Stem cells are undifferentiated cells thought to exist in many tissues with a remarkable potential for self-renewal or differentiation into other specialized cells which support normal development or maintain tissue homeostasis (Morrison et al., 1997; Watt and Hogan, 2000; Donovan and Gearhart, 2001; Spradling et al., 2001). Much interest in stem cell biology has been inspired after human embryonic stem cells (hESCs) were successfully cultured in vitro and generated differentiated cell types found in different germ layers (Shamblott et al., 1998; Thomson et al., 1998). To date, in vitro and in vivo studies on mammalian stem cells have started to reveal the important nature of stem cells and also have led to recognition of the tremendous potential of these cells for regenerative medicine (Fuchs et al., 2004; Wagers and Weissman, 2004). Furthermore, realization of similarities between stem cells and cancer cells has prompted many researchers to look into the existence of potential stem cells in cancer tissues (Reya et al., 2001). Also, decline in the ability of aging tissues to replenish lost cells could be partly due to loss of stem cells and/or reduction of stem cell activities (Van Zant and Liang, 2003). Therefore, the molecular mechanisms controlling stem cell self-renewal and proliferation are of great interest not only to using stem cells for regenerative medicine but also to understanding cancer biology and aging. These mechanisms are just beginning to be elucidated but are still largely a fascinating mystery. Conceivably, studies on stem cells will benefit patients with diseases such as cancer, Parkinson's disease, Alzheimer's disease, spinal cord injury, diabetes, and liver injury as the molecular mechanisms of their regulation are elucidated.
Two mechanisms have been proposed to control stem cell self-renewal: intrinsic mechanism (asymmetric cell division) and extrinsic mechanism (the niche). In the stem cell literature, stem cell division is often called asymmetric cell division, which is correct based on the fact that two stem cell daughters often have different cell fates. It may be different mechanistically from the asymmetric cell division in Drosophila neurogenesis. In Drosophila neurogenesis, asymmetric cell division is the primary mechanism for generating different cell fates (Roegiers and Jan, 2004). In the central nervous system, a neuroblast divides asymmetrically along the apical–basal axis to produce a larger neuroblast and a smaller ganglion mother cell (GMC), which in turn divides asymmetrically to generate neurons and glia. In the peripheral nervous system, sensory organ precursor (SOP) cells also divide asymmetrically to generate a complete sensory organ. Asymmetric localization of intrinsic cellular cell fate determinants such as Numb and Prospero has been elegantly demonstrated to control the asymmetric cell division in both the central and peripheral nervous systems (Roegiers and Jan, 2004). To date, there are no reports showing that any components needed for asymmetric cell division in Drosophila neurogenesis are also involved in controlling cell fate determination of adult stem cells. Therefore, it remains uncertain whether this kind of asymmetric cell division plays any role in determining cell fates in adult stem cells. However, it has become clear that the fate of adult stem cells is controlled by their surrounding microenvironment, also known as the “niche”. Maybe, progenitor cells with a short-term self-renewal ability, such as Drosophila neuroblasts and SOPs, can reliably use asymmetric localization of cell fate determinants to ensure self-renewal for limited division cycles, and adult stem cells with a long-term self-renewal ability have to rely on instructive signals from niches to guarantee self-renewal division throughout the lifetime.
The niche is often referred to as the physical structure that is made of signaling cells, functioning to maintain stem cell self-renewal and preventing stem cell differentiation (Watt and Hogan, 2000; Spradling et al., 2001; Lin, 2002; Fuchs et al., 2004). The niche concept was first proposed in hematopoietic stem cells (HSCs) over 25 years ago (Schofield, 1978). It was postulated that some extrinsic signals from the niche were important for self-renewal of HSCs. Studies on stem cells in Drosophila have provided novel insights into niche structure and function (Cox et al., 1998; Xie and Spradling, 1998, 2000; King and Lin, 1999; Kiger et al., 2000, 2001; Tran et al., 2000; Tulina and Matunis, 2001; Song et al., 2002). The distal tip cell functions as a niche for controlling germline stem cells (GSC) self-renewal and proliferation in Caenorhabditis elegans (Austin and Kimble, 1987; Henderson et al., 1994). Recently, the niches for the stem cells in the mouse testis, the bone marrow, the skin, the intestinal crypt, and the central nervous system have also been identified (Brinster and Zimmermann, 1994; Calvi et al., 2003; Zhang et al., 2003; Fuchs et al., 2004). Knowledge of niche signals is crucial for expanding stem cells in culture for cell replacement therapy. However, most of the niche signals remain to be identified and the function of many identified niche signals remain poorly understood in mammalian systems. Therefore, studies on stem cells in different systems will not only greatly enhance our understanding of general stem cell biology but will also benefit future regenerative medicine.
Stem cells are inherently difficult to study in vivo in their niche in complex mammalian tissues, because they are rare and often lack molecular markers (Morrison et al., 1997; Spradling et al., 2001). The Drosophila stem cell systems provide excellent models to investigate stem cell biology in vivo at the molecular and cellular level (Kiger and Fuller, 2001; Xie and Spradling, 2001; Lin, 2002). Both GSCs and somatic stem cells (SSCs) are located in the apical tip of the ovary and the testis. The Drosophila ovary and testis are well-characterized structures, and direct physical interactions can be seen between stem cells and the cells forming the niche. Powerful genetic and cell biological tools in combination with easily identified GSCs in the Drosophila testis and ovary have led to much exciting progress in understanding how stem cell self-renewal and differentiation are controlled. In order for readers to better understand stem cells in the Drosophila reproductive systems, we begin by describing ovarian and testicular stem cell systems and methodological approaches to identify and study stem cells and their niche. Then, we briefly summarize what extrinsic signals and intrinsic factors control stem cell self-renewal, differentiation, and proliferation since several comprehensive reviews have been published within the past several years (Lin, 1998; Kiger and Fuller, 2001; Xie and Spradling, 2001; Spradling et al., 2001; Lin, 2002; Gilboa and Lehmann, 2004a). In the end, we will discuss new progress in stem cell plasticity, implications of Drosophila stem cell studies on mammalian systems, and future challenges.
ANATOMY AND PROPERTIES OF STEM CELL NICHES IN THE DROSOPHILA OVARY AND TESTIS
The Drosophila ovary is composed of 14–16 ovarioles, each of which has two or three GSCs and SSCs located in the tip of the ovariole called the germarium, which is divided into four regions: 1, 2a, 2b, and 3 (Spradling, 1993; Xie and Spradling, 2001; Lin, 2002; Fig. 1A). In region 1, cap cells directly contact one of the disc-like terminal filament (TF) cells anteriorly, two or three GSCs posteriorly, and inner germarial sheath (IGS) cells laterally. A GSC with a round fusome (also known as a spectrosome) is anchored anteriorly to cap cells and divides to generate a GSC and a cystoblast. The fusome is a germ cell-specific structure that is rich in membrane skeletal proteins such as α and β spectrin and Hu-li Tai Shao (Hts) (Lin et al., 1994; de Cuevas et al., 1997). The cystoblast also carries a spectrosome and can undergo four rounds of synchronous division with incomplete cytokinesis to form a 16-cell cyst with a branched fusome that connects individual cystocytes. Region 2a contains round-shaped 16-cell cysts that interact extensively with cellular processes of IGS cells (Schultz et al., 2002), whereas region 2b is filled with lens-shaped 16-cell cysts covered by follicle cell progenitors. Two or three SSCs in the 2a/2b boundary are anchored to the posterior IGS cells through DE-cadherin-mediated cell adhesion (Song and Xie, 2002). These SSCs give rise to somatic follicle cells encapsulating the 16-cell cysts and stalk cells connecting two adjacent egg chambers. Lastly, region 3 usually contains a stage 1 egg chamber, which has 15 nurse cells and one oocyte surrounded by differentiated follicle cells. It takes a cystoblast approximately 4 days to exit the germarium and bud off to form a stage 2 egg chamber, which needs approximately another 4 days to develop until stage 14 to form a mature egg (Spradling, 1993). Therefore, GSCs and SSCs need to be regulated in a coordinated manner to efficiently generate egg chambers.
