Stem cell dynamics in response to nutrient availability

Summary

When nutrient availability becomes limited, animals must actively adjust their metabolism to allocate limited resources and maintain tissue homeostasis [1–3]. However, it is poorly understood how tissues maintained by adult stem cells respond to chronic changes in metabolism. To begin to address this question, we fed flies a diet lacking protein (protein starvation) and assayed both germline and intestinal stem cells. Our results revealed a decrease in stem cell proliferation and a reduction in stem cell number; however, a small pool of active stem cells remained. Upon re-feeding, stem cell number increased dramatically, indicating that the remaining stem cells are competent to respond quickly to changes in nutritional status. Stem cell maintenance is critically dependent upon intrinsic and extrinsic factors that act to regulate stem cell behaviour [4]. Activation of the insulin/IGF signalling (IIS) pathway in stem cells and adjacent support cells in the germ line was sufficient to suppress stem cell loss during starvation. Therefore, our data indicate that stem cells can directly sense changes in the systemic environment to coordinate their behaviour with the nutritional status of the animal, providing a paradigm for maintaining tissue homeostasis under metabolic stress.

Results

Starvation causes a decrease in stem cell maintenance and proliferation

In response to extreme changes in the environment, such as fluctuations in temperature or food availability, some organisms are able to delay developmental and/or reproductive programs until favorable conditions resume (reviewed in [1, 2]). The ability of the nematode Caenorhabditis elegans (C. elegans) to enter into a dauer diapause in response to adverse conditions is perhaps one of the most well characterized examples at the molecular level (reviewed in [3]). However, it is poorly understood how chronic changes in nutrient availability and metabolism affect stem cell behaviour and how tissue maintenance is coordinated with an altered metabolic state. To begin to explore this question, we adapted a starvation paradigm to Drosophila melanogaster, an organism in which both stem cell behaviour and responses to changes in nutrition have been well studied.

In the Drosophila testis, germline stem cells (GSCs) and somatic stem cells, called cyst stem cells (CySCs), reside at the tip of the testis adjacent to a group of somatic cells known as the apical hub. Hub cells express and secrete the self-renewal factor Unpaired (Upd), which activates the JAK-STAT pathway in adjacent stem cells to specify stem cell maintenance [5–7]. GSCs divide with invariant asymmetry to generate one daughter cell that maintains contact with the hub and retains stem cell identity, while the other daughter cell loses contact with the hub and initiates differentiation as a gonialblast. The gonialblast will undergo 4 rounds of mitotic divisions with incomplete cytokinesis to generate a cyst of 16 interconnected spermatogonia that develop in synchrony. CySCs also self-renew and generate hub cells and cyst cells that are important for regulating maintenance and differentiation of the germ line, respectively (Figure 1A).

Figure 1. Starvation causes loss of male GSCs, which is reversed upon re-feeding.

(A) Schematic representation of the apical tip of the Drosophila testis. One cyst of germ cells is depicted progressing through mitotic amplification, meiosis and spermatid elongation. (B–C) Immunofluorescence images of testes stained with antibodies to Vasa (green) to mark germ cells and Fasiclin III (FasIII; red) to mark the hub. 4', 6-diamidino-2-phenylindole (DAPI) was used to highlight DNA (blue). Testes are from flies fed for 20 days (B) or starved for 20 days on 10% sucrose (C). (D–E) Immunofluorescence image of wild-type GFP⁺ clones in testes from flies fed (D) or starved (E) for 12 days. Testes are stained with antibodies to FasIII (red); GFP (clones), and DAPI (blue). In each testis, the number of GFP-positive GSCs was counted (arrow indicating GSC), and the number of marked cysts arising from the marked GSC was counted (arrows indicate marked clones numbered 1, 2 and 3) to assay proliferation. (F) Starvation paradigm used. Flies were analyzed following 15 or 20 days of starvation on 10% sucrose and compared to flies fed standard cornmeal molasses medium. Re-fed flies were starved for 15 days then fed for 5 days. (G) Quantification of GSCs, counted at 1 day, 15 and 20 days in fed and starved flies, and in re-fed flies (starved 15 days fed five days). Error bars: 95% confidence interval. Double asterisk: statistically significant difference using Student’s t-test (p<0.001). (H, I) Immunofluorescence images of testes stained with antibodies against Vasa to mark germ cells (green), FasIII (hub, red), and DAPI (blue). Testes are from flies starved for 15 days (H) or starved for 15 days then re-fed for 5 days (I). Scale bars, 20 µm.

