|Home | About | Journals | Submit | Contact Us | Français|
The ability of adult stem cells to maintain their undifferentiated state depends upon residence in their niche. While simple models of a single self-renewal signal are attractive, niche-stem cell interactions are likely to be more complex. Many niches have multiple cell types, and the Drosophila testis is one such complex niche with two stem cell types, germline stem cells (GSCs) and somatic cyst progenitor cells (CPCs). These stem cells require chemokine activation of Jak/STAT signaling for self-renewal. We identified the transcriptional repressor Zfh-1 as a presumptive somatic target of Jak/STAT signaling, demonstrating that it is necessary and sufficient to maintain CPCs. Surprisingly, sustained zfh-1 expression or intrinsic STAT activation in somatic cells caused neighboring germ cells to self-renew outside their niche. In contrast, germline-intrinsic STAT activation was insufficient for GSC renewal. This data reveals unexpected complexity in cell interactions in the niche, implicating CPCs in GSC self-renewal.
Adult stem cells contribute a steady source of new cells to maintain tissues of many types. The potential to use these stem cells in regenerative medicine, however, is hampered by a lack of knowledge in how they are normally regulated within their niches. Study of Drosophila male and female gonads has greatly increased our general knowledge about how niches regulate stem cells, contributing such ideas as localized domains competent for self-renewal, mechanisms of asymmetric stem cell divisions, and the ability of differentiating cells to “de-differentiate” (reviewed in Fuller and Spradling, 2007) While these advances have provided useful paradigms for understanding niches in higher organisms, many niches remain poorly understood because they are inherently more complex, containing multiple cell types. Understanding how multiple cell types interact to create complex niche environments in which stem cell populations can function is a largely unexplored frontier.
The Drosophila testis niche supports two stem cell populations, and as such it has potential to provide insights into how complex niches function. Each germline stem cell (GSC) division is accompanied by divisions of two cyst progenitor cells (CPCs). The differentiating daughters of the CPC division are called cyst cells, and they invest themselves around the differentiating daughter of the GSC, called the gonialblast. The gonialblast undergoes transit amplifying (TA) divisions, and then differentiates, all the while surrounded by these two cyst cells. Since differentiation of the germ cells is clearly dependent upon crosstalk with the cyst cells at several different stages (Fabrizio et al., 2003; Kiger et al., 2000; Matunis et al., 1997; Tran et al., 2000), it is essential that GSCs and CPCs are both maintained in the niche, and that there is close to a 1:2 ratio of GSCs to CPCs. Mechanisms regulating this balance are unknown. There are hints that the Drosophila ovarian niche may achieve this balance via GSC dependence on a signal from the accompanying somatic stem cell population (Decotto and Spradling, 2005).
In testes, both the CPCs and GSCs cluster around a group of nondividing somatic cells called the hub (Hardy et al., 1979). The hub is the source of the self-renewal signal Unpaired (Upd) (Figure 1B), a ligand that activates Jak/STAT signaling in surrounding cells—both somatic and germ cells. Upd availability is restricted such that only cells near the hub significantly activate Jak/STAT signaling and thereby adopt stem cell fate (Kiger et al., 2001; Tulina and Matunis, 2001). Daughters that are displaced from the hub via oriented cell divisions move out of the signal's influence and differentiate (Yamashita et al., 2003). STAT is intrinsically required in GSCs for their self-renewal, and is also thought to be required for CPC self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001). Nevertheless, the molecular mechanism of self-renewal for either lineage is poorly understood in part because no targets of STAT function in the stem cells have been identified.
Upd misexpression causes all daughters of GSC divisions to retain stem cell fate (Kiger et al., 2001; Tulina and Matunis, 2001); it has therefore been assumed that direct activation of STAT signaling in germ cells by Upd is sufficient to specify self-renewal. Upd misexpression, however, also generates excess CPCs, which embrace the germ cells(Tulina and Matunis, 2001). Hence, it is unclear whether each lineage independently self-renews in response to ectopic Upd, or whether renewal of one lineage relies on the other, presumably via GSC-CPC crosstalk.