In the ovary, there exist two stem cell niches in the germarium: the GSC niche and the SSC niche. Existing experimental evidence supports that cap cells are the major component of the GSC niche. First, the number of GSCs is closely correlated with the number of cap cells but not with the number of TF cells and IGS cells (Xie and Spradling, 2000). Laser ablation of the TF cells increases the rate of GSC division but does not affect GSC maintenance, suggesting that TF cells affect GSC proliferation but not self-renewal (Lin and Spradling, 1993). Second, GSCs are directly juxtaposed to and are anchored to cap cells by adherens junctions, and disruption of adherens junctions between GSCs and cap cells results in GSC loss (Song et al., 2002). Third, GSC formation is correlated well with the formation of cap cells but not with that of TF cells and IGS cells (Zhu and Xie, 2003). Finally, cap cells express genes that are known to be important for maintaining GSCs, such as dpp, gbb, hh, piwi, and Yb (Cox et al., 1998; Xie and Spradling, 1998, 2000; King and Lin, 1999; Cox et al., 2000; King et al., 2001; Song et al., 2004). Taken together, these findings argue strongly that cap cells are a major component of the GSC niche. On the other hand, the evidence supports that IGS cells and cap cells function as a SSC niche. The SSCs are anchored to the posterior IGS cells through DE-cadherin–mediated cell adhesion, and loss of the adhesion jeopardizes SSC self-renewal, suggesting that the proximal IGS cells are at least a part of the SSC niche for anchoring SSCs (Song and Xie, 2002). Despite that cap cells are a few cells away from SSCs, they express Hh and Wg, two growth factors that are required for controlling SSC maintenance and proliferation, supporting the idea that they are also a part of the SSC niche (Forbes et al., 1996a; King et al., 2001; Zhang and Kalderon, 2001; Song and Xie, 2003). So far, in mammalian systems, stem cell niches have been identified by their direct contact with resident stem cells. It would be more complicated to define stem cell niches if niche cells can provide crucial long-range signals from a distance to control stem cell function.
In the apical tip of the Drosophila testis, two types of stem cells, GSCs and SSCs (also known as cyst progenitor cells), are responsible for producing differentiated germ cells and somatic cyst cells, respectively (Fuller, 1993; Kiger and Fuller, 2001). In the adult testis, seven to nine GSCs can be reliably identified by their attachment to the hub (Hardy et al., 1979; Lindsley and Tokuyasu, 1980; Yamashita et al., 2003). Like GSCs and cystoblasts in the female, male GSCs and their differentiated daughters, gonialblasts, also contain a spectrosome (Hime et al., 1996). Unlike in female GSCs where the spectrosome is located anteriorly, the spectrosome in male GSCs are randomly localized. Male GSCs normally divide asymmetrically giving rise to one stem cell that remains in contact with the hub and one gonialblast that is displaced away from the hub (Hardy et al., 1979; Lindsley and Tokuyasu, 1980; Yamashita et al., 2003). As a GSC divides to produce a gonialblast, the two neighboring SSCs also divide to generate two cyst cells that envelop the gonialblast (Hardy et al., 1979; Gonczy and DiNardo, 1996). Similar to a cystoblast, a gonialblast also undergoes four rounds of synchronous division with incomplete cytokinesis to form a 16-cell germ cell cluster with intermediate stages of two cells, 4 cells and 8 cells, in which there is the branched fusome that connects individual germ cells. After the 16-cell spermatogonial cyst enters premeiotic S-phase, the cyst is now called a spermatocyte, which undergoes cell volume expansion, meiosis, and spermatid differentiation. The two somatic cyst cells surrounding the gonialblast do not divide but continue to grow in size and remain to envelop the underlying germ cells as they undergo mitotic and meiotic divisions and dramatic growth. In the testis, the hub serves as a niche for GSCs, because they directly contact the hub cells. Supporting this idea, unpaired (upd) encodes a secreted growth factor that is expressed in the hub cells and directly acts on the GSCs to control their self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001). In contrast to the current knowledge of the GSC niche, little is known about the SSC niche in the testis. However, based on known structures of stem cells in Drosophila, we can speculate that the hub may also serve as a SSC niche. If the hub is also a SSC niche, both GSCs and SSCs could share a common niche in the testis (Fig. 1B). We can speculate that sharing a niche or a key niche component between two stem cell types in the same tissue or organ may help coordinate their functions. It would be interesting to see whether this is the case for two stem cell types in some mammalian tissues, such as mesenchymal stem cells (MSCs) and HSCs in the bone marrow.
Stem cell niches exhibit a physical asymmetry in the tissue relative to stem cells and their progeny and can exist independently of stem cells. The niche asymmetry allows only one daughter cell to stay in the niche to regenerate a stem cell and the other daughter to move away and differentiate (Fig. 1A,B). For example, a GSC divides to generate one daughter which remains in the niche and retains stem cell identity and the other daughter is displaced away from the niche and differentiates into a cystoblast (in the ovary) or a gonialblast (in the testis). At least, in the mammalian bone marrow and the intestinal crypt, the stem cells stay in the niche and the differentiated daughters move away from it (Roth et al., 1991; Calvi et al., 2003; Zhang et al., 2003; He et al., 2004). The GSC niches in the ovary and testis still maintain their integrity for a period of time in the complete absence of germ cells (Margolis and Spradling, 1995; Gonczy and DiNardo, 1996; Xie and Spradling, 2000; Kai and Spradling, 2003). In the ovary, after the loss of IGS cells caused by the absence of germ cells, SSCs are relocated to the empty GSC niche and proliferate in response to Hh from TF/cap cells, and then the niche degenerates 2 weeks later. However, the GSC niche cannot reprogram incoming SSCs into GSCs (Kai and Spradling, 2003). Similarly, the testis formed from the gonad without germ cells also contains hub cells and SSCs (Gonczy and DiNardo, 1996). The SSCs can also proliferate without germ cells but their progeny often adopt the hub cell fate, suggesting that the GSC niche may also change with time. These findings indicate that the GSC niches are also quasistable without stem cells. These findings also raise similar concern about loss of stem cells and their niches in some diseased tissues. Perhaps, simultaneous transplantation of stem cells and their niches are needed to repair the diseased tissues.