To examine how stem cells respond to chronic nutrient deprivation, flies were raised under standard conditions and then switched to a diet lacking protein (protein starvation) for either 15 or 20 days. Testes of starved flies became progressively thinner over time (Figure 1B, C), and a significant decrease in the average number of GSCs per testis was observed in flies starved for 20 days (4.9 ± 0.3, n=134) when compared to testes from fed males (7.5 ± 0.3, n=123). An extension of the starvation paradigm to 32 days did not lead to an additional significant decline in the average number of GSCs/testis (20 days, 4.9 ± 0.3, n=134; 32 days, 4.4 ± 0.7, n=19). A similar decline in early cyst cells, including CySCs, was also observed upon starvation, from an average of 15 ± 0.6 per testis in starved animals (n=36) compared to an average of 28.3 ± 1.0 per testis in fed flies (n=32) (Figure S1A–D), suggesting that CySC maintenance was also affected by the chronic lack of protein in the diet. TUNEL staining to detect apoptotic cells did not reveal an increase in programmed cell death in testes from starved flies, and overexpression of the anti-apoptotic protein p35 in germ cells did not block the loss of GSCs in response to starvation (Figure S1E–H). Therefore, stem cell loss in response to protein deprivation appears to be due to direct differentiation, rather than apoptosis, although cell death due to necrosis could not be excluded.

Males fed a diet lacking protein showed a dramatic reduction in spermatogenesis (Figure 1 B, C), which could be due to a decrease in the rate of GSC proliferation, in addition to fewer GSCs. To assay GSC proliferation, ex vivo incorporation of EdU, a thymidine analog, was used to label cells in S-phase, and the percentage of EdU⁺ GSCs was calculated (S-phase index; see Experimental Procedures). The S-phase index for GSCs in fed animals was 28% (n=30), which dropped to 17% upon starvation for 20 days (n=32). As an additional strategy to assay the proliferation of GSCs, wild-type, marked (GFP⁺) GSCs were generated using FRT-mediated mitotic recombination [8], and the number of GFP⁺ germline cysts derived from the marked stem cell was quantified as an indication of the number of times the GSC divided. A significant decrease in the average number of cysts derived from GFP⁺ GSCs was observed in testes from starved males (2.2±0.4, n=23) when compared to the number of cysts in testes from fed males (4.7±0.5, n= 28) (Figure 1D, E), which is consistent with a decrease in GSC proliferation.

Effects of starvation on GSCs are reversible

The decrease in GSCs upon starvation is reminiscent of the adult reproductive diapauses (ARD) observed in C. elegans, during which many morphological changes occur, including an arrest in germline proliferation [9]. Once the animals are shifted back to favorable conditions, the soma is remodeled and germline proliferation resumes. To determine whether the effects of starvation on male GSCs are reversible, flies that were starved for 15 days were moved onto food for 5 days (re-fed) (Figure 1F).

Upon re-feeding, testes increased in size, and GSCs and spermatogonia repopulated the apical tips (Figure 1G–I). The average number of GSCs in re-fed flies (8.4 ± 0.3, n=119) was significantly higher than the average number present in testes from flies starved for 15 days (5.6 ± 0.3, n=104) (Figure 1G). In addition, the number of early cyst cells increased to levels observed in fed animals (Figure S1C, D). Therefore, when starved flies are moved back onto a protein-containing food source, somatic and germline stem cells in the testis are able to recover and respond quickly to changes in the nutritional status of the animal.

Rapid replacement of lost stem cells upon re-feeding could occur by two mechanisms: symmetric division of remaining stem cells [10] or de-differentiation of progenitor cells back to a stem cell state [11, 12]. To investigate the origin of the GSCs that are recovered following refeeding, testes from fed, starved and re-fed flies were assayed for symmetric division and de-differentiation of spermatogonia. Centrosomes in GSCs are oriented throughout the cell cycle, such that the mitotic spindle will be perpendicular to the hub to ensure asymmetric cell division [13]. Mis-positioned centrosomes would likely be observed during symmetric GSC division, as both daughter cells would maintain contact with the hub. In addition, mis-positioned centrosomes have been noted in GSCs arising from spermatogonia [12]. The relative contribution of symmetric divisions and de-differentiating germ cells was difficult to assess, as no significant increase in mis-positioned centrosomes was observed in testes from flies starved for 17 days and re-fed for 2 days, when compared to age-matched, fed controls (Figure S2A–D), although symmetric divisions were detected (Figure S2C). To assay whether GSC replacement in testes from re-fed males could be the consequence of dedifferentiation, spermatogonia in 4–16 cell cysts were permanently marked by expression of lacZ, and the number of β-galactosidase (β-gal)-positive GSCs adjacent to the hub was quantified [12]. No significant difference in the number of β-gal⁺ GSCs was detected in fed, starved, or re-fed males, indicating that GSCs in re-fed males are not derived primarily from de-differentiating spermatogonia in late stage cysts (Figure S2E, F). However, gonialblasts and spermatogonia within 2-cell cysts would not be marked using this technique.

