Nanos Maintains Germline Stem Cell Self-Renewal by Preventing Differentiation

See allHide authors and affiliations

Science  26 Mar 2004:
Vol. 303, Issue 5666, pp. 2016-2019
DOI: 10.1126/science.1093983


Despite much progress in understanding how extrinsic signaling regulates stem cell self-renewal, little is known about how cell-autonomous gene regulation controls this process. In Drosophila ovaries, germline stem cells (GSCs) divide asymmetrically to produce daughter GSCs and cystoblasts, the latter of which develop into germline cysts. Here, we show that removing the translational repressor Nanos from either GSCs or their precursors, the primordial germ cells (PGCs), causes both cell types to differentiate into germline cysts. Thus, Nanos is essential for both establishing and maintaining GSCs by preventing their precocious entry into oogenesis. These functions are likely achieved by repressing the translation of differentiation factors in PGCs and GSCs.

Stem cells are characterized by their ability to produce daughter stem cells for self-renewal and numerous differentiating daughter cells. GSCs of the Drosophila ovary provide an effective model to understand these characteristics: Genetic tools are available to study their properties, and GSCs are readily distinguished from their differentiating daughters by their position in the ovary and by their morphology (Fig. 1A) (1). Moreover, differentiating daughters of GSCs develop into germline cysts that form a linear array of egg chambers according to their birth order, making it easy to trace GSC divisions. GSCs themselves are derived from PGCs that are initially formed in the posterior region of the embryo. During subsequent embryogenesis, PGCs migrate to the embryonic gonads, where they proliferate without differentiation during larval development. Eventually, they become GSCs at the larval/pupal transition (Fig. 1E) (2).

Fig. 1.

Nos expression during PGC development and early oogenesis. (A) Schematic view of a germarium. Two GSCs are located at the anterior tip in contact with somatic cap cells (CpC) at the base of the terminal filament (TF). A GSC divides asymmetrically to produce a daughter GSC that continues to contact the CpC and a cystoblast (CB) located one cell away from cap cells. The cystoblast divides four times with incomplete cytokinesis to form a 16-cell cyst. Spectrosomes (Sp), spherical organelles containing membrane skeletal proteins, mark GSCs and cystoblasts, whereas fusomes (Fu), spectrosome-derived structures, mark dividing cysts. (B and C) A third-instar ovary stained for Nos-Myc (B, red) and Vasa, a germ-cell marker (C, green). Nos is present in all PGCs. (D) Confocal image of a germarium stained for Nos-Myc (red) and spectrosomes/fusomes (green) (16). Nos is present in all GSCs, much reduced in dividing cysts, present at a high level in newly formed 16-cell cysts, and then down-regulated in older 16-cell cysts. (E) Strategy of hs-nos rescuing experiments. (Top) Key stages of GSC development. (Bottom) 1L, 2L, and 3L denote first-, second-, and third-instar larval stages, respectively.

The Nanos (Nos) family of translational repressors has evolutionarily conserved functions in PGC migration and survival during early embryogenesis (38). During Drosophila embryogenesis, maternally derived Nos complexes with Pumilio (Pum) and binds to the Nos response elements in the 3′ untranslated region of target mRNAs to repress their translation (913). For example, the repression of cyclin B mRNA translation is required to prevent migrating PGCs from undergoing mitosis (12). However, the function of Nos in PGC proliferation, GSC establishment, and oogenesis still remains largely unexplored. In adult ovaries of nos mutants, most germaria display few or no germ cells (14, 15). The number of germ cells decreases with age. It has been hypothesized that nos is required for germline cyst differentiation (14) or for GSC survival (15). However, it is not known whether this adult nos phenotype reflects the direct involvement of nos in oogenesis, an indirect consequence of germline defects that occur before oogenesis, or both. The involvement of Nos in pre-oogenic germline development is suggested by the high level of Nos expression in all PGCs in the third-instar larval ovary (Fig. 1, B and C). In addition, Nos could function in GSC division, because a relatively high level of Nos is present in GSCs in the adult germarium (Fig. 1D). Finally, Nos could also function in subsequent oogenic processes, because Nos is dynamically expressed in dividing cysts and 16-cell cysts (Fig. 1D).

To analyze the requirement of nos for GSC division directly, we constructed a nos transgene driven by the hsp70 promotor (hs-nos) that rescues nos mutant phenotype (nos18/nos53) (fig. S1) (16). Using this transgene, we first tested whether Nos is required during oogenesis by supplying Nos during preadult development and then withdrawing Nos after eclosion (Fig. 1E, bottom, scheme a). We observed a process of GSC depletion following heat-shock withdrawal (Fig. 2, A to D). Two and four days after withdrawal, the number of normal (Class I) (Fig. 2A) ovarioles was sharply reduced from 66% to 36% and 17%, respectively. By day 6, the number dropped to less than 1%, a frequency similar to non–heat-shocked hs-nos;nos18/nos53 control females. In contrast, the number of nonrescued (Class III) (Fig. 2C) ovarioles steadily increased over time, from 9% before the withdrawal to 60% 8 days after withdrawal. This increase is not due to aging, because in a control experiment where we continued heat shock up to 8 days, most ovarioles displayed normal (Class I) morphology, whereas nonrescued (Class III) ovarioles never exceeded 16% (Fig. 2E). This dynamic pattern of GSC depletion indicates that Nos is required for GSC self-renewal during oogenesis.