METHODOLOGY OF STUDYING STEM CELLS IN THE DROSOPHILA OVARY AND TESTIS
There are several obvious advantages for studying stem cells in the Drosophila ovary and testis. First, the stem cells can be reliably identified and distinguished from their surrounding cells by molecular markers and/or lineage tracing. The most challenging problem in studying stem cells in many systems, particularly mammalian systems, is that they are rare and difficult to isolate and identify in their natural context (Morrison et al., 1997; Spradling et al., 2001). Drosophila female GSCs can be reliably identified by their direct association with cap cells and their anteriorly located spectrosome (Xie and Spradling, 2001; Lin, 2002; Fig. 2A). LacZ expression with hedgehog (hh)-lacZ and engrailed (en)-lacZ lines can reliably identify TF/cap cells (Forbes et al., 1996a, b; King et al., 2001), whereas the 1444 (lacZ) enhancer trap line identifies cap cells and IGS cells (Margolis and Spradling, 1995; Xie and Spradling, 2000). Even though the ovarian SSCs lack unique molecular markers and distinctive morphology, they can still be identified using lineage tracing (Margolis and Spradling, 1995; Zhang and Kalderon, 2001; Song and Xie, 2002, 2003). Lineage tracing uses FLP (a DNA recombinase) -mediated FRT (FLP recognition target) recombination to mark dividing cells by expression of LacZ or GFP (unmarked cells are LacZ- or GFP-negative). In both the testis and ovary, stem cells and their early progeny are mitotically active and, therefore, can be marked by lineage tracing. The marked stem cell progeny are transient residents and move away from the germarium after 4 days; however, the marked stem cells remain in their niches and continuously produce marked progeny after 4 days. For example, 1 week after lineage labeling, a GFP-labeled SSC in the germarium can be identified by their position (the 2a/2b boundary), low Fasciclin III (Fas 3) expression (vs. high Fas 3 expression in differentiated follicle cells), and the generation of labeled differentiated follicle cells lying posteriorly (Zhang and Kalderon, 2001; Song and Xie, 2002, 2003; Fig. 2B). The lineage tracing technique has also been used successfully to demonstrate that insulin-secreting β cells are maintained by self-duplication but not by adult stem cells in the pancreas (Dor et al., 2004). Male GSCs are also easily identified by the presence of a spectrosome and direct contact with the hub cells and expression of a germ cell-specific marker Vasa (Hay et al., 1988; Lasko and Ashburner, 1988; Kiger et al., 2000; Tran et al., 2000; Fig. 2C). The GSCs and gonialblasts express escargot mRNAs and lacZ enhancer trap lines such as M5-4, M34a, and S1-3 (Kiger et al., 2000, 2001; Tran et al., 2000; Tulina and Matunis, 2001), whereas the 2-cell to early 16-cell spermatogonial cells have branched fusomes (Hime et al., 1996). The hub cells show high Fas 3, DE-cadherin, and Armadillo expression (Yamashita et al., 2003) and can also be identified by the enhancer trap line 254 (Gonczy and DiNardo, 1996). In the testis, SSCs and somatic cyst cells also can be identified by enhancer trap lines such as M12-41 and 1-en-11, whereas the enhancer trap line P-573 only labels mature somatic cyst cells (Gonczy and DiNardo, 1996). The GFP or LacZ-labeled SSCs in the testis using lineage tracing can be identified by their direct contact with hub cells and elongated morphology (Gonczy and DiNardo, 1996; Kawase and Xie, unpublished results; Fig. 2D). Easy identification of the stem cells makes the Drosophila ovary and testis ideal systems for studying stem cell biology in vivo.
The second advantage is that versatile genetic tools in Drosophila make it feasible to manipulate gene function in the stem cells and their niches. The FLP-mediated FRT recombination technique has been used to construct mutant stem cell clones to determine the function of a particular gene in controlling stem cell self-renewal, division, and differentiation (Xu and Rubin, 1993; Chou and Perrimon, 1996; Perrimon, 1998; Xie and Spradling, 1998). For example, the marked mutant stem cell clones homozygous for loss-of-function mutations in the ovary and testis can be identified by loss of arm-lacZ expression; wild-type GSCs and homozygous mutant GSCs can be studied side-by-side in the same testis or ovariole (Xie and Spradling, 1998; Kiger et al., 2001; Tulina and Matunis, 2001). The rates of stem cell loss for wild-type control and mutant stem cell clones can be compared to determine the role of a particular gene in stem cell self-renewal, whereas comparison of numbers of progeny produced by wild-type control and mutant stem cells can help determine whether a gene is required for stimulating stem cell division. The cell-specific gal4 can be used to express UAS constructs for gene overexpression and to express UAS-RNAi constructs for gene knockdown in stem cells and niches. For example, in the germarium, en-gal4 can drive gene expression specifically in TF/cap cells (King et al., 2001), and c587-gal4 can express a UAS construct in IGS cells, SSCs, and follicle progenitors (Song et al., 2004). In the testis, upd-gal4 and c587-gal4 can be used to drive gene expression in the hub and in somatic cyst cells, respectively (Kawase et al., 2004). nos-gal4VP16 can drive gene expression specifically in the male and female germ line, including GSCs (Van Doren et al., 1998).
Third, straightforward genetic screens can be used to identify genes that are important for stem cell self-renewal, division, and differentiation in Drosophila. Mutations affecting the function of stem cells in either ovary or testis lead to sterility. Genetic screens can be carried out to isolate mutants that lose male or female fertility prematurely. Mutations affecting the normal development of GSC and SSC progeny but not stem cells themselves also render females or males sterile, but they can be eliminated from the screens based on the existence of intact stem cells. Overexpression screens can also be carried out using germ cell- or niche-specific drivers and P insertion lines carrying UAS sites at one of the P ends (Rorth, 1996). Such screens have been used successfully to isolate genes that influence GSC maintenance and differentiation in the testis (Schulz et al., 2004). Finally, enhancer and suppressor screens can be applied using sensitized genetic backgrounds. For example, a suppressor screen for piwi has identified many new loci that genetically interact with piwi (Smulders-Srinivasan and Lin, 2003).
Fourth, a genomics approach can be applied to identify intrinsic factors in the stem cells and extrinsic signals from the niche cells. The Drosophila genome has been completed, and the genes identified and/or predicted can be printed onto one slide for whole-genome gene expression analysis (Arbeitman et al., 2004). In the testis, upd-gal4-driven GFP expression in the testis can help purify the hub cells, and c587-gal4-driven GFP expression can be applied to purify SSCs and somatic cyst cells (Kawase et al., 2004). In the ovary, en-gal4-driven GFP expression can be used to purify TF/cap cells, and c587-gal4 can be applied to purify IGS cells, SSCs, and follicle progenitor cells (Song et al., 2004). vas-GFP expression can be used to purify female GSCs from dpp overexpression-induced GSC-like tumors and male GSCs from upd overexpression-induced GSC-like tumors (Kawase et al., 2004; Song et al., 2004). The purified stem cells and niche cells can be used in a genomics approach to identify genetic pathways that operate in them to control stem cell behavior. By comparing gene expression profiles between purified wild-type or mutant stem cells or niche cells, changes in gene expression affected by a particular mutation can be identified to determine how one gene affects stem cell or niche function.