Intestinal stem cells respond similarly to starvation and re-feeding

Given that GSCs are the only cells in the organism that can pass genetic information on to the next generation, the germ line may have adapted unique strategies to protect stem cells in response to extreme environmental changes. Alternatively, the plasticity of the germ line to nutrient availability may represent a conserved strategy utilized by other tissues to maintain a small pool of stem cells available to respond once favorable conditions resume. To determine whether the germ line is unique in its ability to respond to a chronic decrease in protein availability, we also examined the response of stem cells in the Drosophila midgut to starvation.

Tissue homeostasis in the midgut is maintained by pluripotent intestinal stem cells (ISCs), which are distributed along the basement membrane [14, 15]. Division of an ISC gives rise to one daughter cell that retains stem cell fate and another daughter cell that becomes an enteroblast (EB), both expressing a transcription factor called Escargot (Esg) (Figure 2A). Thus, expression of Esg is often used as a surrogate marker for ISCs and EBs. Daughter enteroblasts do not divide again and differentiate into either large, polyploid enterocytes that constitute the majority of the gut epithelium or small, diploid enteroendocrine cells [14, 15].

Figure 2. Starvation induces loss of ISCs/EBs in the midgut, which is reversed upon re-feeding.

(A) Schematic representation of cell types in the Drosophila posterior midgut. Division of an ISC produces one daughter cell that retains stem cell fate and another daughter cell that becomes an EB, both expressing Escargot (Esg). EBs do not divide again and differentiate into either large, polyploid enterocytes that constitute the majority of the gut epithelium or small, diploid enteroendocrine cells that express Prospero (Pros). (B) Quantification of ISCs/EBs using Esg-GFP. GFP⁺ ISCs/EBs were counted in midguts collected from newly eclosed males that were fed 1–2 days, then continuously fed or starved for 15 or 19 days, and from re-fed flies (starved 15 days then fed 4 days) as described in Experimental Procedures. n= total number of guts examined. Error bars represent 95% confidence interval of the mean. Asterisks indicate statistically significant difference using Student’s t-test (p<0.001). (C–E) Immunofluorescence images of posterior midguts from esg-GFP flies stained with antibodies against GFP to mark ISCs/EBs (green), antibodies against Arm to outline ISC/EB boundaries (red), and DAPI to mark DNA (blue). Guts are from flies fed for 15 days (C), starved for 15 days (D) or starved for 15 days then re-fed for 4 days (E). Scale bars, 20 µm. (F) Quantification of the average number of pHH3⁺ cells per posterior midgut under each feeding paradigm. Results are presented as 95% confidence interval of the mean. “n” in parentheses indicated total number of guts counted.

One to two day-old flies expressing green fluorescent protein under control of the escargot promoter (esg-GFP) were shifted to starvation conditions for 15 days, and the number of GFP⁺ ISCs/EBs was quantified (see Experimental Procedures for details). A significant decrease in the average number of GFP⁺ cells was observed in guts from flies starved for 15 days, when compared to guts from fed controls (Figure 2B–D). Upon re-feeding for 4 days, guts increased in size, and the average number of GFP⁺ cells increased significantly (Figure 2B, E), when compared to the average number present in guts from flies starved for 15 days (Figure 2B). Importantly, the average number of GFP⁺ cells was comparable to the average number in age-matched controls that were never starved (Figure 2B).

As ISCs are the only proliferative cells in the midgut, we assayed ISC behavior upon starvation and in response to re-feeding by labeling with the mitotic marker phosphorylated histone H3 (pHH3) (Figure 2F). In fed flies, the average number of pHH3⁺ cells per posterior midgut was 22.8 ± 2.0 (n=27), while in starved flies the average number was 6.0 ± 1.0 (n=24). Upon re-feeding starved flies for 24 hrs, the average number of mitotic ISCs per gut increased to 40.0 ± 4.8 (n=22). Therefore, our data suggest that both GSCs and ISCs respond to a chronic lack of protein by reducing the pool of stem cells available for tissue homeostasis and repair. However, once normal growth conditions resume, the number of stem cells rapidly increases, returning to the approximate number available before onset of starvation.

