cell-like germ cells in the testis. As both Jak-Stat and Bmp signaling pathways are required in GSCs for their maintenance, they must be integrated and interpreted collectively in GSCs. There are several possible ways both signaling transduction pathways could interact with each other. First, Jak-Stat and Bmp signaling pathways regulate each other in GSCs. In the mammalian system, Bmps can regulate Stat function by controlling the activity of the FRAP serine-threonine kinase in neural stem cells (Rajan et al.,2003). This may also happen in the GSCs of the testis. It is possible that Jak-Stat signaling regulates Bmp signaling through a yet unidentified mechanism. Second, both signaling pathways activate their own transcription factors, which together activate the expression of the genes that are important for maintaining GSCs in the undifferentiated or self-renewal state. Third, both signaling pathways could activate expression of different genes that are important for maintaining GSCs while repressing different genes that cause GSC differentiation. These different scenarios await to be investigated in the future.

We thank D. Drummond-Barbosa, S. Cohen, D. Harrison, M. Hoffmann, S. Kobayashi, D. McKearin, A. Spradling, P. ten Dijke, K. Wharton and the Drosophila stock center for reagents. We also thank the Xie laboratory members for critical comments on manuscripts; J. Haynes for help with the manuscript preparation; and K. Zueckert-Gaudenz, J. Haug, C. Sonnenbrot and X. Song for technical support. This work is supported by Stowers Institute for Medical Research.