With the available powerful genetic tools and easily identified stem cells in the Drosophila ovary and testis, the function of a particular gene in stem cell regulation can be studied with great precision, and the genes affecting stem cells can be effectively identified through straightforward genetic screens. In addition, many known pathways and their downstream components have been identified and well studied in Drosophila, and their role in stem cell regulation can be directly tested. Therefore, the Drosophila reproductive systems offer a unique opportunity to study two types of stem cells, GSCs and SSCs, at the molecular and cellular level, and to dissect genetic networks controlling stem cell self-renewal and differentiation. Similar genetic tools and strategies can be applied in mice to facilitate studies of stem cell regulation.
SIGNALING PATHWAYS THAT REGULATE SELF-RENEWAL, DIVISION, AND DIFFERENTIATION OF OVARIAN AND TESTICULAR STEM CELLS
BMP, Hh, and Piwi-Mediated Signaling Pathways Are Involved in the Control of Ovarian GSC Self-Renewal
The most well-characterized signaling pathways for controlling GSC maintenance in the Drosophila ovary are the bone morphogenetic protein (BMP) -like pathways (Table 1). In the ovary, dpp and gbb are expressed in cap cells and other somatic cells of the germarium, and dpp and gbb mutant ovaries lose their GSCs prematurely, indicating that both BMPs are required in the surrounding somatic cells for GSC self-renewal (Xie and Spradling, 1998, 2000; Song et al., 2004). GSCs mutant for BMP signal transducers (i.e., punt, tkv, mad, and Med) have shorter life spans and divide slower than wild-type GSCs, suggesting that BMPs directly act on GSCs to control stem cell self-renewal and stimulate their division (Xie and Spradling, 1998). Overexpression of dpp but not gbb completely blocks cystoblast differentiation, resulting in formation of GSC-like tumors, indicating that dpp signaling is sufficient to sustain GSC self-renewal (Xie and Spradling, 1998; Song et al., 2004). Given the previous finding that Gbb and Dpp likely use common signal transducers for their signal transduction (Haerry et al., 1998; Khalsa et al., 1998), it will be important to reveal why dpp but not gbb is sufficient to block germ cell differentiation when overexpressed. The answer to this question will help understand how two BMP molecules contribute differently to GSC self-renewal.
Table 1. Extrinsic Signals and Intrinsic Factors That Are Required for Regulating Stem Cell Function in the Drosophila Reproductive Systema
GSC anchorage in the niche (Song et al., 2002); SSC anchorage in the niche (Song and Xie, 2002).
Possibly, GSC anchorage and spindle orientation (Yamashita et al., 2003).
GSC maintenance, differentiation and survival (Tazuke et al., 2002; Gilboa et al., 2003).
GSC differentiation and survival (Tazuke et al., 2002).
GSC maintenance and cyst differentiation (Lin and Spradling, 1997; Forbes and Lehmann, 1998; Gilboa and Lehmann, 2004; Wang and Lin, 2004).
GSC maintenance and/or survival (Styhler et al., 1998).
Sxl, otu and orb
Cyst differentiation (Bopp et al., 1993; Christerson and McKearin, 1994; Lantz et al., 1994; Rodesch et al., 1997).
Bam and Bgcn
Cystoblast differentiation (McKearin and Spradling, 1990; Lavoie et al., 1999; Ohlstein and McKearin, 1997; Ohlstein et al., 2000).
Spermatogonial cell development (Gonczy et al., 1997).
BMP signaling controls GSC self-renewal by repressing bam expression. bam is necessary and sufficient to cause cystoblast differentiation (McKearin and Spradling, 1990; McKearin and Ohlstein, 1995; Ohlstein and McKearin, 1997). bam is repressed in GSCs through a transcriptional silencer in the bam promoter (Chen and McKearin, 2003b). Dpp overexpression blocks germ cell differentiation by repressing bam expression in the germ cells (Xie and Spradling, 1998; Chen and McKearin, 2003b; Song et al., 2004). Furthermore, remaining GSCs in dpp and gbb mutant ovaries and the marked GSCs mutant for punt and Med up-regulate bam expression (Song et al., 2004). pMad and Dad, two indicators of BMP signaling activities, primarily accumulate in GSCs but diminish rapidly in cystoblasts where bam transcription starts (Chen and McKearin, 2003a; Gilboa et al., 2003; Kai and Spradling, 2003; Song et al., 2004). Moreover, bacterially expressed Mad and Med proteins can bind to the bam silencer in vitro (Chen and Mckearin, 2003a; Song et al., 2004), suggesting that BMP signaling directly represses bam transcription in GSCs. Further experiments are needed to demonstrate that Mad and Med protein complexes bind to the bam silencer in vivo. These findings led to a simple model that BMP signals from the GSC niche directly act on GSCs to cause accumulation of Mad–Med complexes in the nucleus to directly repress bam expression and thereby maintain GSC self-renewal (Fig. 3; Table 1).
Several other important genes for controlling ovarian GSC self-renewal, such as piwi, hh, and Yb, are also expressed in TF/cap cells (Forbes et al., 1996a; Cox et al., 1998; King and Lin, 1999; Table 1). Mutations in piwi and Yb cause rapid depletion of GSCs, whereas Yb and piwi overexpression increases GSC-like or cystoblast-like germ cells (Lin and Spradling, 1997; Cox et al., 1998, 2000; King and Lin, 1999; King et al., 2001). Yb is a novel intracellular protein (King and Lin, 1999), whereas Piwi is the founding member of the piwi family genes containing conserved PAZ and Piwi domains, which can bind to RNAs (Cox et al., 1998). Of interest, Yb regulates expression of piwi and hh in TF/cap cells (King et al., 2001). In addition to the requirement of Yb and piwi functions in the somatic cells for GSC maintenance, Piwi is also expressed in GSCs and regulates GSC division (King et al., 2001). hh overexpression can slightly increase GSC-like or cystoblast-like germ cells, and removal of the hh inhibitory receptor patched (ptc) from GSCs also partially rescues piwi mutant phenotypes, although the GSCs mutant for ptc appear to function normally. This result suggests that Hh signaling may play a redundant role with Piwi-mediated signaling in ensuring GSC self-renewal. As BMP signaling pathways are essential for controlling GSC self-renewal, Piwi somehow works within or with BMP signaling pathways to regulate GSC function (Fig. 3). One possibility is that Piwi is involved in regulating the production, secretion, processing, and/or activation of BMP molecules. Another possibility is that Piwi participates in the control of an unidentified niche signal that is also important for controlling GSC self-renewal.
