Direct Control of Germline Stem Cell Division and Cyst Growth by Neural Insulin in Drosophila

See allHide authors and affiliations

Science  12 Aug 2005:
Vol. 309, Issue 5737, pp. 1071-1073
DOI: 10.1126/science.1111410


Stem cells reside in specialized niches that provide signals required for their maintenance and division. Tissue-extrinsic signals can also modify stem cell activity, although this is poorly understood. Here, we report that neural-derived Drosophila insulin-like peptides (DILPs) directly regulate germline stem cell division rate, demonstrating that signals mediating the ovarian response to nutritional input can modify stem cell activity in a niche-independent manner. We also reveal a crucial direct role of DILPs in controlling germline cyst growth and vitellogenesis.

Germline and somatic stem cells support oogenesis throughout adult life in Drosophila (1) (fig. S1). Germline stem cells (GSCs) reside within a specialized niche where they are exposed to a unique combination of signals required for stem cell function (2, 3). However, GSCs are also controlled by tissue-extrinsic signals, such as Drosophila insulin-like peptides (DILPs), which mediate the ovarian response to nutrition (4). On a protein-rich diet, germline and somatic stem cells have high division rates, and their progeny exhibit high division and development rates. On a protein-poor diet or under reduced insulin signaling, rates of division and development are reduced, and progression through vitellogenesis is blocked (4). It remains unclear, however, how DILPs control the response of GSCs in coordination with their differentiating progeny and with follicle cells.

In adult females, DILPs are produced in two clusters of medial neurosecretory cells in the brain (5), and stage 10 follicle cells express dilp5 mRNA (6). Ablation of brain DILP-producing cells results in reduced egg production rates and a partial block in vitellogenesis (6). To examine the role of the brain DILP-producing cells in previtellogenic stages, we ablated them and measured follicle cell proliferation rates (7) (fig. S2). Females missing brain DILP-producing cells (ablated) have a severely impaired ability to up-regulate follicle cell proliferation in response to a protein-rich diet (Fig. 1). The rate of germline development is reduced in coordination with follicle cell divisions because no abnormalities are observed in previtellogenic egg chambers (8). Ablation of DILP-producing cells reduces the size of eclosed adults and delays development (9). Ablated females in which these developmental defects are rescued by an hs-dilp2 transgene expressed during larval stages show a reduced follicle cell proliferation rate, comparable to that of nonrescued, ablated females (Table 1). Thus, the impaired response to a protein-rich diet is not a secondary consequence of the developmental defects. Moreover, the 2.3-fold delay caused by ablation of brain DILP-producing cells (7) is very similar to that caused by blocking reception of DILP signals by the germ line (below). This indicates that the brain is the major source of DILPs that determine the rate of egg chamber development with little, if any, contribution from dilp5-expressing follicle cells.

Fig. 1.

Brain DILP-producing cells are required for the ovarian response to nutrition. At different times after heat shock, ovaries were dissected and immunostained for visualization of β-galactosidase (β-gal, green) and cell membranes (1B1, red). (A and B) Examples of egg chambers containing follicle cell clones (arrowheads) at 2 days after heat shock in (A) control or (B) ablated females on a rich food source are shown at the same magnification. (C) The number of follicle cells per clone was counted, and the average clone sizes in control or ablated females on rich or poor food sources were plotted. Bars show standard deviation. The responses of control and ablated females to a rich source are significantly different from each other (P < 0.001). Doubling times are shown next to corresponding curves. Scale bar, 10 μm.

Table 1.

The impaired ovarian response to nutrition in females lacking brain DILP-producing cells is not due to developmental defects.

Strain (View inline) Status of DILP-producing cells Mitotic index
Control Present 2.05% (9,080)View inline
Ablated Absent 0.89% (11,631)View inline
Ablated, developmentally rescued Absent 0.84% (12,962)
  • View inline* The total number of follicle cells analyzed is shown in parentheses.

  • View inline The mitotic indices for “ablated” versus “ablated, rescued development” females were not significantly different from each other, whereas these numbers were significantly different from the mitotic index for control females (P < 0.001).