As a way to identify targets of STAT activation, whether direct or indirect, and new regulatory circuits operative in this complex niche, we undertook transcriptional profiling experiments. By comparing testes with ectopic upd expression to normal testes, we identified RNAs predicted to be enriched in the excess GSCs and CPCs (Terry et al., 2006). One such gene was zfh-1, encoding a transcriptional repressor with multiple zinc fingers and a homeodomain (Fortini et al., 1991). zfh-1 is required for numerous developmental fate decisions during embryogenesis, including specification of the somatic gonadal precursors (Broihier et al., 1998). Of the vertebrate homologs, δEF1/zfhx1a and SIP1/zfhx1b, zfhx1b is thought to be the zfh-1 ortholog (Liu et al., 2006), and in heterozygous state is the cause of human Mowat-Wilson syndrome, a form of Hirschsprung disease associated with severe mental retardation (Zweier et al., 2002). Here we examined the role of zfh-1 in the Drosophila testis niche. We found that zfh-1 is activated by STAT signaling, and it plays a central role in the maintenance of undifferentiated CPC fate via its role as a transcriptional repressor. Our work also revealed an unsuspected role for CPCs assisting in GSC self-renewal.
In various tissues, it has been shown that signaling by Upd leads to activation and accumulation of STAT protein (Chen et al., 2002). In the testis, STAT accumulation is restricted mostly to germline and somatic cells next to the hub (Figure 1A), consistent with hub cells being the source of Upd (Figure 1B) (Kiger et al., 2001; Tulina and Matunis, 2001). As a consequence, the germ cells next to the hub take on GSC fate, and stat mutant GSCs differentiate (Kiger et al., 2001; Tulina and Matunis, 2001). While it has been assumed that CPCs also intrinsically require STAT activation, this has not been tested directly. We found that STAT is indeed intrinsically required in CPCs, either for survival or self-renewal, as stat null CPC clones were rapidly lost (Table 1). Similar data has been obtained by C. Brawley, M. de Cuevas and E. Matunis (personal communication). The fact that stat was essential in the somatic lineage and that its accumulation was restricted to cells near the hub suggested that key self-renewal genes would be similarly restricted in their expression, and Zfh-1, identified in our expression profiling experiments (Terry et al., 2006), met this criteria.
Prior lineage-tracing experiments demonstrated that the CPCs are among the first tier of somatic cells surrounding the hub (Gonczy and DiNardo, 1996). We found that Zfh-1 accumulated to high levels in these cells (Figure 1C, green), as shown by double staining with Traffic jam (Tj), a pan-somatic cell marker (Li et al., 2003). In newly formed cyst cell daughters, that is, those associated with a gonialblast, Zfh-1 levels decayed to about half that observed in CPCs (Figure 1C′, arrows; Table S1). In older cyst cells, those associated with TA spermatogonia, Zfh-1 levels dropped to four- to five-fold below that in CPCs (Figure 1C′, arrowheads; Table S1). The enriched expression of Zfh-1 in CPCs, and its decay in daughter cells that move away from the influence of Upd, is consistent with Zfh-1 being a genetic target of STAT activation in CPCs. We indeed found that persistent STAT activation led to accumulation of high levels of Zfh-1 in all somatic cells (Figure 1D; see also Figure 6D). However, Zfh-1 decreased only modestly upon STAT inactivation; thus there may also be other inputs to Zfh-1 expression (see Supplemental Data).
We tested whether zfh-1 is required in CPCs by generating zfh-1 mutant clones, counting the proportion of testes with at least one remaining mutant stem cell 2, 4, and 8 days later. We found that zfh-1 mutant CPCs were recovered at negligible levels compared to control clones by even two days after clone induction (Table 1). zfh-1 mutant clones were recovered efficiently in GSCs, where zfh-1 is not expressed, at each time point (Table 1). Thus, zfh-1 is required in CPCs.