A study on stet function in the ovary has revealed that a signal(s) from IGS cells is critical for cystoblast differentiation (Schultz et al., 2002). stet encodes a rhomboid-like molecule that functions as a protease to cleave and release one of the Egfr ligands, Spitz (Spi; Urban et al., 2001). Normally, IGS cells send their cellular processes to wrap around underlying differentiated germ cell cysts. In stet mutant germaria, these IGS cellular processes are dramatically reduced and an elevated number of single germ cells with a spectrosome accumulate. Consistent with the idea that stet is involved in activating the Egfr signaling pathway, the active form of MAP kinase is present in IGS cells but is reduced in stet mutant IGS cells (Schultz et al., 2002). Currently, it remains unclear which Egfr ligand regulated by Stet in the germ cells signals directly to IGS cells for controlling the formation of IGS cellular processes and production of an IGS signal(s) that in turn regulates germ cell differentiation.
BMP, Hh, and Wg Signaling Pathways Are Required for Controlling SSC Self-Renewal in the Ovary
gbb is expressed in the somatic cells of the germarium, and mutations in gbb cause severe follicle cell proliferation defects (Kirilly and Xie, unpublished results). Furthermore, SSCs mutant for punt, tkv, mad, and Med are lost rapidly, indicating that BMP signaling is required for controlling SSC self-renewal. Of interest, two major signaling molecules, Hh and Wg, are expressed in TF/cap cells and are required for controlling SSC self-renewal and proliferation (Forbes et al., 1996a; King et al., 2001; Zhang and Kalderon, 2001; Song and Xie, 2003). Mutations in hh and wg appear to slow down follicle cell production (Forbes et al., 1996a; Song and Xie, 2003). hh overexpression or removal of the inhibitory hh receptor, ptc, causes excessive proliferation of SSCs/follicle progenitor cells (Forbes et al., 1996a; King et al., 2001; Zhang and Kalderon, 2001), whereas the SSCs mutant for smo and Ci are lost rapidly and proliferate slowly (Zhang and Kalderon, 2001). Similarly, excessive Wg signaling by removal of two negative regulators, Axin and sgg/zw3, causes excessive proliferation and abnormal differentiation, whereas mutations in dsh and arm accelerate SSC loss (Song and Xie, 2003). These findings indicate that Hh and Wg are major signaling pathways for controlling SSC self-renewal and proliferation. During the Drosophila imaginal development, Dpp, Hh, and Wg regulate each other's expression to control cell proliferation and differentiation (Jiang and Struhl, 1996; Chen and Baker, 1997). In the future, it will be very important to determine how BMP, Hh, and Wg signaling pathways interact with each other and are integrated to control SSC self-renewal and proliferation. Also, Yb is expressed in TF/cap cells and regulates SSC proliferation through modulating hh expression (King et al., 2001). In addition, Yb regulates expression of Piwi to control GSC self-renewal. Therefore, Yb is a master gene in TF/cap cells that coordinates the function of both stem cells to efficiently assemble egg chambers (Fig. 3; Table 1).
JAK-STAT and BMP Signaling Pathways Control Male GSC Self-Renewal
The first identified signaling pathway for male GSCs is the JAK-STAT pathway. Upd from the hub activates the JAK-STAT pathway in GSCs and promotes their self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001). In Drosophila, Hopscotch (Hop), a homologue of JAK, phosphorylates a transcriptional activator, STAT, leading to the activation of its target gene expression. GSCs in hop mutant testes or GSCs mutant for stat are lost rapidly, whereas overexpression of upd prevents proper differentiation of gonialblasts and results in an accumulation of GSC-like or gonialblast-like cells (Kiger et al., 2001; Tulina and Matunis, 2001). Of interest, differentiated germ cells can be reactivated by JAK-STAT signaling to become GSCs (Brawley and Matunis, 2004). These findings demonstrate that Upd from the hub cells plays an instructive role in controlling GSC self-renewal (Fig. 4; Table 1). In the ovary, JAK-STAT signaling is required for controlling follicle cell differentiation and patterning (McGregor et al., 2002; Xi et al., 2003). However, it remains unclear whether JAK-STAT signaling is also required for controlling GSCs in the ovary. STAT3 is required for maintaining long-term self-renewal of mouse ESCs (Niwa et al., 1998; Matsuda et al., 1999). In the future, effort needs to be directed toward identification of STAT target genes in testicular GSCs, which may help understand how JAK-STAT signaling controls stem cell self-renewal from Drosophila to mammals.
Like in the ovary, BMP signaling is also essential for maintaining GSCs in the male (Shivdasani and Ingham, 2003; Kawase et al., 2004; Schulz et al., 2004). Hub cells and somatic cyst cells express gbb at high levels and dpp at much lower levels (Kawase et al., 2004). Consequently, gbb is essential for GSC maintenance, but dpp is not essential and plays a less important role in testicular GSCs. punt mutant GSCs are not lost due to apoptosis, suggesting that BMP signaling controls GSC self-renewal. Similar to ovarian GSCs, GSCs mutant for BMP downstream components such as punt, tkv, mad, and Med are lost faster than wild-type GSCs, supporting that BMPs directly act on GSCs to control their self-renewal (Kawase et al., 2004). Like ovarian GSCs, BMP signaling also represses bam expression in GSCs and gonialblasts. Of interest, the Med mutant GSC loss phenotype can only be partially rescued by a mutation in bam, suggesting that BMP signaling must control GSC self-renewal through repressing bam-dependent and bam-independent differentiation pathways (Kawase and Xie, unpublished results). dpp overexpression fails to completely suppress spermatogonial cell differentiation, which is in contrast with complete suppression of cystoblast differentiation in the ovary. In addition, BMP signaling has another role in restricting spermatogonial cell proliferation (Matunis et al., 1997). Therefore, BMP signaling plays a permissive role in controlling male GSC self-renewal in contrast to an instructive role in controlling female GSC self-renewal (Fig. 4; Table 1). Because BMP and JAK-STAT signaling pathways are required for controlling male GSC self-renewal, they must somehow interact with each other. In mice, BMP signaling suppresses differentiation and sustains ESC self-renewal with STAT3 by inducing expression of Id proteins and antagonizing MAPK signaling (Ying et al., 2003; Qi et al., 2004). BMPs signal through a SMAD or a STAT pathway to regulate cell fate in the stem cells of the mouse central nervous system (Rajan et al., 2003). In the future, the information gained from studies of how BMP and JAK-STAT signaling pathways are integrated in male GSCs will help provide insight into stem cell self-renewal in mammalian systems. In the testis, piwi is expressed in early germ cells, hub cells, and somatic cyst cells of the adult testis, and piwi mutant testes also lose GSCs prematurely (Lin and Spradling, 1997). It remains to be determined whether Piwi regulates testicular GSC maintenance like in the ovary.
An unknown signal emanates from SSCs and somatic cyst cells and plays a critical role in the regulation of gonialblast differentiation. The production of the signal is controlled by the Egfr signaling pathway (Kiger et al., 2000; Tran et al., 2000; Fig. 4; Table 1). Egfr is a tyrosine kinase receptor, which activates the MAPK cascade through regulating Raf and Ras activities (Freeman, 2002). In Egfr and raf mutants, GSC-like or gonialblast-like single germ cells are greatly increased at the testis tip (Kiger et al., 2000; Tran et al., 2000). Genetic mosaic analyses indicate that Egfr and Raf function is required in SSCs/somatic cyst cells but not in germ cells. Consistently, active MAP-kinase is detected in the somatic cyst cells neighboring GSCs. Moreover, the GSC population in the raf or Egfr mutant testis remains active longer than in the wild-type. Lastly, stet is expressed in germ cells and is also required for controlling gonialblast differentiation (Schulz et al., 2002). stet mutant testes carry more spectrosome-containing single germ cells than wild-type. Egfr signaling appears to also have a function in the control of spermatogonial cell proliferation. These observations support the idea that somatic cyst cells are functionally equivalent to IGS cells to control germ cell differentiation. It would be interesting to identify the unknown Egfr ligand in the germ cells that regulates the function of SSCs/cyst cells (Fig. 4; Table 1).