Insulin signalling acts cell-autonomously to regulate male GSC behavior

Insulin/IGF signalling (IIS) is a well-characterized regulator of nutrient signalling and longevity, and emerging evidence from a number of systems indicates that IIS activity plays an important role in regulating the behaviour of tissue stem cells in multiple species [16–24]. Therefore, we speculated that insulin signalling could provide a link between changes in nutritional status and altered stem cell behaviour.

The D. melanogaster genome encodes a single insulin-like receptor (dInR), activation of which culminates in phosphorylation and inactivation of the transcription factor dFOXO. In addition, there are seven insulin-like peptides (dILPs) in three of these, dILP2, dILP3 and dILP5, are expressed in insulin producing cells (IPCs) in the brain [25, 26] and secreted into the haemolymph where they coordinate the response of cells throughout the organism to nutritional conditions. Immunofluorescence microscopy revealed expression of dInR at the tip of the testis in GSCs, as well as in spermatogonia and early cyst cells (Figure 3A–C), indicating that both early germline and somatic cells are competent to respond directly to dILPs and initiate signalling via dInR. In addition, strong expression was detected in the hub, with enrichment at the interface between the hub and GSCs (Figure 3A, C), suggesting that hub cells could respond directly to insulin signalling and act as a sensor of dILPs to coordinate changes in stem cell behaviour in response to metabolic flux.

Figure 3. dInR is required for maintenance of male GSCs.

(A–C) Immunofluorescence image of testes in which spermatogonia are expressing GFP (green; A,B), stained with antibodies that recognize dInR (red; A,C). dInR is expressed in germ cells including GSCs (GFP⁺, arrows), in somatic cells (GFP⁻, arrowheads) and in the hub (asterisk). Genotype: w; UAS-GFP^mCD8; nanos-GAL4. Scale bars, 20 µm. (D–G) Staining of control (FRT82B) (D,F) or dInR³³⁹/dInR³³⁹ (E,G) marked (GFP⁺) clones at 3 days (D,E) or 10 days (F,G) post-heat shock (PHS). FasIII (red) marks the hub, GFP (green) marks clones, and DAPI (blue) marks DNA. GFP channel alone is shown in insets. GSC clones (GFP⁺, arrows) are not maintained at 10 days PHS if homozygous for dInR³³⁹ (G). Scale bars, 10 µm. (H) Quantification of the number of testes that contain one or more GFP⁺ GSCs at 3 or 10 days PHS, compared to the total number of testes scored (shown in parentheses), as percentages for each genotype.

To test whether GSCs respond directly to insulin signalling, FRT-mediated mitotic recombination was used to generate positively-marked GSCs that are homozygous for either of two mutant alleles of dInR: dInR^E19, a hypomorphic allele and dInR³³⁹, a null allele [25, 27]. Heat-shocks were used to induce recombination in 1–2 day old males, and testes were dissected either 3 or 10 days post-heat shock (PHS). Although both wild-type and dInR mutant GSC clones were observed at 3 days PHS (Figure 3D, E and H), dInR mutant GSCs were not maintained (Figure 3F–H). At 10 days PHS, the percentage of testes containing wild-type marked GSCs was 45%, and marked GSCs continued to divide to produce marked germ cells along the length of the testis. In contrast, at 10 days PHS only 20% of testes examined contained a marked GSC homozygous for dInR^E19 and only 2% contained a marked GSC homozygous for dInR³³⁹ (Figure 3H), indicating a critical cell-autonomous requirement of dInR for the maintenance of male GSCs. The absolute survival of germ cells, however, does not require dInR, as rare mutant germ cell cysts continued to develop (data not shown).

Expression of dILP5 and dILP2, the most abundantly expressed dILPs in IPCs, decreased in flies starved for 20 days (Figure S3A), as assayed by qRT-PCR. In contrast, dILP3 expression was increased (Figure S3A), which is consistent with the reported increase in dILP3 as a response to loss of dILPs 2 and 5 [28]. We also assayed expression of the translational inhibitor 4E-binding protein (4E-BP), as it is a well-characterized target of dFOXO [29]. Consistent with a drop in insulin signalling and activation of dFOXO, 4E-BP transcript levels increased in the heads of flies starved for 15 or 20 days (Figure S3B). Re-feeding caused an increase in dILP expression levels (Figure S3A) and a decrease in 4E-BP (Figure S3B), reversing the gene expression changes that occurred during starvation.