In order to determine the fate of zfh-1 mutant CPCs, we examined testes one day after clone induction, since most zfh-1 mutant CPCs were lost by two days. We directly identified mutant CPCs by loss of Zfh-1 protein expression (Figure 1E, arrow). As expected, in control testes, all CPCs expressed Zfh-1 (0/10 testes had a Zfh-1-negative CPC). In contrast, experimental testes frequently exhibited one or more CPCs that were Zfh-1-negative (12/24 testes). To test whether the zfh-1 mutant CPCs failed to self-renew, and instead underwent differentiation, we examined expression of the differentiation gene eyes absent (eya). Eya is normally undetectable in CPCs or their immediate daughter cyst cells associated with gonialblasts (Figure 1F, arrowheads). Eya accumulates at low levels in cyst cells accompanying groups of TA spermatogonia (Figure 1F, arrows), and at high levels in cyst cells accompanying spermatocytes (not shown) (Fabrizio et al., 2003). We identified mutant CPCs among first-tier Tj-positive somatic cells by lack of Zfh-1 accumulation (Figure 1G, arrows). Most such cells (15/18 scored) now precociously expressed low levels of Eya (Figure 1G, arrows, Figure 1G′ shows Eya only), while all other non-mutant (Zfh-1-expressing) CPCs exhibited negligible Eya accumulation (Figure 1G and 1G′, arrowheads). This indicated that zfh-1 loss in a CPC caused the stem cell to initiate its differentiation program; thus zfh-1 is required for CPC self-renewal. While zfh-1 mutant CPCs clearly start differentiating, we do not know their eventual fate (see Supplemental Information for additional analysis and discussion).
We observed that Zfh-1 protein accumulation decreased in the cyst cell daughter of the CPC (Figure 1C). Since zfh-1-deficient CPCs differentiated, we wondered whether the natural decay of Zfh1 from daughter cyst cells was essential for their differentiation. To test this, we used the GAL4 UAS system to artificially maintain zfh-1 expression, employing the cyst cell driver EyaA3 GAL4. This driver contains an eya regulatory element that causes GAL4 to be expressed precociously compared to normal eya gene expression. It drives low levels of expression in CPCs (Figure 2A, arrowheads), and high levels of expression in cyst cells (Figure 2A, arrows). To circumvent the lethality observed due to Zfh-1 misexpression in other tissues earlier in development, we restricted GAL4 induction to adults by including a temperature-sensitive allele of GAL80, the GAL4 inhibitor (McGuire et al., 2004). Thus, only when adults were up-shifted, would Zfh-1 expression be sustained in CPC daughters.
Normally, histologically undifferentiated cells are restricted to the apex of the testis; these cells fluoresce brightly with DNA stains (Figure 2B, bracket). By contrast, after ten or more days of sustained zfh-1 expression, the entire testes became filled with undifferentiated cells (Figure 2C), many of which were Tj-positive somatic cells (Figure 2D, green). Surprisingly, Vasa staining revealed that excess small, early-stage germ cells were usually intermingled with the excess somatic cells (Figure 2D, red). The excess germ cells were primarily found together in pairs, although individual cells and larger groupings of spermatogonia were also present. Occasionally, excess somatic cells accumulated without accompanying germ cells (not shown). No phenotype was observed upon zfh-1 misexpression in germ cells (data not shown), but we did observe a similar phenotype using a second cyst cell driver, c784 (data not shown)(Hrdlicka et al., 2002).
Zfh-1 and its vertebrate homologues act as active repressors, inhibiting transcription by recruiting the co-repressor C-terminal binding protein (CtBP) (Postigo and Dean, 1999). We sought to determine whether repression was key in generating this gain-of-function phenotype. To test this, we expressed a form of zfh-1 in cyst cells in which the CtBP binding site was mutated (PLDLS > ASASA) to abrogate interaction between CtBP and Zfh-1 (Postigo and Dean, 1999). We found that this mutant could not induce accumulation of excess early-stage cells (Figure 2E), even though the protein was stably expressed in cyst cell nuclei (Figure 2F, arrows). In these testes, the cyst cell differentiation marker Eya was expressed normally (data not shown). Therefore, sustained Zfh-1 exerts its effects by transcriptional repression, likely utilizing the co-repressor CtBP. Of note, we also found that CtBP was required for maintenance of CPCs, as CtBP mutant CPC clones were not recovered (Table 1). This is consistent with an essential role for repression by Zfh-1/CtBP in maintaining CPC fate, though we cannot rule out the possibility that CtBP is required independently of Zfh-1.