Several signaling pathways, namely BMP, Wg, Hh, JAK-STAT, and Egfr, are required in the Drosophila reproductive systems to control the regulation of GSCs and/or SSCs (Table 1). Of interest, most of their counterparts in mammalian systems have already been shown to control stem cell regulation. BMP signaling directly represses activities of stem cells in the intestine crypt and the hair follicle and controls ES cell and spermatogonial stem cell self-renewal (Zhao et al., 1996; Zhang et al., 2003; Kobielak et al., 2003; Ying et al., 2003; Haramis et al., 2004; He et al., 2004; Qi et al., 2004). Wnt-like signaling pathways have been shown to control self-renewal of HSCs, intestinal stem cells and possibly hair follicle stem cells (Korinek et al., 1998; Reya et al., 2003; Fuchs et al., 2004; He et al., 2004). The JAK-STAT pathway has been shown to control mouse ES cell self-renewal in mice in junction with BMP signaling (Niwa et al., 1998; Matsuda et al., 1999; Ying et al., 2003). Piwi-like genes have been shown to regulate stem cells in plants and C. elegans (Bohmert et al., 1998; Cox et al., 1998; Moussian et al., 1998). Future studies on different signaling pathways controlling different stem cell types will surely reveal more parallels in stem cell regulation in Drosophila and mammals.
INTRACELLULAR MECHANISMS THAT REGULATE OVARIAN AND TESTICULAR STEM CELLS
Two classes of intrinsic factors, self-renewal factors and differentiation-promoting factors, have been identified to have opposite effects on ovarian GSCs. The two best studied ovarian GSC self-renewal factors are Pumilio (Pum) and Nanos (Nos), which are highly expressed in GSCs and cystoblasts. pum and nos mutant ovaries lose GSCs prematurely, indicating the requirement of nos and pum for GSC maintenance (Lin and Spradling, 1997; Forbes and Lehmann, 1998; Bhat, 1999). The functions of Nos and Pum are to prevent GSCs and PGCs (primordial germ cells) from precociously entering the oogenic pathway (Gilboa and Lehmann, 2004b; Wang and Lin, 2004). vasa (vas) is likely required for ovarian GSC self-renewal, because vas mutant germaria contain a few degenerate or growth-arrested germ cells (Styhler et al., 1998). vas encodes a germline-specific translation initiation factor eIF4A (Hay et al., 1988; Lasko and Ashburner, 1988). In the embryo, Pum and Nos work together to repress hunchback translation (Barker et al., 1992). Because Nanos, Pumilio, and Vas are probably involved in maintaining GSCs by regulating translation (Fig. 3), it is important to identify their target genes to gain more insight into GSC regulation. In addition to the requirement of BMP downstream components inside ovarian GSCs for responding to BMP signals (Xie and Spradling, 1998), ovarian GSCs express DE-cadherin and its interacting partner Armadillo (Arm) to form adherens junctions that anchor GSCs to cap cells (Song et al., 2002). The junctions, established during niche formation, are required for maintaining GSCs, because removal of DE-cadherin and Arm function from GSCs causes them to leave the niche and differentiate into cystoblasts. Although the possibility that DE-cadherin–mediated signaling directly controls GSC self-renewal cannot be ruled out, the primary function of DE-cadherin/Arm is likely to keep GSCs juxtaposed to the BMP signaling source cap cells, which are essential for controlling GSC self-renewal (Fig. 3; Table 1).
Several differentiation genes for ovarian GSCs have also been identified, including bam, benign germ cell neoplasia (bgcn), orb, ovarian tumor (otu), and Sex lethal (Sxl). Mutations in bgcn and bam prevent cystoblast differentiation and cause formation of GSC-like or cystoblast-like tumors, indicating that they promote cystoblast differentiation (McKearin and Ohlstein, 1995; Lavoie et al., 1999; Ohlstein et al., 2000). Furthermore, bam and bgcn genetically interact with each other and forced bam expression can drive wild-type GSCs to differentiation but cannot induce bgcn mutant germ cells to differentiate, suggesting that Bam and Bgcn may physically interact with each other to form a protein complex (Lavoie et al., 1999; Ohlstein et al., 2000). bgcn is expressed in GSCs, cystoblasts and early cystocytes, whereas bam is expressed in cystoblast and mitotic cysts. Bam is a novel cytoplasmic protein (McKearin and Ohlstein, 1995), whereas Bgcn belongs to the DExH-box family of RNA-dependent helicases, but lacks critical residues for ATPase and helicase functions, which is probably involved in RNA binding (Ohlstein et al., 2000). Bam functions redundantly with Smurf in cystoblasts and descendents to extinguish BMP signaling activities by blocking BMP signaling downstream of BMP receptor activation (Casanueva and Ferguson, 2004). smurf encodes a ubiquitin ligase, which presumably degrades phosphorylated Mad to restrict BMP signaling (Bonni et al., 2001; Podos et al., 2001; Zhang et al., 2001). Bam interacts with one of the vesicle transport proteins, Ter94, an AAA ATPase, to perhaps regulate membrane trafficking (Leon and McKearin, 1999). These findings suggest that Bam/Bgcn complexes function in multiple pathways to control cystoblast differentiation. In contrast to bam and bgcn mutants, mutations in orb, otu, and Sxl cause the excessive accumulation of GSC-like or cystoblast-like single germ cells mixed with early mitotic germline cysts, indicating that these three genes are required but not essential for cystoblast differentiation. Orb and Sxl, which both can bind mRNAs to regulate their translation or stability, are expressed in GSCs, cystoblasts, and developing cysts (Bopp et al., 1993; Christerson and McKearin, 1994; Lantz et al., 1994). otu encodes two protein isoforms that regulate accumulation of Sxl in germ cells (Bopp et al., 1993; Rodesch et al., 1997). It is likely that these differentiation genes are involved in regulating the stability or translation of mRNAs encoding proteins that are important for cystoblast differentiation. Clearly, identification of Orb and Sxl targets will help better understand germ cell differentiation.
In both the ovary and the testis, Zero population growth (Zpg; innexin-4), a functional homologue of the vertebrate gap junction connexins, is expressed on cytoplasmic membranes of GSCs and differentiated germ cells (Tazuke et al., 2002). zpg mutant ovaries have a few single germ cells with a spectrosome or lack germ cells including GSCs, indicating that it is required for GSC maintenance and differentiation (Tazuke and al., 2002; Gilboa et al., 2003). zpg functions in a parallel pathway with bam to regulate germ cell differentiation, because zpg and zpg; bam double-mutant germaria also contain few undifferentiated single germ cells (Gilboa et al., 2003). Adult zpg-null testes contain only small numbers of early germ cells resembling stem cells or early spermatogonia, indicating that Zpg is required for male GSC or gonialblast differentiation and also survival of early spermatogonial cells (Tazuke et al., 2002). Zpg gap junctions likely allow transfer of unknown factors from somatic support cells to germ cells, which are important for GSC maintenance as well as germ cell differentiation.