Constitutive activation of insulin signalling suppresses GSC loss during starvation

Given that starvation leads to decreased dILP expression and that GSCs require insulin signalling for maintenance, we wanted to directly test whether decreased insulin signalling could be responsible for the loss of GSCs observed in starved males. The bipartite GAL4-UAS system was used to express a constitutively active form of dInR (dInR^CA) within germ cells and hub cells to determine whether expression of this activated form of dInR is sufficient to suppress GSC loss during starvation. While expression of dInR^CA in early germ cells (nanosGAL4) led to a partial rescue of GSC loss in flies starved for 20 days, expression of dInR^CA in both early germ cells and hub cells (updGAL4; nanosGAL4) resulted in a significant suppression of GSC loss when compared to outcrossed controls (Figure 4 and Supplemental Table S1). Similar data were obtained when an activated form of the Drosophila PI3 kinase (Dp110^CAAX), which acts downstream of dInR, was expressed in both hub and early germ cells (Figure 4 and Supplemental Table S1). Therefore, InR signalling can directly regulate GSC maintenance under normal, homeostatic conditions, while hyperactivation of InR signalling in stem cells and adjacent support cells is sufficient to block GSC loss in response to starvation. Interestingly, our data indicate that although GSCs can receive and respond to InR signalling directly, hub cells likely play an important role in sensing dILP levels and, through secondary signals, coordinating changes in GSC behaviour in response to nutritional status.

Figure 4. Simultaneous expression of activated dInR or Dp110 in the hub and germ cells suppresses GSC loss upon starvation.

Average number of GSCs per testis from flies fed (dark green) or starved (light green) for 20 days were plotted. Studies are listed across the bottom. Genotypes, average GSC numbers, and the total number of testes analyzed for each study are summarized in Supplemental Table S1. We used nanos-GAL4 to drive expression of transgenes (UAS-dInR^CA and UAS-Dp110^CAAX) in germ cells; upd-GAL4, in hub cells; and a combination of the two drivers (upd, nanos-GAL4) to confer simultaneous expression in both germ cells and hub cells. The difference between fed and starved flies with simultaneous expression of dInR^CA or Dp110^CAAX in the hub and germ cells is significantly smaller than for either of their control studies [analyzed by two way analysis of variance (ANOVA), p<0.005]. Asterisks indicate statistically significant differences at p<0.005. Error bars represent 95% confidence interval of the mean.

Discussion

As stem cells play a critical role during development, tissue homeostasis and wound repair, it is likely that stem cell behaviour is regulated either directly or indirectly by systemic signals to coordinate an appropriate, tissue-specific response to metabolic stress. Our data are consistent with a model whereby insulin signalling may act as a conserved mechanism to regulate the decline in stem and progenitor cells upon chronic nutrient deprivation to match the new baseline metabolic state.

Drosophila oogenesis is also sensitive to nutritional conditions [16], and dInR regulates female GSC proliferation directly and maintenance indirectly via the niche [17, 30]. However, no evidence for a cell-autonomous requirement for dInR in the maintenance of female GSCs or ICSs has been reported. Therefore, although insulin signalling appears to regulate both germline and intestinal stem cell proliferation, our study provides the first evidence that insulin signalling is required autonomously for stem cell maintenance.

Interestingly, not all stem cells are lost as a consequence of protein deprivation. Similarly, a fixed number of GSCs appeared to remain after 70 days of starvation during C. elegans ARD [9]. Whether the remaining stem cells possess a more resilient stress response pathway and represent a more robust stem cell pool has yet to be determined; however, our data indicate that resistance of a small pool of stem cells to nutritional stress is not unique to the germ line. Preservation of a minimum number of stem cells to ensure adequate recovery of a tissue upon restoration of favourable conditions could prove to be an evolutionary strategy that is utilized by animals to survive conditions of extreme environmental stress. We predict that several mechanisms must be in place in order to preserve a small, but active, pool of tissue stem cells to maintain tissue homeostasis during chronic changes in metabolism. Loss of stem cells in response to changes in nutrition could be stochastic or occur by a selection process during which some stem cells are culled, while the most robust stem cells are maintained. In addition, mechanisms must be in place to protect the remaining stem cells in such a way that they are able to respond rapidly to provide a pool of progenitor cells to coordinate tissue regeneration with improved conditions. Future studies elucidating conserved mechanisms by which stem cell behaviour is regulated in response to stress will provide insights into the therapeutic use of stem cells in regenerative medicine, particularly for individuals who suffer from metabolic diseases.

Highlights

• Starvation results in a decline in germline and intestinal stem cells

• Loss of germline stem cells is suppressed by activation of InR signaling

• InR signalling acts autonomously in male GSCs to regulate stem cell behaviour

• Adult stem cell behaviour is coordinated with the metabolic state of the organism