Accumulation of excess somatic cells suggested that sustained zfh-1 prevented differentiation into a cyst cell. To test this, we examined expression of the somatic differentiation gene eya (Fabrizio et al., 2003). In testes with sustained zfh-1 expression, most excess somatic cells had low or negligible Eya accumulation (Figure 3A and A′, 3B and B′, arrowheads show Tj-positive, Eya-negative cells), suggesting that zfh-1 prevented somatic differentiation. Consistent with this, Wingless (Wg) protein, which accumulates to a significant degree only in CPCs and early cyst cells near the hub (Figure 3C), accumulated to high levels in cells with sustained Zfh-1 (Figure 3D, D′ shows Wg only). Finally, in wild-type testes, only the CPCs undergo mitosis; daughter cyst cells, upon their birth, immediately withdraw from the cell cycle. In testes expressing sustained zfh-1 we found many examples of dividing somatic cells, even far from the hub (Figure 3E, arrows), suggesting that the excess somatic cells also retained this stem cell characteristic. Thus, sustained zfh-1 prevented differentiation of most somatic cells into mature cyst cells, and these cells retained stem cell characteristics. Taken together, these data suggest that zfh-1 is required to maintain CPCs; and its decay is also essential to promote lineage differentiation.
In this experiment, even though zfh-1 expression was being driven in cyst cells, occasional cyst cells showed little Zfh-1 accumulation (data not shown). These cells were invariably Eya+ and associated with TA spermatogonia (Figure 3A′, arrows; Figure 3B′ shows co-stain with Bam to mark TA spermatogonia, arrows show Eya-positive cells). We speculate this could be due to active degradation of Zfh-1 mRNA or protein initiated by the cyst cell differentiation program. Consistent with this possibility, mammalian zfhx1b was recently shown to be under control of the miR-200 family of microRNAs (Christoffersen et al., 2007).
The non-autonomous effect of somatically-sustained Zfh-1 on germline cells was striking. It is known that defects in germline encystment can lead to proliferation of early-stage germ cells (Sarkar et al., 2007; Schulz et al., 2002). This is not the case here, as the germ cells were intermingled with somatic cells, and E-Cadherin distribution, which normally accumulates on somatic membranes (Figure 4A) (Tazuke et al., 2002), showed that somatic extensions did encyst individual or pairs of germ cells (Figure 4B).
To establish the identity of these germ cells, we first tested whether they differentiated normally by examining expression of Bam, a gene necessary and sufficient for germline differentiation (Gonczy et al., 1997; McKearin and Ohlstein, 1995; Schulz et al., 2004; Song et al., 2004). Bam is normally repressed in GSCs and gonialblasts, and is induced at the two to four-cell spermatogonial stage (Figure 4C) (Brawley and Matunis, 2004; Schulz et al., 2004). In testes with sustained zfh-1 expression, we did observe germ cell groupings that expressed a Bam-GFP reporter (Figure 4D), suggesting some germ cells differentiated. However, many germ cells, including individual and pairs of germ cells, did not express the reporter (Figure 4D). Therefore, sustained somatic zfh-1 expression non-autonomously inhibited germ cell differentiation.
Given this, we used three independent assays to determine whether the germ cells retained stem cell character. First, we examined the expression of the M5-4 lacZ enhancer trap, which is expressed in GSCs and gonialblasts, and occasionally in two cell cysts (Figure 4E) (Gonczy and DiNardo, 1996). Many of the germ cells in testes with sustained zfh-1 expression were positive for M5-4 lacZ (Figure 4F). Second, most germ cells cycled as single cells in locations far removed from the hub, as shown by the M-phase marker phospho-histone H3 (Figure 4G, arrows). This is consistent with GSC or gonialblast identity, and inconsistent with TA spermatogonia, in which germ cells undergo mitosis synchronously. Finally, we examined the appearance of the fusome by adducin staining. In normal testes, the fusome appears spherical in GSCs and gonialblasts, and becomes branched in larger cysts of spermatogonia (not shown). We found that the zfh-1-induced excess pairs of germ cells typically shared a spherical fusome between them (Figure 4H, arrows), again consistent with GSC or gonialblast identity. Such germ cell pairs are similar to GSC-gonialblast pairs, which normally remain associated for an extended period due to a delay in cytokinesis (Hardy et al., 1979). Therefore, sustained zfh-1 expression in somatic cells induced the accumulation of undifferentiated germ cells, and these cells retained stem cell character. The non-autonomous effect of Zfh-1-expressing somatic cells in generating excess GSCs/gonialblasts suggests that a zfh-1-dependent signal from CPCs can cause GSCs to undergo symmetric divisions, such that both daughters of a stem cell division self-renew.