In addition to Zpg, the components in JAK-STAT and BMP signaling pathways are also required intrinsically to control testicular GSC self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001; Shivdasani and Ingham, 2003; Kawase et al., 2004). Because STAT, Mad, and Med are essential for GSC self-renewal, both BMP and JAK-STAT signaling pathways must be integrated at or after the level of target gene transcription. The other intrinsic factors that are required for controlling spindle orientation such as DE-cadherin, Centrosomin (Cnn), and APC will be discussed later (Yamashita et al., 2003). As discussed earlier, bam and bgcn play crucial roles in regulating stem cell-cystoblast transition in the ovary. Surprisingly, both of them are dispensable for the stem cell–gonialblast transition in the testis because bam and bgcn mutant gonialblasts differentiate normally (Gonczy et al., 1997). Instead, both bam and bgcn act in the germ line to restrict proliferation of spermatogonial cells. Therefore, Bgcn and Bam proteins have shifted their primary function in stem cell differentiation in the ovary to restriction of spermatogonial proliferation in the testis.
In ovarian SSCs, downstream components of Hh, BMP, and Wg pathways are required for responding to Hh and Wg for controlling self-renewal (Forbes et al., 1996a; Zhang and Kalderon, 2001; Song and Xie, 2003; Kirilly and Xie, unpublished results). Because transcriptional factors responding to these pathways are required in SSCs, the integration of the pathways likely takes place at or after the transcriptional of their target genes. Future effort needs to be directed toward identification of their target genes in SSCs to better understand how these three pathways control SSC self-renewal. DE-cadherin is expressed in the junctions between SSCs and posterior IGS cells and is required for maintaining SSCs (Song and Xie, 2002). Because the Wg signaling pathway and DE-cadherin–mediated cell adhesion require Arm, it will be interesting to know whether Wg signaling and DE-cadherin–mediated cell adhesion modulate each other in controlling SSC maintenance.
Among a few identified intrinsic factors important for GSC or SSC self-renewal, GSCs in Drosophila males and females require gap junction component Zpg and adherens junction components, DE-cadherin and Arm, for their maintenance (Table 1). Although roles of Pum and Nos in controlling testicular GSCs have not been shown in Drosophila, Nos-like genes in mice are required for maintaining spermatogonial stem cells in mice (Tsuda et al., 2003), and a Pum-like protein, FBF, is required for maintaining GSCs in C. elegans (Crittenden et al., 2002). At this point, it is not very clear how many self-renewal factors are shared by Drosophila female and male GSCs. However, it is clear that the mechanisms regulating germ cell differentiation are quite different between the sexes. Bam, Bgcn, Sxl, and Orb are required for cystoblast differentiation but not for gonialblast differentiation. Future studies will reveal how stem cell self-renewal and differentiation programs are shared among stem cells in different systems.
MECHANISMS CONTROLLING ASYMMETRIC DIVISION OF STEM CELLS IN THE DROSOPHILA OVARY AND TESTIS
A stem cell normally undergoes stereotypic asymmetric cell division to generate a self-renewing stem cell that remains in the niche and a differentiated daughter that moves away from the niche in Drosophila as well as in mammalian systems. Conceivably, spindle orientation of the stem cell is responsible for placing one self-renewing daughter in the niche and the daughter destined to differentiate outside the niche. In the ovary, the spectrosome is located on the anterior side of the GSC and is the focal point where one spindle pole is anchored. The stem cell spindle is always oriented in such a way that the anterior daughter remains in the position of the parent GSC and the posterior daughter commits differentiation (Lin and Spradling, 1995). Supporting the idea that the spectrosome plays an instructive role in orienting the stem cell spindle, elimination of spectrosome structures in hts mutant germaria leads to random orientation of the spindle of the dividing GSC (Deng and Lin, 1997). However, although the orientation of the stem cell spindle is randomized, no GSC loss or gain is observed in the hts mutant germaria. Perhaps due to the limited niche size, one of the two GSC daughters has been forced to stay outside the niche; therefore, no extra stem cells are observed. However, a stem cell can generate two stem cells if both daughters are situated in the niche after its division due to loss of one stem cell (Xie and Spradling, 2000).
Spindle orientation has been demonstrated recently to determine asymmetric division of GSCs in the testis. Testicular GSCs always orient their division perpendicular to the hub so that only one of the two daughter cells stays in the niche (Yamashita et al., 2003). Cnn, a centrosomal component, is required for normal astral microtubule formation (Li and Kaufman, 1996). In the cnn mutant testis, both centrosome associations with the cortex and orientation of the spindle perpendicular to the hub are disrupted and, consequently, more GSCs are crowded into the niche (Yamashita et al., 2003). The localized DE-cadherin proteins at the cortex of GSCs provide anchorage sites for β-catenin and APC2 (Yamashita et al., 2003). APC2, one of the Drosophila APC homologues, is concentrated around the junction between GSCs and hub cells and is required for controlling correct spindle orientation of GSCs because APC2 mutant testes show similar defects in spindle attachment and orientation as cnn mutants do. Of interest, APC1, the other Drosophila APC homologue, is associated with centrosomes of the GSC, and mutations in APC1 also disrupt GSC spindle orientation. These findings indicate that spindle orientation is important for maintaining stem cell identity in addition to signaling from the niche and that adherens junction-associated APC2 and centrosome components such as APC1 and Cnn are important for controlling spindle orientation (Fig. 4). Because APCs are known to be important for Wnt signaling, it will be interesting to determine whether Wnt signaling is important for orienting stem cell division in the testis. Of interest, spindle orientation regulated by DE-cadherin and APCs also plays an essential role in controlling asymmetric cell division in Drosophila neurogenesis (Roegiers and Jan, 2004). Therefore, it will also be interesting to see whether Cnn and APC proteins are involved in controlling stem cell spindle orientation in the Drosophila ovary and in mammalian systems.
The asymmetry of the niche structure (the physical organization of the niche) in the tissue is established during niche formation and continues to be maintained in adult animals. The best-studied example is the GSC niche in the Drosophila ovary. By the end of the third-instar larval stage, 9–12 disc-like TFs have already formed and cap cells begin to sequentially form at the anterior end of the gonad (King, 1970; Godt and Laski, 1995; Chen et al., 2001; Cohen et al., 2002; Zhu and Xie, 2003). Cofilin, Dwnt4, and focal adhesion kinase are required for the formation of TFs and perhaps cap cells (Chen et al., 2001; Cohen et al., 2002). During the larval–pupal transitional stage, the anterior PGCs that are adjacent to TFs/cap cells become GSCs, while the rest of the PGCs express bam and differentiate into mitotic cysts (Zhu and Xie, 2003; Asaoka and Lin, 2004). dpp is expressed in the TF and cap cells, and directly signals to promote PGC and GSC proliferation (Zhu and Xie, 2003). Sometimes, one PGC can generate two or more GSCs by placing its daughters in the newly formed niche. In the adult ovary, both dpp and gbb are involved in regulating GSC self-renewal and proliferation (Xie and Spradling, 1998; Song et al., 2004). Possibly, gbb is also involved in the regulation of stem cell establishment. However, it remains to be seen whether stem cells in the Drosophila testis use a similar strategy to build their niches.