  • To examine whether DILPs control the rate of germline development directly or indirectly, we created germline cysts unable to respond to DILPs, by inducing Drosophila insulin receptor (dinr) mutant clones using the flipase (FLP)/FLP-recognition target (FRT) technique (Fig. 2A). Germline cysts homozygous for dinr339, a genetic null allele (10), had normal morphology and correct cell number (8); however, 83% of these cysts were developmentally delayed, showing markedly decreased size relative to neighboring wild-type egg chambers (7) (Fig. 2B and Table 2). Further quantification of these data showed a 2.4-fold delay in the development of dinr339 cysts (7) (table S1). Similar results were obtained for germline cysts homozygous for dinrE19 and dinr353, which are viable hypomorphic alleles (10); 78% and 64% of dinrE19 and dinr353 cysts, respectively, were developmentally delayed (Fig. 2, C and D, and Table 2). These results reveal that dinr function is required cell autonomously for a normal rate of germline cyst development. Thus, the rate of cyst development is regulated by a DILP signal that is received directly.

    Fig. 2.

    dinr is cell-autonomously required in the germ line for normal rates of GSC division, cyst development, and progression through vitellogenesis. (A) The FLP/FRT technique (18) was used to generate mosaic ovarioles containing dinr homozygous mutant cells in the context of nearby heterozygous tissue. Flies carrying a wild-type dinr allele (+) in trans to a mutant or control dinr allele (dinr*) were heat-shocked (hs) to induce FLP-mediated recombination between the FRT sites. Stem cell–derived dinr homozygous mutant clones were recognized by the absence of β-gal or green fluorescent protein (GFP) markers. (B to D) Homozygous (B) dinr339, (C) dinrE19, and (D) dinr353 germline cysts (arrowheads) are developmentally delayed relative to heterozygous control cysts. (B), (C), and (D) are shown at the same magnification. (E) Ovariole containing a germ line that is a fully dinrE19 homozygous mutant (and surrounded by wild-type follicle cells), showing a degenerating egg chamber (asterisk) that has failed to undergo vitellogenesis. (F) Example of mosaic germaria used to determine the relative division rate of dinrE19 homozygous mutant (mut) and control (wt) GSCs (Table 2). (G) Mosaic ovariole showing similar numbers of homozygous dinrE19 mutant (GFP-negative) and control (GFP-positive) follicle cells. 1B1 antibodies (blue) highlight cell membranes, whereas antibodies to β-gal (red) or GFP (green) mark control cysts. (E) and (G) are shown at the same magnification. Scale bars, 10 μm.

    Table 2.

    DILPs directly regulate the rates of germline cyst development and of GSC division.

    Strain Percentage of marker-negative cysts showing delayed development (View inline) Relative GSC division rateView inline
    Control 0% (96)View inline 0.90 (419)View inline
    dinr 339 83% (118)View inline 0.31 (693)View inline
    dinr E19 77% (92) 0.65 (321)
    dinr 353 64% (123) 0.55 (461)
  • View inline* The total number of marker-negative germline cysts analyzed is shown in parentheses.

  • View inline The percentage of delayed cysts for each of the dinr mutants was statistically different from that of the control (P < 0.001 for each case).

  • View inline Number of marker-negative cystoblasts and cysts divided by number of marker-positive cystoblasts and cysts (7).

  • View inline§ The total number of cystoblasts and cysts counted is shown in parentheses.

  • View inline The relative division rate for each of the dinr mutants was significantly different from that of controls (P < 0.001 for dinr353 and dinr339, and P < 0.04 for dinrE19).

  • Progression through vitellogenesis requires DILP signaling (4, 11); however, it has been unclear whether this role is direct. Reduced juvenile hormone levels are present in homozygous viable dinr mutants, and the block in yolk uptake in these mutants can be partially bypassed by treatment with methoprene, a juvenile hormone analog (12), suggesting an indirect role for DILPs in promoting vitellogenesis. To specifically address whether direct activation of germline cysts by DILPs is required for vitellogenesis, we analyzed mosaic ovarioles in which the entire germ line was homozygous dinr mutant for the ability of their egg chambers to undergo vitellogenesis. All egg chambers containing dinr339 (n = 11 egg chambers) or dinrE19 (n = 18 egg chambers) homozygous mutant cysts failed to progress through vitellogenesis and degenerated (Fig. 2E). In the case of dinr353, the allele with the higher level of dinr activity, only one out of six egg chambers containing homozygous mutant cysts failed to undergo vitellogenesis (8). These results suggest that the level of insulin signaling within the germ line controls vitellogenesis, revealing a direct role for DILPs in this process. Moreover, complete loss of dinr function in the germ line results in a complete block in vitellogenesis, whereas this block is partial upon ablation of brain DILP-producing cells (6). Thus, DILP5 expressed in stage 10 follicle cells likely signals in combination with brain DILPs to regulate vitellogenesis.