The phenotype we observed with sustained zfh-1 expression was reminiscent of that elicited by misexpression of upd, the ligand that activates the Jak/STAT pathway (Kiger et al., 2001; Tulina and Matunis, 2001). In fact, when we misexpressed upd or zfh-1 under the same experimental conditions (using eyaA3-GAL4 GAL80ts), the resulting phenotypes were indistinguishable (Figure 5A, B, compare to Figure 2C, D). With upd misexpression, we observed STAT protein accumulation in both the excess somatic and germ cells, as expected since the extracellular ligand can signal to both lineages (Figure 5C, C′; arrows show some STAT-positive germ cells; arrowheads, some STAT-positive somatic cells).
One reason for the similar phenotypes elicited by zfh-1 and upd could be that zfh-1 caused induction of ectopic upd expression, which, in turn, would generate ectopically activated STAT in both somatic and germline lineages. However, sustained zfh-1 did not induce upd mRNA expression outside of its normal domain in hub cells (Figure 5D, arrowhead; inset: DNA stain shows this testis has accumulation of excess early-stage cells). A second possibility was that STAT was being activated by a related ligand or another signaling pathway. However, neither the excess somatic (arrowheads) nor germ cells (arrows) away from the hub accumulated STAT (Figure 5E, E′); STAT accumulation was observed solely in its normal domain in and around hub cells (Figure 5E′, bracket; note: there is some spurious STAT staining in the sheath that does not co-label as Tj-positive somatic cells or Vasa-positive germ cells). Therefore, sustained somatic zfh-1 expression does not cause the STAT accumulation that is normally observed with pathway activation, either intrinsically in somatic cells, or in the induced excess germ cells.
The lack of STAT accumulation in the non-autonomously-induced GSCs is a surprising result, indicating that GSCs can self-renew without high levels of STAT activation. While we cannot rule out the possibility that STAT is required only at low levels in the zfh-1-induced excess GSCs, these results led us to question the assumption that activation of STAT in germ cells is sufficient to cause continual GSC self-renewal. To address this, we expressed hopTumL, an activated allele of jak (Hanratty and Dearolf, 1993), intrinsically in germ cells. As expected, STAT accumulated in germ cells away from the hub (Figure 6B, B′, arrows), demonstrating that STAT was indeed activated in cells distant from the normal domain. Surprisingly, however, the gonial region appeared normal with no evidence of increased GSC self-renewal (Figure 6A, B). Thus, intrinsic activation of STAT in germline cells was not sufficient for self-renewal outside the niche.
Since expressing the ligand is sufficient to induce excess GSC self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001), our results now suggest that a key event is actually activation of STAT in the somatic lineage. To test this, we intrinsically expressed hopTumL in somatic cells. This indeed caused the testes to fill with early-stage cells (Figure 6C); these cells included CPCs, as most somatic cells expressed Zfh-1 (Figure 6D, green), as well as GSCs, as most germ cells did not express Bam and cycled as individual cells away from the hub (data not shown). Since hopTumL expression was restricted to somatic cells, ectopic STAT accumulation was pronounced in somatic cells outside of the normal domain (Figure 6E, E′, arrows). Thus, intrinsic activation of STAT in somatic cells was sufficient to provide a non-autonomous, self-renewing signal to neighboring germ cells. This signal likely was induced by Zfh-1, as Zfh-1 accumulated strongly in the excess somatic cells. We did observe some low level STAT accumulation in germ cells in this instance (Fig 6E, E′, arrowheads). This is unlikely to influence our conclusion, as even higher levels of STAT in germ cells (when STAT was intrinsically induced in germ cells, Figure 6B′) did not lead to excess renewal. Thus, in the testis niche, while STAT is essential for GSC self-renewal, there is likely to be an assisting signal delivered by adjacent cyst progenitor cells.