AMAZING PLASTICITY OF DIFFERENTIATED OVARIAN AND TESTICULAR GERM CELLS IN DROSOPHILA
The stem cell number in a niche is relatively stable, although stem cells have a limited lifespan, suggesting that the lost stem cells must be efficiently repopulated. Indeed, in the ovary, an empty GSC niche space left by a lost GSC can be repopulated by a daughter of the remaining GSCs in the same niche (Xie and Spradling, 2000). Simply, one of the remaining GSCs changes its division orientation to generate two daughters, which both stay in the niche and become GSCs due to the available niche space left by a lost stem cell. For the ovarian SSCs, the empty niche space left by natural or mutation-induced stem cell loss must also be repopulated by the progeny of the other SSCs in the same germarium, but it remains unclear how this is accomplished (Margolis and Spradling, 1995; Zhang and Kalderon, 2001; Song and Xie, 2002). It still remains uncertain whether the niche has the ability to reprogram differentiated cells into stem cells because both the GSC daughters may not be inherently different.
The convincing evidence supporting this idea comes from the observations that differentiated mitotic germ cell cysts can revert back to stem cells in the ovary as well as in the testis (Brawley and Matunis, 2004; Kai and Spradling, 2004). In the adult ovary, overexpression of dpp can also revert cysts back into single germ cells by breaking existing branched fusomes in the cysts and closing out ring canals (Kai and Spradling, 2004). In the second-instar female gonad, which is exposed to a pulse of bam expression, all PGCs prematurely develop into two-cell, four-cell, and eight-cell germline cysts, but the single germ cells re-form later by breaking branched fusomes and sealing ring canals connecting individual germ cells in germline cysts. Of interest, the gonads go on to develop normally and form a normal number of functional GSCs, resulting in normal fertility. Because Dpp signaling activities are high in all the PGCs before the end of the third-instar larval stage, it is likely that Dpp signaling is responsible for reversion of germline cysts into PGCs in the developing gonad. In the Drosophila testis carrying a stat temperature-sensitive mutation, GSCs can be induced to differentiate into spermatogonial germ cell clusters at a restrictive temperature (29°C) (Brawley and Matunis, 2004). After being shifted back into the permissive temperature (18°C), spectrosome-containing single GSCs start to appear again around the hub. The newly formed GSCs are generated by breaking germline cysts that are close to the hub. Of interest, the spermatocytes that finish all mitotic divisions cannot revert back into GSCs even if functional JAK-STAT signaling is regained. In both the ovary and the testis, only early mitotic germline cysts appear to have the capacity to revert back into GSCs if the niche signals are provided. These results demonstrate that differentiated germline cysts can revert back to stem cells and further suggest that niche signals maintain GSCs possibly by two distinct mechanisms: one is to prevent GSCs from differentiation, and the other is to reprogram accidentally differentiated germ cells back into GSCs to maintain stem cell homeostasis. It has also been proposed that the breaking up of early spermatogonial clusters represents an emergency way to increase the number of stem cells in mammalian testes (de Rooij, 2001), but it has not been determined whether single cells pinched off from the clusters are functional stem cells. If early spermatogonial cells can revert back to functional stem cells, the well-established stem cell transplantation assay in mammalian testes may not effectively determine stem cell number. Because these observations are made on GSCs in Drosophila and mice, it will be fundamentally important to investigate whether early differentiated cells can be reprogrammed back into stem cells by niche signals in other stem cell systems.
CONCLUSIONS AND FUTURE DIRECTIONS
Stem cells in the Drosophila reproductive systems represent powerful stem cell systems for dissecting niche function and genetic pathways that control stem cell self-renewal, proliferation, and differentiation. With easily identified GSCs and simple tissue structures in the Drosophila ovary and testis, their niches have been well defined, and some fundamental principles about the niche have been learned. For example, the stem cell niche exhibits structural asymmetry in the tissue, in which spindle orientation can ensure a stem cell to undergo asymmetric cell division. Stem cells use cadherin-mediated cell adhesion to stay in the niche to receive multiple instructive signals for controlling their self-renewal. The knowledge gained from the studies of Drosophila stem cell niches has been providing guiding principles for defining stem cell niches in mammalian systems. By using a combination of genetic, molecular and cell biological approaches in Drosophila, several important signaling pathways from the niche have been identified for their ability to control stem cell self-renewal and proliferation. As expected, these signaling pathways are also involved in the regulation of stem cells in mammalian systems. Further studies of these signaling pathways and their downstream target genes in Drosophila will surely provide more insight into the molecular mechanisms governing stem cell self-renewal and differentiation in general. Furthermore, new genetic and molecular screens in Drosophila will identify more genes that play important roles in controlling stem cell function. In this exciting era of stem cell biology, C. elegans and mammalian systems have also made rapid progress in understanding the functions of stem cells and their niches. The knowledge gained from comparative studies in multiple systems will further enhance our ability to define the common mechanisms and strategies governing stem cell self-renewal and differentiation.
Recent studies on stem cells in Drosophila reproductive systems, such as spindle orientation in asymmetric cell division and plasticity of differentiated mitotic germline cysts, have raised many interesting questions for mammalian and other systems. Is cadherin-mediated cell adhesion a general mechanism for anchoring stem cells? Can two or more types of stem cells share a niche in a tissue? How are two or more types of stem cells regulated in a coordinated manner in a tissue? Can some niche cells function to control stem cell self-renewal from a distance? Can differentiated cells be reverted back into stem cells? Is spindle orientation important for stem cell self-renewal? How is spindle orientation of a stem cell controlled? The answers to these questions in mammalian systems will surely provide more insight into stem cell biology in general.
Although we have learned a great deal from studying stem cells and their niche in Drosophila, this is only the beginning of a long journey leading to understanding stem cell biology. Our current understanding and knowledge has further stimulated many more interesting and deeper questions about stem cells in Drosophila reproductive systems, some of which are discussed here. How are niches formed? What are the molecular and genetic mechanisms controlling niche formation? How are mitotic germline cysts reprogrammed back into GSCs? What additional signaling pathways are required for stem cell self-renewal and differentiation? How are multiple signaling pathways integrated in stem cells at the molecular level? What are additional intrinsic factors that control stem cell function? How do intrinsic factors interact with different signaling pathways in controlling stem cell function? How do self-renewal–promoting factors and differentiation-promoting factors strike a balance between self-renewal and differentiation? The answers to some of these questions will provide novel insight into how stem cell self-renewal and differentiation are controlled in general. The Drosophila stem cell systems will continue to make more exciting contributions to stem cell biology for a long time to come.
We thank the Xie laboratory members and three anonymous reviewers for critical comments and suggestions and J. Haynes for help with the manuscript preparation.