    We next asked whether DILPs control GSC division rate directly by binding to receptors on their surface (a cell-autonomous requirement for dinr in GSCs) or indirectly by regulating signals produced by niche cells (a non–cell-autonomous requirement). We analyzed dinr mosaic ovarioles containing one wild-type and one mutant GSC (Fig. 2F) and counted the number of wild-type versus mutant cystoblasts and cysts present in their germaria. [The relative numbers of labeled versus unlabeled cysts was not affected by early germline cyst death (table S2).] Because each cystoblast or cyst corresponds to one GSC division (fig. S1C), the ratio of mutant to wild-type cystoblasts and cysts is a measure of their relative division rates (7). For dinr339 homozygous mutant GSCs, we found a relative division rate of 0.31, whereas, for wild-type GSCs, it was 0.90 (Table 2). Similarly, the relative division rates of dinr353 and dinrE19 GSCs were 0.55 and 0.65, respectively (Table 2). Thus, dinr homozygous mutant GSCs divide more slowly than wild-type GSCs, and GSC division rate appears sensitive to the level of dinr activity. These results demonstrate that GSCs directly receive the DILP signal to regulate their division rate without mediation by the stem cell niche.

    Germline and somatic cells respond to nutritional status in a coordinated manner; however, it is unclear whether somatic cells receive the DILP signal directly (a cell-autonomous role of dinr in follicle cell proliferation) as does the germ line, or indirectly through secondary signals (a non–cell-autonomous role). We measured the percentages of dinr mutant and control follicle cells in mosaic ovarioles carrying one wild-type and one dinr mutant somatic stem cell. If follicle cells receive the DILP signal directly, the reduced level of insulin signaling in dinr mutant follicle cells should result in lower rates of proliferation (i.e., fewer mutant than control follicle cells should be observed), whereas if they receive the signal indirectly, the proliferation rates should be similar. In dinrE19 mosaic ovarioles, 51% of follicle cells were wild-type and 49% were mutant (n = 7113 follicle cells), indicating similar proliferation rates (Fig. 2G). dinr mutant follicle cells appeared to enter the endoreplicative cycle normally, but pycnotic (degenerating) nuclei and cell death were observed within dinrE19 and dinr339 mutant follicle cell clones starting at stage 8 (8). These results reveal that although a reduction in dinr activity delays germline cyst development cell autonomously, it does not cause a cell-autonomous reduction in follicle cell proliferation rate. Furthermore, in ovarioles carrying a fully dinr mutant germ line (Fig. 2E), excess follicle cells were not observed, showing that proliferation of surrounding wild-type follicle cells remains coordinated with germline growth. These results suggest that follicle cells respond indirectly to increased DILP levels through a secondary signal from the germ line. Similar degrees of coordination between germ line and soma have been observed in the presence of developmentally delayed dMyc mutant germline clones (13).

    These data demonstrate that tissue-extrinsic DILP signals can directly modify GSC proliferative activity, acting in parallel to signals from their niche. We also provide evidence that, in addition to its previously reported indirect roles in Drosophila and mammals through secondary hormonal signals (12, 14), insulin signaling plays a crucial direct role during Drosophila oogenesis in regulating not only GSC division rate but also germline cyst development rate and progression through vitellogenesis. Insulin may, therefore, have important direct roles in mammalian oogenesis. Finally, our data suggest that the coordinated response of germline and somatic cells to nutrition involves communication between these tissues. These results have broad significance, in light of the long-known effects of nutrition on human fertility (15) and of the high degree of conservation of insulin signaling functions (16, 17).

    Supporting Online Material

    Materials and Methods

    Figs. S1 and S2

    Tables S1 and S2

    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article