This study revealed two novel aspects of stem cell control. First, we identified a key somatic self-renewal factor, and suggest that this factor maintains stem cell character through transcriptional repression. Second, in this niche which supports two stem cell populations, we found that one stem cell type can influence the self-renewal state of other, providing a potential mechanism for coordinating the two populations.
We found that Zfh-1 function is required to block differentiation of CPCs. zfh-1 may have conserved function as a stem cell factor, since it is expressed in somatic stem cell types in the Drosophila ovary (J.L.L. and S.D., unpublished), and the mammalian homologue zfhx1b is expressed in human ES cells (Boyer et al., 2005). As zfh-1 is also required to specify the CPC lineage (Broihier et al., 1998), it joins a growing list of genes that act within a cell lineage to both to specify the lineage and later to maintain that cell type in the stem cell state (Lang et al., 2005).
In the testis, the enrichment of Zfh-1 in CPCs and the fact that zfh-1 can be somatically induced in response to STAT activation suggested that zfh-1 is a target of the STAT self-renewal signal in CPCs. Consistent with this, Zfh-1 accumulation is reduced when STAT function is compromised (Table S2). While we do suspect there will be other STAT targets in CPCs—two candidates are the CPC-enriched genes SOCS36e and CG2264 (M. Issogonis, N. Tulina, and E. Matunis, personal communication)(Terry et al., 2006)—zfh-1 is likely a key target, since sustained expression of zfh-1 elicits the same phenotype as somatic activation of STAT (the maintenance of stem cell fate in somatic daughter cells). However, we also noted that Zfh-1 did not disappear upon stat inactivation, suggesting that another input(s) to CPCs remains to be identified.
Our data suggest strongly that the key activity of Zfh-1 is transcriptional repression. This is definitively the case from our gain-of-function test, and given the requirement we found for the co-repressor CtBP in the somatic lineage, likely also so for Zfh-1 in CPCs in the niche. This is significant as recent embryonic stem cell studies suggest that a primary function for the self-renewal factors Nanog, Oct4, and Sox2 is to repress master regulators of differentiation pathways (Boyer et al., 2005). Zfh-1 and its mammalian homologue zfhx1b repress various differentiation genes, including mef2 for muscle differentiation (Postigo et al., 1999), brachyury for mesoderm formation (Verschueren et al., 1999), and E-cadherin for epithelial differentiation (Comijn et al., 2001). In the testis niche, eya may be a target of Zfh-1 repression, since we find it is rapidly activated when zfh-1 is removed. Similarly, the Eya transcription factor is an essential differentiation factor in the somatic lineage (Fabrizio et al., 2003). Thus, Zfh-1 shares functional similarity with the ES cell self-renewal factors in blocking differentiation.
Among other potential targets for Zfh-1 in the CPCs is the repression of genes regulated by EGFR/raf signaling. Prior work showed that EGFR signaling is required in somatic cells, and the transcriptional branch of the EGFR pathway culminates in the activation of ETS transcription factors, regulating gene expression in the somatic cells (Kiger et al., 2000; Schulz et al., 2002; Tran et al., 2000). Interestingly, transcriptional repression by members of the Zfh-1 family can be overcome by ETS proteins acting in synergy with the transcription factor c-Myb (Postigo et al., 1997). Perhaps in the testis niche Zfh-1 prevents differentiation by inhibition of ETS targets until an essential ETS synergizing factor is produced, or until Zfh-1 levels simply decay in CPC daughters.