Fig. 2.

The continuous requirement of Nos for GSC self-renewal. (A to C) Representative Class I to III hs-nos;nos18/nos53 ovarioles (normal, partial rescue, and nonrescued phenotypes, respectively) stained for germ cells (green), spectrosomes/fusomes (red), and the outline of somatic cells (red). Bars, 10 μm. (D) The distribution of Class I to III ovarioles 0 to 8 days after heat-shock withdrawal. Open circles represent Class I ovarioles from hs-nos;nos18/nos53 females without any heat shock. (E) The distribution of Class I to III ovarioles from hs-nos;nos18/nos53 0- to 8-day-old control adult females with heat shock up to day 8.

The number of partially rescued (Class II) (Fig. 2B) ovarioles increased from 25% at day 0 to 52% at day 2 and 62% at day 6, but decreased afterward (Fig. 2, B and D). Thus, these ovarioles likely represent an intermediate state between Class I and Class III. Egg chambers and mature oocytes in Class II and III ovarioles were morphologically normal, suggesting that Nos is not required for post–stem-cell events of oogenesis (Fig. 2, B and C).

In the above experiments, Nos levels were modulated in both somatic and germline cells. To determine whether Nos functions cell-autonomously, we removed Nos only from GSCs. We generated nos mutant GSC clones using the FLP-FRT technique in a dominant female sterile ovoD1 mutant background (16, 17). This transgenic ovoD1 allele functions germline-autonomously to block oogenesis mostly within the germarium, resulting in the development of rudimentary germaria (Fig. 3A). Consequently, egg chambers that develop normally beyond the germarial stage should be derived from nos GSCs that no longer carry ovoD1. Because it takes 7 days for a GSC to develop into an egg chamber (18), and another 3 days to complete oogenesis (19), egg chambers derived from non–self-renewing GSCs should exist between 7 and 10 days after clone induction. By contrast, only self-renewing GSC-derived clones can persist beyond 10 days after induction. The nos mutant egg chambers were present at 8 but not 14 days after induction. Furthermore, at day 8, 85% of the nos mutant clone–containing ovarioles had only one egg chamber (Fig. 3, D and E). This contrasts with the continuous division of wild-type GSC clones that generated a full complement of germ cells in the germarium and multiple egg chambers at day 8 (Fig. 3, B and E). These observations indicate that Nos is required cell-autonomously for GSC self-renewal and that Nos maintains GSCs by preventing their differentiation.

Fig. 3.

nos is required cell-autonomously for GSC self-renewal but not for cyst differentiation. (A) to (D) are DAPI (4′,6′-diamidino-2-phenylindole) images. (A) A typical chromosome 3R transgenic ovoD1 mutant ovary. All ovarioles have a rudimentary germarium (rG) but none has egg chambers. (B) A wild-type GSC clone generated a fully developed germarium (G) and five developing egg chambers numbered according to the birth order. (C) A nos18 clonal ovariole contains only a rG and two morphologically normal egg chambers. (D) A mature nos18 oocyte displays normal morphology. (E) The distribution of the number of egg chambers per ovariole from wild-type (WT) and nos18 (NOS) clones.

Consistent with the conclusion that Nos is not required for post-GSC oogenesis, the germline nos egg chambers displayed no obvious morphological defects (Fig. 3, C and D) (16).

In addition to its function in GSC self-renewal, Nos may be required for pre-oogenic PGC development, because many ovarioles of the newly eclosed nos mutant flies lack germ cells and because there is a high level of Nos expression in wild-type PGCs (Fig. 1, B and C, and fig. S2A). To test this possibility, hs-nos;nos18/nos53 flies were heat shocked starting at various stages of development and continuing to 4-day-old adults (Fig. 1E, bottom, scheme b). The ovaries were then examined for rescue of germline defects by quantifying the number of Class I to III ovarioles (Figs. 2, A to C, and Fig. 4A). These experiments revealed that zygotic Nos is required at two distinct phases during pre-oogenic PGC development. The first phase is from the embryonic stage to the first-instar larval stage, as indicated by a sharp decrease in the rescuing ability of Nos when expressed starting at the first-instar larval stage instead of the embryonic stage. The number of normal ovarioles in the resulting adult females dropped from 79% to 39%. The second phase is from the late third-instar larval stage to the pupal stage that spans GSC establishment. This is reflected by another drop in the rescuing ability when Nos expression was delayed from the third-instar larval stage to the adult stage, with the number of normal ovarioles decreased from 40% to 13%.