The ability of sustained zfh-1 expression to non-autonomously cause continual GSC self-renewal is a surprising result that questions the idea that germline Jak/STAT activation is instructive for GSC self-renewal (Kiger et al., 2001; Tulina and Matunis, 2001). Our data shows that Jak/STAT activation in the GSCs is not sufficient for GSC renewal outside the niche; rather, Jak/STAT activation in CPCs (with Zfh-1 activation) is necessary for niche independence. Our gain-of-function experiment suggests several possibilities for normal GSC renewal. One is that Zfh-1 normally inhibits a differentiation signal sent from somatic to germ cells, and as a consequence of sustaining Zfh-1, GSCs stay in a “default” stem cell state. This option is supported by the ability of GSCs to proliferate in the absence of EGFR activation in somatic cells (Kiger et al., 2000; Tran et al., 2000). A second possibility is that both STAT activation and a zfh-1-dependent signal delivered by somatic cells are required for GSC renewal. For instance, the zfh-1-dependent signal could activate a transcription factor in germ cells that assisted phosphorylated STAT in GSC gene regulation. In this scenario, a requirement for STAT phosphorylation might be bypassed if enough of the second factor is produced (as when we sustain zfh-1 expression in CPC daughters). Interestingly, mouse ES cells require two signals for self-renewal: LIF activation of STAT3 and BMP activation of SMADs (Ying et al., 2003). A requirement for BMP signaling in testis GSCs has already been demonstrated (Kawase et al., 2004; Shivdasani and Ingham, 2003), and it has been suggested that Jak/STAT signaling leads to production of the BMP ligand Dpp (Wang et al., 2008). However, a BMP ligand could only constitute part of the signal, as overexpression of BMP does not cause the same phenotype as does sustaining zfh1 expression (Kawase et al., 2004; Shivdasani and Ingham, 2003).
A requirement for a second signal assisting in GSC self-renewal has precedent in the Drosophila ovarian niche. There, a BMP was the first signaling pathway discovered to act in GSC self renewal (Xie and Spradling, 1998). However, newly identified somatic escort stem cells (ESCs) surround GSCs, just as CPCs do in the testis. Like CPCs, ESCs intrinsically require STAT, as STAT loss non-autonomously causes GSC loss, demonstrating the requirement for an ESC-dependent signal in GSC renewal (Decotto and Spradling, 2005). In both the female and male niches, if somatic stem cell types are indeed required for GSC renewal, this would provide a means of balancing the two stem cell populations, since GSCs would not be able to survive without CPCs/ESCs, and an overabundance of CPCs/ESCs might increase the available GSC “renewal” signal, causing additional germ cells to become GSCs via dedifferentiation or symmetric stem cell division. Proof of such a requirement for CPCs in the testis will require removing zfh-1 from many or most CPCs; since zfh-1 mutant CPCs are lost so rapidly, we have not been able to follow the fate of neighboring GSCs. Our attempts to knockdown zfh-1 in all CPCs by dsRNA transgenesis have thus far been unsuccessful in achieving significant reduction of Zfh-1 protein.
Our work suggests the existence of a hub to CPC to GSC self-renewal relay signal: Upd, secreted by hub cells, activates Zfh-1 in CPCs, which in turn causes a signal to be sent to the GSCs resulting in their self-renewal. Whether this relay is a required component of the GSC renewal, or simply an amplification of a hub renewal signal, it demonstrates a higher degree of complexity in cell-cell interactions than has been previously found in a stem cell niche. Recent work in other niches increasingly points toward the existence of (and a need for) such complex interactions. For example, in the Drosophila ovary, a feedback loop between stem cells and niche cells has recently been discovered (Song et al., 2007; Ward et al., 2006). Similar to the Drosophila testis, the mammalian hair follicle niche supports two stem cell populations—melanocyte and hair follicle stem cells (Nishimura et al., 2002). The potential for coordination between these two populations has not been explored, and could be relevant to the prevention of melanocyte stem cell loss, which results in hair graying (Nishimura et al., 2005). Neural stem cells were recently found to be much more diverse than expected, and their identity is dependent on their location (Merkle et al., 2007), implying that niche signals for these stem cells must do more than just keep them in an undifferentiated state. Finally, hematopoietic stem cells reside in two distinct niches, associated with either osteoblasts or endothelial cells (Perry and Li, 2007). Both niches require a second cell type, CXCL12-abundant reticular cells, for stem cell maintenance (Sugiyama et al., 2006). How signals from these different cell types interact to coordinate self-renewal is completely unknown. Models based on the self-renewal relay described here will be a starting point for beginning to explore the complex cell interactions in these niches.
MARCM clones were generated from stocks obtained from Gary Struhl and Victor Hatini. zfh-1 mutant alleles used were zfh-175.26 and zfh-165.34 (Ruth Lehmann), both strong EMS-induced alleles without detectable Zfh-1 protein, and they produce the same phenotype in trans with each other or over a deficiency (Broihier et al., 1998). Flies of the genotype yw, hs-flp, Tub Gal4, UAS GFP; FRT 82B zfh-175.26 or 65.34/FRT 82B Gal80 were used for clone induction. Stat85c9 (Erica Bach), Stat06346 and CtBPDE10 (Ken Cadigan), all strong or null alleles, were used for clones. Clones were induced by three 1hr 38° heat shocks separated by 1hr at room temperature. Zfh-1 clones one day after clone induction were visualized by anti-Zfh-1 stain (due to a delay in GFP accumulation, Figure 1E,G); clones analyzed at 2, 4, and 8 days were visualized by GFP stain (Table 1). UAS zfh-1 and UAS zfh-1 CIDm (PLDLS > ASASA mutant) were made by Antonio Postigo and obtained from Bloomington. GAL4 was induced in adults by using GAL80ts (Bloomington), growing flies at 18°, then transferring adult flies to 29°. Flies were kept at 29° for 10-20 days to allow phenotype development. EyaA3 GAL4 is a 5.8Kb EcoRI genomic fragment flanking the start of the type II eya cDNA (RA: FBgn0000320) (Zimmerman et al., 2000) cloned into pH-GAL4 (a gift from S. Barolo and J. Posakony). GAL4 c784 was from Bloomington. Nanos Gal4 VP16 on chromosome II (Erica Selva) was used to drive germline expression. UAS hoptumL was from Norbert Perrimon.
Immunostaining was performed as previously described (Terry et al., 2006). Antibodies were rabbit anti-vasa (Ruth Lehmann, 1:5000), goat anti-vasa (Santa Cruz, 1:400), mouse anti-zfh-1 (Zhi-Chun Lai 1:500), rabbit anti-zfh-1 (Ruth Lehmann, 1:5000), guinea pig anti-traffic jam (Dorothea Godt, 1:10,000), mouse anti-fasciclin 3 (Developmental Studies Hybridoma Bank—DSHB, 1:50), rabbit anti-phosphohistone H3 (Upstate Biotech, 1:500), mouse anti-wingless (DSHB, 1:500), mouse anti-eya (DSHB, 1:20), rat anti-Ecadherin (DSHB, 1:20), mouse anti-IBI (adducin, DSHB, 1:20), mouse anti-βgal (Sigma, 1:1000), rabbit anti-STAT (Erica Bach, 1:1000), rabbit anti-GFP (Molecular Probes, 1:1000), and rabbit anti cleaved caspase-3 (Cell Signaling, 1:200). Secondary antibodies conjugated to A488, Cy3, or Cy5 (Molecular Probes and Jackson Immunologicals) were used at 1:400. Testes were stained 5 minutes with Hoechst 33342 (Sigma) at 1.0 μg/mL.
Synthesis of digoxigenin-labeled upd probe and in situ hybridization was performed as described (Terry et al., 2006), except testes were fixed in 4% formaldehyde with 0.1% NaDOC, and hybridization was at 65°.
Thanks to Nancy Bonini for genomic eya clones. Jim Fabrizio and Matt Wallenfang made the EyaA3 Gal4 line. We thank Zhi-Chun Lai, Ruth Lehmann, Erica Bach, Dorothea Godt, Ken Cadigan, Gary Struhl, Scott Barolo, Jim Posakony, Erica Selva, Victor Hatini, and Norbert Perrimon for antibodies and fly stocks. Thanks to Steve Wasserman, members of the DiNardo lab, and an anonymous reviewer for suggestions on the manuscript. J.L.L. was supported by NRSA postdoctoral fellowship F32 GM074549 from the NIH. Work in S.D.'s laboratory is supported by NIH GM60804.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.