Fig. 4.

nos prevents PGC differentiation. (A) Distribution of Class I to III ovarioles in hs-nos;nos18/nos53 females heat shocked according to Fig. 1E, bottom, scheme b. Also shown is the distribution of ovarioles from flies that were never heat shocked (Never) (16). (B) Distribution of single germ cells (presumably PGCs) and 2-, 4-, 8- and 16-cell cysts in wild-type (WT) and hs-nos;nos third-instar larvae heat shocked according to Fig. 1E, bottom, scheme c. (C, D, F, and G) Confocal images of wild-type [(C) and (F)] and nos mutant [(D) and (G)] third-instar larval ovaries stained for germ cells (green), the outline of somatic cells (red), and spectrosomes (Sp)/fusomes (Fu) (red). TF, terminal filaments. (E and H) Drawings tracing germ cells (green) and fusomes (red) in (D) and (G), respectively. PGCs/cystoblasts, light green; cysts, green. The number in a cyst indicates the germ cell number in that cyst (determined by confocal three-dimensional reconstruction). D, degenerating cyst. The bar in (C) = 20 μm [(C) to (E)] and 10 μm [(F) to (H)].

The requirement of Nos for PGC development could be achieved by one of the following two mechanisms: Nos could support PGC proliferation and/or survival, or Nos could prevent PGCs from premature differentiation. To distinguish between these possibilities, we supplied nos18/nos53 mutants with Nos by heat shock from the embryonic to the first-instar larval stage, and analyzed the PGC phenotype in third-instar larval ovaries (Fig. 1E, bottom, scheme c). In hs-nos;nos/+ control ovaries heat shocked the same way, 96% of PGCs proliferated without differentiation, as indicated by the presence of spectrosomes, an intracellular organelle specific for PGCs, GSCs, and cytoblasts (Fig. 4, B, C, and F). Only 4% of two- or four-cell cysts were present, as indicated by the presence of early fusomes, an intercellular organelle present only in germline cysts. These few cysts never contacted the forming terminal filament/cap cells, consistent with a recent report that PGCs contacting the niche cells will receive somatic signals and become GSCs (20). Eight- or 16-cell cysts were never observed. In contrast, in hs-nos;nos18/nos53 third-instar larval ovaries, 73% of PGCs differentiated into germline cysts, with only 27% remaining as single cells (Fig. 4B). All stages of differentiating cysts were readily identified, even though their fusomes appear to be somewhat defective (Fig. 4, D, E, G, and H). Moreover, most germ cells in contact with forming terminal filament/cap cells were in differentiating cysts (Fig. 4, G and H), indicating that PGCs directly differentiated into germline cysts in nos mutants, even in the presence of the stem cell niche.

The premature differentiation of PGCs in nos mutant ovaries appears to occur often as early as the first-instar larval stage. This conclusion is based on three observations. First, there was little difference in the rescuing ability of hs-nos when Nos was provided continuously starting at the first-, second-, or third-instar stages, suggesting that mutant PGCs initiate differentiation at the first-instar stage. Second, it takes 3 days for a GSC to develop into a 16-cell cyst (21). Assuming nos PGCs differentiate at a similar rate, the presence of 16-cell cysts in the third-instar larval stage indicates that their differentiation must start during the first-instar larval stage. Third, the number of germ cells/cysts in the third-instar larval ovaries that had been depleted of Nos at the second-instar larval stage was only about 45, which roughly corresponds to the number of PGCs in the wild-type second-instar larvae derived from 12 PGCs in the embryonic gonad (22). This suggests that, as soon as Nos is depleted from the second-instar larvae, PGCs start to differentiate. Because third-instar larval ovaries have no follicle cell, the differentiated cysts failed to form egg chambers and eventually degenerated.

Because maternal Nos partners with Pumilio (Pum) during embryogenesis (11) and because pum is also required for GSC maintenance (14, 23, 24), we investigated whether the two collaborate in controlling GSC self-renewal. Although pum and nos phenotypes are not identical, their defects are fundamentally similar. In both mutants, most ovarioles were devoid of germ cells (fig. S2, A and B). A small number of ovarioles contained some germ cells, including GSCs (fig. S2, D and E). Normal-looking ovarioles were only occasionally seen (fig. S2, G and H). Furthermore, the phenotype of the nos-pum double mutant was very similar to that of either pum or nos mutants (fig. S2, C, F, and I). These results suggest that nos and pum may function as partners in larval and adult germline development.

In summary, we have demonstrated that Nos is directly and continuously required for both the establishment and the self-renewal of GSCs. Nos does so by preventing PGCs and GSCs from prematurely entering the oogenic pathway. This study indicates that translational regulation maintains the fate of stem cells and their precursors by preventing the precocious activation of a normal differentiation pathway. Nos and Pum may function together in these processes. It will be interesting to see whether the Nos-mediated translational mechanism is regulated by niche signals; such regulation has so far been shown only at the transcriptional level (25, 26). Because Nos is known to have an evolutionarily conserved role in germline development, the novel function of Nos in stem cell division may also be conserved during evolution.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article